Method for combined analysis of hyaluronic acid and multi-active ingredients in medical and cosmetic adjuvants

By using a two-dimensional chromatography system based on a dual-column switching valve and FPGA parallel spectrum interpretation, the problem of simultaneous analysis of hyaluronic acid and active ingredients in medical aesthetic excipients has been solved. This system achieves efficient and accurate determination of molecular weight distribution and concentration, overcoming the long analysis cycle and data processing bottlenecks in existing technologies.

CN122709649BActive Publication Date: 2026-11-10中检华通威国际检验(苏州)有限公司
View PDF 1 Cites 0 Cited by

Patent Information

Application Number
CN202611201237.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2026-08-10
Publication Date
2026-11-10
Estimated Expiration
2046-08-10

AI Technical Summary

Technical Problem

In existing methods for analyzing medical aesthetic excipients, the molecular weight distribution of hyaluronic acid and the concentration of active ingredients are determined separately, resulting in long analysis cycles, high sample consumption, and an inability to reveal the coexistence relationship between hyaluronic acid and active ingredients in the molecular weight dimension. Two-dimensional chromatography systems suffer from problems such as solvent incompatibility, phase shift of slices, and overlapping co-elution spectra that cannot be resolved in real time.

Method used

A two-dimensional chromatography system based on a dual-column switching valve and FPGA parallel spectrum interpretation was adopted. The system uses a size exclusion column and a hydrophilic interaction column connected in series, and an online solvent conversion unit is set between the switching valve and the hydrophilic interaction column. Combined with the FPGA parallel spectrum interpretation unit for signal processing, the molecular weight distribution of hyaluronic acid and the concentration of multiple active ingredients can be determined simultaneously.

Benefits of technology

This method enables simultaneous analysis of hyaluronic acid molecular weight distribution and the concentration of multiple active ingredients, shortening the analysis cycle, reducing sample consumption, and improving calculation accuracy and data processing efficiency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122709649B_ABST
    Figure CN122709649B_ABST
Patent Text Reader

Abstract

The present application relates to the technical field of data analysis and processing, and further relates to a combined analysis method for hyaluronic acid and multiple active ingredients in medical and beauty adjuvants, comprising the following steps: step 1, mixing medical and beauty adjuvant samples with mobile phase diluent to obtain a test solution, and injecting the test solution into a two-dimensional chromatography system based on a double-column switching valve and FPGA parallel spectrum analysis after filtering through a microporous filter membrane; step 2, separating the test solution through a size exclusion chromatography column according to the order of hydrodynamic volume from large to small; and step 3, extracting the number average molecular weight, weight average molecular weight and molecular weight dispersity of hyaluronic acid from the first-dimensional flow curve digital signal by a result extraction module. The present application realizes the simultaneous completion of the characterization of the molecular weight distribution of hyaluronic acid and the determination of the concentration of multiple active ingredients in a single injection, and solves the problems of solvent incompatibility, slice phase offset and overlapping spectrum in real-time analysis in the prior art.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the field of data analysis and processing technology, specifically relating to a combined analytical method for hyaluronic acid and multiple active ingredients in medical aesthetic excipients. Background Technology

[0002] Medical aesthetic excipients are functional formulations with hyaluronic acid as the core matrix and compounded with various active ingredients. Their quality control requires simultaneous acquisition of the molecular weight distribution parameters of hyaluronic acid and the concentration values ​​of each active ingredient. Existing analytical methods typically separate the characterization of hyaluronic acid's molecular weight from the determination of the content of active ingredients. The molecular weight distribution of hyaluronic acid is generally determined using size exclusion chromatography with a tandem differential refractive index detector or a multi-angle laser scattering detector, while the content of active ingredients is generally determined item by item using reversed-phase liquid chromatography or hydrophilic interaction chromatography with a tandem ultraviolet detector. These two independent methods each require separate sample pretreatment, injection, and data processing workflows, resulting in long analysis cycles, large sample consumption, and an inability to reveal the coexistence relationship between hyaluronic acid and active ingredients in the molecular weight dimension. In recent years, two-dimensional liquid chromatography (2D-LC) technology has been introduced into the field of complex sample analysis. By setting a switching valve at the outlet of the first-dimensional column to transfer the effluent in segments to the second-dimensional column for orthogonal separation, separation information in both dimensions can be obtained simultaneously in a single injection. However, existing 2D-LC systems still face three shortcomings when applied to the analysis of medical aesthetic excipients. First, when aqueous effluent from size exclusion columns directly enters hydrophilic interaction columns, solvent incompatibility issues arise. Hydrophilic interaction columns require an initial mobile phase with an organic solvent content exceeding 70%. Aqueous effluent disrupts the retention mechanism, preventing effective retention of active ingredients. Second, the difference in dead volume between the detection and sampling branches causes a systematic phase shift between the slicing time and the detection signal. Without correction, the correspondence between effluent fragments and molecular weight scales will be inaccurate, affecting the accuracy of hyaluronic acid molecular weight distribution calculations. Third, multiple active ingredients may co-elute in second-dimensional chromatography. Traditional single-wavelength quantitative methods cannot accurately resolve overlapping peaks, while conventional software-based multivariate curve resolution algorithms are limited by the serial computing architecture of general-purpose processors, making it difficult to complete real-time decoupling of full-wavelength spectral data within the slice modulation cycle, thus failing to meet the timeliness requirements of online coupled analysis. Summary of the Invention

[0003] The main objective of this invention is to provide a combined analytical method for hyaluronic acid and multiple active ingredients in medical aesthetic excipients. This method enables the simultaneous characterization of the molecular weight distribution of hyaluronic acid and the determination of the concentration of multiple active ingredients in a single injection, and solves the problems of solvent incompatibility, phase shift of sections, and overlapping co-elution spectra in existing technologies that prevent real-time analysis.

[0004] To solve the above problems, the technical solution of the present invention is implemented as follows: A method for the combined analysis of hyaluronic acid and multiple active ingredients in medical aesthetic excipients, comprising the following steps: Step 1: Add the medical aesthetic excipient sample to the mobile phase dilution solution and mix well to obtain the test solution. After filtering the test solution through a microporous membrane, inject it into a two-dimensional chromatography system based on a dual-column switching valve and FPGA parallel spectrum interpretation. Step 2: The test solution is separated by a size exclusion column in descending order of fluid dynamic volume. An evaporative light scattering detector continuously detects the effluent from the size exclusion column and outputs a simulated signal of the first-dimensional effluent curve. The FPGA parallel resolution unit samples the simulated signal of the first-dimensional effluent curve to obtain a digital signal of the first-dimensional effluent curve. The FPGA parallel resolution unit performs first-dimensional coupled slice window recognition on the digital signal of the first-dimensional effluent curve and continuously sends switching pulses to the ten-way two-position electric switching valve within the first-dimensional coupled slice window to segment the effluent into effluent fragments. After second-dimensional separation by a hydrophilic column, the effluent fragments are scanned at full wavelength by a diode array detector to output a digital measured absorbance vector. The stepwise stripping spectral decoupling pipeline in the FPGA parallel resolution unit performs multi-active component stepwise spectral stripping on the digital measured absorbance vector to obtain the response intensity value of each active component. The data cube assembly module assembles the response intensity values ​​of each active component into a three-dimensional data cube of molecular weight-retention time-component according to molecular weight coordinates, retention time, and component number. Step 3: The result extraction module extracts the number-average molecular weight, weight-average molecular weight, and molecular weight dispersion of hyaluronic acid from the digital signal of the first-dimensional effluent curve, and extracts the concentration value of each active ingredient from the three-dimensional data cube of molecular weight-retention time-component, and outputs it as the combined analysis result.

[0005] Furthermore, the two-dimensional chromatography system based on a dual-column switching valve and FPGA parallel spectrum interpretation includes a one-dimensional infusion pump, an injector, a size exclusion column, an evaporative light scattering detector, a 10-port 2-position electric switching valve, sampling loops, an online solvent conversion unit, a two-dimensional infusion pump, a hydrophilic interaction column, a diode array detector, and an FPGA parallel spectrum interpretation unit. The one-dimensional infusion pump, injector, and size exclusion column are connected sequentially. The one-dimensional infusion pump delivers an aqueous phase containing volatile buffer salts to the injector as the one-dimensional mobile phase. The outlet of the size exclusion column is connected to the inlet of the evaporative light scattering detector and the injection port of the 10-port 2-position electric switching valve via a three-way tubing. The 10-port 2-position electric switching valve has two sampling loops of the same volume, and its outlet port is connected to the inlet of the hydrophilic interaction column via the online solvent conversion unit. The two-dimensional infusion pump delivers a water-acetonitrile mixture with a higher acetonitrile volume ratio than the one-dimensional mobile phase to the online solvent conversion unit as the two-dimensional mobile phase. The diode array detector is connected to the outlet of the hydrophilic interaction column.

[0006] Furthermore, the diode array detector includes a diode array photoelectric conversion module and a multi-channel analog-to-digital conversion module. The multi-channel analog-to-digital conversion module converts the absorbance signals of each wavelength channel into digitally measured absorbance vectors. The FPGA parallel spectrum resolution unit is connected to the analog output terminal of the evaporative light scattering detector through the analog-to-digital conversion channel. The FPGA parallel spectrum resolution unit is connected to the data output terminal of the multi-channel analog-to-digital conversion module through a high-speed digital data interface. The digital pulse output port of the FPGA parallel spectrum resolution unit is connected to the drive interface of the ten-way two-position electric switching valve.

[0007] Furthermore, the on-chip block random access memory of the FPGA parallel spectrum interpretation unit is pre-written with the pure reference spectra of each active ingredient, the concentration standard curves of each active ingredient, the molecular weight standard curve of hyaluronic acid, the response correction curve of hyaluronic acid, and the phase compensation amount. The hyaluronic acid response correction curve is obtained by calibrating the evaporative light scattering response value of a series of hyaluronic acid mass concentration standards on a two-dimensional chromatography system based on a dual-column switching valve and FPGA parallel spectrum interpretation with known mass concentrations. The phase compensation amount is jointly calibrated by the dead volume of the detection branch between the outlet of the size exclusion chromatography column and the detection cell of the evaporative light scattering detector, the dead volume of the sampling branch between the outlet of the size exclusion chromatography column and the inlet of the sampling loop, and the one-dimensional mobile phase flow rate.

[0008] Furthermore, in step 2, the first-dimensional slicing window consists of two sub-windows: a hyaluronic acid elution window and a preset active ingredient elution window. The peak detection state machine in the FPGA parallel spectral unit sends the current sampled value of the first-dimensional elution curve digital signal and the previous sampled value into a subtractor to obtain a difference value. When the difference value changes from negative to positive and exceeds a preset noise threshold, the current sampling time is determined to be the start time of the hyaluronic acid elution window. When the difference value changes from positive to negative, the current sampling time is determined to be the peak time of the hyaluronic acid elution window. When the absolute value of the difference value is continuously lower than the preset noise threshold and maintains a preset hold count, the current sampling time is determined to be the end time of the hyaluronic acid elution window. The retention time interval of each active ingredient on the size exclusion column is used as the start and end times of the preset active ingredient elution window and is pre-written into the on-chip block random access memory. The peak detection state machine sequentially writes the start time, peak time, end time of the hyaluronic acid elution window, and the start and end times of the preset active ingredient elution window into the elution window time register group.

[0009] Furthermore, in step 2, the valve-controlled timing generator in the FPGA parallel despectroscopy unit uses the starting time of each sub-window in the first-dimensional combined slicing window, superimposed with the phase compensation amount, as the starting time of each sub-window. It continuously outputs switching times within the first-dimensional combined slicing window according to a preset modulation period until the current switching time reaches the ending time of each sub-window, superimposed with the phase compensation amount. At each switching time, the valve-controlled timing generator sends one switching pulse to the drive interface of the ten-way two-position electric switching valve through the digital pulse output port. The ten-way two-position electric switching valve alternates between the sampling position and the injection position each time it receives a switching pulse, causing the two sampling loops to alternately be in the sampling state and the injection state. The sampling loop in the sampling state continuously retains one effluent fragment from the size exclusion column for one modulation period. The effluent fragment in the sampling loop in the injection state is mixed with the two-dimensional mobile phase by the online solvent conversion unit and then enters the hydrophilic interaction column for second-dimensional separation. The online solvent conversion unit adjusts the acetonitrile volume ratio in the effluent fragment to ensure that the effluent fragment meets the initial retention conditions of the hydrophilic interaction column.

[0010] Furthermore, in step 2, the valve-controlled timing generator writes the phase difference between each switching moment and the peak moment of the hyaluronic acid elution window as the molecular weight positioning phase difference of the corresponding effluent fragment into the elution window timing register group. Since the phase compensation amount has been pre-calibrated to a unified phase reference to calibrate the transmission time difference between the detection branch and the sampling branch, a one-to-one correspondence is formed between the molecular weight positioning phase difference of each effluent fragment and the molecular weight scale indicated by the peak moment, so that each effluent fragment is locked to a certain molecular weight range on the size exclusion chromatography column, and the center molecular weight scale of the locked molecular weight range is used as the molecular weight coordinate of the corresponding effluent fragment.

[0011] Furthermore, in step 2, the stepwise stripping spectral decoupling pipeline includes a correlation ranking controller and multiple tandem stripping stages, the total number of stripping stages equal to the total number of active ingredient pure reference spectra; before stripping begins, the correlation ranking controller performs normalized cross-correlation calculations on each active ingredient pure reference spectrum and the digitized measured absorbance vector to obtain the initial correlation value of each active ingredient, and generates an active ingredient stripping sequence index table according to the order of the initial correlation values ​​from largest to smallest. Each tandem stripping stage sequentially retrieves the corresponding active ingredient pure reference spectrum from the on-chip block random access memory according to the active ingredient stripping sequence index table; the digitized measured absorbance vector serves as the first tandem stripping stage. The initial input residual absorbance vector for each stripping stage; the hardware matching core in each stripping stage performs a normalized cross-correlation operation on the input residual absorbance vector and the active ingredient pure reference spectrum called by the stage, and outputs the response intensity value of the active ingredient at the current retention time point. The hardware stripping core in each stripping stage removes the spectral contribution of the active ingredient of the stage from the input residual absorbance vector according to the response intensity value and the active ingredient pure reference spectrum called by the stage to obtain a new residual absorbance vector and pass it to the next stripping stage; each stripping stage runs in parallel in a pipeline manner. When the first stripping stage processes the data of the current effluent fragment, the second stripping stage processes the data of the previous effluent fragment at the same time.

[0012] Furthermore, in step 2, the data cube assembly module writes the response intensity values ​​of each active ingredient output from the stepwise stripping spectral decoupling pipeline into the on-chip dual-port random access memory according to three dimensions: molecular weight coordinates, retention time, and ingredient number. The molecular weight coordinates are calculated by combining the molecular weight positioning phase difference of each effluent fragment written into the elution window time register group by the valve-controlled timing generator with the standard curve of hyaluronic acid molecular weight. After writing, the data stored in the on-chip dual-port random access memory constitutes a three-dimensional data cube of molecular weight-retention time-component.

[0013] The present invention provides a method for the coupled analysis of hyaluronic acid and multiple active ingredients in medical aesthetic excipients, which has the following beneficial effects: The method provided by the present invention constructs a two-dimensional chromatographic system by connecting size exclusion chromatography and hydrophilic interaction chromatography in series with a ten-way two-position electric switching valve, and sets an online solvent conversion unit between the switching valve and the hydrophilic interaction chromatography column. This solves the solvent incompatibility problem between the first-dimensional aqueous phase effluent and the initial retention conditions of the second-dimensional hydrophilic interaction chromatography, so that the active ingredients are effectively retained and separated on the second-dimensional chromatographic column. This enables the characterization of the molecular weight distribution of hyaluronic acid and the determination of the concentration of multiple active ingredients to be completed simultaneously in a single injection, which significantly shortens the analysis cycle and reduces the sample consumption. This invention pre-calibrates and writes phase compensation values ​​into the FPGA parallel interpretation unit, aligning the transmission time difference between the detection and sampling branches to a unified phase reference. This ensures that each effluent fragment is accurately locked to a specific molecular weight range on the size exclusion column, eliminating the systematic offset of molecular weight coordinates caused by differences in tubing dead volume and improving the accuracy of calculating the number-average molecular weight, weight-average molecular weight, and molecular weight dispersion of hyaluronic acid. Furthermore, this invention constructs a step-by-step spectral decoupling pipeline within the FPGA parallel interpretation unit. Utilizing the parallel processing capabilities of the hardware pipeline, it completes full-wavelength spectral decoupling of all active ingredients within each modulation cycle. This overcomes the bottleneck of general-purpose processors' inability to perform real-time interpretation within a modulation cycle, enabling simultaneous acquisition and analysis of two-dimensional chromatographic data. This achieves a fully online and automated process from sample injection to the output of coupled analysis results. Attached Figure Description

[0014] Figure 1 The diagram shows the principle of two-dimensional chromatographic orthogonal separation contour lines obtained after the two-dimensional chromatographic system based on dual-column switching valve and FPGA parallel spectrum interpretation provided in the embodiment of the present invention is used to perform coupled analysis on medical aesthetic excipient samples. Figure 2 This is a schematic diagram of the FPGA parallel spectrum descaling unit performing peak detection and combined slice window recognition on the first-dimensional outflow curve digital signal, as provided in an embodiment of the present invention. Figure 3 A schematic diagram illustrating the phase compensation and slice timing in a two-dimensional chromatography system based on a dual-column switching valve and FPGA parallel spectroscopy, provided in an embodiment of the present invention. Figure 4 The molecular weight distribution curve of hyaluronic acid provided in the embodiments of the present invention. Detailed Implementation

[0015] A method for the combined analysis of hyaluronic acid and multiple active ingredients in medical aesthetic excipients, comprising the following steps:

[0016] Step 1: Add the medical aesthetic excipient sample to the mobile phase dilution solution and mix well to obtain the test solution. After filtering the test solution through a microporous membrane, inject it into a two-dimensional chromatography system based on a dual-column switching valve and FPGA parallel spectrum interpretation. Step 2: The test solution is separated by a size exclusion column in descending order of fluid dynamic volume. An evaporative light scattering detector continuously detects the effluent from the size exclusion column and outputs a simulated signal of the first-dimensional effluent curve. The FPGA parallel resolution unit samples the simulated signal of the first-dimensional effluent curve to obtain a digital signal of the first-dimensional effluent curve. The FPGA parallel resolution unit performs first-dimensional coupled slice window recognition on the digital signal of the first-dimensional effluent curve and continuously sends switching pulses to the ten-way two-position electric switching valve within the first-dimensional coupled slice window to segment the effluent into effluent fragments. After second-dimensional separation by a hydrophilic column, the effluent fragments are scanned at full wavelength by a diode array detector to output a digital measured absorbance vector. The stepwise stripping spectral decoupling pipeline in the FPGA parallel resolution unit performs multi-active component stepwise spectral stripping on the digital measured absorbance vector to obtain the response intensity value of each active component. The data cube assembly module assembles the response intensity values ​​of each active component into a three-dimensional data cube of molecular weight-retention time-component according to molecular weight coordinates, retention time, and component number. Step 3: The result extraction module extracts the number-average molecular weight, weight-average molecular weight, and molecular weight dispersion of hyaluronic acid from the digital signal of the first-dimensional effluent curve, and extracts the concentration value of each active ingredient from the three-dimensional data cube of molecular weight-retention time-component, and outputs it as the combined analysis result.

[0017] Medical aesthetic excipients are a special type of formulation that falls between cosmetics and medical devices. Their core matrix is ​​typically hyaluronic acid, upon which various biologically active auxiliary ingredients are often added, depending on the product's positioning. These include niacinamide, ascorbic acid and its derivatives, arbutin, tranexamic acid, oligopeptides, epidermal growth factor, dipotassium glycyrrhizate, and asiaticoside, to achieve synergistic functions such as moisturizing, whitening, anti-inflammation, repair, or anti-aging. Because hyaluronic acid itself is a polydisperse macromolecule, its molecular weight distribution directly determines the product's permeability, water retention, and viscoelasticity in the dermis. Furthermore, the various coexisting small molecule or polypeptide active ingredients each have independent mass concentration indicators. Therefore, for a finished or intermediate medical aesthetic excipient, complete release analysis requires obtaining not only the number-average molecular weight, weight-average molecular weight, and molecular weight dispersion of hyaluronic acid, but also the mass concentration of each active ingredient—this is the technical objective of coupled analysis. The following section details the entire preparation process from sample delivery to entry into the two-dimensional chromatography system.

[0018] Medical aesthetic excipients commonly come in three forms: gel-like serums, aqueous ampoules, and emulsions. Gel-like serums, due to their high hyaluronic acid concentration, exhibit rheological behavior similar to shear-thinning fluids; direct injection can contaminate the flow path and clog the column sieve. Aqueous ampoules typically contain significant amounts of alcohol solvents and penetrants; direct injection can create a significant refractive index difference and solvent front with the aqueous mobile phase of the size exclusion column, causing baseline disturbances near the hyaluronic acid elution window. Emulsions contain emulsifiers and an oil phase; oil droplets entering the size exclusion column can irreversibly adsorb onto the cross-linked dextran or cross-linked polyhydroxymethyl acrylate backbone of the column packing. Considering the characteristics of these three types of samples, dilution and filtration must be performed before injection into the two-dimensional chromatography system. The purpose is not to reduce the sample concentration to accommodate the detector's dynamic range, but rather to ensure that the sample's viscosity, ionic strength, and solvent composition are as consistent as possible with the subsequent one-dimensional mobile phase, thereby eliminating the interference of the injection solvent effect on the first-dimensional peak shape. This consideration is repeatedly emphasized in Snyder et al.'s summary of the solvent effect in liquid chromatography. The inconsistency between the composition of the sample solvent and the mobile phase is the primary cause of frontal distortion and peak broadening.

[0019] refer to Figure 1 , Figure 1 The horizontal axis represents the first dimension of retention time, i.e., the retention time in size exclusion chromatography, measured in minutes; the vertical axis represents the second dimension of retention time, i.e., the retention time in hydrophilic interaction chromatography, measured in seconds. The density of the contour lines represents the detection response intensity at the corresponding positions; the denser the contour lines, the stronger the response. Figure 1 A continuous band extending along the horizontal axis appears in the left-hand region, covering a range of approximately 4 to 11 minutes. The vertical axis is concentrated within an extremely short retention range of 2 to 4 seconds. This band corresponds to hyaluronic acid, a polydisperse macromolecule that elutes sequentially in descending order of hydrodynamic volume on size exclusion columns. Hyaluronic acid segments of different molecular weights occupy different retention time positions in the first dimension, thus forming a broad band along the horizontal axis. Hyaluronic acid is almost not retained on hydrophilic interaction columns because its molecular weight is too large to effectively penetrate the water-rich layer on the hydrophilic interaction stationary phase surface for partitioning; therefore, it is concentrated near the dead time position along the vertical axis. Figure 1 Multiple independent spots appeared in the right-hand region, distributed within a range of approximately 16 to 24 minutes on the horizontal axis and approximately 5 to 20 seconds on the vertical axis. These spots correspond to various active ingredients, including nicotinamide, ascorbic acid phosphate, arbutin, tranexamic acid, and dipotassium glycyrrhizate. The active ingredients are small molecules with a hydrodynamic volume much smaller than hyaluronic acid, resulting in longer retention times on size-exclusion columns, hence their appearance on the right side of the horizontal axis. The different active ingredients have varying retention times on hydrophilic interaction columns due to their polarity differences, thus separating them on the vertical axis. Figure 1 The distribution band of hyaluronic acid and the spot cluster of active ingredients occupy completely different regions on the two-dimensional plane and do not overlap, which intuitively reflects the positive and complementary relationship between the two separation modes of size exclusion chromatography and hydrophilic interaction chromatography. Figure 1 The dashed boxes in the figure mark the hyaluronic acid elution window and the active ingredient elution window, respectively. The hyaluronic acid elution window corresponds to the first-dimensional retention time interval of approximately 3.5 minutes to 11.5 minutes on the horizontal axis, and the active ingredient elution window corresponds to the first-dimensional retention time interval of approximately 16 minutes to 24 minutes on the horizontal axis. These two windows are the target intervals for the subsequent FPGA parallel spectral descrambling unit to perform the first-dimensional combined slice window identification.

[0020] The specific dilution process is as follows. Take samples of medical aesthetic excipients. to Accurately weigh to Add the mobile phase diluent to the mark in the volumetric flask to obtain the initial diluent. The sample volume and volumetric flask specifications here can be adjusted flexibly according to the nominal concentration of hyaluronic acid in the sample. If the nominal concentration of hyaluronic acid in the sample is within... The following methods can be used to increase the sample weight to... And changed to Volumetric flask; if the nominal hyaluronic acid content in the sample is within The above can be followed first. right Perform a first dilution at the specified ratio, then transfer a portion from the diluent. join in A second dilution was performed in a volumetric flask to obtain the final test solution. The reason for the two-step dilution is that hyaluronic acid forms a semi-dilute solution with entangled molecular chains under high concentration conditions. Direct single-step high-fold dilution will create a local concentration gradient instantaneously, causing the molecular chains to slowly return to their free coil conformation. If the sample is injected before equilibrium is reached, the first-dimensional elution curve will show a shift in peak position and a broadening of peak width. The two-step stepwise dilution allows the molecular chains sufficient relaxation time after each dilution.

[0021] The composition of the mobile phase diluent is strictly consistent with the one-dimensional mobile phase used in subsequent size exclusion column chromatography. The one-dimensional mobile phase is an aqueous phase containing volatile buffer salts, with a typical formulation as follows: to Ammonium acetate aqueous solution, or to The pH value of the ammonium formate aqueous solution is controlled at . to The core reason for choosing a volatile buffer salt system instead of a conventional phosphate system lies in the detection mechanism of the evaporative light scattering detector. In the detection cell, the mobile phase, after atomization, needs to be completely evaporated in a heated drift tube. The remaining solute particles are then collected by a photomultiplier tube after laser scattering. Non-volatile buffer salts would form a constant background of particles in the drift tube, completely drowning out the sample signal. Volatile ammonium salts, on the other hand, have a drift tube temperature set at [temperature value missing]. to It can be completely decomposed into ammonia, carbon dioxide, or formic acid and discharged with the carrier gas without forming background. The pH value is chosen in the weakly acidic range rather than neutral or weakly alkaline because each disaccharide repeating unit on the hyaluronic acid backbone carries one carboxyl group. Under weakly acidic conditions, the carboxyl groups are partially protonated, and the molecular chain has moderate flexibility, making it less likely to cause ion repulsion or adsorption on the surface of the size-exclusion chromatography column packing material; if the pH value is too high... In summary, complete dissociation of the carboxyl group causes excessive extension of the molecular chain, resulting in an artificially high hydrodynamic radius, and the linear range of the standard curve will deviate towards higher molecular weights.

[0022] The selection of microporous filter membranes is based on pore size. to Polyethersulfone (PES) or regenerated cellulose (RCC) membranes are preferred. Polyvinylidene fluoride (PVDF) and nylon membranes should be avoided. PVDF membranes exhibit significant hydrophobic adsorption of some oil-soluble excipients and small-molecule peptides, leading to a decrease in the recovery rate of active ingredients in the analyte. Nylon membranes release small amounts of oligomers under weakly acidic conditions, contaminating the chromatographic column, and also exhibit hydrogen-bonded adsorption of arbutin-like phenolic compounds. Regenerated cellulose membranes, due to their highly hydrophilic surface, show no significant adsorption of hyaluronic acid and the aforementioned active ingredients, making them the most versatile choice. Filtration occurs during… Completed using a plastic syringe and a needle filter, discard the previous filter. Filtrate, collect the following to The filtrate is transferred to a brown injection bottle as the injection solution. The purpose of discarding the first filtrate is to ensure that the filter membrane surface is wetted by the first sample and reaches adsorption equilibrium, so that the target analytes in the subsequent filtrate will not be lost due to non-specific adsorption by the filter membrane.

[0023] After filtration, the analyte solution is immediately introduced into a two-dimensional chromatography system based on a dual-column switching valve and FPGA parallel spectrometry. The injection is performed by the system's injector, with the injection volume set to [value missing]. to The selection of the injection volume needs to consider both the lower limit of response of the evaporative light scattering detector to hyaluronic acid and the upper limit of column efficiency of the size-exclusion column. The lower limit of response of the evaporative light scattering detector to large carbohydrate molecules is approximately... The concentration of hyaluronic acid in the test solution after dilution of typical medical aesthetic excipients is within the range of to Between, therefore This is sufficient to guarantee the signal-to-noise ratio required for molecular weight distribution calculations; if the injection volume is further increased to... In summary, the initial bandwidth of the injection plug at the column head will exceed the dead volume of the column. This significantly reduces the resolution of size exclusion columns, causing partial overlap between the hyaluronic acid elution window and the elution window of low molecular weight active ingredients.

[0024] The injector uses a six-way injection valve with automatic cleaning function. After each injection, the injection needle is automatically rinsed with mobile phase diluent both inside and outside. to To prevent cross-contamination, it is recommended to administer two consecutive blank injections before injecting each batch of samples. The blank injections should use pure mobile phase diluent to observe baseline stability and confirm system cleanliness. Following this, administer one injection of a known concentration of hyaluronic acid reference solution and one injection of a mixed reference solution of multiple active ingredients to confirm the system suitability parameters for the day, including retention time drift, peak area repeatability, and column efficiency. Only after these suitability parameters have stabilized should the test solution be injected. This procedure follows the standard practice for system suitability testing in the General Principles of the Chinese Pharmacopoeia.

[0025] Before the flow path enters the chromatographic column, it is also necessary to pay attention to the online degassing of the one-dimensional mobile phase. After the one-dimensional mobile phase is prepared, it first undergoes... The regenerated cellulose membrane is used to remove insoluble particles via vacuum filtration, and then connected to the system's online vacuum degasser. The negative pressure of the degasser is set at... to During this process, continuous degassing is performed. The reason for emphasizing degassing is that the evaporative light scattering detector is extremely sensitive to dissolved air in the mobile phase. Dissolved oxygen will precipitate microbubbles in the pressure reduction zone. These microbubbles, after being mixed into the atomization chamber, scatter the laser along with the sample particles, which manifests as sharp burr noise on the first-dimensional effluent curve. This noise can be misjudged as the true peak by subsequent peak detection.

[0026] In optional implementations, sample pretreatment can be adjusted according to the specific form of the medical aesthetic excipients. For high-viscosity gel samples, after adding the mobile phase diluent, the sample can be placed in a container... vortex in water bath Temperature is used to reduce the apparent viscosity caused by molecular chain entanglement, making the dilution process more uniform; for emulsion samples containing emulsifiers, they can be added before dilution. to Shake once in hexane and allow to stand for separation. Discard the upper oil phase and then dilute with the aqueous phase. This liquid-liquid partitioning pretreatment, although sacrificing some oil-soluble vitamins with high solubility in hexane, can significantly improve the lifespan of subsequent size-exclusion columns. For samples containing protein-active components (such as epidermal growth factor or recombinant collagen peptides), to avoid protein denaturation due to shearing during dilution, the diluted analyte can be... Let stand in the refrigerator Then, the sample is injected again, and the protein is allowed to stand at low temperature to return to its native folded state from the shear-induced partially unfolded state.

[0027] Another alternative implementation involves adding an ultrafiltration pre-concentration step before microporous membrane filtration, using a molecular weight cutoff of [missing information]. to Centrifugal ultrafiltration tubes, with Centrifugation The retentate was collected as a dedicated analyte for hyaluronic acid analysis, and the filtrate was collected as a dedicated analyte for small molecule active ingredient analysis. The two analytes were injected twice, one for molecular weight distribution and the other for concentration determination. This split-pathway analysis method is suitable for special products with extremely low hyaluronic acid concentrations but extremely high active ingredient concentrations (such as so-called "water-light essence" with nicotinamide as the main ingredient). This is because size exclusion chromatography columns can only clearly detect the weak signal of hyaluronic acid under overload injection conditions, while small molecule active ingredients experience severe overload on the column, leading to severe peak splitting in second-dimensional hydrophilic interaction chromatography columns. Split-pathway analysis decouples the two types of components at the injection source, using two injections of one sample to expand the dynamic range of concentration, at the cost of doubling the analysis time.

[0028] Finally, it's important to note that the tools used to remove samples from the sample container must also be carefully controlled. Hyaluronic acid is quite sensitive to adsorption from metal needle tips, especially stainless steel needle tips, which can slowly release iron ions under weakly acidic conditions and complex with the carboxyl groups of hyaluronic acid, causing a falsely high molecular weight in some areas. Therefore, it is recommended to use glass pipettes or disposable polypropylene pipette tips for sampling, avoiding stainless steel needles. If laboratory conditions only allow for the use of stainless steel needles, a [preparedness / measurement] can be made before each sampling. Rinse the needle with disodium ethylenediaminetetraacetate solution Rinse with pure water This is to passivate the active metal sites on the needle surface.

[0029] After a series of operations including dilution, filtration, injection, and degassing, the test solution enters the one-dimensional flow path of the two-dimensional chromatography system based on a dual-column switching valve and FPGA parallel spectrometry in a homogeneous solvent composition, stable temperature, and bubble-free flow path. This lays the foundation for subsequent separation by size exclusion columns, acquisition of the first-dimensional elution curve, and identification of the first-dimensional coupled slice window. It also eliminates potential sources of systematic error for subsequent second-dimensional separation by hydrophilic interaction columns and full-wavelength scanning by a diode array detector. All the parameters in the above operations are adjustable. In practical applications, appropriate adjustments can be made based on the specific product's hyaluronic acid molecular weight level, the number and types of active ingredients, and the laboratory instrument configuration, without affecting the complete implementation of this step as part of the overall technical solution.

[0030] After the test solution enters the two-dimensional chromatography system based on a dual-column switching valve and FPGA parallel spectrum interpretation, it first undergoes size exclusion chromatography to complete the first dimension of separation. The size exclusion chromatography column should be selected to cover the typical molecular weight range of hyaluronic acid in medical aesthetic excipients. Typically, the molecular weight distribution of medical-grade hyaluronic acid ranges from... The low molecular weight segment extends to Even higher molecular weight ranges require separation, so a single column is often insufficient, necessitating the use of two or three columns in series with different size exclusion limits. A typical configuration is to use columns with size exclusion limits of... The limit of the column and the resistance of the array is The columns are connected in series, and the packing material inside the columns is cross-linked copolymerized hydroxymethyl acrylate or cross-linked dextran with a particle size of [missing information]. to The column specifications are The basic physical process of size exclusion is that sample molecules flow out first due to the difference in accessible pores in the porous stationary phase based on their hydrodynamic volume. Large molecules flow out first because they have fewer accessible pores, while small molecules flow out later because they have more accessible pores. The order of their flow out strictly follows the hydrodynamic volume from largest to smallest, rather than the molecular weight from largest to smallest. This is the fundamental reason why narrowly dispersed standards must be used to establish a standard curve for subsequent molecular weight calibration.

[0031] The flow rate of the one-dimensional mobile phase was controlled between 0.4 mL / min and 0.8 mL / min, and the column temperature was set between 30 °C and 35 °C. A column temperature higher than room temperature can significantly reduce the viscosity of the mobile phase, thereby reducing the column pressure, and shorten the recovery relaxation time of the molecular chains, ensuring that the hydrodynamic volume of each eluted molecule corresponds to the steady-state conformation. Choosing too low a flow rate will cause increased axial diffusion and peak broadening, while choosing too high a flow rate will exceed the optimal point of the van der Munther equation and reduce column efficiency. The above flow rate range is a compromise obtained by balancing the two broadening mechanisms.

[0032] The post-column flow path of the size exclusion column is immediately connected to a low dead volume tee connector. The two outlets of the tee connect to the inlet of the evaporative light scattering detector (ELSD) and the injection port of the 10-port 2-position electrically operated switching valve, respectively, forming two parallel flow paths: a detection path and a sampling path. It is important to note that the tee connector performs spatial parallel splitting of the column effluent, rather than temporal sequential distribution: the same elution volume of effluent is simultaneously split into two streams upon reaching the tee connector. These two streams are identical in composition and generated simultaneously, and then flow independently. One stream entering the detection path is consumed during the nebulization and evaporation process of the ELSD, while the other stream entering the sampling path is retained intact and enters the sampling loop. The fact that the sample in the detection path is consumed does not change the integrity or availability of the sample in the sampling path, nor does it impose any constraint on the timing of the sampling path. This configuration of parallel splitting with destructive detector bypass monitoring is a mature and common setup in the field of liquid chromatography. It is often seen in the arrangement of parallel ultraviolet detectors in liquid chromatography-mass spectrometry systems, as well as in the arrangement of splitting to the detector in preparative liquid chromatography to guide fraction collection.

[0033] The split ratio between the detection and sampling branches is determined by the flow resistance ratio of the two branches. To ensure a stable and controllable split ratio, a fused silica flow-limiting capillary with an inner diameter of 0.10 mm to 0.13 mm and a length of 0.3 m to 1.2 m is installed on the detection branch as a flow-limiting device. The outlet of the flow-limiting capillary is connected to the nebulizer inlet of the evaporative light scattering detector (ELSD). By selecting the inner diameter and length of the flow-limiting capillary, the flow ratio between the detection and sampling branches is made to fall within the range of 1:3 to 1:4, meaning that approximately 20% to 25% of the effluent enters the ELSD for concentration monitoring, and the remaining 75% to 80% of the effluent enters the sampling branch for subsequent second-dimensional separation. This ratio design is a compromise; if too much flow is diverted to the ELSD, the amount of sample entering the sampling loop will decrease, causing a drop in the second-dimensional signal; if too much flow is diverted to the sampling loop, the signal of the ELSD will be too weak, affecting the signal-to-noise ratio of the subsequent first-dimensional peak detection. The split ratio is confirmed by measuring and weighing the effluent from the two branches at regular intervals after the system is assembled. The deviation between the measured value and the design value should be within 5%. If the deviation exceeds this range, the length of the flow-limiting capillary tube should be replaced and readjusted.

[0034] In the sampling branch, a delay capillary is connected in series between the three-way connector and the inlet port of the ten-way two-position electric switching valve. The purpose of the delay capillary is to artificially increase the transmission time of the sampling branch, ensuring that the arrival time of the effluent of the same elution volume at the sampling loop inlet is strictly later than its arrival time at the evaporative light scattering detector detection cell, thus reserving sufficient time margin for real-time peak detection and valve control decisions by the FPGA parallel spectral analysis unit. The delay capillary has an inner diameter of 0.25 mm, a length of 1.5 m to 4.0 m, and an internal volume of 75 μL to 200 μL, and is wound into a spiral shape with a winding diameter of 10 mm to 20 mm. The reason for choosing a spiral winding instead of a straight tube arrangement is that the laminar flow in the spiral tube will generate Dean's secondary flow due to centrifugal force. The secondary flow has a forced radial mixing effect on the velocity distribution on the tube cross-section, which can reduce the Taylor-Aris axial dispersion coefficient to less than one-third of that of a straight tube of the same volume. Therefore, while obtaining the required delay, the additional band broadening is controlled within 15% of the volume of one effluent segment. The rationale for choosing a smaller inner diameter of 0.25 mm is that, under the condition that the internal volume of the delay capillary is fixed, the volume variance of Taylor-Aris diffusion is proportional to the square of the inner diameter. Reducing the inner diameter can significantly reduce the band broadening while maintaining the same delay, and the resulting increase in post-column pressure drop does not exceed 0.2 MPa under the above specifications, and has no substantial impact on the working pressure of the size exclusion column.

[0035] The response process of the evaporative light scattering detector (ELS) involves three cascaded steps: the mobile phase is nebulized by an inert gas (usually high-purity nitrogen) to form an aerosol mist; the droplets enter a heated drift tube where the solvent is completely evaporated, leaving only solute particles; these particles enter the detection cell, are irradiated by a laser beam, and the scattered light is collected by a photomultiplier tube. For detector parameter settings, the nitrogen pressure is set to 25 psi to 35 psi, the drift tube temperature to 60°C to 90°C, and the photomultiplier tube gain is adjusted between levels 1 and 5 based on the nominal concentration of hyaluronic acid in the test solution. The response of the ELS detector follows a power-law relationship with the solute mass. ,in This represents the evaporative light scattering response value. For the instantaneous mass entering the detection pool, A proportionality coefficient related to detector geometry and nitrogen flow rate. The response index typically ranges between 1.2 and 1.8. This is precisely because... Since it is not equal to 1, directly using the evaporation light scattering response value as the mass concentration for molecular weight distribution integration will produce systematic bias. This is the reason for performing two power-law inverse transforms on the hyaluronic acid response correction curve.

[0036] It should be noted that the evaporative light scattering (ELS) detector process is destructive; the solute entering the detection cell cannot be recovered after atomization and evaporation. However, as mentioned earlier, only one effluent stream from the detection branch after parallel splitting at the three-way connector enters the ELS detector. The other effluent stream from the sampling branch does not pass through the ELS detector. Therefore, the destructive detection of the ELS detector only consumes the portion of the sample used for monitoring. The effluent fragment retained by the sampling loop is always the undetected, intact original effluent. There is no causal relationship between the destructive nature of the ELS detector and the availability of the effluent fragment from the sampling branch.

[0037] The voltage signal output by the evaporative light scattering detector is transmitted via a coaxial shielded cable to the analog-to-digital conversion channel of the FPGA parallel spectrum deconversion unit in the form of a first-dimensional effluent curve analog signal. The analog-to-digital conversion channel has a bit depth of no less than 16 bits and a sampling frequency of [missing information]. to Adjustable. The upper limit of the sampling frequency is limited by the peak width of the size exclusion column; the full width at half maximum (FWHM) of the hyaluronic acid peak on a size exclusion column is typically [value missing]. to According to the coverage of each peak at least The Nyquist criterion for sampling points is taken as follows: It is sufficient to retain all shape information; higher sampling frequencies are mainly used to reduce the truncation error of point-by-point numerical integration.

[0038] After the digital signal of the first-dimensional elution curve obtained from sampling enters the FPGA parallel spectrum interpretation unit, the first-dimensional coupled slicing window identification is first performed. The first-dimensional coupled slicing window consists of two sub-windows: a hyaluronic acid elution window and a preset active ingredient elution window. The reason for dividing the entire first-dimensional retention time axis into these two segments instead of covering the entire retention time in a uniform modulation mode is that there is usually a large retention time gap between the elution positions of hyaluronic acid and small molecule active ingredients on size exclusion columns. Only the mobile phase flows through this gap. If slicing continues in the gap, a large number of empty slices will be generated. Empty slices still need to complete a complete second-dimensional gradient cycle on the second-dimensional hydrophilic interaction column, wasting analysis time. By opening the slicing according to the two intervals where the sample actually flows out, not only can all target components be covered, but the overall analysis time can also be compressed to about one-third of that in continuous slicing mode. to .

[0039] refer to Figure 3 , Figure 3The system includes three parallel horizontal time axes, corresponding from top to bottom to the detection branch, sampling branch, and molecular weight coordinates, respectively. The detection branch time axis plots the response curve of hyaluronic acid on the evaporative light scattering detector, with the start, peak, and end times of the hyaluronic acid elution window marked below the curve. The sampling branch time axis uses equally spaced vertical lines to mark the times of each switching pulse emitted by the valve-controlled timing generator; the time interval between two adjacent switching pulses is the modulation period. The effluent intercepted between every two adjacent switching pulses constitutes one effluent segment, and each effluent segment is intercepted alternately by sampling ring A and sampling ring B.

[0040] There is a time offset between the start time of the detection branch and the first switching pulse of the sampling branch; this offset is the phase compensation amount. . Figure 3 The first switching pulse of the sampling branch is located to the right of the starting point of the detection branch, which intuitively reflects... To meet this design constraint—that effluent of the same elution volume is always detected by the evaporative light scattering detector before reaching the sampling loop inlet—the FPGA parallel spectral analysis unit thus has the time condition to perform slicing control on the effluent that has not yet reached the sampling loop based on the detection results. Phase compensation amount The physical meaning is: the time when the effluent of the same elution volume reaches the sampling loop inlet, minus the time when it reaches the detection cell of the evaporative light scattering detector. This value is obtained by actual measurement and calibration of the tracer after the system is assembled, and is kept positive and not less than 5s by connecting a delay capillary in series in the sampling branch. If the transmission time difference between the detection branch and the sampling branch is not corrected, the correspondence between the slicing time and the peak time detected by the evaporative light scattering detector will produce a systematic shift, resulting in inaccurate calibration of the molecular weight coordinates. When calculating the emission time of each switching pulse, the valve-controlled timing generator uses the starting time detected by the evaporative light scattering detector plus the phase compensation amount as the time of the first switching pulse, and then according to the modulation period. Continuously increasing.

[0041] Figure 3 The molecular weight axis at the bottom is marked with the molecular weight scale corresponding to each effluent fragment. The molecular weight scale is determined by the phase difference of the molecular weight positioning of each effluent fragment, recorded by the valve-controlled timing generator. The value was obtained by conversion from the standard curve of hyaluronic acid molecular weight, where , The switching moment of the effluent fragment, This represents the peak time of the hyaluronic acid elution window. Since the molecular weight positioning phase difference is used during the data cube assembly stage after all data acquisition is complete, the fact that the peak time is later than the earlier slice times does not affect the accuracy of the molecular weight coordinate assignment. Each elution fragment is thus locked to a specific molecular weight range on the size exclusion column, and the center molecular weight scale of that range is used as the molecular weight coordinate to write into the molecular weight-retention time-component three-dimensional data cube.

[0042] The hyaluronic acid elution window is obtained by real-time peak detection of the digital signal of the first-dimensional elution curve. The core action of real-time peak detection is to perform a first-order difference on the digital signal of the first-dimensional elution curve, that is, to subtract the current sample value from the previous sample value to obtain the difference value. Let the first-dimensional elution curve digital signal be at the [missing value]. The value at each sampling time is Then the difference value ,in For sampling sequence number, For the first Voltage values ​​at each sampling time, For the first Voltage values ​​at each sampling time, For the first The difference value at each sampling time. When The value changes from negative to positive and exceeds the preset noise threshold. The current sampling time is determined to be the start time of the hyaluronic acid elution window. A preset noise threshold is defined as a safety threshold that is more than three times the baseline noise standard deviation; a typical value is three to five times the baseline noise standard deviation. When a value changes from positive to negative, the current sampling time is determined to be the peak time of the hyaluronic acid elution window. The physical meaning of this criterion is that the first-dimensional elution curve reaches its maximum value at that moment. Continuously less than the preset noise threshold And maintain the preset hold count. Each sampling time To prevent misjudgments caused by small local fluctuations, the typical value for the continuous counting threshold is [value missing]. to The current sampling time is determined to be the end point of the hyaluronic acid elution window. The reason for using continuous counting instead of single-point determination is that the first-dimensional elution curve often shows slight fluctuations caused by mobile phase pulsation at the end of the descending branch. Single-point determination would prematurely cut off the elution window and lose the tail distribution of the low molecular weight side of hyaluronic acid, which would have a particularly significant impact on the calculation of molecular weight dispersion.

[0043] The start and end times of the preset active ingredient elution window do not depend on real-time peak detection, but are directly read from the retention time interval pre-stored in the on-chip block random access memory of the FPGA parallel resolution unit. This retention time interval was determined by sequentially injecting each active ingredient single standard into a two-dimensional chromatography system based on a dual-column switching valve and FPGA parallel resolution when establishing this analytical method. Considering the diurnal drift of retention time, the preset active ingredient elution window is extended before and after the single standard retention time. to As a safety margin, after writing the start time, peak time, and end time of the hyaluronic acid elution window, as well as the start and end times of the preset active ingredient elution window, into the elution window time register group, the boundaries of the first-dimensional combined slicing window are all determined.

[0044] After the first dimension combined slicing window is determined, switching pulses are continuously sent within the window to control the ten-way two-position electric switching valve to perform the slicing action.

[0045] The slicing process involves a critical time alignment issue. After the same elution volume of effluent is split into two parallel streams at the T-junction, the travel time along the detection branch to the evaporative light scattering detector's detection cell is not equal to the travel time along the sampling branch to the sampling loop inlet. A phase compensation amount is defined. Subtract the former from the latter, that is ,in For phase compensation amount, The time when the effluent of the same elution volume reaches the inlet of the sampling loop. This refers to the moment when the effluent of the same elution volume reaches the detection cell of the evaporative light scattering detector. According to this definition, A positive value indicates that the sampling branch lags behind the detection branch, meaning that a certain volume of effluent is detected by the evaporative light scattering detector before reaching the sampling loop inlet; A negative value indicates that the sampling branch precedes the detection branch, meaning that the elution volume of effluent reaches the sampling loop inlet first and is then detected.

[0046] This invention explicitly requires that the two-dimensional chromatography system based on a dual-column switching valve and FPGA parallel spectrometry must meet the following requirements during assembly. Positive value and A design constraint of no less than 5 seconds. The necessity of this constraint lies in the fact that only when... When the value is positive, the evaporative light scattering detector detects a certain elution volume only before that elution volume reaches the sampling loop inlet. Only then does the FPGA parallel spectral analysis unit have the time condition to implement slicing control of the effluent section that has not yet reached the sampling loop based on the detection result; if... If the value is negative or zero, the elution volume has already flowed through the sampling loop inlet before or simultaneously with detection. If a switching pulse is then issued based on the detection result, the controlled object no longer exists, and real-time valve control cannot be established in terms of timing. To ensure... The value remains constant and has sufficient time margin. The aforementioned delay capillary is connected in series in the sampling branch. The inner diameter and length of the delay capillary are selected to ensure... It falls within the range of 5s to 20s.

[0047] The lower limit of no less than 5 seconds is determined by the system's control delay budget. The analog-to-digital conversion channel sampling frequency of the FPGA parallel spectrum descaling unit is 10Hz to 100Hz. The peak detection state machine can complete the differential and decision within one sampling cycle, corresponding to a delay of 10ms to 100ms. Both the peak detection state machine and the valve control timing generator are implemented in hardware logic, and the delay of combinational logic and register transmission is on the order of microseconds, which can be ignored compared to the previous item. The mechanical action time of the electric actuator of the ten-way two-position electric switching valve from receiving the switching pulse to completing the valve core switching is taken as 150ms according to the technical parameters of commercial two-position multi-way electric switching valves. The maximum control delay obtained by adding the above three items is approximately 250ms. The lower limit is 5s, which has a margin of more than 20 times relative to the maximum control delay, which is sufficient to cover short-term fluctuations in the velocity of the one-dimensional mobile phase, quantization errors at the sampling time, and batch differences in valve action time.

[0048] The values ​​were obtained through actual tracer calibration after system assembly, without relying on theoretical calculations of the pipeline volume. The calibration process is as follows: The hydrophilic interaction column was temporarily replaced with a zero-dead-volume straight-through connector, and the 10-port 2-position electric switching valve was fixed in the injection position, allowing the effluent from the sampling branch to pass sequentially through the sampling loop, the online solvent conversion unit, and the straight-through connector to reach the diode array detector. A tracer solution that can be detected by the evaporative light scattering detector and has strong absorption in the ultraviolet region was injected through the injector. A caffeine aqueous solution with a mass concentration of 0.1 mg / mL could be used as the tracer. The peak times of the tracer on the evaporative light scattering detector and the diode array detector were recorded simultaneously. The peak time of the diode array detector was subtracted from the known volume between the sampling loop inlet and the diode array detection cell by the corresponding flow rate to obtain the transmission time. The peak time of the evaporative light scattering detector is... The difference between the two is The calibration is repeated three times, and the average value is written to the on-chip block random access memory of the FPGA parallel spectrum resolution unit. This calibration is performed once after system assembly, and remains valid thereafter as long as the piping configuration and one-dimensional mobile phase velocity remain unchanged. If the one-dimensional mobile phase velocity is changed, the calibration is adjusted according to the inverse relationship between the velocities. Alternatively, the calibration can be performed once more.

[0049] As an auxiliary check during the design phase, It can also be estimated from the pipeline geometry parameters, and the estimation formula is: ,in The volume of the sampling branch between the outlet of the tee connector and the inlet of the sampling loop, including the delay capillary. The internal volume of the detection branch between the outlet of the tee connector and the detection cell of the evaporative light scattering detector, including the flow-limiting capillary, For the one-dimensional mobile phase velocity, This represents the flow fraction diverted to the sampling branch. Take 0.5 mL / min, Take 0.8, Take 15μL, Taking 160 μL as an example, the flow rate of the detection branch was 0.1 mL / min, the flow rate of the sampling branch was 0.4 mL / min, the transmission time of the detection branch was 9.0 s, and the transmission time of the sampling branch was 24.0 s. It is 15.0s, which satisfies the requirement. Design constraint of not less than 5 seconds. The value range is from 5 μL to 20 μL. The value ranges from 90 μL to 200 μL. It is important to emphasize that this estimation formula is only for selecting delayed capillaries; the actual system usage... All numerical values ​​shall be based on the actual measured calibration results of the aforementioned tracer.

[0050] It should also be noted that the slicing action is triggered by the start time of the hyaluronic acid elution window, not the peak time. The start time is located in the rising phase of the hyaluronic acid peak, which is much earlier than the peak time. Therefore, before the peak arrives, the slicing action has already been carried out continuously for several cycles according to the modulation period. In this invention, the peak time is only used to determine the molecular weight positioning phase difference of each effluent fragment. The molecular weight positioning phase difference is a coordinate assignment operation performed on the already acquired data during the data cube assembly stage. It belongs to the data post-processing action and does not participate in any real-time valve control decision. Therefore, it is not subject to the time sequence constraint of detection after control.

[0051] The slicing start time is the sum of the phase compensation amount and the start time of each sub-window. Subsequent slicing actions follow a preset modulation period. Repeat within the first-dimensional combined slice window until the current switching time reaches the end time of each sub-window and the phase compensation amount is superimposed, then stop. The preset modulation period is the fixed time interval between two adjacent switching pulses, typically set to a value of [value missing]. to The modulation period is selected according to the Murphy-Schure-Foley rule, requiring the first-dimensional peak to be sampled at least three times within one modulation period to ensure that the first-dimensional separation information is not lost due to undersampling; at the same time, the modulation period cannot be too short, because the minimum time required for the second-dimensional hydrophilic interaction column to complete one gradient cycle and equilibrate to the initial conditions is the lower limit of the modulation period. If the second-dimensional column specifications are... Sub-2 micron filler, second dimension gradient from Acetonitrile decreased Acetonitrile rose again to A complete cycle of acetonitrile can be compressed to If a longer second-dimensional column is used, the modulation period will be extended accordingly.

[0052] refer to Figure 2 , Figure 2 In the figure, (a) is the first-dimensional effluent curve output by the evaporative light scattering detector. The horizontal axis is the retention time in minutes, and the vertical axis is the evaporative light scattering response in millivolts. Figure 2 In (a) of the image, a broad main peak appears on the left, with its apex located at approximately 7 minutes. This main peak corresponds to hyaluronic acid. Multiple narrow peak groups appear on the right, with retention times ranging from approximately 17 to 22 minutes, corresponding to various active ingredients. The hyaluronic acid elution window is defined by three key moments: the start time, the peak apex time, and the end time. The start time corresponds to the rising edge of the hyaluronic acid peak, the peak apex time corresponds to the maximum value of the hyaluronic acid peak, and the end time corresponds to the point where the trailing edge of the hyaluronic acid peak falls back to the baseline. The preset active ingredient elution window is located to the right of the hyaluronic acid elution window, and its start and end times are predetermined by the retention time range of each active ingredient on the size exclusion column. The hyaluronic acid elution window and the preset active ingredient elution window together constitute the first-dimensional coupled slicing window. There is a gap between the two windows where no sample flows out. The valve-controlled timing generator does not output switching pulses during this gap, thus avoiding the generation of empty slices. Figure 2 In (a) of the diagram, the positions of the modulation slices are marked with equally spaced short vertical lines inside the two windows. The time interval between two adjacent short vertical lines is the modulation period. Each modulation cycle corresponds to one effluent fragment. Figure 2 In the diagram, (b) represents the sequence of difference values ​​obtained after performing a first-order difference operation on the digital signal of the first-dimensional outflow curve. The horizontal axis represents the retention time, and the vertical axis represents the difference value. ,in , For the first The response value at each sampling time point, For the first The response value at each sampling time. Figure 2 In (b) of the diagram, the preset noise threshold is marked with a horizontal dashed line. and When the difference value Change from negative to positive and exceed At that time, the peak detection state machine determines that the current moment is the start time of the hyaluronic acid elution window; when The moment when a value changes from positive to negative is considered the peak time; when Persistently below And maintain the preset hold count. The endpoint time is determined after each sampling. Figure 2 The corresponding intervals of the hyaluronic acid elution window and the active ingredient elution window in (b) are related to... Figure 2 (a) is strictly aligned, and the difference value can be visually observed to rise rapidly across the hyaluronic acid peak front. Crossing the zero line at the peak, falling back along the trailing edge and remaining below it. The entire process until the endpoint is determined.

[0053] Each time a switching pulse reaches the drive interface of the 10-way 2-position electric switching valve, the valve switches between the sampling and injection positions once. Two identical sampling rings are installed on the 10-way 2-position electric switching valve; the volume of the sampling rings is typically [missing information]. to The sampling loop volume is selected in conjunction with the modulation period and the one-dimensional flow phase velocity to be exactly equal to or slightly greater than the modulation period and the flow rate of the one-dimensional flow phase. ,in This represents the split ratio for the flow to the sampling branch. Upon arrival of each switching pulse, the sampling loop in the sampling position switches to the injection position, and the effluent retained in its interior during the previous modulation cycle is flushed into the hydrophilic interaction column by the two-dimensional mobile phase. Simultaneously, the other sampling loop, previously in the injection position, switches to the sampling position and begins accepting the effluent from the next modulation cycle. The two sampling loops alternate between sampling and injection, ensuring uninterrupted retention of the first-dimensional effluent and preventing sample loss.

[0054] The effluent fragments trapped in the sampling loop cannot directly enter the hydrophilic interaction column because the first-dimensional mobile phase is aqueous, while the hydrophilic interaction column requires an acetonitrile volume ratio greater than 70% in the initial mobile phase to ensure effective hydrophilic retention of small molecule active ingredients at the column head. If the effluent fragments enter the hydrophilic interaction column directly with pure aqueous phase, the sample will penetrate directly through the column head without retention, resulting in broad peaks with superimposed solvent fronts, and the second-dimensional separation will completely fail. Therefore, an online solvent switching unit is installed between the sample outlet port of the 10-way 2-position electric switching valve and the inlet of the hydrophilic interaction column.

[0055] The online solvent conversion unit consists of two parts connected in series: a T-shaped mixing chamber and a vortex mixing capillary. The T-shaped mixing chamber is a three-way structure made of stainless steel or polyetheretherketone (PEEK). The inner diameter of each of the three channels is 0.15 mm to 0.20 mm, and the internal volume of the chamber is 2 μL to 8 μL. One inlet comes from a segment of the aqueous effluent discharged from the sampling loop at the injection position, and the other inlet comes from a high acetonitrile ratio replenishment mobile phase delivered by a two-dimensional infusion pump. The outlet end is connected to the vortex mixing capillary. The vortex mixing capillary is a polyetheretherketone (PEEK) capillary with an inner diameter of 0.17 mm, a length of 200 mm to 400 mm, and an internal volume of 4.5 μL to 9.0 μL. The capillary has a 90-degree to 120-degree bend every 20 mm to 30 mm along its axial direction, with the bend planes of adjacent bends at 90 degrees to each other. This causes the fluid inside the capillary to generate alternating Dean's secondary flows as it passes through each bend, thus achieving radial forced mixing of the two liquid streams within a very short capillary length. The sum of the internal volumes of the T-shaped mixing chamber and the vortex mixing capillary is 6.5 μL to 17 μL, much smaller than the volume of a single effluent segment; therefore, the additional band broadening introduced by the mixing process is negligible.

[0056] The flow rate ratio of the two inlet streams is set as follows. The sampling loop in the injection position is backflushed and emptied by one branch of the two-dimensional infusion pump at a flow rate of 0.10 mL / min to 0.25 mL / min. The replenishing mobile phase is delivered into the T-shaped mixing chamber by the other branch of the two-dimensional infusion pump at a flow rate of 1.25 mL / min to 1.40 mL / min. The acetonitrile volume ratio of the replenishing mobile phase is 95%. The total flow rate after the two streams are combined is 1.50 mL / min, meaning that the dilution factor of the sampling loop outflow falls within the range of 6 to 15 times.

[0057] Taking a sampling loop purging flow rate of 0.15 mL / min and a replenishing mobile phase flow rate of 1.35 mL / min as an example, the total flow rate after mixing is 1.50 mL / min. The volume ratio of acetonitrile after mixing is 95% multiplied by 1.35 and then divided by 1.50, which equals 85.5%, meeting the requirement of an acetonitrile volume ratio greater than 70% for the initial retention conditions of hydrophilic interaction chromatography columns. If the active ingredient in the medical aesthetic excipient formulation is highly polar and requires a lower initial acetonitrile volume ratio to avoid excessive retention, the sampling loop purging flow rate can be increased to 0.25 mL / min, at which point the volume ratio of acetonitrile after mixing decreases to 79.2%. If the active ingredient is less polar and requires a higher initial acetonitrile volume ratio, the sampling loop purging flow rate can be reduced to 0.10 mL / min, at which point the volume ratio of acetonitrile after mixing increases to 88.7%. The above ratio range can cover the initial retention conditions required by common active ingredients in medical aesthetic excipients on hydrophilic interaction chromatography columns. The volume ratio of acetonitrile after mixing is allowed to have a difference of no more than 5 percentage points from the initial volume ratio of acetonitrile in the second gradient, because the effluent fragments only form a solvent intensity step at the column head with a volume equal to the volume of the sampling loop. This step is immediately flattened by the initial mobile phase after the gradient elution begins. Within the above-mentioned difference range, the peak shape symmetry factor of the active ingredient can be kept within 1.5.

[0058] The online monitoring and feedback adjustment of the acetonitrile volume ratio after mixing is achieved as follows. In the acetonitrile-water binary system, within the acetonitrile volume ratio range above 60%, the viscosity monotonically decreases with increasing acetonitrile volume ratio. Therefore, under constant flow rate in the second dimension, there is a monotonic correlation between the inlet pressure of the hydrophilic interaction column and the volume ratio of the mixed acetonitrile. A pressure sensor is installed at the outlet of the two-dimensional infusion pump, and the output of the pressure sensor is sent to the FPGA parallel resolution unit after analog-to-digital conversion. During the method establishment phase, three to five water-acetonitrile mixtures with known acetonitrile volume ratios are sequentially passed through the hydrophilic interaction column, and the corresponding inlet pressures are recorded. A pressure-composition comparison table between inlet pressure and acetonitrile volume ratio is established and written to the on-chip block random access memory. During operation, the FPGA parallel spectral analysis unit reads the pre-column pressure once at the beginning of each modulation cycle, looks up the current acetonitrile volume ratio after mixing in a table, and if the deviation between the found value and the set value exceeds 2 percentage points, the flow rate of the supplementary mobile phase channel in the two-dimensional infusion pump is adjusted by increments or decrements of 0.02 mL / min according to the sign of the deviation, so that the acetonitrile volume ratio after mixing returns to near the set value. The response period of this feedback loop is consistent with the modulation cycle, which is sufficient to compensate for long-term drift of pump flow rate and slow changes in pipeline resistance.

[0059] The volatile buffer salts in the one-dimensional mobile phase are carried into the second dimension along with the eluent fragments. Their effect on the retention mechanism in hydrophilic interaction chromatography is handled as follows: The concentration of ammonium acetate in the one-dimensional mobile phase is 50 mmol / L to 100 mmol / L, or the concentration of ammonium formate is 20 mmol / L to 50 mmol / L. The eluent fragments are diluted 6 to 15 times in the T-shaped mixing chamber, thus reducing the concentration of volatile buffer salts entering the hydrophilic interaction column to approximately 3.3 mmol / L to 16.7 mmol / L. Ammonium acetate and ammonium formate are commonly used mobile phase additives in hydrophilic interaction chromatography, with conventional concentrations ranging from 5 mmol / L to 20 mmol / L. Their function is to inhibit the ion exchange of residual silanol groups on the stationary phase surface with basic active components and to stabilize the thickness of the water-rich layer on the stationary phase surface. Therefore, the volatile buffer salts in the one-dimensional mobile phase, after being carried into the second dimension, not only do not cause interference but also function as conventional additives in the second dimension, improving the reproducibility of retention times in the hydrophilic interaction column. This is the second consideration for selecting volatile ammonium salts as the one-dimensional mobile phase in this invention. The first consideration is to adapt to the atomization evaporation mechanism of the evaporative light scattering detector.

[0060] The risk of precipitation of volatile buffer salts under high acetonitrile volume ratio conditions needs to be controlled. The solubility of ammonium acetate in a water-acetonitrile mixture with an acetonitrile volume ratio of 85% is approximately 30 mmol / L to 40 mmol / L; exceeding this concentration will cause crystallization in the tubing and column head. Therefore, this invention specifies that the concentration of ammonium acetate in the one-dimensional mobile phase should not exceed 100 mmol / L, and the dilution factor in the T-shaped mixing chamber should not be less than 6 times, ensuring that the concentration of ammonium acetate entering the hydrophilic interaction column does not exceed 16.7 mmol / L, leaving sufficient safety margin. If ammonium formate is used as the buffer salt in the one-dimensional mobile phase, the above safety margin is further increased because the solubility of ammonium formate in high acetonitrile systems is higher than that of ammonium acetate. After daily operation, the entire flow path is flushed with a 50:50 water-acetonitrile mixture for 20 minutes to further eliminate the possibility of long-term salt accumulation in the flow path.

[0061] After passing through the online solvent conversion unit, the effluent fragments enter the hydrophilic interaction column to begin the second-dimensional separation. The second-dimensional gradient uses an acetonitrile-water binary system, with the acetonitrile volume ratio linearly decreasing from 90% to 50%. The gradient time is matched with the available time window after deducting the column equilibration time from the modulation cycle. The second-dimensional flow rate is controlled between 1.0 mL / min and 1.5 mL / min, and the column temperature is between 35℃ and 45℃.

[0062] The effluent from the hydrophilic column was fed into a diode array detector for full-wavelength scanning. The diode array detector used a deuterium lamp as the light source, and the scanning wavelength range was [missing value]. to , wavelength resolution is or The scanning frequency is to Each scan time yields a vector containing the absorbance values ​​for all wavelength channels, called the digitized absorbance vector. The dimension of the digitized absorbance vector is equal to the number of wavelength grid points, expressed in terms of resolution. Wavelength range to The vector dimension is calculated to be The diode array detector sends the digitized measured absorbance vector to the FPGA parallel spectral decoding unit via a high-speed digital data interface. This high-speed digital data interface is typically a gigabit Ethernet or USB 3.0 interface, with a data transfer rate of [missing information - likely a value] per second. The above is sufficient to support vector streams that retain hundreds of time points in each slice.

[0063] After the digitally measured absorbance vector enters the FPGA parallel spectral decoupling pipeline for stepwise stripping of multiple active ingredients, the multi-active ingredient stepwise spectral stripping begins. Multiple active ingredients often exhibit overlapping retention times in the two-dimensional chromatograms of the same effluent fragment, especially among similar types of polyphenolic antioxidants or multiple vitamin derivatives. At a certain retention time point, the digitally measured absorbance vector acquired by the diode array detector is the result of a weighted superposition of the pure reference spectra of multiple active ingredients according to their respective instantaneous concentrations. Directly performing single-wavelength quantification on the vector results in significant errors during component co-elution. Therefore, it is necessary to utilize multi-wavelength full-spectrum information to decouple each component; this is the motivation for introducing the stepwise stripping spectral decoupling pipeline.

[0064] The sequential stripping spectral decoupling pipeline operates in two stages: the first stage determines the stripping sequence, and the second stage executes the sequential stripping. The first stage is initiated once before each effluent fragment is processed. For each active ingredient, its pure reference spectrum is obtained during the method setup stage by scanning a single standard at the same wavelength grid. After being normalized to a unit vector with a L2 norm equal to 1, the spectrum is stored in the on-chip block random access memory of the FPGA parallel spectral decoupling unit.

[0065] The normalized cross-correlation operation in this invention refers to the inner product operation that normalizes only the reference spectral vector to the L2 norm while keeping the original amplitude of the measured absorbance vector unchanged. Let the first... The normalized pure reference spectral vector of the active ingredient is: , The L2 norm is equal to 1, and the digitally measured absorbance vector is denoted as . Then the first Initial correlation values ​​of the active ingredients From inner product The calculation yielded, where for and The dot product of vectors. Because The L2 norm is equal to 1. Geometrically, this is... exist The orthogonal projection length in the direction.

[0066] It should be noted that, Cosine similarity differs from cosine similarity in mathematical meaning. The expression for cosine similarity is: Divide by The second norm and The product of the second norms, since the denominator contains The L2 norm and cosine similarity completely normalize the amplitude of the measured vectors, retaining only the directional angle information between the two vectors. Therefore, they can only reflect the similarity of the spectral shape and cannot reflect the abundance of the component at the current retention time point. This invention uses... Normalization is applied only to the reference spectrum, not to the measured vectors. equal Multiplying by the second norm and The cosine of the angle contains both the amplitude information of the measured signal and the degree of matching of the spectral shape. Under the condition of good spectral shape matching, i.e., the cosine of the angle is close to 1, It is directly proportional to the concentration of the active ingredient at the current retention time point, therefore It is the correct measure of abundance.

[0067] At the same time, due to Normalized to a unit vector with a norm of 1. Mathematically, it is exactly equal to The only independent variable pair The regression coefficients obtained by least squares fitting, i.e. That is to put along The magnitude that should be deducted for directional decomposition. This makes the sorting process... The response intensity values ​​actually subtracted in each subsequent stripping stage are completely consistent in physical dimensions and numerical definitions, avoiding logical disconnect caused by inconsistencies between the sorting and subtraction criteria. The above-described processing flow, which uses orthogonal projection length as the selection criterion and successively subtracts the contribution of selected atoms, falls under the category of matching pursuit algorithms in signal processing. This algorithm uses the largest projection value as the selection criterion for each step, and its convergence and the monotonically decreasing property of the residual norm have mature theoretical conclusions. This invention fixes the dictionary atoms of this algorithm to the pure reference spectra of each active ingredient and implements it using an FPGA hardware pipeline, enabling real-time spectral decomposition within the modulation cycle.

[0068] according to The stripping order is obtained by sorting the active ingredients from highest to lowest abundance. The necessity of stripping in descending order of abundance lies in the fact that if a component with a weak signal and inaccurate estimated response intensity value is stripped first, its estimation error will remain in the residual vector in spectral form and be amplified in the matching of subsequent high-abundance components. Conversely, if components with strong signals and obvious shape characteristics are stripped first, only relatively small higher-order errors remain in the residual, and their impact on subsequent components is controllable. The sorting results are then written into an active ingredient stripping order index table.

[0069] The second stage processes the active ingredients sequentially according to the active ingredient stripping order index table. The digitized measured absorbance vector at each retention time point is used as the initial input for the first stripping stage, denoted as the residual absorbance vector. In the first During primary stripping, Values ​​range from 1 to , For the number of active ingredient types, read the index specified in the peeling order index table from the on-chip random access memory. Grade A active ingredient pure reference spectrum ,in For the first step of stripping the sequential index table Bit, Indicates the active ingredient number that should be used at this level.

[0070] Calculate the response intensity value ,in This represents the response intensity value of the active ingredient at the current retention time point. To input the residual absorbance vector for this level, This refers to the normalized reference spectral vector of the pure active ingredient with a L2 norm equal to 1, which is called at this level. This is the inner product of the two vectors. This inner product operation is the same as the normalized cross-correlation operation used in the first stage to determine the stripping order; the only difference is that the projected vector in the first stage was the digitized measured absorbance vector, while the projected vector in this stage is the residual absorbance vector input to this stage. Because... The L2 norm is equal to 1. Geometrically, exist The orthogonal projection length in the direction is mathematically equivalent to the length of the projection in the direction of the orthogonal projection. The only independent variable pair The regression coefficients are obtained by performing a least-squares fit. Then the residuals are updated. This means subtracting the spectral contribution of the active ingredient at this stage from the current residual.

[0071] The stripping iteration has been advanced to the next stage. After the stage is completed, the final residual It should contain only baseline noise and trace amounts of unregistered impurities. The output of each stage... By reversing the index table of the stripping order back to the original number of the active ingredient, the response intensity value of each active ingredient at the current retention time point is obtained. The above-mentioned step-by-step stripping process is implemented in a pipeline manner inside the FPGA, that is, the first stage processes the current effluent fragment. At the retention time point, the second level simultaneously processes the current effluent fragment. Each retention time point is processed sequentially. Pipeline parallelism is a key advantage of FPGAs over general-purpose central processing units (CPUs) for real-time spectrum interpretation. Traditional CPUs process each retention time point serially, requiring sequential completion of each time point. The inner product sum The next update will have a processing delay. The linear increase in speed makes it difficult to keep up with the scan frame rate of the diode array detector; while the FPGA pipeline will... Each peeling stage is laid out as follows Each hardware path is independent, and each path is only responsible for a fixed inner product and update. The overall throughput is... It is unrelated and can stably keep up with the measured absorbance vector flow of the diode array detector for tens of frames per second.

[0072] While the spectral decoupling pipeline continuously outputs response intensity values, the data cube assembly module writes these values ​​into the on-chip dual-port random access memory (DRAM) along three dimensions. The first dimension is the molecular weight coordinates; the second dimension is the retention time, i.e., the retention time of the current response intensity value on the second-dimensional chromatogram corresponding to the current effluent fragment; and the third dimension is the component number. The determination of the molecular weight coordinates depends on the phase difference of molecular weight positioning recorded by the valve-controlled timing generator for each effluent fragment. Its definition is ,in The molecular weight positioning phase difference refers to the phase difference between the effluent fragment switching moment and the peak moment of the hyaluronic acid elution window. The time of effluent segment switching, This represents the peak time of the hyaluronic acid elution window. Since the phase compensation has been pre-calibrated to a unified phase reference to adjust the transmission time difference between the detection and sampling branches, the difference between the switching time and the peak time corresponds one-to-one with the elution position of the corresponding molecular weight component on the size exclusion column. Each elution fragment is thus locked to a specific molecular weight range on the size exclusion column. The width of this range is determined by the modulation period and the one-dimensional mobile phase flow rate. The molecular weight scale at the center of the locked molecular weight range is used as the molecular weight coordinate of the elution fragment. The molecular weight coordinate values ​​are derived from the hyaluronic acid molecular weight standard curve, which is established by fitting a third-order polynomial between the retention time and logarithmic molecular weight of a series of dextran or narrow-distribution hyaluronic acid reference standards measured under the same chromatographic conditions. As elution fragments are processed sequentially, the molecular weight-retention time-component three-dimensional data cube is filled piece by piece along the molecular weight coordinates, forming a complete data structure after all elution fragments have been processed.

[0073] The results extraction module is activated after data acquisition, extracting the number-average molecular weight, weight-average molecular weight, and molecular weight dispersion of hyaluronic acid, as well as the concentration values ​​of each active ingredient. The relevant parameters of hyaluronic acid are calculated directly from the digital signal of the first-dimensional elution curve, without going through the molecular weight-retention time-composition three-dimensional data cube. This is because the response of hyaluronic acid in the diode array detector mainly originates from... to The weak absorption of nearby carboxyl groups and glycosidic bonds results in a signal much lower than that of co-existing components, making the inversion of hyaluronic acid molecular weight distribution from diode array data insufficiently robust. The digital signal of the first-dimensional effluent curve, acquired by an evaporative light scattering detector, is stable in response to large carbohydrate molecules and is more suitable as the source signal for molecular weight distribution calculation.

[0074] The extraction process consists of three steps. The first step involves reading the response values ​​of all sampling points within the hyaluronic acid elution window from the digital signal of the first-dimensional elution curve. in For the first in the window sampling points For the first The evaporative light scattering response values ​​at each sampling point. Since the evaporative light scattering detector's response to mass follows a power-law relationship, it can be directly used... Calculating molecular weight distribution based solely on mass concentration underestimates the contribution of the low-concentration range, resulting in an underestimation of molecular weight dispersion. Therefore, it is necessary to adjust the molecular weight distribution according to the hyaluronic acid response correction curve. Perform inverse transformation ,in For the first The quality response value of each sampling point. This is the inverse function of the hyaluronic acid response calibration curve, i.e., a mapping from the response value to the mass concentration. The hyaluronic acid response calibration curve is obtained during the method establishment phase by injecting a series of hyaluronic acid mass concentration standards into the same system, including at least [number missing]. , , , , , There are a total of 6 calibration points. For each calibration point, the peak response value is taken, and the response value and mass concentration are double logarithmic and then linearly fitted to obtain the result. When actually looking up the table, use retrieval Right now ,in The intercept is a double logarithmic fit. The slope of the double logarithmic fit corresponds to the aforementioned response index. The second step involves determining the sampling time for each sampling point according to the hyaluronic acid molecular weight standard curve. Convert to the corresponding molecular weight scale The molecular weight standard curve for hyaluronic acid is also denoted as . Forms among them , , , The coefficients of the third-order polynomial, This refers to the molecular weight value. To preserve time values. Step 3 is based on... and Perform numerical integration according to the following expression. The number average molecular weight was obtained. The weight-average molecular weight was obtained, among which Number-average molecular weight, which is the average molecular weight of hyaluronic acid molecules weighted by the number of molecules. The weight-average molecular weight is the average molecular weight of hyaluronic acid molecules weighted by mass. Molecular weight dispersion is determined by... Divide by This reflects the breadth or narrowness of the molecular weight distribution.

[0075] The concentration values ​​for each active ingredient were extracted from a three-dimensional data cube of molecular weight-retention time-component composition. For the... The total response area is obtained by summing the response intensity values ​​of all layers corresponding to the active ingredient along the molecular weight coordinate and retention time direction. in For the first The total response area of ​​the active ingredients For the first The active ingredients in the first The molecular weight coordinates and the first The significance of summing the response intensity values ​​at each retention time along two dimensions is to combine the responses of the active ingredient at all peak positions within the first-dimensional combined slicing window into a single total value. This is based on the pre-written values ​​in the on-chip block random access memory. The concentration standard curve of the active ingredient will Convert to concentration ,in For the first Concentration values ​​of various active ingredients For the first A standard curve function for the concentration of each active ingredient. During the method establishment phase, the standard curve for the concentration of each active ingredient is prepared using single-standard references for each active ingredient at least... The total response area obtained from the concentration point injection is fitted to the known concentration using linear regression or weighted linear regression.

[0076] The number-average molecular weight, weight-average molecular weight, molecular weight dispersion, and concentration values ​​of each active ingredient of hyaluronic acid constitute the final analytical results, which are output to an external host computer via a communication interface. The communication interface can be a gigabit Ethernet, serial port, or USB interface. The output data format uses Structured Markup Language or a parameter-value pairing format commonly used in manufacturing to ensure compatibility with different host computer software. In practical use, the results can also include quality control parameters, including the peak height and half-width at half-maximum (WHM) of the hyaluronic acid elution window during this run, the signal-to-noise ratio of the total response area of ​​each active ingredient, and the retention time drift from the system suitability test on the same day, serving as a basis for assessing the reliability of the analytical results.

[0077] In alternative implementations, the above process allows for the following substitutions. First, the size exclusion column can be replaced with a field flow fractionation device for molecular weights exceeding [a certain value]. The ultra-high molecular weight hyaluronic acid field flow fractionation column exhibits lower shear compared to size exclusion columns, resulting in less damage to the molecular chains. Secondly, the evaporative light scattering detector can be replaced with a differential refractive index detector or a combination of a multi-angle laser scattering detector and a differential refractive index detector. The multi-angle laser scattering detector can directly measure the gyration radius of the eluted molecules and thus directly calculate the absolute molecular weight, eliminating the need for establishing a molecular weight standard curve. Thirdly, the T-shaped mixing chamber in the online solvent transfer unit can be replaced with a vortex mixer integrated within a microfluidic chip. Chip integration further reduces the dead volume to [a smaller value]. The following steps improve the initial stability of the second-dimensional gradient. Fourth, hydrophilic interaction columns can be replaced with reversed-phase or mixed-mode columns. Reversed-phase columns offer higher selectivity for medical aesthetic excipients with predominantly hydrophobic active ingredients; mixed-mode columns are more versatile when the active ingredients are both polar and nonpolar. Fifth, diode array detectors can be connected in series with fluorescence detectors. For active ingredients with natural fluorescence, such as vitamin B2 and certain plant extracts, the detection limit of the fluorescence detector is 2 to 3 orders of magnitude lower, serving as a highly sensitive supplementary channel. Sixth, the stripping order in the stepwise stripping spectral decoupling pipeline can be sorted not by abundance, but by the sharpness of the wavelength characteristic peak or by the expected retention time of the active ingredient in the preset active ingredient elution window. Different sorting rules are suitable for sample systems of varying complexity, and can be adjusted specifically according to the particular medical aesthetic excipient formulation in practical applications. Seventh, the integration of the molecular weight distribution of hyaluronic acid in the results extraction module can be combined with moving average or three-point Simpson integration rules to further reduce the truncation error of numerical integration. The above optional implementation methods can be used individually or in combination without affecting the integrity of the overall technical solution.

[0078] refer to Figure 4 , Figure 4 The horizontal axis below represents the common logarithm of the molecular weight. The horizontal axis at the top represents the molecular weight scale in kilodaltons, and the vertical axis represents the differential mass fraction. . Figure 4 Two curves are plotted. The first curve is the uncorrected curve, obtained by directly using the original response value of the evaporative light scattering detector as the ordinate. The second curve is the corrected curve, obtained by applying mass response correction to the response values ​​at each sampling point using the hyaluronic acid response correction curve. The response of the evaporative light scattering detector follows a power-law relationship with the solute mass. ,in This represents the evaporative light scattering response value. For instantaneous mass, This is the proportionality coefficient. In response to the index and It is usually between 1.2 and 1.8. Because Since the original response value is not equal to 1, it cannot be directly equated to mass concentration. The peak shape of the curve before correction exhibits systematic distortion relative to the true distribution, with the high concentration range being exaggerated and the low concentration range being compressed. After inverse transformation of the hyaluronic acid response correction curve, the response values ​​at each sampling point are converted into mass response values, and the peak shape of the corrected curve is restored to the true differential mass fraction distribution. Figure 4 After correction, the curve shows one main peak and one shoulder peak on the low molecular weight side. The main peak is located at... At approximately 5.6, corresponding to a molecular weight of approximately 400 kilodaltons, the shoulder peak is located at... At approximately 4.8, corresponding to a molecular weight of approximately 60 kilodaltons, it indicates that the hyaluronic acid in this medical aesthetic excipient has a bimodal distribution. Figure 4 The number-average molecular weights are marked with vertical lines. and weight average molecular weight The location. The average molecular weight of hyaluronic acid molecules is calculated by weighting the number of molecules. The calculation yielded, where For the first The corrected quality response value of each sampling point. For the first The molecular weight corresponding to each sampling point This represents the logarithmic molecular weight step size. The average molecular weight of hyaluronic acid molecules is calculated by weighting the molecular weight of the molecules. Calculated. Molecular weight dispersion is derived from... Divide by The result reflects the breadth of the molecular weight distribution; the closer the value is to 1, the narrower the distribution. Figure 4 The molecular weight dispersion indicated in the label is approximately 1.65, which means that the molecular weight distribution of hyaluronic acid in this medical aesthetic excipient is of medium width, consistent with the typical characteristics of medical-grade hyaluronic acid.

[0079] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for the combined analysis of hyaluronic acid and multiple active ingredients in medical aesthetic excipients, characterized in that, Includes the following steps: Step 1: Add the medical aesthetic excipient sample to the mobile phase dilution solution and mix well to obtain the test solution. After filtering the test solution through a microporous membrane, inject it into a two-dimensional chromatography system based on a dual-column switching valve and FPGA parallel spectrum interpretation. Step 2: The test solution is separated by a size exclusion column in descending order of fluid dynamic volume. An evaporative light scattering detector continuously detects the effluent from the size exclusion column and outputs a simulated signal of the first-dimensional effluent curve. The FPGA parallel resolution unit samples the simulated signal of the first-dimensional effluent curve to obtain a digital signal of the first-dimensional effluent curve. The FPGA parallel resolution unit performs first-dimensional coupled slice window recognition on the digital signal of the first-dimensional effluent curve and continuously sends switching pulses to the ten-way two-position electric switching valve within the first-dimensional coupled slice window to segment the effluent into effluent fragments. After second-dimensional separation by a hydrophilic column, the effluent fragments are scanned at full wavelength by a diode array detector to output a digital measured absorbance vector. The stepwise stripping spectral decoupling pipeline in the FPGA parallel resolution unit performs multi-active component stepwise spectral stripping on the digital measured absorbance vector to obtain the response intensity value of each active component. The data cube assembly module assembles the response intensity values ​​of each active component into a three-dimensional data cube of molecular weight-retention time-component according to molecular weight coordinates, retention time, and component number. The post-column flow path of the size exclusion chromatography column is immediately connected to a low dead volume tee connector. The two outlets of the tee are connected to the inlet of the evaporative light scattering detector and the injection port of the ten-way two-position electric switching valve, respectively, forming two parallel flow paths: the detection branch and the sampling branch. On the sampling branch, a delay capillary tube is connected in series between the three-way connector and the sampling port of the ten-way two-position electric switching valve, so that the time when the effluent of the same elution volume reaches the sampling loop inlet is strictly later than the time when it reaches the detection cell of the evaporative light scattering detector. The specifications of the delayed capillary are an inner diameter of 0.25 mm, a length of 1.5 m to 4.0 m, and an internal volume of 75 μL to 200 μL, wound into a spiral shape with a coil diameter of 10 mm to 20 mm; The phase compensation amount is the time when the effluent of the same elution volume reaches the sampling loop inlet minus the time when it reaches the detection cell of the evaporative light scattering detector. The phase compensation amount is obtained by actual measurement and calibration of the tracer after the system is assembled, and the phase compensation amount is kept positive and not less than 5s by connecting a delay capillary in series on the sampling branch. In step 2, the first-dimensional slicing window consists of two sub-windows: a hyaluronic acid elution window and a preset active ingredient elution window. In step 2, the valve-controlled timing generator in the FPGA parallel spectrum descaling unit uses the starting time of each sub-window in the first-dimensional combined slicing window, superimposed with the phase compensation amount, as the slice start time corresponding to each sub-window. According to the preset modulation period, it continuously outputs the switching time in the first-dimensional combined slicing window until the current switching time reaches the end time of each sub-window and superimposed with the phase compensation amount. At each switching time, the valve-controlled timing generator sends one switching pulse to the drive interface of the ten-way two-position electric switching valve through the digital pulse output port. In step 2, the valve-controlled timing generator writes the phase difference between each switching moment and the peak moment of the hyaluronic acid elution window as the molecular weight positioning phase difference of the corresponding elution fragment into the elution window timing register group; the phase compensation amount has been pre-calibrated to a unified phase reference to calibrate the transmission time difference between the detection branch and the sampling branch; the difference between the switching moment and the peak moment corresponds one-to-one with the elution position passed by the corresponding molecular weight component of the elution fragment on the size exclusion column; each elution fragment is locked to a certain molecular weight range on the size exclusion column, and the center molecular weight scale of the locked molecular weight range is used as the molecular weight coordinate of the corresponding elution fragment; The peak moment is only used to determine the molecular weight localization phase difference of each effluent fragment; the molecular weight localization phase difference is used in the data cube assembly stage after all data acquisition is completed, and does not participate in any real-time valve control decisions; The working process of the stepwise stripping spectral decoupling pipeline is divided into two stages. The first stage is to determine the stripping order, and the second stage is to execute the stepwise stripping. The first stage is started once before each effluent fragment is processed. For each active ingredient, its pure reference spectrum is obtained by scanning the single standard reference under the same wavelength grid during the method establishment stage, and then normalized to a unit vector with a L2 norm of 1 and stored in the on-chip block random access memory of the FPGA parallel spectral decomposition unit. In step 2, the stepwise stripping spectral decoupling pipeline includes a correlation sorting controller and multiple tandem stripping stages. The total number of stripping stages is equal to the total number of pure reference spectra of active ingredients. Before stripping begins, the correlation sorting controller performs normalized cross-correlation calculations on each pure reference spectrum of active ingredients with the digitally measured absorbance vector to obtain the initial correlation value of each active ingredient. It then generates an active ingredient stripping sequence index table in descending order of the initial correlation values. Each tandem stripping stage sequentially retrieves the corresponding pure reference spectrum of active ingredients from the on-chip block random access memory according to the active ingredient stripping sequence index table. The digitally measured absorbance vector serves as the initial input residual absorbance vector for the first tandem stripping stage. In each stripping stage, the hardware matching core performs a normalized cross-correlation operation on the input residual absorbance vector and the reference spectrum of the active ingredient called by that stage, outputting the response intensity value of the active ingredient at the current retention time. The hardware stripping core in each stage removes the spectral contribution of the active ingredient from the input residual absorbance vector based on the response intensity value and the reference spectrum of the active ingredient called by that stage, obtaining a new residual absorbance vector, which is then passed to the next stripping stage. Each stripping stage operates in parallel in a pipeline manner; while the first stripping stage processes the data of the current effluent fragment, the second stripping stage simultaneously processes the data of the previous effluent fragment. Normalized cross-correlation is an inner product operation that normalizes the reference spectral vector to the L2 norm while keeping the original amplitude of the measured absorbance vector unchanged. Output at each level By reversing the index table according to the stripping order, we can obtain the response intensity values ​​of each active ingredient at the current retention time point; where... This represents the response intensity value of the active ingredient at the current retention time point. In step 2, the data cube assembly module writes the response intensity values ​​of each active ingredient output from the stepwise stripping spectral decoupling pipeline into the on-chip dual-port random access memory according to three dimensions: molecular weight coordinates, retention time, and ingredient number. The retention time is the retention time of the current response intensity value on the second-dimensional chromatogram corresponding to the current effluent fragment. The molecular weight coordinates are calculated by combining the molecular weight positioning phase difference of each effluent fragment written into the elution window time register group by the valve-controlled timing generator with the standard curve of hyaluronic acid molecular weight. After the writing is completed, the data stored in the on-chip dual-port random access memory constitutes a three-dimensional data cube of molecular weight-retention time-component. Step 3: The result extraction module extracts the number-average molecular weight, weight-average molecular weight, and molecular weight dispersion of hyaluronic acid from the digital signal of the first-dimensional effluent curve, and extracts the concentration value of each active ingredient from the three-dimensional data cube of molecular weight-retention time-component, and outputs it as the combined analysis result.

2. The method as described in claim 1, characterized in that, The two-dimensional chromatography system based on a dual-column switching valve and FPGA parallel spectrometry includes a one-dimensional infusion pump, an injector, a size exclusion column, an evaporative light scattering detector, a 10-port 2-position electric switching valve, sampling loops, an online solvent conversion unit, a two-dimensional infusion pump, a hydrophilic interaction column, a diode array detector, and an FPGA parallel spectrometry unit. The one-dimensional infusion pump, injector, and size exclusion column are connected sequentially. The one-dimensional infusion pump delivers an aqueous phase containing volatile buffer salts to the injector as the one-dimensional mobile phase. The outlet of the size exclusion column is connected to the inlet of the evaporative light scattering detector and the injection port of the 10-port 2-position electric switching valve via a three-way tubing. The 10-port 2-position electric switching valve has two sampling loops of the same volume. The outlet of the 10-port 2-position electric switching valve is connected to the inlet of the hydrophilic interaction column via the online solvent conversion unit. The two-dimensional infusion pump delivers a water-acetonitrile mixture with a higher acetonitrile volume ratio than the one-dimensional mobile phase to the online solvent conversion unit as the two-dimensional mobile phase. The diode array detector is connected to the outlet of the hydrophilic interaction column.

3. The method as described in claim 2, characterized in that, The diode array detector includes a diode array photoelectric conversion module and a multi-channel analog-to-digital conversion module. The multi-channel analog-to-digital conversion module converts the absorbance signals of each wavelength channel into digitally measured absorbance vectors. The FPGA parallel spectrum resolution unit is connected to the analog output terminal of the evaporative light scattering detector through the analog-to-digital conversion channel. The FPGA parallel spectrum resolution unit is connected to the data output terminal of the multi-channel analog-to-digital conversion module through a high-speed digital data interface. The digital pulse output port of the FPGA parallel spectrum resolution unit is connected to the drive interface of the ten-way two-position electric switching valve.

4. The method as described in claim 2, characterized in that, The on-chip block random access memory of the FPGA parallel spectrum interpretation unit is pre-written with the pure reference spectra of each active ingredient, the concentration standard curves of each active ingredient, the molecular weight standard curve of hyaluronic acid, the response correction curve of hyaluronic acid, and the phase compensation amount. The hyaluronic acid response correction curve is obtained by calibrating the evaporation light scattering response values ​​of a series of hyaluronic acid mass concentration standards between known mass concentrations and the measured values ​​on a two-dimensional chromatography system based on a dual-column switching valve and FPGA parallel spectrum interpretation. The phase compensation amount was determined by calibrating the dead volume of the detection branch between the size exclusion column outlet and the evaporative light scattering detector detection cell, the dead volume of the sampling branch between the size exclusion column outlet and the sampling loop inlet, and the one-dimensional mobile phase flow rate.

5. The method as described in claim 1, characterized in that, In step 2, the peak detection state machine in the FPGA parallel spectral analysis unit sends the current sampled value of the first-dimensional elution curve digital signal and the previous sampled value into a subtractor to obtain a difference value. When the difference value changes from negative to positive and exceeds a preset noise threshold, the current sampling time is determined to be the start time of the hyaluronic acid elution window. When the difference value changes from positive to negative, the current sampling time is determined to be the peak time of the hyaluronic acid elution window. When the absolute value of the difference value is continuously lower than the preset noise threshold and maintains a preset hold count, the current sampling time is determined to be the end time of the hyaluronic acid elution window. The retention time interval of each active ingredient on the size exclusion column is used as the start and end times of the preset active ingredient elution window and is pre-written into the on-chip block random access memory. The peak detection state machine sequentially writes the start time, peak time, end time of the hyaluronic acid elution window and the start and end times of the preset active ingredient elution window into the elution window time register group.

6. The method as described in claim 5, characterized in that, In step 2, the 10-way 2-position electric switching valve alternately switches between the sampling position and the injection position each time it receives a switching pulse, so that the two sampling loops alternately are in the sampling state and the injection state. The sampling loop in the sampling state continuously retains one effluent fragment of the size exclusion column for one modulation cycle to obtain one effluent fragment. The effluent fragment in the sampling loop in the injection state is mixed with the two-dimensional mobile phase by the online solvent conversion unit and then enters the hydrophilic interaction column for second-dimensional separation. The online solvent conversion unit adjusts the acetonitrile volume ratio in the effluent fragment so that the effluent fragment meets the initial retention conditions of the hydrophilic interaction column.

Citation Information

Patent Citations

  • Liquid phase chromatography data processing system based on FPGA

    CN101769901A