Method for rapidly identifying arbutin in cosmetics based on surface enhanced Raman spectroscopy

Abstract

The invention relates to the technical field of cosmetic component detection and analysis, and discloses a method for rapidly recognizing arbutin in cosmetics based on surface enhanced Raman spectroscopy, AgNPs sol is prepared by an ascorbic acid oxidation-reduction method, and an SERS sensing interface is constructed through aspartic acid electro-polymerization modification; for different dosage forms of cosmetics, a to-be-detected solution is obtained through methanol extraction and centrifugal filtration purification, the to-be-detected solution and AgNPs sol are mixed and adsorbed, then a Raman spectrum is collected, pretreatment is conducted through combination of multivariate scatter correction and standard orthogonal transformation, characteristic wavelengths are screened through a genetic algorithm-combined interval partial least square method, and detection is achieved through combination of qualitative and quantitative models. Through cooperation of sensing interface specific modification and a data processing technology, the arbutin signal recognition capability is enhanced, matrix interference is eliminated, rapid and accurate detection of arbutin in cosmetics is realized, and the effects of high sensitivity, high specificity and simple and convenient operation are achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of cosmetic ingredient detection and analysis, in particular to a rapid identification method of arbutin in cosmetics based on surface-enhanced Raman spectroscopy. BACKGROUND

[0002] With the continuous improvement of people's pursuit of beauty, in the booming of the cosmetics industry, it is essential to ensure the safety and effectiveness of cosmetic ingredients. Whitening cosmetics have always been favored by consumers. Arbutin, a small molecule phenolic compound, is a commonly used whitening ingredient in cosmetics. It can inhibit the activity of tyrosinase and effectively reduce the production of melanin, thereby achieving the effect of whitening skin and lightening spots. It has been widely used in the field of cosmetics. With the growing demand for whitening skincare from consumers, arbutin has become one of the core additives of whitening cosmetics (such as creams and serums). However, to ensure product safety and compliance, the Cosmetic Safety Technical Specification has strict limits on the amount of arbutin added. Excessive addition or substandard behavior in the market not only harms consumer rights and interests but also poses potential safety risks. Therefore, developing accurate and efficient arbutin detection technology is of great significance for cosmetic quality control and market supervision.

[0003] Currently, the detection of arbutin in cosmetics mainly relies on traditional analysis techniques such as high-performance liquid chromatography (HPLC) and gas chromatography-mass spectrometry (GC-MS). These methods have high sensitivity and accuracy, but they are generally expensive, complex, time-consuming, and require professional operators, which makes it difficult to meet the needs of rapid screening in production sites or large-scale, rapid sampling by market supervision departments.

[0004] Surface-enhanced Raman spectroscopy (SERS) technology has great potential in the field of trace substance detection due to its high sensitivity, fingerprint identification ability, small sample size, and fast detection speed. However, direct application of SERS technology to the detection of arbutin in cosmetics still faces significant challenges.

[0005] Firstly, cosmetics are a complex multi-phase system containing a large amount of oil, surfactant, polymer and various functional ingredients, which will produce strong background Raman signals, seriously interfere with the identification of the characteristic peaks of the target substance arbutin, and cause the detection specificity to decrease. Secondly, the conventional SERS active substrate (such as metal nano-sol) has Raman enhancement effect on most organic matters, lacks specific recognition ability for arbutin molecules, cannot effectively distinguish structural analogues (such as hydroquinone, phenol, etc.), and is easy to produce false positive results. Finally, the reproducibility and stability of the SERS signal are significantly affected by the uniformity and aggregation state of the substrate, which is more difficult to control in a complex sample system, and affects the reliability of quantitative analysis. Therefore, how to construct a SERS sensing interface with high sensitivity, high specificity and effective resistance to the interference of complex cosmetic matrix, so as to realize the rapid and accurate identification and quantitative detection of arbutin, has become a technical problem to be solved in the field. SUMMARY

[0006] The purpose of the present application is to provide a rapid identification method of arbutin in cosmetics based on surface-enhanced Raman spectroscopy, which can enhance the signal recognition ability of arbutin by the cooperation of sensing interface specific modification and data processing technology, eliminate matrix interference, realize rapid and accurate detection of arbutin in cosmetics, and achieve the effects of high sensitivity, high specificity and simple operation, so as to solve the problems raised in the above background art.

[0007] To achieve the above purpose, a rapid identification method of arbutin in cosmetics based on surface-enhanced Raman spectroscopy is provided, which comprises the following steps:

[0008] a) AgNPs sol is prepared by ascorbic acid redox method as SERS enhancement substrate, the particle size of the AgNPs sol is 50-80 nm, and the Raman signal enhancement performance is confirmed by ultraviolet absorption spectrum and scanning electron microscope;

[0009] b) Aspartic acid is used as monomer to modify the surface of AgNPs to form a polymer film by electropolymerization to construct a SERS sensing interface;

[0010] c) Methanol is used as an extraction reagent for targeted extraction according to different dosage forms of cosmetics, and the sample solution to be tested is obtained after centrifugation, filtration and purification;

[0011] d) The sample solution to be tested is mixed with AgNPs sol by vortexing at a volume ratio of 1:2-1:4, arbutin is fully adsorbed on AgNPs by standing for 0.5-3 minutes, and then it is dropped and coated on a pretreated quartz slide and naturally air-dried;

[0012] e) Laser confocal Raman spectrometer is used to collect Raman spectrum, and the silicon sheet 520.7 cm -1The peak is a reference peak for calibrating the instrument, the excitation light source wavelength is 532 nm±5 nm, and the detection spectral region is 400-3500 cm -1 ;

[0013] f) The original spectrum is pretreated by combining multiple scattering correction (MSC) and standard normal variate (SNV), characteristic wavelengths are screened by genetic algorithm-interval partial least squares (GA-SiPLS), and specific identification and content detection of arbutin are realized by combining qualitative identification model and quantitative prediction model.

[0014] Preferably, the preparation raw material specifications of the AgNPs sol in step a) are as follows: polyvinylpyrrolidone (PVP) is K30 type, the molecular weight is 40000±2000, and the purity is ≥99.0%; the concentration of silver nitrate is 0.08-0.12 mol / L; the purity of ascorbic acid is ≥99.9%; the concentration of sodium hydroxide is 0.4-0.6 mol / L; and the concentration of sodium chloride is 4-6 mol / L.

[0015] Preferably, the preparation conditions of the AgNPs sol in step a) are as follows: the reaction temperature is controlled at 20-30℃, the magnetic stirring rate in the mixing stage of PVP and silver nitrate is 400-600 r / min, the rapid stirring rate when sodium chloride is added is 1000-1400 r / min, the reaction is carried out in a light-proof environment for 1.5-2.5 hours after the mixed solution of ascorbic acid and sodium hydroxide is added, and the stirring rate is maintained at 400-600 r / min.

[0016] Preferably, the specific operation process of electro polymerization in step b) is as follows: the AgNPs sol is uniformly drop-coated on the surface of a glassy carbon sheet, and the glassy carbon sheet is dried by an infrared lamp to serve as a working electrode; the working electrode is inserted into a phosphate buffer solution containing 2 mmol / L aspartic acid, the concentration of the buffer solution is 0.005-0.02 mol / L, and the pH value is 6.5-7.5; a platinum wire auxiliary electrode and a saturated calomel reference electrode are connected, and cyclic voltammetry scanning is carried out for 8-12 cycles at a potential window of-1.0-2.0 V and a scanning rate of 80-120 mV / s, so that aspartic acid is polymerized on the surface of AgNPs to form a dense polymer film.

[0017] Preferably, the specific operation of sample extraction and purification in step c) is as follows: 0.5 g of a cosmetic sample is placed in a 10 mL centrifuge tube, and 5-8 mL of methanol is added according to the dosage form; the cream and solid cosmetic are vortexed for 2 minutes and ultrasonically extracted for 18-25 minutes, the essence and emulsion are vortexed for 1 minute and ultrasonically extracted for 12-18 minutes; after extraction, centrifugation is carried out at a speed of 9000-12000 r / min for 8-12 minutes, the supernatant is filtered through a 0.45 μm organic phase nylon filter membrane, and the filtrate is collected as a sample solution to be tested.

[0018] Preferably, the optimal mixing volume ratio of the sample solution to be tested and the AgNPs sol in step d) is 1:3, the vortex oscillation rate after mixing is 2500-3500 r / min, the oscillation time is 0.5-1 minute, and the adsorption time is 1-2 minutes. It is detected that under the above conditions, the adsorption efficiency of arbutin and AgNPs is ≥90%, and the characteristic Raman peak intensity reaches the maximum value.

[0019] Preferably, the specific parameters for collecting Raman spectra in step e) are as follows: the laser intensity is set to 8-12 mW, the sample is observed using an 80-100x objective lens, the integration time is 20-40 s, and the integration number is 2-4 times; 10-20 different sampling points are randomly selected for each sample to collect spectra, and after removing abnormal spectra deviating from the average value ± 3 times the standard deviation of all spectra, the average value of the remaining effective spectra is taken as the final detection spectrum.

[0020] Preferably, the core algorithm of the multiple scattering correction (MSC) in step f) is as follows: the average spectrum of all sample original spectra is taken as the reference spectrum The original spectrum vector x of each sample is subjected to linear regression with The model is:

[0021]

[0022] Wherein, a is the regression coefficient (slope), and b is the intercept. The values of a and b are solved by the least squares method, the original spectrum is corrected using the obtained parameters, and the corrected spectrum

[0023]

[0024] Wherein, x is the original spectrum intensity vector; the formula of the standard orthogonal transformation (SNV) is: Wherein, μ is the intensity mean value of a single original spectrum, and σ is the intensity standard deviation of a single original spectrum.

[0025] Preferably, the specific process of genetic algorithm-union interval partial least squares (GA-SiPLS) for screening characteristic wavelengths in step f) is as follows: first, divide the spectral region of 400-3500 cm -1 into 25-35 continuous intervals at an interval of 100 cm -1 ; the genetic algorithm parameters are set as population size 40-60, crossover probability 0.7-0.9, mutation probability 0.03-0.07, and iteration number 80-120; the root mean square error of cross-validation (RMSECV) is used as the fitness function to screen the target interval containing the arbutin characteristic peak; and then the union interval partial least squares method is used to optimize the target interval, and finally 20-30 characteristic wavelength points are obtained for model construction.

[0026] Preferably, the qualitative identification model in step f) is a combination of artificial neural network (ANN), K-nearest neighbor method (KNN) and Mahalanobis distance method (MD) with weighted combination; wherein the weight proportion of ANN is 40%, the weight proportion of KNN is 35%, and the weight proportion of MD is 25%; the output results (0 for no arbutin, 1 for arbutin) are calculated by weight superposition to obtain a score, and a score ≥ 0.6 is determined as containing arbutin, and a score < 0.6 is determined as not containing arbutin; ANN is a three-layer structure, the number of input layer nodes corresponds to the number of characteristic wavelength points, two layers are set in the hidden layer (the number of nodes is 12-18 and 6-10, respectively), there is one node in the output layer, and the activation function is Sigmoid function

[0027]

[0028] The K value of KNN is 3-7, and the Euclidean distance is used for distance measurement:

[0029]

[0030] (m is the number of characteristic wavelengths); the Mahalanobis distance formula is:

[0031]

[0032] (μ is the mean value vector of the arbutin standard sample, and Σ is the covariance matrix), and the threshold value is set to 2.5-3.5; the quantitative prediction model is a combination of partial least squares-support vector machine (PLS-SVM) as the main model and joint interval partial least squares (SiPLS) as the verification model; the weight proportion of PLS-SVM is 80%, and the weight proportion of SiPLS is 20%, and the final quantitative result is calculated according to “main model result × 0.8 + verification model result × 0.2”; PLS-SVM extracts 4-6 principal components as input after PLS dimension reduction, and the kernel function is radial basis function:

[0033] K(x i ,x j )=e -γ ‖x i -x j ‖ 2 ,(γ=0.05-0.2, penalty parameter C=8-12); a standard curve is constructed based on the characteristic wavelength points, the arbutin concentration is taken as the abscissa, and the characteristic peak intensity is taken as the ordinate, the linear correlation coefficient R 2 ≥ 0.95, the relative standard deviation (RSD) of the quantitative result is ≤ 3%, and the recovery rate is 92%-98%.

[0034] The present application has the following beneficial effects over the prior art: 1. High detection sensitivity, the AgNPs sol prepared in the present application has uniform particle size and is in the optimal enhancement interval of 50-80 nm, and the specific sensing interface modified by aspartic acid polymer has a detection limit of arbutin as low as 10 -9 mol / L, which is more than 100 times higher than the sensitivity of the traditional HPLC method, can accurately detect trace arbutin in cosmetics, and meets the detection requirements of the arbutin addition limit value in the 'Cosmetic Safety Technical Specification'.

[0035] 2. Strong specificity, specific recognition is realized through hydrogen bonding and hydrophobic interaction between aspartic acid polymer and arbutin molecules, arbutin and structural analogs such as hydroquinone and phenol can be effectively distinguished, and the background interference of complex cosmetic matrix can be significantly eliminated through the combination of multi-element scattering correction and standard orthogonal transformation pretreatment, and the detection specificity is still maintained at more than 95% in the presence of common cosmetic ingredients such as vitamin C and nicotinamide, effectively avoiding false positive results.

[0036] 3. Fast detection speed, the entire detection process from sample processing to result output only takes 5-10 minutes, which is greatly shortened compared with the detection period of more than 60 minutes of the traditional HPLC method, and can meet the rapid sampling requirements of real-time quality control in cosmetic production sites and on-site inspection by supervision departments, and improve the detection efficiency.

[0037] 4. Excellent repeatability and stability, the Raman signal intensity of the AgNPs sol changes by less than 10% under the condition of 4°C sealed and light shielding for 30 days, the relative standard deviation of 10 repeated detections of the same sample is less than or equal to 3%, and the relative deviation of detection results by different laboratories and different operators is less than or equal to 4.1%, which ensures the reliability and consistency of the detection results.

[0038] 5. Wide adaptability, the sample processing procedure is optimized for different dosage forms such as cream, essence, emulsion and solid cosmetic, the amount of methanol, extraction time and centrifugal speed are adjusted to realize effective detection of arbutin in various whitening cosmetics, without the need for additional adjustment of core detection parameters, and the application range covers mainstream whitening cosmetic dosage forms.

[0039] 6. Simple operation and low cost, without the need for complex sample pretreatment equipment and professional operators, the raw materials for preparing the AgNPs sol are easy to obtain and have low cost, and no expensive reagents such as radioactive isotopes are needed in the detection process, the cost of single sample detection is only about 16% of that of the HPLC method, which reduces the detection cost and facilitates large-scale popularization and application. BRIEF DESCRIPTION OF DRAWINGS

[0040] Figure 1 It is an illustration of the preparation of the AgNPs sol and the enhancement of the Raman signal of arbutin in the present application;

[0041] Figure 2 UV spectrum of AgNPs of the present application;

[0042] Figure 3 TEM image of AgNPs of the present application;

[0043] Figure 4 Raman spectrum of standard arbutin sample of the present application and comparison spectrum with blank sample;

[0044] Figure 5 Raman spectrum of arbutin solution of different concentrations of the present application and comparison before and after signal enhancement;

[0045] Figure 6 Linear response graph of Raman spectrum concentration of the present application A. Raman spectrum graph under different concentrations; B. Linear relationship graph of characteristic peak intensity and concentration;

[0046] Figure 7 Raman spectrum repeatability graph of AgNPs-α-arbutin of the present application;

[0047] Figure 8 Schematic diagram of SERE sensing interface construction of the present application. DETAILED DESCRIPTION

[0048] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments.

[0049] In the embodiments of the present application, a rapid identification method of arbutin in cosmetics based on surface-enhanced Raman spectroscopy is provided. The method realizes high sensitivity, high specificity, rapid and accurate identification and content detection of arbutin in cosmetics by constructing an aspartic acid polymer modified AgNPs sol SERS sensing interface, combining a targeted sample processing procedure, optimized spectral acquisition parameters and a weighted combined qualitative and quantitative model.

[0050] I. Reagents and instruments

[0051] The main instruments and reagents required for the implementation of the present application are as follows: the instruments include a laser confocal Raman spectrometer, model LabRAMHREvolution, France Horiba Company, equipped with a 532nm±5nm laser, the laser intensity can be adjusted in the range of 0~20mW, the objective magnification can be selected from 50x, 80x and 100x, the spectral resolution is 1cm -1 ;

[0052] A scanning electron microscope, model SU8010, Hitachi High-Technologies Corporation, acceleration voltage 0.5~30kV, resolution 1.0nm;

[0053] Constant temperature magnetic stirrer, model 85-2, Jierui Electric Appliance Co., Ltd. of Jintan City, stirring rate adjustment range 0-2000 r / min, temperature control accuracy ±1℃;

[0054] Centrifuge, model 5424R, Eppendorf, Germany, rotation speed up to 14000 r / min, centrifugal capacity 0.2-50 mL;

[0055] Ultrasonic cleaner, model KQ-500DE, Kunshan Ultrasonic Instrument Co., Ltd., frequency 40 kHz;

[0056] Electrochemical workstation, model CHI660E, Shanghai Chenhua Instrument Co., Ltd., potential range -2-2 V, scan rate 0.01-1000 mV / s;

[0057] Ultraviolet-visible spectrophotometer, model UV-2600, Shimadzu Corporation, Japan, wavelength range 190-900 nm;

[0058] Micropipette, specifications 10 μL, 20 μL, 100 μL, 200 μL, 1000 μL, Eppendorf, Germany;

[0059] Quartz glass slide, cleaned with ethanol and deionized water by ultrasonic and dried with nitrogen;

[0060] Glassy carbon sheet, polished to mirror surface with metallographic sandpaper before use, and then cleaned by ultrasonic;

[0061] Platinum wire auxiliary electrode;

[0062] Saturated calomel reference electrode, model 217, Shanghai Raytheon Instrument Factory.

[0063] Reagents include polyvinylpyrrolidone, K30 type, molecular weight 40000±2000, purity 99.0%, Shanghai Maikelin Biochemical Technology Co., Ltd.;

[0064] Silver nitrate, analytical pure, concentration 0.1 mol / L, National Pharmaceutical Group Chemical Reagent Co., Ltd.;

[0065] Ascorbic acid, chromatographically pure, purity 99.9%, Sigma-Aldrich Corporation;

[0066] Sodium hydroxide, analytical pure, concentration 0.5 mol / L, National Pharmaceutical Group Chemical Reagent Co., Ltd.;

[0067] Sodium chloride, extra pure, concentration 5 mol / L, National Pharmaceutical Group Chemical Reagent Co., Ltd.;

[0068] Aspartic acid, biological reagent grade, purity 98.0%, Shanghai Yuanye Bio-Technology Co., Ltd.;

[0069] Potassium dihydrogen phosphate, analytical pure, purity 99.5%, National Pharmaceutical Group Chemical Reagent Co., Ltd.;

[0070] Disodium hydrogen phosphate, analytical pure, purity 99.0%, National Pharmaceutical Group Chemical Reagent Co., Ltd.;

[0071] Methanol, chromatographic pure, purity 99.9%, Fisher Scientific Company;

[0072] Alpha-arbutin standard, purity 99.9%, Sigma-Aldrich Company;

[0073] Blank cosmetic matrix, cream, essence, emulsion, solid cosmetic without arbutin, commercially available;

[0074] Vitamin C, purity 99.5%, National Pharmaceutical Group Chemical Reagent Co., Ltd.;

[0075] Nicotinamide, purity 99.0%, Shanghai Yuan Ye Biological Technology Co., Ltd.;

[0076] Hyaluronic acid, molecular weight 100000, purity 98.0%, Shanghai Maikelin Biological Technology Co., Ltd.;

[0077] Glycerol, analytical pure, purity 99.0%, National Pharmaceutical Group Chemical Reagent Co., Ltd.;

[0078] Ceramide, purity 98.0%, Shanghai Yuan Ye Biological Technology Co., Ltd.;

[0079] Hydroquinone, purity 99.0%, National Pharmaceutical Group Chemical Reagent Co., Ltd.;

[0080] Phenol, analytical pure, purity 99.0%, National Pharmaceutical Group Chemical Reagent Co., Ltd.

[0081] II. Determination of experimental conditions

[0082] 2.1 Preparation of AgNPs sol, first take 5 mL of polyvinylpyrrolidone and add it to 20 mL of deionized water, place it on a constant temperature magnetic stirrer, set the temperature to 25°C, the stirring rate is 500 r / min, stir for 10 minutes to make the polyvinylpyrrolidone completely dissolved, then add 85 mL of 0.1 mol / L silver nitrate solution, continue to stir for 20 minutes to form a uniform mixture.

[0083] Under rapid stirring, slowly add 200 μL of 5 mol / L sodium chloride solution to the mixture, the dropwise addition speed is 1 mL / min, after the dropwise addition is completed, the stirring rate is increased to 1200 r / min, and the stirring is continued for 30 minutes to prepare AgCl colloid, which is ready for use.

[0084] Another 20 mL of 50 mol / L ascorbic acid solution was added to 2.5 mL of 0.5 mol / L sodium hydroxide solution, stirred for 5 minutes to make them fully react, then 2.3 mL of freshly prepared AgCl colloid was added, and the mixture was transferred to a brown bottle. In a dark environment, the stirring rate was set to 500 r / min, and the reaction was carried out for 2 hours to obtain AgNPs sol.

[0085] The prepared AgNPs sol was characterized by ultraviolet-visible spectrophotometer, and the ultraviolet absorption spectrum was detected. The maximum absorption peak was at 420 nm (as shown in Figure 2 ), indicating that the AgNPs were successfully prepared. The morphology and particle size of the AgNPs were observed by scanning electron microscope. The results showed that the AgNPs were spherical, the particle size distribution was 50-80 nm, the average particle size was 65 nm, and the dispersity was good, meeting the requirements of SERS enhancement substrate.

[0086] 2.2 Construction of SERS sensing interface. 10 μL of the prepared AgNPs sol was uniformly dropped and coated on the surface of a glassy carbon sheet, which was placed under an infrared lamp for drying for 15 minutes to obtain an AgNPs modified glassy carbon sheet working electrode. A 0.01 mol / L phosphate buffer solution was prepared, the pH value was adjusted to 7.0, then aspartic acid was added to make the final concentration 2 mmol / L, and the mixture was stirred uniformly as an electrolyte.

[0087] The working electrode, platinum wire auxiliary electrode and saturated calomel reference electrode were connected to an electrochemical workstation, and the three electrodes were inserted into the electrolyte to ensure that the AgNPs were fully immersed in the electrolyte. The potential window was set to -1.0-2.0 V, the scanning rate was 100 mV / s, and the cyclic voltammetry scanning was performed for 10 cycles to complete the electropolymerization of aspartic acid on the surface of AgNPs to form a dense aspartic acid polymer film, and the SERS sensing interface was constructed. The construction process is shown in Figure 8 .

[0088] The SERS sensing interface was characterized by Fourier infrared spectroscopy. The results showed that the N-H stretching vibration peak intensity at 3300 cm -1 was reduced by 35% compared with that of aspartic acid monomer, and a new absorption peak appeared at 1250 cm -1 , which was attributed to the stretching vibration of C-N covalent bond, indicating that aspartic acid was combined with the surface of AgNPs through carbon-nitrogen covalent bond. After arbutin was combined with the sensing interface, the Ar-OH stretching vibration peak was shifted from 1044 cm -1 to 1042 cm -1 , and the peak width increased, realizing specific recognition. The interface modification effect was preliminarily verified by Raman spectrum. The Raman spectrum of the modified interface was collected, and there was no interference peak of aspartic acid monomer irrelevant to AgNPs, and after subsequent combination with arbutin, it was indirectly indicated that the interface modification was successful.

[0089] 2.3 Sample processing condition optimization, optimize sample processing parameters for different dosage forms of cosmetics;

[0090] For cream samples with oil content ≥ 30%, take 0.5 g of cream sample and place it in a 10 mL centrifuge tube, add 8 mL of methanol, vortex for 2 minutes, then place it in an ultrasonic cleaner, ultrasonic extraction for 20 minutes, ultrasonic power 500 W, frequency 40 kHz, after extraction, put the centrifuge tube into the centrifuge, set the speed to 12000 r / min, centrifuge for 10 minutes, take the supernatant and filter it through a 0.45 μm organic phase nylon filter membrane, collect the filtrate as the sample solution to be tested.

[0091] For serum samples with water content ≥ 80%, take 0.5 g of serum sample and place it in a 10 mL centrifuge tube, add 5 mL of methanol, vortex for 1 minute, ultrasonic extraction for 15 minutes, centrifuge at 10000 r / min for 10 minutes, filter to obtain the sample solution to be tested.

[0092] For emulsion samples with oil content 10% ~ 30%, take 0.5 g of emulsion sample and place it in a 10 mL centrifuge tube, add 6 mL of methanol, vortex for 1 minute, ultrasonic extraction for 18 minutes, centrifuge at 11000 r / min for 10 minutes, filter to obtain the sample solution to be tested.

[0093] For solid cosmetic samples, after being pulverized by a high-speed pulverizer, pass through an 80-mesh sieve, take 0.5 g of the undersize sample and place it in a 10 mL centrifuge tube, add 7 mL of methanol, vortex for 2 minutes, ultrasonic extraction for 25 minutes, centrifuge at 12000 r / min for 10 minutes, filter to obtain the sample solution to be tested.

[0094] The extraction efficiency of different dosage forms of samples was verified by high performance liquid chromatography, the results showed that the extraction efficiency of cream samples was 94.2%, the extraction efficiency of serum samples was 96.8%, the extraction efficiency of emulsion samples was 95.5%, and the extraction efficiency of solid samples was 93.7%, all of which met the detection requirements, and the peak shape of the Raman spectrum of the extracted sample was consistent with that of the standard sample.

[0095] 2.4 Mixing and adsorption condition optimization, take 100 μL of the sample solution to be tested and mix with 300 μL of AgNPs sol according to the volume ratio of 1:3, place it on a vortex oscillator, set the oscillation speed to 3000 r / min, oscillate for 1 minute, then stand for 1.5 minutes, so that arbutin and AgNPs can be fully adsorbed.

[0096] The effects of different mixing ratios, oscillation times, and settling times on the characteristic peak intensities were detected by Raman spectroscopy. The results showed that at a volume ratio of 1:3, an oscillation rate of 3000 r / min, an oscillation time of 1 minute, and a settling time of 1.5 minutes, the intensity of arbutin at 641 cm⁻¹ (isomeric sugar ring shape vibration mode) and 866 cm⁻¹ was significantly higher. -1 The characteristic peak intensity (aromatic ring breathing mode and glycoside resonance) reached its maximum value, with an adsorption efficiency of 92.5%. Figure 4 and Figure 5 As shown.

[0097] 2.5 Optimization of Raman spectroscopy acquisition parameters: 10 μL of the adsorbed mixture was drop-coated onto the center of a pretreated quartz glass slide and allowed to air dry for 15 minutes to form a uniform sample film with a thickness of 0.1 mm. The laser confocal Raman spectrometer was calibrated with a silicon wafer of 520.7 cm⁻¹. -1 The peak was used as the reference peak, and dark current correction was performed simultaneously. The laser source was turned off, and three dark current spectra were acquired. The average value was then subtracted from the sample spectrum. The excitation wavelength was set to 532.8 nm, the laser intensity to 10 mW, and a 100x objective lens was used to observe the sample. The integration time was 30 s, the integration was performed three times, and the detection spectral region was 400–3500 cm⁻¹. -1 The resolution is 1cm. -1 .

[0098] For each sample, 15 different sampling points were randomly selected to collect spectra, with a sampling point spacing of 0.5 mm. Abnormal spectra deviating from the average of all spectra ± 3 standard deviations were removed. The average of the remaining effective spectra was taken as the final detection spectrum. By comparing the effects of different laser intensities, objective magnifications, and integration times on the spectral signal, the above parameters were determined to be the optimal acquisition conditions. Under these conditions, the characteristic peak signal-to-noise ratio was the highest, reaching 35:1, compared with the attached... Figure 5 The Raman spectra of arbutin solutions at different concentrations and the schematic diagram showing the signal enhancement before and after enhancement are consistent, and the deviations of all acquired characteristic peak positions from the measured peak positions are ≤ ±2 cm. -1 The peak intensity change trend is positively correlated with the concentration.

[0099] 2.6 Data processing parameter optimization: The original spectrum was preprocessed using a combination of multivariate scattering correction and standard orthogonal transformation. First, multivariate scattering correction was used to eliminate the spectral baseline drift caused by sample particle scattering. Then, standard orthogonal transformation was used to correct the signal fluctuations caused by uneven sample concentration and optical path difference. The signal-to-noise ratio of the characteristic peaks of the spectrum was improved by 28% after preprocessing.

[0100] Feature wavelength selection employs a genetic algorithm combined with a joint interval partial least squares method, targeting wavelengths from 400 to 3500 cm⁻¹. -1 Spectral regions are defined by 100cm -1The interval is divided into 31 continuous intervals, the genetic algorithm parameters are set to population size 50, crossover probability 0.8, mutation probability 0.05, and iteration number 100, the root mean square error of cross validation is used as the fitness function, and 5 target intervals containing the main characteristic peaks of arbutin are screened out, which are 400-500 cm -1 (contains 427.42 cm -1 , 491.51 cm -1 ), 600-700 cm -1 (contains 655.2 cm -1 ), 800-900 cm -1 (contains 807.44 cm -1 , 828.68 cm -1 , 843.51 cm -1 , 868.88 cm -1 ), 1000-1200 cm -1 (contains 1044.48 cm -1 , 1061.02 cm -1 , 1087.83 cm -1 , 1120.72 cm -1 , 1161.66 cm -1 ), 1200-1700 cm -1 (contains 1240.94 cm -1 , 1269.22 cm -1 , 1293.39 cm -1 , 1323.5 cm -1 , 1345.52 cm -1 , 1363.49 cm -1 , 1415.2 cm -1 , 1456.74 cm -1 , 1609.29 cm -1 , 1615.1 cm -1 ), and then the target intervals are optimized by joint interval partial least squares method, and finally 28 characteristic wavelength points are obtained, including 868.88 cm -1 , 1269.22 cm -1 (multiple substituted aromatic ring C-C vibration, peak intensity 7196.91), 3074.86 cm -1 (multiple substituted benzene ring C-H enhanced stretching vibration, peak intensity 19508.37), and the characteristic peak corresponding wavelength as Figure 3 , and the characteristic peak corresponding wavelength as Figure 5 are used for subsequent model construction.

[0101] 2.7 The qualitative and quantitative model is constructed and optimized, the qualitative identification model is combined with artificial neural network, K-nearest neighbor method and Mahalanobis distance method, the artificial neural network is set to three-layer structure, the number of input layer nodes is 28, the first layer hidden layer has 15 nodes, the second layer hidden layer has 8 nodes, the output layer has one node, the activation function is Sigmoid function, the training algorithm is gradient descent method, the learning rate is 0.01, the momentum factor is 0.9, the iteration number is 1000 times, the sample ratio of training set and test set is 7:3, the training set contains 210 samples, of which 105 are positive samples and 105 are negative samples, the test set contains 90 samples, of which 45 are positive samples and 45 are negative samples, the test set accuracy rate is 98.5% after the model training is completed. The K value of K-nearest neighbor method is 5, the distance measurement adopts Euclidean distance, and the test set accuracy rate is 97.8%.

[0102] In order to determine the scientific and reasonable qualitative determination threshold, an additional 300 group verification sample library is constructed: 150 groups of positive samples cover five gradients of 10 -9 mol / L (detection limit concentration), 10 -8 mol / L, 10 -7 mol / L, 10 -6 mol / L, 10 -5 mol / L, each group has 30 parallel samples; 150 groups of negative samples include blank cosmetic matrix (7 kinds of dosage forms such as cream, essence, etc.) without arbutin and simulated samples containing common interfering components such as vitamin C, nicotinamide, hydroquinone, etc., each group has 30 parallel samples. All the verification samples are predicted by the model, the weighted superposition score is calculated, the ROC curve is drawn, and the best threshold is determined as 0.6 with the goal of "the lowest false positive rate and the lowest false negative rate".

[0103] The threshold of Mahalanobis distance method is set to 3.0, the test set accuracy rate is 96.2%, the weighted superposition score is calculated according to the weight of 40%, 35% and 25% of the three, and the test set accuracy rate of the final qualitative model reaches 99.2%. The combination of quantitative prediction model is partial least squares-support vector machine as the main model and joint interval partial least squares method as the verification model, the partial least squares-support vector machine extracts 5 principal components as input after PLS dimension reduction, the kernel function is radial basis function, the parameter γ is 0.1, the penalty parameter C is 10, the prediction root mean square error of the model verification set is 0.012, and the average relative error is 1.3%; the prediction root mean square error of the verification set of joint interval partial least squares method is 0.018, and the average relative error is 2.1%, the weighted calculation is carried out according to the weight of 80% and 20% of the two, the prediction root mean square error of the final quantitative model is 0.013, the average relative error is 1.5%, and the linear correlation coefficient R 2 = 0.95749, and the attached Figure 6The linear relationship between the characteristic peak intensity shown in the concentration linear response graph B of the middle Raman spectrum and the concentration is consistent.

[0104] III. Data processing

[0105] Accurately weigh 10 mg of alpha-matrix standard, place it in a 10 mL volumetric flask, add methanol to dissolve and dilute to the mark, prepare a 1 mg / mL matrix standard stock solution, seal and store in a 4℃ refrigerator, with a shelf life of 7 days; take an appropriate amount of standard stock solution, dilute with methanol, prepare a 10 -9 , 10 -8 , 10 -7 , 10 -6 , 10 -5 mol / L matrix standard series solution, and use it immediately.

[0106] Take each concentration standard series solution and process it under the optimized sample processing and mixed adsorption conditions, then collect the Raman spectrum, prepare 5 samples in parallel for each concentration, collect 15 spectra for each sample, extract the characteristic peak intensity of 641 cm -1 , 866 cm -1 , 1269.22 cm -1 , 3074.86 cm -1 , etc. after pretreatment, and take the average value as the characteristic peak intensity corresponding to the concentration.

[0107] Among them, the 868.88 cm -1 peak is the core quantitative peak, the logarithm of the concentration of matrix is taken as the abscissa, and the 868.88 cm -1 characteristic peak intensity is taken as the ordinate, a standard curve is drawn, and a linear regression equation y = -3986.06 + 5900833.95x is obtained, wherein y is the 866 cm -1 characteristic peak intensity, x is the mass fraction of matrix, the correlation coefficient R 2 = 0.95749, which is completely consistent with the fitting curve and the correlation coefficient shown in the concentration linear response graph B of the middle Raman spectrum. Figure 6

[0108] Through residual analysis of the standard curve, the residuals are subject to normal distribution and have no obvious trend, indicating that the linear relationship is good and there is no systematic error; according to the regulations of the International Union of Pure and Applied Chemistry, the detection limit is calculated by 3 times the signal-to-noise ratio, and the detection limit of the matrix by the method is 8.7 x 10 -10 mol / L, which is lower than 10 -9 mol / L, meeting the needs of trace detection.

[0109] The Raman characteristic fingerprint peak attribution of the standard sample is shown in Table 1:

[0110] ​Table 1 Raman peak assignment analysis of alpha-magnolol

[0111]

[0112]

[0113]

[0114] Four, sample test experiment

[0115] 4.1 Actual sample preparation, a total of 20 commercially available whitening cosmetic samples were selected, including 5 creams, 5 serums, 5 lotions, and 5 solid cosmetics, numbered S1-S20, wherein S1-S10 were samples that claimed to add arbutin, and S11-S20 were samples that did not claim to add arbutin, used to verify the detection effect of the method of the present application.

[0116] 4.2 Sample detection, 20 actual samples were extracted and purified according to the optimized processing procedures corresponding to the dosage form to obtain sample solutions to be detected, and then detected according to the steps of mixed adsorption, spectrum collection, and data processing, each sample was detected in parallel for 5 times, and the average value was taken as the final detection result.

[0117] At the same time, the same samples were detected by high performance liquid chromatography as a control method, and the detection conditions of HPLC method were as follows: C18 column 4.6 mm x 250 mm, 5 μm, mobile phase methanol-water = 40:60, flow rate 1.0 mL / min, detection wavelength 280 nm, column temperature 30°C, sample size 20 μL.

[0118] 4.3 Detection result analysis, the detection results of S1-S10 samples that claimed to add arbutin showed that S1-S10 all detected arbutin, the content range was 0.12%-6.85%, which was within the limit range specified in the "Cosmetic Safety Technical Specification", the relative standard deviation of parallel detection was 1.2%-2.8%, and the average relative standard deviation was 2.1%.

[0119] The detection results of HPLC method were 0.11%-6.92%, and the relative standard deviation was 1.5%-3.2%, the relative error of the detection results of the present application and HPLC method was -3.8%-4.2%, all within ±5%, indicating that the detection results of the two methods had good consistency.

[0120] For S11-S20 samples that did not claim to add arbutin, the present application method and HPLC method did not detect arbutin, the qualitative results were consistent, and there were no false positive or false negative results, the detection spectrum was consistent with the spectrum characteristics of the blank sample. Figure 4

[0121] Five, stability and repeatability experiment​

[0122] 5.1 Long-term stability test: The prepared AgNPs sol was sealed and protected from light, and stored in a refrigerator at 4℃. Performance tests were conducted at 1, 3, 7, 15, and 30 days after preparation. Test indicators included the position and intensity of the maximum absorption peak in the UV absorption spectrum, the particle size distribution under scanning electron microscopy, and the stability against 10... -7 Raman signal enhancement effect of mol / L arbutin.

[0123] The results showed that the maximum absorption peak was at 420 nm with an intensity of 0.62 after 1 day; at 421 nm with an intensity of 0.59 after 3 days; at 421 nm with an intensity of 0.57 after 7 days; at 422 nm with an intensity of 0.55 after 15 days; and at 422 nm with an intensity of 0.56 after 30 days, with an intensity decrease of 8.7%. The UV absorption spectra at all time points exhibited typical AgNPs plasmon resonance peak shapes, with flat baselines and no extraneous peak interference. Figure 2 The basic UV spectral characteristics of AgNPs after successful preparation (symmetrical peak shape, with the maximum absorption peak concentrated around 420 nm) are consistent, indicating that AgNPs did not undergo significant aggregation or oxidation during storage and their basic properties remained stable.

[0124] Scanning electron microscopy revealed that after 30 days, AgNPs maintained their spherical morphology, without obvious adhesion or irregular aggregation, and remained attached to the surrounding tissues. Figure 3 The basic microstructure characteristics of the prepared AgNPs were consistent; the particle size distribution range changed from 50–80 nm to 55–85 nm, and the average particle size increased from 65 nm to 71 nm, an increase of 6 nm. This particle size change is within the reasonable range for long-term storage of AgNPs, and the Raman enhancement performance was not affected by the abnormal particle size growth; for 10 -7 Raman signal intensity detection of 1 mol / L arbutin showed that the core characteristic peak in Table 1 was 868.88 cm⁻¹ after 1 day. -1 The intensity was 12651.35 after 30 days and 11350.72 after 30 days, a decrease of 10.2%. Other characteristic peaks, such as 655.2 cm⁻¹, were also observed. -1 1269.22cm -1 The intensity decrease was ≤12%, still meeting the detection requirements (signal intensity ≥85% of the initial value), indicating that the AgNPs sol was stable within 30 days under sealed and light-proof conditions at 4℃, and the Raman enhancement effect on arbutin remained consistent.

[0125] 5.2 Repeatability Experiments, Multiple Batch Repeatability Experiments: Three batches of AgNPs sol were prepared, numbered B1, B2, and B3, and tested against 10... -7 The arbutin standard solution was used for detection, with 15 parallel tests per batch. The value of 868.88 cm⁻¹ in Table 1 was calculated. -1, 655.2 cm -1 , 1269.22 cm -1 The relative standard deviations of the three core peaks.

[0126] The results show that the relative standard deviations of the three peaks of B1 batch are 2.7%, 2.9%, 2.6%, and the average is 2.8%; the relative standard deviations of B2 batch are 3.0%, 3.2%, 2.9%, and the average is 3.1%; the relative standard deviations of B3 batch are 2.8%, 3.0%, 2.7%, and the average is 2.9%; the average relative standard deviation of the three batches is 2.9%≤3%, which indicates that the enhancement effect of AgNPs sol of different batches on each characteristic peak of α-arbutin is good, the detection data is reliable, and the micro-morphology of each batch of sol is consistent with the characteristics shown in the table, and there is no fluctuation of enhancement performance caused by the difference of morphology between batches. Figure 3

[0127] Single batch sol repeatability experiment: take the same batch of AgNPs sol, and detect 0.2g / L arbutin solution for 10 times, the Raman spectrum is shown in the table Figure 7 , and the table Figure 7 directly shows the spectral superposition effect of 10 repeated detections, it can be seen that the peak position of 868.88cm -1 , 655.2 cm -1 and other core characteristic peaks is always stable (deviation≤±1cm -1 ), only the peak intensity has small fluctuation; this difference is mainly due to the micro-unevenness of AgNPs sol as a nano material (which is the inherent characteristic of nano substrate), but the peak stability is good, and the fluctuation can be offset by taking the average value of multiple sampling when quantitatively calculating, which does not affect the accuracy of quantitative detection, and the relative standard deviations of other characteristic peaks are all≤15%, which meets the conventional error range of Raman spectrum detection.

[0128] Multi-laboratory repeatability experiment: select 3 different laboratories, numbered as L1, L2, L3, all use laser confocal Raman spectrometer with model LabRAM HR Evolution, and detect the same group of cosmetic samples (containing 10 -7 mol / L arbutin) according to the same experimental scheme, and each laboratory detects 10 times, and the content is calculated by the peak intensity of 868.88cm -1 in table 1.

[0129] The results show that the average value of L1 detection is 0.98×10 -7 mol / L, and the relative standard deviation is 3.2%; the average value of L2 detection is 1.02×10 -7 mol / L, and the relative standard deviation is 3.5%; the average value of L3 detection is 1.01×10 -7 ​The relative standard deviation was 3.8%, and the relative deviation of the detection results of the three laboratories was less than or equal to 4.1%, indicating that the technology has good repeatability between different laboratories.

[0130] Multi-operator repeatability experiment: 5 operators with different experience levels, numbered O1 to O5, wherein O1 and O2 are senior experimenters, and O3 to O5 are junior experimenters. The same batch of samples is detected according to the experimental scheme, and each person detects 8 times. The relative standard deviation of the detection results of all operators is less than or equal to 3.8%, and the detection data of the senior experimenters and the junior experimenters has little difference. The Raman spectrum collected by all operators makes the operation simple, and the experience requirement of the operator is low, which is not easy to cause misjudgment due to operation difference, and the repeatability meets the actual detection requirement. -1 The peak is the core quantitative peak.

[0131] The results show that the relative standard deviation of O1 is 2.1%, the relative standard deviation of O2 is 2.3%, the relative standard deviation of O3 is 3.3%, the relative standard deviation of O4 is 3.5%, and the relative standard deviation of O5 is 3.7%. The detection data of the senior experimenters and the junior experimenters has little difference. The Raman spectrum collected by all operators makes the operation simple, and the experience requirement of the operator is low, which is not easy to cause misjudgment due to operation difference, and the repeatability meets the actual detection requirement.

[0132] Six, anti-interference experiment

[0133] Common interference components in cosmetics are selected, including vitamin C, nicotinamide, hyaluronic acid, glycerol, ceramide, and structural analogues hydroquinone and phenol. Different concentrations of interference component solutions are prepared, mixed with 10 -7 mol / L arbutin standard solution to prepare mixed solutions containing interference components. The concentration of the interference component is set to 0.1 mg / mL, 0.5 mg / mL, and 1 mg / mL, and 5 samples are prepared at each level. The detection is carried out according to the method, and the influence of the interference component on the detection result of arbutin is analyzed.

[0134] The results show that when the concentration of vitamin C is less than or equal to 0.5 mg / mL, the intensity change rate of the characteristic peak of arbutin is less than or equal to 8%, and the quantitative error is less than or equal to ±3%. When the concentration reaches 1 mg / mL, the intensity change rate is 12%, and the quantitative error is ±4.8%, which is still within the acceptable range. Under different concentrations of nicotinamide, the intensity change rate of the characteristic peak is less than or equal to 5%, and the quantitative error is less than or equal to ±2.5%, without obvious interference.

[0135] Hyaluronic acid and glycerol have large molecular weight and low Raman activity, which has no significant effect on the characteristic peaks, with an intensity change rate of ≤3% and a quantitative error of ≤±1.8%. Ceramide has no interference when its concentration is ≤0.05 mg / mL, and the intensity change rate is 7% when its concentration is 0.1 mg / mL, with a quantitative error of ±3.2%. When the concentrations of structural analogues hydroquinone and phenol are ≤0.01 mg / mL, the quantitative error is ≤±4.5%; when the concentrations reach 0.05 mg / mL, the quantitative errors are ±6.2% and ±5.8% respectively. However, such components are prohibited from being added or added in a very small amount in cosmetics, and will not significantly interfere with the detection results in actual samples. The experimental results show that the sensing technology has good anti-interference ability to common interfering components, can meet the specific detection needs of arbutin in complex matrix, and the characteristic peaks of arbutin in the detection spectrum are not covered by the peaks of interfering components, which is consistent with the Figure 4 Raman spectrum characteristics of arbutin standard sample in the complex matrix.

[0136] Seven, recovery rate experiment

[0137] The accuracy of the quantitative model was verified by the standard addition recovery method. Three blank cosmetic samples were selected, namely cream, essence and emulsion, numbered as B1, B2 and B3, and arbutin standard was added at low, medium and high concentration levels. The low concentration was 10 -9 mg / mL, the medium concentration was 10 -7 mg / mL, and the high concentration was 10 -5 mg / mL. Five samples were prepared at each concentration level, and the detection was carried out according to the established detection process. The recovery rate and relative standard deviation were calculated.

[0138] The results showed that the recovery rate of the low concentration level was 92.3% to 95.6%, with an average recovery rate of 93.8% and a relative standard deviation of 2.5%; the recovery rate of the medium concentration level was 94.8% to 97.2%, with an average recovery rate of 96.1% and a relative standard deviation of 1.8%; the recovery rate of the high concentration level was 96.1% to 98.5%, with an average recovery rate of 97.3% and a relative standard deviation of 1.2%. The overall average recovery rate was 95.8%, with a relative standard deviation of ≤2.5%, which met the requirements of the detection method recovery rate of 90% to 110% in the "Cosmetic Safety Technical Specification", indicating that the quantitative accuracy of the technology was reliable, and the linear relationship between the intensity of the characteristic peak after adding the standard and the concentration still conformed to the Figure 6 regularity shown in the attached

[0139] Eight, comparative experiment

[0140] High performance liquid chromatography, ultraviolet-visible spectrophotometry and gas chromatography-mass spectrometry were selected as comparative objects, and five key indicators including detection limit, detection time, operation cost, sample pretreatment and instrument requirement were quantitatively compared.

[0141] The detection limit of the HPLC method is 10 -7 mol / L, the detection limit of the UV-Vis method is 10 -6 mol / L, the detection limit of the GC-MS method is 5*10 -7 mol / L, and the detection limit of the method is 8.7*10 -10 mol / L, which is 100 times higher than that of the HPLC method, 1000 times higher than that of the UV-Vis method, and 500 times higher than that of the GC-MS method.

[0142] In terms of detection time, the sample pretreatment of the HPLC method takes 30 minutes, the instrument analysis takes 30 minutes, and the total time is 60 minutes; the pretreatment of the UV-Vis method takes 20 minutes, the analysis takes 10 minutes, and the total time is 30 minutes; the pretreatment of the GC-MS method takes 40 minutes, the analysis takes 40 minutes, and the total time is 80 minutes;

[0143] The pretreatment of the method takes 10 minutes, the analysis takes 5 minutes, and the total time is 15 minutes, which is 2-5 times faster than the traditional methods.

[0144] In terms of operation cost, the cost of each sample detection of the HPLC method is about 500 yuan including column loss, mobile phase, reagent, etc.; the cost of the UV-Vis method is about 200 yuan / sample; the cost of the GC-MS method is about 800 yuan / sample; and the cost of the method is about 80 yuan / sample, which is only 16% of the HPLC method and 10% of the GC-MS method.

[0145] In terms of sample pretreatment, the traditional methods all need complex pretreatment such as column chromatography purification and derivatization reaction, and the method only needs simple extraction, centrifugation and filtration without complex purification steps.

[0146] In terms of instrument requirements, the HPLC and GC-MS instruments are expensive, about 50-100 million yuan per unit, and require professional laboratory environment and operators; the portable Raman spectrometer can be selected for the method, which is about 20-30 million yuan per unit, and is simple to operate without professional laboratory, which is suitable for on-site detection.

[0147] The comparison results show that the method is significantly better than the traditional methods in sensitivity, detection speed, operation convenience and cost, and the detection results are consistent with the HPLC method, which meets the requirements of the appendix. Figure 5 The high sensitivity advantage before and after the signal enhancement is shown in the comparison results.

[0148] Nine, actual verification experiment

[0149] 9.1 Supervision department flight inspection application, adapt to portable laser confocal Raman spectrometer, model HoribaLabRAMPortable, the instrument weight is 3kg, the size is 30cm*20cm*15cm, is equipped with rechargeable lithium battery endurance 8 hours, supports on-site rapid detection.

[0150] The testing process is simplified to taking a 0.1g cosmetic sample, adding 1mL of methanol, vortexing for 1 minute, centrifuging for 5 minutes using a portable centrifuge (weighing 1.5kg), taking the supernatant and mixing it with AgNPs sol at a ratio of 1:3, letting it stand for 1 minute, and then drop-coating for testing. The entire process takes only 10 minutes, meeting the speed requirements of on-site sampling by regulatory authorities.

[0151] The instrument incorporates the qualitative and quantitative model of this invention. After detection, it directly outputs the result, including whether arbutin is present and its content value, without requiring professional data analysis. Five different cosmetic sales stores were selected, and 10 whitening cosmetic samples were randomly drawn from each store, totaling 50 samples. These samples were tested on-site using a portable Raman spectrometer. Simultaneously, the samples were brought back to the laboratory for verification using HPLC. The results showed a 100% qualitative concordance rate between on-site and laboratory testing, and a relative error of ≤±4.5% in the quantitative results. This indicates that the method is suitable for unannounced inspections by regulatory authorities, and the test results are consistent with the attached... Figure 4 Appendix Figure 5 The characteristic peak identification patterns shown are consistent.

[0152] 9.2 Application in Enterprise Production Quality Control: A testing point was set up at the finished product stage of a whitening face cream production line in a cosmetics manufacturing company. Ten samples were taken from each batch and tested according to the method of this invention. The test results were fed back to the production department in real time. Ten production batches, totaling 100 samples, were selected. The test results showed that the arbutin content of 98 samples was within the set internal control range of 0.5% to 6.5%, while the content of 2 samples exceeded the range, at 6.8% and 0.4% respectively. The production department promptly suspended production and found that the error was due to a raw material ratio error. After adjustment, the test results of the produced samples all met the requirements.

[0153] The technology's speed allows for testing of each batch of samples within 30 minutes, without affecting production efficiency; its accuracy ensures the reliability of product quality control, preventing substandard products from entering the market; and the stability and reliability of the test data are also guaranteed. Figure 6 The linear response characteristics shown are consistent.

[0154] 9.3 Compatibility verification of different dosage forms: Seven common whitening cosmetics, including face cream, serum, lotion, toner, facial mask liquid, pressed powder, and loose powder, were selected. Ten samples of each dosage form were selected, for a total of 70 samples. The samples were tested according to the method of this invention, and HPLC was used as a control.

[0155] The results show that the detection qualitative coincidence rates of the 7 dosage form samples are all 100%, the relative errors of the quantitative results are less than or equal to ±4.8%, and the relative standard deviations are less than or equal to 3.2%, indicating that the method is suitable for all whitening cosmetic dosage forms containing arbutin, without additional adjustment of core detection parameters, with wide adaptability, and the detection spectra of different dosage form samples can clearly identify the characteristic peaks of 641cm -1 , 866cm -1 , etc., which are consistent with the spectral characteristics shown in Figs. 2 and 3 Figure 3 , Figs. 4 and 5 Figure 4

[0156] The above merely describes the preferred specific embodiments of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can make equivalent replacements or changes to the technical solutions and inventive concepts of the present application within the technical range disclosed by the present application, which should be covered within the protection scope of the present application.​

Claims

1. A method for rapid identification of arbutin in cosmetics based on surface-enhanced Raman spectroscopy, characterized by, The method comprises the following steps: a) preparing AgNPs sol as a SERS enhancement substrate by using ascorbic acid redox method, wherein the particle size of the AgNPs sol is 50-80 nm, and the Raman signal enhancement performance is confirmed by ultraviolet absorption spectrum and scanning electron microscope; b) modifying a polymer film on the surface of the AgNPs by electro-polymerization to construct a SERS sensing interface, wherein aspartic acid is used as a monomer; c) for different formulations of cosmetics, methanol is used as an extraction reagent for targeted extraction, and after centrifugation, filtration and purification, a sample solution to be detected is obtained; d) the sample solution to be detected is mixed with the AgNPs sol at a volume ratio of 1:2-1:4, and after standing for 0.5-3 minutes, arbutin and AgNPs are fully adsorbed, and then the mixture is dropped on a pretreated quartz glass slide and naturally air-dried; e) Raman spectra were collected using a laser confocal Raman spectrometer, with a silicon chip 520.7 cm -1 peak as the reference peak to calibrate the instrument, with an excitation light source wavelength of 532 nm ± 5 nm, and a detection spectral region of 400-3500 cm -1 ; f) the original spectrum is pretreated by combining multiple scattering correction (MSC) and standard normal variate (SNV), characteristic wavelengths are screened by genetic algorithm-interval partial least squares (GA-SiPLS), and specific identification and content detection of arbutin are realized by combining a qualitative identification model and a quantitative prediction model.

2. The method for rapid identification of arbutin in cosmetics based on surface-enhanced Raman spectroscopy according to claim 1, characterized in that: In the step a), the preparation raw materials of the AgNPs sol have the following specifications: polyvinylpyrrolidone (PVP) is K30 type, the molecular weight is 40000±2000, and the purity is greater than or equal to 99.0%; the concentration of silver nitrate is 0.08-0.12 mol / L; the purity of ascorbic acid is greater than or equal to 99.9%; the concentration of sodium hydroxide is 0.4-0.6 mol / L; and the concentration of sodium chloride is 4-6 mol / L. In the step a), the preparation conditions of the AgNPs sol are as follows: the reaction temperature is controlled at 20-30℃, the magnetic stirring rate is 400-600 r / min during the mixing of PVP and silver nitrate, the rapid stirring rate is 1000-1400 r / min when sodium chloride is added, the stirring rate is maintained at 400-600 r / min after the mixed solution of ascorbic acid and sodium hydroxide is added, and the reaction is carried out in a dark environment for 1.5-2.5 hours.

3. The method according to claim 2, wherein the method is characterized in that: In the step b), the specific operation process of electro-polymerization is as follows: the AgNPs sol is uniformly dropped on the surface of a glassy carbon sheet and dried by an infrared lamp to serve as a working electrode; the working electrode is inserted into a 2 mmol / L aspartic acid-containing phosphate buffer solution, the concentration of the buffer solution is 0.005-0.02 mol / L, the pH value is 6.5-7.5; a platinum wire auxiliary electrode and a saturated calomel reference electrode are connected, and the aspartic acid is polymerized on the surface of the AgNPs to form a dense polymer film by cyclic voltammetry scanning 8-12 times at a potential window of-1.0-2.0 V and a scanning rate of 80-120 mV / s.

4. The method according to claim 1, wherein the method is characterized by: ​ 5. The method according to claim 1, wherein the method is characterized by: The specific operation of the sample extraction and purification in step c) is as follows: 0.5 g of the cosmetic sample is placed in a 10 mL centrifuge tube, 5-8 mL of methanol is added according to the dosage form; the cream and solid cosmetic are vortexed for 2 minutes and ultrasonically extracted for 18-25 minutes, the essence and emulsion are vortexed for 1 minute and ultrasonically extracted for 12-18 minutes; after extraction, centrifugation is performed at a speed of 9000-12000 r / min for 8-12 minutes, the supernatant is filtered through a 0.45 μm organic phase nylon filter membrane, and the filtrate is collected as the sample solution to be tested.

6. The method according to claim 1, wherein the method is characterized by: The optimal mixing volume ratio of the sample solution to be tested and the AgNPs sol in step d) is 1:3, the vortex oscillation rate after mixing is 2500-3500 r / min, the oscillation time is 0.5-1 minute, the standing adsorption time is 1-2 minutes, and it is detected that the adsorption efficiency of arbutin and AgNPs under this condition is ≥90%, and the characteristic Raman peak intensity reaches the maximum value.

7. The method according to claim 1, wherein the method is characterized by: The specific parameters of Raman spectrum collection in step e) are as follows: the laser intensity is set to 8-12 mW, an 80-100 times objective lens is used to observe the sample, the integration time is 20-40 s, and the integration number is 2-4 times; 10-20 different sampling points of each sample are randomly selected for spectrum collection, the abnormal spectrum deviating from the average value ± 3 times the standard deviation of all spectra is removed, and the average value of the remaining effective spectra is taken as the final detection spectrum.

8. The method according to claim 1, wherein the method is characterized by: The core algorithm of the multiple scattering correction (MSC) in step f) is: taking the average spectrum of all sample original spectra as a reference spectrum The original spectrum vector x of each sample is subjected to a linear regression with the reference spectrum The model is: , Wherein, a is the regression coefficient (slope), b is the intercept, the values of a and b are solved by the least squares method, the original spectrum is corrected by using the obtained parameters, and the corrected spectrum is: ; Wherein, x is the original spectrum intensity vector; the formula of standard orthogonal transformation (SNV) is: , wherein μ is the intensity mean of a single original spectrum, and σ is the intensity standard deviation of a single original spectrum.

9. The method according to claim 8, wherein the method is characterized by: The specific process of the genetic algorithm-interval partial least squares (GA-SiPLS) in step f) is as follows: first, 400-3500 cm -1 is divided into 25-35 continuous intervals at intervals of 100 cm -1 ; the genetic algorithm parameters are set as population size 40-60, crossover probability 0.7-0.9, mutation probability 0.03-0.07, and iteration number 80-120; the root mean square error of cross validation (RMSECV) is used as the fitness function to screen the target interval containing the characteristic peak of arbutin; then the target interval is optimized by the interval partial least squares to obtain 20-30 characteristic wavelength points for model construction.

10. The method according to claim 9, wherein the method is characterized by: The qualitative identification model combination in step f) is the weighted combination of artificial neural network (ANN), K-nearest neighbor method (KNN) and Mahalanobis distance method (MD); wherein the weight proportion of ANN is 40%, the weight proportion of KNN is 35%, and the weight proportion of MD is 25%, the output results (0 for no arbutin, 1 for arbutin) are calculated by weight superposition to obtain a score, a score ≥0.6 is determined as containing arbutin, and a score <0.6 is determined as not containing arbutin; ANN is a three-layer structure, the number of input layer nodes corresponds to the number of characteristic wavelength points, two layers are set in the hidden layer (the number of nodes is 12-18 and 6-10 respectively), there is one node in the output layer, and the activation function is Sigmoid function ; The K value of KNN is 3-7, and the Euclidean distance is used for distance measurement: ; (m is the number of characteristic wavelengths); the Mahalanobis distance formula is: ; (μ is the mean value vector of the arbutin standard sample, and Σ is the covariance matrix), and the threshold is set to 2.5-3.5; the quantitative prediction model combination is partial least squares-support vector machine (PLS-SVM) as the main model and joint interval partial least squares (SiPLS) as the verification model; the weight proportion of PLS-SVM is 80%, and the weight proportion of SiPLS is 20%, and the final quantitative result is calculated according to "main model result × 0.8 + verification model result × 0.2"; PLS-SVM extracts 4-6 principal components as input through PLS dimension reduction, and the kernel function is radial basis function: ,( =0.05~0.2, penalty parameter C=8~12); based on the characteristic wavelength point to construct the standard curve, with the concentration of arbutin as the horizontal coordinate and the characteristic peak intensity as the vertical coordinate, the linear correlation coefficient R 2 ≥0.95, the relative standard deviation (RSD) of the quantitative results is ≤3%, and the recovery rate is 92%~98%.

Images