A method for simultaneously detecting multiple quinolone antibiotics in milk based on surface-enhanced Raman spectroscopy
By combining surface-enhanced Raman spectroscopy with a magnetic Fe3O4@COF@Ag substrate and a machine learning algorithm, the problem of simultaneous detection of multiple quinolone antibiotics in existing technologies has been solved, achieving rapid and sensitive detection results.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-05-11
- Publication Date
- 2026-03-20
AI Technical Summary
Existing detection methods are difficult to distinguish and detect multiple quinolone antibiotics simultaneously, and the procedures are complex and time-consuming, which cannot meet the need for efficient detection of quinolone antibiotics in dairy products.
Surface-enhanced Raman spectroscopy combined with a magnetic Fe3O4@COF@Ag substrate and machine learning algorithms was used to enrich quinolone antibiotics on the magnetic substrate. A linear model was constructed using Raman spectral characteristic peaks, and the model was then classified and identified using machine learning algorithms.
It enables rapid and sensitive detection of multiple quinolone antibiotics in milk, with detection limits as low as 5.61×10⁻⁹M for ciprofloxacin, 1.44×10⁻⁸M for levofloxacin, and 1.56×10⁻⁸M for norfloxacin, thus improving the selectivity and efficiency of detection.
Smart Images
Figure CN116678864B_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to a method for simultaneously detecting multiple quinolone antibiotics in milk based on surface-enhanced Raman spectroscopy, belonging to the field of analysis and detection. BACKGROUND
[0002] Quinolones are a widely used antibiotic that can effectively treat bacterial infections in dairy cows. As people's demand for dairy products has grown, the misuse of antibiotics in the dairy industry has led to poor quality milk and the emergence of drug-resistant bacterial strains, which has become so serious. Many countries have changed the maximum allowable content of antibiotics in dairy products to prevent the situation from worsening. Ciprofloxacin is commonly used to treat gram-negative bacilli and gram-positive bacilli, while norfloxacin and levofloxacin are commonly used to treat bronchitis, cowpox virus infection and pneumonia. However, the use of excessive quinolone antibiotics to treat dairy cows through intravenous injection or feed additives can lead to antibiotic residues in milk, which can have adverse effects on the liver and kidneys of the human body, so the detection of quinolone antibiotics in dairy products is particularly important.
[0003] Currently, the commonly used detection methods for quinolone drugs are liquid chromatography mass spectrometry and gas chromatography mass spectrometry. Both methods can effectively detect quinolone antibiotics in dairy products. In addition, immunological methods can also be used for detection. This method uses specific antibodies to bind with quinolones to form antigen-antibody complexes. However, due to the high similarity of the molecular structure of quinolone antibiotic molecules, it is more difficult to distinguish at low concentrations, and currently these methods cannot simultaneously distinguish and detect different types of quinolone antibiotics. SUMMARY
[0004] [TECHNICAL PROBLEM]
[0005] Currently, the main methods for detecting quinolone antibiotics are high-performance liquid chromatography and electrophoresis, etc. These detection methods are mostly complex and time-consuming, requiring sample pretreatment, specialized equipment and manual labor, and cannot simultaneously distinguish and detect multiple quinolone antibiotics. The method of the present application solves the problem of selectivity of SERS technology in detecting quinolone antibiotics, and through the enrichment effect of the magnetic substrate, the sensitivity and detection limit have certain advantages over traditional methods; through machine learning algorithm for accurate classification, greatly improving the selectivity of SERS method in application to quinolone antibiotic detection, realizing simultaneous detection of multiple quinolone antibiotics in milk.
[0006] [TECHNICAL SCHEME]
[0007] A method for simultaneously detecting multiple quinolone antibiotics in milk based on surface-enhanced Raman spectroscopy, the method comprising the following steps:
[0008] (1) ethylene glycol, sodium acetate and ferric chloride hexahydrate are used to prepare magnetic Fe3O4 particles by a single-step solvothermal method; then Fe3O4 and COF precursors are dissolved in acetonitrile, and Fe3O4@COF materials are synthesized at room temperature by using acetic acid as a catalyst; the Fe3O4@COF materials are collected by a magnet; finally, a hydrothermal reaction is carried out between silver nitrate as a silver source, sodium citrate as a reducing agent and the Fe3O4@COF materials to obtain a Fe3O4@COF@Ag substrate;
[0009] (2) a quinolone antibiotic mother liquor is diluted to prepare quinolone antibiotic standard solutions with different concentration gradients, then a mixed solution of ultrapure water and milk is added, and the mixture is uniformly mixed to obtain a quinolone antibiotic-milk mixture; methanol, water and the obtained quinolone antibiotic-milk mixture are mixed, and the supernatant is collected by centrifugation to obtain a sample solution;
[0010] (3) the sample solution obtained in step (2) is mixed with the Fe3O4@COF@Ag substrate obtained in step (1) to obtain a test sample solution, Raman detection is carried out, and a surface-enhanced Raman spectrum of the sample solution is collected; a linear model is constructed by using characteristic peaks at 1390 cm -1 -1398cm -1 in the surface-enhanced Raman spectrum and antibiotic concentrations; and a classification and identification model is constructed by training the surface-enhanced Raman spectrum data through a machine learning algorithm (PCA-k-NN, PCA-SVM, PCA-Decision Tree).
[0011] In an embodiment of the present application, the method further comprises: a to-be-tested milk sample is pretreated by adding methanol and water according to the process of step (2), and then uniformly mixed with the Fe3O4@COF@Ag substrate solution to obtain a to-be-tested sample; a surface-enhanced Raman spectrum is measured; and the concentrations of various quinolone antibiotics in the to-be-tested sample are obtained according to the linear model and the classification and identification model in step (3).
[0012] In an embodiment of the present application, in step (1), the mass ratio of sodium acetate to ferric chloride hexahydrate is (1-3): 1. Specifically, 1.5: 1 can be selected.
[0013] In an embodiment of the present application, in step (1), the amount of ethylene glycol relative to the total mass of sodium acetate and ferric chloride hexahydrate is (10-25) mL / g. Specifically, 12.5 mL / g can be selected.
[0014] In an embodiment of the present application, in step (1), the conditions of the single-step solvothermal method are as follows: heating at 150-200℃ for 10-15 hours.
[0015] In one embodiment of the present invention, in step (1), the mass ratio of Fe3O4, m-phenylenediamine, and 2,5-divinyl terephthalaldehyde is (3-5):1:(2-4). Specifically, 3.5:1:2.6 may be selected.
[0016] In one embodiment of the present invention, in step (1), the concentration of m-phenylenediamine relative to acetonitrile is 0.1-1.0 mg / mL; specifically, 0.4-0.5 mg / mL may be selected.
[0017] In one embodiment of the present invention, in step (1), the amount of acetic acid relative to m-phenylenediamine is 0.5-1.0 mL / mg; specifically, 0.6 mL / mg may be selected.
[0018] In one embodiment of the present invention, in step (1), the synthesis process of Fe3O4@COF is as follows: the reaction is carried out at room temperature for 24 hours, the mixture is collected with a magnet, and then washed with ethanol and acetone to obtain the product Fe3O4@COF material.
[0019] In one embodiment of the present invention, in step (1), the mass ratio of Fe3O4@COF to silver nitrate is (1-5):(2-3.5). Specifically, 2.5:2 is optional.
[0020] In one embodiment of the present invention, in step (1), the concentration of Fe3O4@COF in the hydrothermal reaction is 0.03-0.3 mg / mL; specifically, 0.125 mg / mL may be selected.
[0021] In one embodiment of the present invention, in step (1), the mass ratio of Fe3O4@COF to sodium citrate is (1-5):(3-6). Specifically, 2.5:4 is optional.
[0022] In one embodiment of the present invention, in step (1), the synthesis conditions of Fe3O4@COF@Ag are as follows: incubation at room temperature with stirring for 10 minutes, addition of reducing agent and heating to boiling for 30 minutes to obtain Fe3O4@COF@Ag substrate.
[0023] In one embodiment of the present invention, the quinolone antibiotics in step (2) include any one or more of ciprofloxacin, norfloxacin, and levofloxacin.
[0024] In one embodiment of the present invention, the method for preparing the standard solution of quinolone antibiotics in step (2) is as follows: Norfloxacin, ciprofloxacin and levofloxacin are dissolved in ultrapure water to prepare 30-70 mg / L norfloxacin solution, 30-150 mg / L ciprofloxacin solution and 30-150 mg / L levofloxacin solution; then the three are mixed in equal volumes.
[0025] In an embodiment of the present application, the volume ratio of the quinolone antibiotic standard solution to the mixed solution of ultrapure water and milk in step (2) is 1:(1-3).
[0026] In an embodiment of the present application, the concentration of each antibiotic in the quinolone antibiotic-milk mixed solution in step (2) is 1x10 -5 M-1x10 -8 M.
[0027] In an embodiment of the present application, in step (2), the volume ratio of methanol, water, and quinolone antibiotic-milk solution is 1:(4-6):1.
[0028] In an embodiment of the present application, in step (2), centrifugation is performed at a speed of 12000 rpm for 10 minutes, and the supernatant is taken as the sample solution.
[0029] In an embodiment of the present application, in step (3), the Raman detection conditions are as follows: Raman detection is performed using a Raman spectrometer, the excitation light source wavelength of the Raman spectrometer is 532 nm, the integration time is 20 s, and the laser power is 5%.
[0030] In an embodiment of the present application, the preparation method of the standard curve model in step (3) comprises the following steps: mixing the sample solution obtained in step (2) and the Fe3O4@COF@Ag substrate obtained in step (1) uniformly according to a volume ratio of 1:1 to obtain a test sample solution; then performing Raman detection to obtain a Raman spectrum; and finally constructing a linear model, i.e., a standard curve model, using the peak intensity at 1390 cm -1 -1398 cm -1 -1 of the Raman spectrum and the concentration of quinolone antibiotics.
[0031] In an embodiment of the present application, the SERS spectrum data preprocessing method used by the machine learning algorithm classification model in step (3) is to take the first derivative spectrum of the SERS spectrum and perform normalization preprocessing; and the kernel function used in the PCA-SVM algorithm is an RBF function.
[0032] In an embodiment of the present application, in the construction process of the standard curve, the sample solution and the substrate are mixed by vortex oscillation for 2 minutes.
[0033] In an embodiment of the present application, the standard curve model in step (3) is as follows: ciprofloxacin: I = 9786.72 log 10 C + 78800.69, the correlation coefficient R 2 = 0.991, levofloxacin: I = 10073.87 log 10 C + 77258.73, the correlation coefficient R 2= 0.992, Norfloxacin: I = 10594.44 log 10 C + 82147.29, correlation coefficient R 2 = 0.997, unit: mol / L; I is the Raman intensity of three antibiotics at 1390cm -1 -1398cm -1 characteristic peak intensity of symmetric stretching vibration of O-C-O.
[0034] The application provides application of the above method in the field of food detection.
[0035] [beneficial effects]:
[0036] 1. The application uses the magnetic Fe3O4@COF@Ag Raman enhancement substrate for detection, and the magnetism is conducive to the enrichment of antibiotic molecules. Moreover, the framework material COF can make the nano-silver grow to form a dense nano-gap, provide more strong hot spot areas, and be more conducive to the detection of quinolone antibiotics.
[0037] 2. The application discloses a sensitive label-free SERS method, which combines the Fe3O4@COF@Ag substrate with a machine learning algorithm, and greatly improves the detection selectivity of SERS on similar molecules with high structural similarity.
[0038] 3. The Fe3O4@COF@Ag substrate in the application can be used as a Raman enhancement substrate to detect ciprofloxacin, levofloxacin and norfloxacin in milk, and the detection time is short, and the detection limit is as low as ciprofloxacin: 5.61x10 -9 M, levofloxacin: 1.44x10 -8 M, and norfloxacin: 1.56x10 -8 M. This has important significance for supervising the residues of quinolone antibiotics in food. BRIEF DESCRIPTION OF DRAWINGS
[0039] Figure 1 It is a flowchart for detecting quinolone antibiotics based on surface-enhanced Raman spectroscopy.
[0040] Figure 2 It is a linear model obtained by fitting the relationship between the concentrations of three antibiotics in the system in Example 1 and the Raman intensity at 1390cm -1 -1398cm -1 .
[0041] Figure 3 It is the influence of different proportions of methanol and test solution during pretreatment on the SERS characteristic peak intensity in Example 3.
[0042] Figure 4The classification results and decision boundaries of the original SERS spectral data used in Example 1 are shown. Two-dimensional PCA diagrams were obtained after standardizing the original SERS spectra of the three antibiotics. Red dots represent norfloxacin, purple dots represent levofloxacin, and pink dots represent ciprofloxacin. (a) PC1×PC2; (b) PC1×PC3; (c) PC2×PC3; (d) PCA-k-NN classification decision boundary; (e) PCA-SVM classification; (f) PCA-decision tree classification.
[0043] Figure 5 This is the confusion matrix of the classification results using the original SERS spectral data in Example 1. Confusion matrix diagrams for the classification results of the three antibiotics: (a) PCA-k-NN classification; (b) PCA-SVM classification; (c) PCA-decision tree classification.
[0044] Figure 6 The classification results and decision boundaries of the standardized first-order derivative SERS spectral data used in Example 1 are shown. Two-dimensional PCA diagrams were obtained after standardizing the first-order derivative SERS spectra of the three antibiotics. Red dots represent norfloxacin, purple dots represent levofloxacin, and pink dots represent ciprofloxacin. (a) PC1×PC2; (b) PC1×PC3; (c) PC2×PC3; (d) PCA-k-NN classification decision boundary; (e) PCA-SVM classification; (f) PCA-decision tree classification.
[0045] Figure 7 The results of the experiment on the interference of other possible antibiotics in milk on SERS detection in Example 1 (different antibiotics at 1390 cm⁻¹) -1 (SERS intensity histogram).
[0046] Figure 8 The SERS spectra of different COF substrates in Comparative Example 1 are compared. Detailed Implementation
[0047] The embodiments of the present invention will be described in detail below with reference to examples. However, those skilled in the art will understand that the following examples are only for illustrating the present invention and should not be regarded as limiting the scope of the present invention.
[0048] Example 1: Preparation of Fe3O4@COF@Ag substrate
[0049] The preparation method of Fe3O4@COF@Ag substrate includes the following steps:
[0050] Synthesis of magnetic Fe3O4 NPs: 25 mL of ethylene glycol, 1.2 g of sodium acetate and 0.8 g of iron chloride hexahydrate were added into a teflon-lined 50 mL autoclave sterilized, shaken thoroughly for 30 minutes to make it completely dissolved; then heated in an oven at 200 °C for 10 hours. The product was washed with ethanol and then dried in a vacuum environment at 50 °C, evaporating the excess ethanol.
[0051] Synthesis of Fe3O4@COF: 15 mg of Fe3O4 NPs were dissolved in acetonitrile for 5 minutes, then 4.32 mg of m-phenylenediamine, 11.172 mg of 2,5-divinyl terephthaldehyde and 3 mL of acetic acid were added. Due to the electrostatic interaction between Fe3O4 NPs, the Schiff base reaction accelerated the growth of the COF shell on the Fe3O4 NPs. After 10 minutes, brown solids began to deposit at the bottom of the test tube. The reaction continued for 24 hours, Fe3O4@COF was collected with a magnet and washed with ethanol and acetone.
[0052] Synthesis of Fe3O4@COF@Ag: 25 mg of Fe3O4@COF and 20 mg of silver nitrate were added to 200 mL of ultrapure water, stirred for 10 minutes, then heated the reaction with sodium citrate (1%, 4 mL) as a reducing agent. After adding the reducing agent, it was heated to boiling for 30 minutes to obtain the Fe3O4@COF@Ag substrate (solution).
[0053] Example 2 Construction of standard curve
[0054] Construction of linear determination model:
[0055] (1) Preparation of sample solution: 5 mg of norfloxacin, 5 mg of ciprofloxacin and 10 mg of levofloxacin were dissolved in a beaker containing 100 mL of ultrapure water, respectively, to obtain the corresponding antibiotic solution, and mixed in equal volume to obtain a quinolone antibiotic standard solution; then add the mixed solution of ultrapure water and milk (the volume ratio of ultrapure water to milk in the mixed solution is 1:1) according to the volume ratio of 1:1 to obtain a series of concentrations (1 x 10 -5 M-1 x 10 -8 M) of quinolone antibiotic-milk solution with a total volume of 100 mL to the required experiment; mix methanol, water and the above prepared quinolone antibiotic-milk solution according to the volume ratio of 1:6:1, centrifuge at 12000 rpm for 10 minutes, and take the supernatant as the sample solution.
[0056] (2) Mix the sample solution and Fe3O4@COF@Ag substrate according to the volume ratio of 1:1, vortex for 2 minutes, and measure the Raman spectrum;
[0057] SERS spectrum was measured: the scanning condition was that the laser light source wavelength was 532 nm, the integral time was 20 s, the laser power was 10%, and the corresponding SERS spectrum was measured;
[0058] (3) Constructing a linear determination model: a curve of the relationship between different quinolone antibiotic concentrations and the characteristic peak intensity of the 1390 cm -1 -1398 cm -1 O-C-O symmetric stretching vibration in the Raman spectrum is shown in the figure. Figure 2
[0059] Ciprofloxacin: I = 9786.72 log 10 C + 78800.69, the correlation coefficient R 2 = 0.991, Levofloxacin: I = 10073.87 log 10 C + 77258.73, the correlation coefficient R 2 = 0.992, Norfloxacin: I = 10594.44 log 10 C + 82147.29, the correlation coefficient R 2 = 0.997, unit mol / L. The detection limit is as low as Ciprofloxacin: 5.61 x 10 -9 M, Levofloxacin: 1.44 x 10 -8 M, Norfloxacin: 1.56 x 10 -8 M.
[0060] Example 3: Explore the influence of different spectral pretreatment methods on the classification results
[0061] In order to compare the classification results of the normalized first derivative spectrum and the original spectrum, the same PCA-k-NN, PCA-SVM and PCA-decision tree method were used for classification. The two-dimensional PCA and classification decision boundary are shown in the figure. Figure 4 The classification accuracy of the normalized original spectrum is PCA-k-NN (81%), PCA-SVM (74%) and PCA-decision tree (74%), respectively. Figure 5 The confusion matrix analysis in the figure shows that the three models perform worst in identifying norfloxacin, and the PCA-decision tree classification only accurately predicts 33% of the norfloxacin in the training set. PCA-k-NN and PCA-SVM misclassify 30% and 36% of levofloxacin and 36% of ciprofloxacin as norfloxacin, respectively. The classification results of the normalized first derivative spectrum are shown in the figure. Figure 6 All three methods can classify quinolone antibiotics with 100% accuracy, so the spectral data is taken to the first derivative and standardized as the spectral pretreatment method in the present application.
[0062] Example 4: Explore the interference of other antibiotics that may exist in milk
[0063] To assess the impact of other types of antibiotics on SERS detection, referring to Example 1, antibiotics intended for antibacterial purposes, such as penicillins, tetracyclines, and aminoglycosides, were used as experimental interfering agents. The SERS intensity histogram for the interfering experiments is shown below. Figure 7 As shown. The antibiotic concentration was fixed at 1×10⁻⁶. -5 M, the SERS signal of other antibiotics was significantly weaker than that of ciprofloxacin, norfloxacin, and levofloxacin, 1390cm -1 The SERS signal at the location ranged from 1.18% to 6.53% of the strongest signal. Therefore, the results indicate that other antibiotics that may be present in these milk samples did not significantly interfere with the detection of quinolone antibiotics, verifying that the SERS substrate used has good anti-interference ability for the detection of quinolone antibiotics.
[0064] Example 5: Detection in actual milk samples
[0065] Referring to Example 1, under the aforementioned optimal conditions, quinolone antibiotic solutions of different concentrations were spiked into milk using the standard addition method. Following step (2) in Example 2, methanol and water were added to pretreat the antibiotic solutions. The solutions were then mixed with the substrate for Raman spectroscopy detection. The predicted concentration was obtained by fitting an equation, and the recovery rate was finally calculated. The detection results are shown in Table 1. The recovery rate and relative standard deviation in the detection results are within a reasonable range, indicating that this method is accurate and feasible.
[0066] Table 1. Recovery rate test of quinolone antibiotics in actual milk samples.
[0067]
[0068]
[0069] Comparative Example 1
[0070] Referring to Example 1, 11.172 mg of 2,5-divinyl terephthalaldehyde was replaced with 12.26 mg of trialdehyde phloroglucinol, while keeping other parameters unchanged, to prepare the corresponding COF-2 substrate.
[0071] Referring to Example 1, 4.32 mg of m-phenylenediamine was replaced with 14.06 mg of 1,3,5-tris(4-aminobenzene)benzene, while other parameters remained unchanged, to obtain the corresponding COF-3 substrate.
[0072] Referring to Example 1, 4.32 mg of m-phenylenediamine and 11.172 mg of 2,5-divinyl terephthaldehyde were replaced by 5.45 mg of 2,5-dimethyl-1,4-phenylenediamine and 12.62 mg of triformylphloroglucinol, and the rest remained unchanged to prepare the corresponding COF-4 substrate.
[0073] The above different substrates were subjected to SERS spectral detection according to Example 2, and the corresponding spectral data are shown in FIG. 2. Figure 8 The COF-1 substrate in the figure refers to the substrate prepared in Example 1.
[0074] It can be seen that different COF precursors for preparing different substrates have an impact on the enhancement performance. By comparing the SERS spectra of 10-5M levofloxacin aqueous solution of the four substrates, it can be found that the combination of COF-1 substrate and levofloxacin is more conducive to enhancement, and the main characteristic peaks of levofloxacin are all enhanced, and the intensity is much higher than that of the other three COF substrates.
[0075] The above examples are not intended to limit the scope of the present application, and the described steps are not intended to limit the execution order. Those skilled in the art can make obvious improvements to the present application in combination with the existing common knowledge, which also falls within the protection scope defined by the claims of the present application.
Claims
1. A method for simultaneous detection of multiple quinolone antibiotics in milk based on surface-enhanced Raman spectroscopy, characterized in that, Includes the following steps: (1) Magnetic Fe3O4 particles were prepared by a single-step solvothermal method using ethylene glycol, sodium acetate, and ferric chloride hexahydrate; then Fe3O4 was reacted with m-phenylenediamine, 2,5- Divinyl terephthalaldehyde and acetic acid were dissolved in acetonitrile, and Fe3O4@COF material was synthesized at room temperature using acetic acid as a catalyst. The Fe3O4@COF material was collected using a magnet. Finally, Fe3O4@COF material was subjected to a hydrothermal reaction with silver nitrate as the silver source and sodium citrate as the reducing agent to obtain Fe3O4@COF@Ag substrate. The mass ratio of Fe3O4, m-phenylenediamine, and 2,5-divinyl terephthalaldehyde is (3-5):1:(2-4); the concentration of m-phenylenediamine relative to acetonitrile is 0.1-1.0 mg / mL; the amount of acetic acid relative to m-phenylenediamine is 0.5-1.0 mL / mg; the synthesis process of Fe3O4@COF is as follows: the reaction is carried out at room temperature for 24 hours. (2) Prepare standard solutions of quinolone antibiotics with different concentration gradients, then add a mixture of ultrapure water and milk, mix well to obtain quinolone antibiotic-milk mixture; then mix methanol, water and the obtained quinolone antibiotic-milk mixture, centrifuge and collect the supernatant to obtain the corresponding sample solution; The quinolone antibiotics include any one or more of ciprofloxacin, norfloxacin, and levofloxacin; (3) The sample solution obtained in step (2) is mixed with the Fe3O4@COF@Ag-based sample obtained in step (1) to obtain a test sample solution. Raman detection is performed, and the surface-enhanced Raman spectrum of the sample solution is collected. The 1390 cm⁻¹ value in the surface-enhanced Raman spectrum is used to analyze the surface-enhanced Raman spectrum. -1 -1398 cm -1 A linear model was constructed by relating the characteristic peaks to antibiotic concentration; and a classification and recognition model was constructed by training the model using surface-enhanced Raman spectroscopy data with a machine learning algorithm.
2. The method according to claim 1, characterized in that, The method further includes: adding methanol and water to the milk sample to be tested according to the process in step (2) for pretreatment, then mixing it evenly with Fe3O4@COF@Ag substrate solution to obtain the sample to be tested, measuring the surface-enhanced Raman spectrum, and obtaining the concentration of various quinolone antibiotics in the sample to be tested according to the linear model and classification recognition model in step (3).
3. The method according to claim 1, characterized in that, In step (1), the mass ratio of sodium acetate to ferric chloride hexahydrate is (1-3):1; the amount of ethylene glycol relative to the total mass of sodium acetate and ferric chloride hexahydrate is (10-25) mL / g; the conditions for the single-step solvothermal method are: heating the reaction at 150-200°C for 10-15 hours.
4. The method according to claim 1, characterized in that, In step (1), the mass ratio of Fe3O4@COF to silver nitrate is (1-5):(2-3.5); the concentration of Fe3O4@COF in the hydrothermal reaction is 0.03-0.3 mg / mL; the mass ratio of Fe3O4@COF to sodium citrate is (1-5):(3-6); the synthesis conditions of Fe3O4@COF@Ag are: incubation at room temperature with stirring for 10 minutes, addition of reducing agent and heating to boiling for 30 minutes to obtain Fe3O4@COF@Ag substrate.
5. The method according to claim 1, characterized in that, The preparation method for the quinolone antibiotic standard solution in step (2) is as follows: Norfloxacin, ciprofloxacin, and levofloxacin are dissolved in ultrapure water to prepare 30-70 mg / L norfloxacin solution, 30-150 mg / L ciprofloxacin solution, and 30-150 mg / L levofloxacin solution, respectively; then the three solutions are mixed in equal volumes; the volume ratio of the quinolone antibiotic standard solution to the mixture of ultrapure water and milk is 1:(1-3); the concentration of each antibiotic in the quinolone antibiotic-milk mixture is 1×10⁻⁶. -5 M-1×10 -8 M.
6. The method according to claim 1, characterized in that, In step (2), the volume ratio of methanol, water, and quinolone antibiotic-milk solution is 1:(4-6):
1.
7. The method according to any one of claims 1-6, characterized in that, In step (3), the SERS spectral data preprocessing method used in the machine learning algorithm classification model is to take the first derivative spectrum of the SERS spectrum and perform normalization preprocessing; the kernel function used in the PCA-SVM algorithm is the RBF function.
8. The application of the method according to any one of claims 1-7 in the field of food testing.