Raman spectrum method for direct detection of anions, detection substrate and recognition method

Abstract

The invention provides a Raman spectrum method for direct detection of anions, a detection substrate and an identification method, silver nanoparticles reduced by sodium borohydride and calcium ions are fused as aggregation agents to obtain an efficient'hot spot ', accurate identification of various anions is realized, and by capturing a unique SERS fingerprint spectrum of target molecules, the detection substrate can be used for detecting the anions directly. The stability and sensitivity of the detection method are deeply evaluated, a machine learning technology is introduced, particularly for distinguishing different anions, fluorine ions in real urine samples from people in different regions are successfully detected, fluctuation of other components in the urine can be ignored, and the detection accuracy is improved. According to the method, the characteristic fingerprints of fluorine ions in urine of different crowds are rapidly and accurately captured, subtle differences between the characteristic fingerprints are sensitively recognized, the method has wide applicability, and changes are brought to multiple fields such as food safety monitoring, environment quality detection and medical diagnosis.

Description

Technical Field

[0001] The present invention relates to the field of rapid Raman detection, and in particular to a Raman spectroscopy method for directly detecting anions, a detection substrate and an identification method. Background Art

[0002] Anions widely exist in nature and are applied in the fields of industry, agriculture, medicine and catalysis. For example, sulfite, sulfate, nitrite, nitrate, phosphate and fluoride ions are often used as key additives. The water environment is an important resource for human survival and development, and its quality directly affects the balance of the ecosystem and human health. With the acceleration of the industrialization process, the problem of water pollution is becoming increasingly serious, especially the pollution of anions, which has become a serious challenge that cannot be ignored. The accumulation of harmful anions in water not only reduces the water quality standard, but may also be amplified step by step through the ecological chain and finally enter the human food chain, posing a potential and profound irreversible threat to human health. For example, excessive intake of fluoride in drinking water may cause skeletal fluorosis and neurological dysfunction; while sulfite, as a food preservative, can effectively prevent food browning, but its excessive use is closely related to health problems such as chronic respiratory diseases and cardiovascular diseases. In addition, nitrite, as a common additive in meat products, can maintain color, improve flavor and extend the shelf life, but its excessive intake will endanger health, oxidize normal hemoglobin in the blood to methemoglobin, induce methemoglobinemia, and may even induce serious diseases such as gastric cancer, esophageal cancer and colon cancer.

[0003] In view of the above hazards, the development of efficient, sensitive and rapid anion detection technologies in the human body is of profound significance for protecting human health. The research and development of sensors that can accurately and quantitatively detect low-concentration ions can not only monitor harmful anions (fluoride ions) in the body in real time, timely warn of potential health risks, but also, with the help of machine learning algorithms, intelligently distinguish the fluoride exposure levels of different populations, and thus provide a scientific basis for personalized health management and disease prevention. Such technological innovation can not only improve the level of public health, but also play an important role in many fields such as medical diagnosis and environmental protection, escorting the health and well-being of mankind.

[0004] At present, the main techniques for detecting anions cover a variety of methods such as spectrophotometry, potentiometric titration, colorimetry, ion chromatography, and ion-selective electrode method. However, when these methods are applied to the detection of anion samples, they all face a series of challenges and limitations. First of all, the sample preparation process is often complex and time-consuming, increasing the overall difficulty and cost of detection. Secondly, the sensitivity of some methods is limited, making it difficult to capture the presence of trace anions and affecting the accuracy of detection. Moreover, the selectivity problem is also a major concern, that is, these methods may not be able to effectively distinguish the target anion from other similar ions in complex matrices, resulting in result deviations. In addition, interfering factors such as coexisting ions and matrix effects often have an adverse impact on the detection results, further restricting the scope of application of these methods. At the same time, the analysis cycle is long, making it difficult to meet the requirements of rapid detection. More critically, the reliability of the detection results is often restricted by multiple factors such as laboratory environmental conditions, instrument performance, and the proficiency of technical personnel, increasing the uncertainty of the results. Although these methods play an important role in the field of anion detection, their applications are limited due to their limitations.

[0005] Surface-Enhanced Raman Scattering (SERS) has attracted people's attention due to its good stability and the sensitivity that can reach single-molecule detection. Since the discovery of this technology, it has been rapidly and widely applied in many fields such as food, medicine, biology, and chemistry. SERS introduces active nanosubstrates as key elements. Surface-enhanced Raman spectroscopy significantly enhances the Raman signal of the substance to be detected through the "hot spot" effect on the nanosubstrate.

[0006] At present, SERS has become an ideal technology for detecting anions in aqueous environments, which can avoid the interference of water and provide fingerprint signals even at the single-molecule level. For example, Li et al. combined 1,4-diketone-3,6-diphenylpyrrole (DPP)-based compounds with 1-iodobutane (DPP1) to selectively detect fluoride ions by SERS. The detection system was able to detect inorganic fluoride as low as 1.0 μmol / L (0.019 mg / L), far lower than the drinking water standard recommended by the Public Health Service. Kitaw et al. used silver nanospheres coated with polyethyleneimine (AgNS@PEI) as a surface-enhanced Raman scattering (SERS) substrate to successfully achieve sensitive and rapid detection of sulfite residues in beer samples. The experimental results showed that this SERS substrate had excellent sensitivity, selectivity, repeatability, and stability for sulfur dioxide, with a detection limit of 0.48 mg / L, meeting the requirements of actual sample analysis. However, this method detected sulfur dioxide by converting the sulfite analysis sample into sulfur dioxide, rather than directly detecting sulfite. Zhang et al. (2020) developed a composite SERS probe based on 4-aminothiophenol-modified hydroxyapatite nanotubes / silver nanoparticles for in-situ derivatization and selective determination of nitrite ions in meat products. The detection limit could reach 0.51 μg / L, and it could be applied to the detection of nitrite ions in sausages and luncheon meats, with a relative standard deviation of less than 10.3%. The results were consistent with those analyzed by ultraviolet-visible spectroscopy.

[0007] The above several methods all have good stability for anions in aqueous environments and can obtain Raman signals. However, there are problems such as cumbersome operation and high cost, and appropriate substrates or probes need to be selected according to the actual situation. The direct determination method can only detect one anion and cannot achieve simultaneous detection of multiple anions. Currently, there has been no publicly reported technology for simultaneously detecting multiple anions using a single SERS substrate.

[0008] Artificial Intelligence (AI) incorporates a variety of Machine Learning (ML) algorithms, which demonstrate powerful capabilities in data processing and analysis. Machine learning algorithms can be mainly divided into two major categories: supervised learning and unsupervised learning. In supervised learning, the algorithm relies on pre-given labels to train the model, so as to learn how to classify or predict new data. It includes Linear Discriminant Analysis (LDA), Support Vector Machine (SVM), and Artificial Neural Network (ANN), etc. These models strive to build a model that can accurately predict or classify new data under the guidance of known labels through continuous optimization. Unsupervised learning is a more free learning method that does not rely on any predefined labels. In unsupervised learning, the computer needs to explore the internal structure and patterns in the data by itself, and then classify or cluster the data. K-means clustering, hierarchical clustering, Principal Component Analysis (PCA), and T-distributed Stochastic Neighbor Embedding (t-SNE) are commonly used methods in unsupervised learning. These methods can reveal the hidden information in the data and provide a new perspective for data analysis and mining. For multivariate data such as Raman spectra, the above machine learning algorithms can exert their powerful automatic learning ability to build prediction and recognition models from the data. These models can not only improve the efficiency of fingerprint spectrum recognition, but also avoid the occurrence of overfitting phenomenon through a more accurate decision boundary. At the same time, the output results of these models are also more intuitive and easy to interpret, providing strong decision-making support for scientific researchers. This combination not only improves the analysis efficiency and accuracy of SERS spectral data, but also broadens the application scope of SERS technology. Summary of the Invention

[0009] The object of the present invention is to provide a Raman spectroscopy detection substrate for direct anion detection, which combines sodium borohydride-reduced silver nanoparticles and calcium ions as aggregating agents to create efficient "hot spots" and achieve precise recognition of various anions.

[0010] The object of the present invention is achieved by the following technical solutions: A Raman spectroscopy detection substrate for direct anion detection, and the preparation method of the detection substrate is as follows: Add an aqueous solution of sodium borohydride with a concentration of 0.3 - 0.4 mol / L to deionized water, and stir to disperse sodium borohydride evenly in water; Add 0.03 - 0.04 mol / L of silver nitrate, and continue to stir until the reaction is complete; Centrifuge at 5000 - 5500 rpm for 20 min, and remove the supernatant to obtain silver sol; Mix the silver sol with a 0.5 - 2 mmol / L CaCl₂ solution, and incubate at room temperature for ≥30 min to obtain a uniform and stable spherical SERS detection substrate Ag@BOCNPs.

[0011] As a more preferred technical solution of the present invention, the volume ratio of water to sodium borohydride is 490:5.

[0012] As a more preferred technical solution of the present invention, the calcium chloride solution is 1 mmol / L.

[0013] As a more preferred technical solution of the present invention, the concentration of sodium borohydride is 0.35 mol / L.

[0014] As a more preferred technical solution of the present invention, the concentration of silver nitrate is 0.0389 mol / L.

[0015] As a more preferred technical solution of the present invention, the volume ratio of the silver sol to the calcium chloride solution is 5:0.1.

[0016] As a more preferred technical solution of the present invention, the volume ratio of silver nitrate to sodium borohydride is 1:99.

[0017] As a more preferred technical solution of the present invention, the centrifugation temperature is 25°C, and the rotation speed is 5200 rpm for centrifugation. If the rotation speed is too high, the nanoparticles will adhere to the wall, and if the rotation speed is too low, the nanoparticles cannot be separated.

[0018] Another object of the present invention is to provide a Raman spectroscopy detection method for direct anion detection, including the following steps: Step 1: Mix the test solution with the detection substrate Ag@BOCNPs, and stir evenly to obtain a mixture. The anions in the test solution adsorb on the surface of Ag@BOCNPs to form SERS active hot spots.

[0019] Step 2: Shake the mixture well and put it into a Raman spectrometer for SERS detection.

[0020] As a more preferred technical solution of the present invention, the volume ratio of the test solution to the detection substrate Ag@BOCNPs is 1:1.

[0021] Another object of the present invention is to provide a method for identifying anions in a complex system, comprising the steps of: The PCA method is used to perform dimensionality reduction processing on the raw data obtained by the Raman spectroscopy detection method for direct anion detection, and a feature vector that can maximize the retention of key information in the raw data is extracted; The reduced-dimensional data is input into the SVM classifier for training and testing, and the data set is divided into a 70% training set and a 30% test set; The test set is used to evaluate the model performance and the classification results are quantified by the confusion matrix.

[0022] The beneficial effects of the present invention are as follows: Ag@BOCNPs are silver nanoparticles successfully prepared by reducing silver nitrate with sodium borohydride, and calcium ions are added as aggregating agents to aggregate on the surface of the nanoparticles to generate "hot spots". Ag@BOCNPs are used as a SERS detection platform to perform label-free SERS detection on different types of anions, and eight types of anions are successfully and specifically distinguished by combining machine learning algorithms. This achievement is of milestone significance in the field of anion recognition in complex systems.

[0023] The present invention successfully developed an Ag@BOCNPs-SERS detection platform based on silver nanoparticles (AgNPs) and sodium borohydride reduction method, and innovatively introduced calcium ions as aggregating agents to induce the formation of "hot spots" on the surface of nanoparticles, significantly enhancing the SERS signal. The platform demonstrated excellent label-free SERS detection capabilities for a variety of anions, with good stability, sensitivity, and reproducibility.

[0024] More importantly, the present invention successfully applied the Ag@BOCNPs-SERS detection platform to the analysis of real samples, which can effectively distinguish the fluoride ion signals of different fluoride exposure levels, especially in distinguishing the fluoride ion signals in the urine of people with different fluoride exposure levels, showing extremely high sensitivity and accuracy. And combined with the SVM model, the model achieved an accuracy rate of 100%. The SVM classifier can accurately distinguish urine samples from people in areas with different fluoride exposure levels. It not only verifies the great potential of the platform in the fields of environmental monitoring, public health and clinical diagnosis, but also provides strong technical support for the early screening and risk assessment of fluoride poisoning, further verifying the potential and value of the SERS detection platform in clinical applications.

[0025] In summary, the Ag@BOCNPs-SERS detection platform constructed in the present invention, combined with machine learning algorithms, not only realizes the efficient and specific detection of multiple anions, but also is successfully applied to the analysis of fluoride ions in real samples, opening up a new way for the application of SERS technology in the analysis of complex systems, and has important scientific value and application prospects. Description of the Drawings

[0026] Figure 1 TEM image of Ag@BONPs without added calcium ions.

[0027] Figure 2 TEM image of Ag@BOCNPs with added calcium ions.

[0028] Figure 3 Particle size change of Ag@BOCNPs before and after adding calcium ions.

[0029] Figure 4 Zeta potential change of Ag@BOCNPs before and after adding calcium ions.

[0030] Figure 5 For Figure 5 SERS spectra of 1 μg / mL fluoride ion aqueous solution under different detection conditions: (a) Ag@BOCNPs substrate, (b) pure fluoride ion solution, (c) Ag@BOCNPs + fluoride ion.

[0031] Figure 6 SERS spectrum of 1 μg / mL sulfite ion aqueous solution.

[0032] Figure 7 SERS spectrum of 1 μg / mL sulfate ion aqueous solution.

[0033] Figure 8 SERS spectrum of 1 μg / mL nitrite ion aqueous solution.

[0034] Figure 9 SERS spectrum of 1 μg / mL nitrate ion aqueous solution.

[0035] Figure 10 SERS spectrum of 1 μg / mL phosphate ion aqueous solution.

[0036] Figure 11 SERS spectrum of 1 μg / mL bromide ion aqueous solution.

[0037] Figure 12 SERS spectrum of 1 μg / mL iodide ion aqueous solution.

[0038] Figure 13PCA analysis results for 8 different anions.

[0039] Figure 14 LDA analysis results for 8 different anions.

[0040] Figure 15 t-SNE analysis results for 8 different anions.

[0041] Figure 16 SERS spectra of sulfite ions at different concentrations (0.01 - 0.05 mg / mL) in acetonitrile.

[0042] Figure 17 Histogram I of the intensity ratio change of sulfite ion concentration 924 / I 2255 (n = 5).

[0043] Figure 18 SERS spectra of real urine from different populations.

[0044] Figure 19 SERS intensity heatmap of populations with different fluoride exposure levels.

[0045] Figure 20 PCA analysis of populations with different fluoride exposure levels.

[0046] Figure 21 SVM analysis of populations with different fluoride exposure levels. Detailed implementation manners

[0047] The present invention will be further described in detail below in combination with the specific implementation manners. The given embodiments are only for clarifying the present invention and should not be construed as a limitation to the present invention.

[0048] For those not specified in the embodiments regarding specific technologies or conditions, they shall be carried out according to the technologies or conditions described in the literature in this field or according to the product specifications. For reagents or instruments not specified by the manufacturer, they are all conventional products that can be obtained through commercial purchase. The anion solutions used were purchased from Shanghai Macklin Biochemical Co., Ltd. (Shanghai, China) and Shanghai Aladdin Biochemical Technology Co., Ltd. (Shanghai, China) respectively, and stored at 4 - 8°C. The SERS tester used was manufactured by WITec Alpha300R (Germany).

[0049] The present invention is based on SERS technology. This method can quickly detect anions in water, with a detection time (<5 min), and can distinguish different anion solutions, having good specificity and being able to perform quantitative analysis on sulfite ions. It eliminates the disadvantages of long detection time and complex operation in traditional anion detection.

[0050] Example 1 Prepare silver sol Ag@BONPs according to the method of reducing anions with sodium borohydride. Use sodium borohydride as a reducing agent to reduce silver nitrate to silver, and then form silver nanoparticles. Figure 1 The transmission electron microscope (TEM) image intuitively shows the morphology of Ag@BONPs (silver nanoparticles without added calcium ions).

[0051] Example 2 Adding calcium ions as an aggregating agent to the nanoparticles is an effective method, aiming to promote the aggregation between the nanoparticles, thereby increasing the formation of "hot spots". Hot spots refer to areas where the nanoparticles are very close to each other. Due to the enhanced effect of the electromagnetic field in these areas, the Raman scattering signal can be greatly enhanced. The addition of calcium ions interacts with the surface of the nanoparticles through electrostatic or coordination effects, resulting in an enhanced attraction between the nanoparticles and thus aggregation. Figure 2 The transmission electron microscope (TEM) image shows the morphology of Ag@BOCNPs (silver nanoparticles with added calcium ions) intuitively. It can be observed that after adding calcium ions, the nanoparticles do aggregate to form larger particle clusters, which is beneficial to the formation of hot spots and signal enhancement.

[0052] A method for label-free detection of anions in water using SERS technology specifically includes the following steps: Step 1: Synthesize the SERS substrate by the method of reducing silver ions with sodium borohydride: Add 5 mL of 0.35 M sodium borohydride to 490 mL of water. Sodium borohydride is used as a reducing agent here for subsequent reduction of silver ions. Stir the mixture at 1650 rpm for 8 minutes to ensure the uniform dispersion of sodium borohydride in water. Then add 5 mL of 0.0389 mol / L silver nitrate, increase the stirring rate to 2050 rpm, and continue stirring for 18 minutes. The silver ions are reduced to silver atoms and gradually aggregate into nanoparticles. Remove the supernatant by centrifugation (5200 rpm, 25°C, 20 minutes) to obtain silver sol. Mix the centrifuged silver sol with calcium chloride solution (1 mmol / L) at a volume ratio of 5:0.1 at room temperature for 60 minutes. The obtained uniform and stable spherical Ag@BOCNPs are the key substrates for SERS detection.

[0053] Step 2: Prepare the anion solution as the sample to be tested, mix the solution to be tested with Ag@BOCNPs at a volume ratio of 1:1, and stir evenly. The analyte will adsorb on the surface of Ag@BOCNPs to form SERS active hot spots.

[0054] Step 3: Shake the mixture well for SERS detection. SERS analysis is performed using a Raman spectrometer (Wintec, Germany).

[0055] The experimental parameters were as follows: Laser control: 532 nm (30 mW), scanning time: 15 s, accumulation: 1 time.

[0056] The particle sizes of the nanoparticles before and after adding calcium ions were detected. Figure 3 It was shown that the particle size increased after adding calcium ions due to the aggregation of the nanoparticles.

[0057] The Zeta potential of the nanoparticles before and after adding calcium ions was detected. Figure 4 It was found that the absolute value of the Zeta potential changed after adding calcium ions, and the decrease in the absolute value of the Zeta potential reflected the reduction of the surface charge density of the nanoparticles. The Zeta potential is an important parameter characterizing the stability of the colloidal dispersion system. The smaller its absolute value, the smaller the electrostatic repulsion between the particles, and the easier it is for aggregation to occur. Through the above examples, it was proved that the Ag@BOCNPs enhanced substrate was successfully prepared, and this substrate could be used for SERS detection.

[0058] In other methods of the present invention for label-free detection of anions in water using SERS technology, it was verified that the difference was only in the addition of 0.1 mL, 1 mL, 3 mL, and 6 mL of sodium borohydride. A small amount of sodium borohydride added would cause incomplete reduction of silver nitrate, and silver nitrate would be contained in the substrate, affecting the signal of the substance to be detected. The range of sodium borohydride in the present invention is between 1 - 5 mL.

[0059] In other methods of the present invention for label-free detection of anions in water using surface-enhanced Raman spectroscopy technology, it was verified that the difference was only in the addition of 0.03 and 0.04 mol / L of silver nitrate. Therefore, the concentration of silver nitrate in the present invention is between 0.03 - 0.04 mol / L.

[0060] In other methods of the present invention for label-free detection of anions in water using SERS technology, it was verified that the difference was only in the concentration of calcium chloride solution being 0.1, 0.5, 2, and 3 mmol / L. Too high concentration (2 mmol / L) or too low concentration (0.5 mmol / L) would affect the formation of hot spots, or no hot spots would be formed at 3 mmol / L. Therefore, the concentration of calcium chloride solution in the present invention was determined to be between 0.5 - 2 mmol / L.

[0061] In other methods of the present invention for label-free detection of anions in water using SERS technology, it was verified that the difference was only in the volume ratio of silver nitrate to sodium borohydride being 1:95 and 1:105. Therefore, the volume ratio of silver nitrate to sodium borohydride in the present invention is 1:(95~105). In other methods of the present invention for label-free detection of anions in water using surface-enhanced Raman spectroscopy, it has been verified that the only difference lies in centrifugation at speeds of 5000 and 5500 rpm. At a high speed, the nanoparticles will adhere to the wall, and at a low speed, the nanoparticles cannot be separated. Therefore, the rotation speed of the present invention is between 5000 and 5500.

[0062] Example 3 SERS detection was performed on fluoride ions in an aqueous solution. Figure 5 The SERS spectra of an aqueous fluoride ion solution (1 μg / mL) in different states are shown. The results show that no obvious SERS signal was observed for Ag@BOCNPs itself, which proves that this substrate material will not generate background signal interference to the analyte during the detection process, ensuring the accuracy of the detection results. Subsequently, the SERS detection results of an aqueous fluoride ion solution (1 μg / mL) without Ag@BOCNPs were obtained. Similarly, no characteristic signal peaks appeared, indicating that the Raman scattering signal of the aqueous fluoride ion solution is very weak and difficult to detect without an enhancement substrate. However, when Ag@BOCNPs were added to the aqueous fluoride ion solution (1 μg / mL) and SERS detection was performed, obvious characteristic signal peaks appeared at 848, 924, 1131, 1201, 1379, 1606 cm -1 This marked a significant enhancement of the Raman signal of fluoride ions. This result not only proves the effectiveness of Ag@BOCNPs as a SERS enhancement substrate but also demonstrates its potential in practical applications.

[0063] Example 4 The difference between this example and Example 3 is that the analyte sample is sulfite ions. As Figure 6 shown, 622 cm -1 is the symmetric bending vibration of O-S-O (δsym O-S-O), and 924 cm -1 is the symmetric stretching vibration of S-O (νsym S-O), making the signal more directly reflect its existence state and improving the accuracy and reliability of the detection.

[0064] Example 5 The difference between this example and Example 3 is that the analyte sample is sulfate ions. As Figure 7 shown, an obvious characteristic signal appeared at 986 cm -1 which is the signal of the symmetric stretching (ν1) of sulfate anions, proving the sensitivity of SERS technology to its structure.

[0065] Example 6 The difference between this example and Example 3 is that the analyte sample is nitrite ions. As Figure 8As shown, nitrite ions exhibit distinct characteristic signals at wavenumbers such as 931, 1392, 1436, and 1607 cm -1 etc. Among them, 1392 corresponds to the stretching vibration of the N-N bond, and 1436 corresponds to the in-plane vibration of the N-N bond.

[0066] Example 7 The difference between this example and Example 3 is that the sample to be tested is nitrate ions. As Figure 9 shown, the results show that it only generates a significant in-plane symmetric stretching vibration signal at 1066 cm -1 . This indicates that the binding mode between anions and silver nanoparticles is highly specific and is not affected by the presence of the same or similar external ion populations.

[0067] Example 8 The difference between this example and Example 3 is that the sample to be tested is phosphate ions. As Figure 10 shown, 931 cm -1 corresponds to VsP-(OH), and 1087 cm -1 corresponds to VsPO2 of HPO 4- .

[0068] Example 9 The difference between this example and Example 3 is that the sample to be tested is bromide ions. As Figure 11 shown, bromide ions only exhibit a significant characteristic signal at 1609 cm -1 . This unique vibration mode serves as the "identity marker" of bromide ions.

[0069] Example 10 The difference between this example and Example 3 is that the sample to be tested is iodide ions. As Figure 11 shown, iodide ions exhibit more complex signal characteristics. They show obvious characteristic signals at both 1438 cm -1 and 1647 cm -1 . These signals together constitute the unique fingerprint of iodide ions in the SERS spectrum. This difference in signal characteristic peaks not only reflects the high sensitivity and high selectivity of SERS technology in anion detection but also provides a new method for quickly and accurately distinguishing different anions. By comparing the SERS spectra of different ions, the target ion can be clearly identified and qualitatively analyzed, which is of great significance for research and applications in fields such as environmental monitoring, food safety, and biomedicine.

[0070] Example 11 The SERS spectra of 20 random groups of sulfite ions, sulfate ions, nitrite ions, nitrate ions, phosphate ions, fluoride ions, bromide ions, and iodide ions were analyzed using the machine learning algorithm PCA. As a powerful dimensionality reduction technique, during the PCA process, it was observed that the first principal component (PC1) was particularly crucial. It can usually capture the largest variability in the dataset, that is, the main differences between the SERS spectra of different anions. In Figure 13 the display, it can be seen that based on the projection results of PC1, the sample points of different anions are clearly separated, forming non-overlapping 95% confidence ellipses. The first two PCs account for 57.6% of the variance (32.9% and 24.7%). This phenomenon fully demonstrates the effectiveness of PCA in capturing the characteristic differences of anion SERS spectra and also verifies the powerful ability of the machine learning model in specific discrimination tasks.

[0071] Example 12 The difference between this example and Example 11 is that the machine learning algorithm is LDA. Different from the PCA method, LDA can not only effectively utilize the characteristic peak positions of each anion in the SERS spectrum as composite variables, but also further reveals the overall information structure and differences between classes in the spectral data through its unique dimensionality reduction and classification strategies. Especially when dealing with high-dimensional data, it can significantly reduce the computational cost and improve the classification efficiency. LDA processing was performed on 20 random groups of SERS spectral data of 8 different anions. As Figure 14 shown, the results of LDA analysis are presented in the form of a point set in a multi-dimensional space. Each point represents a spectral sample, and its coordinates in each dimension reflect the performance of the sample at the characteristic peak positions. This result not only intuitively demonstrates the excellent ability of LDA in distinguishing different anion SERS spectra but also verifies the effectiveness of the characteristic peak positions as composite variables. More importantly, it shows that LDA can intelligently identify the subtle differences in the spectral data, thus achieving specific classification of 8 anions.

[0072] Example 13 The difference between this example and Example 11 is that the machine learning algorithm is t-SNE. In Figure 15shows the results of non-linear dimensionality reduction processing of Raman signals in the range of 400 - 1800 cm⁻¹, which are effectively mapped onto a two-dimensional space, specifically through two main components of t-SNE (tSNE1 and tSNE2). The t-SNE map clearly reveals a significant separation between SERS spectra of different anion types. The spectral data points of each anion category are closely clustered together, forming separate clusters without any overlap, which fully demonstrates that the detection platform has extremely high specificity and accuracy in differentiating different anions.

[0073] Example 14 The quantitative ability of this SERS detection platform was deeply evaluated. To achieve precise quantification of sulfite ions, acetonitrile was selected as the internal standard substance, and its characteristic peak at 2255 cm⁻¹ was used to normalize the SERS spectra to eliminate the influence of experimental condition fluctuations on the measurement results. As Figure 16 shown, during the quantification process, the signal peak at 924 cm⁻¹ was focused on. This peak corresponds to the symmetric stretching vibration of the S - O bond (νsym S - O), which is a characteristic spectral band of sulfite ions. By collecting SERS spectra of sulfite ions at different concentrations (0.01 mg / mL to 0.05 mg / mL), it was observed that as the concentration of sulfite ions gradually increased, the intensity of the 924 cm⁻¹ signal peak showed an obvious increasing trend.

[0074] To further quantify the relationship between this concentration and peak intensity, a histogram of the ratio of sulfite ion concentration to internal standard peak intensity (I 924 / I 2255 ) was constructed. Figure 17 The analysis results showed that the correlation equation was y = 0.01013x - 0.0156 (R 2 = 0.9957). The ratio of I 924 / I 2255 peak intensity and the concentration of sulfite ions showed a good linear growth relationship, and this finding verified the accuracy and reliability of the SERS detection platform in quantitative detection.

[0075] To deeply explore the broad prospects of this SERS detection platform in clinical applications, experimental analysis was specifically carried out on urine samples from different people. This study aimed to evaluate the platform's ability to identify and quantify the fluoride ion content in urine, and thus evaluate different fluoride exposure levels in the human body. In Figure 18It shows the comparison results of the SERS fingerprint spectra of urine from different people and aqueous solutions of fluoride ions. It is found that by comparing the characteristic signals of fluoride ions (1131 cm⁻¹ and 1201 cm⁻¹), the fluoride exposure levels in the human body can be distinguished. It is found that the characteristic signals of fluoride ions (1131 cm⁻¹ and 1201 cm⁻¹) are not detected in artificial urine, while people with different exposure levels show different peak intensities at the characteristic peaks of 1131 cm⁻¹ and 1201 cm⁻¹, which reflects different fluoride exposure levels in the urine of different individuals. This finding is directly related to the degree of fluoride pollution in the region and the exposure risk of the population.

[0076] To further visually show the differences in the intensities of SERS characteristic peaks at different fluoride ion levels, the heat map in Figure 19 is made. The heat map intuitively reflects the changes in the intensities of fluoride ion characteristic peaks with the depth of color, making the differences between different samples obvious at a glance. This visualization method not only enhances the readability of the data but also provides strong support for subsequent data analysis.

[0077] Given the significant differences in the fluoride ion concentrations in the urine of different populations, the present invention provides a more efficient and accurate method for discrimination and classification. Machine learning algorithms are introduced, especially the combined application of PCA (Principal Component Analysis) and SVM (Support Vector Machine). First, the PCA method is used to perform dimensionality reduction on the original data to extract the most representative feature vectors, which can retain the key information in the original data to the greatest extent. Figure 20 As shown by the dimensionality reduction by PCA, the complex data set is successfully simplified into a low-dimensional feature space, laying a solid foundation for subsequent classification tasks. Next, the data after dimensionality reduction is input into the SVM classifier for training and testing. The data set is divided into a 70% training set and a 30% test set. As a powerful supervised learning algorithm, SVM can find the optimal hyperplane in the high-dimensional space to separate samples of different classes. Finally, the test set is used to evaluate the performance of the model and generate a confusion matrix. As shown in Figure 21 According to the results of the confusion matrix, the model achieves an accuracy rate of 100%.

[0078] The present invention promotes the application of SERS in the field of anion detection. This platform demonstrates excellent capabilities for label-free SERS detection of various anions, featuring good stability, sensitivity, and reproducibility. By combining machine learning algorithms, specific differentiation of eight different anions is achieved, which is of milestone significance in the field of anion recognition in complex systems. In this patent, the SVM classifier can accurately distinguish urine samples of people with different levels of fluoride exposure. This discovery not only verifies the great potential of this platform in the fields of environmental monitoring, public health, and clinical diagnosis, but also provides strong technical support for the early screening and risk assessment of fluorosis. It further validates the potential and value of this SERS detection platform in clinical applications.

[0079] The detection method provided by the present invention has important prospects in the detection of actual samples.

[0080] The present invention innovatively develops a method for directly detecting anions using SERS technology. By integrating sodium borohydride-reduced silver nanoparticles and calcium ions as aggregators, efficient "hot spots" are created to achieve precise recognition of various anions. Through capturing the unique SERS fingerprint spectra of target molecules, the stability and sensitivity of this detection method are deeply evaluated. To further improve the recognition accuracy, machine learning techniques are introduced, especially for the differentiation of different anions. On this basis, the influence of changes in sulfite ion concentration on its SERS peak intensity is specifically explored, providing a scientific basis for optimizing detection conditions. In addition, this innovative method is applied to actual scenarios, and fluoride ions in real urine samples of different people are successfully detected. Combining artificial intelligence algorithms, principal component analysis (PCA) is used to effectively reduce the dimensionality of complex SERS data, and then accurate analysis and prediction are carried out through support vector machine (SVM). This method can ignore the fluctuations of other components in urine, quickly and accurately capture the characteristic fingerprint spectra of fluoride ions in the urine of different people, and sensitively identify the subtle differences between them. This platform not only demonstrates significant advantages of high sensitivity, rapid response, and high cost-effectiveness, but also has wide applicability, bringing changes to multiple fields such as food safety monitoring, environmental quality detection, and medical diagnosis.

[0081] Although the embodiments of the present invention are described and illustrated above, the above-described embodiments are merely illustrative and not restrictive of the present invention. Within the scope of the present invention, those skilled in the art can change, modify, substitute, and transform the above embodiments. Finally, it should be noted that: the above embodiments are only used to illustrate rather than limit the technical solutions of the present invention. Although the present invention has been described in detail with reference to the above examples, those of ordinary skill in the art should understand that: the present invention can still be modified or partially substituted, and all of them should be covered by the scope of the claims of the present invention.

Claims

1. A Raman spectroscopy detection substrate for direct anion detection, and the preparation method of the detection substrate is as follows: Add sodium borohydride with a concentration of 0.3 - 0.4 mol / L to water, and stir to uniformly disperse sodium borohydride in water; Add silver nitrate with a concentration of 0.03 - 0.04 mol / L, and continue to stir until the reaction is complete; Centrifuge at a rotation speed of 5000 - 5500 rpm for 20 minutes, and remove the supernatant to obtain silver sol; Mix the silver sol with a calcium chloride solution with a concentration of 0.5 - 2 mmol / L at room temperature for more than 30 minutes to obtain a uniform and stable spherical SERS detection substrate Ag@BOCNPs.

2. The Raman spectroscopic detection substrate for direct anion detection according to claim 1, characterized in that: The volume ratio of the water to sodium borohydride is 490:(5 - 15).

3. The Raman spectroscopy detection substrate for direct anion detection according to claim 1, characterized in that: The concentration of the calcium chloride solution is 1 mmol / L.

4. The Raman spectroscopy detection substrate for direct anion detection according to claim 1, wherein: The concentration of the sodium borohydride is 0.35 mol / L.

5. The Raman spectroscopy detection substrate for direct anion detection according to claim 1, wherein: The concentration of the silver nitrate is 0.0389 mol / L.

6. The Raman spectroscopic detection substrate for direct anion detection according to claim 1, wherein: The volume ratio of the silver nitrate to sodium borohydride is 1:(95 - 105).

7. The Raman spectroscopy detection substrate for direct anion detection according to claim 1, wherein: The centrifugation temperature is 25°C and the centrifugation speed is 5200 rpm.

8. A Raman detection method for anions, characterized in that, It includes the following steps: Step 1: Mix the solution to be detected with the detection substrate Ag@BOCNPs, and stir evenly to obtain a mixture. Anions in the solution to be detected adsorb on the surface of Ag@BOCNPs to form SERS active hot spots; Step 2: Shake the mixture well and put it into a Raman spectrometer for SERS detection.

9. The Raman spectroscopic detection substrate for direct anion detection according to claim 6, wherein: The volume ratio of the solution to be detected to the detection substrate Ag@BOCNPs is 1:

1.

10. A method for anion recognition in a complex system, characterized in that, It includes the following steps: Use the principal component analysis (PCA) method to perform dimensionality reduction on the spectrum obtained by the Raman detection method for anions as described in Claim 8, and extract eigenvectors that can maximize the retention of key information in the spectrum; Input the data after dimensionality reduction into a support vector machine (SVM) classifier for training and testing, and divide this data set into a 70% training set and a 30% test set; Use the test set to evaluate the performance of the model and generate a confusion matrix.

Images