A cobalt complex, a composite SERS substrate and application thereof in detection of pesticide residues in fruits and vegetables

Abstract

The present application relates to a kind of cobalt complex, composite SERS substrate and its application in detecting pesticide residues of fruits and vegetables, belong to the field of biotechnology.The present application provides a kind of cobalt complex, and is loaded on nano-silver to obtain composite SERS substrate, utilizes the composite SERS substrate to detect pesticide residues on fruits and vegetables, not only can realize the precise identification and qualitative analysis of the surface of fruits and vegetables thiram, imazalil, imidacloprid three kinds of pesticide residues simultaneously, and simple when detecting, detection sensitivity is high.

Description

Technical Field

[0001] This invention belongs to the field of biotechnology, and in particular relates to a cobalt complex, a composite SERS substrate, and its application in the detection of pesticide residues in fruits and vegetables. Background Technology

[0002] Pesticides are widely used in agricultural production to control pests, diseases, and weeds and ensure crop yields. However, improper application can easily lead to excessive pesticide residues in agricultural products, soil, and water. Therefore, establishing efficient and sensitive pesticide residue detection technologies is of great significance for ensuring the quality and safety of agricultural products and protecting consumer health.

[0003] When surface-enhanced Raman scattering (SERS) technology is applied to the detection of pesticide residues in fruits and vegetables, problems such as complex pretreatment and poor compatibility of the SERS substrates used for detection still exist. Currently, SERS substrates used for fruit and vegetable safety detection can generally be divided into two categories: general-purpose SERS materials and specific SERS materials. General-purpose SERS materials (such as gold and silver nanoparticles and their different morphologies) have advantages such as simple preparation, high SERS activity, and significant enhancement effect, and are currently the most widely used substrate type. However, they have poor selectivity for homologues with similar structures and Raman activities, making it difficult to achieve simultaneous analysis of multiple harmful substances; the SERS signal of the fruit matrix itself is prone to overlap with the target spectrum, affecting the accuracy of quantification; and some pesticides lack structures that can bind or adsorb onto the substrate, making it difficult to directly identify them using general-purpose substrates. Specific SERS materials, by modifying Raman-labeled molecules and molecular recognition molecules (such as antibodies, aptamers, molecularly imprinted polymers, etc.) on the enhanced substrate, construct SERS probes, which can achieve specific capture and trace detection of target molecules, effectively overcome matrix interference, and significantly improve the selectivity and sensitivity of detection. Molecular recognition (such as antibodies and aptamers) can achieve target-specific capture and effectively overcome matrix interference, but it has limitations such as poor stability, high cost, and non-uniform recognition sites. Furthermore, characteristic peaks easily overlap with matrix spectra, requiring additional modification, which is detrimental to achieving the repeatability and stability requirements of quantitative detection.

[0004] To overcome the challenges of SERS detection, the "functionalized substrate enrichment" approach has gradually become a hot research topic for assisting SERS pesticide residue detection. Compared with traditional physical adsorption substrates, functionalized substrates can achieve selective enrichment of target pesticides through molecular recognition, effectively avoiding matrix interference and significantly improving detection sensitivity and selectivity. However, due to the limitations of the properties of functional materials, existing methods still suffer from problems such as heterogeneous recognition sites, complex synthesis processes, narrow target spectrum, and poor compatibility with SERS substrates.

[0005] Meanwhile, in SERS pesticide residue detection, although the development of novel SERS substrates has significantly improved detection performance and data repeatability, it has also led to a substantial increase in the amount of SERS spectral data, making data processing complex and time-consuming. Furthermore, the spectral overlap of different molecules further limits its practical application, generating a large amount of complex spectral data. Rapid and accurate analysis and processing of this data remains a major challenge for SERS technology.

[0006] Therefore, it is of great significance to provide a SERS substrate and detection method that can simultaneously identify multiple pesticide residues in complex matrices, with simple pretreatment and high sensitivity. Summary of the Invention

[0007] In view of this, the purpose of this invention is to provide a cobalt complex, a composite SERS substrate and its application in the detection of pesticide residues in fruits and vegetables, which can simultaneously achieve accurate identification and qualitative analysis of three pesticide residues on the surface of fruits and vegetables: thiram, imazalil and imidazole, and the detection pretreatment is simple and the detection sensitivity is high.

[0008] To achieve the above-mentioned objectives, the present invention provides the following technical solution: This invention provides a cobalt complex, wherein the cobalt complex is a 2-hydroxy-1-naphthaldehyde ethylenediamine cobalt complex, and its structural formula is shown in Formula I: Formula I.

[0009] The present invention provides a composite SERS substrate comprising silver nanoparticles loaded with the cobalt complex.

[0010] This invention provides a method for preparing the composite SERS substrate, comprising the following steps: The cobalt complex was dissolved in anhydrous ethanol and then mixed with a nano-silver solution and incubated by shaking. The molar ratio of the cobalt complex to the nano-silver was 1:(8~12) during mixing. The nano-silver solution was prepared by reducing silver nitrate with sodium citrate.

[0011] Preferably, the incubation is performed at 180-220 rpm for 1-3 minutes.

[0012] This invention provides the application of any one of the following in the detection of pesticide residues in fruits and vegetables, wherein the pesticide includes one or more of imazalil, thiram, and thiram; (1) The cobalt complex; (2) The composite SERS substrate; (3) The composite SERS substrate prepared by the method described above.

[0013] This invention also provides a method for detecting pesticide residues in fruits and vegetables, comprising the following steps: Fruit and vegetable samples were extracted in acetonitrile solution to obtain the sample solution to be tested; The sample solution to be tested was mixed with the composite SERS substrate, and the mixture was dropped onto the silicon wafer for Raman spectroscopy detection. The Raman spectrum of the sample to be tested is compared with the Raman spectrum of the pesticide standard to determine whether the substances in the sample solution match the pesticide standard. The pesticides include one or more of imazalil, thiram, and thiram.

[0014] Preferably, the detection parameters for the Raman spectroscopy detection are: excitation source wavelength of 785 nm, laser power of 100 mW, integration time of 10 s, and scanning wavenumber range of 300~2000 cm⁻¹. -1 .

[0015] Preferably, the comparison includes comparing the characteristic peaks of the Raman spectrum of the test sample with those of the pesticide standard; if the characteristic peaks are consistent, the substance in the test sample solution is determined to match the pesticide standard; the characteristic peak of the imazalil is 672 cm⁻¹. -1 799cm -1 878cm -1 920cm -1 1172cm -1 1339cm -1 1483cm -1 The characteristic peak of the imidazolium is 624 cm⁻¹. -1 878cm -1 1044cm -1 1086cm -1 The characteristic peak of the thiamethoxam is 378 cm⁻¹. -1 555cm -1 912cm -1 and 1380cm -1 .

[0016] Preferably, the comparison further includes comparing the Raman spectrum of the sample to be tested with the Raman spectrum of the pesticide standard using a support vector machine. The comparison method includes the following steps: Raman spectra of pesticide standards and Raman spectra of the sample to be tested were extracted from the 400-1800 cm⁻¹ range. -1 The fingerprint region is preprocessed to obtain preprocessed full-spectrum data. The full-spectrum data of the pretreated pesticide standards were divided into training and testing sets in a 7:3 ratio. Establish a support vector machine model: using preprocessed full-spectrum data as input features and pesticide category as output label, a classification model is established using radial basis kernel function; optimize the model with training set and verify the model performance with test set. When the classification accuracy on test set is not less than 85%, the model is considered to have effective pesticide classification ability. The preprocessed full-spectrum data of the sample to be tested is input into the validated model, and the output category is used as the final judgment result.

[0017] Preferably, the fruit or vegetable includes lychee.

[0018] The beneficial effects of this invention are: Outstanding multi-component detection capability: This invention, through the rational design of Schiff base cobalt complexes with functionalizability and multidentate coordination, and further constructs a nano-silver cobalt complex composite SERS substrate, generates specific binding with different types of pesticide molecules through interactions such as hydrogen bonding and π-π stacking. This enables the simultaneous and accurate identification and qualitative analysis of three pesticide residues—thiram, imazalil, and imidazole—on the surface of fruits and vegetables. This overcomes the limitation of traditional SERS substrates, which can usually only detect a single pesticide, and greatly improves detection throughput and efficiency.

[0019] High detection sensitivity and significantly improved signal-to-noise ratio: In the Schiff base cobalt metal complex-nano silver bimetallic composite SERS substrate constructed in this invention, the Schiff base cobalt metal complex can regulate the electronic structure and local electromagnetic field distribution on the surface of nano-silver. While forming high-density "hot spots", it promotes the directional enrichment of pesticide molecules in the hot spot region through coordination, thereby simultaneously achieving electromagnetic enhancement and chemical enhancement effects, improving the overall SERS response intensity, and thus significantly enhancing the Raman signal of the target pesticide. This enables highly sensitive detection of trace pesticide residues. Compared with traditional single metal nanoparticle substrates, it has higher signal stability and lower detection limit, and can still effectively detect pesticide residues at 0.5 mg / L.

[0020] The invention features simplified pretreatment and high detection efficiency, making it suitable for rapid on-site screening. In the composite SERS substrate of this invention, the Schiff base structure provides multi-site weak interactions (hydrogen bonds, π-π stacking, and coordination), which can selectively capture target pesticides in complex fruit peel matrices, reduce background interference, and allow samples to obtain stable and reliable Raman signals without deep pretreatment. Therefore, using the composite SERS substrate of this invention for detection eliminates the need for complex extraction or column chromatography purification steps. Detection can be completed simply by treating the surface of fruits and vegetables, significantly shortening the detection cycle and improving detection efficiency, making it suitable for rapid screening and large-scale applications.

[0021] Furthermore, after obtaining the Raman spectra, this invention combines a nonlinear classification model—Support Vector Machine (SVM)—to perform feature extraction and pattern recognition on the enhanced SERS spectra. SVM achieved 100% precision and recall on the three pesticide samples, effectively eliminating confusion between similar categories. The macro-average F1 score (Macro-F1) was improved to 100% compared to 91.90% for Partial Least Squares Discriminant Analysis (PLS-DA), indicating that combining the functionalized SERS substrate with the SVM data analysis model not only enables rapid detection of pesticide residues but also effectively distinguishes multiple pesticides through intelligent spectral recognition, significantly improving the automation and reliability of the detection process. Attached Figure Description

[0022] Figure 1 The molecular structure diagram of the 2-hydroxy-1-naphthaldehyde ethylenediamine cobalt complex.

[0023] Figure 2 Single-crystal simulation (black) and powder PXRD (red) spectra of the 2-hydroxy-1-naphthaldehyde ethylenediamine cobalt complex.

[0024] Figure 3 The UV stability spectrum of the 2-hydroxy-1-naphthaldehyde ethylenediamine cobalt complex.

[0025] Figure 4 The image shows the UV spectrum of the composite SERS substrate with silver nanoparticles.

[0026] Figure 5 Infrared comparison of 2-hydroxy-1-naphthaldehyde ethylenediamine cobalt complex (left image) and composite SERS substrate (right image).

[0027] Figure 6 The images show the characteristic Raman spectra of three pesticide components (thiram, imidacloprid, and prochloraz) and their mixed systems (a mixture of thiram, imidacloprid, and prochloraz) on a composite SERS substrate.

[0028] Figure 7 The image shows a comparison of the confusion matrices between the PLS-DA model (left side) and the SVM model (right side) based on Raman spectroscopy in the identification of three pesticides (thiram, imazalil, and prochloraz). Detailed Implementation

[0029] This invention provides a cobalt complex, wherein the cobalt complex is a 2-hydroxy-1-naphthaldehyde ethylenediamine cobalt complex, and its structural formula is shown in Formula I: Formula I.

[0030] In this invention, the cobalt complex is preferably 2-hydroxy-1-naphthaldehyde ethylenediamine and C4H6CoO4. The product is obtained by mixing 4H₂O (cobalt acetate tetrahydrate) in DMF (N,N-dimethylformamide) solution and reacting at 75-85℃ (e.g., 75℃, 78℃, 80℃, 82℃, or 85℃) for 70-74 h (e.g., 70 h, 72 h, or 74 h). The mixture contains 2-hydroxy-1-naphthaldehyde ethylenediamine and C₄H₆CoO₄. The molar ratio of 4H2O is preferably 1:(1.5~2.5), more preferably 1:1.5, 1:2 or 1:2.5. The mixing is preferably of C4H6CoO4. A DMF solution containing 4H₂O is added dropwise to a DMF solution of 2-hydroxy-1-naphthaldehyde ethylenediamine. After the reaction is complete, the preferred steps also include filtration and washing with anhydrous ethanol to obtain the cobalt complex.

[0031] This invention relates to the C4H6CoO4 The sources of 4H2O and 2-hydroxy-1-naphthaldehyde ethylenediamine are not particularly limited, and conventional commercially available products in the art can be used. Preferably, 2-hydroxy-1-naphthaldehyde ethylenediamine can also be prepared by a self-made method. The present invention does not particularly limit the self-made method, and conventional preparation methods in the art can be used.

[0032] The present invention provides a composite SERS substrate comprising silver nanoparticles loaded with the cobalt complex.

[0033] This invention provides a method for preparing the composite SERS substrate, comprising the following steps: The cobalt complex is dissolved in anhydrous ethanol and then mixed with a nano-silver solution, followed by shaking and incubation. The molar ratio of the cobalt complex to the nano-silver during mixing is 1:(8~12), preferably 1:8, 1:9, 1:9.45, 1:10, 1:10.6, 1:11 or 1:12. The nano-silver solution is silver nitrate prepared by sodium citrate reduction.

[0034] In this invention, the concentration of the cobalt complex during mixing is preferably 1.0 × 10⁻⁶. -4 mol / L. The present invention does not specifically limit the specific steps for preparing silver nitrate using the sodium citrate reduction method; conventional preparation steps in the art can be used. In some embodiments, silver nitrate can be dissolved in ultrapure water, heated to boiling, and 1% sodium citrate solution can be added dropwise while stirring, maintaining boiling for 0.8~1.2h (e.g., 0.8h, 1h, or 1.2h) to obtain a nano-silver solution. The preferred mass ratio of silver nitrate to sodium citrate is 18:(15~25), more preferably 18:20. After adding the sodium citrate solution, the final concentration of silver nitrate is preferably 0.18 g / L.

[0035] In this invention, the incubation is preferably performed at 180-220 rpm for 1-3 minutes, more preferably at 180 rpm, 200 rpm, or 220 rpm for 1 minute, 2 minutes, or 3 minutes. The incubation is preferably carried out in the dark at room temperature, where room temperature refers to any temperature between 25°C and 35°C. Incubation allows the cobalt complex molecules to fully contact and adsorb onto the surface of the silver nanoparticles. After incubation, the process preferably includes adding anhydrous ethanol and centrifuging twice to remove unbound or weakly adsorbed cobalt complex molecules, resulting in a composite SERS substrate, which is a liquid.

[0036] The cobalt complex described in this invention can be functionalized and multidentately coordinated to construct a composite SERS substrate. This substrate can not only interact with different types of pesticide molecules through hydrogen bonding and π-π stacking to selectively capture target pesticides in complex fruit peel matrices and reduce background interference, but also enhance the overall SERS response intensity and strengthen the Raman signal of the target pesticide through electromagnetic and chemical enhancement effects. This enables simultaneous and accurate identification and qualitative analysis of three pesticide residues—thiram, imazalil, and imidazole—on the surface of fruits and vegetables. This overcomes the limitation of traditional SERS substrates, which typically only detect single pesticides, significantly improving detection throughput and efficiency while avoiding cumbersome pretreatment processes and increasing detection sensitivity.

[0037] This invention provides the application of any one of the following in the detection of pesticide residues in fruits and vegetables, wherein the pesticide includes one or more of imazalil, thiram, and thiram; (1) The cobalt complex; (2) The composite SERS substrate; (3) The composite SERS substrate prepared by the method described above.

[0038] This invention also provides a method for detecting pesticide residues in fruits and vegetables, comprising the following steps: Fruit and vegetable samples are extracted in acetonitrile solution. Preferably, the fruit and vegetable samples include fruit peels. The material-to-liquid ratio of the fruit and vegetable samples to the acetonitrile solution is preferably 1 mg: (1~2) mL, more preferably 1 mg: 1 mL. The extraction is preferably performed by shaking for 1~2 min, such as 1 min, 1.5 min or 2 min. The shaking speed is preferably 2000~3000 rpm, more preferably 3000 rpm. After extraction, the sample solution to be tested is obtained.

[0039] The sample solution to be tested is mixed with a composite SERS substrate. The preferred volume ratio of the sample solution to the composite SERS substrate is 1:(1~1.5), more preferably 1:1, 1:1.2, or 1:1.5. The resulting mixture is then dropped onto a silicon wafer for Raman spectroscopy detection. The preferred detection parameters for the Raman spectroscopy are: excitation wavelength of 785 nm, laser power of 100 mW, integration time of 10 s, and scanning wavenumber range of 300~2000 cm⁻¹. -1 ; The Raman spectrum of the sample to be tested is compared with the Raman spectrum of the pesticide standard to determine whether the substances in the sample solution match the pesticide standard. The pesticides include one or more of imazalil, thiram, and thiram.

[0040] The preferred comparison method includes comparing the characteristic peaks of the Raman spectrum of the sample to be tested with those of the Raman spectrum of the pesticide standard. If the characteristic peaks are consistent, it is determined that the substances in the sample solution match the pesticide standard. The characteristic peak of imazalil is preferably 672 cm⁻¹. -1 (Benzene ring bending vibration), 799cm -1 (CH out-of-plane curvature), 878cm -1 (CH out-of-plane bending), 920cm -1 (Ring stretching vibration), 1172cm -1 (CN stretching coupling CH bending), 1339cm -1 (CH2 shear vibration or CN extension), 1483cm -1 (Imidazole ring skeletal breathing vibration), the characteristic peak of the imidazolium is preferably 624 cm⁻¹. -1 (In-plane / out-plane deformation vibration of benzene ring), 878cm -1 (CH out-of-plane curvature), 1044cm -1 (CH in-plane curvature), 1086cm -1 (CO stretching or CN stretching), the characteristic peak of the thiram is preferably 378 cm⁻¹. -1 (CS stretching vibration), 555cm -1 (SS stretching vibration), 912cm -1 (CH bending or CN stretching) and 1380cm -1 (CH3 symmetric deformation coupled CN stretching).

[0041] Under 785 nm excitation conditions, all three pesticides yielded stable Raman enhanced signals with high signal-to-noise ratios. By comparing the characteristic peaks of prochloraz, imazalil, and thiram, the three pesticides could be effectively distinguished from a mixture of pesticide residues.

[0042] The comparison preferably further includes comparing the Raman spectrum of the sample to be tested with the Raman spectrum of the pesticide standard using a support vector machine. The comparison method preferably includes the following steps: Raman spectra of pesticide standards and Raman spectra of the sample to be tested were extracted from the 400-1800 cm⁻¹ range. -1 The fingerprint region is preprocessed to obtain preprocessed full-spectrum data. Preferred preprocessing includes Savitzky-Golay smoothing, baseline correction, and normalization. More preferably, it also includes first-order derivative processing to enhance peak resolution or eliminate linear baselines. Whether to add first-order derivative processing depends on data quality. This invention does not specifically limit the preprocessing steps; conventional preprocessing steps in the art can be used.

[0043] The full-spectrum data of the pretreated pesticide standards were divided into training and testing sets in a 7:3 ratio. Establish a support vector machine model: using preprocessed full-spectrum data as input features and pesticide category as output label, a classification model is established using radial basis kernel function; The model is optimized using the training set; the preferred optimization method includes using grid search combined with 5-fold cross-validation to optimize parameters C and γ, wherein the search range of C is 2. -5 ~2 15 The search range for γ is 2. -15 ~2 3 ; The model performance was validated using a test set. The optimal evaluation criteria for model performance were accuracy, sensitivity, specificity, F1 score, and confusion matrix. A model was considered to have effective pesticide classification ability when the classification accuracy on the test set was not lower than 85%. The preprocessed full-spectrum data of the sample to be tested is input into the validated model, and the output category is used as the final judgment result.

[0044] After obtaining the Raman spectrum, this invention further combines a nonlinear classification model—Support Vector Machine (SVM)—to extract features and recognize patterns in the enhanced SERS spectrum. SVM achieves 100% precision and recall across all three classes of samples, effectively eliminating confusion between similar categories. The macro-average F1 score (Macro-F1) is 100% compared to PLS-DA's 91.90%, demonstrating that combining a functionalized SERS substrate with the SVM data analysis model not only enables rapid detection of pesticide residues but also effectively distinguishes between multiple pesticides through intelligent spectral recognition, significantly improving the automation and reliability of the detection process.

[0045] In this invention, the preferred fruits and vegetables include lychee. The composite SERS substrate of this invention can accurately identify three pesticides, namely prochloraz, imazalil, and thiram, in the extract containing the lychee surface matrix, under the premise of simple extraction of lychee surface by acetonitrile. This achieves the effect of sensitive and convenient simultaneous detection of multiple pesticides.

[0046] The technical solutions provided by the present invention will be described in detail below with reference to the embodiments, but they should not be construed as limiting the scope of protection of the present invention.

[0047] Unless otherwise specified, the following embodiments are all conventional methods.

[0048] Unless otherwise specified, all materials and reagents used in the following examples are commercially available.

[0049] Example 1 Preparation and characterization of cobalt complexes 1. Synthesis of ligand H2L 10 mmol (1.2212 g) of 2-hydroxy-1-naphthaldehyde and 5 mmol (0.4506 g) of ethylenediamine were dissolved in 40 mL of anhydrous ethanol, refluxed at 60 °C for 5 h, cooled, filtered, and washed with ethanol to obtain a yellow powder solid, which was the ligand H2L (2-hydroxy-1-naphthaldehyde ethylenediamine), with a yield of 60%. The reaction formula is shown in Formula II below. Calcd. (%) for C 24 H 20 N2O2(368.15): C,78.24; H, 5.47; N, 7.60. Found (%): C, 77.59; H, 5.02; N, 7.53. FT-IR (KBr,cm -1 ): 3434 (s), 2925 (w), 1639 (s), 1543 (s), 1450 (m), 1259 (w), 1150 (m), 997 (w), 875 (w), 746 (w). 1H NMR (400 MHz, DMSO) δ 14.19 (d, 2H), 9.17 (d, 2H), 8.04 (d, 2H), 7.74 (d, 2H), 7.64 (d, 2H), 7.41 (d, 2H), 7.20 (t, 2H), 6.75 (d, 2H), 4.03 (d, 4H).

[0050] Formula II 2. Synthesis of 2-hydroxy-1-naphthaldehyde ethylenediamine cobalt complex [CoL] C4H6CoO4 0.2 mmol of 4H₂O was dissolved in 3 mL of DMF, and then added dropwise to 5 mL of DMF solution containing 0.1 mmol of ligand H₂L. After mixing thoroughly, the sample vial was placed in an oven and reacted at 80 °C for 72 h. After filtration and washing with anhydrous ethanol, yellow prismatic crystals were obtained, which were the 2-hydroxy-1-naphthaldehyde ethylenediamine cobalt complex [CoL] (hereinafter referred to as the cobalt complex), with a yield of 51%. Calcd. (%) for C 24 H 18 CoN2O2(425.07): C, 67.77; H, 4.27; N, 6.59. Found(%): C, 67.72; H, 4.26; N, 6.58. FT-IR (KBr, cm -1 ): 3430 (s), 2924 (w), 1608 (s), 1530 (m), 1430 (w), 1355 (w), 1192 (w), 1096 (w), 828 (m), 738 (m).

[0051] Cobalt complex crystals were tested on an Agilent Technologies single-crystal diffractometer using MoKα radiation with the conditions λ = 0.71073 Å and T = 293(2) K. The structure was obtained by direct method followed by mixed hydrogenation, where anisotropic thermal parameters were used for non-hydrogen atoms and isotropic thermal parameters were used for hydrogen atoms. The X-ray diffraction (PXRD) pattern of the cobalt complex crystals was determined on a Rigaku D / max-2550 diffractometer.

[0052] The obtained crystallographic data are shown in Table 1, and the molecular structure diagram of the cobalt complex is shown in Table 2. Figure 1 Some bond lengths and bond angles are listed in Table 2. The results indicate that the structure of the complex belongs to the monoclinic crystal system, with space group P2. 1 / c The central metal Co1 is coordinated with two oxygen atoms (O1, O2) and two nitrogen atoms (N1, N2) in the ligand, respectively, with a coordination number of four, forming a twisted planar quadrilateral structure of CoN2O2. The Co-O bond lengths are 1.844(3) Å and 1.875(3) Å, respectively, and the Co-N bond lengths are 1.880(3) Å and 1.842(3) Å, respectively.

[0053] Powder X-ray diffraction (PXDR) was performed on the cobalt complex, and the diffraction data obtained were compared with those obtained by simulation using crystal data. The two methods showed good agreement. Figure 2 As shown in the figure, the synthesized cobalt complex has good phase purity.

[0054] Table 1 Crystallographic data and structural refinement parameters of cobalt complexes

[0055] Table 2. Partial bond lengths [Å] and bond angles [°] of the complexes.

[0056] Example 2 Stability study of the cobalt complex stock solution prepared in Example 1 Method: Preparation of phosphate buffer solution (PB): Dissolve Na₂HPO₄ and NaH₂PO₄ separately in ultrapure water (resistivity ≥ 18.2 MΩ·cm), and then prepare a solution with a concentration of 0.01 mol·L⁻¹. -1 PB (pH=7.0).

[0057] The cobalt complex obtained in Example 1 was dissolved in dimethyl sulfoxide (DMSO) and prepared to a concentration of 1×10⁻⁶. -3 mol·L -1 A stock solution of the cobalt complex was prepared. To ensure the stability of the prepared solution during the experiment, the stability of the complex was characterized by UV-Vis absorption spectroscopy. First, 30 μL of the cobalt complex stock solution was added to 5 mL of PB buffer, mixed thoroughly, and the UV-Vis absorbance was measured at room temperature at 0 h, 24 h, and 48 h.

[0058] Results: The UV spectrum of the cobalt complex in solution is shown in the figure. Figure 3 From the approximate peak shapes in the ultraviolet spectrum, it can be observed that the peaks of the complex did not exhibit a blue shift or a red shift (the ultraviolet spectra at the three time points overlapped). Furthermore, no new absorption peaks appeared in the test graphs, indicating that the cobalt complex DMSO solution of this invention is stable in PB at room temperature, laying the foundation for subsequent experiments.

[0059] Example 3 Preparation of nano-silver-supported cobalt complex composite SERS substrate Preparation of nano-silver solution: The sodium citrate reduction method was used. 18 mg of AgNO3 was accurately weighed and dissolved in ultrapure water, then diluted to a 100 mL volumetric flask. The solution was transferred to a 250 mL round-bottom flask and heated to boiling under reflux and stirring. 2 mL of 1% sodium citrate solution was added dropwise while stirring, maintaining boiling for 1 hour. The solution gradually turned grayish-green, yielding the AgNPs solution (nano-silver solution). After natural cooling, it was stored at 4°C in the dark for later use.

[0060] The synthesized 2-hydroxy-1-naphthaldehyde ethylenediamine cobalt complex (the cobalt complex prepared in Example 1) was dissolved in anhydrous ethanol to prepare a solution of 1.0 × 10⁻⁶ ppm. -3 Prepare a stock solution of cobalt complex at mol / L. Take 1 mL of the prepared silver nanoparticle solution into a centrifuge tube, add 100 μL of the cobalt complex stock solution, and bring the final concentration of the cobalt complex to 1.0 × 10⁻⁶ mol / L. -4 A solution of mol / L was placed on a small vortex mixer and incubated at 200 rpm in the dark for 2 min at room temperature to ensure that the cobalt complex molecules fully contacted and adsorbed onto the surface of the silver nanoparticles. After incubation, anhydrous ethanol was added and the mixture was centrifuged twice to remove unbound or weakly adsorbed cobalt complex molecules, resulting in a silver nanoparticle-supported cobalt complex composite SERS substrate (liquid state, hereinafter referred to as the composite SERS substrate). This substrate should be prepared and used immediately and should not be stored for a long period.

[0061] To verify the construction effect of the composite SERS substrate, UV-Vis was used to characterize the silver nanoparticles (AgNPs) and the composite SERS substrate (AgNPs@Complex), and FTIR was used to characterize the cobalt complex (cpmplex) and the composite SERS substrate (AgNPs@Complex). UV-Vis results ( Figure 4 The results showed that the surface plasmon resonance peak of the functionalized silver nanoparticles (AgNPs@Complex) exhibited a redshift and broadening, indicating that the cobalt complex successfully adsorbed and altered the local chemical environment of the silver nanoparticle surface. FTIR analysis ( Figure 5 Further confirmation revealed that the characteristic vibrational peaks of the cobalt complex were retained in the composite SERS substrate, with a slight shift in the C=N stretching vibration peak, indicating an interaction between the imine groups in the cobalt complex ligands and the surface of the silver nanoparticles. These results collectively demonstrate the successful construction of the silver nanoparticle-supported cobalt complex composite SERS substrate.

[0062] Example 4 Preparation of SERS substrates functionalized with silver nanoparticles and cobalt complexes Preparation of nano-silver solution: The sodium citrate reduction method was used. 9 × 10⁻⁶ AgNO₃ was weighed... -5 Dissolve 1 mol of AgNPs in ultrapure water and dilute to 100 mL in a volumetric flask. Transfer the solution to a 250 mL round-bottom flask and heat to boiling under reflux and stirring. Add 1.5 mL of 1% sodium citrate solution dropwise while stirring, and maintain boiling for 0.8 h. The solution gradually turns grayish-green, yielding the AgNPs solution (nano silver solution). After natural cooling, store in a 4°C refrigerator protected from light for later use.

[0063] The synthesized 2-hydroxy-1-naphthaldehyde ethylenediamine cobalt complex (the cobalt complex prepared in Example 1) was dissolved in anhydrous ethanol to prepare a solution of 1.0 × 10⁻⁶ ppm. -3Prepare a stock solution of cobalt complex at mol / L. Take 1 mL of the prepared silver nanoparticle solution into a centrifuge tube, add 100 μL of the cobalt complex stock solution, and bring the final concentration of the cobalt complex to 1.0 × 10⁻⁶ mol / L. -4 The solution was incubated at 180 rpm for 1 min in a small vortex mixer at room temperature in the dark to allow the cobalt complex molecules to fully contact and adsorb onto the surface of the silver nanoparticles. After incubation, anhydrous ethanol was added and the mixture was centrifuged twice to remove unbound or weakly adsorbed cobalt complex molecules, resulting in a composite SERS substrate (liquid state, hereinafter referred to as composite SERS substrate) supported on silver nanoparticles.

[0064] Example 5 Preparation of SERS substrates functionalized with silver nanoparticles and cobalt complexes Preparation of nano-silver solution: The sodium citrate reduction method was used. 12 × 10⁻⁶ AgNO₃ was weighed... -5 Dissolve 1 mol of AgNPs in ultrapure water and dilute to 100 mL in a volumetric flask. Transfer the solution to a 250 mL round-bottom flask and heat to boiling under reflux and stirring. Add 2.5 mL of 1% sodium citrate solution dropwise while stirring, and maintain boiling for 1.2 h. The solution gradually turns grayish-green, yielding the AgNPs solution (nano silver solution). After natural cooling, store in a 4°C refrigerator protected from light for later use.

[0065] The synthesized 2-hydroxy-1-naphthaldehyde ethylenediamine cobalt complex (the cobalt complex prepared in Example 1) was dissolved in anhydrous ethanol to prepare a solution of 1.0 × 10⁻⁶ ppm. -3 Prepare a stock solution of cobalt complex at mol / L. Take 1 mL of the prepared silver nanoparticle solution into a centrifuge tube, add 100 μL of the cobalt complex stock solution, and bring the final concentration of the cobalt complex to 1.0 × 10⁻⁶ mol / L. -4 The solution was incubated at 220 rpm for 3 min at room temperature in the dark using a small vortex shaker to ensure that the cobalt complex molecules were fully contacted and adsorbed onto the surface of the silver nanoparticles. After incubation, anhydrous ethanol was added and the mixture was centrifuged twice to remove unbound or weakly adsorbed cobalt complex molecules, resulting in a composite SERS substrate (liquid state, hereinafter referred to as composite SERS substrate) supported on silver nanoparticles.

[0066] Example 6 SERS detection and verification of composite SERS substrate 1. Detection of pesticide residues in litchi samples (1) Preparation of standard samples Weigh 10 mg of imazalil, thiram and thiram solid standards respectively, place them in 10 mL volumetric flasks, and prepare solutions with a mass concentration of 1000 mg·L⁻¹ using methanol as solvent. -1Standard stock solutions of prochloraz, imazalil, and thiram were prepared, and the concentrations of prochloraz, imazalil, and thiram were all 1000 mg·L⁻¹. -1 A standard stock solution of a mixed pesticide (prochloraz, imazalil, and thiram) was prepared. The obtained pesticide standard stock solutions were diluted to prepare 0.5, 5, and 10 mg / L solutions. -1 A standard working solution of pesticide at a certain concentration.

[0067] (2) Preparation and detection methods of actual samples Fresh lychee samples purchased from supermarkets were selected. They were first soaked in deionized water for 1.5 hours to remove surface impurities, then rinsed thoroughly with clean water and air-dried at room temperature. The lychees were grouped (using imazalil at concentrations of 0.5, 5, and 10 mg / L). -1 Groups, imazalil 0.5, 5, 10 mg / L -1 Groups, 0.5, 5, and 10 mg / L of thiram -1 Groups, mixed pesticides at doses of 0.5, 5, and 10 mg / L -1 The litchi was divided into groups, each sprayed with a standard working solution of the corresponding type and concentration of pesticide. The pesticide standard working solution was evenly sprayed onto the surface of the litchi and allowed to dry naturally at room temperature to allow the pesticide to be fully adsorbed and stabilized on the surface of the peel. The resulting sample of the peel was used as the spiked litchi sample for actual testing.

[0068] Before testing, 5 mg of spiked lychee sample was placed in a clean container, and 5 mL of acetonitrile was added as the extraction solvent. The mixture was shaken at 3000 rpm for 1 min to wash away pesticide residues on the surface. The resulting extract was used as the sample solution to be tested.

[0069] 20 μL of the sample solution was added dropwise to 20 μL of the composite SERS substrate and mixed. The mixture was then dropped onto a silicon wafer for Raman spectroscopy detection at room temperature. The detection parameters were: 785 nm excitation source, 100 mW laser power, 10 s integration time, and a scan wavenumber range of 300–2000 cm⁻¹. -1 The spectra were acquired at room temperature using a micro Raman spectrometer (ATR8300 fully automatic focusing laser micro Raman spectroscopy scanning imager from Aopu Tiancheng Company). Five spectra were acquired each time, and the average value of the five spectra was taken as the original spectrum of the sample.

[0070] (3) Results and Discussion Using the composite SERS substrate constructed in Example 3, standard working solutions of three pesticides—thiram, imazalil, and prochloraz—and a mixture of the three pesticides were detected. The results showed that, under 785 nm excitation conditions, stable Raman enhanced signals with high signal-to-noise ratios were obtained for all three pesticides at different spiking concentrations. The spiking concentration for each pesticide was 10 mg·L⁻¹. -1Raman spectra of litchi samples are as follows Figure 6 As shown in the figure, the composite SERS substrate of the present invention can effectively distinguish the three pesticide residue mixtures. Furthermore, in the SERS detection, the detection limits for the standard solutions of thiram, imidacloprid, and acetamiprid were all 0.5 mg / L, lower than the national maximum residue limit of 5 mg / L for pesticides on litchi.

[0071] On the constructed composite SERS substrate, thiram, imazalil, and prochloraz all exhibited stable Raman enhanced signals with fingerprint characteristics. Under the condition of multi-component coexistence, the characteristic peaks still maintained good separation, indicating that the substrate can achieve simultaneous recognition of multiple pesticide molecules. The chemical properties of the cobalt complex enable it to interact chemically with functional groups (such as amino, imidazolium, and dithiocarbamate groups) in pesticide molecules through aromatic rings and nitrogen-oxygen coordination sites.

[0072] Specific interactions not only enhance the selective adsorption of pesticide molecules but also improve their enrichment capacity in SERS hotspot regions, thus significantly increasing signal intensity and giving them a clear advantage in the detection of low-concentration pesticides. In particular, pesticides such as imidacloprid and thiram contain nitrogen groups that can coordinate with cobalt ions, while thiram contains sulfur-rich dithiocarbamate groups. These factors enable cobalt complexes to generate strong adsorption and interactions with these pesticide molecules. Furthermore, the cobalt ion center itself has a strong electron-donating ability, enabling it to interact with pesticide molecules through electron transfer effects. This not only enhances the sensitivity of the SERS signal but also makes SERS detection more sensitive by altering the electronic structure of pesticide molecules.

[0073] Example 7 Combined SERS detection and machine learning methods for pesticide residue detection This invention further combines machine learning methods to model and analyze the full-spectrum Raman data (raw Raman spectra) collected in Example 6, in order to effectively distinguish different pesticide categories. Specifically, two classification algorithms, Partial Least Squares Discriminant Analysis (PLS-DA) and Support Vector Machine (SVM), are used to construct the discriminant model.

[0074] First, extract the original Raman spectrum from 400 to 1800 cm⁻¹. -1 In the fingerprint region, the extracted spectrum is preprocessed, including Savitzky-Golay smoothing, baseline correction, normalization, and optional first-order derivative processing (selected depending on the data), resulting in preprocessed full-spectrum data. The preprocessed full-spectrum data is then divided into training and testing sets in a 7:3 ratio, and model optimization is performed using 5-fold cross-validation on the training set.

[0075] For the PLS-DA model, the preprocessed full-spectrum intensity matrix is ​​used as the independent variable X, and the pesticide category label is used as the dependent variable Y. The search range for the number of latent variables is set to 1~10. The discriminant model is established by selecting the number of latent variables corresponding to the lowest classification error rate through cross-validation. For the test set samples, the category corresponding to the maximum predicted value is used as the judgment result.

[0076] For the SVM model, preprocessed full-spectrum data is used as input features, and pesticide category is used as output label. A radial basis function kernel is used to establish the classification model, and grid search combined with 5-fold cross-validation is used to optimize parameters C and γ, where the search range of C is 2. -5 ~2 15 The search range for γ is 2. -15 ~2 3 The test set is used to validate the performance of the optimized model and output the class classification results.

[0077] Model performance was evaluated using accuracy, sensitivity, specificity, F1 score, and confusion matrix; a model was considered to have effective pesticide classification ability when the classification accuracy on the test set was not less than 85%.

[0078] Comparing the performance of the two models reveals that the SVM model exhibits smaller errors in cross-validation and requires fewer feature variables; it achieves 100% classification accuracy on both the training and test sets, while the PLS-DA model fails to achieve perfect classification. Figure 7 As shown, PLS-DA's recall rates for imidacloprid and imidacloprid were 87.64% and 87.91%, respectively, indicating a certain degree of false negatives. Furthermore, its precision was below 90% for both classes, suggesting false positives in these two categories. In contrast, SVM achieved 100% precision and recall across all three classes, effectively eliminating confusion between similar categories. The macro-F1 score improved from 91.90% for PLS-DA to 100% for SVM, further validating the significant advantage of the nonlinear classification model (SVM) in the Ben-Raman spectroscopy recognition task.

[0079] Another 120 litchi samples were prepared according to steps (1) and (2) of Example 6, and spiked litchi samples were used as test samples. The Raman spectra of the test samples were obtained, and the Raman spectra were intercepted from 400 to 1800 cm⁻¹. -1In the fingerprint region, the extracted spectrum is preprocessed, including Savitzky-Golay smoothing, baseline correction, normalization, and optional first-order derivative processing (selected depending on the data), resulting in preprocessed full-spectrum data. This preprocessed full-spectrum data is then input into the validated SVM model. The results show that the SVM achieves 100% precision and recall on all three classes of samples, consistent with the results on the test set.

[0080] The above results demonstrate that combining functionalized SERS substrates with SVM data analysis models can not only enable rapid detection of pesticide residues, but also effectively distinguish multiple pesticides through intelligent spectral recognition, significantly improving the automation level and reliability of detection, and providing a complete and efficient technical solution for food safety testing.

[0081] As can be seen from the above embodiments, the composite SERS substrate constructed by combining 2-hydroxy-1-naphthaldehyde ethylenediamine cobalt complex and metal nanoparticles in this invention, relying on its excellent molecular enrichment ability, enhanced surface reactivity, and effective charge transfer effect, can significantly improve the Raman signal intensity and signal-to-noise ratio of pesticide molecules, thereby improving the sensitivity, selectivity, and accuracy of detection. Especially under complex matrix conditions, this composite SERS substrate can significantly enhance the characteristic peak responses of different pesticide molecules, providing high-quality spectral data for subsequent accurate identification. On this basis, combining a nonlinear classification model (SVM) for feature extraction and pattern recognition of the enhanced SERS spectrum can further overcome the difficulty of distinguishing multiple pesticide spectra with high overlap, achieving high-precision differentiation and accurate quantitative analysis of different types of pesticide residues. This method has both a solid theoretical foundation and good experimental feasibility, and demonstrates good stability and scalability in practical applications, making it suitable for rapid detection and intelligent screening of multiple pesticide residues in fields such as food safety and environmental monitoring.

[0082] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A cobalt complex, characterized in that, The cobalt complex is a 2-hydroxy-1-naphthaldehyde ethylenediamine cobalt complex, the structural formula of which is shown in Formula I: Formula I.

2. A composite SERS substrate, characterized in that, Including silver nanoparticles loaded with the cobalt complex of claim 1.

3. The method for preparing the composite SERS substrate according to claim 2, characterized in that, Includes the following steps: The cobalt complex was dissolved in anhydrous ethanol and then mixed with a nano-silver solution and incubated by shaking. The molar ratio of the cobalt complex to the nano-silver was 1:(8~12) during mixing. The nano-silver solution was prepared by reducing silver nitrate with sodium citrate.

4. The preparation method according to claim 3, characterized in that, The incubation is performed at 180-220 rpm for 1-3 minutes.

5. The application of any one of the following in the detection of pesticide residues in fruits and vegetables, characterized in that: The pesticides include one or more of imazalil, thiram, and thiram; (1) The cobalt complex according to claim 1; (2) The composite SERS substrate of claim 2; (3) The composite SERS substrate prepared by the method of claim 3 or 4.

6. A method for detecting pesticide residues in fruits and vegetables, characterized in that, Includes the following steps: Fruit and vegetable samples were extracted in acetonitrile solution to obtain the sample solution to be tested; The sample solution to be tested is mixed with the composite SERS substrate described in claim 2, and the mixture is dropped onto a silicon wafer for Raman spectroscopy detection. The Raman spectrum of the sample to be tested is compared with the Raman spectrum of the pesticide standard to determine whether the substances in the sample solution match the pesticide standard. The pesticides include one or more of imazalil, thiram, and thiram.

7. The method according to claim 6, characterized in that, The detection parameters for the Raman spectroscopy detection are as follows: excitation source wavelength of 785 nm, laser power of 100 mW, integration time of 10 s, and scanning wavenumber range of 300~2000 cm⁻¹. -1 .

8. The method according to claim 6, characterized in that, The comparison includes comparing the characteristic peaks of the Raman spectrum of the test sample with those of the pesticide standard. If the characteristic peaks match, the substance in the test sample solution is considered to be compatible with the pesticide standard. The characteristic peak of the imazalil is 672 cm⁻¹. -1 799cm -1 878cm -1 920cm -1 1172cm -1 1339cm -1 1483cm -1 The characteristic peak of the imidazolium is 624 cm⁻¹. -1 878cm -1 1044cm -1 1086cm -1 The characteristic peak of the thiamethoxam is 378 cm⁻¹. -1 555cm -1 912cm -1 and 1380cm -1 .

9. The method according to claim 6, characterized in that, The comparison also includes comparing the Raman spectrum of the sample to be tested with the Raman spectrum of the pesticide standard using a support vector machine. The comparison method includes the following steps: Raman spectra of pesticide standards and Raman spectra of the sample to be tested were extracted from the 400-1800 cm⁻¹ range. -1 The fingerprint region is preprocessed to obtain preprocessed full-spectrum data. The full-spectrum data of the pretreated pesticide standards were divided into training and testing sets in a 7:3 ratio. Establish a support vector machine model: using preprocessed full-spectrum data as input features and pesticide category as output label, a classification model is established using radial basis kernel function; optimize the model with training set and verify the model performance with test set. When the classification accuracy on test set is not less than 85%, the model is considered to have effective pesticide classification ability. The preprocessed full-spectrum data of the sample to be tested is input into the validated model, and the output category is used as the final judgment result.

10. The application according to claim 5 or the method according to any one of claims 6 to 9, characterized in that, The fruits and vegetables mentioned include lychees.

Images