Bi2Se3 / Ti3C2Tx sweat heavy metal electrochemical sensor and preparation method thereof

Abstract

The invention relates to the technical field of electrochemical sensing, and discloses a Bi2Se3 / Ti3C2Tx sweat heavy metal electrochemical sensor and a preparation method thereof, and the preparation method comprises the following steps: preparing a Bi2Se3 / Ti3C2Tx composite material from a Ti3C2Tx nanosheet and a Bi2Se3 nanoflower material through an electrostatic self-assembly method, and modifying the Bi2Se3 / Ti3C2Tx composite material on the surface of a working electrode platform by adopting a micro dispensing method to prepare the lead-cadmium ion electrochemical sensor. Connecting the lead-cadmium ion electrochemical sensor into a circuit, performing electrochemical quantitative determination by adopting differential pulse voltammetry, and calculating the contents of Pb < 2 + > and Cd < 2 + > in sweat according to a peak current value; a machine learning method is used, a hybrid model fusing a least square lifting algorithm and a northern eagle optimization algorithm is adopted, and key experimental parameters of the sensor for detecting Pb < 2 + > and Cd < 2 + > are optimized.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of electrochemical sensing technology, in particular, to a Bi2Se3 / Ti3C2T x Electrochemical sensor for heavy metal ions in sweat and preparation method thereof. BACKGROUND

[0002] As a biological fluid, the concentration of heavy metal ions in sweat is proven to be significantly higher than that in blood and urine, especially the concentration of elements such as nickel (Ni), lead (Pb) and chromium (Cr) can reach several to dozens of times that of blood and urine. Therefore, monitoring heavy metal ions such as lead ions (Pb 2+ ) and cadmium ions (Cd 2+ ) in sweat has become an effective and non-invasive technical means to assess the level of human heavy metal exposure.

[0003] Electrochemical sensing technology, especially anodic stripping voltammetry (ASV), has attracted widespread attention in the field of heavy metal ion detection due to its convenience, low cost and potential for real-time monitoring. However, the concentration of Pb 2+ and Cd 2+ in sweat is extremely low, which poses extremely high requirements for the sensitivity, selectivity and anti-interference of the sensor.

[0004] In the prior art, the selection and construction of electrode materials are the core of determining the detection performance. Early studies used mercury electrodes, which were limited in application due to biological toxicity. Subsequent studies explored electrode materials such as gold and bismuth, which are more biocompatible. However, gold electrodes are high in cost, which is not conducive to commercialization. Sensors based on bismuth thin films have the problem of insufficient sensitivity when detecting trace amounts of Pb 2+ , Cd 2+ in actual sweat, making it difficult to observe obvious stripping peak signals.

[0005] In recent years, new nanomaterials such as graphene and MXene have been introduced to improve sensor performance. For example, existing technologies have developed graphene / polyaniline / polystyrene composite electrodes for detecting Pb 2+ and Cd 2+ ; another technology uses nitrogen-doped carbon-coated Ti3C2T x -MXene heterostructures to achieve detection of Pb 2+ and Cd 2+ by promoting interfacial charge transfer and enhancing metal ion adsorption. Although these materials show potential, they still generally face problems such as limited electrode active specific surface area, low interfacial mass transfer efficiency, background current interference, and complicated material synthesis and modification process when detecting trace heavy metal ions in complex sweat matrix, which restricts further improvement of detection sensitivity, stability and accuracy.

[0006] In addition, the electrochemical detection process involves multiple key parameters such as material ratio, immobilization process, deposition potential and deposition time, and there is a complex nonlinear interaction between these parameters. The traditional single-variable optimization strategy is difficult to efficiently and accurately determine the global optimal detection condition, resulting in a long sensor development cycle, high cost, and difficulty in ensuring its performance to be optimal. At the same time, the signal output by the electrochemical sensor in the detection process is easily disturbed by multiple coexisting substances in the sweat, making the signal background complex and difficult to analyze. SUMMARY

[0007] The application provides a Bi2Se3 / Ti3C2T x The sweat heavy metal electrochemical sensor and the preparation method thereof are prepared by combining Bi2Se3 nanoflower and Ti3C2T x MXene nanosheet to construct a synergistic sensitive interface, and a machine learning algorithm is introduced to globally optimize the detection conditions and signal analysis, so that the technical problems of low sensitivity, poor anti-interference ability and insufficient detection accuracy of the existing sweat heavy metal electrochemical sensor in the detection of trace Pb 2+ , Cd 2+ and other heavy metals due to insufficient active sites of electrode materials, low mass transfer efficiency, large background interference and experience-dependent optimization of detection parameters are solved.

[0008] According to one aspect of the application, a Bi2Se3 / Ti3C2T x The preparation method of the sweat heavy metal electrochemical sensor comprises the following steps: S100, Ti3C2T x nanosheet and Bi2Se3 nanoflower material are prepared into Bi2Se3 / Ti3C2T x composite material by an electrostatic self-assembly method; S200, the Bi2Se3 / Ti3C2T x composite material is modified on the surface of a working electrode platform to obtain a lead-cadmium ion electrochemical sensor; S200, the lead-cadmium ion electrochemical sensor is connected to a circuit, and differential pulse voltammetry is used for electrochemical quantitative determination, and the content of Pb 2+ and Cd 2+ in sweat is calculated according to the peak current value; S300, a hybrid model fused with a least square boosting algorithm and a northern hawk optimization algorithm is used to optimize the key experimental parameters of the sensor for detecting Pb 2+ and Cd 2+ .

[0009] Furthermore, in step S100, the working electrode platform adopts a 300-type screen-printed electrode, which is a three-electrode system consisting of a working electrode, a reference electrode, and a counter electrode; the working electrode and the counter electrode are composed of carbon paste, and the reference electrode uses a mixture of silver and silver chloride; the working electrode platform is directly connected to an electrochemical workstation for analysis.

[0010] Furthermore, in step S100, Ti3C2T x The preparation process is as follows: S111, 1000mg-2000mg of lithium fluoride is slowly added and dissolved in 20mL-40mL of 9M hydrochloric acid under stirring to obtain a LiF / HCl mixed solution; S112, 800mg-1300mg of titanium aluminum carbide MAX precursor is slowly poured into the LiF / HCl mixed solution, and the mixture is reacted at 35℃-45℃ with continuous stirring for 20h-30h to obtain an etching suspension; S113, after the reaction is complete, the etching suspension is washed repeatedly with deionized water at a speed of 4000rpm-6000rpm for 3min-8min; S114, after washing three times, the mixture is centrifuged at a speed of 9000rpm-12000rpm until the pH of the supernatant reaches 6 to obtain the sample; S115, the sample is placed in... After pretreatment in a freeze dryer at 80℃ for 1.5-2.5 hours, the product is freeze-dried for 20-30 hours to obtain powdered Ti3C2T. x -MXene products were collected and stored in a vacuum drying oven for later use.

[0011] Further, the preparation of Bi₂Se₃ in step S100 is specifically as follows: S121, dissolve 1000mg-2500mg of sodium hydroxide in 50mL-90mL of high-purity water, and magnetically stir for 20min-35min at 60℃-80℃ and 400r / min-550r / min; S122, add 500mg-750mg of selenium dioxide and 1000mg-2500mg of bismuth nitrate pentahydrate to the solution sequentially, and continue magnetic stirring for 20min-35min at 60℃-80℃ and 400r / min-550r / min; S123, add 2000mg-2500mg of gallic acid monohydrate and 4000mg-6000mg of Triton to the mixed solution, and stir at 60℃-80℃. Stir magnetically until homogeneous at 0℃ and 400-550 rpm; S124. Transfer the solution to a high-pressure reactor and react at 150-250℃ for 20-30 hours; S125. After the reaction is complete, cool the reactor to room temperature, transfer the solution to centrifuge tubes, and centrifuge at 7000-9000 rpm for 4-8 minutes; S126. Filter off the supernatant, and then centrifuge using high-purity water and anhydrous ethanol in the same manner as in step S125, washing each sample 3 times to obtain Bi2Se3 nanomaterials; S127. Dissolve the Bi2Se3 nanomaterials in 100-200 mL of ultrapure water and sonicate for 0.8-1.5 hours to obtain a Bi2Se3 solution for later use.

[0012] Further, take multiple portions of the Bi2Se3 solution obtained in step S127, each 2 mL, and vacuum dry at 50℃-70℃ for 40h-55h; after complete drying, weigh to prepare Bi2Se3 solutions with a concentration of 4mg / mL-6mg / mL.

[0013] Furthermore, in step S100, Bi2Se3 / Ti3C2T x The composite material was prepared as follows: Six composite solutions were prepared with Bi₂Se₃:Mxene ratios of 1:1, 2:1, 3:1, 4:1, 5:1, and 6:1 by mass. Then, 0.2 mL of isopropanol was added to each of the six composite solutions at a ratio of 0.1 mL Nafion solution to 0.2 mL of isopropanol. The mixtures were ultrasonically mixed for 0.8-1.5 hours to obtain Bi₂Se₃ / Ti₃C₂T x Dispersion.

[0014] Further, the preparation of the lead-cadmium ion electrochemical sensor in step S100 specifically includes: S131, treating the bare screen-printed carbon electrode with O2 plasma at a power of 120W-180W for 25s-35s; S132, adding 2μL, 3μL, 4μL, 5μL, and 6μL of Bi2Se3 / Ti3C2T at a concentration of 5mg / mL respectively. x S133. Disperse the solution onto the working electrode; S134. Dry the electrode in an oven at 55℃-65℃ for 0.3h-1h, and finally store it at 3℃-6℃ for later use; S135. Prepare and store the MXene electrode and Bi2Se3 electrode as controls using the same method as steps S131-S133.

[0015] Further, step S200 specifically involves preparing Pb solutions with concentrations of 0 ppb, 5 ppb, 10 ppb, 20 ppb, 40 ppb, 60 ppb, 80 ppb, 100 ppb, and 150 ppb, respectively. 2+ Cd 2+ Artificial sweat with pH=5.5 was used, and the electrode was tested using differential pulse voltammetry, and the differential pulse voltammetry curve was recorded.

[0016] Further, step S300 specifically involves: constructing a model using four machine learning methods—LSBoos-NGO, Support Vector Machine (SVM), Random Forest (RF), and Decision Tree Regression (DTR)—and comparing it with the optimized experimental model using a single-variable multiple regression method; and analyzing the Ti3C2T complex. x The ratio of MXene to Bi2Se3, the volume of the modified electrode material, the voltage for depositing Pb and Cd ions, and the deposition time were optimized.

[0017] According to another aspect of the present invention, a Bi2Se3 / Ti3C2T is also provided. x The sweat heavy metal electrochemical sensor uses the aforementioned Bi2Se3 / Ti3C2T x An electrochemical sensor for heavy metals in sweat was prepared using a specific method.

[0018] The present invention has the following beneficial effects: 1. Significant improvement in detection sensitivity was achieved at the sensitive interface level: Bi2Se3 / Ti3C2T prepared by electrostatic self-assembly method x Composite materials are not simply a mixture of materials; Bi₂Se₃ nanoflowers possess a large specific surface area, providing abundant adsorption sites for metal ions; while two-dimensional Ti₃C₂T x Nanosheets possess excellent electrical conductivity, acting as efficient electron transport channels. When combined with nanosheets, a synergistic effect is formed at the interface; Ti3C2T xThis accelerated the electron transfer rate from Bi₂Se₃ to the electrode surface, while Bi₂Se₃ enhanced the system's affinity for Pb. 2+ and Cd 2+ The enrichment capacity of the materials enhances the electrochemical response signal, thereby solving the problem of insufficient sensitivity caused by insufficient active sites in the electrode material and low interfacial mass transfer efficiency.

[0019] 2. At the detection process level, enhanced anti-interference capability and accuracy are achieved: Differential pulse voltammetry is used for measurement, which has high sensitivity and resolution and can effectively suppress background current interference; The introduced machine learning hybrid model integrates least squares lifting and Northern Eagle optimization algorithms. Before detection, the model can intelligently and globally optimize key parameters such as deposition potential and deposition time to find the optimal detection conditions that are difficult to obtain by traditional "trial and error methods", maximizing the signal-to-noise ratio from the source; After detection, the powerful nonlinear fitting capability of the algorithm can accurately analyze the characteristic peak signal of the target ion from the complex voltammetric signal generated by the complex matrix of sweat. This intelligent optimization and analysis process systematically solves the problems of poor anti-interference capability and low detection accuracy caused by non-optimal parameter configuration and sweat matrix interference.

[0020] 3. In terms of the completeness of the technical solution, a closed-loop optimization system from materials to algorithms is formed: This method constructs a full-chain solution of "high-performance sensitive interface preparation - high-precision detection method application - intelligent parameter and signal optimization"; materials are the foundation of performance, detection methods are the means, and machine learning algorithms serve as the intelligent brain to ensure that the former two always work in the optimal state. This system-level design enables the sensor to maintain stable and reliable high-performance output when facing complex real samples such as sweat.

[0021] In addition to the objectives, features, and advantages described above, the present invention has other objectives, features, and advantages. The invention will now be described in further detail with reference to the figures. Attached Figure Description

[0022] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings: Figure 1 This is a Zeta potential diagram characterizing the surface electrical properties of a material according to a preferred embodiment of the present invention, wherein... Figure 1 (a) is the sample Ti3C2T x Zeta potential, Figure 1 (b) shows the Zeta potential of the Bi2Se3 sample; Figure 2These are scanning electron microscope (SEM) and energy dispersive spectroscopy (EDS) images of the microstructure and elemental composition of the material according to a preferred embodiment of the present invention. Figure 2 (a) is a SEM image of Bi2Se3. Figure 2 (b) is Bi2Se3 / Ti3C2T x SEM image; Figure 2 (c)- Figure 2 (g) is Bi2Se3 / Ti3C2T x EDS plot; Figure 3 These are X-ray diffraction (XRD) and X-ray photoelectron spectroscopy (XPS) patterns characterizing the crystal structure and surface chemical state of the material according to a preferred embodiment of the present invention, wherein... Figure 3 (a) is Bi2Se3 / Ti3C2T x XRD pattern of the material Figure 3 (b) is Bi2Se3 / Ti3C2T x XPS full spectrum of the material Figure 3 (c)- Figure 3 (f) represents Bi2Se3 / Ti3C2T x High-resolution XPS spectra of C 1s, Ti 2p, Bi 4f and Se3d in the middle; Figure 4 This is a graph illustrating the system evaluation of the electrochemical properties of the electrode interface and electron transport kinetics according to a preferred embodiment of the present invention, wherein... Figure 4 (a) bare SPCE, Bi2Se3 / SPCE, Ti3C2T x / SPCE、Bi2Se3 / Ti3C2T x EIS diagram of the / SPCE electrode and equivalent circuit diagram of electrode EIS. Figure 4 (b) is Bi2Se3 / Ti3C2T x / SPCE electrode in [Fe(CN)6] 3- / 4- CV plot in solution Figure 4 (c) Bare SPCE electrode in [Fe(CN)6] 3- / 4- A graph showing the relationship between the redox peak current and the square root of the scan rate in CV solution. Figure 5 This is a multi-index radar chart of a preferred embodiment of the present invention for system evaluation and intuitive comparison of the predictive performance of different machine learning models, wherein... Figure 5 (a) is a radar chart comparing different MAE models. Figure 5 (b) Comparison radar charts of different MAPE models Figure 5 (c) Radar chart comparing the MSE of different models. Figure 5 (d) is a radar chart comparing the RMSE of different models. Figure 5(e) represents different models R 2 Compare radar charts; Figure 6 This is a comprehensive analytical diagram of the performance of machine learning models in a preferred embodiment of the present invention, providing a complete evaluation, explanation, and comparison. Figure 6 (a) R for different algorithms 2 -MAE performance curve comparison chart. Figure 6 (b) is the SHAP analysis diagram. Figure 6 (c) is a feature importance plot for SHAP analysis. Figure 6 (d) shows the results of the 5-fold cross-validation. Figure 6 (e) Overall radar chart comparing the performance of different models; Figure 7 This is the core graph of the LSBoost-NGO machine learning model used in the evaluation of the preferred embodiment of the present invention, which shows the predictive performance and goodness of fit. Figure 7 (a) shows the training set results for the LSBoost-NGO model. Figure 7 (b) is the training set density map of the LSBoost-NGO model. Figure 7 (c) shows the test set results for the LSBoost-NGO model. Figure 7 (d) is the test set density map of the LSBoost-NGO model; Figure 8 This is a partial dependency graph of a preferred embodiment of the present invention. Figure 8 (a)- Figure 8 (d) represents Bi2Se3 and Ti3C2T in the LSBoost-NGO model, respectively. x Partial dependence of peak current on ratio, sediment volume, sediment potential and sedimentation time; Figure 9 This is a quantitative evaluation chart of sensor performance according to a preferred embodiment of the present invention, wherein... Figure 9 (a) Pb concentration range of 0-150 ppb 2+ and Cd 2+ The DPASV test results Figure 9 (b) and Figure 9 (c) are Cd respectively 2+ and Pb 2+ Linear regression plot of concentration and its respective peak current; Figure 10 This is an evaluation of Bi2Se3 / Ti3C2T according to a preferred embodiment of the present invention. x The key performance test diagrams for the reliability of basic electrochemical sensors in practical application scenarios are shown. Figure 10 (a) and Figure 10 (b) is Bi2Se3 / Ti3C2T x Results of interference immunity test of basic electrochemical sensor;Figure 10 (c) is Bi2Se3 / Ti3C2T x The results of a seven-day stability test of the basic electrochemical sensor are shown in the figure. Detailed Implementation

[0023] The embodiments of the present invention will be described in detail below with reference to the accompanying drawings. However, the present invention can be implemented in many different ways as defined and covered below. Unless otherwise stated, the raw materials and reagents used in the following embodiments are commercially available or can be prepared by known methods.

[0024] Bi2Se3 / Ti3C2T in this embodiment x The preparation method of a sweat heavy metal electrochemical sensor includes the following steps: S100, preparing Ti3C2T x Bi2Se3 / Ti3C2T nanosheets and Bi2Se3 nanoflowers were prepared by electrostatic self-assembly. x The composite material, Bi2Se3 / Ti3C2T, was prepared using a micro-droplet coating method. x A lead-cadmium ion electrochemical sensor was fabricated by modifying the surface of the working electrode platform with a composite material. The lead-cadmium ion electrochemical sensor was connected to a circuit, and electrochemical quantitative determination was performed using differential pulse voltammetry. The Pb concentration in sweat was calculated based on the peak current value. 2+ and Cd 2+ The content of Pb; S300, using machine learning methods, a hybrid model integrating the least squares lifting algorithm and the Northern Eagle optimization algorithm is used to optimize the sensor's response to Pb. 2+ and Cd 2+ Key experimental parameters for detection. This invention relates to Bi₂Se₃ / Ti₃C₂T. x A method for fabricating a sweat heavy metal electrochemical sensor, specifically a Bi2Se3 / Ti3C2T sensor prepared via electrostatic self-assembly. x Composite materials are not simply a mixture of materials; Bi₂Se₃ nanoflowers possess a large specific surface area, providing abundant adsorption sites for metal ions; while two-dimensional Ti₃C₂T x Nanosheets possess excellent electrical conductivity, acting as efficient electron transport channels. When combined with nanosheets, a synergistic effect is formed at the interface; Ti3C2T x This accelerated the electron transfer rate from Bi₂Se₃ to the electrode surface, while Bi₂Se₃ enhanced the system's affinity for Pb. 2+ and Cd 2+The enrichment capacity of the materials enhances the electrochemical response signal, thus solving the problem of insufficient sensitivity caused by insufficient active sites in the electrode material and low interfacial mass transfer efficiency. Differential pulse voltammetry is used for measurement, which has high sensitivity and resolution and can effectively suppress background current interference. The introduced machine learning hybrid model, integrating least squares lifting and Northern Eagle optimization algorithms, intelligently optimizes key parameters such as deposition potential and deposition time globally before detection, finding the optimal detection conditions that are difficult to obtain through traditional trial and error methods, maximizing the signal-to-noise ratio from the source. After detection, the powerful nonlinear fitting capability of the algorithm can accurately analyze the characteristic peak signal of the target ion from the complex voltammetric signal generated by the complex matrix of sweat. This intelligent optimization and analysis process systematically solves the problems of poor anti-interference ability and low detection accuracy caused by suboptimal parameter configuration and sweat matrix interference. This method constructs a complete chain solution encompassing "high-performance sensitive interface preparation—high-precision detection method application—intelligent parameter and signal optimization." Materials form the foundation of performance, detection methods are the means, and machine learning algorithms act as the intelligent brain, ensuring that the former two always operate at their optimal state. This system-level design enables the sensor to maintain stable, reliable, and high-performance output when dealing with complex real-world samples such as sweat. The preparation method of this invention constructs Bi2Se3 / Ti3C2T... x By incorporating a composite sensitive interface and introducing machine learning for intelligent optimization, a synergistic effect is achieved, ultimately systematically solving the problem of detecting trace Pb in environments with high background interference and low concentrations of sweat, as described in existing technologies. 2+ Cd 2+ The core technical problems of low sensitivity, poor anti-interference and insufficient accuracy faced by the past have provided an effective technical means to achieve accurate and reliable non-invasive health monitoring.

[0025] In this embodiment, the working electrode platform in step S100 uses a type 300 screen-printed electrode. The type 300 screen-printed electrode consists of a working electrode, a reference electrode, and a counter electrode, forming a three-electrode system. The working electrode and counter electrode are composed of carbon paste, and the reference electrode uses a mixture of silver and silver chloride. The working electrode platform is directly connected to an electrochemical workstation for analysis. The use of a type 300 screen-printed electrode integrating the working electrode, reference electrode, and counter electrode replaces the bulky three independent electrode system in traditional electrochemical analysis. This results in a compact sensor structure and significantly reduced size, which not only facilitates operation but, more importantly, lays the hardware foundation for the future development of miniaturized, portable, and even wearable detection devices, directly meeting the potential requirements of sweat detection for convenience and real-time performance. The three-electrode system design separates the current path between the working electrode and the counter electrode from the potential reference circuit between the working electrode and the reference electrode. The reference electrode, composed of silver and silver chloride, provides a highly stable potential reference, effectively avoiding potential drift caused by counter electrode polarization or solution impedance, thus ensuring precise and controllable potential applied to the working electrode. The carbon paste working electrode and counter electrode provide good conductivity. This stable system fundamentally guarantees the stability and reproducibility of the differential pulse voltammetry measurement signal, providing a reliable guarantee for subsequent accurate quantitative analysis. Screen-printed electrodes can be manufactured using large-scale, low-cost printing processes. Carbon paste and silver / silver chloride are mature and relatively low-cost electrode materials, making the fabricated sensor suitable for single use, avoiding cross-contamination, and cost-controllable. This perfectly meets the stringent requirements for device cost and economy in public health screening applications.

[0026] In this embodiment, Ti3C2T in step S100 x The preparation process is as follows: S111, 1000mg-2000mg of lithium fluoride is slowly added and dissolved in 20mL-40mL of 9M hydrochloric acid under stirring to obtain a LiF / HCl mixed solution; S112, 800mg-1300mg of titanium aluminum carbide MAX precursor is slowly poured into the LiF / HCl mixed solution, and the mixture is reacted at 35℃-45℃ with continuous stirring for 20h-30h to obtain an etching suspension; S113, after the reaction is complete, the etching suspension is washed repeatedly with deionized water at a speed of 4000rpm-6000rpm for 3min-8min; S114, after washing three times, the mixture is centrifuged at a speed of 9000rpm-12000rpm until the pH of the supernatant reaches 6 to obtain the sample; S115, the sample is placed in... After pretreatment in a freeze dryer at 80℃ for 1.5-2.5 hours, the product is freeze-dried for 20-30 hours to obtain powdered Ti3C2T. x-MXene products were collected and stored in a vacuum drying oven for later use. A mixed solution of lithium fluoride and hydrochloric acid was used to gently etch the titanium aluminum carbide MAX phase precursor. Hydrochloric acid dissolved the aluminum (Al) atomic layer, while the hydrofluoric acid (HF) generated in situ by lithium fluoride (LiF) selectively etched away the aluminum layer, while simultaneously releasing lithium ions (Li... + This allows it to embed between layers and form intercalations, effectively weakening the van der Waals forces between MXene layers. This synergistic etching-intercalation mechanism paves the way for subsequent centrifugal washing and freeze-drying to obtain structurally complete Ti3C2T with fewer layers. x The nanosheets laid the foundation, directly determining the structural prerequisite for the high specific surface area and abundant active sites of the final composite electrode material. The set centrifugation speed range (4000rpm-6000rpm for initial washing, 9000rpm-12000rpm for fine separation) and pH=6 as the washing endpoint systematically removed byproducts from the etching reaction and residual acidic electrolytes, avoiding the adverse effects of impurity ions on the electrochemical performance of MXene. Subsequent freeze-drying, through the solid-state sublimation of water, minimized the irreversible recombination and aggregation of MXene nanosheets caused by the surface tension of water during drying, thus better preserving their exfoliated state in solution and obtaining a fluffy, easily redispersible powder product. This provided structurally intact MXene raw material for subsequent effective composite with Bi₂Se₃ nanoflowers. Detailed ranges were given for key parameters such as etching temperature (35-45℃), etching time (20-30h), centrifugation conditions, and freeze-drying conditions. These key variables affecting the morphology, number of layers, and surface functional groups of the final product were controlled within optimized ranges. This precise control avoided material property differences caused by process fluctuations, ensuring the quality of the prepared Ti3C2T. x -MXene products exhibit consistent electrochemical properties, thus ensuring the stability and reproducibility of the sensor performance constructed from them. In step S100, Ti3C2T... x A specific preparation method, through optimized etching-intercalation chemistry and refined post-processing, successfully yielded high-purity, low-layer, and structurally well-preserved MXene nanosheet powder, providing a foundation for the subsequent construction of high-performance Bi2Se3 / Ti3C2T nanosheets. x Composite sensitive interfaces provide material assurance, which in turn helps to solve the problem of insufficient sensor sensitivity.

[0027] In this embodiment, the preparation of Bi2Se3 in step S100 is specifically as follows: S121, dissolve 1000mg-2500mg of sodium hydroxide in 50mL-90mL of high-purity water, and magnetically stir for 20min-35min at 60℃-80℃ and 400r / min-550r / min; S122, add 500mg-750mg of selenium dioxide and 1000mg-2500mg of bismuth nitrate pentahydrate to the solution sequentially, and continue magnetic stirring for 20min-35min at 60℃-80℃ and 400r / min-550r / min; S123, add 2000mg-2500mg of gallic acid monohydrate and 4000mg-6000mg of Triton to the mixed solution, and stir at 60℃- Stir magnetically at 80℃ and 400-550 rpm until homogeneous; S124. Transfer the solution to a high-pressure reactor and react at 150-250℃ for 20-30 hours; S125. After the reaction is complete, cool the reactor to room temperature, transfer the solution to centrifuge tubes, and centrifuge at 7000-9000 rpm for 4-8 minutes; S126. Filter off the supernatant, and then centrifuge with high-purity water and anhydrous ethanol in the same manner as in step S125, washing each sample 3 times to obtain Bi2Se3 nanomaterials; S127. Dissolve the Bi2Se3 nanomaterials in 100-200 mL of ultrapure water and sonicate for 0.8-1.5 hours to obtain a Bi2Se3 solution for later use. After sequentially adding sodium hydroxide, selenium dioxide, and bismuth nitrate pentahydrate to form a precursor, gallic acid and Triton X-100 were introduced as composite structure directing agents. Prior to the hydrothermal reaction, multiple phenolic hydroxyl groups on the gallic acid molecule could react with Bi... 3+Ions coordinate to effectively control the nucleation and growth rate of Bi2Se3; while the nonionic surfactant Triton adsorbs onto the surface of the crystal nucleus through intermolecular forces, changing the surface energy of different crystal planes; in the subsequent high-temperature and high-pressure hydrothermal environment, these two directing agents work together to guide two-dimensional primary nanostructures (such as nanosheets) to bend and self-assemble around specific crystal nuclei, eventually forming flower-like Bi2Se3 nanoflowers composed of finer nanounits; compared with solid particles, this three-dimensional hierarchical structure exposes more active sites, greatly increasing the contact area with target metal ions, and providing a key morphological advantage for improving the enrichment efficiency and sensitivity of the sensor. Following the hydrothermal reaction, centrifugation within a set speed range and alternating washing with high-purity water and anhydrous ethanol effectively removed residual ions, unreacted precursors, and structure-directing agents, yielding pure Bi₂Se₃ nanomaterials. Anhydrous ethanol washing not only removed some organic impurities but also benefited subsequent drying steps due to its volatility. Finally, the obtained nanomaterials were redispersed in ultrapure water and subjected to ultrasonic treatment to break up the soft agglomerates of the nanoflowers, forming a relatively stable colloidal dispersion (Bi₂Se₃ solution). This ensured that the Bi₂Se₃ nanoflowers remained in a uniform, monodisperse state for use, preparing them for further processing with Ti₃C₂T₂. x The nanosheets were well-suited for effective electrostatic self-assembly, avoiding the problem of uneven composite interfaces caused by agglomeration. Specific regulations on key parameters such as hydrothermal temperature (150℃-250℃), time (20h-30h), and centrifugation conditions effectively controlled the key variables affecting the crystallinity, size, and morphology of the nanoflowers. This precise control ensured that Bi₂Se₃ nanoflowers prepared in different batches exhibited consistent morphology and electrochemical properties, providing a foundation for the stability and reproducibility of sensor performance. The specific preparation method for Bi₂Se₃ in step S100, through hydrothermal synthesis regulated by a structure-directing agent and systematic post-processing, successfully obtained high-purity, easily dispersed Bi₂Se₃ nanoflowers with a three-dimensional hierarchical porous structure. This morphology increased the specific surface area and the number of active sites, enabling efficient enrichment of trace heavy metal ions at the composite sensitive interface, thus helping to address the problem of insufficient sensor sensitivity.

[0028] In this embodiment, multiple portions of the Bi2Se3 solution obtained in step S127, each 2 mL, were taken and vacuum dried at 50℃-70℃ for 40-55 hours. After complete drying, the portions were weighed to prepare Bi2Se3 solutions with concentrations of 4 mg / mL-6 mg / mL. By taking a fixed volume (2 mL) of Bi2Se3 solution dispersion and vacuum drying, all solvents and any remaining volatile components were completely removed, resulting in pure, solvent-free Bi2Se3 solids. By accurately weighing this portion of solids, the actual mass concentration of Bi2Se3 in the original dispersion could be accurately calculated, providing mass data support for the subsequent preparation of Bi2Se3 solutions with known and precise concentrations (4 mg / mL-6 mg / mL). This solved the problem of uncontrolled feed ratios caused by unknown raw material concentrations during the material composite process. The hydrothermal synthesis and dispersion processes may exhibit batch-to-batch variations, leading to fluctuations in the actual solid content of the Bi₂Se₃ solution obtained in different batches in step S127. This step, as a crucial standardization and calibration step, eliminates such initial concentration fluctuations. Regardless of the concentration of the dispersion prepared in the preceding steps, the drying-weighing-re-volume adjustment process in this step ensures that a new Bi₂Se₃ stock solution with highly consistent and precisely controllable concentration can be obtained. This guarantees that the amount of Bi₂Se₃ added is precisely known during subsequent electrostatic self-assembly with MXene, thereby ensuring the uniformity of Bi₂Se₃ and Ti₃C₂T in the final composite material. x The mass ratios (e.g., 1:1 to 6:1) are accurate, reliable, and reproducible. This was achieved through optimization of the Bi₂Se₃ and Ti₃C₂T ratios. x To improve sensor performance, this step involves obtaining a Bi₂Se₃ solution with a defined concentration. This allows for compounding with MXene at specific mass ratios (e.g., 1:1, 2:1, etc.), providing a fundamental and quantifiable guarantee for optimizing the compounding ratio. The process of drying, weighing, and re-preparing the standard solution ensures precise calibration of the Bi₂Se₃ active material mass and standardization of the solution concentration. This pretreatment step ensures precise control of the subsequent material compounding ratio, reproducibility of the experimental process, and ultimately, performance optimization, guaranteeing the reliability and comparability of sensor performance from the outset.

[0029] In this embodiment, Bi2Se3 / Ti3C2T in step S100 x The composite material was prepared as follows: Six composite solutions were prepared with Bi₂Se₃:Mxene ratios of 1:1, 2:1, 3:1, 4:1, 5:1, and 6:1 by mass. Then, 0.2 mL of isopropanol was added to each of the six composite solutions at a ratio of 0.1 mL Nafion solution to 0.2 mL of isopropanol. The mixtures were ultrasonically mixed for 0.8-1.5 hours to obtain Bi₂Se₃ / Ti₃C₂T xDispersions. Six complexes of Bi2Se3 to MXene with mass ratios ranging from 1:1 to 6:1 were prepared to construct a systematic gradient experimental sequence; due to the interaction between Bi2Se3 nanoflowers (primarily providing heavy metal ion enrichment sites) and Ti3C2T... x Nanosheets (primarily providing high-speed electron conduction channels) play different roles and make different contributions in composite materials. There exists an optimal ratio between them, resulting in a best match between interfacial contact area, electron transport paths, and the number of active sites, thus producing the strongest synergistic enhancement effect. By systematically examining the performance of composites with different ratios, it is possible to scientifically and efficiently screen for Pb-resistant composites. 2+ Cd 2+ The optimal composite ratio with the highest sensitivity was determined to address the technical problem of suboptimal performance that may result from single materials or random composites. Nafion solution and isopropanol were added at a fixed ratio (Nafion solution: isopropanol = 0.1 mL: 0.2 mL) and subjected to ultrasonic treatment. The ultrasonic energy ensured that the two nanomaterials were fully dispersed in the liquid phase and uniformly mixed at the nanoscale, preventing agglomeration and facilitating the formation of a uniform composite interface. Nafion acted as a binder, firmly adhering the composite material to the electrode surface and helping to form a dense and stable sensitive film. Isopropanol adjusted the slurry viscosity, making it suitable for subsequent micro-drop coating processes. The final Bi2Se3 / Ti3C2T... x The dispersion is essentially a ready-to-use, performance-controllable electrode modification slurry. It incorporates two previously prepared basic nanomaterials (Bi₂Se₃ nanoflowers and Ti₃C₂T₅ nanomaterials). x The nanosheets are seamlessly integrated with the final electrode fabrication process, achieving not only simple physical mixing of materials but also, through optimized proportions and ultrasonic dispersion, nanoscale composite and interface engineering of functional components. This directly prepares the foundation for the next step of constructing a high-performance sensitive interface on the electrode using a micro-droplet coating method, and is a key step in realizing the improvement of sensor sensitivity and selectivity in this invention. In step S100, Bi2Se3 / Ti3C2T... x The preparation method of the composite material determined the optimal composite ratio through systematic ratio screening, and prepared a uniform and stable electrode-modified dispersion through a standardized slurry process. This directly determines the microstructure and composition of the composite sensitive interface, and is crucial for achieving the composite of Bi2Se3 and Ti3C2T. x The two complement each other's strengths and work synergistically to systematically solve the problem of insufficient sensor sensitivity.

[0030] In this embodiment, the preparation of the lead-cadmium ion electrochemical sensor in step S100 specifically includes: S131, treating the bare screen-printed carbon electrode with O2 plasma at a power of 120W-180W for 25s-35s; S132, adding 2μL, 3μL, 4μL, 5μL, and 6μL of Bi2Se3 / Ti3C2T at a concentration of 5mg / mL respectively. x The dispersion was applied to the working electrode; S133, the electrode was dried in an oven at 55℃-65℃ for 0.3h-1h, and finally stored at 3℃-6℃ for later use; S134, the MXene electrode and Bi2Se3 electrode used as controls were prepared and stored using the same method as steps S131-S133. The bare screen-printed carbon electrode was treated with O2 plasma at a specific power of 120W-180W and a treatment time of 25s-35s. Through the high-energy bombardment and chemical modification of the plasma, oxygen-containing polar functional groups (such as carboxyl and hydroxyl groups) were effectively introduced into the surface of the electrode carbon paste, thereby significantly improving the hydrophilicity of the electrode surface and increasing its surface energy, making the subsequent addition of Bi2Se3 / Ti3C2T... x Aqueous dispersions can better wet and spread on the electrode surface; the increase in surface functional groups provides more anchoring points for nanocomposite materials, and through stronger interfacial interactions (such as hydrogen bonds), ensures a firm bond between the sensitive membrane and the electrode substrate, effectively preventing membrane detachment during use and improving the mechanical stability and lifespan of the sensor. Five different volumes of composite dispersion, ranging from 2 μL to 6 μL, were added dropwise to systematically study the effect of sensitive membrane thickness (or loading) on ​​detection performance. Membrane thickness is related to the total number of active sites, the length of the electron transport path, and the mass transfer rate of reactants (metal ions) into the membrane. An excessively thin membrane may lead to insufficient active sites and a low signal; an excessively thick membrane may increase electron transport impedance and mass transfer resistance, thus reducing the response. By preparing sensors with different modification amounts and comparing their performance, the optimal modification volume that produces the strongest electrochemical response signal can be scientifically determined, thereby avoiding performance loss due to improper modification and ensuring that the sensor sensitivity reaches its optimal state. Pure MXene and pure Bi₂Se₃ electrodes were used as controls, and prepared and stored using the same methods as the composite electrode. Strict adherence to the scientific principles of the controlled variable method ensured that all electrodes differed only in the active material composition, while all other factors potentially affecting performance, such as the substrate, pretreatment process, modification solvent, and drying conditions, remained consistent. Therefore, the observed superior performance of the composite electrode compared to any single-material electrode can be uniquely and convincingly attributed to the combination of Bi₂Se₃ and Ti₃C₂T. xThe synergistic effect resulting from the combination of the two materials is not due to interference caused by different preparation conditions. The sensor preparation method in step S100 enhances the film-substrate bonding through plasma treatment, determines the optimal performance conditions through systematic optimization of the modification amount, and demonstrates the synergistic advantages of the composite material through rigorous control experiments. This series of operations together ensures that the prepared sensor has reliable reproducibility, optimized sensitivity, and verifiable performance improvement, laying the foundation for accurate and stable detection applications in the future.

[0031] In this embodiment, step S200 specifically involves preparing Pb solutions with concentrations of 0 ppb, 5 ppb, 10 ppb, 20 ppb, 40 ppb, 60 ppb, 80 ppb, 100 ppb, and 150 ppb, respectively. 2+ Cd 2+ Artificial sweat, pH 5.5, was used, and differential pulse voltammetry was performed on the fabricated electrode to record the differential pulse voltammetric curve. Multiple concentration gradients of Pb covering the range of 0 ppb to 150 ppb were prepared. 2+ Cd 2+ Standard solutions were used to obtain the electrochemical response signals (i.e., peak current values ​​on differential pulse voltammetry curves) corresponding to different concentrations of the target analyte. By recording this series of concentration-response data, Pb could be plotted. 2+ and Cd 2+ Each sensor has its own calibration curve (usually a fitted curve of peak current versus concentration). When detecting unknown samples, only the generated electrical signal needs to be measured to calculate the accurate concentration of the target analyte in the sample, thus solving the problem of quantitative analysis required by the sensor. Artificial sweat was used as a matrix to prepare the standard solution, and the pH was set to 5.5 to simulate the acid-base environment of real sweat. This ensured that the calibration process was performed under conditions highly relevant to practical applications, effectively assessing the potential interference from common electrolytes and organic matter in sweat. The fact that a clear and distinguishable response signal starting from a low concentration (5 ppb) was obtained even in such a complex matrix validated the prepared Bi₂Se₃ / Ti₃C₂T x When faced with actual samples, the sensor is capable of detecting trace amounts of Pb. 2+ Cd 2+The high sensitivity and good resistance to matrix interference demonstrate the feasibility of its application in actual sweat detection. Differential pulse voltammetry (DPV) is used for testing. By applying a stepped voltage superimposed pulse and measuring the current difference before and after the pulse, interference from background charging current can be suppressed. This efficient signal-to-noise ratio improvement makes the method particularly suitable for detecting the weak Faradaic current generated by trace heavy metal ions in sweat. This allows for clear differentiation of the characteristic dissolution peaks of the target analyte from complex background signals, enhancing the resolution and accuracy of the detection results and ensuring high sensitivity and high reliability. Step S200 involves preparing a series of standard solutions of artificial sweat matrix at various concentrations and testing them using differential pulse voltammetry to establish a calibration curve for quantitative analysis. The sensor's sensitivity to trace Pb is verified under near-realistic detection conditions. 2+ Cd 2+ Its high sensitivity and anti-interference capability provide a methodological basis and performance data support for transforming the sensor from a functional material into a reliable quantitative analysis tool.

[0032] In this embodiment, step S300 specifically involves: constructing a model using four machine learning methods—LSBoos-NGO, Support Vector Machine (SVM), Random Forest (RF), and Decision Tree Regression (DTR)—and comparing it with the optimized experimental model of single-variable multiple regression; and analyzing the Ti3C2T complex. xThe ratio of MXene to Bi₂Se₃, the volume of the modified electrode material, the voltage for depositing Pb and Cd ions, and the deposition time were optimized. The four key parameters—composite material ratio, modification volume, deposition potential, and deposition time—which exhibit nonlinear interactions, were synergistically optimized, overcoming the limitations of traditional "single-variable rotation" optimization methods. Machine learning algorithms (such as LSBoost-NGO, SVM, and RF) can automatically learn and model the complex nonlinear mapping relationship between these parameters and detection signals (such as peak current) by analyzing large amounts of experimental data. This allows them to overcome local extrema and efficiently and accurately search for parameter combinations that achieve the global optimum for objective functions such as detection sensitivity. This solves the technical problems of low optimization efficiency, susceptibility to local optima, and difficulty in handling parameter interactions inherent in traditional methods, ensuring the sensor operates at its optimal state. This approach not only employs an advanced hybrid algorithm (LSBoost-NGO) but also incorporates various mainstream machine learning models such as Support Vector Machines and Random Forests. It systematically compares these models with traditional single-variable multiple regression methods, forming a rigorous algorithm performance evaluation framework. By comparing the prediction accuracy and optimization performance of different models on the same dataset, it objectively selects the optimal algorithm best suited for the specific application scenario. This is not only to obtain the best parameters but also, methodologically, to demonstrate the significant advantages and necessity of introducing complex machine learning models compared to traditional statistical methods in handling such high-dimensional, nonlinear optimization problems. Deeply embedding machine learning into the sensor optimization process enhances intelligence, moving away from relying on repeated trial and error based on operator experience. Instead, it uses data-driven models for scientific decision-making, improving the efficiency and reproducibility of the optimization process. Furthermore, it holds the potential to discover even better parameter combinations beyond human experience, thereby unlocking higher sensor performance potential and shifting sensor development strategies from experience-driven to intelligence-driven. More specifically, this includes: S301. Experimental Design and Data Acquisition (Dataset Construction): S3011. Determine the optimization variables and their ranges, and clarify the four key optimization parameters and their scope of consideration: (1) Material ratio (X1): Bi2Se3 and Ti3C2T x The mass ratio (e.g., 1:1, 2:1, 3:1, 4:1, 5:1, 6:1).

[0033] (2) Modification volume (X2): The volume of the composite dispersion dropped onto the working electrode (e.g., 2 μL, 3 μL, 4 μL, 5 μL, 6 μL).

[0034] (3) Deposition potential (X3): applied for Pb enrichment 2+ and Cd 2+ The voltage value (within a range, such as -1.2V to -0.8V, Ag / AgCl).

[0035] (4) Deposition time (X4): The duration of the enrichment step (e.g., 60s to 300s).

[0036] S3012. Generate experimental scheme: Use experimental design methods (such as central composite design, Box-Behnken design, or optimal Latin hypercube sampling) to generate a set of experimental points with a reasonable number and uniform distribution in the parameter space. Each combination is a specific set of (X1, X2, X3, X4) conditions.

[0037] S3013. Perform the experiment and record the response value: Prepare the sensor and perform differential pulse voltammetry (DPV) detection according to each experimental condition combination generated in step one. Record the output response value (Y) for each experiment, specifically: Y Pb Pb 2+ The characteristic dissolution peak current value.

[0038] Y Cd Cd 2+ The characteristic dissolution peak current value.

[0039] The goal is to maximize Y Pb and Y Cd .

[0040] S302, Construction and Training of Machine Learning Models: S3021. Data Preprocessing: Randomly divide the collected dataset (input variable X and output response Y) into a training set (approximately 70%-80%) and a test set (approximately 20%-30%). Standardize or normalize the data to eliminate the influence of units.

[0041] S3022, Model Training and Hyperparameter Tuning: (1) The four machine learning algorithms LSBoost-NGO, SVM, RF and DTR were trained using the training set data respectively.

[0042] (2) For each algorithm, its hyperparameters (e.g., the kernel function and penalty coefficient C of SVM, the number of decision trees n_estimators of RF, etc.) are adjusted by methods such as cross-validation to make the model achieve optimal performance on the training data. The Northern Eagle Optimization (NGO) algorithm in LSBoost-NGO is used for this purpose.

[0043] S3023. Model Performance Evaluation and Comparison: (1) Use test set data that was not used in training to evaluate the predictive performance of each trained model. Commonly used evaluation metrics include the coefficient of determination (R²). 2), root mean square error (RMSE), etc.

[0044] (2) The prediction accuracy of the four machine learning models is quantitatively compared with that of the single variable multiple regression model (as a representative of traditional methods), so as to objectively demonstrate the superiority of machine learning methods.

[0045] S303, Parameter Optimization and Verification: S3031. Finding the optimal condition: Treat the best-performing machine learning model (such as the LSBoost-NGO model with the highest prediction accuracy) as the accurate "surrogate model." Use an optimization algorithm (such as NGO or other global optimization algorithms) to search on this surrogate model for the prediction of the response value Y. Pb and Y Cd The combination of input variables (X1, X2, X3, X4) that is maximized simultaneously is the globally optimal detection condition predicted by the model.

[0046] S3032. Experimental Verification: Conduct actual experiments in the laboratory under the optimal conditions predicted by the model to measure the real DPV response. Compare the measured values ​​with the model predictions to verify the reliability and accuracy of the optimization results.

[0047] Step S300 is a closed-loop process: data is acquired through carefully designed experiments, multiple machine learning models are trained using the data and the optimal model is selected, a global optimization search is performed based on the optimal model, and finally, the optimization results are verified through experiments. This method systematically replaces the inefficient "trial and error method" and achieves rapid, accurate, and global optimization of multiple key detection parameters.

[0048] Bi2Se3 / Ti3C2T in this embodiment x The sweat heavy metal electrochemical sensor uses the aforementioned Bi2Se3 / Ti3C2T x An electrochemical sensor for heavy metals in sweat was prepared using a specific method.

[0049] In implementation, a Bi2Se3 / Ti3C2T is provided. x An electrochemical sensor for heavy metals in sweat and its preparation method, which deeply integrates the innovative design of MXene-based composite materials with intelligent optimization of machine learning algorithms, is developed to achieve trace Pb detection. 2+ and Cd 2+ The detection provides a systematic solution, including the following steps: 1) Ti3C2T x Nanosheets and Bi2Se3 nanoflowers were prepared into a composite material by electrostatic self-assembly, and the Bi2Se3 / Ti3C2T composite material was prepared by micro-droplet coating. x A lead-cadmium ion electrochemical sensor was fabricated by modifying the surface of the working electrode platform with composite materials.

[0050] 2) Connect the sensor to the circuit and use differential pulse voltammetry (DPV) for electrochemical quantitative determination. Calculate the Pb content in sweat based on the peak current value. 2+ and Cd 2+ content.

[0051] 3) Using machine learning methods, a hybrid model (LSBoost-NGO model) combining the Least Squares Boosting Algorithm (LSBoost) and the Northern Eagle Optimization Algorithm (NGO) was employed to optimize the sensor's response to Pb. 2+ and Cd 2+ Key experimental parameters for detection.

[0052] The electrode used is a Type 300 screen-printed electrode, consisting of a working electrode, a reference electrode, and a counter electrode, forming a three-electrode system. The working and counter electrodes are made of carbon paste, while the reference electrode uses a mixture of silver and silver chloride. This electrode module can be directly connected to an electrochemical workstation for analysis.

[0053] During the DPV test, the starting potential was set to -1.4V, the ending potential was set to -0.6V, the amplitude was set to 5mV, and the frequency was set to 15Hz.

[0054] The LSBoost algorithm is an ensemble learning method, while the NGO optimization algorithm can automatically optimize the model's hyperparameters, helping the model find better solutions. The Pb and Cd detection models built using the LSBoost-NGO algorithm can handle complex multivariate nonlinear relationships.

[0055] The specific implementation steps are as follows: (1) Preparation of Bi₂Se₃: 1.800 g of sodium hydroxide was dissolved in 70 mL of high-purity water and magnetically stirred for 25 min at 70 °C and 450 r / min. 665 mg of selenium dioxide and 1940 mg of bismuth nitrate pentahydrate were added sequentially to the solution, and magnetic stirring was continued for 25 min at 70 °C and 450 r / min. 2.35 g of gallic acid monohydrate and 5.0 g of Triton were added to the mixed solution, and magnetically stirred until homogeneous under the same conditions. The solution was transferred to a 100 mL high-pressure reactor and reacted at 200 °C for 24 h. After the reaction was complete, the reactor was cooled to room temperature, and the solution was transferred to centrifuge tubes and centrifuged at 8000 r / min for 5 min. The supernatant was filtered off, and the solution was then centrifuged again using high-purity water and anhydrous ethanol, washing three times each. The prepared Bi₂Se₃ was dissolved in 150 mL of ultrapure water and sonicated for 1 h to ensure homogeneity. Take three 2 mL portions of the above solution into centrifuge tubes and dry them under vacuum at 60 °C for 48 h. After complete drying, weigh the tubes, subtract the mass of the empty centrifuge tubes, and divide by the volume to prepare a Bi₂Se₃ solution with a concentration of 5 mg / mL.

[0056] (2) Preparation of Bi2Se3 / Ti3C2T x 1.6 g of lithium fluoride was slowly added and dissolved in 30 mL of 9M hydrochloric acid under stirring. Then, 1 g of titanium aluminum carbide MAX precursor was slowly poured into a LiF / HCl mixed solution, and the mixture was reacted at 40 °C with continuous stirring for 24 h. After the reaction was complete, the etching suspension was repeatedly washed with deionized water, centrifuged at 5000 rpm for 5 min, washed three times, and then centrifuged at 10000 rpm until the pH of the supernatant reached 6. The sample was then placed in... After pretreatment at 80℃ for 2 hours using a freeze dryer, the product was freeze-dried for 24 hours to finally obtain powdered Ti3C2T. x The MXene product was stored in a vacuum drying oven for further room temperature storage and use. Six complex solutions were prepared with mass ratios of Bi2Se3:MXene of 1:1, 2:1, 3:1, 4:1, 5:1, and 6:1. The above two substances were added according to the ratio of 0.1 mL Nafion solution and 0.2 mL isopropanol, and the mixture was sonicated for 1 hour to make it homogeneous.

[0057] (3) Preparation of electrochemical sensor: The bare screen-printed carbon electrode was first treated with O2 plasma at a power of 150W for 30s. Then, 2μL, 3μL, 4μL, 5μL and 6μL of Bi2Se3 / Ti3C2T at a concentration of 5mg / mL were added dropwise respectively. xThe dispersion was applied to the working electrode and then dried in an oven at 60°C for half an hour. Finally, it was stored at 4°C for later use. The MXene and Bi₂Se₃ electrodes, serving as controls, were prepared and stored using the same method.

[0058] (4) Electrochemical test: Prepare Pb solutions with concentrations of 0 ppb, 5 ppb, 10 ppb, 20 ppb, 40 ppb, 60 ppb, 80 ppb, 100 ppb, and 150 ppb. 2+ Cd 2+ Artificial sweat (pH=5.5) was used. The fabricated electrodes were tested using differential pulse voltammetry, and the differential pulse voltammetry curves were recorded.

[0059] (5) Machine Learning-Assisted Optimization of Experimental Conditions: Four machine learning methods—LSBoos-NGO, Support Vector Machine (SVM), Random Forest (RF), and Decision Tree Regression (DTR)—were used to construct models, which were compared with traditional single-variable multiple regression optimization models. The Ti3C2T complex... x The ratio of MXene to Bi2Se3, the volume of the modified electrode material, the voltage for depositing Pb and Cd ions, and the deposition time were optimized.

[0060] The results of this invention are Figure 1 It can be seen that Ti3C2T in acid solution x The zeta potentials of Bi₂Se₃ and Ti₃C₂T₃ are +12.8 mV and -23.2 mV, respectively. Because their potentials are opposite, when the two are mixed and ultrasonically tested, Bi₂Se₃ can be uniformly coated onto Ti₃C₂T₃ through electrostatic adsorption. x Thin film surface.

[0061] Figure 2 SEM images show that Bi2Se3 exhibits a flower-like structure with thin, lamellar aggregates. Bi2Se3 / Ti3C2T x In composite materials, Ti3C2T x The thin film adhered to the Bi₂Se₃ surface, exhibiting a larger lateral dimension and increased specific surface area, which facilitates the exposure of active sites and mass transfer. EDS spectroscopy confirmed the uniform distribution of Bi, Se, Ti, and C elements, indicating successful composite formation.

[0062] Figure 3 The XRD pattern shows that the composite material contains both Ti3C2T. x The material exhibits characteristic peaks similar to those of Bi₂Se₃, with good crystallinity and no impurity peaks. XPS spectra confirm the presence of C, Ti, Bi, Se, and surface functional groups. Notably, the Se-O bonds indicate strong interfacial coupling, signifying successful composite formation.

[0063] Figure 4The EIS plot shows that Bi2Se3 modification of the electrode maximizes Rct, while Ti3C2T x Modification significantly reduced Rct, while the composite material's Rct fell between the two, confirming the successful construction of the sensor. CV showed a linear relationship between peak current and the square root of the scan rate (Rc). 2 =0.99), indicating that the reaction is diffusion-controlled. The electrochemical active area of ​​the composite electrode reaches 0.3052 cm². 2 It is 2.43 times that of a bare electrode, providing more active sites.

[0064] Figure 5 The radar chart in the image shows that the LSBoost-NGO model has the best performance (R). 2 =0.99394, the lowest error index), the RF model is the worst (R 2 =0.24065), with SVM, MLR, and DTR performance in the middle range.

[0065] Figure 6 SHAP analysis reveals the influence of Pb 2+ The key factors for peak current are, in order: Cd 2+ Peak current, deposition time, potential, and material ratio were considered. The average MAE of the five-fold cross-validation was 5.2581 with a standard deviation of 0.56438, indicating that the model has good robustness.

[0066] Figure 7 This indicates that after 60 iterations, the R-values ​​of the LSBoost-NGO model on the training and test sets are significantly improved. 2 The error indices are 0.998 and 0.939 respectively, indicating low accuracy and a high degree of agreement between the predicted and actual values. This demonstrates the model's high precision and strong generalization ability, making it suitable for predicting Pb in practical applications. 2+ Peak current.

[0067] Figure 8 This indicates that Bi2Se3 and Ti3C2T x The optimal synthesis ratio was 1.48, the optimal deposition potential was -1.248V, the deposition time was 255s, and the optimal material volume for the modified electrode was 4.505μL.

[0068] Figure 9 Prove Bi2Se3 / Ti3C2T x The sensor can detect Pb within the range of 0-150 ppb. 2+ with cd 2+ Cd 2+ The linear equation y = 0.038x + 0.953 (R) 2 =0.995), sensitivity 0.038 μA·ppb -1 The detection limit is 2.88 ppb.

[0069] Pb 2+ The linear equation y = 0.163x + 0.953 (R²) 2 =0.996), sensitivity 0.163 μA·ppb -1 The detection limit is 3.45 ppb, which meets the actual needs of sweat testing.

[0070] Figure 10 Bi2Se3 / Ti3C2T x Sensor anti-interference tests showed that lactic acid and uric acid affect Pb. 2+ / Cd 2+ Peak current has little effect; glucose reduces current; K + / Na + The peak shift and current reduction exceeding 10% were caused by the ion shielding effect. After storage at 4°C for 7 days, the effect on Cd... 2+ and Pb 2+ The detection activities remained at 91.7% and 90.6%, respectively, indicating good stability.

[0071] This invention, Bi2Se3 / Ti3C2T x Technical advantages of the sweat heavy metal electrochemical sensor and its preparation method: 1. This invention uses Ti3C2T x -MXene and Bi2Se3 nanoflower composite modification of screen-printed electrodes, utilizing Ti3C2T x The high specific surface area and excellent conductivity of Bi₂Se₃, along with its high affinity for heavy metal ions, allow it to form a stable composite structure through electrostatic self-assembly, significantly improving the sensor's sensitivity and the number of active sites. Compared to traditional Au or pure Bi electrodes, the electrode material cost is significantly reduced, and the fabrication process is simple, making it suitable for large-scale production.

[0072] 2. This invention introduces the LSBoost-NGO machine learning algorithm to globally optimize experimental conditions, overcoming the limitations of traditional OFAT and OED methods. This algorithm can automatically adjust the Ti3C2T... x Several key parameters, including the Bi2Se3 ratio, deposition potential, deposition time, and modification volume, were considered, while also taking into full account Cd. 2+ For Pb 2+ The interference effects of detection were mitigated, enabling the detection of Pb in complex systems. 2 + and Cd 2+ High-precision prediction, model fit R 2 The accuracy reached 0.99394, with a prediction error significantly lower than other machine learning methods.

[0073] 3. This invention relates to the treatment of Pb in artificial sweat. 2+ and Cd 2+It exhibits excellent detection performance, among which Pb 2+ The linear detection range is 0–150 ppb, and the detection limit is 3.45 ppb; Cd 2+ The linear detection range is 10–150 ppb, with a detection limit of 2.88 ppb. Furthermore, the sensor maintains good selectivity in the presence of interfering substances (such as lactic acid, glucose, uric acid, KCl, and NaCl) and retains over 90% activity stability within 7 days, meeting the needs of practical sweat detection.

[0074] 4. This invention combines screen-printed electrodes with a micro electrochemical platform, realizing the portability and wearability of the sensor, and providing a fast, reliable and economical detection method for heavy metal pollution in the environment and personal health monitoring.

[0075] Matters not covered in this invention are common knowledge.

[0076] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0077] The embodiments described above are merely examples of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention.

[0078] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A Bi2Se3 / Ti3C2T x A method for preparing a sweat heavy metal electrochemical sensor, characterized by, The method comprises the following steps: S100、Ti3C2T x nanosheets and Bi2Se3 nanoflower materials are prepared into Bi2Se3 / Ti3C2T x composite material, Bi2Se3 / Ti3C2T x composite material is modified on the surface of the working electrode platform to obtain a lead and cadmium ion electrochemical sensor; S200. Connect the lead-cadmium ion electrochemical sensor to the circuit, and use the differential pulse voltammetry method for electrochemical quantitative determination. Calculate the Pb content in sweat based on the peak current value. 2+ and Cd 2+ The content; S300、Using the method of machine learning, a hybrid model combining the least square boosting algorithm and the northern hawk optimization algorithm is used to optimize the key experimental parameters of Pb 2+ and Cd 2+ detection by sensors.

2. The Bi2Se3 / Ti3C2T x A method for preparing a sweat heavy metal electrochemical sensor, characterized by, The working electrode platform in step S100 adopts a 300 type screen-printed electrode, which is composed of a working electrode, a reference electrode and a counter electrode to form a three-electrode system; the working electrode and the counter electrode are composed of carbon paste, and the reference electrode uses a mixture of silver and silver chloride; the working electrode platform is directly connected to an electrochemical workstation for analysis.

3. The Bi2Se3 / Ti3C2T of claim 1 x A method for preparing a sweat heavy metal electrochemical sensor, characterized by, In step S100, Ti3C2T x The preparation process is as follows: S111, 1000mg-2000mg of lithium fluoride is slowly added and dissolved in 20mL-40mL of 9M hydrochloric acid under stirring to obtain a LiF / HCl mixed solution; S112, 800mg-1300mg of titanium aluminum carbide MAX precursor is slowly poured into the LiF / HCl mixed solution, and the reaction is carried out at 35℃-45℃ under continuous stirring for 20h-30h to obtain an etching suspension; S113, after the reaction is completed, the etching suspension is repeatedly washed by centrifugation at a speed of 4000rpm-6000rpm for 3min-8min using deionized water; S114, after washing for three times, the sample is obtained by centrifugation at a speed of 9000rpm-12000rpm until the pH value of the supernatant reaches 6; S115, placing the sample in 80 °C freeze dryer for 1.5-2.5 h pretreatment and 20-30 h freeze drying to obtain Ti3C2T powder x MXene product and stored in a vacuum drying box for use.

4. The Bi2Se3 / Ti3C2T of claim 3 x A method for preparing an electrochemical sensor of sweat heavy metals, characterized by, The preparation of Bi2Se3 in step S100 is specifically as follows: S121, 1000mg-2500mg of sodium hydroxide is dissolved in 50mL-90mL of high-purity water, and the solution is magnetically stirred at 60℃-80℃ and a speed of 400r / min-550r / min for 20min-35min; S122, 500mg-750mg of selenium dioxide and 1000mg-2500mg of bismuth nitrate pentahydrate are sequentially added to the solution, and the solution is continuously magnetically stirred at 60℃-80℃ and a speed of 400r / min-550r / min for 20min-35min; S123, 2000mg-2500mg of gallic acid monohydrate and 4000mg-6000mg of triton are added to the mixed solution, and the solution is magnetically stirred at 60℃-80℃ and a speed of 400r / min-550r / min until it is uniform; S124, the solution is transferred to a high-pressure reaction kettle, and the reaction is carried out at 150℃-250℃ for 20h-30h; S125, after the reaction is completed, the reaction kettle is cooled to room temperature, the solution is introduced into a centrifuge tube, and the centrifuge tube is placed in a centrifuge and centrifuged at a speed of 7000r / min-9000r / min for 4min-8min; S126, the supernatant is filtered out, and the same method as step S125 is used to centrifuge the solution using high-purity water and anhydrous ethanol, respectively, and each sample is washed for 3 times to obtain Bi2Se3 nanomaterials; S127, the Bi2Se3 nanomaterials are dissolved in 100mL-200mL of ultrapure water, ultrasonically treated for 0.8h-1.5h, and uniformly mixed to obtain a Bi2Se3 solution, which is ready for use.

5. The Bi2Se3 / Ti3C2T of claim 4. x A method for preparing an electrochemical sensor of sweat heavy metals, characterized by, A plurality of Bi2Se3 solutions obtained in step S127 are taken, each 2mL, vacuum dried at 50℃-70℃ for 40h-55h; after complete drying, the weight is measured, and a Bi2Se3 solution with a concentration of 4mg / mL-6mg / mL is prepared.

6. The Bi2Se3 / Ti3C2T of claim 5 x A method for preparing an electrochemical sensor of sweat heavy metals, characterized by, Bi2Se3 / Ti3C2T x The preparation of the composite material is specifically: Bi2Se3:Mxene were 1:1, 2:1, 3:1, 4:1, 5:1, 6:1 by mass ratio, 6 portions of composite solution were prepared, and 0.2 mL of isopropyl alcohol was added to the 6 portions of composite solution at the same time according to the ratio of adding 0.1 mL of Nafion solution, and ultrasonic mixing was performed for 0.8-1.5 h to obtain Bi2Se3 / Ti3C2T x dispersion.

7. The Bi2Se3 / Ti3C2T of claim 6 x A method for preparing a sweat heavy metal electrochemical sensor, characterized by, The preparation of the lead-cadmium ion electrochemical sensor in step S100 is specifically as follows: S131, the bare screen-printed carbon electrode is first treated with O2 plasma at a power of 120 W-180 W for 25 s-35 s; S132, respectively drop 2 μL, 3 μL, 4 μL, 5 μL and 6 μL of Bi2Se3 / Ti3C2T x dispersion onto the working electrode; S133, drying in an oven at 55 ℃-65 ℃ for 0.3 h-1 h, and finally storing at 3 ℃-6 ℃ for standby; S134, the MXene electrode and the Bi2Se3 electrode as controls are prepared and stored in the same way as steps S131-S133.

8. The Bi2Se3 / Ti3C2T of any one of claims 1-7. x A method for preparing an electrochemical sensor of sweat heavy metals, characterized by, Step S200 is specifically: preparing artificial sweat containing Pb 2+ , Cd 2+ with concentrations of 0 ppb, 5 ppb, 10 ppb, 20 ppb, 40 ppb, 60 ppb, 80 ppb, 100 ppb, and 150 ppb, respectively, pH = 5.5, and performing differential pulse voltammetry test on the prepared electrode to record the differential pulse voltammetry curve.

9. The Bi2Se3 / Ti3C2T of any one of claims 1-7, wherein the Bi2Se3 / Ti3C2T is a Bi2Se3 / Ti3C2T MXene nanocomposite. x A method for preparing an electrochemical sensor of sweat heavy metals, characterized by, Step S300 is specifically as follows: four machine learning methods, LSBoos-NGO, support vector machine SVM, random forest RF and decision tree regression DTR, are used to construct models, and the models are compared with an optimized experimental method model of single variable multiple regression. The ratio of Ti3C2T x The ratio of MXene to Bi2Se3, the volume of modified electrode material, the voltage for depositing Pb ions and Cd ions, and the deposition time were optimized.

10. A Bi2Se3 / Ti3C2T x An electrochemical sensor of sweat heavy metals characterized in that, Bi2Se3 / Ti3C2T x A method for preparing an electrochemical sensor of sweat heavy metals is prepared.

Images