A method of screening a g-c3n4-supported single-atom catalyst

Abstract

The present application relates to the technical field of catalyst screening, and particularly relates to a method for screening g-C3N4-supported single-atom catalysts. The specific method comprises the following steps: placing a doping element at a doping site of a g-C3N4 substrate, and performing structure optimization to screen pre-screening catalysts that meet electronic localization; using density functional theory and an XGBoost model, taking significant key features as inputs, and forming energy as outputs to screen initial screening catalysts; performing CO2 reduction reaction catalytic activity evaluation on the initial screening catalysts by using adsorption free energy change to screen re-screening catalysts; and performing hydrogen evolution reaction catalytic activity evaluation on the re-screening catalysts to screen catalysts for syngas production, CO2 reduction reaction catalysts, and hydrogen evolution reaction catalysts. The method of the present application solves the problems of low screening efficiency, long time consumption, and low prediction accuracy of existing single-atom catalyst screening methods, which affect the reliability of screening results.

Description

Technical Field

[0001] The present invention relates to the technical field of catalyst screening, and in particular to a method for screening g-C3N4-supported single-atom catalysts. Background Art

[0002] Currently, existing single-atom catalyst screening methods mainly rely on the following technologies: experimental screening, DFT-based computational screening, and pure machine learning screening.

[0003] Experimental screening methods require extensive synthesis and characterization of materials, which is time-consuming and costly. Repeated testing under multiple experimental conditions is often necessary, and the screening results are subject to uncertainty. Consequently, experimental screening methods suffer from low efficiency, an inability to process a large number of potential catalyst combinations, and an inability to quickly identify highly effective catalyst candidates.

[0004] DFT-based computational screening is a widely used method in theoretical calculations, capable of accurately predicting the electronic structure and reactivity of materials. However, the DFT calculation process is complex and consumes a large amount of computing resources. Therefore, DFT-based computational screening, which requires high-precision ab initio calculations for each material combination, is prohibitively expensive and slow to screen large numbers of single-atom catalysts, making it difficult to quickly identify highly efficient catalysts from large-scale material libraries.

[0005] Pure machine learning screening primarily utilizes machine learning models to identify potential catalyst combinations from large-scale data sets. However, pure machine learning methods rely on high-quality training datasets. Data sets in materials science are often scarce, particularly high-quality, comprehensive experimental data. Consequently, pure machine learning screening suffers from a lack of sufficient high-quality data, which can easily lead to model overfitting and inaccurate predictions.

[0006] Currently, g-C3N4, due to its unique two-dimensional structure, particularly its abundant six-fold cavities and non-equivalent distribution of nitrogen atoms, has become an ideal support for single-atom catalysts, particularly in CO2 reduction reactions and hydrogen evolution reactions. This structure provides stable anchoring sites for single atoms, ensuring their efficient fixation and good electron localization. The large surface area of ​​g-C3N4 also enhances catalytic activity and effectively improves reaction selectivity. However, existing single-atom catalyst screening methods suffer from low screening efficiency, long processing times, and low prediction accuracy, which can easily affect the reliability of screening results. Summary of the Invention

[0007] In order to solve the problems of low screening efficiency, long time consumption and low prediction accuracy of existing single-atom catalyst screening methods, which affect the reliability of screening results, the purpose of the present application is to provide a method for screening g-C3N4 supported single-atom catalysts, which improves the screening efficiency of g-C3N4 system catalysts and improves the reliability of the prediction results.

[0008] In order to achieve the above-mentioned purpose, the technical scheme of the present application is as follows.

[0009] The present application provides a method for screening g-C3N4 supported single-atom catalysts, comprising the following steps:

[0010] According to the two-dimensional structure of the g-C3N4 substrate, the doping sites of the g-C3N4 substrate are obtained; the doping elements are placed in the doping sites of the g-C3N4 substrate, and the structure is optimized to screen out catalysts that meet the electronic localization, and obtain a pre-screening result; the formation energy of part of the catalysts in the pre-screening result is calculated, the key feature data related to the formation energy is collected, the formation energy-key feature data set is established, and the first prediction model is constructed, so that the first prediction model can learn the nonlinear relationship between the key features and the formation energy; the feature importance analysis is performed on the first prediction model, and the feature importance pie chart is generated; the significance analysis is performed on the feature importance pie chart to determine the significant key features; the catalysts with negative formation energy are screened out by using the significant key features, and the preliminary screening result is obtained; the adsorption free energy change of the catalysts in all preliminary screening results is calculated, the core feature data related to the adsorption free energy change is collected, the adsorption free energy change-core feature data set is established, and the second prediction model is constructed, so that the second prediction model can learn the nonlinear relationship between the key features and the adsorption free energy change; the SHAP value is calculated according to the training result of the second prediction model, the feature marginal contribution analysis is performed on the core features according to the SHAP value, and the significant core features are determined; based on the second prediction model, the adsorption free energy change and the limiting potential of the preliminary screening catalyst are predicted by using the significant core features, which are used for evaluating the CO2 reduction reaction catalytic activity of the preliminary screening catalyst, and screening out the catalysts with absolute value of adsorption free energy change ≤0.8eV and limiting potential ≤0.8V, to obtain the re-screening result; the hydrogen evolution reaction catalytic activity of the catalysts in the re-screening result is evaluated, to screen out the catalysts suitable for synthesis gas production, CO2 reduction reaction or hydrogen evolution reaction.

[0011] Preferably, the doping sites of the g-C3N4 substrate are 2 C substitution sites, 3 N substitution sites and 2 H interstitial sites. Preferably, the doping elements are at least one of Zr, Ti, Mn, Fe, Co, Ni, Cu, Zn, Ru, Rh, Pd, Pt, Al, Li, B, Si, S, O, P and Cl.

[0012] Preferably, the key features related to formation energy are doping site, atomic radius, valence electron number, electron affinity, chemical potential, atomic number, electronegativity, first ionization energy, atomic weight and inner electron number; the significant key features are doping site and atomic radius.

[0013] Preferably, the method for constructing the first prediction model is: according to the formation energy-key feature data set, using root mean square error, coefficient of determination and mean absolute error as evaluation indexes, the hyperparameters of the model are optimized, thereby learning the nonlinear relationship between the key features and the formation energy, and the first prediction model with the best prediction accuracy is constructed and trained.

[0014] Preferably, the method for constructing the second prediction model is: according to the adsorption free energy change-core feature data set, using root mean square error, coefficient of determination and mean absolute error as evaluation indexes, the hyperparameters of the model are optimized, thereby learning the nonlinear relationship between the core features and the adsorption free energy change, and the second prediction model with high interpretability is constructed and trained.

[0015] Preferably, the core features related to the adsorption free energy change include single atom features and catalyst features; the single atom features are atomic radius, atomic number, single atom chemical potential, valence electron, first ionization energy and electron affinity; the catalyst features are charge transfer amount, binding energy, formation energy, Fermi level and band gap; the significant core features are band gap and binding energy.

[0016] Preferably, the method for screening the catalysts suitable for synthesis gas production, CO2 reduction reaction or hydrogen evolution reaction is: according to the competitive relationship of the adsorption free energy change of the CO2 reduction reaction and the hydrogen evolution reaction, the catalysts in the complex screening result are evaluated in terms of hydrogen evolution reaction catalytic activity, CO2 reduction reaction activity and synthesis gas production applicability; the adsorption free energy change of the CO2 reduction reaction is denoted as ΔG CO2RR; the adsorption free energy change of the hydrogen evolution reaction is denoted as ΔG H; when the ΔG CO2RR of the catalyst is lower than the ΔG H, the catalyst is a CO2 reduction catalyst that tends to promote the CO2 reduction reaction; when the ΔG H of the catalyst is lower than the ΔG CO2RR, the catalyst is a hydrogen evolution catalyst that tends to promote the hydrogen evolution reaction; when the ΔG CO2RR of the catalyst is close to the ΔG H, the catalyst is a synthesis gas catalyst that promotes both the CO2 reduction reaction and the hydrogen evolution reaction.

[0017] The application provides a system for screening g-C3N4 solid-supported single atom catalysts, which comprises a processor and a memory, and the memory stores a computer program capable of running on the processor; when the computer program is executed by the processor, the steps of the method for screening g-C3N4 solid-supported single atom catalysts are implemented.

[0018] The application provides a computer-readable storage medium, wherein a data processing program is stored on the computer-readable storage medium, and the data processing program realizes steps of the method for screening a g-C3N4 solid-supported monatomic catalyst when executed by a processor.

[0019] Advantages of the application:

[0020] 1. The method of the application is mainly used for screening a g-C3N4 solid-supported monatomic catalyst. First, a doping site is determined based on a two-dimensional structure of g-C3N4, and a candidate catalyst satisfying electronic localization is screened through structure optimization. Then, a prediction model is established by using a machine learning technology, and the formation energy and the adsorption free energy change of the catalyst are predicted by using key feature data, so as to evaluate the catalytic activity of the catalyst for CO2 reduction reaction. Finally, the catalytic activity of the catalyst for hydrogen evolution reaction is evaluated, so as to screen a catalyst suitable for different catalytic reactions. The method of the application can quickly screen a potential high-efficiency catalyst by combining efficient prediction of the machine learning model, and can further reduce the screening range and significantly reduce the sample amount of experimental and calculation verification by combining feature importance analysis, thereby improving the screening efficiency of the g-C3N4 system catalyst and improving the reliability of the prediction result.

[0021] 2. In the screening process, the prediction model is combined with key features affecting the performance of the catalyst, and important features such as formation energy and adsorption energy are identified by SHAP analysis to contribute to the catalytic performance, thereby effectively improving the accuracy of screening. Compared with the traditional screening method, the method of the application can effectively avoid the data deficiency problem in the pure machine learning method, thereby ensuring the reliability of the prediction result.

[0022] 3. The screening method of the application can set different screening conditions according to different reaction requirements, such as CO2 reduction reaction, hydrogen evolution reaction and synthesis gas production. For example, by using the standard of limiting potential and free energy change, a catalyst with high reaction rate and good selectivity can be screened. For multi-reaction catalyst design, such as synthesis gas production, the application can screen a catalyst capable of simultaneously catalyzing CO2 reduction and hydrogen evolution reaction in the same system, thereby realizing efficient utilization of reaction conditions.

[0023] 4. The method of the application is not only limited to CO2 reduction reaction, but also can be applied to efficient screening of catalysts for other gas-solid interface catalytic reactions, such as hydrogen evolution reaction and synthesis gas production. The method of the application has high scalability, and provides a broad application prospect for design of new catalysts. BRIEF DESCRIPTION OF DRAWINGS

[0024] Figure 1is the electron localization function map of g-C3N4 and the schematic diagram of substitution sites and interstitial sites. Among them, (a) is the electron localization function map of g-C3N4; (b) is the schematic diagram of substitution sites and interstitial sites of g-C3N4; the orange ball represents the N atom; the blue ball represents the C atom.

[0025] Figure 2 E calculated by DFT and E predicted by XGBoost model f E calculated by DFT and E predicted by XGBoost model f E calculated by DFT and E predicted by XGBoost model f E calculated by DFT and E predicted by XGBoost model f E calculated by DFT and E predicted by XGBoost model f Feature importance analysis results. Among them, (a) is the determination coefficient, mean absolute error and root mean square error value of various machine learning models; (b) is E predicted by XGBoost model f E calculated by DFT and E predicted by XGBoost model f Comparison results of test set of calculated value; (c) is E based on XGB model f Feature importance analysis results. Among them, Train R 2 represents the determination coefficient of the training set; Train MAR represents the mean absolute error of the training set; Train MSR represents the root mean square error of the training set; Test R 2 represents the determination coefficient of the test set; Test MAR represents the mean absolute error of the test set; Test MSR represents the root mean square error of the test set. The full name of machine learning in English is Machine Learning, which is abbreviated as ML. XGB, ABR, KNR, LR, MLP, PLS, RR, SVR, RFR and GBR respectively represent different machine learning algorithms or models. Doping site, Radii, Out_e, Aff, Chemi_pot, Atom_N, N m , I1 respectively represent the key features related to the formation energy. Among them, Doping site represents the doping site; Radii represents the atomic radius; Out_e represents the number of outer electrons; Aff represents the electron affinity; Chemi_pot represents the chemical potential; Atom_N represents the atomic number; N m represents electronegativity; I1 represents the first ionization energy; Atom_W represents the atomic weight; In_e represents the number of inner electrons.

[0026] Figure 3Schematic diagram of the mechanism and reaction pathway of the CO2 reduction reaction on the surface of the SAC@g-C3N4 catalyst and the limiting potential diagram of the CO2 reduction reaction. Among them, (a) is a schematic diagram of the mechanism and reaction pathway of the CO2 reduction reaction on the surface of the SAC@g-C3N4 catalyst; (b) is the limiting potential diagram of the CO2 reduction reaction.

[0027] Figure 4 The training set of the ΔG predicted value predicted by the XGBoost prediction model, the test set of the ΔG calculated value calculated by DFT, and the comparison results of the training set of the ΔG predicted value and the test set of the ΔG calculated value, and the ΔG feature importance analysis results based on the XGBoost prediction model. Among them, (a) is the determination coefficient, mean absolute error and root mean square error value of various machine learning models; (b) is the comparison result of the training set of the ΔG predicted value predicted by the XGBoost prediction model and the test set of the ΔG calculated value calculated by DFT; (c) is the ΔG feature importance analysis result based on the XGB prediction model.

[0028] Figure 5 The SHAP summary diagram of the XGB prediction model for ΔG1, ΔG2, ΔG3 and ΔG4. Among them, (a) is the SHAP summary diagram of the XGB prediction model for ΔG1; (b) is the SHAP summary diagram of the XGB prediction model for ΔG2; (c) is the SHAP summary diagram of the XGB prediction model for ΔG3; (d) is the SHAP summary diagram of the XGB prediction model for ΔG4.E g , Radii, Out_e, Aff, μ, Atom_N, E b , E f , E fermi , Bader, I1 respectively represent the key features related to the formation energy. Among them, E g represents the band gap; Radii represents the atomic radius; Out_e represents the number of outer electrons; Aff represents the electron affinity; μ represents the single atom chemical potential; Atom_N represents the atomic number; E b represents the binding energy; E f represents the formation energy; E fermi represents the Fermi level; Bader represents the charge transfer amount; I1 represents the first ionization energy.

[0029] Figure 6 The competition relationship diagram of the CO2 reduction reaction and the hydrogen evolution reaction on the surface of the SAC@g-C3N4 catalyst. Among them, CO2RR represents the CO2 reduction reaction, and HER represents the hydrogen evolution reaction.

[0030] Figure 7 The principle diagram for screening the single atom catalyst supported by g-C3N4 provided by an embodiment of the present application. DETAILED DESCRIPTION

[0031] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0032] Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative work shall fall within the scope of protection of the present invention.

[0033] Density Functional Theory, abbreviated as DFT. Vienna Ab initio Simulation Package, abbreviated as VASP. Generalized Gradient Approximation, abbreviated as GGA. Perdew-Burke-Ernzerhof Functional, abbreviated as PBE. VESTA, abbreviated as Visualization for Electronic and Structural Analysis. Projected Augmented Wave, abbreviated as PAW. g-C3N4 stands for graphitic carbon nitride or graphite-like carbon nitride. Single-Atom Catalysts, abbreviated as SAC. CO2RR stands for carbon dioxide reduction reaction; HER stands for hydrogen evolution reaction.

[0034] The present invention mainly provides a method for screening single-atom catalysts for CO2 reduction reaction, hydrogen evolution reaction and synthesis gas generation on g-C3N4 through density functional theory and machine learning.

[0035] In this paper, the Vienna ab initio simulation package was used to calculate the electronic properties of all structures. The PBE functional under the generalized gradient approximation was used to describe the electronic interaction, and VESTA was used for model construction. The pseudopotential used the projected augmented wave method with an energy cutoff of 500 eV to ensure a balance between the accuracy of the calculation results and the calculation time. The k-point grid was 2×2×1 for the integration of the Brillouin zone, and the z direction was increased. The geometric relaxation process of the structure is carried out with a residual force per atom less than The convergence standard of electron energy is 10 -5eV. In addition, van der Waals interactions are treated using DFT-D3 semi-empirical correction to ensure the accuracy of weak interactions.

[0036] The formation energy of the catalyst is calculated by density functional theory to evaluate the thermodynamic stability of the doping atoms on g-C3N4. The formula for calculating the formation energy is:

[0037] For substitution model: E f = E(X-g-C3N4) - E(g-C3N4) - μ(X) + μ(N) + μ(C); for interstitial model: E f = E(X-g-C3N4) - E(g-C3N4) - μ(X); where, E f represents the formation energy;

[0038] E(X-g-C3N4) represents the total energy of the g-C3N4 system doped with element X; E(g-C3N4) is the total energy of pure g-C3N4, μ(X) is the chemical potential of a single doping element X, μ(N) is the chemical potential of a single N atom, and μ(C) is the chemical potential of a single C atom.

[0039] The binding energy E b of the doping element introduced into the g-C3N4 system is calculated as follows:

[0040] E b = E(X-g-C3N4) - E(X) - E(g-C3N4); where, E b represents the binding energy; E(X-g-C3N4) represents the total energy of the g-C3N4 system doped with element X; E(X) represents the energy without structural optimization after removing g-C3N4 from E(X-g-C3N4); E(g-C3N4) represents the energy without structural optimization after removing X from E(X-g-C3N4). According to the definition of binding energy, the more negative the E b value, the stronger the binding between the doping element X and the substrate.

[0041] Adsorption energy calculation is used to evaluate the adsorption of CO2 and H2 on the surface of the catalyst, to ensure that the catalyst has good adsorption capacity for target molecules. The formula for calculating the adsorption energy is:

[0042] E ads = E(Y-X-g-C3N4) - E(X-g-C3N4) - E Y ; where, E ads represents the adsorption energy;

[0043] E(Y-X-g-C3N4) represents the total energy of the adsorbed substance Y on the g-C3N4 system doped with element X; E(X-g-C3N4) represents the total energy of the g-C3N4 system doped with element X, E Y represents the total energy of the adsorbed substance Y in a free state. For example, the adsorbed substance Y is CO2 or H2.

[0044] According to the reaction path of CO2RR and HER, the thermodynamic feasibility of the reaction is described using the change amount of Gibbs free energy. The zero-point energy and the change amount of entropy are calculated by the vibration frequency, and the calculation formula of the change amount of Gibbs free energy is: ΔG = E ads - TΔS + ΔZPE; wherein, ΔG represents the change amount of Gibbs free energy; E ads represents the adsorption energy; T represents the temperature; ΔS represents the change amount of entropy; and ΔZPE represents the zero-point energy. For CO2RR and HER, the change amount of entropy and the zero-point energy of the reaction intermediate are calculated using the vibration frequency data, and the change amount of entropy of the gaseous molecule comes from the NIST database. All calculations are carried out at a constant pressure of 0 GPa and a temperature of 298.15 K.

[0045] In addition, on the basis of a large number of density functional theory calculation results, the application constructs a plurality of machine learning models, including extreme gradient boosting regression, English full name XGBoost Regression, for short XGB; K-nearest neighbor regression, English full name K-Nearest Neighbors Regression, for short KNR; support vector regression, English full name Support Vector Regression, for short SVR; random forest regression, English full name Random Forest Regression, for short RFR; multilayer perceptron, English full name, Multilayer Perceptron, for short MLP; gradient boosting regression, English full name Gradient Boosting Regression, for short GBR; linear regression, English full name Linear Regression, for short LR; partial least squares regression, English full name Partial Least Squares Regression, for short PLSR; ridge regression, English full name Ridge Regression, for short RR; AdaBoost regression, AdaBoost Regression, for short ABR. The application combines density functional theory DFT and machine learning, aiming to accelerate the screening process of single-atom catalysts, and overcomes many defects of existing screening methods, and has the significant advantages of high-efficiency screening capability, accuracy and reliability, strong pertinence and wide application range. The method of the application uses high-quality DFT data as a training set, and the machine learning model can effectively predict the performance of new catalyst combinations, especially in the application potential in CO2 reduction reaction, hydrogen evolution reaction and synthesis gas generation. It is worth noting that the XGB model successfully predicts the catalyst activity, and combines the SHAP analysis method to further analyze the specific contribution of the key features to the catalyst activity, helping to optimize the screening results. SHAP, English full name SHapley Additive Explanations, Chinese explanation is SHapley Additive Explanations.

[0046] Feature engineering: the key features related to the formation energy have 10, respectively, the doping site, the atomic radius, denoted as Radii; the number of valence electrons, denoted as Out_e; the electron affinity, denoted as Aff; the chemical potential, denoted as Chemi_pot; the number of atoms, denoted as Atom_N; the electronegativity, denoted as N m; first ionization energy, denoted as I1; atomic weight, denoted as Atom_W; number of inner electrons, denoted as In_e. The core features related to the change in adsorption free energy have 11, which are divided into single-atom features and catalyst features. The single-atom features include atomic radius, denoted as Radii; atomic number, denoted as Atom_N; single-atom chemical potential, denoted as mu; number of valence electrons, denoted as Out_e; first ionization energy, denoted as I1; electron affinity, denoted as Aff. The catalyst features include charge transfer amount, denoted as bader; binding energy, denoted as E b ; formation energy, denoted as E f ; Fermi level, denoted as E fermi ; band gap, denoted as E g .

[0047] Model training: The data set is divided into training set and test set according to the ratio of 8:2. Through 500 times of iteration training, it is ensured that the model has high precision and generalization ability. The performance of the model is evaluated by root mean square error, determination coefficient and mean absolute error, and the specific calculation formula is as follows:

[0048]

[0049] Among them, R 2 represents the determination coefficient; MSR represents the root mean square error; MAR represents the mean absolute error; y i represents the DFT calculation value obtained by calculation; represents the machine learning prediction value obtained by prediction; represents the average value of DFT calculation value; n is the sample size. The closer the values of MSR and MAR are to 0, the closer the value of R 2 is to 1, indicating that the prediction performance of the machine learning model is better.

[0050] The technical solutions of the application will be further described below through specific embodiments.

[0051] In each of the following embodiments, the method is a conventional method unless otherwise specified; the reagents and materials can be purchased on the market unless otherwise specified.

[0052] As Figure 7 , a method for screening a single-atom catalyst supported by g-C3N4, comprising the following steps:

[0053] Step 1: According to the two-dimensional structure of the g-C3N4 substrate, the doping site of the g-C3N4 substrate is obtained; the doping element is placed at the doping site of the g-C3N4 substrate, and the structure is optimized to screen out a pre-screening catalyst that meets the electron localization, and a pre-screening result is obtained.

[0054] The doping sites of the g-C3N4 substrate are located at substitution sites or interstitial sites, and the doping sites of the g-C3N4 substrate have 7, such as Figure 1 The doping sites of the g-C3N4 substrate are at least one of 2 C substitution sites, 3 N substitution sites, and 2 H interstitial sites. The doping element is a metal element or a non-metal element, wherein the metal element is at least one of Zr, Ti, Mn, Fe, Co, Ni, Cu, Zn, Ru, Rh, Pd, Pt, Al, and Li; and the non-metal element is at least one of B, Si, S, O, P, and Cl. The pre-screening catalyst has 140 kinds. The g-C3N4-supported monatomic catalyst is denoted as SAC@g-C3N4. g-C3N4 is an ideal carrier for monatomic catalysts and can be used to improve the performance of CO2 reduction reactions.

[0055] As shown in Figure 1 (a) of the drawing, the electron localization function diagram of the g-C3N4 substrate is used to reveal the local electron distribution characteristics of atoms or molecules. The full name of the electron localization function is Electron Localization Function, abbreviated as ELF. The value range of the electron localization function diagram is usually between 0 and 1, and different value ranges correspond to different electron localization. Among them, ELF=1 corresponds to complete localization, that is, the electron distribution is high, which usually corresponds to the red area. ELF=0.5 corresponds to the pair probability of the electron gas type, that is, the electron here is a bonding electron, which usually corresponds to the yellow area. ELF=0 corresponds to complete delocalization of electrons, that is, there is no electron here, which usually corresponds to the blue area. By analyzing the electron localization function diagram, the local distribution of valence electrons can be obtained directly, and then the bonding characteristics between atoms or molecules can be judged. For example, in a metal bond, the electron shows a relatively uniform distribution characteristic.

[0056] Due to its unique two-dimensional structure, especially the rich six-fold cavity and non-equivalent nitrogen atom distribution, g-C3N4 becomes an ideal carrier for monatomic catalysts. This structure provides stable anchoring sites for monatomic catalysts, ensures efficient immobilization and good electron localization, and at the same time, the large specific surface area of g-C3N4 also makes the catalytic reaction more active, which can effectively improve the selectivity of the reaction. As shown in Figure 1 (b) of the drawing, g-C3N4 has 5 substitution sites and 2 interstitial sites, and the 5 substitution sites are respectively denoted as C1, C2, N1, N2, and N3; and the 2 interstitial sites are respectively denoted as H1 and H2. The doping element is a metal element and a non-metal element.

[0057] According to the doping elements and the substitution sites and interstitial sites of g-C3N4, g-C3N4 single-atom catalysts of different doping types are screened out. According to the 5 substitution sites and 2 interstitial sites of g-C3N4, and in combination with 14 metal elements and 6 non-metal elements, 140 SAC@g-C3N4 catalysts of different doping types are systematically screened out, and the catalysts have good electron localization.

[0058] By doping metal and non-metal single atoms on the g-C3N4 substrate, the electronic structure and catalytic activity can be effectively regulated. The doping of metal elements can generally enhance the electron transfer ability of g-C3N4, accelerate the adsorption and conversion process of reactants on the catalyst surface. The doping of non-metal elements is mainly used to optimize the distribution of active sites and improve the selectivity and stability of the reaction. In addition, the doping elements can provide stable single-atom structure in the six-fold cavity of g-C3N4, prevent agglomeration, and reduce the energy barrier in the CO2 reduction reaction, ensuring efficient catalytic reaction. This single-atom catalyst based on the structure of g-C3N4 theoretically provides an efficient and low-energy solution for CO2 reduction reaction, and shows great potential for converting CO2 into high-value chemicals.

[0059] Step 2, calculate the formation energy of part of the pre-screening catalysts in the pre-screening result, collect key feature data related to the formation energy to establish a formation energy-key feature data set, and construct a first prediction model, so that the first prediction model can learn the nonlinear relationship between the key features and the formation energy; perform feature importance analysis on the first prediction model to generate a feature importance pie chart; perform significance analysis on the feature importance pie chart to determine significant key features; use the significant key features to screen out catalysts with negative formation energy to obtain the preliminary screening result. The specific method is as follows:

[0060] Set hyperparameters: according to experience or experiment, set the hyperparameters of the first XGBoost prediction model, such as n_estimators, which represents the number of trees; learning_rate, which represents the learning rate; max_depth, which represents the maximum depth of the tree; reg_alpha and reg_lambda, which represent the regularization parameters, etc. In the embodiment of the present application, these parameters are set to n_estimators=10, learning_rate=0.781, max_depth=5, reg_alpha=2 and reg_lambda=1.

[0061] The method for constructing the first XGBoost prediction model is: according to the formation energy-key feature dataset, using the root mean square error, the determination coefficient and the mean absolute error as evaluation indexes, the hyperparameters of the model, such as the depth of the tree, the learning rate, the number of weak learners, etc. are optimized, so as to learn the nonlinear relationship between the key features and the formation energy, and to construct and train the first XGBoost prediction model with the best prediction accuracy.

[0062] The key features related to the formation energy have 10, which are respectively the doping site, the atomic radius, denoted as Radii; the number of valence electrons, denoted as Out_e; the electron affinity, denoted as Aff; the chemical potential, denoted as Chemi_pot; the number of atoms, denoted as Atom_N; the electronegativity, denoted as N m ; the first ionization energy, denoted as I1; the atomic weight, denoted as Atom_W; the number of inner electrons, denoted as In_e, to construct the feature pool.

[0063] The formation energies of some catalysts calculated by the density functional theory are used, and the 10 key features related to the formation energy of some catalysts are collected to form a dataset. According to the ratio of 8:2, the dataset is divided into a training set and a test set. The training set is used to train the first XGBoost prediction model, and the test set is used to evaluate the prediction performance of the first XGBoost prediction model.

[0064] Model training: the first XGBoost prediction model is trained using the training set data through the calculation formulas of the root mean square error, the determination coefficient and the mean absolute error, so that it can learn the nonlinear relationship between the key features and the formation energy. The prediction performance of the first XGBoost prediction model is evaluated on the test set, and the determination coefficient, the mean absolute error and the root mean square error are calculated. These indexes are used to measure the accuracy and reliability of the model prediction.

[0065] Model evaluation: according to the first XGBoost prediction model constructed, the key features of the test set are used as the input independent variables, and the formation energy is used as the output dependent variable. The formation energy prediction value predicted by the first XGBoost prediction model is obtained through the first XGBoost prediction model constructed. And based on the feature importance pie chart obtained from the formation energy prediction value, the significance analysis is carried out, and the significant key features are obtained. As shown in the (c) figure of Figure 2 , the doping site and the atomic radius are identified as significant key features, accounting for 52.2% and 33.7% respectively.

[0066] In addition, the calculated value of the formation energy calculated by DFT is taken as the abscissa, and the formation energy prediction value predicted by the first XGBoost prediction model is taken as the ordinate, and the Figure 2The comparison results are shown in Figure (b). Based on the comparison results, the prediction ability of the first XGBoost prediction model is intuitively evaluated, and the reliability and generalizability of the model are verified.

[0067] In order to ensure the stability and high efficiency of SAC@g-C3N4 catalyst in long-term use, the embodiment of the present invention uses XGBoost model to accurately predict the formation energy. f The prediction shows extremely high accuracy and reliability. f The key features related to E were successfully established using the XGBoost model. f The nonlinear relationship between the values ​​was analyzed. Then, based on the XGBoost model's predictions and feature importance analysis, the structure of the SAC@g-C3N4 catalyst was optimized to improve its stability and performance. Based on feedback from actual applications and new data, the XGBoost model's hyperparameters and feature set were continuously adjusted and optimized to improve its prediction accuracy and generalization ability.

[0068] like Figure 2 As shown in Figure (a), XGB, ABR, KNR, LR, MLP, PLS, RR, SVR, RFR, GBR, etc. represent different machine learning algorithms or models. In the embodiment of the present invention, the XGBoost model is mainly used, referred to as XGB. Figure 2 Figure (b) shows the E predicted by the XGBoost model. f The value is consistent with the DFT calculated E f The comparison of the calculated values ​​shows that the data points are closely distributed along the diagonal, which shows that the XGBoost model has a highly accurate prediction ability. f The value has a higher precision.

[0069] R on the training and test sets 2, MAR and MSR scores are 0.9664 / 0.9120, 0.5218eV / 0.7354eV and 0.5403eV / 0.8348eV respectively, indicating that the XGBoost model has demonstrated excellent fitting ability and prediction accuracy when processing complex data. This excellent performance is attributed to the reasonable setting of hyperparameters such as n_estimators=10, learning_rate=0.781 and max_depth=5. These parameters help the model achieve a balance between complexity and overfitting risk. In addition, the regularization parameters reg_alpha=2 and reg_lambda=1 effectively suppress the complexity of the model, avoid overfitting, and thus improve the generalization ability of the model. Through these fine parameter tuning, the XGBoost model is able to capture E f The complex nonlinear relationships in the catalyst are stabilized and accurately predicted, ensuring the accurate modeling of the SAC@g-C3N4 catalyst.

[0070] In addition, the feature importance analysis of the XGBoost model further revealed that E f The main driving factors of Figure 2 As shown in Figure (c), the doping site and atomic radius are the factors that affect E f The two most critical features of the single-atom catalyst account for 52.2% and 33.7% respectively. This shows that the combination of different doping sites and atoms directly determines the stability and performance of the single-atom catalyst. f With its powerful prediction and feature importance analysis capabilities, the XGBoost model provides strong support for the structural optimization of SAC@g-C3N4 catalyst and its CO2RR application.

[0071] In step 2, the screening condition for catalyst screening using formation energy is: screening out catalysts with negative formation energy, indicating that the system is thermodynamically stable. The screening result is: from the 140 pre-screened catalysts in step 1, 35 thermodynamically stable catalysts were obtained through the formation energy screening in step 2. The embodiment of the present invention performs a significance analysis based on the feature importance pie chart predicted by the first XGBoost prediction model to obtain significant key features; the prediction results of the predicted catalysts are obtained by the first XGBoost prediction model, and compared with the formation energy calculation results obtained by density functional theory calculations to verify the reliability and generalizability of the model.

[0072] Step 3, calculate the adsorption free energy change of all catalysts in the primary screening results, collect the core feature data related to the adsorption free energy change to establish the adsorption free energy change-core feature data set, and construct a second prediction model, so that the second prediction model can learn the nonlinear relationship between the key features and the adsorption free energy change; calculate the SHAP value according to the training result of the second prediction model, and perform feature marginal contribution analysis on the core features according to the SHAP value to determine the significant core features; based on the second prediction model, the significant core features are used to predict the adsorption free energy change and the limiting potential of the primary screening catalyst, which is used to evaluate the CO2 reduction reaction catalytic activity of the primary screening catalyst, and the catalyst with an absolute value of adsorption free energy change ≤0.8 eV and a limiting potential ≤0.8 V is screened out to obtain the secondary screening result.

[0073] Step 3.1, construct a second prediction model, so that the second prediction model can learn the nonlinear relationship between the key features and the adsorption free energy change.

[0074] The method of constructing the second XGBoost prediction model is: according to the adsorption free energy change-core feature data set, using root mean square error, determination coefficient and mean absolute error as evaluation indexes, the hyperparameters of the model, such as tree depth, learning rate, weak learner number, etc. are optimized, so as to learn the nonlinear relationship between the core features and the adsorption free energy change, and to construct and train the second XGBoost prediction model with high interpretability. The specific method is as follows:

[0075] Collect 11 core feature data related to the adsorption free energy change of the 35 SAC@g-C3N4 catalysts screened in step 2, which are divided into single atom feature data and catalyst feature data. Among them, the single atom features include atomic radius, denoted as Radii; atomic number, denoted as Atom_N; single atom chemical potential, denoted as μ; valence electron, denoted as Out_e; first ionization energy, denoted as I1; and electron affinity, denoted as Aff. The catalyst features include charge transfer amount, denoted as bader; binding energy, denoted as E b ; formation energy, denoted as E f ; Fermi level, denoted as E fermi ; band gap, denoted as E g .

[0076] The adsorption free energy change of 35 SAC@g-C3N4 catalysts is calculated by density functional theory as the target variable; 11 core feature data related to the adsorption free energy change of all preliminary screening catalysts are collected to form an adsorption free energy change-core feature data set, and the data set is divided into a training set and a test set according to a ratio of 8:2. The training set is used to train the second XGBoost prediction model, and the model parameters are adjusted to optimize the prediction performance. The test set is used to evaluate the prediction performance of the second XGBoost prediction model.

[0077] Model training: The second XGBoost prediction model is trained using the training set data through the calculation formulas of root mean square error, determination coefficient and mean absolute error, so that it can learn the nonlinear relationship between core features and adsorption free energy change. As shown in (a) and (b) of FIG. 1, Figure 4 The R values of the second XGBoost prediction model on the training set and the test set are 0.9664 and 0.9120 respectively, which are obviously better than those of other models, and show higher prediction accuracy and stability. 2

[0078] Model evaluation: The prediction performance of the first XGBoost prediction model is evaluated on the test set, and the determination coefficient, mean absolute error and root mean square error are calculated to ensure that the second XGBoost prediction model has good prediction accuracy and stability on the training set and the test set.

[0079] According to the model evaluation results of the second XGBoost prediction model, a feature importance pie chart is obtained; the feature importance pie chart is analyzed for significance to obtain significant core features. As shown in (c) of FIG. 1, Figure 4 Through the feature importance pie chart of the second XGBoost prediction model, four significant core features affecting the CO2RR activity are revealed: band gap, binding energy, atomic number and atomic radius. Among them, the band gap of 32.4% and the binding energy of 24.1% have the greatest influence on the thermodynamic stability and electronic transmission capacity. Smaller band gap makes electron transition and conduction more smooth, effectively improving catalytic efficiency, and binding energy balances the adsorption of reactants and the desorption of products, optimizing the reaction process.

[0080] Step 4.2, according to the results of training the second prediction model, the SHAP value is calculated, and the core features are analyzed for feature marginal contribution according to the SHAP value to determine the significant core features.

[0081] In the embodiment of the application, the SHAP method is used to calculate the marginal contribution value of each core feature to the prediction result, that is, the SHAP value, as shown in (a) of FIG. 2, Figure 5 ​and draw SHAP summary plots to show the significance of each core feature on the prediction result. Analyze the distribution and position of different features in the SHAP summary plot to determine the influence of each core feature on the prediction of adsorption free energy. According to the positive and negative and size of SHAP value, explain the positive or negative influence of each feature on the prediction result. Analyze the mechanism of different features in the process of CO2 adsorption, hydrogenation catalysis and desorption.

[0082] The embodiment of the present application combines the second XGBoost prediction model with the SHAP method to obtain the specific marginal contribution of each core feature in the catalytic process, which is used to accurately predict the CO2 reduction reaction activity of SAC@g-C3N4 catalysts, and further screen the type of dopant. At the same time, the catalytic performance of the screened catalysts is evaluated and verified to obtain the accuracy of the above screening method. The embodiment of the present application successfully predicts the change amount of adsorption free energy of 35 SAC@g-C3N4 catalysts, and effectively evaluates the CO2RR activity.

[0083] In order to more accurately quantify the contribution of each feature to the prediction result, as Figure 5 , the significance of 11 core features closely related to the geometric and electronic properties affecting the adsorption free energy is evaluated through the SHAP summary plot. In Figure 5 , all features are arranged in descending order according to the sum of SHAP values, and the distribution of SHAP values is displayed on the horizontal axis. Each point represents a catalyst, and the color of the point represents the numerical size of the corresponding catalyst feature, and the position of the point represents the SHAP value of the corresponding catalyst feature. It is worth noting that among all the features, the binding energy, atomic radius, ionization energy and band gap are usually located at the top of the summary plot and have a wide distribution, which indicates that they have a great influence on the prediction of adsorption free energy.

[0084] In the adsorption process of CO2, as Figure 5 (a) figure, the larger binding energy has a positive SHAP value, which means that the binding energy of the atom is positively correlated with the predicted adsorption free energy value. The binding energy reflects the interaction strength between the catalyst surface and the reactant, so the larger binding energy is consistent with the higher predicted adsorption free energy value.

[0085] In the hydrogenation catalysis process of CO2, as Figure 5In Figures (b) and (c), smaller ionization energies have negative SHAP values, which means that the ionization energy of the atom is positively correlated with the predicted adsorption free energy value. The ionization energy reflects the ability of the catalyst to lose electrons. Higher ionization energies are not conducive to electron transfer and thus hinder the catalytic hydrogenation of CO2, resulting in a larger ΔG. Similarly, larger binding energy, chemical potential, and atomic radius correspond to negative SHAP values, indicating that there is a negative correlation between these parameters and the predicted adsorption free energy value. This means that enhanced binding energy and increased atomic radius may lead to a significant increase in the stability of the reactant on the catalyst surface, thereby reducing its adsorption free energy.

[0086] In addition, during the desorption process of CO2, Figure 5 Figure (d) shows an opposite relationship with CO2 adsorption, with larger binding energy corresponding to negative SHAP values, indicating that weaker interactions make CO more easily released on the catalyst surface, thereby improving the desorption efficiency. It is worth noting that, as Figure 5 , the band gap values ​​of purple and dark blue balls appear in the region where ΔG is close to 0, which indicates that a moderate band gap is more conducive to the transmission of electrons, thereby reducing ΔG and promoting the occurrence of CO2RR.

[0087] In step 4.3, based on the second prediction model, the significant core features are used to predict the adsorption free energy change and limiting potential of the primary screened catalysts, which are used to evaluate the catalytic activity of the primary screened catalysts in the CO2 reduction reaction, and to screen out catalysts with an absolute value of the adsorption free energy change ≤0.8eV and a limiting potential ≤0.8V to obtain the re-screening results.

[0088] In this example, the catalytic activity of 35 thermodynamically stable g-C3N4 single-atom catalysts in the CO2 reduction reaction was evaluated to further screen the type of dopant. When evaluating the catalytic activity of the CO2 reduction reaction, the change in adsorption free energy is a commonly used important indicator because it can reliably describe the activity of different materials. The specific method is:

[0089] Draw a reaction mechanism diagram: Figure 3 Figure (a) shows the mechanism and reaction pathway of the CO2 reduction reaction on the SAC@g-C3N4 catalyst surface. The figure clearly identifies the four key steps of CO2 adsorption, COOH intermediate formation, CO formation, and CO desorption. The changes in adsorption free energy are designated as ΔG1, ΔG2, ΔG3, and ΔG4, respectively.

[0090] CO2 adsorption: CO2 molecules bind to the active sites on the surface of SAC@g-C3N4 catalyst, forming adsorbed CO2. This process can be represented as CO2 + * + 2(H + +e - ) → CO2* + 2(H + +e - ); where * represents the active sites on the surface of SAC@g-C3N4 catalyst that can bind to CO2. COOH intermediate formation: Adsorbed CO2 is reduced to COOH intermediate through hydrogenation reaction on the surface of SAC@g-C3N4 catalyst. The specific reaction is CO2* + 2(H + +e - ) → COOH* + (H + +e - ). CO formation: COOH intermediate continues to be reduced to CO through hydrogenation reaction. The reaction equation is COOH* + (H + +e - ) → CO* + H2O(l). CO desorption: The generated CO desorbs from the surface of SAC@g-C3N4 catalyst. The expression is CO* + H2O → CO + * + H2O(l).

[0091] Calculate the change in adsorption free energy: For each key step, calculate the corresponding change in adsorption free energy, denoted as ΔG1, ΔG2, ΔG3, and ΔG4, respectively. These changes in adsorption free energy are calculated by density functional theory or obtained through experimental measurements.

[0092] Analyze the change in adsorption free energy: Analyze the change in adsorption free energy of each key step to determine whether the binding strength of CO2 to the surface of SAC@g-C3N4 catalyst is moderate. According to the Sabatier principle, a catalyst with an absolute value of ΔG close to 0 is selected to ensure both effective adsorption of CO2 and smooth release of generated CO. During the CO2 reduction reaction process, the binding strength of CO2 to the surface of SAC@g-C3N4 catalyst is crucial for each reaction step. If the binding strength is too weak, the initial adsorption and activation will become difficult, while if the binding strength is too strong, the active sites will be occupied, thereby hindering the release of CO. Therefore, an ideal CO2 reduction reaction catalyst should follow the Sabatier principle; the Sabatier principle is also known as the methanation principle. This means that the adsorption of CO2 should be close to thermoneutrality, i.e., the absolute value of ΔG should be as close to 0 as possible. Only in this case, the SAC@g-C3N4 catalyst can effectively adsorb CO2 and smoothly release the generated CO, thereby achieving efficient CO2 reduction reaction.

[0093] Drawing the limiting potential diagram: according to the change amount of free energy of the potential determining step, the limiting potential is calculated to evaluate the CO2 reduction reaction activity. The calculation formula of the limiting potential is: UL = -ΔG max / e, wherein UL represents the limiting potential; ΔG max represents the maximum value of the change amount of adsorption free energy; e represents the charge amount of an electron. The drawn limiting potential diagram is shown in FIG. Figure 3 (b) of the accompanying drawings, which shows the limiting potential values of different SAC@g-C3N4 catalysts. Comparing the limiting potential: in the limiting potential diagram, the UL values of different SAC@g-C3N4 catalysts are compared. The catalysts with lower UL values are selected because these catalysts have lower limiting potential when performing the CO2 reduction reaction, and require less additional energy.

[0094] Screening the candidate dopant: according to the conditions that the absolute value of ΔG is close to 0 and the UL value is low, the candidate dopant with excellent catalytic activity and higher reaction efficiency is screened out. Figure 3 FIG. (b) of the accompanying drawings shows the UL values of 35 SAC@g-C3N4 catalysts. It can be clearly seen that when the |ΔG| value is less than 0.8 eV, 11 candidate dopants can be screened out, which are Zr(H1), Zr(H2), Ti(H1), Ti(H2), Co(H1), Ni(H2), Si(H1), P(H1), P(H2), Al(H1) and Al(H2). It can be clearly seen from FIG. Figure 3 (b) of the accompanying drawings that the overpotential of the 11 candidate dopants is low, which means that these dopants require less additional energy when performing the CO2 reduction reaction, and therefore have excellent catalytic activity and higher reaction efficiency. The present application performs CO2 reduction reaction screening, and the screening condition is: according to the absolute value of the change amount of adsorption free energy, the catalysts with ΔG absolute value ≤0.8 eV and limiting potential ≤0.8 V are screened out. This condition indicates that the catalysts meet the Sabatier principle in the CO2RR process, and have effective adsorption and desorption behavior. Screening result: from the 35 catalysts, 11 catalysts are obtained through ΔG absolute value and limiting potential screening.

[0095] Step 4: The catalysts in the re-screening result are evaluated for hydrogen evolution reaction catalytic activity to screen out catalysts suitable for synthesis gas production, CO2 reduction reaction or hydrogen evolution reaction.

[0096] The method for screening the catalysts suitable for synthesis gas production, CO2 reduction reaction or hydrogen evolution reaction is: according to the competitive relationship of the adsorption free energy change amount of the CO2 reduction reaction and the hydrogen evolution reaction, the catalysts in the complex screening results are subjected to hydrogen evolution reaction catalytic activity evaluation, CO2 reduction reaction activity evaluation and synthesis gas production suitability analysis; the adsorption free energy change amount of the CO2 reduction reaction is denoted as ΔG_CO2RR; the adsorption free energy change amount of the hydrogen evolution reaction is denoted as ΔG_H; when the ΔG_CO2RR of the catalyst is lower than the ΔG_H, the catalyst is a CO2 reduction reaction catalyst that tends to promote the CO2 reduction reaction; when the ΔG_H of the catalyst is lower than the ΔG_CO2RR, the catalyst is a hydrogen evolution catalyst that tends to promote the hydrogen evolution reaction; when the ΔG_CO2RR of the catalyst is close to the ΔG_H, the catalyst is a synthesis gas catalyst that has the functions of promoting the CO2 reduction reaction and the hydrogen evolution reaction.

[0097] As Figure 6 , some catalysts exhibit high conversion ability in CO2RR, some catalysts show excellent catalytic activity in HER, and some catalysts show excellent catalytic performance in both CO2RR and HER.

[0098] Screening results: according to the HER performance, the remaining 11 single-atom catalysts are divided into three categories: one category is a catalyst suitable for synthesis gas production, which has the catalytic activity of CO2RR and HER; one category is a high-efficiency catalyst for CO2 reduction reaction, which is mainly used for high-efficiency catalysis of CO2 reduction reaction; and the other category is a catalyst that is biased towards hydrogen evolution reaction, which is more suitable for the catalytic performance of hydrogen reaction. These results provide a clear direction for the development of catalysts with different catalytic activities, and are suitable for different industrial application scenarios of carbon dioxide emission reduction, clean energy conversion and synthesis gas production.

[0099] The above are only preferred embodiments of the present application and are not used to limit the present application, and any modification, equivalent replacement and improvement made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for screening g-C3N4-supported single-atom catalysts, characterized in that: The following steps are involved: Based on the two-dimensional structure of the g-C3N4 substrate, the doping sites of the g-C3N4 substrate are obtained; the doping elements are placed at the doping sites of the g-C3N4 substrate and the structure is optimized to screen out catalysts that meet the electron localization requirements and obtain pre-screening results; Calculating the formation energy of some catalysts in the pre-screening results, collecting key feature data related to the formation energy to establish a formation energy-key feature data set, and constructing a first prediction model so that the first prediction model can learn the nonlinear relationship between the key features and the formation energy; Perform feature importance analysis on the first prediction model and generate a feature importance pie chart; Perform significance analysis on the feature importance pie chart to identify significant key features; Using significant key features, we screened out catalysts with negative formation energy and obtained preliminary screening results; Calculating the adsorption free energy changes of the catalysts in all primary screening results, collecting core feature data related to the adsorption free energy changes to establish an adsorption free energy change-core feature dataset, and constructing a second prediction model that can learn the nonlinear relationship between the key features and the adsorption free energy changes; Calculate the SHAP value based on the results of the second prediction model training, and perform feature marginal contribution analysis on the core features based on the SHAP value to determine the significant core features; Based on the second prediction model, the significant core features were used to predict the adsorption free energy change and limiting potential of the pre-screened catalysts. These were used to evaluate the catalytic activity of the pre-screened catalysts for the CO2 reduction reaction. Catalysts with an absolute value of adsorption free energy change ≤ 0.8 eV and a limiting potential ≤ 0.8 V were selected to obtain the re-screening results. The catalytic activity of the catalysts in the re-screening results is evaluated for hydrogen evolution reaction to screen out catalysts suitable for synthesis gas production, CO2 reduction reaction or hydrogen evolution reaction.

2. The method for screening g-C3N4-supported single-atom catalysts according to claim 1, characterized in that: The doping sites of g-C3N4 substrate are 2 C substitution sites, 3 N substitution sites and 2 H interstitial sites.

3. The method for screening g-C3N4-supported single-atom catalysts according to claim 1, characterized in that: The doping element is at least one of Zr, Ti, Mn, Fe, Co, Ni, Cu, Zn, Ru, Rh, Pd, Pt, Al, Li, B, Si, S, O, P and Cl.

4. The method for screening g-C3N4-supported single-atom catalysts according to claim 1, characterized in that: The key features related to formation energy are doping site, atomic radius, number of valence electrons, electron affinity, chemical potential, atomic number, electronegativity, first ionization energy, atomic weight and number of internal electrons; the significant key features are doping site and atomic radius.

5. The method for screening g-C3N4-supported single-atom catalysts according to claim 1, characterized in that: The method for constructing the first prediction model is: Based on the formation energy-key feature dataset, the root mean square error, determination coefficient and mean absolute error are used as evaluation indicators to optimize the model's hyperparameters, thereby learning the nonlinear relationship between key features and formation energy, and constructing and training the first prediction model with the best prediction accuracy.

6. The method for screening g-C3N4-supported single-atom catalysts according to claim 1, characterized in that: The method for constructing the second prediction model is: Based on the adsorption free energy change-core feature dataset, the root mean square error, determination coefficient and mean absolute error are used as evaluation indicators to optimize the model's hyperparameters, thereby learning the nonlinear relationship between the core features and the adsorption free energy change, and constructing and training a second prediction model with high interpretability.

7. The method for screening g-C3N4-supported single-atom catalysts according to claim 1, characterized in that: The core features related to the change in adsorption free energy include single-atom features and catalyst features; The single-atom characteristics are atomic radius, atomic number, single-atom chemical potential, valence electrons, first ionization energy and electron affinity; the catalyst characteristics are charge transfer amount, binding energy, formation energy, Fermi level and band gap; the significant core characteristics are band gap and binding energy.

8. The method for screening g-C3N4-supported single-atom catalysts according to claim 1, characterized in that: The method for screening catalysts suitable for synthesis gas production, CO2 reduction reaction or hydrogen evolution reaction is: Based on the competitive relationship between the adsorption free energy changes of the CO2 reduction reaction and the hydrogen evolution reaction, the catalysts selected in the re-screening results were evaluated for their catalytic activity in the hydrogen evolution reaction, their activity in the CO2 reduction reaction, and their suitability for synthesis gas production. The adsorption free energy change of CO2 reduction reaction is recorded as ΔG_CO2RR; the adsorption free energy change of hydrogen evolution reaction is recorded as ΔG_H; When the ΔG_CO2RR of the catalyst is lower than ΔG_H, the catalyst is a CO2 reduction reaction catalyst that tends to promote the CO2 reduction reaction; When the ΔG_H of the catalyst is lower than the ΔG_CO2RR, the catalyst is a hydrogen evolution catalyst that tends to promote the hydrogen evolution reaction; When the ΔG_CO2RR of the catalyst is close to ΔG_H, the catalyst is a synthesis gas catalyst that promotes both the CO2 reduction reaction and the hydrogen evolution reaction.

9. A system for screening g-C3N4-supported single-atom catalysts, characterized in that: The system comprises: processor; a memory having stored thereon a computer program executable on the processor; Wherein, when the computer program is executed by the processor, the steps of the method for screening g-C3N4-supported single-atom catalysts as described in any one of claims 1 to 8 are implemented.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a data processing program, which, when executed by a processor, implements the steps of the method for screening g-C3N4-supported single-atom catalysts as described in any one of claims 1 to 8.

Images