A method, device and medium for predicting the photocatalytic water splitting performance of 2D-COF materials

Abstract

The present invention discloses a method, device and medium for predicting the performance of photocatalytic water splitting of 2D-COF materials, which relate to the fields of catalytic technology and material information technology. The method comprises: first, converting the material structure into crystal structure data represented by atomic species and coordinates. In view of the large number of 2D-COF material types, a large number of material structure data files are produced using automated file generation technology. Then, material structure characteristic data are generated through experimental testing. Next, the structural data file is subjected to first-principles calculations to generate performance characteristic data. Finally, machine learning technology is used to realize the prediction of the catalytic properties of 2D-COF materials. Since the catalytic performance indicators of photocatalytic water splitting include three types of indicators for distinguishing material electronic structure values, reaction free energy values ​​and overpotential values, the above process can be repeated to perform three predictions, and excellent photocatalytic water splitting 2D-COF materials can be screened out.

Description

Technical Field

[0001] The present invention relates to the fields of catalytic technology and material information technology, and in particular to a method, device and medium for predicting the photocatalytic water splitting performance of a 2D-COF material. Background Art

[0002] Today, human society is facing severe challenges such as energy shortages and global warming. To address the energy shortages caused by the non-renewable combustion of fossil fuels and the environmental damage caused by their combustion products, the development of clean, renewable energy alternatives to fossil fuels is particularly urgent. Hydrogen energy is a highly promising clean energy source, producing only water as a combustion product, without causing environmental pollution. The development and utilization of hydrogen energy can not only alleviate the energy crisis but also effectively reduce greenhouse gas emissions, thereby slowing global warming. Industrial hydrogen production is primarily achieved through natural gas reforming, but this method still relies on fossil fuels and cannot fundamentally solve the energy shortage problem.

[0003] Two-dimensional covalent organic frameworks (2D-COFs) are multifunctional materials with crystalline properties and inherent porosity. They possess advantages such as large specific surface area, low backbone density, strong stability, designable structure, and high carrier transport rates. These advantages have led to their widespread application in chemical energy fields such as catalysis, gas adsorption, sensors, energy conversion, and storage, making them potential energy carriers to replace fossil fuels. 2D-COF materials exhibit diverse microstructures and can be composed of building blocks of various symmetries, sizes, and functions connected by chemical bonds. Although a large number of 2D-COF materials have been studied in experimental synthesis, performance testing, and catalytic applications, there is still a lack of low-cost, efficient methods and technologies for predicting the catalytic properties of 2D-COF materials. Traditional catalytic performance prediction methods have common shortcomings and limitations, such as long cycles, high costs, and high energy consumption, which restrict the advancement of prediction work on excellent catalysts and thus slow down the discovery of catalysts.

[0004] Traditional computational, experimental, and machine learning approaches to predict material properties rely on analyzing the material's structure and imposing reasonable initial conditions to predict its characteristics and properties. These methods are primarily based on prior knowledge and technical proficiency, but these methods are often limited by high costs and operator subjectivity, limiting the discovery of superior 2D-COF materials. Summary of the Invention

[0005] The purpose of the present invention is to provide a method, device and medium for predicting the photocatalytic water splitting performance of 2D-COF materials, which can solve the problems of insufficient existing 2D-COFs data, high cost of catalytic performance prediction, long operation cycle and single results.

[0006] To achieve the above object, the present invention provides the following solutions:

[0007] A method for predicting the photocatalytic water splitting performance of a 2D-COF material, comprising:

[0008] Establish a 2D-COF photocatalytic water splitting thermodynamic performance data set I; the 2D-COF photocatalytic water splitting thermodynamic performance data set I is obtained by screening 2D-COF structure CIF file 1 and 2D-COF structure CIF file 2; the data in the 2D-COF structure CIF file 1 are randomly and uniformly sampled from the 2D-COF structure CIF total file; the data in the 2D-COF structure CIF file 2 are the remaining samples not sampled in the 2D-COF structure CIF total file;

[0009] According to the kinetic judgment conditions of the photocatalytic water splitting performance, the 2D-COF photocatalytic water splitting thermodynamic performance data set I was screened to obtain 2D-COF structures without photocatalytic performance and 2D-COF structures with photocatalytic performance, and the 2D-COF structure with photocatalytic performance was determined as 2D-COF structure CIF file 3;

[0010] Establish a 2D-COF photocatalytic water splitting thermodynamic performance data set II; the 2D-COF photocatalytic water splitting thermodynamic performance data set II is obtained by screening 2D-COF structure CIF file 4 and 2D-COF structure CIF file 5; the data in the 2D-COF structure CIF file 4 are randomly and uniformly sampled from the 2D-COF structure CIF file 3; the data in the 2D-COF structure CIF file 5 are the remaining samples not sampled in the 2D-COF structure CIF file 3;

[0011] The 2D-COF photocatalytic water splitting thermodynamic performance data set II was subjected to secondary screening, and the data that met the secondary screening conditions were determined to have excellent photocatalytic water splitting performance 2D-COF; the secondary screening conditions included condition 1 and condition 2; the condition 1 was the free energy value ΔG of the 2D-COF hydrogen evolution reaction. H* Is the absolute value less than the photoelectric potential energy value eU? e , the free energy value of 2D-COF oxygen evolution reaction ΔG OH* Is the absolute value less than the photoelectric potential energy value eU? h ; The condition 1 is the 2D-COF hydrogen evolution reaction overpotential energy value η HER Is the value less than the photoelectric potential energy value eU? e , the potential energy value of the oxygen evolution reaction of 2D-COF η OER Is the value less than the photoelectric potential energy value eU? h .

[0012] Optionally, the method for establishing the 2D-COF photocatalytic water splitting thermodynamic performance data set I includes:

[0013] The 2D-COF structure CIF file 1 is processed using an automated input file generation program to generate an electronic structure data set 1-A, and structural features and chemical features are collected to obtain a structural feature data set 1-B and a chemical feature data set 1-C, respectively;

[0014] Aggregating the electronic structure dataset 1-A, the structural feature dataset 1-B, and the chemical feature dataset 1-C, and performing machine learning on the aggregated 2D-COF catalytic performance total dataset 1-D using a GBR algorithm to obtain a prediction model; the prediction model includes a mathematical model A for predicting catalytic performance and a mathematical model B for predicting catalytic performance;

[0015] Collecting structural features and chemical features of the 2D-COF structure CIF file 2 to obtain a structural feature dataset 1-E and a chemical feature dataset 1-F, respectively;

[0016] Inputting the structural feature dataset 1-E and the chemical feature dataset 1-F as initial data into the catalytic performance prediction mathematical model A to generate an electronic structure dataset 1-G; and inputting the structural feature dataset 1-E and the chemical feature dataset 1-F as initial data into the catalytic performance prediction mathematical model B to generate an electronic structure dataset 1-H;

[0017] The 2D-COF photocatalytic water splitting thermodynamic performance data set I was constructed based on the electronic structure data set 1-A, the electronic structure data set 1-G and the electronic structure data set 1-H.

[0018] Optionally, the kinetic judgment conditions for the photocatalytic water splitting performance specifically include: the conduction band bottom energy level value E in the 2D-COF band structure CBM Is it greater than -4.44, the value of the valence band top energy level E in the 2D-COF band structure? VBM Is it less than -5.67, and E CBM -E VBM Check whether the value is between 2.2 and 3.26.

[0019] Optionally, the judgment process of the secondary screening is:

[0020] Based on the thermodynamic performance data set II of 2D-COF photocatalytic water splitting, the absolute value of the Gibbs free energy of hydrogen evolution |ΔG H* | and the absolute value of the Gibbs free energy of oxygen evolution |ΔG OH*Whether the value of | satisfies the photocatalytic complete water splitting kinetic conditions at the same time: |ΔG H* |<E CBM +4.44 and |ΔG OH* |<E VBM -5.67;

[0021] If yes, then filter out the data that meets the conditions and create a 2D-COF structure CIF file 6;

[0022] Determine again the hydrogen evolution overpotential η of each structure of the 2D-COF structure CIF file 6 HER and oxygen evolution overpotential η OER Whether the value satisfies the conditions for excellent photocatalytic complete water splitting kinetics: η HER <E CBM +4.44 and η OER <E VBM -5.67;

[0023] If so, it is determined to be a 2D-COF with excellent photocatalytic complete water splitting performance.

[0024] Optionally, the generation process of the electronic structure dataset 1-A is:

[0025] The 2D-COF structure CIF file 1 is automatically generated through an input file generation program to automatically generate scientific calculation input files for the material calculation software VASP in large quantities;

[0026] Submit the scientific calculation input file in the server directory and run the material calculation software VASP, and obtain a large number of output files after the calculation is completed;

[0027] The output file generated after the calculation is run through the automated electronic structure data reading program to automatically read the electronic structure data in large quantities and generate the electronic structure data set 1-A.

[0028] Optionally, the electronic structure data set 1-A includes: the valence band top energy level value E in the 2D-COF band structure VBM , the conduction band bottom energy level value E in the 2D-COF band structure CBM And the band gap value E in the 2D-COF band structure g .

[0029] The present invention also provides an electronic device comprising a memory and a processor, wherein the memory is used to store a computer program, and the processor runs the computer program to enable the electronic device to execute the above-mentioned method for predicting the photocatalytic water splitting performance of the 2D-COF material.

[0030] The present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the method for predicting the photocatalytic water splitting performance of the 2D-COF material as described above.

[0031] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0032] The present invention discloses a method, device and medium for predicting the photocatalytic complete water splitting performance of 2D-COF materials. The method comprises: first, converting the material structure into crystal structure data represented by atomic species and coordinates. In view of the large number of 2D-COF material types, a large number of material structure data files are produced using automated file generation technology. Then, material structure characteristic data are generated through experimental testing. Next, the structure data file is subjected to first-principles calculations to generate performance characteristic data. Finally, machine learning technology is used to predict the catalytic properties of 2D-COF materials. Since the catalytic performance indicators of photocatalytic complete water splitting include three types of indicators for distinguishing material electronic structure values, reaction free energy values ​​and overpotential values, the above process can be repeated to perform three predictions, and excellent photocatalytic complete water splitting 2D-COF materials can be screened out. Therefore, the present invention can solve the problems of too little existing 2D-COFs data, high cost of catalytic performance prediction, long operation cycle and single results. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0034] Figure 1 Generate a flow chart for the thermodynamic performance dataset I of 2D-COF photocatalytic water splitting in this example;

[0035] Figure 2 A flow chart for generating a 2D-COF structure library that satisfies the thermodynamic conditions for photocatalytic water splitting in this embodiment;

[0036] Figure 3 Generate a flow chart for the 2D-COF photocatalytic water splitting kinetic performance data set II in this example;

[0037] Figure 4 Flowchart for obtaining the 2D-COF structure library that meets the photocatalytic kinetic conditions and the excellent photocatalytic complete water splitting 2D-COF structure in this embodiment;

[0038] Figure 5 This is a flow chart of GBR machine learning training in this embodiment;

[0039] Figure 6 A diagram showing the energy band structure distribution of some 2D-COFs predicted to have photocatalytic water splitting capabilities in this embodiment;

[0040] Figure 7 Figure 2 shows the energy band structure distribution of some 2D-COFs predicted to have photocatalytic water splitting capabilities in this embodiment;

[0041] Figure 8 Schematic diagram of the fitting results of multiple machine learning algorithms in this embodiment. DETAILED DESCRIPTION

[0042] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0043] The purpose of the present invention is to provide a method, device and medium for predicting the photocatalytic water splitting performance of 2D-COF materials, which can solve the problems of insufficient existing 2D-COFs data, high cost of catalytic performance prediction, long operation cycle and single results.

[0044] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0045] like Figures 1-8 As shown, the present invention provides a method for predicting the photocatalytic water splitting performance of 2D-COF materials, and specifically discloses the following processing steps:

[0046] In the first step, a small sample is randomly and uniformly extracted from the total structure file library consisting of the 2D-COF material crystal structure file CIF to construct file library 1;

[0047] The extracted structural data sets are used to automatically generate input files for the material calculation software VASP in large quantities through an automatic input file generation program;

[0048] Submit the above scientific calculation input files in the server directory and run the material calculation software VASP. After the calculation is completed, a large number of output files are obtained;

[0049] The output file generated after the calculation is run through the automated electronic structure data reading program to automatically read the electronic structure data in large quantities and generate the electronic structure data set 1-A;

[0050] Collect the material structure features in the small sample database to obtain the material structure feature dataset 1-B;

[0051] Collect the material chemical features in the small sample database to obtain the material chemical feature dataset 1-C;

[0052] The electronic structure data set 1-A includes: the valence band top energy level value E in the 2D-COF band structure VBM , the conduction band bottom energy level value E in the 2D-COF band structure CBM , the band gap value E in the 2D-COF band structure g ;

[0053] The material structure feature dataset 1-B includes: the number of N atoms, the number of C atoms, the number of H atoms, the number of oxygen atoms, the total number of non-H atoms in the unit cell, the total number of atoms in the unit cell, the number of non-H atoms in the node, the number of node atoms, the number of non-H atoms in the linker, the number of linker atoms, the number of non-H atoms in the covalent bond, the number of covalent bond atoms, the number of N atoms in the node, the number of N atoms in the linker, the number of C atoms in the node, the number of C atoms in the linker, the number of O atoms in the node, the number of O atoms in the linker, the number of N atoms in the covalent bond, and the number of O atoms in the covalent bond.

[0054] The material chemical characteristic data set 1-C includes: total period number, total unit cell mass, average node mass, average connector mass, number of covalent bond C, average covalent bond mass, total unit cell electronegativity, average node electronegativity, connector electronegativity, covalent bond electronegativity, average node p z Number of electrons, average p of linker z Number of electrons, covalent bond p z Number of electrons, total lone pairs and conjugated electrons, node 1 polarization area, node 2 polarization area, linker molecule polarization area, covalent bond individual molecule polarization area, unit cell total polarization area;

[0055] Furthermore, the above electronic structure dataset 1-A, material structure dataset 1-B, and material chemistry dataset 1-C were aggregated to obtain the 2D-COF catalytic performance total dataset 1-D;

[0056] The catalytic performance dataset 1-D includes: the number of N atoms, the number of C atoms, the number of H atoms, the number of oxygen atoms, the total number of non-H atoms in the unit cell, the total number of atoms in the unit cell, the number of non-H atoms in the node, the number of node atoms, the number of non-H atoms in the linker, the number of linker atoms, the number of non-H atoms in the covalent bond, the number of covalent bond atoms, the number of N atoms at the node, the number of N atoms in the linker, the number of C atoms at the node, the number of C atoms in the linker, the number of O atoms at the node, the number of O atoms in the linker, the number of N atoms in the covalent bond, the number of covalent bond O atoms, the total number of periods, the total mass of the unit cell, the average mass of the node, the average mass of the linker, the number of covalently bonded C atoms, the average mass of the covalent bond, the total electronegativity of the unit cell, the average electronegativity of the node, the electronegativity of the linker, the electronegativity of the covalent bond, the average p-value of the node z Number of electrons, average p of linker z Number of electrons, covalent bond p z Number of electrons, total number of lone pairs and conjugated electrons, node 1 polarization area, node 2 polarization area, linker molecule polarization area, covalent bond individual molecule polarization area, unit cell total polarization area, valence band top energy level value E in 2D-COF band structure VBM , the conduction band bottom energy level value E in the 2D-COF band structure CBM , the band gap value E in the 2D-COF band structure g .

[0057] Furthermore, the GBR algorithm was used to perform machine learning on the 2D-COF catalytic performance dataset 1-D to obtain the predictable valence band top energy level E in the 2D-COF band structure. VBM Mathematical model A and the valence band top energy level E in the 2D-COF band structure VBM Predictable mathematical model B.

[0058] Gradient Boosting Regression (GBR) is a powerful machine learning algorithm for regression problems. Based on the gradient boosting framework, it builds a robust model by integrating multiple weak learners (typically decision trees). GBR has shown excellent performance in practical applications, particularly when dealing with nonlinear relationships and high-dimensional data.

[0059] The core idea of ​​GBR is to train multiple weak learners (such as decision trees) in sequence, with each new learner trying to correct the errors of the previous learner. The specific steps are as follows:

[0060] Initialize the model: Initialize with a simple model (such as a constant model).

[0061] Calculate residuals: Calculate the error (residual) between the predicted value of the current model and the true value.

[0062] Train a weak learner: Use the residual as the target variable and train a new weak learner.

[0063] Update model: Add the newly trained weak learners to the existing model and update the model's predictions.

[0064] Repeat the above steps of calculating residuals, training weak learners, and updating the model until the predetermined number of iterations is reached or the performance of the model is no longer significantly improved.

[0065] Furthermore, the CIF files of the 2D-COF material crystal structures that were not extracted in the first step are used to construct a file library 2.

[0066] The structural features of the above-mentioned 2D-COF materials are collected to obtain the material structural feature dataset 1-E;

[0067] The chemical characteristics of the above 2D-COF materials were collected to obtain material chemical characteristics dataset 1-F;

[0068] The material structure characteristic data set 1-E and the material chemical characteristic data set 1-F are input as initial data into the mathematical model A in the first step to generate the electronic structure data set 1-G;

[0069] The material chemical characteristic data set 1-E and the material chemical characteristic data set 1-F are input as initial data into the mathematical model B in the first step to generate the electronic structure data set 1-H;

[0070] The material structure feature dataset 1-E includes: the number of N atoms, the number of C atoms, the number of H atoms, the number of oxygen atoms, the total number of non-H atoms in a unit cell, the total number of atoms in a unit cell, the number of non-H atoms in a node, the number of node atoms, the number of non-H atoms in a linker, the number of linker atoms, the number of non-H atoms in a covalent bond, the number of covalent bond atoms, the number of N atoms at nodes, the number of N atoms in a linker, the number of C atoms at nodes, the number of C atoms in a linker, the number of O atoms at nodes, the number of O atoms in a linker, the number of N atoms in a covalent bond, and the number of O atoms in a covalent bond.

[0071] The material chemical feature data set 1-F includes: total period number, total unit cell mass, average node mass, average connector mass, number of covalent bond C, average covalent bond mass, total unit cell electronegativity, average node electronegativity, connector electronegativity, covalent bond electronegativity, average node p z Number of electrons, average p of linker z Number of electrons, covalent bond p z Number of electrons, total lone pairs and conjugated electrons, node 1 polarization area, node 2 polarization area, linker molecule polarization area, covalent bond individual molecule polarization area, unit cell total polarization area;

[0072] Furthermore, the electronic structure dataset 1-A, the electronic structure dataset 1-E and the electronic structure dataset 1-F in the above process were collected to generate the total dataset I of the thermodynamic properties of 2D-COF photocatalytic water splitting.

[0073] In the second step, based on the kinetic judgment conditions of photocatalytic water splitting performance, it is judged whether the conditions are met in the 2D-COF photocatalytic water splitting thermodynamic performance data set I, and 2D-COF without photocatalytic performance is predicted and the 2D-COF structure CIF file library 3 is generated.

[0074] The thermodynamic judgment conditions for the photocatalytic water splitting performance include the conduction band bottom energy level value E in the 2D-COF band structure. CBM Is it greater than -4.44, the value of the valence band top energy level E in the 2D-COF band structure? VBM Is it less than -5.67, and E CBM -E VBM Check whether the value is between 2.2 and 3.26.

[0075] Furthermore, if the conditions are met, a 2D-COF structure CIF file library 3 is generated, otherwise it is predicted that the 2D-COF does not have the photocatalytic full water splitting performance;

[0076] The third step is to generate a 2D-COF structure CIF file library 4 by random uniform sampling from the 2D-COF structure CIF file library 3;

[0077] The above 2D-COF structure CIF file library 4 is used through an automated input file generation program to automatically generate scientific calculation input files for the material calculation software VASP in large quantities;

[0078] The output file generated after the calculation is run through the automated catalytic kinetics Gibbs free energy data reading program to automatically read the Gibbs free energy data in large quantities and generate the catalytic kinetics data set 4-A;

[0079] Collect the material structure features in the small sample database to obtain the material structure feature dataset 4-B;

[0080] Collect the material chemical features in the small sample database to obtain the material chemical feature dataset 4-C;

[0081] The catalytic kinetics data set 3-C includes: 2D-COF hydrogen evolution reaction free energy value ΔG H* , 2D-COF hydrogen evolution reaction free energy value ΔG OH* ;

[0082] The material structure feature data set 3-D includes: the number of N atoms, the number of C atoms, the number of H atoms, the number of oxygen atoms, the total number of non-H atoms in the unit cell, the total number of atoms in the unit cell, the number of non-H atoms in the node, the number of node atoms, the number of non-H atoms in the linker, the number of linker atoms, the number of non-H atoms in the covalent bond, the number of covalent bond atoms, the number of node N atoms, the number of linker N atoms, the number of node C atoms, the number of linker C atoms, the number of node O atoms, the number of linker O atoms, the number of covalent bond N atoms, and the number of covalent bond O atoms;

[0083] The material chemical feature data set 3-E includes: total period number, total unit cell mass, average node mass, average connector mass, number of covalent bond C, average covalent bond mass, total unit cell electronegativity, average node electronegativity, connector electronegativity, covalent bond electronegativity, average node p z Number of electrons, average p of linker z Number of electrons, covalent bond p z Number of electrons, total number of lone pairs and conjugated electrons, node 1 polarization area, node 2 polarization area, linker molecule polarization area, covalent bond individual molecule polarization area, unit cell total polarization area;

[0084] Furthermore, the above catalytic kinetics dataset 4-A, material structure dataset 4-B, and material chemistry dataset 4-C were aggregated to generate the 2D-COF catalytic performance dataset 4-D.

[0085] The catalytic performance dataset 3-F includes: the number of N atoms, the number of C atoms, the number of H atoms, the number of oxygen atoms, the total number of non-H atoms in the unit cell, the total number of atoms in the unit cell, the number of non-H atoms in the node, the number of node atoms, the number of non-H atoms in the linker, the number of linker atoms, the number of non-H atoms in the covalent bond, the number of covalent bond atoms, the number of N atoms at the node, the number of N atoms in the linker, the number of C atoms at the node, the number of C atoms in the linker, the number of O atoms at the node, the number of O atoms in the linker, the number of N atoms in the covalent bond, the number of covalent bond O atoms, the total number of periods, the total mass of the unit cell, the average mass of the node, the average mass of the linker, the number of covalently bonded C atoms, the average mass of the covalent bond, the total electronegativity of the unit cell, the average electronegativity of the node, the electronegativity of the linker, the electronegativity of the covalent bond, the average p-value of the node z Number of electrons, average p of linker z Number of electrons, covalent bond p z Number of electrons, total number of lone pairs and conjugated electrons, node 1 polarization area, node 2 polarization area, linker molecule polarization area, covalent bond individual molecule polarization area, total unit cell polarization area, 2D-COF hydrogen evolution reaction free energy value ΔG H* , 2D-COF hydrogen evolution reaction free energy value ΔG OH* .

[0086] Furthermore, the GBR algorithm was used to perform machine learning on the 2D-COF catalytic performance dataset 4-D to obtain a predictable 2D-COF hydrogen evolution reaction free energy value ΔG H* Mathematical model C and the free energy value ΔG of the 2D-COF hydrogen evolution reaction OH* Mathematical model D.

[0087] GBR is a powerful machine learning algorithm for regression problems. Based on the Gradient Boosting framework, it builds a powerful model by integrating multiple weak learners (typically decision trees). GBR has shown excellent performance in practical applications, especially when dealing with nonlinear relationships and high-dimensional data.

[0088] The core idea of ​​GBR is to train multiple weak learners (such as decision trees) in sequence, with each new learner trying to correct the errors of the previous learner. The specific steps are as follows:

[0089] Initialize the model: Initialize with a simple model (such as a constant model).

[0090] Calculate residuals: Calculate the error (residual) between the predicted value of the current model and the true value.

[0091] Train a weak learner: Use the residual as the target variable and train a new weak learner.

[0092] Update model: Add the newly trained weak learners to the existing model and update the model's predictions.

[0093] Repeat the above steps of calculating residuals, training weak learners, and updating the model until the predetermined number of iterations is reached or the performance of the model is no longer significantly improved.

[0094] Furthermore, the 2D-COF material crystal structure CIF files that are not extracted in the above process are used to construct a 2D-COF structure CIF file library 5.

[0095] The structural features of the above-mentioned 2D-COF materials are collected to obtain material structural feature dataset 5-A;

[0096] The chemical characteristics of the above 2D-COF materials were collected to obtain material chemical characteristics dataset 5-B;

[0097] The material structure characteristic data set 5-A and the material chemical characteristic data set 5-B are input as initial data into the mathematical model C in the first step to generate the catalytic kinetic data set 5-C;

[0098] The material structure characteristic data set 3-H and the material chemical characteristic data set 3-I are used as initial data to input into the mathematical model D in the first step to generate the catalytic kinetic data set 5-D;

[0099] The material structure feature dataset 3-H includes: the number of N atoms, the number of C atoms, the number of H atoms, the number of oxygen atoms, the total number of non-H atoms in the unit cell, the total number of atoms in the unit cell, the number of non-H atoms in the node, the number of node atoms, the number of non-H atoms in the linker, the number of linker atoms, the number of non-H atoms in the covalent bond, the number of covalent bond atoms, the number of N atoms in the node, the number of N atoms in the linker, the number of C atoms in the node, the number of C atoms in the linker, the number of O atoms in the node, the number of O atoms in the linker, the number of N atoms in the covalent bond, and the number of O atoms in the covalent bond.

[0100] The material chemical characteristic data set 3-I includes: total period number, total unit cell mass, average node mass, average connector mass, number of covalent bond C, average covalent bond mass, total unit cell electronegativity, average node electronegativity, connector electronegativity, covalent bond electronegativity, average node p z Number of electrons, average p of linker z Number of electrons, covalent bond p z Number of electrons, total lone pairs and conjugated electrons, node 1 polarization area, node 2 polarization area, linker molecule polarization area, covalent bond individual molecule polarization area, unit cell total polarization area;

[0101] Furthermore, catalytic kinetic data set 4-A, catalytic kinetic data set 5-C and catalytic kinetic data set 5-D in the above process were collected to generate 2D-COF photocatalytic water splitting kinetic performance data set II.

[0102] The fourth step is to determine whether conditions one and two are met in the 2D-COF photocatalytic water splitting kinetic performance dataset II based on the kinetic judgment conditions of photocatalytic water splitting performance, and to predict 2D-COF with excellent photocatalytic water splitting performance and 2D-COF without photocatalytic performance.

[0103] Furthermore, if condition 1 is met, condition 2 is further judged, otherwise it is predicted that the 2D-COF does not have the photocatalytic full water splitting performance;

[0104] Furthermore, if the second condition is met, it is predicted to be a 2D-COF with excellent photocatalytic complete water splitting performance; otherwise, it is predicted to be a 2D-COF without photocatalytic complete water splitting performance;

[0105] The kinetic judgment condition of the photocatalytic water splitting performance includes the free energy value of the 2D-COF hydrogen evolution reaction ΔG H* Is the absolute value less than the photoelectric potential energy value eU? e , the free energy value of 2D-COF oxygen evolution reaction ΔG OH*Is the absolute value less than the photoelectric potential energy value eU? h ;

[0106] The second kinetic judgment condition for the photocatalytic water splitting performance includes the overpotential energy value η of the 2D-COF hydrogen evolution reaction. HER Is the value less than the photoelectric potential energy value eU? e , the potential energy value of the oxygen evolution reaction of 2D-COF η OER Is the value less than the photoelectric potential energy value eU? h .

[0107] Based on the above technical solution, a specific embodiment is provided to explain the contents of the above steps in detail.

[0108] Step 1: Establish the thermodynamic performance data set I of 2D-COF photocatalytic water splitting.

[0109] Step 1.1. Create 2D-COF electronic structure dataset 1-A, including:

[0110] (1) Randomly and uniformly extract a small sample structure file library 1 from the total structure file library of 2D-COF material crystal structure files CIF;

[0111] (2) Use shell scripts to automatically generate scientific computing input files for the materials calculation software VASP;

[0112] (3) Use shell script to read the calculation results (OUTCAR file) to obtain the conduction band bottom energy (E) of each structure above. CBM ), valence band top energy (E VBM ) and the band gap (E g ) Numerically establish 2D-COF electronic structure dataset 1-A.

[0113] A high-throughput computing platform was established using shell scripts, and the above data information was obtained through the VASP computing package;

[0114] Step 1.2: Establish 2D-COF structural feature dataset 1-B, including:

[0115] Number of nitrogen atoms, number of carbon atoms, number of hydrogen atoms, number of oxygen atoms, total number of non-hydrogen atoms in the unit cell, total number of atoms in the unit cell, number of non-hydrogen atoms in the nodes, number of node atoms, number of non-hydrogen atoms in the linker, number of linker atoms, number of non-hydrogen atoms in covalent bonds, number of covalent bond atoms, number of node nitrogen atoms, number of linker nitrogen atoms, number of node carbon atoms, number of linker carbon atoms, number of node oxygen atoms, number of linker oxygen atoms, number of covalently bonded nitrogen atoms, number of covalently bonded oxygen atoms.

[0116] Among them, this embodiment mainly selects the atomic number of H, C, N, and O as the main features, and uses a shell script to read the atomic number of each element in the CIF file.

[0117] In this embodiment, the number of atoms of C, H, O, and N is mainly selected as the structural features. The crystallographic information file CIF takes the following COF.cif file as an example:

[0118]

[0119]

[0120]

[0121]

[0122] Step 1.3: Create a 2D-COF chemical feature dataset 1-C, including:

[0123] Total number of periods, total mass of unit cell, average mass of node, average mass of connector, number of covalent bond C, average mass of covalent bond, total electronegativity of unit cell, average electronegativity of node, electronegativity of connector, electronegativity of covalent bond, average p of node z Number of electrons, average p of linker z Number of electrons, covalent bond p z Number of electrons, total number of lone pairs and conjugated electrons, node 1 polarization area, node 2 polarization area, linker molecule polarization area, covalent bond individual molecule polarization area, and total unit cell polarization area.

[0124] Among them, the above chemical characteristics were collected using Excel files.

[0125] Step 1.4: Merge the 1-A, 1-B, and 1-C datasets to create the 1-D dataset. Use the total dataset to construct a table document, and store the table document in .csv format to facilitate data reading, thereby obtaining the total dataset.

[0126] Step 1.5: Construction of GBR algorithm:

[0127] The general approach of boosting methods is to learn a series of classifiers, where each classifier in this series gives greater weight to examples misclassified by the previous classifier. Specifically, after a classifier is learned, the weights of the training examples that resulted in the misclassification are increased, and the weights are recalculated for each training example before learning the next classifier. This training process is repeated several times. The final classifier is synthesized from this series of classifiers. In this process, each training example is assigned a corresponding weight. If a training example is misclassified by a classifier, its weight is increased accordingly, so that the classifier gives greater weight to the information represented by that example in the next learning cycle. GBR is a classification method that combines weak classifiers in some way to produce a strong classifier with significantly improved classification performance. This method transforms crude empirical rules into highly accurate predictions. The strong classifier classifies data through a majority voting mechanism among the weak classifiers. It has been theoretically proven that any weak classification algorithm can be effectively transformed or boosted into a strong learning classification algorithm. This algorithm is essentially a simple weak classifier boosting process that, through continuous training, improves its classification ability. The whole process is as follows:

[0128] ① First, obtain the first weak classifier h1 by learning N training data;

[0129] ② Combine the data that h1 misclassified with other new data to form a new sample with N training data, and obtain the second weak classifier h2 by learning this sample;

[0130] ③ Add the data that h1 and h2 have misclassified to other new data to form another new sample with N training data. By learning this sample, we get the third weak classifier h3;

[0131] ④Finally, the enhanced strong classifier h final =MajorityVote(h1,h2,h3). This means that the data must be classified into a certain category by a majority vote of h1, h2, and h3.

[0132] Based on the above method, a mathematical model A of catalytic performance is obtained;

[0133] The catalytic performance mathematical model B was obtained based on the above method.

[0134] Step 1.6: Construct file library 2 using the CIF files of the 2D-COF material crystal structure that were not extracted in the first step.

[0135] Step 1.7: Create a 2D-COF structural feature dataset 1-E, including:

[0136] Number of N atoms, number of C atoms, number of H atoms, number of oxygen atoms, total number of non-H atoms in the unit cell, total number of atoms in the unit cell, number of non-H atoms in nodes, number of node atoms, number of non-H atoms in linkers, number of linker atoms, number of non-H atoms in covalent bonds, number of covalent bond atoms, number of node N atoms, number of linker N atoms, number of node C atoms, number of linker C atoms, number of node O atoms, number of linker O atoms, number of covalent bond N atoms, number of covalent bond O atoms.

[0137] Step 1.8. Create 2D-COF chemical feature dataset 1-F, including:

[0138] Total number of periods, total mass of unit cell, average mass of node, average mass of connector, number of covalent bond C, average mass of covalent bond, total electronegativity of unit cell, average electronegativity of node, electronegativity of connector, electronegativity of covalent bond, average p of node z Number of electrons, average p of linker z Number of electrons, covalent bond p z Number of electrons, total number of lone pairs and conjugated electrons, node 1 polarization area, node 2 polarization area, linker molecule polarization area, covalent bond individual molecule polarization area, and total unit cell polarization area.

[0139] Step 1.9: Input the above material structure characteristic dataset 1-E and material chemical characteristic dataset 1-F as initial data into the mathematical model A obtained in step 1.5 to generate the 2D-COF electronic structure dataset 1-G, including: conduction band bottom energy (E CBM );

[0140] Step 1.10: Input the above material structure characteristic dataset 1-E and material chemical characteristic dataset 1-F as initial data into the mathematical model B obtained in step 1.5 to generate the 2D-COF electronic structure dataset 1-H, including: valence band top energy (E VBM ), band gap (E g );

[0141] Step 1.11: Merge the 1-D, 1-E, 1-F, and 1-H datasets to create a comprehensive dataset I for the thermodynamic properties of 2D-COF photocatalytic water splitting. Use the comprehensive dataset to construct a spreadsheet document, which is stored in .csv format for easy data reading, to obtain the comprehensive dataset.

[0142] Step 2: Screening 2D-COF structures that meet the thermodynamic conditions for photocatalytic water splitting. Figure 2 )

[0143] Step 2.1, determine the conduction band bottom energy (E CBM ), valence band top energy (E VBM ), band gap (E g) value simultaneously satisfies the thermodynamic conditions of photocatalytic complete water splitting, namely, E CBM >-4.44, E VBM <-5.67 and 2.2<E VBM <3.26.

[0144] Step 2.2: Screen out 2D-COF structures that meet the thermodynamic conditions for photocatalytic water splitting and establish a 2D-COF structure CIF file library 3.

[0145] Step 3: Establish the 2D-COF photocatalytic water splitting kinetic performance data set II. Figure 3 )

[0146] Step 3.1: Establish 2D-COF catalytic kinetics data set 3-A, including:

[0147] (1) Randomly and uniformly extract a small sample structure file library 4 from the 2D-COF material crystal structure file CIF structure file library 3;

[0148] (2) Use shell scripts to automatically generate scientific computing input files for the materials calculation software VASP;

[0149] (3) Use the shell script to read the calculation results (OUTCAR file) to obtain the hydrogen evolution Gibbs free energy (ΔG) for each structure mentioned above. H* ), oxygen evolution Gibbs free energy (ΔG OH* ) Numerically establish 2D-COF catalytic kinetics dataset 3-A.

[0150] A high-throughput computing platform was established using shell scripts, and the above data information was obtained through the VASP computing package;

[0151] Step 3.2: Establish 2D-COF structural feature dataset 3-B, including:

[0152] Number of nitrogen atoms, number of carbon atoms, number of hydrogen atoms, number of oxygen atoms, total number of non-hydrogen atoms in the unit cell, total number of atoms in the unit cell, number of non-hydrogen atoms in the nodes, number of node atoms, number of non-hydrogen atoms in the linker, number of linker atoms, number of non-hydrogen atoms in covalent bonds, number of covalent bond atoms, number of node nitrogen atoms, number of linker nitrogen atoms, number of node carbon atoms, number of linker carbon atoms, number of node oxygen atoms, number of linker oxygen atoms, number of covalently bonded nitrogen atoms, number of covalently bonded oxygen atoms.

[0153] Among them, this embodiment mainly selects the atomic number of H, C, N, and O as the main features, and uses a shell script to read the atomic number of each element in the CIF file.

[0154] In this embodiment, the number of atoms of C, H, O, and N is mainly selected as the structural features. The crystallographic information file CIF takes the following COF.cif file as an example:

[0155]

[0156]

[0157]

[0158]

[0159] Step 3.3: Create a 2D-COF chemical feature dataset 3-C, including:

[0160] Total number of periods, total mass of unit cell, average mass of node, average mass of connector, number of covalent bond C, average mass of covalent bond, total electronegativity of unit cell, average electronegativity of node, electronegativity of connector, electronegativity of covalent bond, average p of node z Number of electrons, average p of linker z Number of electrons, covalent bond p z Number of electrons, total number of lone pairs and conjugated electrons, node 1 polarization area, node 2 polarization area, linker molecule polarization area, covalent bond individual molecule polarization area, and total unit cell polarization area.

[0161] Among them, the above chemical characteristics were collected using Excel files.

[0162] Step 3.4: Merge the 3-A, 3-B, and 3-C datasets to create a 3-D dataset. Use the dataset to construct a table document, and store the table document in .csv format to facilitate data reading, thereby obtaining the total dataset.

[0163] Step 3.5, Construction of GBR algorithm ( Figure 5 ):

[0164] The general approach of boosting methods is to learn a series of classifiers, where each classifier in this series gives greater weight to examples misclassified by the previous classifier. Specifically, after a classifier is learned, the weights of the training examples that resulted in the misclassification are increased, and the weights are recalculated for each training example before learning the next classifier. This training process is repeated several times. The final classifier is synthesized from this series of classifiers. In this process, each training example is assigned a corresponding weight. If a training example is misclassified by a classifier, its weight is increased accordingly, so that the classifier gives greater weight to the information represented by that example in the next learning cycle. GBR is a classification method that combines weak classifiers in some way to produce a strong classifier with significantly improved classification performance. This method transforms crude empirical rules into highly accurate predictions. The strong classifier classifies data through a majority voting mechanism among the weak classifiers. It has been theoretically proven that any weak classification algorithm can be effectively transformed or boosted into a strong learning classification algorithm. This algorithm is essentially a simple weak classifier boosting process that, through continuous training, improves its classification ability. The whole process is as follows:

[0165] ① First, obtain the first weak classifier h1 by learning N training data;

[0166] ② Combine the data that h1 misclassified with other new data to form a new sample with N training data, and obtain the second weak classifier h2 by learning this sample;

[0167] ③ Add the data that h1 and h2 have misclassified to other new data to form another new sample with N training data. By learning this sample, we get the third weak classifier h3;

[0168] ④Finally, the enhanced strong classifier h final =MajorityVote(h1,h2,h3). This means that the data must be classified into a certain category by a majority vote of h1, h2, and h3.

[0169] Based on the above method, a mathematical model C of catalytic performance was obtained;

[0170] The catalytic performance mathematical model D was obtained based on the above method.

[0171] Step 3.6: Construct database 5 using the CIF files of the 2D-COF material crystal structures not extracted from database 3.

[0172] Step 3.7: Create a 2D-COF structural feature dataset 3-E, including:

[0173] Number of N atoms, number of C atoms, number of H atoms, number of oxygen atoms, total number of non-H atoms in the unit cell, total number of atoms in the unit cell, number of non-H atoms in nodes, number of node atoms, number of non-H atoms in linkers, number of linker atoms, number of non-H atoms in covalent bonds, number of covalent bond atoms, number of node N atoms, number of linker N atoms, number of node C atoms, number of linker C atoms, number of node O atoms, number of linker O atoms, number of covalent bond N atoms, number of covalent bond O atoms.

[0174] Step 3.8: Create 2D-COF chemical feature dataset 3-F, including:

[0175] Total number of periods, total mass of unit cell, average mass of node, average mass of connector, number of covalent bond C, average mass of covalent bond, total electronegativity of unit cell, average electronegativity of node, electronegativity of connector, electronegativity of covalent bond, average p of node z Number of electrons, average p of linker z Number of electrons, covalent bond p z Number of electrons, total number of lone pairs and conjugated electrons, node 1 polarization area, node 2 polarization area, linker molecule polarization area, covalent bond individual molecule polarization area, and total unit cell polarization area.

[0176] Step 3.9: Input the above material structure feature dataset 3-E and material chemical feature dataset 4-F as initial data into the mathematical model A obtained in step 3.5 to generate the 2D-COF electronic structure dataset 3-G, including: hydrogen evolution Gibbs free energy (ΔG H* );

[0177] Step 3.10: Input the above material structure characteristic dataset 3-E and material chemical characteristic dataset 3-F as initial data into the mathematical model B obtained in step 3.5 to generate the 2D-COF electronic structure dataset 3-H, including: oxygen evolution Gibbs free energy (ΔG OH* );

[0178] Step 3.11: Merge the 3-D, 3-E, 3-F, and 3-H datasets to establish a total dataset II for the 2D-COF photocatalytic water splitting kinetics performance. Use the total dataset to construct a spreadsheet document, and store the spreadsheet document in .csv format to facilitate data reading, thereby obtaining the total dataset.

[0179] Step 4: Screening 2D-COF structures that meet the kinetic conditions for photocatalytic complete water splitting. Figure 4 )

[0180] Step 4.1. Determine the absolute value of the Gibbs free energy of hydrogen evolution of 2D-COF (|ΔG H* |), the absolute value of the Gibbs free energy of oxygen evolution (|ΔG OH*|) value simultaneously satisfies the kinetic conditions for photocatalytic complete water splitting, namely |ΔG H* |<E CBM +4.44 and |ΔG OH* |<E VBM -5.67;

[0181] Step 4.2, screen out 2D-COF structures that meet the kinetic conditions of photocatalytic complete water splitting, and establish a 2D-COF structure CIF file library 6;

[0182] Step 4.3, determine the hydrogen evolution overpotential (η) of each structure in the 2D-COF structure CIF file library 6 HER ) and oxygen evolution overpotential (η OER ) value simultaneously satisfies the excellent photocatalytic complete water splitting kinetics condition, namely η HER <E CBM +4.44 and η OER <E VBM -5.67.

[0183] Step 4.4: Obtain an excellent photocatalytic water splitting 2D-COF structure that meets the requirements.

[0184] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0185] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only intended to help understand the core concept of the present invention. At the same time, those skilled in the art will find that the specific implementation methods and application scopes may vary based on the concept of the present invention. In summary, the contents of this specification should not be construed as limiting the present invention.

Claims

1. A method for predicting the photocatalytic water splitting performance of 2D-COF materials, characterized in that: include: Establish a 2D-COF photocatalytic water splitting thermodynamic performance data set I; the 2D-COF photocatalytic water splitting thermodynamic performance data set I is obtained by screening 2D-COF structure CIF file 1 and 2D-COF structure CIF file 2; the data in the 2D-COF structure CIF file 1 are randomly and uniformly sampled from the 2D-COF structure CIF total file; the data in the 2D-COF structure CIF file 2 are the remaining samples not sampled in the 2D-COF structure CIF total file; According to the kinetic judgment conditions of the photocatalytic water splitting performance, the 2D-COF photocatalytic water splitting thermodynamic performance data set I was screened to obtain 2D-COF structures without photocatalytic performance and 2D-COF structures with photocatalytic performance, and the 2D-COF structure with photocatalytic performance was determined as 2D-COF structure CIF file 3; Establish a 2D-COF photocatalytic water splitting thermodynamic performance data set II; the 2D-COF photocatalytic water splitting thermodynamic performance data set II is obtained by screening 2D-COF structure CIF file 4 and 2D-COF structure CIF file 5; the data in the 2D-COF structure CIF file 4 are randomly and uniformly sampled from the 2D-COF structure CIF file 3; the data in the 2D-COF structure CIF file 5 are the remaining samples not sampled in the 2D-COF structure CIF file 3; The 2D-COF photocatalytic water splitting thermodynamic performance data set II was subjected to secondary screening, and the data that met the secondary screening conditions were determined to have excellent photocatalytic water splitting performance 2D-COF; the secondary screening conditions included condition 1 and condition 2; the condition 1 was the free energy value ΔG of the 2D-COF hydrogen evolution reaction. H* Is the absolute value less than the photoelectric potential energy value eU? e , the free energy value of 2D-COF oxygen evolution reaction ΔG OH* Is the absolute value less than the photoelectric potential energy value eU? h ; The condition 1 is the 2D-COF hydrogen evolution reaction overpotential value η HER Is the value less than the photoelectric potential energy value eU? e , the overpotential value η of the oxygen evolution reaction of 2D-COF OER Is the value less than the photoelectric potential energy value eU? h ; The method for establishing the 2D-COF photocatalytic water splitting thermodynamic performance data set I includes: The 2D-COF structure CIF file 1 is processed using an automated input file generation program to generate an electronic structure data set 1-A, and structural features and chemical features are collected to obtain a structural feature data set 1-B and a chemical feature data set 1-C, respectively; Aggregating the electronic structure dataset 1-A, the structural feature dataset 1-B, and the chemical feature dataset 1-C, and performing machine learning on the aggregated 2D-COF catalytic performance total dataset 1-D using a GBR algorithm to obtain a prediction model; the prediction model includes a mathematical model A for predicting catalytic performance and a mathematical model B for predicting catalytic performance; Collecting structural features and chemical features of the 2D-COF structure CIF file 2 to obtain a structural feature dataset 1-E and a chemical feature dataset 1-F, respectively; Inputting the structural feature dataset 1-E and the chemical feature dataset 1-F as initial data into the catalytic performance prediction mathematical model A to generate an electronic structure dataset 1-G; and inputting the structural feature dataset 1-E and the chemical feature dataset 1-F as initial data into the catalytic performance prediction mathematical model B to generate an electronic structure dataset 1-H; The 2D-COF photocatalytic water splitting thermodynamic performance data set I was constructed based on the electronic structure data set 1-A, the electronic structure data set 1-G and the electronic structure data set 1-H.

2. The method for predicting the photocatalytic water splitting performance of 2D-COF materials according to claim 1, characterized in that: The kinetic judgment conditions of the photocatalytic water splitting performance specifically include: the value of the bottom energy level of the conduction band in the 2D-COF band structure E CBM Is it greater than -4.44, the value of the valence band top energy level E in the 2D-COF band structure? VBM Is it less than -5.67, and E CBM -E VBM Check whether the value is between 2.2 and 3.

26.

3. The method for predicting the photocatalytic water splitting performance of 2D-COF materials according to claim 1, characterized in that: The judgment process of the secondary screening is as follows: Based on the thermodynamic performance data set II of 2D-COF photocatalytic water splitting, the absolute value of the Gibbs free energy of hydrogen evolution |ΔG H* | and the absolute value of the Gibbs free energy of oxygen evolution |ΔG OH* Whether the value of | satisfies the photocatalytic complete water splitting kinetic conditions at the same time: |ΔG H* |<E CBM +4.44 and |ΔG OH* |<E VBM -5.67; If yes, then filter out the data that meets the conditions and create a 2D-COF structure CIF file 6; Determine again the hydrogen evolution overpotential η of each structure of the 2D-COF structure CIF file 6 HER and oxygen evolution overpotential η OER Whether the value satisfies the conditions for excellent photocatalytic complete water splitting kinetics: η HER <E CBM +4.44 and η OER <E VBM -5.67; If so, it is determined to be a 2D-COF with excellent photocatalytic complete water splitting performance.

4. The method for predicting the photocatalytic water splitting performance of 2D-COF materials according to claim 1, characterized in that: The generation process of the electronic structure dataset 1-A is as follows: The 2D-COF structure CIF file 1 is automatically generated through an input file generation program to automatically generate scientific calculation input files for the material calculation software VASP in large quantities; Submit the scientific calculation input file in the server directory and run the material calculation software VASP, and obtain a large number of output files after the calculation is completed; The output file generated after the calculation is run through the automated electronic structure data reading program to automatically read the electronic structure data in large quantities and generate the electronic structure data set 1-A.

5. The method for predicting the photocatalytic water splitting performance of 2D-COF materials according to claim 4, characterized in that: The electronic structure data set 1-A includes: the valence band top energy level value E in the 2D-COF band structure VBM , the conduction band bottom energy level value E in the 2D-COF band structure CBM And the band gap value E in the 2D-COF band structure g .

6. An electronic device, characterized in that: The electronic device comprises a memory and a processor, wherein the memory is used to store a computer program, and the processor runs the computer program to enable the electronic device to perform the method for predicting the photocatalytic water splitting performance of a 2D-COF material according to any one of claims 1 to 5.

7. A computer-readable storage medium, characterized in that It stores a computer program, which, when executed by a processor, implements the method for predicting the photocatalytic water splitting performance of a 2D-COF material as described in any one of claims 1 to 5.

Images