An actinide coordination bond length prediction method considering multi-stage structural information migration and coordination competition effect, program, device and storage medium

Abstract

The application belongs to the technical field of cross technology of nuclear fuel cycle chemistry, computational chemistry and artificial intelligence, and particularly relates to an actinide coordination bond length prediction method considering multi-stage structure information migration and coordination competition effect, a program, equipment and a storage medium. In the bond length prediction process, the multi-coordination competition effect is explicitly considered, and the coordination environment information formed in the low-precision or intermediate structure is fully utilized, so that the strong dependence on the high-precision fully converged structure is avoided, and the structural response behavior of the target coordination bond under the complex multi-coordination environment can be more truly and efficiently reflected. The application is suitable for structural analysis and rapid screening of multi-coordination metal coordination systems, has the characteristics of clear method flow, strong applicability, high prediction efficiency and the like, and has a good engineering application prospect.

Description

Technical Field

[0001] This invention belongs to the interdisciplinary fields of nuclear fuel cycle chemistry, computational chemistry and artificial intelligence, and specifically relates to a method, program, device and storage medium for predicting actinide coordination bond lengths that considers multi-stage structural information migration and coordination competition effects. Background Technology

[0002] The structural parameters of actinide coordination bonds, especially bond length, are key indicators characterizing the stability, reactivity, and separation selectivity of their complexes, and are of great significance in research fields such as actinide coordination chemistry, nuclear fuel cycle, radioactive element separation, and transition metal catalysis. With the deepening of related research, it has become increasingly clear that actinide metal centers in actual systems often do not form isolated metal-ligand coordination structures with a single ligand, but rather simultaneously coordinate with multiple ligand molecules or solvent molecules, forming complex multi-coordination systems.

[0003] In the aforementioned multi-coordination systems, the geometry of a particular actinide metal-ligand coordination bond is not solely determined by the properties of that ligand itself, but is significantly influenced by the presence, quantity, and spatial distribution of other metal-ligand coordination bonds around the metal center. Spatial repulsion and coordination competition occur between different ligands near the actinide metal center, resulting in significant mutual coupling between the coordination bonds. This phenomenon is commonly referred to as the multi-coordination competition effect. This effect is particularly pronounced in high coordination number systems such as those involving actinides, often leading to a significant change in the bond length of the target metal-ligand coordination bond compared to the single-coordination case.

[0004] On the other hand, in actual structural modeling and quantum chemical calculations, the coordination structure of actinide multi-coordination systems typically undergoes a gradual evolution from an initial configuration to a stable configuration. The fully converged structure obtained through high-precision quantum chemical calculations can accurately reflect the combined effect of multi-coordination competition in the final stable configuration, but its computational cost is high and the modeling cycle is long, making it difficult to directly apply to the rapid prediction of large-scale systems. In contrast, the low-precision or incompletely converged structures obtained in the early or intermediate stages of structural optimization, although still exhibiting some deviations in geometric parameters, already contain the basic spatial information of the target coordination bonds and their surrounding coordination environment, and preliminarily demonstrate the characteristics of the multi-coordination competition effect.

[0005] However, in existing technologies, most methods for predicting the length of actinide metal-ligand coordination bonds still rely solely on modeling the fully convergent structure obtained from high-precision quantum chemical calculations, typically using this structure as the sole object of learning or prediction. This approach not only heavily depends on high-precision calculation results, making it difficult to reduce data construction costs, but also fails to effectively utilize the coordination environment information contained in low-precision or intermediate structures, making it difficult to balance accuracy and computational efficiency in the prediction process.

[0006] Especially in multi-coordination systems where water molecules or other small molecules fill the actinide metal coordination environment, these molecules coordinate with the actinide metal center at the low-precision or intermediate structural stage, significantly affecting the geometric parameters of the target metal-ligand coordination bond. If the prediction method only uses the final high-precision converged structure as input, without establishing an effective correlation between the low-precision structure and the high-precision coordination structure, it will be difficult to accurately reflect the true multi-coordination competition effect while reducing computational costs.

[0007] Therefore, it is necessary to propose a new technical solution in the prediction of actinide metal-ligand coordination bond length. By introducing a high-precision coordination structure as a knowledge source, the multi-coordination competition law contained in the high-precision structure is learned during the training phase, and this high-precision structure knowledge is transferred to the low-precision or intermediate structure representation. Thus, in the prediction phase, only the optimization results of the low-precision structure are needed to explicitly introduce and quantify the multi-coordination competition effect based on the complete multi-coordination structure, thereby achieving efficient and accurate prediction of the target actinide metal-ligand coordination bond length in the multi-coordination system. Summary of the Invention

[0008] The purpose of this invention is to solve the problems of insufficient accuracy and high computational cost in predicting metal-ligand coordination bond lengths in multi-coordination systems due to neglecting coordination competition effects and relying heavily on high-precision quantum chemical calculations. The invention provides a method, program, device, and storage medium for predicting actinide coordination bond lengths that considers multi-stage structural information migration and coordination competition effects.

[0009] A method for predicting actinide coordination bond lengths that considers multi-stage structural information transfer and coordination competition effects includes the following steps:

[0010] Identify the target actinide multicoordination system to be predicted and obtain samples of multicoordination systems with the same actinide element as the metal center;

[0011] For each group of multi-coordination system samples, obtain structural data for the low-convergence stage and the high-convergence stage, select one coordinating atom as the target coordinating atom, and the remaining coordinating atoms as competing coordinating atoms. Construct competitive coordination context information based on the structural data of the low-convergence stage. Calculate the bond length of the target coordinating bond based on the structural data of the low-convergence stage and the high-convergence stage respectively, and determine the bond length error of the target coordinating bond.

[0012] Construct and train a prediction model that can output the bond length error of the target coordination bond based on the input competitive coordination context information;

[0013] For a target actinide multi-coordination system to be predicted, structural data in the low convergence stage is obtained, a coordinating bond is selected as the target coordinating bond, and the bond length and competing coordination context information of the target coordinating bond are determined. The competing coordination context information is input into the trained prediction model to obtain the bond length error of the target coordinating bond. The sum of the bond length of the target coordinating bond and the bond length error is taken as the bond length prediction result of the target coordinating bond. Other coordinating bonds are selected as target coordinating bonds in turn, and the above process is repeated until the bond length prediction results of all coordinating bonds are obtained.

[0014] Furthermore, the structural data includes the spatial coordinates of the metal center and the spatial coordinates of each coordinating atom, thereby calculating the bond length of the coordinate bond between the metal center and each coordinating atom.

[0015] Furthermore, the competitive coordination context information includes the number of competing coordinating atoms, information on competing coordinating bonds, distance information between the target coordinating atom and the competing coordinating atom, angle information between the target coordinating bond and the competing coordinating bond, coordination environment asymmetry parameters, and spatial occlusion parameters.

[0016] Furthermore, the competing coordinate key information for:

[0017]

[0018] in, For the first The first group of multi-coordination system samples The bond length of the competitive coordination bond between the competing coordinating atoms and the metal center. To calculate the average, To calculate the standard deviation; , For the first The number of competing coordinating atoms in a multi-coordination system sample;

[0019] The distance information between the target coordinating atom and the competing coordinating atom for:

[0020]

[0021] in, For the first The first group of multi-coordination system samples The distance between the competing coordinating atom and the target coordinating atom;

[0022] The angle information between the target coordinate bond and the competing coordinate bond for:

[0023]

[0024] in, For the first The first group of multi-coordination system samples The angle formed between the competing coordinate bond and the target coordinate bond at the center of the metal.

[0025] Furthermore, the coordination environment asymmetric parameters for:

[0026]

[0027] in, For the first In the multi-coordination system samples, from the metal center to the first Vectors of competing coordinating atoms;

[0028] The spatial occlusion parameters for:

[0029]

[0030] in, For the first The first group of multi-coordination system samples van der Waals radii of competing coordinating atoms;

[0031] The competitive coordination context information for:

[0032]

[0033] Furthermore, the bond length of the target coordination bond is calculated based on the structural data during the low-convergence phase. The bond lengths of the target coordination bonds are calculated based on the structural data from the high-convergence phase. Determine the bond length error of the target coordinate bond. ;

[0034] Construct a training set, where each group of training samples is... ;

[0035] The prediction model is trained using a training set, enabling it to predict based on the input competitive coordination context information. Output the bond length error of the target coordinate bond .

[0036] Furthermore, the prediction model is specifically as follows:

[0037]

[0038] in, For local interference terms, This is a global statistical item. This refers to three-dimensional environmental statistics. For constant terms;

[0039]

[0040]

[0041]

[0042] in, , , , and These are the weight coefficients to be trained.

[0043] A computer device includes a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the above-described method for predicting actinide coordination bond lengths considering multi-stage structural information migration and coordination competition effects.

[0044] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method for predicting actinide coordination bond lengths considering multi-stage structural information migration and coordination competition effects.

[0045] A computer program product includes computer instructions that, when executed by a processor, implement the steps of the above-described method for predicting actinide coordination bond lengths considering multi-stage structural information migration and coordination competition effects.

[0046] The beneficial effects of this invention are as follows:

[0047] This invention explicitly considers the multi-coordination competition effect during bond length prediction and fully utilizes the coordination environment information already formed in low-precision or intermediate structures, avoiding a strong dependence on high-precision fully converged structures. This allows for a more realistic and efficient reflection of the structural response behavior of target coordination bonds under complex multi-coordination environments. This invention is applicable to the structural analysis and rapid screening of multi-coordination metal coordination systems, featuring a clear methodology, strong applicability, and high prediction efficiency, and has promising engineering application prospects. Attached Figure Description

[0048] Figure 1 This is a diagram of the overall architecture of the present invention. Detailed Implementation

[0049] The present invention will now be further described with reference to the accompanying drawings.

[0050] This invention provides a method for predicting actinide coordination bond lengths that considers multi-stage structural information migration and coordination competition effects. It is applicable to the accurate prediction of target metal-ligand coordination bond lengths when the metal center simultaneously coordinates with multiple ligands or solvent molecules. Figure 1 As shown, it includes the following steps:

[0051] Step 1: Determine the multicoordination system of the target actinide element to be predicted;

[0052] Obtain samples of multi-coordination systems with the same actinide element as the metal center, and obtain low-convergence stage structure data and high-convergence stage structure data for each sample;

[0053] In a multi-coordinate metal coordination system, the metal center forms coordinate bonds with two or more ligand atoms, and the metal center is an actinide element, preferably uranium or plutonium;

[0054] The structural data includes the spatial coordinates of the metal center, the spatial coordinates of each coordinating atom, and the bond lengths of the coordinate bonds between the metal center and each coordinating atom.

[0055] (1) Structured data in the low convergence stage:

[0056] The approximate coordination configuration is obtained by reducing the number of structural optimization iterations or relaxing the structural optimization convergence threshold, and is used to characterize the coordination structure when the coordination structure is not yet fully stable and the coordination competition effect is not yet fully manifested.

[0057] (2) High convergence stage structure data:

[0058] The stable coordination configuration is obtained by quantum chemical structure optimization calculations using strict convergence criteria, and is used to characterize the stable coordination configuration after the multi-coordination competition effect is fully realized.

[0059] In one embodiment of the present invention, the structural data of the multi-coordinate actinide coordination system are obtained through quantum chemical calculations. The structural optimization calculations are performed using density functional theory (DFT), with the exchange-correlation functional chosen as the PBE functional, and a D3BJ dispersion correction introduced to describe the weak interaction effects in the system. The basis set system uses the def2-SVP basis set, and an effective core potential (ECP) including core electrons is introduced for actinide elements to reasonably account for the influence of relativistic effects on the coordination structure. The solvent environment is described using a continuous medium solvent model, preferably an aqueous environment, to reflect the actual coordination behavior of the multi-coordinate system under solution conditions. During the structural optimization process, rotational and translational degrees of freedom constraints are disabled, and the structural evolution calculations are completed using a combination of self-consistent field iteration and geometric optimization.

[0060] To construct multi-stage structural data reflecting the structural evolution process, structural data for both the low-convergence and high-convergence stages were obtained under the same theoretical methods, basis set systems, and solvent models. The low-convergence stage structural data consisted of intermediate structures (the results of the fifth round of structural optimization) obtained under the preset maximum number of iterations in geometric optimization. These structures did not yet meet the final geometric convergence criterion but had already formed stable metal-ligand coordination relationships. The high-convergence stage structural data consisted of fully converged structures obtained after continuing geometric optimization under the same computational conditions until all convergence criteria were met.

[0061] Step 2: For each group of multi-coordination system samples, select one coordinating atom as the target coordinating atom, and the remaining coordinating atoms as competing coordinating atoms;

[0062] Competing coordination bonds can be formed by different types of ligands. These ligands differ in spatial position, coordination direction, and donor atom properties, thus exerting steric repulsion and coordination competition on the target coordination bond under multi-coordination conditions. For the same multi-coordination metal coordination system, by changing the method of selecting the target coordination bond, multi-coordination competition analysis scenarios targeting different target coordination bonds can be constructed.

[0063] Calculate the bond length of the target coordination bond based on the structural data during the low-convergence phase. The bond lengths of the target coordination bonds are calculated based on the structural data from the high-convergence phase. Determine the bond length error of the target coordinate bond. ;

[0064] Construct competitive coordination context information based on structural data from the low-convergence stage. :

[0065]

[0066] For the first The number of competing coordinating atoms in a multi-coordination system sample is used to distinguish the complexity of the competitive environment under different coordination number conditions such as monocoordination, dicoordination, and tricoordination.

[0067] For the first Information on competing coordination bonds in a multi-coordination system is used to characterize the overall spatial proximity of competing coordination bonds.

[0068]

[0069] For the first The first group of multi-coordination system samples The bond length of the competitive coordination bond between the competing coordinating atoms and the metal center. To calculate the average, To calculate the standard deviation;

[0070] For the first The distance information between the target coordinating atom and the competing coordinating atom in a multi-coordination system sample is used to characterize the spatial repulsion or proximity relationship between the target coordinating atom and the competing coordinating atom.

[0071]

[0072] For the first The first group of multi-coordination system samples The distance between the competing coordinating atom and the target coordinating atom;

[0073] For the first The angle information between the target coordination bond and the competing coordination bond in the multi-coordination system samples is used to reflect the angular distribution characteristics of the competing coordination bonds around the metal center;

[0074]

[0075] For the first The first group of multi-coordination system samples The angle formed between the competing coordinate bond and the target coordinate bond at the metal center;

[0076] For the first The coordination environment asymmetry parameter of a group of multi-coordination system samples is used to characterize whether the competing coordination bonds are uniformly distributed in space. If the competing bonds are uniformly distributed in direction, the vector sum is close to 0. If the competing bonds are concentrated on one side, the vector sum becomes larger.

[0077]

[0078] For the first In the multi-coordination system samples, from the metal center to the first Vectors of competing coordinating atoms;

[0079] For the first Spatial occlusion parameters of multi-coordination system samples are used to characterize the degree of spatial shielding of the target coordinating atom by the competing atom;

[0080]

[0081] For the first The first group of multi-coordination system samples van der Waals radii of competing coordinating atoms;

[0082] Step 3: Use machine learning to predict the bond length correction of the target coordination bond during the evolution from a low-convergence approximate configuration to a high-convergence stable configuration;

[0083] Construct a training set, where each group of samples in the training set is... ;

[0084] A prediction model is constructed to realize a deviation compensation prediction model for the migration mapping from low convergence stage structure to high convergence stage structure, and the target coordination bond length is predicted based on the model.

[0085] Training with a training set enables the prediction model to utilize competitive coordination context information from the input low-convergence stage structural data. Output the bond length error of the target coordination bond between the high-convergence stage and the low-convergence stage. ;

[0086] The prediction model uses the structural quality transfer deviation function:

[0087]

[0088] in, For local interference terms, This is a global statistical item. This refers to three-dimensional environmental statistics. For constant terms;

[0089]

[0090]

[0091]

[0092] in, and These are the weighting coefficients for the local interference term; These are the weighting coefficients for global statistical items; and These are the weighting coefficients for the three-dimensional environmental statistics items;

[0093] In one embodiment of the present invention, the aforementioned weight coefficients and constant terms can be solved by machine learning methods such as least squares or regression with regularization terms.

[0094] Step 4: For the target actinide multi-coordination system to be predicted, obtain the structural data of the low convergence stage, select one coordination bond as the target coordination bond, and determine the bond length and competing coordination context information of the target coordination bond; input the competing coordination context information into the trained prediction model to obtain the bond length error, and take the bond length of the target coordination bond and the bond length error as the bond length prediction result of the target coordination bond; select other coordination bonds as target coordination bonds in turn, and repeat the above process until the bond length prediction results of all coordination bonds are obtained.

[0095] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for predicting actinide coordination bond lengths considering multi-stage structural information transfer and coordination competition effects, characterized in that: Identify the target actinide multicoordination system to be predicted and obtain samples of multicoordination systems with the same actinide element as the metal center; For each group of multi-coordination system samples, obtain the structural data of the low convergence stage and the high convergence stage, select one coordinating atom as the target coordinating atom, and the remaining coordinating atoms as competing coordinating atoms, and construct the competitive coordination context information based on the structural data of the low convergence stage. The bond length of the target coordination bond is calculated based on the structural data of the low convergence stage and the high convergence stage, respectively, and the bond length error of the target coordination bond is determined. Construct and train a prediction model that can output the bond length error of the target coordination bond based on the input competitive coordination context information; For the target actinide multicoordination system to be predicted, obtain the structural data in the low convergence stage, select a coordination bond as the target coordination bond, and determine the bond length of the target coordination bond and the competing coordination context information. The competitive coordination context information is input into the trained prediction model to obtain the bond length error of the target coordination bond. The sum of the bond length of the target coordination bond and the bond length error is used as the bond length prediction result of the target coordination bond. Select other coordinate bonds in turn as target coordinate bonds and repeat the above process until the bond length prediction results of all coordinate bonds are obtained.

2. The method for predicting actinide coordination bond lengths considering multi-stage structural information migration and coordination competition effects according to claim 1, characterized in that: The structural data includes the spatial coordinates of the metal center and the spatial coordinates of each coordinating atom, and then the bond lengths of the coordinate bonds between the metal center and each coordinating atom are calculated.

3. The method for predicting actinide coordination bond lengths considering multi-stage structural information migration and coordination competition effects according to claim 1, characterized in that: The competitive coordination context information includes the number of competing coordinating atoms, information on competing coordination bonds, distance information between the target coordinating atom and the competing coordinating atom, angle information between the target coordinating bond and the competing coordinating bond, coordination environment asymmetry parameters, and spatial occlusion parameters.

4. The method for predicting actinide coordination bond lengths considering multi-stage structural information migration and coordination competition effects according to claim 3, characterized in that: The competing coordinate key information for: in, For the first The first group of multi-coordination system samples The bond length of the competitive coordination bond between the competing coordinating atoms and the metal center. To calculate the average, To calculate the standard deviation; , For the first The number of competing coordinating atoms in a multi-coordination system sample; The distance information between the target coordinating atom and the competing coordinating atom for: in, For the first The first group of multi-coordination system samples The distance between the competing coordinating atom and the target coordinating atom; The angle information between the target coordinate bond and the competing coordinate bond for: in, For the first The first group of multi-coordination system samples The angle formed between the competing coordinate bond and the target coordinate bond at the center of the metal.

5. The method for predicting actinide coordination bond lengths considering multi-stage structural information migration and coordination competition effects according to claim 4, characterized in that: The coordination environment asymmetric parameters for: in, For the first In the multi-coordination system samples, from the metal center to the first Vectors of competing coordinating atoms; The spatial occlusion parameters for: in, For the first The first group of multi-coordination system samples van der Waals radii of competing coordinating atoms; The competitive coordination context information for: 。 6. The method for predicting actinide coordination bond lengths considering multi-stage structural information migration and coordination competition effects according to claim 5, characterized in that: The bond length of the target coordination bond is calculated based on the structural data during the low-convergence phase. The bond lengths of the target coordination bonds are calculated based on the structural data from the high-convergence phase. Determine the bond length error of the target coordinate bond. ; Construct a training set, where each group of training samples is... ; The prediction model is trained using a training set, enabling it to predict based on the input competitive coordination context information. Output the bond length error of the target coordinate bond .

7. The method for predicting actinide coordination bond lengths considering multi-stage structural information migration and coordination competition effects according to claim 5, characterized in that: The prediction model is specifically as follows: in, For local interference terms, This is a global statistical item. This refers to three-dimensional environmental statistics. For constant terms; in, , , , and These are the weight coefficients to be trained.

8. A computer device, comprising a memory, a processor, and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When executed by a processor, the computer program implements the steps of the method according to any one of claims 1 to 7.

10. A computer program product comprising computer instructions, characterized in that: When executed by a processor, the computer instructions implement the steps of the method according to any one of claims 1 to 7.

Images