Method for exploring solid electrolyte interphase phase heat transfer properties based on machine learning force field

Abstract

The present application relates to the technical field of simulation, in particular to a method for exploring solid electrolyte interface phase heat transfer properties based on machine learning force field, and specifically comprises the following contents: constructing an initial data set; training a machine learning force field; verifying the accuracy of the force field model; and analyzing the thermal conductivity of the solid electrolyte interface phase using molecular dynamics simulation method. The present application develops a machine learning force field for the solid electrolyte interface phase, and studies the thermal conductivity of the solid electrolyte interface phase in combination with molecular dynamics simulation. Compared with the existing technology, the present application trains an efficient and accurate force field model with less data set, and has the advantages of simplicity, low cost, simple operation, good universality, etc., and provides theoretical guidance for analyzing and researching the heat transfer behavior of solid materials.

Description

Technical Field

[0001] This invention relates to the field of simulation technology, and in particular to a method for investigating the interfacial heat transfer properties of solid electrolytes based on machine learning force fields. Background Technology

[0002] Lithium-ion batteries, as a new type of battery characterized by high energy density, lightweight, and environmental friendliness, have been widely used in mobile communications, electric vehicles, and other fields. During the charging and discharging process of lithium-ion batteries, the chemical reactions between the electrodes and the electrolyte, as well as ion migration, generate a significant amount of heat, leading to an increase in the internal temperature of the battery and thus affecting its performance and lifespan. Therefore, studying the thermal conductivity properties of lithium-ion batteries can provide important theoretical support for battery design and manufacturing, thereby improving battery safety performance. In lithium-ion batteries, the study of heat transfer properties mainly involves three parts: the electrodes, the electrode solution, and the solid-electrolyte interface. The solid-electrolyte interface phase is a thin film formed by the reaction between the electrolyte and the electrode surface. It can limit the reaction between the electrodes and the electrolyte, protecting the electrodes, improving battery efficiency, and extending battery life. However, current research on the heat transfer properties of solid-electrolytes is lacking. Therefore, a deeper understanding of the heat transfer properties of the solid-electrolyte interface at the microscopic level is of great significance for optimizing battery performance and lifespan.

[0003] With the development of computer technology and the construction of theoretical frameworks and simulation methods, computer simulation technology has been widely used to study the thermal transfer properties of materials, and can help experimental scientists understand the microscopic mechanisms of heat transfer properties at the atomic level. Currently, commonly used simulation methods for studying heat transfer in materials mainly include density functional theory (DFT) calculations and molecular dynamics simulations. DFT calculations primarily study the thermal transfer properties of materials by calculating force constants. Although DFT calculations have high accuracy, their computational cost is very high, especially for complex amorphous materials requiring substantial computational resources. Furthermore, the computational cost of DFT typically increases exponentially with the number of atoms, making this method less efficient for large-scale systems. On the other hand, molecular dynamics simulations can calculate thermal conductivity coefficients using temperature gradients caused by thermal packing / heat flow under non-equilibrium simulation conditions. Molecular dynamics simulations have good scalability and can be used for simulation studies of large-scale systems. Their computational cost typically increases linearly with the number of atoms, thus enabling simulations at the million-atom level and reaching microsecond-level timescales. However, because molecular dynamics simulations typically use empirical force fields to describe the interactions between atoms, it is often difficult to obtain accurate simulation results. Summary of the Invention

[0004] This invention aims to at least improve one of the technical problems existing in the prior art. To this end, this invention proposes a method for investigating the heat transfer properties of solid electrolyte interfacial phases based on machine learning force fields.

[0005] A method for investigating the interfacial heat transfer properties of solid electrolytes based on machine learning force fields according to a first aspect of the present invention, comprising:

[0006] Step S1: Obtain the crystal file of the solid electrolyte interface phase material, construct the corresponding unit cell structure model, and optimize the unit cell structure using the Vienna Ab-initio Simulation Package (VASP).

[0007] Step S2: Expand the optimized original cell structure in three directions so that the side length of the box of the expanded supercell structure is greater than 10 angstroms. Use the ab initio simulation software package to perform ab initio molecular dynamics simulation on the supercell structure, record and output the structural information of each frame.

[0008] Step S3: Integrate the structural information of each frame output as a dataset, and then train the machine learning force field of the solid electrolyte interface phase material using the Machine Learning Interatomic Potential (MLIP) software package.

[0009] Step S4: Calculate the phonon spectrum of the solid electrolyte interface phase material using machine learning force field calculation;

[0010] Step S5: Determine whether the phonon spectrum results calculated by machine learning force field are consistent with the density functional theory results to determine the accuracy of the force field. If they are consistent, proceed to step S6; otherwise, proceed to step S2 to perform a new ab initio molecular dynamics simulation to supplement the dataset.

[0011] Step S6: Combine the machine learning force field with the molecular dynamics simulation software package (Large-scale Atomic / Molecular Massively Parallel Simulator, LAMMPS) to perform equilibrium dynamics simulation at a predetermined temperature, and then perform non-equilibrium dynamics simulation. Calculate the thermal conductivity of the solid electrolyte interface phase material based on the temperature gradient data generated by the non-equilibrium dynamics simulation.

[0012] This invention discloses a method for investigating the heat transfer properties of solid-state electrolyte interfacial phases based on machine learning force fields. Based on a developed machine learning force field, the LAMMPS software package is used to perform non-equilibrium dynamic simulations of the solid-state electrolyte interfacial phase material. During the non-equilibrium dynamic simulation, a heat source and a cold source are introduced to opposite sides of the simulation chamber. After the heat transfer process reaches a steady state, the temperatures at different locations in the heat transfer region are statistically analyzed and linearly fitted to obtain temperature gradient curves. Finally, the thermal conductivity of the solid-state electrolyte interfacial phase material is calculated. This invention provides a theoretical model for studying the heat transfer properties of different types of solid-state electrolyte interfacial phase materials and provides theoretical guidance for the design and development of novel high thermal conductivity solid-state electrolyte interfacial phase materials for lithium batteries.

[0013] In one possible implementation of the first aspect, step S4 specifically includes calculating the total energy, the forces on atoms, and the phonon spectrum of the test set system using the MLIP software package based on the machine learning force field, and the calculation results are used to verify the accuracy of the machine learning force field.

[0014] In one possible implementation of the first aspect, the temperatures of the heat source and the cold source in the non-equilibrium dynamic simulation of step S6 are respectively and , where T is the temperature set in the equilibrium dynamics simulation. By constructing heat sources and cold sources, a temperature gradient can be formed within the system.

[0015] In one possible implementation of the first aspect, the ensemble in the non-equilibrium dynamics simulation in step S6 is a micro-canonical ensemble (NVE ensemble), and the temperature control method is the Langevin method. Based on the Langevin temperature control method, energy transfer and distribution within the material can be better controlled in heat conduction simulations, further ensuring energy conservation within the system, thereby better controlling energy changes during heat conduction.

[0016] In one possible implementation of the first aspect, the formula for calculating the thermal conductivity of the solid electrolyte interface phase material in step S6 is: ,

[0017] in It is thermal conductivity. It is a temperature gradient. It is energy flux. It is the cross-sectional area.

[0018] In one possible implementation of the first aspect, step S3 specifically includes dividing the dataset into a training set and a test set in an 8:2 ratio, wherein the training set is used for training the machine learning force field, and the test set is used for validating the machine learning force field. The machine learning force field training process uses the BFGS (Broyden-Fletcher-Goldfarb-Shanno) method for iterative optimization, giving the invention advantages such as fast convergence speed and low computational complexity.

[0019] In one possible implementation of the first aspect, step S2 specifically includes using the VASP software package to perform ab initio molecular dynamics simulations as an NVT ensemble, with a time step of 1.0 fs and simulation temperatures of 50 K, 300 K, 500 K, and 700 K. NVT simulations at different temperatures can be used to generate a series of different structural information.

[0020] In one possible implementation of the first aspect, the structural information in step S2 includes energy, coordinates, force, and stress tensor for machine learning force field training.

[0021] In one possible implementation of the first aspect, the cutoff energy for optimizing the unit cell structure using the ab initio simulation software package in step S1 is 600 eV, and the K-point is... The electronic convergence criterion is Based on this parameter setting, an accurate lattice constant can be obtained.

[0022] A system for investigating the heat transfer properties of a solid electrolyte interfacial phase based on a machine learning force field, according to a second aspect of the present invention, includes calculating the thermal conductivity of the solid electrolyte interfacial phase material using the aforementioned method for investigating the heat transfer properties of a solid electrolyte interfacial phase based on a machine learning force field.

[0023] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0024] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 This is a flowchart of a method for investigating the interfacial heat transfer properties of solid electrolytes based on machine learning force fields, according to an embodiment of the present invention.

[0026] Figure 2This is a comparison chart of density functional theory calculation results and machine learning force field prediction results according to an embodiment of the present invention;

[0027] Figure 3 This is a comparison diagram of phonon dispersion curves calculated by density functional theory and predicted by machine learning force field according to an embodiment of the present invention.

[0028] Figure 4 This is a schematic diagram of the temperature distribution curves of lithium fluoride, a solid electrolyte interface phase material, at different positions in the z-direction under different temperature conditions in an embodiment of the present invention.

[0029] Figure 5 The thermal conductivity of lithium fluoride, the solid electrolyte interface material in this invention, under different temperature conditions is shown in the embodiments of the present invention. Detailed Implementation

[0030] The embodiments of the present invention are described in detail below. The embodiments described with reference to the accompanying drawings are exemplary. It should be understood that the specific embodiments described herein are merely illustrative of this application and are not intended to limit this application.

[0031] It should be noted that when a component is said to be "fixed to" another component, it can be directly on the other component or there may be an intervening component. When a component is said to be "connected to" another component, it can be directly connected to the other component or there may be an intervening component.

[0032] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0033] Example 1

[0034] See Figure 1 As shown, this embodiment provides a method for investigating the heat transfer properties of solid electrolyte interface phases based on machine learning force fields. The study of heat transfer properties of solid electrolyte interface phase materials using molecular dynamics simulations is highly dependent on the accuracy of the force field; however, currently, there is a lack of high-precision force fields to describe the interactions between atoms in solid electrolyte interface phase materials. This embodiment, by combining a developed machine learning force field and molecular dynamics simulations, can accurately calculate the thermal conductivity of solid electrolyte interface phase materials. This includes:

[0035] Step S1: Obtain the crystal file of the solid electrolyte interface phase material, then construct the corresponding unit cell structure model, and optimize the unit cell structure using the VASP software package;

[0036] Step S2: Expand the optimized original cell structure in three directions so that the side length of the box of the expanded supercell structure is greater than 10 angstroms. Then, use the VASP software package to perform ab initio molecular dynamics simulation on the supercell structure, record and output the structural information of each frame.

[0037] Step S3: Integrate the structural information of each frame output as a dataset, and then train the machine learning force field of the solid electrolyte interface phase material using the MLIP software package.

[0038] Step S4: Calculate the phonon spectrum of the solid electrolyte interface phase material using machine learning force field calculation;

[0039] Step S5: Determine whether the phonon spectrum results calculated by machine learning force field are consistent with the density functional theory results to determine the accuracy of the force field. If they are consistent, proceed to step S6; otherwise, proceed to step S2 to perform a new ab initio molecular dynamics simulation to supplement the dataset.

[0040] Step S6: Combine the machine learning force field with the LAMMPS software package to perform equilibrium dynamics simulation at a predetermined temperature, and then perform non-equilibrium dynamics simulation. Calculate the thermal conductivity of the solid electrolyte interface phase material based on the temperature gradient data generated by the non-equilibrium dynamics simulation.

[0041] This invention discloses a method for investigating the heat transfer properties of solid-state electrolyte interfacial phases based on machine learning force fields. Based on a developed machine learning force field, the LAMMPS software package is used to perform non-equilibrium dynamic simulations of the solid-state electrolyte interfacial phase material. During the non-equilibrium dynamic simulation, a heat source and a cold source are introduced to opposite sides of the simulation chamber. After the heat transfer process reaches a steady state, the temperatures at different locations in the heat transfer region are statistically analyzed and linearly fitted to obtain temperature gradient curves. Finally, the thermal conductivity of the solid-state electrolyte interfacial phase material is calculated. This invention provides a theoretical model for studying the heat transfer properties of different types of solid-state electrolyte interfacial phase materials and provides theoretical guidance for the design and development of novel high thermal conductivity solid-state electrolyte interfacial phase materials for lithium batteries.

[0042] It should be noted that, in step S1, when optimizing the unit cell structure of the solid electrolyte interface phase material using the VASP software package, the convergence criterion for the force is 0.001 eV / Å, and the convergence criterion for the electronic step is 1.0 eV / Å. -8 eV, cutoff energy is 600eV, K-point is During the optimization process, the box side length of the unit cell can be changed, that is, the value of the ISIF parameter is 3.

[0043] It should be noted that in step S2, ab initio molecular dynamics simulations of the solid electrolyte interface material are performed using the VASP software package, employing the NVT ensemble, with a cutoff energy of 600 eV and an electronic step convergence criterion of 1.0 eV. -5 eV, point K is The simulation temperatures were 50K, 300K, 500K, and 700K. The supercell structure of the solid electrolyte interface material was compressed and stretched by 10%, and NVT simulations were performed under the above different temperature conditions. The simulation time was 2.0 ps. The structural information (including energy, coordinates, forces, and stress tensors) of each frame in the dynamic simulation trajectory was recorded and saved.

[0044] It should be noted that in step S3, the structural information obtained in step S2 is integrated into a dataset using the MLIP software package, and the dataset is divided into a training set and a test set in an 8:2 ratio. The weights of energy, force, and stress tensors are 1.0, 0.1, and 0.001, respectively. The BFGS method is used for iterative optimization during the machine learning force field training process using the MLIP software package.

[0045] It should be noted that step S4 specifically includes calculating the total energy, the forces on the atoms, and the phonon spectrum of the test set system using the MLIP software package based on the machine learning force field.

[0046] It should be noted that during the non-equilibrium dynamics simulation in step S6, a heat source and a cold source are respectively placed on the left and right sides of the solid electrolyte interface phase material, and the temperatures of the heat source and the cold source are respectively... and , where T is the target temperature under study.

[0047] It should be noted that the ensemble in the non-equilibrium dynamics simulation in step S6 is the NVE ensemble, the time step is 1fs, the simulation time is 10 ns, and the temperature control method used is the Langevin method, which is used to calculate the temperature at different locations in the heat transfer region after the simulation is completed.

[0048] It should be noted that the formula for calculating the thermal conductivity of the solid electrolyte interface phase material in step S6 is as follows: ,

[0049] in It is thermal conductivity. It is a temperature gradient. It is energy flux. It is the cross-sectional area.

[0050] like Figure 2 The figure shown is a comparison between density functional theory calculation results and machine learning force field prediction results. Figure 2Figures (a) and (c) in the figure show the energy and force prediction results of the lithium fluoride system based on the training set data, respectively. Figure 2 Figures (b) and (d) show the energy and force prediction results of the lithium fluoride system in the test set data, respectively. The machine learning force field of the solid electrolyte interface phase material developed in this embodiment can well replicate the density functional theory calculation results in terms of energy and atomic force in both the training and test sets, with a root mean square error of only 2.00 × 10⁻⁶ for energy. -5 eV / atom, while the root mean square error of atomic force is only 4.89 MeV / Å.

[0051] like Figure 3 The figure shows a comparison of phonon dispersion curves calculated by density functional theory (DFT) and predicted by machine learning force fields. The DFT-calculated phonon dispersion curve is represented by a solid line, while the machine learning-predicted phonon dispersion curve is represented by a dashed line. The phonon spectrum calculated by the machine learning force field of the solid electrolyte interface material developed in this invention is almost identical to the DFT calculation result, indicating that the machine learning force field trained in this embodiment has an accuracy close to that of DFT calculations. Therefore, the machine learning force field developed in this invention can accurately describe the phonon vibrations of solid electrolyte interface materials.

[0052] like Figure 4 The figure shows a schematic diagram of the temperature distribution curves of lithium fluoride, a solid electrolyte interface material, at different positions along the z-axis under different temperature conditions. Figure 4 From (a) we can see that the temperature distribution curves of lithium fluoride at different positions along the z-direction at 300K show that the temperature of the lithium fluoride phase at the solid electrolyte interface in the heat transfer region changes linearly with position. Figure 4 From (b), we can see that the temperature distribution curves of lithium fluoride at different positions along the z-direction at 400K show that the temperature of the lithium fluoride phase at the solid electrolyte interface in the heat transfer region changes linearly with position. Figure 4 As shown in (c), the temperature distribution curves of lithium fluoride at different positions along the z-direction at 500K show that the temperature of the lithium fluoride at the solid electrolyte interface phase in the heat transfer region changes linearly with position. The thermal conductivity of the lithium fluoride material can be obtained by fitting the temperature gradient using a program and combining it with the thermal conductivity formula. Figure 5 As shown, the thermal conductivity of lithium fluoride, a solid electrolyte interface material, is measured under different temperature conditions. The calculated thermal conductivity of lithium fluoride at 300, 400, and 500 K is 17.17, 12.67, and 10.07 W / mK, respectively. The trend of thermal conductivity variation is consistent with the results of DFT calculation.

[0053] Example 2

[0054] This embodiment provides a system for investigating the heat transfer properties of solid electrolyte interfacial phases based on machine learning force fields, including calculating the thermal conductivity of solid electrolyte interfacial phase materials using the aforementioned method for investigating the heat transfer properties of solid electrolyte interfacial phases based on machine learning force fields.

[0055] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," "counterclockwise," "axial," "radial," and "circumferential" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention.

[0056] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example.

[0057] Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. The reference to "embodiment" herein means that a specific feature, structure, or characteristic described in connection with an embodiment can be included in at least one embodiment of this application. The appearance of this phrase in various places in the specification does not necessarily indicate the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0058] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. A method for exploring solid-solid interfacial phase heat transport properties based on machine-learned force fields, characterized in that, The method comprises the following steps: Step S1, obtaining a crystal file of a solid electrolyte interface phase material, constructing a corresponding unit cell structure model, and optimizing the unit cell structure by using an ab initio simulation software package; Step S2, expanding the optimized unit cell structure in three directions respectively, making the box side length of the expanded supercell structure greater than 10 angstroms, and performing ab initio molecular dynamics simulation on the supercell structure by using the ab initio simulation software package, and recording and outputting the structure information of each frame; Step S3, integrating the output structure information of each frame as a data set, and then training a machine learning force field of the solid electrolyte interface phase material by using a machine learning force field software package; Step S4, calculating the phonon spectrum of the solid electrolyte interface phase material by using the machine learning force field; Step S5, determining whether the phonon spectrum result calculated by the machine learning force field is consistent with the density functional theory result, if consistent, jumping to step S6, if not consistent, jumping to step S2; Step S6, combining the machine learning force field with a molecular dynamics simulation software package to perform equilibrium dynamics simulation at a predetermined temperature, and then performing non-equilibrium dynamics simulation, and calculating the thermal conductivity of the solid electrolyte interface phase material based on the temperature gradient data generated by the non-equilibrium dynamics simulation.

2. The method of exploring solid-state electrolyte interfacial phase heat transfer properties based on machine-learned force fields of claim 1, wherein, The step S4 specifically comprises calculating the total energy, atomic force and phonon spectrum of the test set system by using the machine learning force field software package based on the machine learning force field.

3. The method of exploring solid-state electrolyte interfacial phase heat transfer properties based on machine-learned force fields of claim 1, wherein, The temperatures of the heat source and the cold source in the non-equilibrium dynamics simulation of step S6 are respectively and where T is the target temperature of the study.

4. The method of exploring solid-state electrolyte interfacial phase heat transfer properties based on machine-learned force fields of claim 1, wherein, In the step S6, the system in the non-equilibrium dynamics simulation is a microcanonical system, and the temperature control method combined is a Langevin method.

5. The method of exploring solid-state electrolyte interfacial phase heat transfer properties based on machine-learned force fields of claim 1, wherein, The formula for calculating the thermal conductivity of the solid electrolyte interface phase material in step S6 is: , wherein is the thermal conductivity, is the temperature gradient, is the energy flux, is the cross-sectional area.

6. The method of exploring solid-state electrolyte interfacial phase heat transfer properties based on machine-learned force fields of claim 1, wherein, The step S3 specifically comprises dividing the data set into a training set and a test set in a ratio of 8:2, and using a BFGS method for iterative optimization during the training process of the machine learning force field.

7. The method of exploring solid-state electrolyte interfacial phase heat transfer properties based on machine-learned force fields of claim 1, wherein, In the step S2, the system for ab initio molecular dynamics simulation by using the VASP software is a canonical system, the time step is 1.0 fs, and the simulation temperatures are 50K, 300K, 500K and 700K.

8. The method of exploring solid-state electrolyte interfacial phase heat transfer properties based on machine-learned force fields of claim 1, wherein, The structure information in the step S2 includes energy, coordinates, force and stress tensor.

9. The method of exploring solid-state electrolyte interfacial phase heat transfer properties based on machine-learned force fields of claim 1, wherein, The cut-off energy in the step S1 is 600 eV, the K-point is , and the electron convergence criterion is .

10. A system for investigating the interfacial heat transfer properties of solid electrolytes based on machine learning force fields, characterized in that, The method comprises adopting the method for exploring the heat transfer properties of the solid electrolyte interface phase based on the machine learning force field according to any one of claims 1-9 to calculate the thermal conductivity of the solid electrolyte interface phase material.

Images