Low power consumption phase change superlattice material design method and system

Abstract

The present application provides a method and system for designing a low-power phase-change superlattice material, which relates to the technical field of phase-change superlattice materials. The method includes: constructing an experimental model and a control model, performing first-principles calculations on phase-change superlattice materials and crystal materials, and generating a machine learning potential using a machine learning potential model according to the calculation results; using the machine learning potential to simulate the melting process of the phase-change superlattice material and the crystal material, and calculating the simulation results to obtain the local melting characterization results of the phase-change superlattice material; when the local melting characteristic is obtained, the stacking order of the phase-change superlattice material is changed until the lowest melting point of the phase-change superlattice material is obtained, and the corresponding material is used as a low-power phase-change superlattice material. By calculating the simulation results to determine whether the material has the local melting characteristic, and obtaining the lowest melting point of the material after changing the stacking order, the reliability of the designed low-power phase-change superlattice material can be improved.

Description

Technical Field

[0001] The present application relates to the technical field of phase change superlattice materials, and in particular to a low-power consumption phase change superlattice material design method and system. Background Art

[0002] Phase change storage technology is a type of non-volatile storage technology with high read and write speeds and high data storage density. Phase change superlattice material is a material with superlattice structure characteristics and is used in phase change storage technology. Since the main problem of phase change storage technology is high power consumption, some phase change storage technologies reduce power consumption by designing phase change superlattice materials with low power consumption characteristics. When designing low-power phase change superlattice materials, one method is experimental trial and error, that is, the composition and proportion of the elements that make up the superlattice are selected based on experience for experimental acquisition, and then the obtained material is tested based on trial and error, for example, the obtained phase change superlattice material is made into a device, and performance testing is performed to verify whether it has low power consumption characteristics.

[0003] However, when the experimental trial and error method is used to design phase change superlattice materials, the trial and error process is costly and has a low success rate, and it is difficult to obtain a phase change superlattice material with low power consumption characteristics. In some embodiments, low-power phase change superlattice materials can also be designed by theoretical simulation methods, and the theoretical simulation method is mainly carried out by first-principles molecular dynamics simulation or finite element simulation. When the theoretical simulation method is used to design phase change superlattice materials, the simulation scale will be limited to a few hundred atomic scales when simulating by first-principles molecular dynamics, and relatively large-scale phenomena cannot be observed. When finite elements are used for simulation, although the scale is large, finite element simulation is completely dependent on the input parameters and the set model. When establishing the model, different approximate conditions will lead to completely different simulation results, making it impossible to guarantee the accuracy of the results, which in turn leads to low reliability of the designed low-power phase change superlattice materials and cannot meet the requirements. Summary of the invention

[0004] The present application provides a low-power phase-change superlattice material design method and system to solve the problem of low reliability of low-power phase-change superlattice materials designed by theoretical simulation methods.

[0005] The first aspect of the present application provides a low-power consumption phase-change superlattice material design method, comprising:

[0006] Constructing an experimental model and a control model; the experimental model is constructed by a phase-change superlattice material with a layered structure, and the control model is constructed by a crystal material with a cubic phase structure;

[0007] Performing first-principles calculations on the phase-change superlattice material and the crystal material to obtain first-principles calculation results;

[0008] Generate a machine learning potential using a machine learning potential model based on the first principle calculation results;

[0009] Using the machine learning potential to simulate the melting process of the phase change superlattice material and the crystal material, so as to obtain simulation results of the phase change superlattice material and the crystal material respectively;

[0010] Calculating the simulation results of the phase-change superlattice material and the crystal material respectively to obtain the local melting characterization result of the phase-change superlattice material;

[0011] If the local melting characterization result shows that it has local melting characteristics, the stacking order of the phase change superlattice material is changed until the lowest melting point of the phase change superlattice material is obtained, and the phase change superlattice material corresponding to the lowest melting point is used as the low-power phase change superlattice material.

[0012] Optionally, the constructing the experimental model includes:

[0013] Obtaining a basic unit of the phase change superlattice material; the basic unit is a crystal material with a layered structure, and the basic unit has a van der Waals gap;

[0014] stacking the basic units to obtain a superlattice material;

[0015] Recording the relative position vectors of bonding atoms and the relative position vectors of van der Waals interstitial atoms in the superlattice material;

[0016] Expanding the layered structure in the superlattice material using a computing tool according to the relative position vector to obtain a phase change superlattice material;

[0017] The experimental model is constructed using the phase change superlattice material.

[0018] Optionally, generating a machine learning potential using a machine learning potential model according to the first principle calculation result includes:

[0019] Generate a data set using the first principle calculation results;

[0020] Divide the data in the dataset into a training set and a test set;

[0021] The machine learning potential model is trained using the data in the training set to generate a machine learning potential.

[0022] Optionally, the first principles calculation results include atomic positions, atomic forces, total structural energy and structural stress; and generating the machine learning potential using the machine learning potential model according to the first principles calculation results further includes:

[0023] The total structural energy and structural stress in the test set are compared with the total structural energy and structural stress in the machine learning potential model to verify the accuracy of the machine learning potential model.

[0024] Optionally, the using the machine learning potential to simulate the melting process of the phase change superlattice material and the crystal material to obtain simulation results of the phase change superlattice material and the crystal material respectively includes:

[0025] Controlling the phase change superlattice material and the crystal material to maintain a first temperature for a first time period, wherein the first temperature is between 298K and 302K, and the first time period is between 9ps and 20ps;

[0026] Raising the temperature to a temperature greater than the melting point of the phase-change superlattice material and the melting point of the crystalline material respectively, so as to melt the phase-change superlattice material and the crystalline material;

[0027] The molecular dynamics trajectories of the phase-change superlattice material and the crystal material during the melting process are recorded respectively to obtain simulation results of the phase-change superlattice material and the crystal material respectively.

[0028] Optionally, the respectively calculating the simulation results of the phase change superlattice material and the crystal material to obtain the local melting characterization result of the phase change superlattice material includes:

[0029] respectively calculating the bond angle deviation changes of the phase-change superlattice material and the crystal material, so as to compare the melting times of the phase-change superlattice material and the crystal material according to the bond angle deviation changes;

[0030] If the melting time of the phase-change superlattice material is greater than the melting time of the crystal material, obtaining the atomic bond angle deviation distribution of the phase-change superlattice material, and outputting a coexistence result according to the atomic bond angle deviation distribution; the coexistence result is the coexistence result of the melted area and the unmelted area of ​​the phase-change superlattice material during the melting process, and the coexistence result includes coexistence and non-coexistence;

[0031] If the coexistence result is the coexistence, the similarity between the local atomic structure and the lattice atomic structure is calculated by using the atomic position smoothing overlap method, so as to output whether the melting zone distribution of the phase change superlattice material is localized according to the similarity;

[0032] If the melting region is localized, respectively calculating the melting localization of the phase change superlattice material and the crystal material to output a melting localization result;

[0033] The local melting characterization result of the phase change superlattice material is output according to the melting locality result.

[0034] Optionally, different phase-change superlattice materials have different stacking orders; and changing the stacking order of the phase-change superlattice materials until the lowest melting point of the phase-change superlattice material is obtained comprises:

[0035] Changing the stacking sequence of the phase-change superlattice material having the local melting characteristic to obtain an optimized phase-change superlattice material;

[0036] Heating the optimized phase-change superlattice material to obtain the melting point of the optimized phase-change superlattice material;

[0037] The stacking sequence is traversed, and the melting point of the corresponding optimized phase-change superlattice material is acquired, so as to output the lowest melting point value according to the melting point of the optimized phase-change superlattice material.

[0038] Optionally, the step of heating the optimized phase-change superlattice material to obtain the melting point of the optimized phase-change superlattice material includes:

[0039] Calculating the atomic distribution change during the heating process using the radial distribution function, so as to output the destruction result of the long-range order of the optimized phase change superlattice material according to the atomic distribution change;

[0040] When the destruction result of the long-range order is destruction, a second temperature of the optimized phase-change superlattice material is acquired, and the second temperature is output as the melting point of the optimized phase-change superlattice material.

[0041] A second aspect of the present application provides a low-power consumption phase-change superlattice material design system, which is applied to the method provided in the first aspect, and the system comprises:

[0042] Construction module: used to construct an experimental model and a control model; the experimental model is constructed by a phase-change superlattice material with a layered structure, and the control model is constructed by a crystal material with a cubic phase structure;

[0043] A calculation module: used for performing first-principles calculations on the phase-change superlattice material and the crystal material to obtain first-principles calculation results; and for calculating the simulation results of the phase-change superlattice material and the crystal material respectively to obtain local melting characterization results of the phase-change superlattice material;

[0044] A generation module is used to generate a machine learning potential using a machine learning potential model according to the calculation result of the first principle;

[0045] A simulation module: used for simulating the melting process of the phase change superlattice material and the crystal material by using the machine learning potential to obtain simulation results of the phase change superlattice material and the crystal material respectively;

[0046] Optimization output module: used to change the stacking order of the phase change superlattice material when the local melting characterization result shows that it has local melting characteristics, until the lowest melting point of the phase change superlattice material is obtained, and use the phase change superlattice material corresponding to the lowest melting point as the low-power phase change superlattice material.

[0047] The present application provides a method and system for designing low-power phase-change superlattice materials, including: constructing an experimental model and a control model; performing first-principles calculations on phase-change superlattice materials and crystal materials, and generating machine learning potentials using a machine learning potential model based on the calculation results; using machine learning potentials to simulate the melting process of phase-change superlattice materials and crystal materials to obtain simulation results; calculating the simulation results to obtain local melting characterization results of phase-change superlattice materials; when having local melting characteristics, changing the stacking order of phase-change superlattice materials until the lowest melting point of the phase-change superlattice materials is obtained, and using the corresponding materials as low-power phase-change superlattice materials. In the simulation process, the method can improve the accuracy of the simulation by expanding the scale of the simulation, thereby improving the reliability of the low-power phase-change superlattice materials designed by the theoretical simulation method. By calculating the simulation results to determine whether the material has the characteristics of local melting, and obtaining the lowest melting point of the material after changing the stacking order, the reliability of the designed low-power phase-change superlattice materials can be further improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solution of the present application, the drawings required for use in the embodiments are briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0049] Figure 1 A schematic diagram of the process flow of the low-power phase-change superlattice material design method provided in this application;

[0050] Figure 2 A schematic diagram of the process of constructing the experimental model provided for this application;

[0051] Figure 3 A schematic diagram of the process of obtaining simulation results of phase change superlattice materials and crystal materials provided in this application;

[0052] Figure 4 A schematic diagram of the process for obtaining the local melting characterization results of phase change superlattice materials provided in this application;

[0053] Figure 5 A schematic diagram of the process of obtaining the minimum melting point of a phase-change superlattice material provided in this application. DETAILED DESCRIPTION

[0054] The following embodiments are described in detail, and examples thereof are shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The implementations described in the following embodiments do not represent all implementations consistent with the present application. They are only examples of systems and methods consistent with some aspects of the present application as detailed in the claims.

[0055] When designing new phase-change superlattice materials with low power consumption characteristics through theoretical simulation methods, it is mainly carried out through first-principles molecular dynamics simulation or finite element simulation. However, when simulating through first-principles molecular dynamics, the simulation scale will be limited to a few hundred atoms, and relatively large-scale phenomena cannot be observed. When simulating using finite elements, although the scale is large, finite element simulation is completely dependent on input parameters and set models. Different approximate conditions when establishing the model will lead to completely different simulation results, making it impossible to guarantee the accuracy of the results, which in turn leads to low reliability of the designed low-power phase-change superlattice materials and failure to meet the requirements.

[0056] In order to solve the problem of low reliability of low-power phase-change superlattice materials designed by theoretical simulation methods, some embodiments of the present application first provide a method for designing low-power phase-change superlattice materials, see Figure 1 , methods include:

[0057] S100: Construct experimental model and control model.

[0058] Among them, the experimental model is constructed with a layered phase change superlattice material to obtain low-power phase change superlattice materials faster. The control model is constructed with a cubic phase structured crystal material, which is a relatively stable form of crystal structure with a high degree of symmetry and can maintain relatively stable performance under various conditions. Therefore, using the cubic phase structured crystal material as a control group can provide a stable benchmark for evaluating the effects of phase change superlattice materials in terms of structural changes. It can be understood that the number of phase change superlattice materials and crystal materials in the model is multiple.

[0059] It should also be noted that when constructing the control model, the anions and cations of the crystal material are first distinguished during the modeling process. The cations and anions each occupy a set of lattice points in the cubic phase structure. Due to the different valence states of the elements, the number of cations may be less than the number of anions. During the modeling process, vacancies should be randomly distributed on the cation sites to keep the stoichiometric ratio of the crystal material from shifting. Since the melting process of the phase change superlattice material and the crystal material needs to be observed in the subsequent process, the modeling scale should be at least above 3000 atoms to prevent the local melting characterization results from being obtained due to the small modeling scale.

[0060] S200: performing first-principles calculations on the phase-change superlattice material and the crystal material to obtain first-principles calculation results.

[0061] In some embodiments, the first principles calculation results include atomic positions, atomic forces, total structural energy and structural stress. These results should include the local atomic structure that the phase change superlattice material may be involved in during its working process. Therefore, the phase change superlattice material in the experimental model and the crystal material in the control model should be used to perform first principles molecular dynamics simulations, and the atomic structure in the trajectory should be intercepted for more accurate static calculations to obtain the first principles calculation results.

[0062] S300: Generate a machine learning potential using a machine learning potential model according to the first principle calculation results.

[0063] The machine learning potential is used to describe the atomic interactions of the phase change superlattice material in the experimental model and the atomic interactions of the crystal material in the control model. Since phase change superlattice materials with multiple components and stacking orders require large-scale molecular dynamics simulations, and the simulation scale using first-principles molecular dynamics is very limited, in the embodiments of the present application, a machine learning potential model is used to generate a machine learning potential based on the first-principles calculation results, and the simulation scale is expanded using the newly generated machine learning potential, and molecular dynamics simulations are performed on phase change superlattice materials to solve the problem of limited scale when using first-principles molecular dynamics simulations.

[0064] S400: Use machine learning potential to simulate the melting process of phase change superlattice materials and crystal materials to obtain simulation results of phase change superlattice materials and crystal materials respectively.

[0065] Since the low power consumption of the phase-change superlattice material depends on whether it has the local melting characteristic during the melting process, in order to obtain the result of whether the phase-change superlattice material has the local melting characteristic, the generated machine learning potential is used to simulate the melting process of the phase-change superlattice material and the crystal material respectively, so as to output the simulation results of the phase-change superlattice material and the crystal material according to the simulation process. In some embodiments, the melting process simulation adopts the canonical ensemble (NVT), in which the number of particles (N), volume (V), and temperature (T) are conserved. Since the simulation scale of the newly generated machine learning potential is larger than the simulation scale of the first-principles calculation, the reliability of the simulation results can be improved.

[0066] S500: Calculating the simulation results of the phase change superlattice material and the crystal material respectively to obtain the local melting characterization result of the phase change superlattice material.

[0067] Among them, the local melting characterization results of the phase change superlattice material include those with local melting characteristics and those without local melting characteristics. After calculating the simulation results of the phase change superlattice material and the crystal material respectively, the local melting characterization results of the phase change superlattice material can be obtained by comparing the calculation results of the simulation results of the phase change superlattice material with the calculation results of the simulation results of the crystal material.

[0068] S600: If the local melting characterization result shows that the material has local melting characteristics, the stacking order of the phase change superlattice material is changed until the lowest melting point of the phase change superlattice material is obtained, and the phase change superlattice material corresponding to the lowest melting point is used as a low-power phase change superlattice material.

[0069] When the local melting characterization result outputted shows that it has local melting characteristics, it can be preliminarily determined that the obtained phase change superlattice material has the characteristics of low power consumption. At this time, the stacking order of the phase change superlattice material is changed, and the melting point of the phase change superlattice with the changed stacking order is outputted until the lowest melting point value of the phase change superlattice material is obtained. The phase change superlattice material corresponding to the lowest melting point value is the low power consumption phase change superlattice material. By using the phase change superlattice material corresponding to the lowest melting point value as the low power consumption phase change superlattice material, the power consumption of the material can be further reduced, and the reliability of the designed low power consumption phase change superlattice material can be improved.

[0070] The method provided in the present application constructs an experimental model by selecting a phase-change superlattice material with a layered structure, and constructs a control model by selecting a crystal material with a cubic phase structure, uses the trained machine learning potential to simulate the melting of the phase-change superlattice material and the crystal material, and calculates the simulation results, and then compares the calculation results of the phase-change superlattice material with the calculation results of the crystal material to determine whether the phase-change superlattice material has the characteristics of local melting, and then changes the stacking order of the phase-change superlattice material when it has the characteristics of local melting to obtain a low-power phase-change superlattice material. In the simulation process, the method can improve the accuracy of the simulation by expanding the scale of the simulation, thereby improving the reliability of the low-power phase-change superlattice material designed by the theoretical simulation method. By calculating the simulation results to determine whether the material has the characteristics of local melting, and obtaining the lowest value of the melting point of the material after changing the stacking order, the reliability of the designed low-power phase-change superlattice material can be further improved.

[0071] See also Figure 2 In some embodiments, constructing the experimental model includes:

[0072] S110: Obtaining a basic unit of a phase change superlattice material.

[0073] Phase-change superlattice materials are materials with superlattice structures, which are formed by alternating stacking of multiple layered crystals. Therefore, before using phase-change superlattice materials to build experimental models, the crystal structures of the individual crystals that make up the phase-change superlattice materials should be obtained first. Since there may be multiple crystal structures with the same component ratio, in some embodiments, crystal materials with layered arrangement characteristics can be selected from the Materials Project database as the basic unit constituting the superlattice structure, that is, the basic unit of the phase-change superlattice material obtained is a crystal material with a layered structure and has a significant van der Waals gap.

[0074] S120: stacking basic units to obtain a superlattice material.

[0075] The basic units of the superlattice structure can be stacked, such as ABC stacking, to obtain a superlattice material. In some embodiments, the superlattice material can also be structurally optimized to reduce the probability of distortion of the superlattice structure.

[0076] S130: Record the relative position vectors of bonding atoms and the relative position vectors of van der Waals interstitial atoms in the superlattice material.

[0077] S140: Expanding the layered structure in the superlattice material using a computing tool according to the relative position vector to obtain a phase change superlattice material.

[0078] In some embodiments, a python program can be used to expand the layered structure in the superlattice material to obtain a larger-scale phase-change superlattice structure stacked by different layered materials. The material with a phase-change superlattice structure is a phase-change superlattice material.

[0079] S150: Constructing experimental models using phase-change superlattice materials.

[0080] At present, general machine learning potentials are still immature. In the embodiments of the present application, new machine learning potentials are generated by first-principles calculation results of phase change superlattice materials and crystal materials, but the machine learning potentials of specific material systems still need to be trained. The machine learning potentials generated by using the machine learning potential model based on the first-principles calculation results include:

[0081] S310: Generate a data set using first-principles calculation results.

[0082] S320: Divide the data in the data set into a training set and a test set.

[0083] The data set for machine learning potential is composed of a large number of first-principles calculation results. In some embodiments, 20% of the data can be randomly selected from the data set as a test set, and the remaining data can be used as a training set.

[0084] S330: Train the machine learning potential model using the data in the training set to generate a machine learning potential.

[0085] After obtaining the machine learning potential, the machine learning potential model should also be tested to evaluate the accuracy of the model. Therefore, generating the machine learning potential using the machine learning potential model based on the first principle calculation results also includes:

[0086] The total structural energy and structural stress in the test set are compared with the total structural energy and structural stress in the machine learning potential model to verify the accuracy of the machine learning potential model. In some embodiments, the phase change superlattice material can also be structurally optimized by machine learning potential, and the radial distribution function of the optimized structure is calculated and compared with the first principle calculation results to verify the influence of the error of the machine learning potential on the phase change superlattice material.

[0087] See also Figure 3 , using machine learning potential to simulate the melting process of phase change superlattice materials and crystal materials to obtain simulation results of phase change superlattice materials and crystal materials respectively, including the following steps:

[0088] S410: Controlling the phase-change superlattice material and the crystal material to maintain a first temperature for a first period of time.

[0089] The first temperature is between 298K and 302K, and the first time duration is between 9ps and 20ps. Exemplarily, the phase change superlattice material and the crystal material can be maintained at room temperature 300K for 10ps to obtain a stable state at room temperature.

[0090] S420: Raising the temperature to a temperature greater than the melting point of the phase-change superlattice material and the melting point of the crystal material, respectively, so as to melt the phase-change superlattice material and the crystal material.

[0091] S430: recording the molecular dynamics trajectories of the phase change superlattice material and the crystal material during the melting process respectively, so as to obtain simulation results of the phase change superlattice material and the crystal material respectively.

[0092] See also Figure 4 , respectively calculating the simulation results of the phase change superlattice material and the crystal material to obtain the local melting characterization results of the phase change superlattice material includes the following steps:

[0093] S510: respectively calculating the bond angle deviation variation of the phase change superlattice material and the crystal material, so as to compare the melting time of the phase change superlattice material and the crystal material according to the bond angle deviation variation.

[0094] In materials undergoing phase change, the atomic bond angle is near 90° or 180°. If the angles of two different chemical bonds of the same atom are between 45° and 135°, it is considered that the bond angle is formed by the distortion of the 90° bond angle. If the angles of two different chemical bonds of the same atom are between 0° and 45° or 135° and 180°, it is considered that the bond angle is formed by the distortion of the 180° bond angle. The degree of local structural distortion during the melting process of phase change superlattice materials and crystalline materials can be evaluated by calculating the bond angle deviation. For a set of bond angles, the bond angle deviation is the standard deviation of the degree of distortion of each bond angle.

[0095] The melting speed of phase-change superlattice materials and crystal materials can be judged by the bond angle deviation. The phase-change superlattice material in the experimental model and the crystal material in the control model are heated at the same time, and the change in the bond angle deviation during the melting process is analyzed. It can be found that the material first maintains a low bond angle deviation state at room temperature. After the temperature is raised to near the melting point, the bond angle deviation increases. After the material is completely melted, the bond angle deviation further increases, and the bond angle deviation changes little after reaching a stable liquid phase state. Therefore, by observing the change trend of the bond angle deviation of the phase-change superlattice material and the crystal material during the heating and melting process, the change in the bond angle deviation of the phase-change superlattice material and the crystal material can be calculated, and the melting time of the phase-change superlattice material and the crystal material can be compared. Phase change superlattice materials with low power consumption characteristics have a longer melting time than cubic phase crystal materials because the liquid phase nucleation in the initial process of local melting takes a certain amount of time to occur. That is, when the melting time of the phase change superlattice material is greater than the melting time of the crystal material, the phase change superlattice material may be a low power consumption phase change superlattice material.

[0096] S520: If the melting time of the phase-change superlattice material is greater than the melting time of the crystal material, obtain the atomic bond angle deviation distribution of the phase-change superlattice material, and output a coexistence result according to the atomic bond angle deviation distribution.

[0097] When the melting time of the phase change superlattice material is greater than the melting time of the crystal material, it means that the component of the phase change superlattice material is an effective component. It should be noted that the coexistence result is the coexistence result of the melting zone and the unmelted zone of the phase change superlattice material during the melting process, and the coexistence result includes coexistence and non-coexistence. When the material melts, the melted area will affect the lattice of the surrounding unmelted area, causing the adjacent unmelted part to be more distorted. The phase change superlattice material with local melting characteristics has a limited melting zone range, and has a lower structural distortion that can be maintained by a part of the unmelted lattice during the melting process. Therefore, by analyzing the statistics of the bond angle deviation, calculating the bond angle deviation of each atom in the material, and obtaining the distribution of the atomic bond angle deviation, the current melting state of the material can be characterized.

[0098] For cubic phase crystal materials, a large number of vacancies are evenly distributed inside the structure, so the degree of structural distortion in various places will increase after heating. In phase change superlattice materials with local melting characteristics, the influence of the melting zone on the overall structure is limited. Only the local structure of atoms around the melting zone will be greatly distorted, and the atomic structure far away from the melting zone still maintains a low structural distortion. Therefore, after obtaining the atomic bond angle deviation distribution of the phase change superlattice material, if a lower bond angle deviation and a higher bond angle deviation are obtained respectively, it means that the melting zone and the unmelted area can coexist during local melting, that is, the output coexistence result is coexistence.

[0099] S530: If the coexistence result is coexistence, the similarity between the local atomic structure and the lattice atomic structure is calculated using the atomic position smoothing overlap method to output whether the melting zone distribution of the phase change superlattice material is localized according to the similarity.

[0100] In order to observe the distribution of the molten zone, it is necessary to distinguish the molten zone atoms from the lattice atoms. In some embodiments, the Smooth Overlap of Atomic Positions (SOAP) method can be used to distinguish them. SOAP can represent the local structure of atoms within the cutoff radius and quantitatively compare the two local structures. SOAP mainly represents the process and mid-range local structures of atoms, uses radial functions to represent the radial distribution function of coordinated atoms, uses spherical harmonic functions to represent the angular distribution of coordinated atoms, and uses the orthogonality of basis functions and the vector composed of the coefficients in front of the basis functions to quantitatively represent the local atomic structure. When the coexistence result is coexistence, it means that the component of the phase change superlattice material is an effective component. At this time, the similarity between the local atomic structure and the lattice atomic structure in the phase change superlattice material is calculated. In some embodiments, the vector composed of the coefficients can be directly dot-multiplied. If the local atomic structure is exactly the same as the lattice atomic structure, the result is 1, that is, the similarity is 1; if the local atomic structure is greatly different from the lattice atomic structure, the value of the dot-multiplication result is smaller, that is, the value of the similarity is smaller, for example, less than 1. Therefore, according to the calculated similarity result, whether the melt zone distribution of the phase change superlattice material is localized can be output, that is, when the similarity result is equal to 1, the melt zone distribution of the output phase change superlattice material is not localized, and when the similarity result is less than 1, the melt zone distribution of the output phase change superlattice material is localized.

[0101] S540: If the melt is localized, respectively calculate the melting localization of the phase change superlattice material and the crystal material to output the melting localization result.

[0102] If the melting area is localized, it means that the components of the phase change superlattice material are effective components. In some embodiments, the localization of material melting can be analyzed by visualization methods and indirect physical quantities, and the melting characteristics of the phase change superlattice material can be analyzed by defining the degree of melting localization. The overall structure of the phase change superlattice material can be divided into several equal volume regions, and the bond angle deviation of all bond angles in each region is calculated. The following calculation is performed to obtain the localization degree of material melting, which can be expressed by the following formula:

[0103] ;

[0104] in, is the melting locality, is the bond angle deviation of the ith region, is the total number of regions. The larger the value of , the stronger the degree of localized melting.

[0105] S550: Outputting the local melting characterization result of the phase change superlattice material according to the melting locality result.

[0106] The phase-change superlattice material that satisfies the above results is compared with the crystal material in the control model. The crystal material is larger than that in the control model. , it means that the phase-change superlattice material has the characteristics of local melting, and the components of the phase-change superlattice material obtained are effective components.

[0107] By obtaining the phase change superlattice material through judging the entire process mentioned above, invalid components can be eliminated, thereby improving the accuracy of the designed phase change superlattice material components that meet the requirements.

[0108] Different phase-change superlattice materials have different stacking orders. Therefore, in some embodiments, phase-change superlattice materials with different melting points are obtained by changing the stacking order of the phase-change superlattice materials. Figure 5 , changing the stacking order of the phase change superlattice material until the lowest melting point of the phase change superlattice material is obtained includes:

[0109] S610: Changing the stacking sequence of the phase change superlattice material having the local melting characteristic to obtain an optimized phase change superlattice material.

[0110] After determining the composition of the phase-change superlattice material with local melting characteristics, the arrangement order of the phase-change superlattice material blocks is changed while keeping the volume and the composition of the blocks unchanged, so as to obtain an optimized phase-change superlattice material with different structures under the same composition. Since the local melting of the phase-change superlattice material is related to the atomic motion at the van der Waals interface, changing the stacking order will cause the local atomic structure on both sides of the van der Waals interface to change, which in turn causes the melting point of the material to change.

[0111] S620: Raising the temperature of the optimized phase-change superlattice material to obtain a melting point of the optimized phase-change superlattice material.

[0112] The melting point of the optimized phase-change superlattice material can be obtained by heating the optimized phase-change superlattice material, which specifically includes the following steps:

[0113] S621: Calculate the atomic distribution changes during the heating process using the radial distribution function, and output the optimized destruction results of the long-range order of the phase change superlattice material according to the atomic distribution changes.

[0114] The radial distribution function can be expressed as:

[0115] ;

[0116] in, is the radial distribution function, is the number of atoms in the volume element, is the atomic density of the system, is the distance between the volume element and the central atom. In a crystal structure or superlattice structure, the atomic arrangement is long-range ordered, so when calculating the radial distribution function, a series of discrete peaks will be obtained, indicating that other atoms around the central atom are regularly distributed in shells at different distances. When the structure is destroyed by temperature, distorted, or even melted, the long-range order of the structure will be destroyed, the distribution of the surrounding atoms will become irregular, and the radial distribution function will tend to be flat in the long-range part, and a relatively regular atomic distribution cannot be obtained. In the process of molecular dynamics simulation with machine learning potential, different optimized phase change superlattice materials can be slowly heated at the same time, and the atomic distribution changes of the radial distribution function during the heating process can be calculated to determine the degree of destruction of the long-range order during the melting process.

[0117] S622: When the destruction result of the long-range order is destruction, obtaining a second temperature of the optimized phase-change superlattice material, and outputting the second temperature as a melting point of the optimized phase-change superlattice material.

[0118] If the long-range ordered structure corresponding to certain stacking order modes is destroyed at a lower temperature, it means that this type of stacking order has a lower melting point. If the material can still maintain good long-range order in the radial distribution function when it is heated to a higher temperature under certain stacking order modes, it means that these structures have a higher melting point. By slowly heating up the material to obtain the second temperature corresponding to the destruction of the long-range order, the melting point of the material with different stacking order modes can be obtained, that is, when the destruction result of the long-range order is destruction, the second temperature obtained is the melting point of the material.

[0119] S630: traverse the stacking sequence and obtain the melting point of the corresponding optimized phase-change superlattice material to output the lowest melting point value according to the melting point of the optimized phase-change superlattice material.

[0120] Phase-change superlattice materials with lower melting points can reduce energy consumption during heating and melting, thereby reducing power consumption of the device. According to the above step S622, after traversing the stacking sequence, the lowest melting point value in the optimized phase-change superlattice material is obtained, and the phase-change superlattice material corresponding to the lowest melting point value is the low-power phase-change superlattice material.

[0121] Some embodiments of the present application also provide a low-power consumption phase-change superlattice material design system, which is applied to the methods provided in some of the above embodiments, and the system includes:

[0122] Building module: used to build experimental models and control models.

[0123] Among them, the experimental model is constructed by layered phase change superlattice materials, and the control model is constructed by cubic phase structured crystal materials.

[0124] Calculation module: used to perform first-principles calculations on phase-change superlattice materials and crystal materials to obtain first-principles calculation results; and to calculate the simulation results of phase-change superlattice materials and crystal materials respectively to obtain local melting characterization results of phase-change superlattice materials.

[0125] Generation module: used to generate machine learning potential using the machine learning potential model based on the first principle calculation results.

[0126] Simulation module: used to simulate the melting process of phase change superlattice materials and crystal materials using machine learning potential to obtain simulation results of phase change superlattice materials and crystal materials respectively.

[0127] Optimization output module: used to change the stacking order of the phase change superlattice material when the local melting characterization result shows that it has local melting characteristics, until the lowest melting point of the phase change superlattice material is obtained, and use the phase change superlattice material corresponding to the lowest melting point as a low-power phase change superlattice material.

[0128] It can be seen from the above technical scheme that the embodiment of the present application provides a low-power phase-change superlattice material design method and system, the method includes: constructing an experimental model and a control model; performing first-principles calculations on phase-change superlattice materials and crystal materials to obtain first-principles calculation results; generating machine learning potentials using a machine learning potential model according to the first-principles calculation results; using machine learning potentials to simulate the melting process of phase-change superlattice materials and crystal materials to obtain simulation results of phase-change superlattice materials and crystal materials respectively; calculating the simulation results of phase-change superlattice materials and crystal materials respectively to obtain local melting characterization results of phase-change superlattice materials; if the local melting characterization results are characterized by local melting, the stacking order of phase-change superlattice materials is changed until the lowest melting point of the phase-change superlattice material is obtained, and the phase-change superlattice material corresponding to the lowest melting point is used as a low-power phase-change superlattice material. In the simulation process, the method can improve the accuracy of the simulation by expanding the scale of the simulation, thereby improving the reliability of the low-power phase-change superlattice material designed by the theoretical simulation method. By calculating the simulation results to determine whether the material has local melting characteristics and obtaining the lowest melting point of the material after changing the stacking order, the reliability of the designed low-power phase change superlattice material can be further improved.

[0129] Similar parts between the embodiments provided in this application can be referenced to each other. The specific implementation methods provided above are only a few examples under the general concept of this application and do not constitute a limitation on the protection scope of this application. For those skilled in the art, any other implementation methods expanded based on the scheme of this application without creative work belong to the protection scope of this application.

Claims

1. A method for designing a low-power phase-change superlattice material, characterized in that: include: Constructing an experimental model and a control model; the experimental model is constructed by a phase-change superlattice material with a layered structure, and the control model is constructed by a crystal material with a cubic phase structure; Performing first-principles calculations on the phase-change superlattice material and the crystal material to obtain first-principles calculation results; Generate a machine learning potential using a machine learning potential model based on the first principle calculation results; Using the machine learning potential to simulate the melting process of the phase change superlattice material and the crystal material, so as to obtain simulation results of the phase change superlattice material and the crystal material respectively; Calculating the simulation results of the phase-change superlattice material and the crystal material respectively to obtain the local melting characterization result of the phase-change superlattice material; If the local melting characterization result shows that the phase change superlattice material has local melting characteristics, the stacking order of the phase change superlattice material is changed until the lowest melting point of the phase change superlattice material is obtained, and the phase change superlattice material corresponding to the lowest melting point is used as the low-power phase change superlattice material; The changing of the stacking order of the phase-change superlattice material until the lowest melting point of the phase-change superlattice material is obtained includes: changing the stacking order of the phase-change superlattice material with local melting characteristics to obtain an optimized phase-change superlattice material; heating the optimized phase-change superlattice material to obtain the melting point of the optimized phase-change superlattice material; traversing the stacking order and obtaining the corresponding melting points of the optimized phase-change superlattice material to output the lowest melting point according to the melting point of the optimized phase-change superlattice material; heating the optimized phase-change superlattice material to obtain the melting point of the optimized phase-change superlattice material includes: calculating the atomic distribution changes during the heating process using a radial distribution function to output the long-range order destruction result of the optimized phase-change superlattice material according to the atomic distribution changes; when the long-range order destruction result is destruction, obtaining the second temperature of the optimized phase-change superlattice material, and outputting the second temperature as the melting point of the optimized phase-change superlattice material.

2. The low power consumption phase change superlattice material design method according to claim 1, characterized in that: The construction of the experimental model comprises: Obtaining a basic unit of the phase change superlattice material; the basic unit is a crystal material with a layered structure, and the basic unit has a van der Waals gap; stacking the basic units to obtain a superlattice material; Recording the relative position vectors of bonding atoms and the relative position vectors of van der Waals interstitial atoms in the superlattice material; Expanding the layered structure in the superlattice material using a computing tool according to the relative position vector to obtain a phase change superlattice material; The experimental model is constructed using the phase change superlattice material.

3. The low power consumption phase change superlattice material design method according to claim 1, characterized in that: The generating of the machine learning potential using the machine learning potential model according to the first principle calculation result comprises: Generate a data set using the first principle calculation results; Divide the data in the dataset into a training set and a test set; The machine learning potential model is trained using the data in the training set to generate a machine learning potential.

4. The low power consumption phase change superlattice material design method according to claim 3, characterized in that: The first principles calculation results include atomic positions, atomic forces, total structural energy and structural stress; The generating of the machine learning potential using the machine learning potential model according to the first principle calculation result also includes: The total structural energy and structural stress in the test set are compared with the total structural energy and structural stress in the machine learning potential model to verify the accuracy of the machine learning potential model.

5. The low power consumption phase change superlattice material design method according to claim 1, characterized in that: The using the machine learning potential to simulate the melting process of the phase change superlattice material and the crystal material to obtain simulation results of the phase change superlattice material and the crystal material respectively includes: Controlling the phase change superlattice material and the crystal material to maintain a first temperature for a first time period, wherein the first temperature is between 298K and 302K, and the first time period is between 9ps and 20ps; Raising the temperature to a temperature greater than the melting point of the phase-change superlattice material and the melting point of the crystalline material respectively, so as to melt the phase-change superlattice material and the crystalline material; The molecular dynamics trajectories of the phase-change superlattice material and the crystal material during the melting process are recorded respectively to obtain simulation results of the phase-change superlattice material and the crystal material respectively.

6. The low power consumption phase change superlattice material design method according to claim 1, characterized in that: The calculating the simulation results of the phase change superlattice material and the crystal material respectively to obtain the local melting characterization result of the phase change superlattice material includes: respectively calculating the bond angle deviation changes of the phase-change superlattice material and the crystal material, so as to compare the melting times of the phase-change superlattice material and the crystal material according to the bond angle deviation changes; If the melting time of the phase-change superlattice material is greater than the melting time of the crystal material, obtaining the atomic bond angle deviation distribution of the phase-change superlattice material, and outputting a coexistence result according to the atomic bond angle deviation distribution; the coexistence result is the coexistence result of the melted area and the unmelted area of ​​the phase-change superlattice material during the melting process, and the coexistence result includes coexistence and non-coexistence; If the coexistence result is the coexistence, the similarity between the local atomic structure and the lattice atomic structure is calculated by using the atomic position smoothing overlap method, so as to output whether the melting zone distribution of the phase change superlattice material is localized according to the similarity; If the melting region is localized, respectively calculating the melting localization of the phase change superlattice material and the crystal material to output a melting localization result; The local melting characterization result of the phase change superlattice material is output according to the melting locality result.

7. The low power consumption phase change superlattice material design method according to claim 1, characterized in that: Different phase-change superlattice materials have different stacking sequences.

8. A low-power phase-change superlattice material design system, characterized in that: The method applied to any one of claims 1 to 7, wherein the system comprises: Construction module: used to construct an experimental model and a control model; the experimental model is constructed by a phase-change superlattice material with a layered structure, and the control model is constructed by a crystal material with a cubic phase structure; A calculation module: used for performing first-principles calculations on the phase-change superlattice material and the crystal material to obtain first-principles calculation results; and for calculating the simulation results of the phase-change superlattice material and the crystal material respectively to obtain local melting characterization results of the phase-change superlattice material; A generation module is used to generate a machine learning potential using a machine learning potential model according to the calculation result of the first principle; A simulation module: used for simulating the melting process of the phase change superlattice material and the crystal material by using the machine learning potential to obtain simulation results of the phase change superlattice material and the crystal material respectively; Optimization output module: used to change the stacking order of the phase change superlattice material when the local melting characterization result shows that it has local melting characteristics, until the lowest melting point of the phase change superlattice material is obtained, and use the phase change superlattice material corresponding to the lowest melting point as the low-power phase change superlattice material.

Images