Prediction method, computer program, prediction device, and method of manufacturing porous material

Abstract

To provide a prediction method, computer program and prediction device for predicting absorption characteristics of a porous material, and a method of manufacturing a porous material.SOLUTION: A prediction method disclosed herein involves having a computer perform processing for acquiring an initial structure of a porous material (S11), determining an arrangement of an adsorbate when the adsorbate is adsorbed to the porous material of the initial structure by a simulation (S14) using the Monte Carlo method based on neural network potential as a force field, and obtaining predicted values of adsorption characteristics of the porous material of a structure in accordance with the obtained arrangement.SELECTED DRAWING: Figure 2

Description

Technical Field

[0001] The present invention relates to a prediction method, a computer program, a prediction device, and a method for manufacturing a porous material.

Background Art

[0002] Conventionally, the physical properties of materials have been predicted by simulations using a computer. For example, Patent Document 1 discloses a physical property prediction method that can easily predict the interfacial energy of a metal material, etc., by simulation based on the molecular dynamics method.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] The technique described in Patent Document 1 does not predict the adsorption characteristics of a porous material.

[0005] An object of the present disclosure is to provide a prediction method and the like that can predict the adsorption characteristics of a porous material.

Means for Solving the Problems

[0006] A prediction method according to an aspect of the present disclosure acquires an initial structure of a porous material, and based on a neural network potential as a force field, obtains the arrangement of an adsorbate when the adsorbate is adsorbed on the porous material of the initial structure by simulation using the Monte Carlo method, and a computer executes a process of obtaining a predicted value of the adsorption characteristics of the porous material having a structure corresponding to the obtained arrangement.

[0007] A computer program according to an aspect of the present disclosure acquires an initial structure of a porous material, and obtains an arrangement of an adsorbed substance when the adsorbed substance is adsorbed onto the porous material having the initial structure by simulation using the Monte Carlo method based on a neural network potential as a force field, and causes a computer to execute a process of obtaining a predicted value of the adsorption characteristics of the porous material having a structure corresponding to the obtained arrangement.

[0008] A prediction device according to an aspect of the present disclosure includes a control unit that executes a process of acquiring an initial structure of a porous material, obtaining an arrangement of an adsorbed substance when the adsorbed substance is adsorbed onto the porous material having the initial structure by simulation using the Monte Carlo method based on a neural network potential as a force field, and obtaining a predicted value of the adsorption characteristics of the porous material having a structure corresponding to the obtained arrangement.

[0009] A method for manufacturing a porous material according to an aspect of the present disclosure includes a step of acquiring initial structures of a plurality of porous materials, a step of, for each porous material, obtaining an arrangement of an adsorbed substance when the adsorbed substance is adsorbed onto the porous material having the initial structure by simulation using the Monte Carlo method based on a neural network potential as a force field, a step of obtaining a predicted value of the adsorption characteristics of the porous material having a structure corresponding to the obtained arrangement, a step of selecting a porous material whose obtained predicted value of the adsorption characteristics satisfies a predetermined condition, and a step of obtaining the selected porous material.

Effects of the Invention

[0010] According to the present disclosure, the adsorption characteristics of a porous material can be predicted.

Brief Description of the Drawings

[0011]

Figure 1

Figure 2

Figure 3

Figure 4

Mode for Carrying Out the Invention

[0012] The present disclosure will be specifically described with reference to the drawings showing its embodiments.

[0013] (First Embodiment) FIG. 1 is a block diagram showing a configuration example of a manufacturing system 100. The manufacturing system 100 includes a prediction device 1 and a manufacturing device 2. The manufacturing system 100 searches for substances that can be manufacturing targets by predicting the physical properties of a plurality of candidate substances for research and development of new substances, alternative substances, etc., and manufactures substances according to the search results. The manufacturing system 100 of the present embodiment manufactures a porous material that satisfies desired adsorption characteristics by predicting the adsorption characteristics of the porous material. In the present specification, the "adsorption characteristics" refer to the characteristics of a porous material to adsorb an adsorbate. The "adsorbate" may be an atom or a molecule.

[0014] Hereinafter, as an example, it is assumed that the water adsorption characteristics of a metal-organic framework (hereinafter also referred to as MOF) are predicted. MOF is a material also called a porous coordination polymer, and has a high-surface-area coordination network structure formed by the interaction (for example, coordination bond) between a metal and an organic ligand. Since MOF exhibits the characteristics of adsorbing or desorbing an adsorbate such as water, gas, or organic molecules due to the above-described structure, it is used as a variety of materials.

[0015] Note that the molecules for which adsorption characteristics are to be predicted are not limited to water, and may be, for example, gases, organic molecules, etc. Examples of gases include carbon dioxide, hydrogen, carbon monoxide, oxygen, nitrogen, hydrocarbons having 1 to 4 carbon atoms, noble gases, hydrogen sulfide, ammonia, sulfur oxides, nitrogen oxides, siloxanes, etc. Examples of organic molecules include hydrocarbons having 5 to 8 carbon atoms, alcohols having 1 to 8 carbon atoms, aldehydes having 1 to 8 carbon atoms, carboxylic acids having 1 to 8 carbon atoms, ketones having 1 to 8 carbon atoms, amines having 1 to 8 carbon atoms, esters having 1 to 8 carbon atoms, amides having 1 to 8 carbon atoms, etc. The organic molecules may contain an aromatic ring.

[0016] The porous material for which physical properties are to be predicted is not limited to MOF, and may be, for example, zeolite, porous silica, porous polymer, etc.

[0017] The prediction device 1 is an information processing device capable of various information processing and information transmission and reception, and is, for example, a personal computer, a server computer, a quantum computer, etc. The prediction device 1 predicts the water adsorption characteristics of a plurality of candidate MOFs. Based on the prediction results, an MOF that can satisfy the desired water adsorption characteristics can be selected from among the candidate substances.

[0018] The manufacturing device 2 manufactures the selected MOF. The manufacturing device 2 includes, for example, a mixing unit (not shown) that mixes the raw materials of the MOF, and manufactures the MOF by mixing the raw materials of the MOF. Note that the manufacturing device 2 may be appropriately configured according to the substance to be manufactured.

[0019] As shown in FIG. 1, the prediction device 1 includes a control unit 11, a storage unit 12, a communication unit 13, an input unit 14, an output unit 15, etc. The prediction device 1 may be configured to be composed of a plurality of computers for distributed processing, may be realized by a plurality of virtual machines provided in one server, or may be realized using a cloud server.

[0020] The control unit 11 includes a processor using one or more CPUs (Central Processing Units), GPUs (Graphics Processing Units), etc. The control unit 11 controls each component and executes processing by using a memory such as a built-in ROM (Read Only Memory) or RAM (Random Access Memory), a clock, a counter, etc. Note that the functional units of the control unit 11 may be realized by software, or part or all of them may be realized by hardware such as an ASIC (Application Specific Integrated Circuit) or an FPGA (Field Programmable Gate Array).

[0021] The storage unit 12 includes a non-volatile memory such as a hard disk, a flash memory, an SSD (Solid State Drive), etc. The storage unit 12 may be an external storage device connected to the prediction device 1. The storage unit 12 stores various computer programs and data referred to by the control unit 11. The storage unit 12 of the present embodiment stores a program 1P for causing a computer to execute processing related to the prediction of physical properties.

[0022] A computer program (program product) including the program 1P may be provided by a non-temporary recording medium 1A that records the computer program in a readable manner. The storage unit 12 stores the computer program read from the recording medium 1A by a reading device (not shown). The recording medium 1A is, for example, a magnetic disk, an optical disk, a semiconductor memory, etc. Also, a computer program may be downloaded from an external server connected to a communication network and stored in the storage unit 12. The program 1P may be a single computer program or may be composed of a plurality of computer programs, and may be executed on a single computer or on a plurality of computers interconnected by a communication network.

[0023] The communication unit 13 includes a communication module for communicating with an external device via a network (not shown). The control unit 11 transmits and receives data to and from the external device via the communication unit 13. The communication unit 13 may be omitted.

[0024] The input unit 14 accepts the input of various data necessary for the implementation of physical property prediction, such as initial information used for physical property prediction, calculation conditions, and required physical properties. The input unit 14 sends the received input content to the control unit 11. The input unit 14 includes, for example, a keyboard, a mouse, a touch panel device built into a display, an interface for importing data from the outside, and the like.

[0025] The output unit 15 outputs various data associated with the implementation of physical property prediction, such as the predicted physical properties and the selection result of candidate substances. The output unit 15 outputs various information according to an instruction from the control unit 11. The output unit 15 includes, for example, a display device.

[0026] The prediction device 1 may be configured to receive an operation through an externally connected computer and output information to be notified to the external computer. In this case, the prediction device 1 may not include the input unit 14 and the output unit 15.

[0027] The prediction method of the present embodiment includes a simulation process using the Monte Carlo (MC) method and a simulation process using the molecular dynamics (MD) method, and these two processes are repeatedly executed. By combining two types of simulation methods, the accuracy of physical property prediction can be improved.

[0028] FIG. 2 is a flowchart showing an example of a prediction processing procedure executed by the prediction device 1. The processing in the following flowcharts is executed by the control unit 11 according to a program 1P stored in the storage unit 12 of the prediction device 1.

[0029] The control unit 11 of the prediction device 1 acquires an initial structure corresponding to the composition of the MOF to be simulated (step S11). The initial structure is provisional crystal structure data determined according to the composition and includes information representing the positions of each atom. For example, the control unit 11 determines the initial structure by generating a candidate MOF on a computer and assigning provisional coordinates to each atom constituting the generated MOF. The initial structure acquired in step S11 may be a structure optimized using an optimization method such as the BFGS method.

[0030] The control unit 11 acquires an initial state for the MOF to be simulated (step S12). The initial state is preset according to the MOF to be simulated. The initial state includes, for example, the pressure (partial pressure) of water and the temperature of the system. The partial pressure may be represented by the relative pressure P / P0 of water vapor (the adsorbed substance in the gaseous state). Here, P indicates the equilibrium pressure of water vapor, and P0 indicates the saturated vapor pressure. A plurality of partial pressures P / P0 may be set, and may be one or more values, for example, between 0.05 and 0.5. The initial state is used as the calculation conditions in a series of simulation steps.

[0031] The control unit 11 acquires a force field (potential) to be applied to the simulation (step S13). In this embodiment, a neural network potential (hereinafter also referred to as NNP) is used as the force field. The neural network potential is a type of machine learning potential in which the interaction of atoms is learned using a neural network and has excellent versatility. The neural network potential can be defined using a known method.

[0032] Based on the acquired initial structure, initial state, and force field, the control unit 11 executes a simulation using the Monte Carlo method (step S14) to calculate the optimal arrangement of the particles. Specifically, based on the acquired initial structure and initial state, the control unit 11 generates a model in which particles are arranged in cells used for the simulation. The cell is defined, for example, as a rectangular parallelepiped or a cube. The size of the cell can be appropriately set according to the total number of MOFs and water molecules arranged in the cell. The control unit 11 evaluates the generated model using the Monte Carlo method to calculate the optimal arrangement of the particles when water is adsorbed or desorbed, and obtains the structure after adsorption or desorption. The optimal arrangement may be the most stable and most feasible arrangement.

[0033] The Monte Carlo method is a technique for numerically executing the calculation of the partition function of statistical mechanics in the configuration space. As the Monte Carlo method, the Monte Carlo method of the grand canonical ensemble (hereinafter also referred to as the GCMC method) is preferable. In the GCMC method, the temperature of the system (cell) is controlled to be constant throughout all simulations, and the system can exchange energy and particles (specifically, atoms and molecules).

[0034] The GCMC method calculates the phase equilibrium state of a cell that is a mixed system of MOF and particles. In the GCMC method, with the volume of the cell kept constant, various thermodynamic equilibrium states are reproduced by repeating the probabilistic movement (translation, rotation), insertion, and deletion of particles.

[0035] The control unit 11 performs Monte Carlo calculations according to predetermined calculation conditions. The calculation conditions may be set manually in advance and stored in the storage unit 12. The calculation conditions may include the number of MC steps, boundary conditions, cutoff distance of interactions, trial probability, etc. The trial probability includes the probabilities of movement operations, insertion operations, and deletion operations. The adoption of each operation may be determined by the Metropolis method. The calculation conditions can be appropriately selected according to the prior art.

[0036] Among the above trial probabilities, the trial probabilities for the movement operations, i.e., translation and rotation, may be set to zero. In the present embodiment, the behavior corresponding to the translation and rotation of the MOF and water molecules can be simulated by the molecular dynamics calculation step described later. Therefore, in the Monte Carlo calculation step, the calculation load can be reduced by setting the probability of adopting the trials for translation and rotation to zero.

[0037] For the Monte Carlo calculation, known Monte Carlo calculation software can be used. The final structures of the MOF and the water molecules arranged in the MOF after a predetermined number of MC steps correspond to the structures after performing the simulation using the Monte Carlo method.

[0038] The control unit 11 acquires predicted values of the water adsorption characteristics of the MOF under each partial pressure and temperature (step S15). The water adsorption characteristics may be the amount of water adsorbed (water adsorption amount), or may be an adsorption isotherm showing the relationship between the water adsorption amount and the partial pressure. The adsorption amount is represented by the mass (g) of water (adsorbate) adsorbed per 1 g of the MOF. The water adsorption amount may include, for example, the water adsorption amount for each MC step, the maximum water adsorption amount among a series of those water adsorption amounts, the average value of a series of water adsorption amounts, etc. The water adsorption amount may be represented by a graph showing the relationship between the water adsorption amount of the MOF and the MC steps. The adsorption isotherm is generated by plotting the average value of the water adsorption amount from a predetermined MC step to the final step for each pressure.

[0039] The control unit 11 performs a simulation using the molecular dynamics method on the structures of the MOF and water molecules after performing the simulation using the Monte Carlo method (step S16), and calculates the optimal arrangement of the particles. The molecular dynamics method is a method of simulating the movement of particles by numerically solving Newton's equations of motion. By the molecular dynamics calculation, considering the structural change of the MOF itself, the structure of the MOF after adsorbing or desorbing water can be relaxed.

[0040] The control unit 11 performs molecular dynamics calculations according to preset calculation conditions. The calculation conditions may include the number of MD steps, ensemble, etc. As the ensemble, the canonical ensemble (NVT ensemble) is preferable. In the canonical ensemble, the number of particles, volume, and temperature of the system are maintained constant in the simulation. The method of the canonical ensemble is not particularly limited, and an appropriate method can be used. Note that the force field obtained in step S13 is also applied to the simulation using the molecular dynamics method. In the present embodiment, as described above, NNP is used. Known molecular dynamics calculation software can be used for molecular dynamics calculations.

[0041] After executing the simulation using the molecular dynamics method, the control unit 11 increments the number of repetitions of the simulation and stores the calculation results by the molecular dynamics calculation in the storage unit 12 (step S17). The calculation results include the number of repetitions, the structure of the MOF after executing the simulation using the molecular dynamics method and the water molecules arranged in the MOF (the positions of the water molecules), and the momentum of the water molecules, etc.

[0042] The control unit 11 determines whether to end the simulation by determining whether the counted number of repetitions has reached the preset number of repetitions (step S18).

[0043] If it is determined that the simulation is not ended because the counted number of repetitions is less than the preset number of repetitions (S18: NO), the control unit 11 applies the positions and momenta of the water molecules stored in step S17 as the initial structure in the next cycle (step S19). The control unit 11 returns the process to step S14 and repeats the simulation using the Monte Carlo method and the molecular dynamics method based on the new initial structure. By passing on the positions and momenta of the water molecules after the molecular dynamics calculation in the current cycle to the next cycle, the continuity of the arrangement and movement direction of the water molecules between cycles can be ensured.

[0044] When it is determined that the simulation is to be terminated because the counted number of repetitions is equal to or greater than the preset number of repetitions (S18: YES), the control unit 11 stores the simulation result in the storage unit 12 (step S20). The control unit 11 stores, for example, the initial structure of the MOF, the intermediate structure, the final structure after the simulation, and the predicted value of the water adsorption characteristics, etc. in the storage unit 12 in association with each other.

[0045] The control unit 11 outputs the result information including the predicted value of the water adsorption characteristics through the output unit 15 (step S21). The control unit 11 displays, for example, the initial structure of the MOF, the water adsorption amount, and the information representing the adsorption isotherm on the display device. Step S21 may be omitted. The control unit 11 ends a series of processes.

[0046] In the above process, the acquisition process of the water adsorption characteristics in step S15 may be performed after the execution of the simulation using the molecular dynamics method in step S16.

[0047] The prediction device 1 executes the above-described series of processes for each of a plurality of MOFs to be simulated, and predicts and stores the water adsorption characteristics for each MOF. Thereby, the water adsorption characteristics of a plurality of MOFs as manufacturing candidates are collected. The prediction device 1 selects the MOF to be manufactured based on the prediction results of the water adsorption characteristics of each collected MOF.

[0048] FIG. 3 is a flowchart showing an example of the selection processing procedure executed by the prediction device 1.

[0049] The control unit 11 of the prediction device 1 acquires the required physical properties required for the MOF to be manufactured (step S31). The required physical properties may include the water adsorption amount. The water adsorption amount as the required physical property may be the lower limit value of the amount of water adsorbed (for example, 0.2 g / g). The required physical properties may further include the upper limit value and the lower limit value of the partial pressure (for example, P / P0 is 0.2 to 0.4). The control unit 11 acquires the required physical properties by receiving an input from the user via, for example, the input unit 14. The control unit 11 may acquire the required physical properties by receiving information transmitted from an externally connected device, or may acquire the required physical properties by reading preset required physical properties from the storage unit 12.

[0050] Based on the simulation results stored in the storage unit 12, the control unit 11 selects an MOF that satisfies the required physical properties from among a plurality of MOFs for which the water adsorption characteristics have been predicted (step S32). Specifically, the control unit 11 extracts an MOF whose predicted water adsorption amount is 0.2 g / g or more when P / P0 is 0.2 to 0.4. The control unit 11 may preferentially select a predetermined number of MOFs in order from those with a high water adsorption amount.

[0051] The control unit 11 outputs the selection result through the output unit 15 (step S33). For each selected MOF, the control unit 11 may output information representing the initial structure of the MOF, the water adsorption amount, and the adsorption isotherm in association with each other. The control unit 11 ends a series of processes.

[0052] In the above process, the control unit 11 may acquire the measured values of the adsorption / desorption characteristics of the MOF selected in step S33, and finally select only the MOFs whose measured values of the adsorption / desorption characteristics satisfy the required physical properties from among the MOFs selected in step S33.

[0053] In the present embodiment, a method for manufacturing a porous material to which the above-described prediction and selection methods are applied can be provided. The method for manufacturing a porous material according to the embodiment is (1) a step of acquiring the initial structures of a plurality of porous materials; (2) For each of the porous materials, a step of obtaining the arrangement of the adsorbed substance when the adsorbed substance is adsorbed onto the porous material of the initial structure by simulation using the Monte Carlo method based on the neural network potential as the force field; (3) A step of obtaining a predicted value of the adsorption characteristics in the porous material having the structure corresponding to the obtained arrangement; (4) A step of selecting a porous material for which the predicted value of the adsorption characteristics obtained satisfies a predetermined condition; (5) A step of obtaining the selected porous material, and includes.

[0054] Among the above steps, steps (1) to (4) correspond to the prediction and selection steps described with reference to FIGS. 2 and 3 above. Step (5) corresponds to the step performed by the manufacturing apparatus.

[0055] In the step of obtaining the selected porous material, for example, when the porous material is MOF, MOF can be obtained by mixing a metal ion source and an organic ligand or a salt thereof using a hydrothermal synthesis method or a solvothermal synthesis method. For example, MOF can be obtained by the production method described in JP-T-2013-512223.

[0056] FIG. 4 is a diagram showing an example of predicted values and experimental values of the adsorption isotherm. The vertical axis of FIG. 4 is the adsorption amount (g / g) of water molecules, and the horizontal axis is the partial pressure P / P0. FIG. 4A shows the adsorption isotherms obtained by the prediction method using the GCMC method of the present embodiment for MOF1, MOF2, MOF3, MOF4, and MOF5, and FIG. 4B shows the experimental values of the adsorption isotherms of the above five types of MOF. All the prediction results by the prediction method of the present embodiment reproduce well the qualitative tendency of the experimental values of the adsorption isotherm, and it was confirmed that the prediction accuracy of the method of the present disclosure is high.

[0057] According to the present embodiment, the behavior of the adsorbed substance and the porous material can be accurately simulated using two types of simulation methods, and the prediction accuracy of the adsorption characteristics can be improved. By combining the simulation of the static properties of substances by the Monte Carlo method and the simulation of the dynamic properties of substances by the molecular dynamics method for structural optimization, it becomes possible to perform a simulation considering the adsorption and desorption of the adsorbed substance and the movement of the porous material. Since the structure of not only the adsorbed substance but also the porous material itself may change according to the adsorption of the adsorbed substance, the effect of improving the prediction accuracy by the combination of the two methods is exhibited particularly in the prediction of substances having flexibility such as MOF.

[0058] In the simulation, by generating a grand canonical ensemble, it is possible to suitably calculate an experimental system in which the adsorbed substance is adsorbed inside the porous material. By using NNP for the force field, it can be widely applied to various substances, and a simulation suitable for the purpose of searching for unknown materials becomes possible. In particular, by using NNP for predicting the behavior of MOF containing various elements, the effect of improving the prediction accuracy is exhibited.

[0059] By performing the simulation of the next cycle using the position and momentum of the adsorbed substance after the end of a series of cycles as initial information, the continuity of the simulation repeating a plurality of cycles can be ensured, and the reliability of the prediction can be improved.

[0060] (Second Embodiment) In the second embodiment, the prediction is made taking into account the volume change accompanying the deformation of the porous material. In the second embodiment, mainly the differences from the first embodiment will be described, and the same reference numerals will be given to the configurations common to the first embodiment, and the detailed description thereof will be omitted.

[0061] Some porous materials expand or contract when the adsorbed substance is adsorbed, causing the framework of the porous material to expand or contract. Therefore, it is preferable to consider the volume change associated with the expansion or contraction of the porous material in the simulation. In one embodiment of the present disclosure, in order to consider the volume change of the porous material, a simulation may be performed by the Gibbs ensemble Monte Carlo method (hereinafter also referred to as the GEMC method), which is a Monte Carlo method of the Gibbs ensemble.

[0062] The Gibbs ensemble is an ensemble that extends the grand canonical ensemble. In the Gibbs ensemble, the system is composed of a plurality of sub-systems, and volume and particle exchange are possible between each sub-system. Similar to the grand canonical ensemble, the temperature of the system is controlled to be constant throughout the simulation.

[0063] In this embodiment, a first cell representing a MOF system as a model and a second cell representing a system of an adsorbed substance (e.g., water molecules) are generated, and in the initial state, the adsorbed substance is placed only in the second cell. In the GEMC method, various thermodynamic equilibrium states are reproduced by repeating the probabilistic movement (translation, rotation) of the adsorbed substance, particle exchange between cells, and expansion and contraction of cells while keeping the total volume and total number of particles of all cells constant. The particle exchange operation between cells corresponds to the insertion and deletion of water. In the particle exchange operation, finally, water molecules are arranged in the first cell so that the chemical potentials of water in both cells match.

[0064] The expansion and contraction operation of the cell corresponds to the volume exchange of the cell. In the expansion and contraction operation, one cell is expanded and the other cell is contracted so that the total volume of the first cell and the second cell remains constant. The prediction device 1 generates random numbers independently for each axial direction (XYZ directions) of the first cell shown in three dimensions, for example, and expands or contracts the length of each side of the first cell according to the generated random numbers. The prediction device 1 adjusts the length of each side of the second cell so that the sum of the volumes of the first cell and the second cell becomes a constant value based on the volume of the first cell after expansion or contraction. The volume of the MOF can be changed by the expansion and contraction operation.

[0065] Also in the GEMC method, calculation conditions including a trial probability are set in the same manner as in the GCMC method described in the first embodiment. The trial probability in the second embodiment includes, for example, probabilities of a movement operation, a particle exchange operation, and an expansion / contraction operation. By appropriately setting the trial probability of the expansion / contraction operation, a desired volume change can be implemented. The trial probability of the movement operation may be zero as in the first embodiment.

[0066] The prediction device 1 executes the prediction process described in the flowchart of FIG. 2 in the same manner as in the first embodiment, except that the GEMC method is used as the Monte Carlo method. The prediction device 1 can cause a volume change of the MOF in the simulation process using the Monte Carlo method in step S14.

[0067] In the above, a configuration for executing simulation processing including a volume change in the simulation step by the GEMC method has been described, but the processing of the volume change may be executed separately. For example, the prediction device 1 may change the volume of the MOF by setting the trial probability of operations other than the expansion / contraction operation to zero and executing a simulation by the GEMC method. Thereafter, the prediction device 1 may execute a simulation by the GCMC method in the same manner as in the first embodiment.

[0068] According to one embodiment by the GEMC method, since the volume change accompanying the expansion or contraction of the porous material is reflected in the simulation, further improvement in prediction accuracy can be achieved.

[0069] Regarding the above embodiments, the following additional remarks are further disclosed. (Additional Remark 1) Obtain the initial structure of the porous material, Based on the neural network potential as a force field, obtain the arrangement of the adsorbed substance when the adsorbed substance is adsorbed on the porous material of the initial structure by simulation using the Monte Carlo method, Obtain a predicted value of the adsorption characteristics of the porous material having a structure according to the obtained arrangement A prediction method in which a computer executes the process. (Additional Remark 2) After obtaining the arrangement of the adsorbed substance by simulation using the Monte Carlo method, the arrangement of the adsorbed substance is further obtained by simulation using the molecular dynamics method. The prediction method according to Supplementary Note 1. (Supplementary Note 3) Repeatedly execute the simulation using the Monte Carlo method and the simulation using the molecular dynamics method. The prediction method according to Supplementary Note 2. (Supplementary Note 4) Based on the position or momentum of the adsorbed substance after the simulation using the molecular dynamics method, execute the simulation using the Monte Carlo method again. The prediction method according to Supplementary Note 2 or Supplementary Note 3. (Supplementary Note 5) The Monte Carlo method is the Gibbs ensemble Monte Carlo method or the grand canonical Monte Carlo method. The prediction method according to any one of Supplementary Notes 1 to 4. (Supplementary Note 6) In the simulation using the Monte Carlo method, change the volume of the porous material of the initial structure. The prediction method according to any one of Supplementary Notes 1 to 5. (Supplementary Note 7) In the simulation using the Monte Carlo method, change the volume of the porous material of the initial structure by changing the volume of each simulation cell so that the sum of the volume of the simulation cell of the porous material and the volume of the simulation cell of the adsorbed substance becomes constant. The prediction method according to any one of Supplementary Notes 1 to 6. (Supplementary Note 8) Set the probability that the translation and rotation trials of the adsorbed substance in the Monte Carlo method are adopted to zero. The prediction method according to any one of Supplementary Notes 1 to 7. (Supplementary Note 9) The adsorption characteristics include the maximum adsorption amount or the adsorption isotherm. The prediction method according to any one of Supplementary Notes 1 to 8.

[0070] Embodiments of the present disclosure are illustrative in all respects and not restrictive. The technical features described in each embodiment can be combined with each other. The scope of the present invention includes all modifications within the scope of the claims and the scope equivalent to the claims. The sequences shown in each embodiment are not limited. Within a non - conflicting range, the order of each processing procedure may be changed and executed, or a plurality of processes may be executed in parallel. The subject of each process is not limited. Within a non - conflicting range, the processes of each device may be executed by other devices.

Description of Reference Numerals

[0071] 100 Manufacturing system 1 Prediction device 11 Control unit 12 Storage unit 13 Communication unit 14 Input unit 15 Output unit 1P Program 1A Recording medium 2 Manufacturing device

Claims

1. Obtain the initial structure of the porous material, Based on the neural network potential as the force field, determine the arrangement of the adsorbate when the adsorbate is adsorbed onto the porous material with the initial structure by simulation using the Monte Carlo method, Obtain the predicted value of the adsorption characteristics of the porous material with the structure corresponding to the determined arrangement A prediction method in which a computer executes the process.

2. After determining the arrangement of the adsorbate by simulation using the Monte Carlo method, further determine the arrangement of the adsorbate by simulation using the molecular dynamics method The prediction method according to Claim 1.

3. Repeatedly execute the simulation using the Monte Carlo method and the simulation using the molecular dynamics method The prediction method according to Claim 2.

4. Based on the position or momentum of the adsorbate after the simulation using the molecular dynamics method, execute the simulation using the Monte Carlo method again The prediction method according to Claim 2.

5. The Monte Carlo method is the Gibbs ensemble Monte Carlo method or the grand canonical Monte Carlo method The prediction method according to Claim 1 or Claim 2.

6. In the simulation using the Monte Carlo method, change the volume of the porous material with the initial structure The prediction method according to Claim 1 or Claim 2.

7. In the simulation using the Monte Carlo method, change the volume of the porous material with the initial structure by changing the volume of each simulation cell so that the sum of the volume of the simulation cell of the porous material and the volume of the simulation cell of the adsorbate is constant The prediction method according to Claim 1 or Claim 2.

8. Set the probability that the translation and rotation trials of the adsorbate in the Monte Carlo method are adopted to zero The prediction method according to Claim 1 or Claim 2.

9. The adsorption characteristics include the maximum adsorption amount or the adsorption isotherm The prediction method according to Claim 1 or Claim 2.

10. Obtain the initial structure of the porous material, Based on the neural network potential as the force field, determine the arrangement of the adsorbate when the adsorbate is adsorbed onto the porous material with the initial structure by simulation using the Monte Carlo method, Obtain the predicted value of the adsorption characteristics of the porous material with the structure corresponding to the determined arrangement A computer program that causes a computer to execute the process.

11. Obtain the initial structure of the porous material, Based on the neural network potential as a force field, the arrangement of the adsorbed substance when the adsorbed substance is adsorbed onto the porous material of the initial structure is obtained by simulation using the Monte Carlo method, The predicted value of the adsorption characteristics of the porous material having a structure corresponding to the obtained arrangement is acquired, A control unit that executes the process is provided, Prediction device.

12. A step of obtaining the initial structures of a plurality of porous materials, For each porous material, a step of obtaining the arrangement of the adsorbed substance when the adsorbed substance is adsorbed onto the porous material of the initial structure by simulation using the Monte Carlo method based on the neural network potential as a force field, A step of obtaining the predicted value of the adsorption characteristics of the porous material having a structure corresponding to the obtained arrangement, A step of selecting a porous material for which the predicted value of the obtained adsorption characteristics satisfies a predetermined condition, Including the step of obtaining the selected porous material, Method for manufacturing a porous material.

Images