Search device, search method, program, and non-temporary computer-readable medium

Abstract

[Problem] To efficiently search for a promoter. [Solution] According to the present invention, a search device comprises a promoter placement optimization unit and a promoter element search unit. For a specific elementary reaction of a reaction that includes a plurality of elementary reactions and uses a catalyst, the promoter placement optimization unit optimizes the placement of a promoter element on the catalyst on the basis of an activation energy acquired using a trained model. The promoter element search unit searches for the promoter element on the basis of the activation energy acquired using the trained model for every type of promoter element.

Description

[Technical Field]

[0001] This disclosure relates to a search device, a search method, a program, and a non-temporary computer-readable medium. [Background technology]

[0002] In chemical reactions, catalysts are crucial substances for altering the reaction rate. Adding promoters to these catalysts can sometimes improve their performance. In computational science, Nudged Elastic Band (NEB) calculations using Density Function Theory (DFT) are sometimes performed to elucidate the mechanisms of catalytic reactions. This method allows for the calculation of the activation energy of specific elementary reactions. Therefore, NEB calculations can be used to calculate the activation energy when promoters are added, allowing for the estimation of the effect of promoter addition.

[0003] The activation energy and absolute reaction kinetics obtained as described above can be used to calculate the reaction rate constants of elementary reactions. Furthermore, it is possible to calculate the yield of the product obtained from a complex catalytic reaction from the obtained reaction rate constants. By using this, it is possible to investigate which elementary reactions affect the desired yield by changing the rate constants of each reaction.

[0004] In this method, in order to search for catalysts whose performance is improved by adding promoters, it is necessary to search for promoters that lower the activation energy of elementary reactions that increase the desired product yield by accelerating the reaction. In conventional techniques, when calculating the activation energy with the addition of promoters, methods such as arranging promoter elements on the catalyst surface or substituting catalyst elements are used. Therefore, since the physical properties of the catalyst change greatly depending on the placement of the promoters, the above calculations are performed by searching for a configuration with a lower activation energy while gradually changing the position and number of promoters.

[0005] However, DFT calculations are extremely costly, and as mentioned above, it is necessary to change the promoter arrangement among a large number of atoms, making the search for promoter arrangements using this method an extremely difficult task. Even if an element that lowers the activation energy is discovered by performing calculations with added promoters, if that reaction is not an elementary reaction that affects the desired product yield, no improvement in catalytic performance can be expected. [Prior art documents] [Non-patent literature]

[0006] [Non-Patent Document 1] K. Shimura, et. al., “Fischer-Tropsch synthesis over alumina supported cobalt catalyst: effect of promoter addition,” Applied Catalysis A: General 494, 1-11, 2015 [Non-Patent Document 2] B. Chen, et. al., “Charge-Tuned CO Activation over a χ-Fe5C2 Fischer-Tropsch Catalyst,” ACS Catalysis, 8, 4, 2709-2714, 2018 [Non-Patent Document 3] B. Zijlstara, et. al., “The Vital Role of Step-Edge Sites for Both CO Activation and Chain Growth on Cobalt Fischer-Tropsch Catalysts Revealed through First-Principles-Based Microkinetic Modeling Including Lateral Interactions,” ACS Catalysis, 10, 16, 9376-9400, 2020 [Non-Patent Document 4] W. Song, et. al., “Combination of Density Functional Theory and Microkinetic Study to the Mn-Doped CeO2 Catalysts for CO Oxidation: A Case Study to Understand the Doping Metal Content,” J. Phys. Chem. C, 122, 44, 25290-25300, 2018 Summary of the Invention Problems to be Solved by the Invention

[0007] According to the present disclosure, there is provided a search device capable of efficiently searching for a promoter that improves target characteristics. Means for Solving the Problems

[0008] According to one embodiment, the search device includes a promoter arrangement optimization unit and a promoter element search unit. The promoter arrangement optimization unit optimizes the arrangement of promoter elements in the catalyst based on the activation energy obtained using a trained model for a specific elementary reaction in a reaction using a catalyst including a plurality of elementary reactions. The promoter element search unit searches for the promoter elements based on the activation energy obtained using the trained model for each type of the promoter elements.

[0009] This trained model may be a model used for a NNP (Neural Network Potential) that outputs energy when the atomic structure of a substance is input. Brief Description of the Drawings

[0010] [Figure 1] A block diagram showing an example of a search device according to one embodiment. [Figure 2] A flowchart showing the processing of a search device according to one embodiment. [Figure 3]A flowchart illustrating the processing of a search device according to one embodiment. [Figure 4] A figure showing an example of the outermost surface and adsorbed molecules according to one embodiment. [Figure 5] A block diagram showing an example of an implementation of a search device according to one embodiment. [Modes for carrying out the invention]

[0011] Embodiments of the present invention will be described below with reference to the drawings. The drawings and descriptions of the embodiments are provided as examples only and do not limit the present invention.

[0012] (First Embodiment) Figure 1 is a schematic diagram of a search device according to one embodiment. The search device 1 comprises an input unit 100, a storage unit 102, an output unit 104, an elementary reaction identification unit 106, a promoter arrangement optimization unit 110, and a promoter element search unit 108. The search device 1 is a device that, upon input of information on adsorbed molecules and a catalyst, infers a promoter element and its arrangement that will improve performance by substituting some atoms of the catalyst with a promoter element and / or adding a promoter element to the catalyst. In addition to the configuration shown in Figure 1, other configurations necessary for the operation of the search device 1 are not shown but can be appropriately provided as needed.

[0013] The input unit 100 is equipped with an input interface for the search device 1 and accepts data input to the search device 1. The search device 1, for example, receives information about adsorbed molecules and catalysts from the user via this input unit 100.

[0014] The memory unit 102 stores data necessary for the operation of the search device 1. For example, the memory unit 102 stores information input from the input unit 100, a program for operating the search device 1, and search results, as well as information such as intermediate values ​​needed during calculations and parameters related to the neural network model used for inference.

[0015] The output unit 104 outputs the search results to an external device or to the storage unit 102. In other words, in this disclosure, output is a concept that includes output to external storage, monitors, etc., as well as output of results to the storage unit 102 of the search device 1.

[0016] The elementary reaction identification unit 106 identifies elementary reactions in a catalytic reaction that affect a desired characteristic (hereinafter referred to as the target characteristic). The elementary reaction identification unit 106 may, for example, identify elementary reactions that affect the target characteristic by reaction rate simulation, or it may identify elementary reactions that affect the target characteristic based on a database that has already been calculated.

[0017] The elementary reaction identification unit 106 may perform reaction rate simulations for multiple elementary reactions by changing the reaction rate constant for each elementary reaction. In this case, the elementary reaction that affects the target characteristics may be identified by how the target characteristics are affected and changed through the catalytic reaction, including the elementary reactions with changed reaction rate constants.

[0018] For example, if the catalytic reaction time is the target characteristic, the reaction rate constant may be changed for each elementary reaction, and an elementary reaction having a reaction rate constant that changes the catalytic reaction time may be identified. For example, if the yield is the target characteristic, the reaction rate constant may be changed while maintaining the equilibrium constant for each elementary reaction, and an elementary reaction that changes the yield may be identified.

[0019] This identification may, for example, involve simulating catalytic reactions by varying the above-mentioned reaction rate constants for multiple elementary reactions, and then identifying the elementary reaction that had the greatest impact based on the simulation results. More specifically, various reaction rate constants may be varied for each elementary reaction, the rate of change of the target characteristics may be calculated, and the highest rate of change may be used as the influence value of that elementary reaction. The elementary reaction identification unit 106 may then compare the influence values ​​for each elementary reaction and identify one or more elementary reactions from the highest influence values.

[0020] The elementary reaction identification unit 106 may perform this reaction rate simulation using a pre-trained model, as described later. That is, the elementary reaction identification unit 106 may calculate the reaction rate parameters for each elementary reaction using a pre-trained model that infers the potential in the promoter placement optimization unit 110.

[0021] The promoter element search unit 108 searches for promoter elements with lower activation energies compared to other promoter elements, based on the activation energies obtained using a trained model for each promoter element, if it wants to promote a specific elementary reaction. For example, the promoter element search unit 108 instructs the promoter arrangement optimization unit 110 to optimize the arrangement of adsorbed molecules and catalysts related to the elementary reaction identified by the elementary reaction identification unit 106, specifying the promoter elements. The promoter element search unit 108 obtains information on the activation energies for each promoter element from the promoter arrangement optimization unit 110 and searches for which elements are suitable for the specific elementary reaction.

[0022] The promoter placement optimization unit 110 optimizes the placement of promoter elements in the catalyst using a trained model for the elementary reactions identified by the elementary reaction identification unit 106 and the promoters selected by the promoter element search unit. This trained model is, for example, a model provided in NNP (Neural Network Potential), in which case the promoter placement optimization unit 110 infers the potential energy surface near the transition state of each elementary reaction using NNP.

[0023] The promoter placement optimization unit 110 receives information about the catalyst in which the promoter element is positioned in the trained model, and the adsorbed molecules adsorbed on the catalyst in the target properties, and infers the activation energy. As an example, the promoter placement optimization unit 110 repeatedly performs inference with various configurations of the promoter element in the initial atomic structure (initial structure in the calculation) in which the position and orientation of the adsorbed molecules on the catalyst are fixed, and optimizes the configuration of the promoter element by obtaining information on the configuration with a lower activation energy than other configurations.

[0024] The promoter placement optimization unit 110 outputs the activation energy in the optimized target element arrangement to the promoter element search unit 108.

[0025] The trained model is a model trained to infer material property information using various elements. In other words, the trained model takes various combinations of elements as input as atomic structures, and uses appropriate material property information such as energy for these atomic structures as training data. The parameters of the model are optimized based on the error between the model's output value and the training data. Furthermore, the trained model may also take the form of inputting boundary conditions as atomic structures. In this case, it may take the form of specifying the unit atomic structure, the pitch of the unit's repetition, and whether it is periodic or free space. That is, the trained model may be trained as a model that can take the same input as a general NNP. The input to the trained model is the element and coordinates (positions) of each atom constituting the material, and the output may be potential (energy), or information necessary to calculate energy, such as wave functions.

[0026] The trained model may be one that has been trained using, for example, results on the potential obtained from DFT calculations or other first-principles calculations as training data.

[0027] In the above, the trained model is assumed to be able to obtain the activation energy by NNP, but it is not limited to this, and for example, it may be a model that obtains physical properties that are correlated with / highly correlated with this activation energy. Examples of such physical properties include intermolecular distance, atomic charge, adsorption energy, vibration frequency, d-band centroid, or the energy of the reaction intermediate. The trained model may also be a model that infers at least one of these pieces of information about the atomic structure. In this case, the promoter arrangement optimization unit 110 obtains the arrangement of promoter elements that exhibit lower physical properties (or higher physical properties if there is a negative correlation) than other promoter elements by optimizing these physical properties.

[0028] The promoter element search unit 108 uses the physical property values ​​output by the promoter arrangement optimization unit 110 to determine elements suitable as promoter elements and outputs them via the output unit 104. When the promoter arrangement optimization unit 110 outputs activation energy values, the promoter element search unit 108 selects and outputs promoter elements with activation energies lower or higher than other promoter elements. When physical property values ​​correlated with activation energy are used as physical property values, the promoter element search unit 108 selects and outputs promoter elements based on those physical property values.

[0029] The promoter element search unit 108 may specify one type of element as the promoter element and have the promoter arrangement optimization unit 110 perform optimization. In this case, the promoter element search unit 108 designates one type of element as the promoter element and has the promoter arrangement optimization unit 110 perform optimization. The promoter arrangement optimization unit 110 performs optimization of the arrangement of the promoter element specified by the promoter element search unit 108. The promoter element search unit 108 changes the promoter element to various elements and has the promoter arrangement optimization unit 110 repeatedly perform optimization to search for promoter elements.

[0030] The promoter element search unit 108 may specify multiple types of elements as promoter elements and have the promoter arrangement optimization unit 110 perform optimization. In this case, the promoter element search unit 108 specifies multiple types of elements as promoter elements and has the promoter arrangement optimization unit 110 perform optimization. The promoter arrangement optimization unit 110 performs optimization of the arrangement of the promoter elements specified by the promoter element search unit 108. The promoter element search unit 108 changes the promoter elements to various combinations of multiple elements and has the promoter arrangement optimization unit 110 repeatedly perform optimization to search for promoter elements.

[0031] Furthermore, if multiple types of promoter elements are specified, the promoter arrangement optimization unit 110 may optimize the arrangement of the multiple types of promoter elements, as well as the ratio in which the multiple types of promoter elements are arranged.

[0032] Common to all of the above, the promoter arrangement optimization unit 110 repeatedly searches for reaction pathways using a trained model, specifying the arrangement of one or more types of promoter elements designated by the promoter element search unit 108. From these searches, it obtains the activation energy or physical properties correlated with the activation energy, and optimizes the arrangement of the promoter elements based on these values.

[0033] The promoter placement optimization unit 110 places one or more promoter elements specified by the promoter element search unit 108 onto the catalyst and performs the above optimization. When optimizing the placement of one promoter element, the promoter placement optimization unit 110 may perform the optimization by grid search, as an example, though not limited to this. When optimizing the placement of multiple promoter elements, the promoter placement optimization unit 110 may perform the optimization by Bayesian optimization or random search, as an example, though not limited to this.

[0034] Furthermore, the promoter placement optimization unit 110 may set the initial structure in the above-described calculation to have a distance of 5 Å or less between the catalyst and the adsorbed molecule. In addition, the promoter placement optimization unit 110 may also set the initial structure to have a distance of 5 Å or less between the adsorbed molecule and the promoter element to be placed.

[0035] As described above, the promoter placement optimization unit 110 may arrange promoter elements by substituting one or more atoms in the atomic structure constituting the catalyst with promoter elements, or by adding one or more promoter elements to the atomic structure constituting the catalyst. The number of promoter elements may not contribute to improving the reaction rate if it is too few or too many. For this reason, the promoter placement optimization unit 110 may, for example, arrange less than 10% of the atoms in the catalyst atomic structure other than the adsorbed molecules input into the trained model as promoter elements. This 10% number of atoms is given as an example and is not limited. If too many promoter elements are added to the atomic structure constituting the catalyst in the calculation, the surface structure may become unstable depending on the type of element, and the surface structure may change significantly compared to when there are no promoters. The number is set to less than 10% to avoid such changes in the surface structure.

[0036] The elementary reaction identification unit 106 can also identify multiple elementary reactions as elementary reactions that contribute significantly to the target characteristics. After the promoter element search unit 108 has searched for promoter elements for a certain elementary reaction as described above, the same process may be performed for different elementary reactions. The promoter element search unit 108 may also perform the search for promoter elements in the order of the elementary reactions that the elementary reaction identification unit 106 has determined to contribute significantly.

[0037] In such cases, the promoter element search unit 108 may, for example, store the search results in the memory unit 102 and use those search results to search for promoter elements in the next elementary reaction. The promoter element search unit 108 may then comprehensively determine the reaction energy, reaction rate, yield, etc., for the multiple elementary reactions stored and output the final promoter element and its arrangement.

[0038] Figure 2 is a flowchart showing the processing of the search device 1 according to one embodiment.

[0039] The search device 1 acquires information about the substance and catalyst to be searched for via the input unit 100 (S100). The acquired information may be stored in the memory unit 102. The information about the substance may be input in the form of a reaction equation.

[0040] The elementary reaction identification unit 106 identifies elementary reactions that affect the target properties in a given combination of substance and catalyst (S102). When information about the substances is input in the reaction equation, it identifies which intermediate reactions in the reaction pathway affect the target properties. The target properties may be, but are not limited to, at least one of the following: reaction rate, reaction time, or yield. For example, the elementary reaction identification unit 106 may identify elementary reactions that affect the yield of the product. Alternatively, as another example, the elementary reaction identification unit 106 may identify elementary reactions that affect the durability of the catalyst.

[0041] For example, when the reaction rate of a solid catalyst is used as the target characteristic, the elementary reactions are identified from the following equation.

number

number

number

number

[0042] Here, k represents the rate constant. Also, k , is the Boltzmann constant, T is the absolute temperature, m is the mass of the gas molecule, P 0 is the reference pressure of 1 atm, A is the catalyst surface area, K eq is the equilibrium constant, R is the gas constant, S gas is the gas entropy, ΔH ads corr is the adsorption energy, H gas 298→T is the change of the gas enthalpy from 298K, E lat is the correction term representing the interaction between adsorbed molecules, q vib,ads is the term introduced because not all entropy is lost when the molecule is adsorbed, Q ‡ is the partition function of the transition state, h is the Planck constant, respectively. The subscripts respectively represent ads: adsorption, des: desorption, surf: surface. The elementary reaction specifying unit 106 calculates the rate constants of adsorption, desorption, and on the surface based on equations (1) to (3). The activation energy ΔE act zpe is calculated based on DFT, and the entropy S and enthalpy H are obtained using a database. q vib, ads is calculated as 1. The partition function Q is obtained by calculating the eigenvalues ω i 2 of the Hessian matrix according to equation (4), and the angular frequency ω i is obtained by calculating in the vibration calculation of DFT. The calculation of DFT may be performed by NNP using a trained model.

[0043] Then, according to the following equation, the rate constant k is converted to the reaction rate r, and the time derivative of the concentration is obtained by summing the reaction rate r for each element.

Number

Number

number

number

[0044] Here, r ads surf This is the reaction rate of the adsorption reaction, r des surf is the reaction rate of the elimination reaction, r surf is the reaction rate of the surface reaction, r j This is the reaction rate of reaction j, N surf / N total θ is the proportion of specific surface sites. * surf θ is the coverage rate of the empty sites. ads surf θ is the coverage rate of the adsorbed molecules. i This is the coverage of chemical species i, ν i, surf ν is the stoichiometric coefficient of chemical species i. i,j Herein are the stoichiometric coefficients of chemical species i and reaction j, and a gas This is calculated as gas partial pressure P / reference pressure P 0 The ratio, n is a coefficient k, which is 2 for dissociative adsorption and 1 otherwise. ads surf k is the rate constant of adsorption. des surf These represent the rate constants of desorption, respectively. Furthermore, the yield of each substance can be obtained by solving the system of ordinary differential equations.

[0045] For identifying elementary reactions, methods such as DRC (Degree of Rate Control), DSC (Degree of Selectivity Control), and DCGC (Degree of Chain-Growth Control) are used. Each is given by the following equations. The following equations describe an example where CO is dissociated to produce CH4, but this reaction can be modified depending on the desired reaction.

number

number

number

[0046] Here, i is the reaction i, r CO CO molecular reaction rate, k i S is the rate constant of reaction i. CH4 is the selection rate of CH4, α is the chain growth probability, K i This represents the reaction equilibrium constant i.

[0047] In actual catalytic reactions, the question of which elementary reaction is rate-determining is debated, but it is difficult to answer this precisely. In other words, there may be multiple slow elementary reactions. By performing reaction rate simulations using the activation energies of elementary reactions obtained from DFT calculations or databases, and varying the reaction rate constants of each elementary reaction, it is possible to evaluate how much each elementary reaction affects the product yield and / or chain growth probability. Using this method, it is possible to identify multiple elementary reactions that affect the yield and / or chain growth probability.

[0048] The elementary reaction identification unit 106 identifies one or more elementary reactions that contribute to the target characteristics based on the above relationship.

[0049] If the elementary reaction identification unit 106 identifies multiple elementary reactions, the promoter element search unit 108 selects one of the identified elementary reactions (S104) and searches for the promoter element (S106). Note that if only one elementary reaction is identified, the process in S104 is not required.

[0050] The promoter element search unit 108 determines whether the search for other elementary reactions has been completed after it has finished searching for the type of promoter element and the arrangement of the promoter element for one elementary reaction (S108). This determination, as with S104, can be omitted if only one elementary reaction has been identified.

[0051] If the search for each elementary reaction is not yet complete (S108: NO), an elementary reaction other than those for which the search has been completed is selected (S104), and the search process continues. If the search for the identified elementary reaction is complete (S108: YES), the promoter element search unit 108 outputs the promoter element and its arrangement (S110), and the process ends. If multiple elementary reactions have been identified, the promoter element search unit 108 appropriately updates the promoter element and its arrangement based on the search results for the promoter elements of the multiple elementary reactions, and outputs the updated results.

[0052] Figure 3 is a flowchart showing the process for searching for promoter elements in Figure 2. We will use Figure 3 to explain the process of S106 in Figure 2.

[0053] First, the promoter element search unit 108 selects a promoter element (S200). For example, if one type of promoter element is to be placed in the catalyst, the promoter element search unit 108 selects one element from the elements that can be applied as a promoter element and sends it to the promoter placement optimization unit 110 to send a request for placement optimization. If multiple types of promoter elements are to be placed, the unit selects multiple types from the elements that can be applied as promoter elements and sends a request for placement optimization.

[0054] The promoter placement optimization unit 110 optimizes the placement of promoter elements relative to the catalyst in order to optimize the placement of promoter elements based on the information on elementary reactions, catalysts, and promoter elements received from the promoter element search unit 108 (S202).

[0055] The promoter placement optimization unit 110 obtains physical properties for the placement of the promoter elements after the placement has been determined (S204). For example, these physical properties are the activation energies in the elementary reactions.

[0056] The promoter placement optimization unit 110 obtains the activation energy based on information about the catalyst and adsorbed molecules with the promoter elements arranged, for example, using the NEB method.

[0057] The promoter placement optimization unit 110, as an example without limitation, sets the initial and final states of the chemical equation of an elementary reaction and obtains the activation energy by optimization calculation using the NEB method. In the NEB method, an appropriate initial path from the initial state (IS) structure to the final state (FS) structure is selected, and optimization is performed so that the activation energy in this transition is low. The promoter placement optimization unit 110 can quickly obtain the energy values ​​at each transition state of the path required for optimization calculation by using NNP with a trained model.

[0058] The promoter placement optimization unit 110 can, as an example without limitation, use the TS structure obtained by the NEB method to optimize the transition state (TS) structure and obtain the transition state energy using a method employing IRC (Intrinsic Reaction Coordinate). This method involves preparing an initial structure close to the transition state, calculating the second derivative of the energy of this structure to obtain the normal vibration, and then searching for the transition state (TS) structure by confirming the imaginary vibration mode. In the IRC calculation, the normal structure optimization is performed from the TS structure in the reaction coordinate direction, and it is confirmed whether it converges to the structure of the target reactant (IS) and product (FS). In this method as well, by using NNP with a trained model in the energy calculation process, it is possible to perform the search at high speed.

[0059] For example, the promoter arrangement optimization unit 110 performs vibrational analysis on the TS structure obtained by NEB calculation, targeting the outermost layer of the catalyst and promoter elements and the adsorbed molecules. Then, using the vibrational states adjacent to the imaginary vibration of the TS structure, it performs optimization using the BFGS (Broyden-Fletcher-Goldfarb-Shanno) method.

[0060] The explanation uses an example of a CO dissociation reaction, where Co is the catalyst and CO is the adsorbed molecule. The promoter arrangement optimization unit 110 extracts one Co atom from the vicinity of CO and replaces this Co atom with the promoter element. Then, the TS structure is optimized and the activation energy is obtained using IRC.

[0061] The promoter placement optimization unit 110 determines whether the optimization is complete after obtaining the activation energy (S206). For example, the promoter placement optimization unit 110 may determine whether the optimization is complete based on whether the extraction from Co atoms surrounding the CO is complete.

[0062] Figure 4 shows an example of the initial structure of the outermost surface of the catalyst Co and the adsorbed molecule Co. In this figure, the promoter arrangement optimization unit 110 selects, for example, the Co atoms shown by the solid lines one by one and replaces them with promoter elements (S202), and calculates the activation energy using the method described above (S204). The process of replacing Co atoms with promoter elements is repeated for the Co atoms shown by the solid lines (S206: NO to S202, S204). Once all Co atoms have been extracted, the promoter arrangement optimization unit 110 outputs the physical properties and arrangement, indicating that the arrangement optimization is complete (S206: YES) (S208). For example, in the case of the atomic structure of the surface as shown in Figure 4, atoms to be replaced may be selected from the six atoms of the catalyst. This arrangement is shown as an example and may differ depending on the atomic arrangement of the catalyst.

[0063] When adding promoter elements to the surface of Co, specify the position to be added, obtain the activation energy, and perform the above calculation (S202~S208). Similarly, when arranging multiple types of promoter elements, select a combination of Co atoms and replace them with a combination of promoter elements, or add a combination of promoter elements to the outermost surface of the Co atoms, and perform the above calculation (S202~S208).

[0064] These processes can be appropriately parallelized. For example, using a GPU (Graphics Processing Unit), it is possible to perform NNP calculations on multiple atomic structures in parallel. Therefore, for example, after specifying multiple arrangements in S202, it is possible to obtain physical property values ​​for these arrangements using parallel calculations. Furthermore, for multiple promoter elements, it is possible to optimize the arrangement of multiple promoter elements in parallel by arranging multiple promoter elements in S202 and processing them in parallel. The accelerator used is not limited to a GPU; other suitable hardware architectures may also be used.

[0065] In the above method, the activation energy is E in the forward direction. act, forward = E TS - E IS In the opposite direction, E act, backward = E TS - E FS It can be done as follows: E TS E is the energy in the transition state. IS , E FS These represent the energy in the initial and final states, respectively. By using NNP to calculate these energies, the calculations can be performed at high speed.

[0066] Furthermore, if the distance between the adsorbed molecule and the promoter element is too great, it becomes difficult to obtain the promoter effect. For this reason, in order to properly execute the optimization calculation in real time, the initial structure may have a distance of 5 Å or less between the adsorbed molecule and the promoter element. Preferably, the distance may be 4 Å or less.

[0067] When specifying a distance, the promoter arrangement optimization unit 110 may perform optimization for catalyst atoms within a predetermined distance from the adsorbed molecule, regardless of the atoms represented by solid lines in Figure 4. In this case, the number of combinations may increase. Therefore, the promoter arrangement optimization unit 110 may perform the arrangement of the substituted or added promoter elements using various optimization methods.

[0068] In the above, we assumed the use of NEB, TS structure optimization, and IRC, but these methods can be arbitrarily chosen. For example, only the NEB method may be used, or only TS structure optimization and IRC may be used, or other appropriate methods may be used.

[0069] When arranging a single promoter element, the promoter arrangement optimization unit 110 may optimize it, for example, by grid search. When arranging multiple promoter elements, the promoter arrangement optimization unit 110 may use, for example, Bayesian optimization, random search, or a genetic algorithm. These are just some examples, and are not limited to these.

[0070] Once the optimization of the arrangement of the promoter elements (or combinations of promoter elements, hereinafter collectively referred to as promoter elements, etc.) is complete, the promoter element search unit 108 determines whether the search for promoter elements is complete (S210). This determination determines, for example, whether the optimization of the arrangement of candidate elements for promoter elements is complete, or, in the case of multiple types of promoter elements, whether the optimization of the arrangement of appropriate combinations is complete.

[0071] If the optimization of the arrangement of promoter elements, etc., has not been completed (S210: NO), the promoter elements, etc., whose arrangement has not yet been optimized are selected (S200), and the process from S202 is repeated. The promoter arrangement optimization unit 110 performs arrangement optimization by changing the promoter elements, etc., without changing the initial arrangement (initial atomic structure) of the catalyst and adsorbed molecules. By not changing the initial arrangement, it is possible to obtain physical properties such as activation energy under the same conditions.

[0072] When the optimization of the arrangement of promoter elements, etc., is completed (S210: YES), the promoter element search unit 108 outputs the necessary information, such as the best physical property value, for example, the promoter element from which the activation energy was obtained, the said physical property value, and the arrangement of the promoter elements, and then terminates processing (S212).

[0073] As described above, according to this embodiment, it is possible to appropriately search for promoter elements in catalytic reactions using NNP. By appropriately using NNP, for example, it is possible to rapidly search for target characteristics in catalytic reactions, such as promoter elements that improve the yield of the product and the arrangement of said promoter elements.

[0074] (Second Embodiment) In the first embodiment described above, the search device 1 was intended to search for promoter elements and their arrangements that improve target characteristics such as product yield using a trained model. The search device 1 may also be trained by active learning in order to further increase the search efficiency and acquire new promoter elements.

[0075] For example, the promoter element search unit 108 may perform active learning of a model for acquiring activation energy using the feature quantities obtained from the type and arrangement of promoter elements acquired during the search for promoter elements, and the activation energy data. This model for acquiring activation energy is, for example, a regression model.

[0076] The promoter placement optimization unit 110 acquires activation energy data for various promoter elements and their arrangements during the search for promoter elements. The promoter element search unit 108 uses this data on various promoter elements and their activation energies to perform active learning of a regression model.

[0077] The promoter arrangement optimization unit 110 trains a regression model, for example, using an appropriate machine learning method, so that when promoter elements for a catalyst are input, the activation energy is output. By using the model generated by this training, it becomes possible to obtain more quickly which promoter elements can be used to lower or raise the activation energy for a catalyst. The input to the regression model may be just the type of promoter element, or the type and arrangement of promoter elements, or it may be any other value obtained by the promoter arrangement optimization unit 110, in addition to the catalyst.

[0078] According to this embodiment, the search device 1 can perform a search and generate a regression model that obtains the activation energy for the promoter element. By using this regression model, it becomes possible to achieve a more efficient search for the promoter element.

[0079] Several implementation examples are given for each of the embodiments described above. The methods used in the search device 1 are not limited to these examples.

[0080] The search device 1 may identify elementary reactions in the elementary reaction identification unit 106 by obtaining the necessary DFT values ​​of the activation energy from literature values ​​and databases.

[0081] The search device 1 may identify elementary reactions by obtaining the necessary values ​​using NNP calculations in the elementary reaction identification unit 106. As described above, the elementary reaction identification unit 106 can also obtain parameters related to the reaction rate simulation for each elementary reaction using NNP calculations. For example, the elementary reaction identification unit 106 can infer various parameters, such as Q in equation (4), which are used in the reaction rate simulation, using a trained model used by the promoter placement optimization unit 110.

[0082] The search device 1 may, for example, search for promoter elements using elementary reactions identified by the user. In this case, the elementary reaction identification unit 106 does not need to be provided in the search device 1, and the search device 1 may accept input of elementary reactions in S100 without executing the process in S102 in Figure 2.

[0083] The search device 1 may use Bayesian optimization for elemental substitution or addition, or random search for elemental substitution or addition, in the promoter placement optimization unit 110. Alternatively, grid search or a genetic algorithm may be used. The placement ratio can be arbitrary, but may be defined, for example, within a range of 10% or less. The promoter placement optimization unit 110 may obtain physical properties from methods such as the NEB method, TS structure optimization + IRC, or any appropriate combination of these methods. The physical properties may be activation energy, or physical properties that have a positive or negative correlation with the activation energy.

[0084] In the elemental discovery unit 108 of the search device 1, if it is desired to promote the identified elementary reaction, an element with a low activation energy may be selected. Alternatively, in the elemental discovery unit 108, an element may be selected such that the physical property value positively correlated with the activation energy is low, or the physical property value negatively correlated with the activation energy is high.

[0085] In each of the embodiments described above, the search device 1 may be implemented by one or more computers. For example, input may be performed by a client on the user's side, and the necessary information may be transmitted from the client to the search device 1. In this case, the search device 1 may be provided as a server that is part of the search system.

[0086] All of the above-mentioned trained models may be concepts that include, for example, models that have been trained as described and then further distilled using general methods.

[0087] In the embodiments described above, some or all of the devices (search device 1) may be composed of hardware, or they may be composed of information processing by software (programs) executed by a CPU (Central Processing Unit) or GPU (Graphics Processing Unit), etc. If they are composed of information processing by software, the software that realizes at least some of the functions of each device in the embodiments described above may be stored on a non-temporary storage medium (non-temporary computer-readable medium) such as a flexible disk, CD-ROM (Compact Disc-Read Only Memory), or USB (Universal Serial Bus) memory, and the information processing of the software may be executed by having a computer read it. Alternatively, the software may be downloaded via a communication network. Furthermore, the information processing may be executed by hardware by implementing the software on a circuit such as an ASIC (Application Specific Integrated Circuit) or FPGA (Field Programmable Gate Array).

[0088] The type of storage medium used to store the software is not limited. The storage medium is not limited to removable media such as magnetic disks or optical disks; it may also be a fixed storage medium such as a hard disk or memory. Furthermore, the storage medium may be located inside or outside the computer.

[0089] Figure 5 is a block diagram showing an example of the hardware configuration of each device (search device 1) in the embodiment described above. Each device may be implemented as a computer 7, for example, comprising a processor 71, a main memory 72 (memory), an auxiliary memory 73 (memory), a network interface 74, and a device interface 75, which are connected via a bus 76.

[0090] The computer 7 in Figure 5 has one of each component, but it may have multiple identical components. Also, although Figure 5 shows one computer 7, the software may be installed on multiple computers, and each of these computers may execute the same or different parts of the software's processing. In this case, it may be a distributed computing configuration in which each computer communicates via a network interface 74 or the like to execute processing. In other words, each device (search device 1) in the above-described embodiment may be configured as a system that realizes its function by having one or more computers execute instructions stored in one or more storage devices. Alternatively, it may be configured so that information transmitted from a terminal is processed by one or more computers located on the cloud, and the processing results are transmitted to the terminal.

[0091] The various calculations performed by each device (search device 1) in the embodiments described above may be executed in parallel using one or more processors, or using multiple computers connected via a network. Alternatively, the various calculations may be distributed to multiple processing cores within a processor and executed in parallel. Furthermore, some or all of the processing and means of this disclosure may be executed by at least one of a processor and a storage device located on a cloud that can communicate with computer 7 via a network. Thus, each device in the embodiments described above may be in the form of parallel computing using one or more computers.

[0092] The processor 71 may be an electronic circuit (processing circuit, processing circuitry, CPU, GPU, FPGA, or ASIC, etc.) including a computer control unit and arithmetic unit. Alternatively, the processor 71 may be a semiconductor device including a dedicated processing circuit. The processor 71 is not limited to an electronic circuit using electronic logic elements, but may also be realized by an optical circuit using optical logic elements. Furthermore, the processor 71 may include computational functions based on quantum computing.

[0093] The processor 71 performs calculations based on data and software (programs) input from various devices within the computer 7, and can output calculation results and control signals to these devices. The processor 71 may also control the various components of the computer 7 by executing the computer 7's OS (Operating System) or applications.

[0094] Each device (search device 1) in the above-described embodiment may be implemented by one or more processors 71. Here, processor 71 may refer to one or more electronic circuits arranged on one chip, or one or more electronic circuits arranged on two or more chips or two or more devices. When multiple electronic circuits are used, each electronic circuit may communicate by wire or wireless.

[0095] The main memory 72 is a storage device that stores instructions executed by the processor 71 and various data, and the information stored in the main memory 72 is read by the processor 71. The auxiliary storage device 73 is a storage device other than the main memory 72. These storage devices refer to any electronic component capable of storing electronic information, and may be semiconductor memory. The semiconductor memory may be either volatile memory or non-volatile memory. In each device (search device 1) in the above-described embodiment, the storage device for storing various data may be implemented by the main memory 72 or the auxiliary storage device 73, or by the built-in memory of the processor 71. For example, the storage unit 102 in the above-described embodiment may be implemented by the main memory 72 or the auxiliary storage device 73.

[0096] Multiple processors may be connected to one memory device, or only one processor may be connected to one memory device. Multiple memory devices may be connected to one processor. In the above-described embodiment, if each device (search device 1) consists of at least one memory device and multiple processors connected to this at least one memory device, the configuration may include at least one of the multiple processors being connected to at least one memory device. This configuration may also be realized by memory devices and processors included in multiple computers. Furthermore, the configuration may include a memory device integrated with a processor (for example, a cache memory including an L1 cache and an L2 cache).

[0097] The network interface 74 is an interface for connecting to the communication network 8 wirelessly or via a wired connection. The network interface 74 can be any appropriate interface, such as one conforming to existing communication standards. Information may be exchanged between the computer 7 and an external device 9A connected via the communication network 8 through the network interface 74. The communication network 8 may be a WAN (Wide Area Network), LAN (Local Area Network), PAN (Personal Area Network), or a combination thereof, as long as information is exchanged between the computer 7 and the external device 9A. An example of a WAN is the Internet; an example of a LAN is IEEE 802.11 or Ethernet®; and an example of a PAN is Bluetooth® or NFC (Near Field Communication).

[0098] The device interface 75 is an interface such as USB that connects directly to the external device 9B.

[0099] External device 9A is a device connected to computer 7 via a network. External device 9B is a device directly connected to computer 7.

[0100] External device 9A or external device 9B may, for example, be an input device. The input device may be, for example, a camera, microphone, motion capture device, various sensors, keyboard, mouse, or touch panel, and will provide the acquired information to computer 7. Alternatively, it may be a device equipped with an input unit, memory, and processor, such as a personal computer, tablet terminal, or smartphone.

[0101] Furthermore, external device 9A or external device 9B may, for example, be an output device. The output device may be a display device such as an LCD (Liquid Crystal Display), CRT (Cathode Ray Tube), PDP (Plasma Display Panel), or organic EL (Electro Luminescence) panel, or it may be a speaker that outputs sound, etc. It may also be a device equipped with an output unit, memory, and processor, such as a personal computer, tablet terminal, or smartphone.

[0102] Furthermore, external device 9A or external device 9B may be a storage device (memory). For example, external device 9A may be network storage, and external device 9B may be storage such as an HDD.

[0103] Furthermore, the external device 9A or external device 9B may be a device that has some of the functions of the components of each device (search device 1) in the embodiment described above. In other words, the computer 7 may transmit or receive some or all of the processing results of the external device 9A or external device 9B.

[0104] Where the expression "at least one of a, b, and c" or "at least one of a, b, or c" (including similar expressions) is used in this specification (including the claims), it includes any of a, b, c, ab, ac, bc, or abc. Furthermore, it may include multiple instances of any element, such as aa, abb, aabbcc, etc. It also includes adding other elements besides the enumerated elements (a, b, and c), such as abcd having d.

[0105] In this specification (including the claims), when expressions such as "data as input / based on / according to / in accordance with data" (including similar expressions) are used, unless otherwise specified, this includes cases where the data itself is used as input, or where the data has been processed in some way (e.g., data with added noise, normalized data, intermediate representations of the data, etc.) is used as input. Furthermore, when it is stated that some result is obtained "based on / according to / in accordance with data", this includes cases where the result is obtained based solely on the data in question, as well as cases where the result is also influenced by other data, factors, conditions, and / or states other than the data in question. Furthermore, when it is stated that "data is output", unless otherwise specified, this includes cases where the data itself is used as output, or where the data has been processed in some way (e.g., data with added noise, normalized data, intermediate representations of the data, etc.) is used as output.

[0106] In this specification (including the claims), the terms “connected” and “coupled” are intended to be non-restrictive terms that include any direct connection / coupling, indirect connection / coupling, electrical connection / coupling, communicative connection / coupling, operational connection / coupling, physical connection / coupling, etc. The terms should be interpreted as appropriate in the context in which they are used, but any form of connection / coupling that is not intentionally or naturally excluded should be interpreted non-restrictively as being included in the terms.

[0107] In this specification (including the claims), when the expression "A configured to B" is used, it may include that the physical structure of element A has a configuration capable of performing operation B, and that the permanent or temporary setting / configuration of element A is configured to actually perform operation B. For example, if element A is a general-purpose processor, it is sufficient that the processor has a hardware configuration capable of performing operation B, and that it is configured to actually perform operation B by the setting of a permanent or temporary program (instruction). Furthermore, if element A is a dedicated processor or dedicated arithmetic circuit, it is sufficient that the circuit structure of the processor is implemented to actually perform operation B, regardless of whether control instructions and data are actually attached.

[0108] Wherever terms meaning "comprising" or "possessing" (e.g., "comprising / including" and "having") are used in this specification (including the claims), they are intended to be open-ended terms, including cases where the subject matter of such terms is not the object of the term. Where the object of such terms meaning "comprising" or "possessing" is an expression that does not specify a quantity or suggests a singular number (an expression with the article "a" or "an"), such expression should be interpreted as not being limited to a specific number.

[0109] In this specification (including the claims), even if expressions such as "one or more" or "at least one" are used in one place, and expressions that do not specify a quantity or suggest a singularity (expressions using the articles a or an) are used in another place, the latter expressions are not intended to mean "one." In general, expressions that do not specify a quantity or suggest a singularity (expressions using the articles a or an) should be interpreted as not necessarily being limited to a specific number.

[0110] In this specification, if a particular configuration of an embodiment is described as yielding a specific advantage or result, it should be understood, unless otherwise stated, that the same advantage or result can also be obtained from one or more other embodiments having the same configuration. However, it should be understood that the presence or absence of such an advantage or result generally depends on various factors, conditions, and / or states, and that the configuration does not necessarily guarantee that the advantage or result can be obtained. The advantage or result can only be obtained from the configuration described in the embodiment when various factors, conditions, and / or states are met, and the advantage or result cannot necessarily be obtained in the invention claimed to define that configuration or a similar configuration.

[0111] In this specification (including the claims), when terms such as "maximize" are used, they include finding the global maximum value, finding an approximation of the global maximum value, finding the local maximum value, and finding an approximation of the local maximum value, and should be interpreted appropriately depending on the context in which the term is used. They also include finding approximations of these maximum values ​​probabilistically or heuristically. Similarly, when terms such as "minimize" are used, they include finding the global minimum value, finding an approximation of the global minimum value, finding the local minimum value, and finding an approximation of the local minimum value, and should be interpreted appropriately depending on the context in which the term is used. They also include finding approximations of these minimum values ​​probabilistically or heuristically. Similarly, when terms such as "optimize" are used, they include finding the global optimal value, finding an approximation of the global optimal value, finding the local optimal value, and finding an approximation of the local optimal value, and should be interpreted appropriately depending on the context in which the term is used. They also include finding approximations of these optimal values ​​probabilistically or heuristically.

[0112] In this specification (including the claims), when multiple hardware components perform a predetermined process, each component may cooperate to perform the predetermined process, or some components may perform all of the predetermined process. Alternatively, some components may perform part of the predetermined process, while other components perform the remainder. In this specification (including the claims), when expressions such as "one or more hardware components perform a first process, and the one or more hardware components perform a second process" are used, the hardware component performing the first process and the hardware component performing the second process may be the same or different. In other words, it is sufficient that the hardware component performing the first process and the hardware component performing the second process are included in the one or more hardware components. Hardware may include electronic circuits or devices containing electronic circuits.

[0113] While embodiments of this disclosure have been described in detail above, this disclosure is not limited to the individual embodiments described above. Various additions, modifications, substitutions, and partial deletions are possible, provided that they do not depart from the conceptual idea and spirit of the present invention derived from the claims and their equivalents. For example, where numerical values ​​or mathematical formulas are used in the description in all of the embodiments described above, they are provided as examples only and are not limited thereto. Also, the order of operations in the embodiments is provided as examples only and is not limited thereto. [Explanation of symbols]

[0114] 1: Exploration device, 100: Input section, 102: Memory section, 104: Output section, 106: Elementary reaction identification unit, 108: Promoter Elemental Search Department, 110: Promoter placement optimization unit

Claims

1. It comprises an elementary reaction identification unit and a promoter placement optimization unit, The elementary reaction identification unit is, We obtained the effect on the target properties of each of the one or more elementary reactions included in a reaction using a catalyst and adsorbed molecules. Based on the influence on the target characteristics, identify the elementary reactions that affect the target characteristics from the one or more elementary reactions. The promoter placement optimization unit is, Based on the elementary reaction identified by the elementary reaction identification unit, the arrangement of the promoter element in the catalyst is optimized based on physical property values ​​obtained by inputting data on the atomic structure of the promoter element, the catalyst, and the adsorbed molecule into the Neural Network Potential (NNP). Exploration device.

2. A promoter element search unit searches for the optimal promoter element for the catalyst based on the physical properties obtained by optimization in the promoter arrangement optimization unit for each of the multiple types of promoter elements, for each of the multiple types of promoter elements. The search device according to claim 1, further comprising:

3. A promoter element search unit searches for a promoter element whose physical property value is lower than that of other promoter elements or whose physical property value is higher than that of other promoter elements, based on the physical property values ​​for multiple types of promoter elements obtained by the promoter arrangement optimization unit. The search device according to claim 1, further comprising:

4. The promoter placement optimization unit is, In the catalyst, the arrangement of one or more promoter elements is specified, and the reaction pathway search using the NNP is repeated multiple times. Optimize the arrangement of the promoter element whose physical property value is lower than that of the other arrangements, or the arrangement of the promoter element whose physical property value is higher than that of the other arrangements. The search device according to claim 1.

5. The promoter placement optimization unit is, The NNP is used to obtain the physical properties of the adsorbed molecules used in the reaction pathway search, with the structures of the adsorbed molecules arranged in the same position. The search device according to claim 4.

6. The aforementioned NNP consists of a neural network model trained on multiple elements. A search device according to any one of claims 1 to 5.

7. The promoter placement optimization unit is, The arrangement of the promoter element is set to be within 5 Å of the position of the adsorbed molecule. A search device according to any one of claims 1 to 5.

8. The promoter placement optimization unit is, The arrangement of one of the promoter elements is optimized using grid search. A search device according to any one of claims 1 to 5.

9. The promoter placement optimization unit is, The arrangement of multiple promoter elements is optimized using at least one of Bayesian optimization and random search. A search device according to any one of claims 1 to 5.

10. The promoter placement optimization unit is, Of the atomic structure input to the NNP, excluding the atoms constituting the adsorbed molecule, less than 10% of the atoms are arranged as the promoter elements. A search device according to any one of claims 1 to 5.

11. The promoter placement optimization unit is, The search apparatus according to any one of claims 1 to 5, wherein some of the atoms of the catalyst are replaced with the promoter element, and / or the promoter element is added to the atoms of the catalyst to arrange the promoter element.

12. The promoter placement optimization unit further: Multiple types of the aforementioned promoter elements are arranged, The ratio of the multiple types of promoter elements and the arrangement of each of the promoter elements are optimized. A search device according to any one of claims 1 to 5.

13. The aforementioned physical properties include at least one of the following: activation energy, physical properties correlated with activation energy, intermolecular distance, atomic charge, adsorption energy, frequency, d-band centroid, or the energy of the reaction intermediate. The search device according to claim 1.

14. The elementary reaction identification unit is, The simulation was performed by changing the reaction rate constant of each of the aforementioned elementary reactions. The parameters of the aforementioned simulation are calculated using the NNP. A search device according to any one of claims 1 to 5.

15. The simulation described above calculates the rate of change of the reaction rate of each of the elementary reactions. The elementary reaction identification unit is, From the one or more elementary reactions mentioned above, identify the elementary reaction with the highest rate of change in the target characteristic. The search device according to claim 14.

16. The aforementioned promoter element search unit is Active learning is used to train a model that predicts activation energy. The search device according to claim 2 or claim 3.

17. The aforementioned target characteristic is reaction time or yield. A search device according to any one of claims 1 to 5.

18. The effect on the target characteristics is the rate of change of the target characteristics. The elementary reaction identification unit is, From the one or more elementary reactions mentioned above, identify the elementary reaction with the highest rate of change in the target characteristic. The search device according to claim 17.

19. Computers We obtained the effect on the target properties of each of the one or more elementary reactions included in a reaction using a catalyst and adsorbed molecules. Based on the influence on the target characteristics, identify the elementary reactions that affect the target characteristics from the one or more elementary reactions. Based on the identified elementary reaction, the arrangement of the promoter element in the catalyst is optimized based on the physical properties obtained by inputting data on the atomic structure of the promoter element, the catalyst, and the adsorbed molecule into the NNP. Search method.

20. On the computer, We obtained the effect on the target properties of each of the one or more elementary reactions included in a reaction using a catalyst and adsorbed molecules. Based on the influence on the target characteristics, identify the elementary reactions that affect the target characteristics from the one or more elementary reactions. Based on the identified elementary reaction, the arrangement of the promoter element in the catalyst is optimized based on the physical properties obtained by inputting data on the atomic structure of the promoter element, the catalyst, and the adsorbed molecule into the NNP. A program that executes a method.

21. On the computer, We obtained the effect on the target properties of each of the one or more elementary reactions included in a reaction using a catalyst and adsorbed molecules. Based on the influence on the target characteristics, identify the elementary reactions that affect the target characteristics from the one or more elementary reactions. Based on the identified elementary reaction, the arrangement of the promoter element in the catalyst is optimized based on the physical properties obtained by inputting data on the atomic structure of the promoter element, the catalyst, and the adsorbed molecule into the NNP. A non-temporary, computer-readable medium that stores a program that executes a method.

Images