Analysis method for material structure and analyzer

Abstract

To easily extract a local structure of a material.SOLUTION: An analysis method 10 for analyzing a structure of a material, includes: extracting a structure factor from a scattering intensity spectrum (SAXS spectrum) obtained by measuring electromagnetic waves or particle beams scattered by which the material is irradiated (step S4); and calculating a local structure factor representing a local structure of the material from the extracted structure factor (step S5). Calculating the local structure factor (step S5) is to calculate a differential structure factor representing a difference between the structure factor and a reference structure factor by dividing the extracted structure factor by the reference structure factor.SELECTED DRAWING: Figure 4

Description

[Technical field]

[0001] The present invention relates to a method and an apparatus for analyzing a material structure. [Background technology]

[0002] Patent Document 1 discloses a method for evaluating the three-dimensional structure of a material containing a resin and particles by using a reverse Monte Carlo simulation, in which the reverse Monte Carlo simulation is performed using a parallel computation process between nodes. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2015-25746 A Summary of the Invention [Problem to be solved by the invention]

[0004] In the method disclosed in Patent Document 1, a structure factor is extracted from a scattering intensity spectrum (hereinafter also referred to as a "SAXS spectrum") measured by a small angle X-ray scattering method. In the method disclosed in Patent Document 1, the extracted structure factor is used to perform a reverse Monte Carlo simulation to obtain a structure factor and a radial distribution function.

[0005] However, although the method disclosed in Patent Document 1 makes it possible to extract the average structure of the entire material from the obtained structure factor and radial distribution function, it is difficult to easily extract the local structure of the material.

[0006] The present invention has been made in view of the above, and has an object to easily extract the local structure of a material. [Means for solving the problem]

[0007] In order to solve the above problems, the analysis method of the present invention is an analysis method for analyzing a structure of a material, which includes extracting a structure factor from a scattering intensity spectrum obtained by measuring electromagnetic waves or particle beams irradiated and scattered by the material, and calculating a local structure factor representing the local structure of the material from the extracted structure factor, wherein calculating the local structure factor is characterized in that a differential structure factor representing the difference between the structure factor and the reference structure factor is calculated by dividing the extracted structure factor by a reference structure factor.

[0008] In addition, an analytical apparatus according to the present invention is an analytical apparatus for analyzing a structure of a material, and includes a setting unit that sets a scattering intensity spectrum obtained by measuring electromagnetic waves or particle beams irradiated to and scattered by the material, an extraction unit that extracts a structure factor from the set scattering intensity spectrum, and a calculation unit that calculates a local structure factor representing the local structure of the material from the extracted structure factor, wherein the calculation unit calculates a differential structure factor representing the difference between the structure factor and the reference structure factor by dividing the extracted structure factor by a reference structure factor, and outputs the calculated differential structure factor as the local structure factor. Effect of the Invention

[0009] According to the present invention, the local structure of a material can be easily extracted. [Brief description of the drawings]

[0010] [Figure 1] FIG. 1 is a diagram showing the configuration of an information processing system including an analysis device according to a first embodiment. [Diagram 2] FIG. 2 is a diagram for explaining the function of the analysis device shown in FIG. 1. [Diagram 3] FIG. 2 is a diagram for explaining differential structure factors calculated by the calculation unit shown in FIG. 1. [Figure 4] 4 is a flowchart showing an analysis method according to the first embodiment. [Diagram 5] FIG. 11 is a diagram for explaining another example of the canonical structure factor. [Figure 6]FIG. 13 is a diagram showing the configuration of an analysis device according to a second embodiment. [Figure 7] FIG. 1 is a diagram illustrating a virtual obstacle that simulates the local structure of a material. [Figure 8] FIG. 13 is a diagram showing the results of verification of the prediction accuracy of a prediction model. [Figure 9] 10 is a flowchart showing an analysis method according to a second embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0011] Hereinafter, embodiments of the present invention will be described with reference to the drawings. In the embodiments, the same reference numerals denote the same configurations or functions, and unless otherwise specified, the description thereof will be omitted.

[0012] [First embodiment] A first embodiment of the present invention will be described with reference to Fig. 1 to Fig. 5. Fig. 1 is a diagram showing the configuration of an information processing system 1 including an analysis device 4 of the first embodiment. Fig. 2 is a diagram explaining the function of the analysis device 4 shown in Fig. 1. Fig. 3 is a diagram explaining the differential structure factor calculated by the calculation unit 44 shown in Fig. 1.

[0013] The information processing system 1 is a system that analyzes the scattering intensity spectrum of a material input by a user using an analysis device 4 and displays the analysis result to the user. This scattering intensity spectrum is obtained by measuring electromagnetic waves or particle beams irradiated and scattered by the material. The electromagnetic waves may be, for example, electromagnetic radiation such as X-rays or gamma rays, or visible light. The particle beams may be, for example, alpha rays, beta rays, electron beams, proton beams, neutron beams, heavy particle beams, etc. In this embodiment, X-rays are taken as an example of the electromagnetic waves or particle beams irradiated and scattered by the material, and a SAXS (Small Angle X-ray Scattering) spectrum is taken as an example of the scattering intensity spectrum.

[0014] The information processing system 1 includes a user terminal 2 and a server 3 having an analysis device 4. The user terminal 2 and the server 3 are connected via a network N so as to be able to communicate with each other.

[0015] The user terminal 2 includes a processing device 21 and a display device 22. The processing device 21 includes a processor and a memory, and the processor executes a program to realize the functions of the user terminal 2. The display device 22 is composed of a display, and displays the processing results of the processing device 21.

[0016] The user terminal 2 accepts the SAXS spectrum and material information input by the user, and transmits it to the analysis device 4 of the server 3. The user terminal 2 receives the analysis results of the analysis device 4 from the server 3, displays them, and presents them to the user.

[0017] The analysis device 4 analyzes the structure of the material to be analyzed from the SAXS spectrum and material information input by the user. The material to be analyzed by the analysis device 4 is not particularly limited as long as it is a material that is generally analyzed by SAXS. The material to be analyzed may be a solid or a liquid. The material to be analyzed may contain solid, liquid or gas particles (including a region (domain) that can be considered as a particle) in the material. The particles contained in the material may be the same material as the base material of the material, or may be a different material. The material to be analyzed may contain a specific structure, such as a lamellar structure or a fractal structure. Specific examples of the material to be analyzed include colloids, metals, cement, oil, rubber, polymers, plastics, and proteins.

[0018] The material information is information about the material to be analyzed, and includes at least one of the size, shape, and amount of added particles contained in the material.

[0019] A SAXS spectrum is a scattering intensity spectrum obtained by measuring the scattered X-rays that are generated when the material to be analyzed is irradiated with X-rays and that are detected in the region where the scattering angle is small (for example, about 5 degrees) or less. As shown in Figure 2, a SAXS spectrum is represented by a graph with the wave number on the horizontal axis and the scattering intensity on the vertical axis. A SAXS spectrum expresses nanoscale periodic structural information such as the size, shape, and arrangement (interparticle distance) of particles contained in a material.

[0020] The SAXS spectrum is expressed as the sum of the scattering intensity from each particle. When the interaction between particles cannot be ignored, such as when the particle concentration is high or the particles are aggregated, the scattering intensity of the SAXS spectrum is divided into a form factor and a structure factor. The form factor represents the scattering intensity caused by the shape or size of the particles among the scattering intensity of the SAXS spectrum. The structure factor represents the scattering intensity caused by the arrangement of the particles among the scattering intensity of the SAXS spectrum. The scattering intensity of the SAXS spectrum can be considered to be proportional to the product of the structure factor and the form factor. The radial distribution function is obtained by performing an inverse Fourier transform on the structure factor.

[0021] The analysis device 4 calculates the local structure factor included in the structure factor of the SAXS spectrum input by the user. The local structure factor is a structure factor that represents the local structure of a material, such as including an aggregated region of particles and / or a depleted region, among the structure factors of the SAXS spectrum.

[0022] The analysis device 4 includes a processing device 40 and a storage device 45. The storage device 45 includes a storage for storing the SAXS spectrum and material information received from the user terminal 2, and the analysis results of the analysis device 4. The processing device 40 includes a processor and a memory, and the processor executes a program to realize each function. As the respective functions, the processing device 40 includes a setting unit 41, a pre-calculation unit 42, an extraction unit 43, and a calculation unit 44.

[0023] The setting unit 41 reads out from the storage device 45 the SAXS spectrum and material information input by the user and sets them.

[0024] The pre-calculation unit 42 calculates the set shape factor and reference structure factor of the SAXS spectrum. The reference structure factor is a structure factor representing a reference structure that is an average structure of the entire material among the structure factors of the SAXS spectrum. In the case of a material having a local structure including particle aggregation regions and / or depletion regions, the reference structure of the material is a structure in which the particles are randomly arranged within the material. In this case, the reference structure factor is a structure factor representing a structure in which the particles are randomly arranged within the material among the structure factors of the SAXS spectrum. Note that a structure in which the particles are randomly arranged within the material refers to a structure in which the particles are uniformly arranged without any regularity within the material.

[0025] The pre-calculation unit 42 can calculate the shape factor of the set SAXS spectrum using various methods. For example, the pre-calculation unit 42 can directly use a SAXS spectrum measured separately under conditions where interactions between particles can be ignored (for example, conditions where particles are sparse) as the shape factor. For example, the pre-calculation unit 42 can calculate the shape factor from the set material information (particle shape, size, and amount added) using a known theoretical formula (for example, a theoretical scattering function). The pre-calculation unit 42 of this embodiment calculates the shape factor based on the set material information.

[0026] The pre-calculation unit 42 can calculate the reference structure factor of the set SAXS spectrum using various methods. For example, the pre-calculation unit 42 can calculate the reference structure factor from the set material information using a known theoretical formula such as the Percus-Yevick equation. For example, the pre-calculation unit 42 can calculate the reference structure factor by generating a three-dimensional model of the material from the set material information and performing a Fourier transform. In any case, the pre-calculation unit 42 of this embodiment calculates the reference structure factor based on the set material information.

[0027] The extraction unit 43 extracts a structure factor from the set SAXS spectrum. Specifically, the extraction unit 43 divides (divides) the scattering intensity of the SAXS spectrum for each wave number of the SAXS spectrum by the shape factor calculated by the pre-calculation unit 42. In this way, the extraction unit 43 extracts a structure factor from the SAXS spectrum.

[0028] The calculation unit 44 calculates a local structure factor from the structure factor extracted by the extraction unit 43. Specifically, the calculation unit 44 divides the structure factor extracted by the extraction unit 43 by the reference structure factor calculated by the pre-calculation unit 42 to calculate a differential structure factor representing the difference between the structure factor and the reference structure factor. The calculation unit 44 then outputs the calculated differential structure factor as a local structure factor. The local structure factor output from the calculation unit 44 is transmitted to the user terminal 2 and displayed on the display device 22.

[0029] The structure factor extracted by the extraction unit 43, i.e., the structure factor extracted by dividing the SAXS spectrum by the shape factor, does not necessarily clearly express the size of the local structure (e.g., the diameter of the local region) as a peak intensity. Furthermore, since the structure factor extracted by the extraction unit 43 does not have a clear standard for intensity, it is difficult to evaluate the absolute value of the peak intensity itself, as shown on the left side of Fig. 3. Therefore, it is difficult to evaluate the amount of local structure (e.g., the ratio of the volume of the local region to the entire region) of the structure factor extracted by the extraction unit 43 by evaluating the absolute value of the peak intensity.

[0030] In contrast, the differential structure factor calculated by the calculation unit 44 divides the structure factor extracted by the extraction unit 43 by the reference structure factor, and therefore the size of the local structure can be clearly expressed as a peak intensity. Furthermore, since the differential structure factor calculated by the calculation unit 44 divides the structure factor extracted by the extraction unit 43 by the reference structure factor, it can be converted into a spectrum centered on intensity 1, as shown on the right side of Fig. 3, and the reference for intensity can be made clear. Therefore, the differential structure factor calculated by the calculation unit 44 makes it possible to evaluate the absolute value of the peak intensity itself, and to evaluate the amount of local structure present.

[0031] FIG. 4 is a flow chart showing the analysis method 10 of the first embodiment.

[0032] The material structure analysis method 10 shown in FIG. 4 is realized by the processing device 40 of the analysis device 4 executing a program that implements the functions of a setting unit 41, a pre-calculation unit 42, an extraction unit 43 and a calculation unit 44.

[0033] In step S1, the processing device 40 reads out from the storage device 45 the SAXS spectrum and material information input by the user and sets them.

[0034] In step S2, the processing device 40 calculates the shape factor of the SAXS spectrum based on the material information.

[0035] In step S3, the processing device 40 calculates a reference structure factor of the SAXS spectrum based on the material information.

[0036] In step S4, the processing device 40 extracts a structure factor from the SAXS spectrum by dividing each scattering intensity in the SAXS spectrum by the shape factor.

[0037] In step S5, the processing unit 40 calculates the differential structure factor by dividing the structure factor by the reference structure factor.

[0038] In step S6, the processing device 40 outputs the differential structure factor as a local structure factor to the storage device 45 for storage. The processing device 40 outputs the local structure factor stored in the storage device 45 to the communication device of the server 3. The communication device of the server 3 transmits the local structure factor output from the processing device 40 to the user terminal 2. The user terminal 2 displays the local structure factor transmitted from the server 3 on the display device 22.

[0039] As described above, the analysis method 10 of the present embodiment is an analysis method for analyzing the structure of a material, and includes steps of extracting a structure factor from a scattering intensity spectrum (SAXS spectrum) obtained by measuring an electromagnetic wave or particle beam irradiated and scattered by the material (step S4), and calculating a local structure factor representing the local structure of the material from the extracted structure factor (step S5). The calculation of the local structure factor (step S5) is to calculate a differential structure factor representing the difference between the structure factor and the reference structure factor by dividing the extracted structure factor by the reference structure factor.

[0040] As a result, the analysis method 10 of this embodiment can relatively easily obtain a local structure factor that clearly expresses the size of the local structure of a material and can evaluate the amount of the local structure. In addition, since the analysis method 10 of this embodiment can regard the differential structure factor as a local structure factor, it can easily extract and evaluate various local structure features by applying an analysis generally performed on a structure factor. Therefore, the analysis method 10 of this embodiment can easily extract the local structure of a material and quantitatively evaluate it.

[0041] Furthermore, in the analysis method 10 of this embodiment, the material is a material containing particles, and the reference structure factor is a structure factor that represents a structure in which the particles are randomly arranged within the material.

[0042] As a result, the analysis method 10 of the present embodiment can extract and evaluate the local structures of various materials, regardless of the presence or absence of orientation of particles contained in the material. Therefore, the analysis method 10 of the present embodiment can easily extract and quantitatively evaluate the local structure of a material, and can also increase versatility.

[0043] Furthermore, the analysis method 10 of this embodiment further includes setting material information including at least one of particle size, shape, and additive amount (step S1), calculating a shape factor based on the set material information (step S2), and calculating a reference structure factor based on the set material information (step S3). Extracting the structure factor (step S4) is to divide the scattering intensity of the scattering intensity spectrum (SAXS spectrum) by the shape factor for each wave number of the scattering intensity spectrum (SAXS spectrum).

[0044] As a result, the analysis method 10 of this embodiment can calculate the shape factor by simply preparing material information without separately measuring the SAXS spectrum under conditions where the interaction between particles can be ignored. Therefore, the analysis method 10 of this embodiment can easily extract the structure factor, and can further easily calculate the local structure factor. Therefore, the analysis method 10 of this embodiment can further easily extract the local structure of the material and quantitatively evaluate it, while improving the convenience for the user.

[0045] The analysis device 4 of this embodiment can also provide the same effects as those of the analysis method 10 of this embodiment.

[0046] FIG. 5 is a diagram for explaining another example of the canonical structure factor.

[0047] In the above description, the reference structure factor is a structure factor that represents a structure in which particles are randomly arranged in a material. However, some materials have a specific particle arrangement in advance, or the particle arrangement of some materials changes to a specific particle arrangement due to deformation during molding or use.

[0048] Therefore, in the analysis method 10 of the present embodiment, the reference structure factor can be a structure factor that represents a structure in which particles are arranged with an orientation in a material. As a result, the analysis method 10 of the present embodiment can more accurately extract the local structure of a material in which particles have a specific orientation, and can perform more accurate quantitative evaluation.

[0049] For example, when metal or rubber materials are stretched, these materials have a structure in which particles are arranged with uniaxial anisotropy in the material, as shown on the left side of FIG. 5. The reference structure of these materials can be said to be a structure in which particles are arranged with uniaxial anisotropy in the material. In this case, the analysis method 10 of the present embodiment can set the reference structure factor to a structure factor representing a structure in which particles are arranged with uniaxial anisotropy in the material. As a result, the analysis method 10 of the present embodiment can more accurately extract the local structure of the material in which particles are arranged with uniaxial anisotropy, and can perform more accurate quantitative evaluation.

[0050] Also, for example, when a metal, polymer, or protein material is compressed from a specific direction, or when a material has a lamellar structure, these materials have a structure in which particles are arranged in layers within the material, as shown on the right side of FIG. 5. The reference structure of these materials can be said to be a structure in which particles are arranged in layers within the material. In this case, the analysis method 10 of the present embodiment can set the reference structure factor to a structure factor that represents a structure in which particles are arranged in layers within the material. As a result, the analysis method 10 of the present embodiment can more accurately extract the local structure of a material in which particles are arranged in layers, and can perform more accurate quantitative evaluation.

[0051] [Second embodiment] A second embodiment of the present invention will be described with reference to Figures 6 to 9. In the second embodiment, the description of the same components as in the first embodiment will be omitted.

[0052] Fig. 6 is a diagram showing the configuration of an analysis device 4 of the second embodiment. Fig. 7 is a diagram explaining a virtual obstacle that simulates the local structure of a material.

[0053] The analysis device 4 of the second embodiment predicts the value of a parameter representing the size or abundance of a local structure from the differential structure factors and the like obtained by the analysis device 4 of the first embodiment.

[0054] The analysis device 4 of the second embodiment includes, as functions included in the processing device 40, a setting unit 41, a pre-calculation unit 42, an extraction unit 43, and a calculation unit 44, similarly to the first embodiment. In addition, the analysis device 4 of the second embodiment includes, as functions included in the processing device 40, a target variable setting unit 46, a principal component analysis unit 47, and a prediction unit 48.

[0055] The objective variable setting unit 46 sets a parameter to be used as an objective variable, that is, a parameter to be predicted, among parameters representing the size or abundance of the local structure. The size of the local structure is, for example, the size of a local region in a material having the local structure. The parameter representing the size of the local structure is, for example, the diameter of the local region. The abundance of the local structure is, for example, the abundance of the local region in a material having the local structure. The parameter representing the abundance of the local structure may be, for example, the ratio of the volume of the local region to the entire region of the material, or the ratio of the volume of the local region to the entire void region in the material (the entire region of the material excluding the region occupied by the particles). In this embodiment, the parameter representing the abundance of the local structure is the ratio of the volume of the local region to the entire void region in the material. These parameters are feature quantities representing the characteristics of the local structure of the material.

[0056] The parameters to be the objective variables are specified by the material information input by the user. The objective variable setting unit 46 reads the material information input by the user from the storage device 45 and sets the parameters to be the objective variables. The parameters to be the objective variables may be set in advance in the analysis device 4 and stored in the storage device 45.

[0057] The principal component analysis unit 47 performs principal component analysis on the structure factor extracted by the extraction unit 43 or the differential structure factor (i.e., local structure factor) calculated by the calculation unit 44. As a result, the principal component analysis unit 47 acquires a principal component vector and a principal component score of the structure factor or the differential structure factor. The principal component vector is a vector whose elements are the first principal component, the second principal component, the third principal component, ..., the n-th principal component (for example, the number of principal components n = 10) calculated by the principal component analysis. The principal component vector indicates wave number components with large variance in the multiple scattering intensities that constitute the structure factor or the differential structure factor. The principal component score indicates the proportion of each principal component included.

[0058] The principal component analysis unit 47 performs principal component analysis by determining the target of principal component analysis (structure factor or differential structure factor) according to the parameter set by the objective variable setting unit 46. Specifically, when the parameter set by the objective variable setting unit 46 is a parameter representing the size of the local structure, the principal component analysis unit 47 performs principal component analysis on the differential structure factor calculated by the calculation unit 44. When the parameter set by the objective variable setting unit 46 is a parameter representing the abundance of the local structure, the principal component analysis unit 47 performs principal component analysis on the structure factor extracted by the extraction unit 43.

[0059] The prediction unit 48 predicts the value of a parameter representing the size or abundance of a local structure. The prediction unit 48 predicts the value of the parameter from the structure factor extracted by the extraction unit 43 or the differential structure factor calculated by the calculation unit 44, using a prediction model 48a that has been trained in advance by machine learning.

[0060] Specifically, the prediction unit 48 reads out the prediction model 48a linked to the parameter set by the objective variable setting unit 46 from among a plurality of prediction models 48a previously stored in the storage device 45 in association with a parameter representing the size or abundance of a local structure. The prediction unit 48 inputs the principal component scores acquired by the principal component analysis unit 47 as explanatory variables into the read out prediction model 48a. The prediction model 48a predicts the value of the parameter set as the objective variable from the principal component scores input as explanatory variables. The prediction unit 48 outputs the parameter value predicted by the prediction model 48a. The parameter value output from the prediction unit 48 is transmitted to the user terminal 2 and displayed on the display device 22.

[0061] The method for constructing the prediction model 48a by machine learning is not particularly limited, but may be the following method.

[0062] First, in order to obtain a SAXS spectrum as data used in machine learning, a three-dimensional model simulating the structure of a material containing particles is created. At this time, as shown in FIG. 7, a local structure is expressed by arranging a virtual obstacle in the three-dimensional model. The size of the virtual obstacle is, for example, the diameter of the virtual obstacle, simulating a parameter representing the size of the local structure. The amount of the virtual obstacle is, for example, the ratio of the volume of the virtual obstacle to the entire void region in the material, simulating a parameter representing the amount of the local structure. The shape of the virtual obstacle is not particularly limited, but may be, for example, a sphere (including an ellipsoidal sphere). Then, a plurality of three-dimensional models are created by changing the respective values ​​of the size and amount of the virtual obstacle. In this way, a plurality of three-dimensional models with different values ​​of the parameter representing the size of the local structure and a plurality of three-dimensional models with different values ​​of the parameter representing the amount of the local structure are created. In addition, a three-dimensional model in which particles are randomly arranged in the material and there is no arrangement of virtual obstacles is created as a three-dimensional model simulating the material structure of the reference structure factor. The model simulating the structure of the material may be a two-dimensional model instead of a three-dimensional model.

[0063] Next, the SAXS spectrum measured when each of the multiple 3D models is irradiated with X-rays is calculated. Then, using a method similar to that of the pre-calculation unit 42, the shape factor and reference structure factor of each of the multiple acquired SAXS spectra are calculated. Then, using a method similar to that of the extraction unit 43, a structure factor is extracted from each of the multiple SAXS spectra. Then, using a method similar to that of the calculation unit 44, a differential structure factor is calculated from each of the multiple structure factors.

[0064] Next, a scaling process such as a centering process is performed on each of the plurality of structure factors and the plurality of differential structure factors to reduce the variability of the data.

[0065] Next, a principal component analysis is performed on each of the plurality of structure factors to obtain a principal component vector and a principal component score. Similarly, a principal component analysis is performed on each of the plurality of differential structure factors to obtain a principal component vector and a principal component score. The number of principal components n may be, for example, n=10.

[0066] Next, the principal component scores are used as explanatory variables, and the values ​​of parameters representing the size or abundance of the local structure used when creating the three-dimensional model are used as objective variables, and the prediction model 48a is made to learn the relationship between the explanatory variables and the objective variables by machine learning. The learning device is not particularly limited, but may be, for example, a random forest or a neural network. In this way, a plurality of prediction models 48a different for each objective variable are constructed.

[0067] Fig. 8 is a diagram showing the verification results of the prediction accuracy of the prediction model 48a. Fig. 8 shows that the closer the numerical value of "test" is to 1, the higher the prediction accuracy is.

[0068] The prediction accuracy of each of the multiple prediction models 48a, which differ for each objective variable, was verified. Specifically, the prediction accuracy of each prediction model 48a was verified when the original data of the explanatory variables was a structure factor, differential structure factor, or SAXS spectrum (including shape factor and structure factor). In this case, the data set was divided by train / test division in order to learn and verify each prediction model 48a for each case.

[0069] As shown by the bold frame in Fig. 8, in the prediction model 48a in which the value of the parameter representing the size of the local structure is the objective variable, the prediction accuracy is higher when the original data of the explanatory variable is a differential structure factor than when the original data of the explanatory variable is a structure factor or a SAXS spectrum. On the other hand, in the prediction model 48a in which the value of the parameter representing the abundance of the local structure is the objective variable, the prediction accuracy is higher when the original data of the explanatory variable is a structure factor than when the original data of the explanatory variable is a differential structure factor or a SAXS spectrum.

[0070] That is, in order to accurately predict the value of a parameter representing the size or abundance of a local structure, it is not sufficient to simply construct a prediction model 48a using a SAXS spectrum containing a mixture of structure and shape factors as the original data of explanatory variables. In order to accurately predict the value of a parameter representing the size or abundance of a local structure, it is important to construct a prediction model 48a by appropriately selecting the original data of explanatory variables according to the parameter that is the objective variable.

[0071] Therefore, in the analysis device 4 of the second embodiment, when the parameter set by the objective variable setting unit 46 is a parameter representing the size of the local structure, the principal component analysis unit 47 performs principal component analysis on the differential structure factor calculated by the calculation unit 44. The prediction unit 48 predicts the value of the parameter representing the size of the local structure using a prediction model 48a in which the value of the parameter representing the size of the local structure is the objective variable. This allows the analysis device 4 of the second embodiment to accurately predict the value of the parameter representing the size of the local structure.

[0072] On the other hand, in the analysis device 4 of the second embodiment, when the parameter set by the objective variable setting unit 46 is a parameter representing the abundance of local structures, the principal component analysis unit 47 performs principal component analysis on the structure factor extracted by the extraction unit 43. The prediction unit 48 predicts the value of the parameter representing the abundance of local structures using a prediction model 48a in which the value of the parameter representing the abundance of local structures is the objective variable. This allows the analysis device 4 of the second embodiment to accurately predict the value of the parameter representing the abundance of local structures.

[0073] FIG. 9 is a flow chart showing an analysis method 10 of the second embodiment.

[0074] The material structure analysis method 10 shown in FIG. 9 is realized by the processing device 40 of the analysis device 4 of the second embodiment executing a program that implements the functions of a setting unit 41, a pre-calculation unit 42, an extraction unit 43, a calculation unit 44, a target variable setting unit 46, a principal component analysis unit 47, and a prediction unit 48.

[0075] In steps S1 to S6, the processing device 40 performs the same processes as those in FIG.

[0076] In step S7, the processing device 40 reads out the material information input by the user from the storage device 45, and sets a parameter to be used as a response variable among the parameters representing the size or abundance of the local structure.

[0077] In step S8, the processing device 40 performs principal component analysis on the differential structure factor or the structure factor according to the parameter set as the objective variable to obtain a principal component vector and a principal component score. Specifically, when the parameter set as the objective variable is a parameter representing the size of the local structure, the processing device 40 performs principal component analysis on the differential structure factor. When the parameter set as the objective variable is a parameter representing the amount of the local structure, the processing device 40 performs principal component analysis on the structure factor.

[0078] In step S9, the processing device 40 reads out from the storage device 45 the trained prediction model 48a associated with the parameter set as the objective variable. Then, the processing device 40 inputs the principal component scores obtained by the principal component analysis as explanatory variables into the read out prediction model 48a. In this way, the processing device 40 predicts the value of the parameter set as the objective variable. The processing device 40 outputs the prediction result to the storage device 45 and stores it therein. The processing device 40 outputs the prediction result stored in the storage device 45 to the communication device of the server 3. The communication device of the server 3 transmits the prediction result output from the processing device 40 to the user terminal 2. The user terminal 2 displays the prediction result transmitted from the server 3 on the display device 22.

[0079] As described above, the analysis method 10 of the second embodiment further includes predicting a value of a parameter representing the size or abundance of a local structure (step S9) in comparison with the analysis method 10 of the first embodiment. Predicting the value of the parameter (step S9) is to predict the value of the parameter from the differential structure factor or the structure factor using a prediction model 48a that has been trained in advance by machine learning.

[0080] As a result, the analysis method 10 of the second embodiment can accurately predict the size or abundance of the local structure even when it is difficult to distinguish the peak intensity originating from the local structure due to a small absolute value, such as when the abundance of the local structure is small. Therefore, the analysis method 10 of the second embodiment can easily and accurately extract the characteristics of the local structure and express them as numerical values. Therefore, the analysis method 10 of the second embodiment can easily and accurately extract the local structure of the material, and can perform more accurate quantitative evaluation.

[0081] The analysis device 4 of the second embodiment can also provide the same effects as the analysis method 10 of the second embodiment.

[0082] Although the embodiments of the present invention have been described in detail above, the present invention is not limited to the above-mentioned embodiments, and various modifications can be made without departing from the spirit of the present invention. In the present invention, the configuration of one embodiment can be added to the configuration of another embodiment, the configuration of one embodiment can be replaced with the configuration of another embodiment, or a part of the configuration of one embodiment can be deleted. [Explanation of symbols]

[0083] 1...information processing system, 10...analysis method, 2...user terminal, 21...processing device, 22...display device, 3...server, 4...analysis device, 40...processing device, 41...setting unit, 42...pre-calculation unit, 43...extraction unit, 44...calculation unit, 45...storage device, 46...objective variable setting unit, 47...principal component analysis unit, 48...prediction unit, 48a...prediction model, N...network

Claims

1. An analysis method for analyzing a structure of a material, comprising the steps of: Extracting a structure factor from a scattering intensity spectrum obtained by measuring an electromagnetic wave or particle beam irradiated and scattered by the material; Calculating a local structure factor representing a local structure of the material from the extracted structure factors; Calculating the local structure factor is to calculate a differential structure factor representing a difference between the structure factor and the reference structure factor by dividing the extracted structure factor by the reference structure factor. A method for analyzing a material structure comprising the steps of:

2. the material is a material including particles, The reference structure factor is a structure factor that represents a structure in which the particles are randomly arranged within the material.

2. The method for analyzing a material structure according to claim 1 .

3. The analysis method further includes predicting a value of a parameter representing a size or abundance of the local structure, Predicting the value of the parameter means predicting the value of the parameter from the differential structure factor or the structure factor using a prediction model that has been trained in advance by machine learning.

2. The method for analyzing a material structure according to claim 1 .

4. An analysis device for analyzing a structure of a material, comprising: A setting unit that sets a scattering intensity spectrum obtained by measuring an electromagnetic wave or a particle beam that is irradiated and scattered by the material; an extraction unit that extracts a structure factor from the set scattering intensity spectrum; A calculation unit that calculates a local structure factor representing a local structure of the material from the extracted structure factor, The calculation unit calculates a differential structure factor representing a difference between the structure factor and the reference structure factor by dividing the extracted structure factor by a reference structure factor, and outputs the calculated differential structure factor as the local structure factor. The material structure analysis device according to the present invention is characterized in that

5. The analysis apparatus further includes a prediction unit that predicts a value of a parameter representing a size or abundance of the local structure, The prediction unit predicts the value of the parameter from the differential structure factor or the structure factor by using a prediction model that has been trained in advance by machine learning.

5. The material structure analysis device according to claim 4.

Images