TEM image analysis device, TEM image analysis method and program

The TEM image analysis device and method effectively analyze electron and positive charge paths within carbon materials, enhancing lithium ion battery performance prediction and structure optimization.

JP2024167659AActive Publication Date: 2024-12-04TOYOTA JIDOSHA KK +1
View PDF 6 Cites 0 Cited by

Patent Information

Application Number
JP2023083884
Authority / Receiving Office
JP · JP
Patent Type
Applications
Current Assignee / Owner
Filing Date
2023-05-22
Publication Date
2024-12-04
Estimated Expiration
2043-05-22

AI Technical Summary

Technical Problem

Existing TEM image analysis techniques fail to distinguish between the movement paths of electrons and positive charges, such as lithium ions, within structures like lithium ion batteries, limiting their analysis capabilities.

Method used

A TEM image analysis device and method that includes an image vectorization unit, line information generation, image inversion, explanatory variable extraction, and performance prediction to analyze the movement paths of electrons and positive charges by converting TEM images into numerical values and identifying their paths within carbon materials.

Benefits of technology

Enables the analysis of electron and positive charge paths, allowing for improved characterization of lithium ion battery performance and prediction of optimal carbon structures for enhanced battery performance.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 2024167659000001_ABST
    Figure 2024167659000001_ABST
Patent Text Reader

Abstract

To analyze a movement route of an electron and a movement route of a positive charge in an analysis object structure.SOLUTION: A TEM image analysis device 1 comprises: a line information generation unit 132 that extracts and quantifies lines included in TEM images of a vectorized analysis object structure; an image inversion unit 133 that performs black-and-white inversion on the vectorized TEM image; and an explanation variable extraction unit 134 that generates a plurality of explanation variables representing a characteristic of entire lines included in the quantified TEM image, and extracts explanation variables involved in objective variables indicative of performances of the analysis object structure, of a plurality of generated explanation variables. The line information generation unit 132 is configured to: extract and quantify the line included in the TEM image not subjected to the black-and-white inversion so as to make a surface of a carbon material serving as a movement route of an electron in an analysis object structure represented by an amount of characteristic; extract and quantify the line included in the TEM image subjected to the black-and-white inversion so as to make a space between two neighboring carbon materials serving as a movement route of a positive charge in the analysis object structure represented by an amount of characteristic.SELECTED DRAWING: Figure 1
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] The present invention relates to a TEM image analysis device, a TEM image analysis method, and a program. [Background technology]

[0002] Patent Document 1 describes a technology for analyzing images. In the technology described in Patent Document 1, cracks on the surface of a structure are detected by analyzing a composite image showing the surface of the structure or an image before synthesis, and crack information showing the feature amount of the crack is generated. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2022-000646 A Summary of the Invention [Problem to be solved by the invention]

[0004] However, the technique described in Patent Document 1 does not analyze the movement paths of electrons and the like moving within a structure. Therefore, by applying the technique described in Patent Document 1, it is not possible to appropriately analyze the movement paths of electrons and positive charges (e.g., lithium ions) within a structure to be analyzed (e.g., negative electrode carbon of a lithium ion battery).

[0005] In view of the above, an object of the present invention is to provide a TEM image analysis device, a TEM image analysis method, and a program capable of analyzing the electron movement path and the positive charge movement path in a structure to be analyzed. [Means for solving the problem]

[0006] (1) One aspect of the present invention is a TEM image analysis device that includes an image vectorization unit that vectorizes a TEM image of a structure to be analyzed, the structure including a carbon material and through which electrons can move; a line information generation unit that extracts and digitizes lines included in the TEM image vectorized by the image vectorization unit; an image inversion unit that inverts the TEM image vectorized by the image vectorization unit; and an explanatory variable extraction unit that generates a plurality of explanatory variables that represent characteristics of the overall lines included in the TEM image digitized by the line information generation unit and extracts explanatory variables related to a target variable that indicates performance of the structure to be analyzed from the plurality of explanatory variables generated, wherein the line information generation unit extracts and digitizes lines included in a TEM image that has not been inverted by the image inversion unit, thereby characterizing a surface of the carbon material that is a path of movement of the electrons in the structure to be analyzed, and extracts and digitizes lines included in a TEM image that has been inverted by the image inversion unit, thereby characterizing a space between two adjacent carbon materials that is a path of movement of positive charges in the structure to be analyzed that move in a direction roughly opposite to that of the electrons. Effect of the Invention

[0007] According to the present invention, it is possible to analyze the electron transfer path and the positive charge transfer path in the structure to be analyzed. [Brief description of the drawings]

[0008] [Figure 1] FIG. 1 is a diagram showing an example of a TEM image analyzer 1 according to a first embodiment. [Diagram 2] 11 is a diagram for explaining an example of processing by an image vectorization unit 131. FIG. [Diagram 3] FIG. 1 is a diagram showing an example of the distribution (shape) of data of explanatory variables (descriptors). [Figure 4] FIG. 1 is a diagram showing an example of a map showing the correlation of a plurality of explanatory variables (descriptors). [Diagram 5] FIG. 1 is a diagram showing an example of screening of explanatory variables (descriptors). [Figure 6]FIG. 1 is a diagram conceptually illustrating a carbon material (graphene sheet) of a negative electrode carbon of a lithium ion battery, the electron transfer path, and the lithium ion transfer path. [Figure 7] FIG. 13 is a diagram showing an example of the structure of anode carbon after surface decoration required to improve the performance of a lithium ion battery, as predicted by the structure prediction unit 136. [Figure 8] 4 is a flowchart for explaining an example of processing executed in the TEM image analysis apparatus 1 of the first embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0009] Hereinafter, embodiments of a TEM image analysis apparatus, a TEM image analysis method, and a program according to the present invention will be described with reference to the drawings.

[0010] First Embodiment FIG. 1 is a diagram showing an example of a TEM image analysis device 1 according to the first embodiment. In the example shown in FIG. 1, the TEM image analysis device 1 analyzes an analysis target structure, which is a structure containing a carbon material and through which electrons can move. In detail, the TEM image analysis device 1 analyzes the negative electrode carbon of a lithium ion battery as the analysis target structure. The TEM image analysis device 1 is configured by a computer and includes an input / output interface 11, a memory 12 such as a ROM, a RAM, a storage, and the like, and a processor 13 such as a CPU. The processor 13 has a function as an image vectorization unit 131, a function as a line information generation unit 132, a function as an image inversion unit 133, a function as an explanatory variable extraction unit 134, a function as a performance prediction unit 135, and a function as a structure prediction unit 136. The image vectorization unit 131 vectorizes a transmission electron microscope (TEM) image of the negative electrode carbon (structure to be analyzed) of a lithium ion battery input to the TEM image analysis device 1 via the input / output interface 11, for example, by using the technology described on the website indicated at the URL below. https: / / qiita.com / yizaki0212 / items / 11a12a06aa703b72fc12 https: / / magazine.techacademy.jp / magazine / 47393

[0011] FIG. 2 is a diagram for explaining an example of processing by the image vectorization unit 131. In detail, FIG. 2(A) shows a TEM image of negative electrode carbon of a lithium ion battery vectorized by the image vectorization unit 131. FIG. 2(B) shows an enlarged view of the rectangular area in FIG. 2(A). FIG. 2(C) shows an image in which lines included in the image shown in FIG. 2(B) are extracted by the line information generation unit 132 described later. FIG. 2(D) shows an image in which the TEM image shown in FIG. 2(A) is inverted in black and white by the image inversion unit 133 described later. FIG. 2(E) shows an enlarged view of the rectangular area in FIG. 2(D). FIG. 2(F) shows an image in which lines included in the image shown in FIG. 2(E) are extracted by the line information generation unit 132.

[0012] In the example shown in FIG. 1, the line information generation unit 132 extracts lines included in the TEM image vectorized by the image vectorization unit 131 (see FIGS. 2(A) and 2(B)), as shown in FIG. 2(C), by using OpenCV, for example, as in the technique described on the website indicated by the following URL: https: / / shikaku-mafia.com / cv2-houghlinesp / Furthermore, the line information generating unit 132 digitizes the extracted lines (see FIG. 2(C)) (specifically, a plurality of straight lines). Specifically, the line information generating unit 132 calculates the start point, end point, angle, and the like of each of the extracted multiple straight lines in order to express the extracted multiple straight lines as numerical values. The image inversion unit 133 inverts the TEM image (see FIG. 2(A)) vectorized by the image vectorization unit 131 into black and white, as shown in FIG. 2(D). The line information generation unit 132 extracts lines included in the TEM image (see FIG. 2(D) and FIG. 2(E)) inverted into black and white by the image inversion unit 133, as shown in FIG. 2(F).

[0013] The explanatory variable extraction unit 134 generates a plurality of explanatory variables (descriptors) that represent the characteristics of the entire lines contained in the TEM image digitized by the line information generation unit 132. The multiple explanatory variables (descriptors) generated by the explanatory variable extraction unit 134 include, for example, "AllAngle", which means a descriptor obtained by averaging the angle values ​​obtained from all vectors in the TEM overall image, "AllLengthDist", which means a descriptor obtained by averaging the length values ​​obtained from all vectors in the TEM overall image, "AllLengthPa50", which means a descriptor obtained by taking the median of the angle values ​​obtained from all vectors in the TEM overall image, "L_medi_length", which means a descriptor obtained by taking the median of the length values ​​obtained from all vectors in the TEM enlarged view, "LR_medi_length", which means a descriptor obtained by taking the median of the length values ​​obtained from all vectors in the black and white inverted view of the TEM overall image, "L_medi_Angle", which means a descriptor obtained by taking the median of the angle values ​​obtained from all vectors in the black and white inverted view of the TEM overall image, "LR_medi_Angle", which means a descriptor obtained by taking the median of the angle values ​​obtained from all vectors in the black and white inverted view of the TEM overall image, and a line extracted from the TEM enlarged view. "H_medi_length" is a descriptor obtained by taking the median of the lengths of all vectors in the extracted image; "HR_medi_length" is a descriptor obtained by taking the median of the lengths of all vectors in the inverted line extraction image of lines extracted from a TEM enlarged image (line extraction image of lines extracted from an inverted TEM enlarged image); "ratio" is a descriptor that means the ratio of the lengths of the long and short sides of the migration path surface of the positive charge in the TEM image; "RBr" is a descriptor that means the number of intersections between extracted lines in the inverted TEM image; "br" is a descriptor that means the number of intersections between extracted lines in the TEM image; "H_medi_Angle" is a descriptor that means the median of the angles of all vectors in the line extraction image from the enlarged TEM image; "HR_medi_Angle" is a descriptor that means the median of the angles of all vectors in the inverted line extraction image of lines extracted from a TEM enlarged image; "L" is a descriptor that means the length of the long side of the migration path surface in the TEM image; "S" is a descriptor that means the length of the short side of the migration path surface in the TEM image;These include "RL", which means the length of the long side of the moving path surface in the black and white inversion of the TEM image, "RS", which means the length of the short side of the moving path surface in the black and white inversion of the TEM image, and "Rratio", which means the ratio of the length of the long side to the short side of the moving path surface in the black and white inversion of the TEM image. Furthermore, the explanatory variable extraction unit 134 extracts explanatory variables (e.g., "AllLengthDist") that are (highly) related to the objective variable indicating the performance (e.g., charge capacity (Capacity), charge transfer resistance Rct, etc.) of the lithium ion battery including the analysis target structure (negative electrode carbon) from the above-mentioned generated explanatory variables (descriptors).

[0014] FIG. 3 is a diagram showing an example of the distribution (shape) of data for explanatory variables (descriptors) "AllAngle," "AllLengthDist," "AllLengthPa50," "L_medi_length," "LR_medi_length," "L_medi_Angle," "LR_medi_Angle," "H_medi_length," and "HR_medi_length." The explanatory variable extraction unit 134 checks the distribution (shape) of explanatory variable (descriptor) data as shown in Fig. 3, and if the distribution of explanatory variable data is concentrated or the explanatory variable data contains extremely abnormal values, it determines that it is necessary to re-extract the explanatory variable data (for example, by using a part other than the part enclosed by the rectangle in the TEM image shown in Fig. 2(A)). As a result, explanatory variables suitable for analyzing the structure to be analyzed (negative electrode carbon of a lithium ion battery) are extracted. Furthermore, the explanatory variable extraction unit 134 performs dimensionality reduction by using, for example, the technology described in the website indicated by the URL below, and extracts explanatory variables that are (highly) related to the objective variable indicating the performance of the structure to be analyzed from among the multiple explanatory variables (descriptors). https: / / www.jmp.com / support / help / en / 16.2 / index.shtml#page / jmp / cluster-variables.shtml

[0015] Figure 4 shows an example of a map showing the correlation of multiple explanatory variables (descriptors). In Figure 4, negative values ​​on the vertical axis in the upper diagram of Figure 4 indicate that the explanatory variables are not similar (not overlapping), and positive values ​​on the vertical axis in the upper diagram of Figure 4 indicate that the explanatory variables are similar (overlapping). The lower diagram of Figure 4 shows "cluster," "number of members," "most representative explanatory variable," "proportion to intra-cluster variation," and "proportion to total variation." The explanatory variable extraction unit 134 searches for explanatory variables (descriptors) having similar (similar) data shapes through clustering of explanatory variables (descriptors) and groups the explanatory variables (descriptors), as shown in Fig. 4. In addition, the explanatory variable extraction unit 134 prevents distortion of weights that may occur in calculations by removing or combining duplicated (similar) explanatory variables (descriptors). Furthermore, the explanatory variable extraction unit 134 performs screening of explanatory variables (descriptors) by using, for example, the technology described in the website indicated by the following URL. https: / / www.jmp.com / support / help / en / 16.2 / index.shtml#page / jmp / example-of-predictor-screening.shtml#

[0016] Fig. 5 is a diagram showing an example of screening of explanatory variables (descriptors). In the example shown in Fig. 5, the explanatory variable extraction unit 134 ranks explanatory variables involved (contributing) to the charge capacity (Capacity) (objective variable) of a lithium ion battery including an analysis target structure (negative electrode carbon) and ranks explanatory variables involved (contributing) to the charge transfer resistance Rct (objective variable) of a lithium ion battery including an analysis target structure (negative electrode carbon). In detail, the explanatory variable extraction unit 134 calculates the contribution degree (Contribution) and the contribution ratio (Portion) of each of a plurality of explanatory variables to each objective variable. The explanatory variable extraction unit 134 uses a random forest (bootstrap forest) to calculate the contribution of each explanatory variable to the objective variable and screens the explanatory variables, as shown in Fig. 5. That is, the explanatory variable extraction unit 134 selects explanatory variables that contribute to the performance of the lithium ion battery before the performance prediction unit 135, which will be described later, predicts the performance of the lithium ion battery (e.g., charge capacity, charge transfer resistance Rct, etc.) using a neural network or the like.

[0017] Through intensive research, the present inventors have found that electrons in the structure to be analyzed (negative electrode carbon of a lithium ion battery) move along the surface of the carbon material of the negative electrode carbon, and that positive charges (lithium ions) moving in a direction roughly opposite to that of the electrons in the structure to be analyzed (negative electrode carbon of a lithium ion battery) move within the space between two adjacent carbon materials (see FIG. 6 ). FIG. 6 is a conceptual diagram showing the carbon material (graphene sheet) of the negative electrode carbon of a lithium ion battery, the electron transfer path, and the lithium ion transfer path. Furthermore, the present inventors discovered that in a TEM image of the structure to be analyzed (negative electrode carbon of a lithium ion battery), the surface of the carbon material of the negative electrode carbon appears black, and the space between two adjacent carbon materials appears white. Furthermore, the inventors have discovered that the electron movement path can be characterized by extracting and analyzing the lines contained in the black parts of the TEM image of the structure to be analyzed (negative electrode carbon of a lithium ion battery), and that the movement path of positive charges (lithium ions) can be characterized by extracting and analyzing the lines contained in the black parts after inverting the TEM image of the structure to be analyzed (negative electrode carbon of a lithium ion battery).

[0018] In view of the above, in the example shown in FIG. 1, the line information generating unit 132 extracts and digitizes lines contained in a TEM image of the structure to be analyzed (negative electrode carbon of a lithium ion battery) that has not been subjected to black-and-white inversion by the image inverting unit 133, thereby characterizing the surface of the carbon material, which is the path of electron movement in the structure to be analyzed. Furthermore, the line information generating unit 132 extracts and digitizes lines contained in the TEM image of the structure to be analyzed that has been subjected to black-white inversion by the image inversion unit 133, thereby characterizing the space between two adjacent carbon materials, which is the movement path of positive charges (lithium ions) in the structure to be analyzed that move in roughly the opposite direction to the electrons. Therefore, in the example shown in FIG. 1, it is possible to analyze the electron transfer path and the positive charge (lithium ion) transfer path in the structure to be analyzed.

[0019] 1, the performance prediction unit 135 predicts the performance (e.g., charging capacity, charge transfer resistance, etc.) of a lithium ion battery including the negative electrode carbon (analysis target structure). In detail, the performance prediction unit 135 predicts the performance (e.g., charging capacity, charge transfer resistance Rct, etc.) of a lithium ion battery including the analysis target structure (negative electrode carbon) based on a TEM image of the negative electrode carbon (analysis target structure) captured by a transmission electron microscope by using a first model in which a first learning was performed using first teacher data that is a set of explanatory variables extracted from a plurality of explanatory variables that represent the overall line characteristics included in a TEM image of the negative electrode carbon of a first learning lithium ion battery for which battery performance verification has actually been performed, and performance obtained by actual measurement, etc. of the first learning lithium ion battery for which battery performance verification has actually been performed. The structure prediction unit 136 predicts the structure of the negative electrode carbon (analysis target structure) of the lithium ion battery. In detail, the structure prediction unit 136 predicts the structure of the negative electrode carbon (analysis target structure) after the surface decoration necessary for improving the performance of the lithium ion battery including the negative electrode carbon (analysis target structure) is performed (optimal carbon structure that can produce the best battery performance) based on the TEM image of the negative electrode carbon (analysis target structure) captured by a transmission electron microscope by using a second model in which second learning is performed using second teacher data that is a set of the structure of the second learning negative electrode carbon (actually produced sample carbon structure) whose surface is decorated using solution plasma described in the WEB site indicated by the following URL, and the performance (e.g., charge capacity, charge transfer resistance Rct, etc.) (obtained by actual measurement or simulation) of the second learning lithium ion battery including the second learning negative electrode carbon whose surface is decorated. https: / / www.engg.nagoya-u.ac.jp / techno / techno2016 / pdf / booth / no05.pdf https: / / www.jstage.jst.go.jp / article / sfj / 58 / 12 / 58_12_810 / _pdf / -char / ja https: / / www.jstage.jst.go.jp / article / sfj / 68 / 3 / 68_147 / _pdf / -char / ja https: / / www.chuden.co.jp / resource / seicho_kaihatsu / kaihatsu / kai_library / news / news_2020 / news_145_01.pdf

[0020] FIG. 7 is a diagram showing an example of the structure of the negative electrode carbon (analysis target structure) after the surface decoration required for improving the performance of the lithium ion battery predicted by the structure prediction unit 136 is performed. The inventors have confirmed that the carbon structure in which graphite is coated with amorphous carbon (carbon structure that can be made using solution plasma) as shown in FIG. 7(A) can reduce the charge transfer resistance Rct of a lithium ion battery including the negative electrode carbon having the carbon structure and improve the charge capacity. The structure prediction unit 136 can predict an optimal carbon structure that can provide the best battery performance by using a second model in which a second learning is performed using a combination of the carbon structure as shown in FIG. 7(A) and the performance of a lithium ion battery including the negative electrode carbon having the carbon structure as second teacher data. In addition, the optimal carbon structure that can provide the best battery performance predicted by the structure prediction unit 136 can be made by using solution plasma.

[0021] The present inventors performed function fitting under the following conditions. - Forecasting method used: Neural network Hidden layer: Sigmoid Tangent Hyperbolic (TanH(3)) Number of boosting models: 20 Learning rate: 0.1 Number of tours: 100 The prediction accuracy of the charge capacity of a lithium-ion battery using the explanatory variable (descriptor 2) extracted by the explanatory variable extraction unit 134 and the prediction accuracy of the charge capacity of a lithium-ion battery using the explanatory variable (descriptor 2) extracted by the explanatory variable extraction unit 134 using explanatory variables (descriptor 1, descriptor 3) different from the explanatory variable (descriptor 2) extracted by the explanatory variable extraction unit 134 (average value: 0.24954136, maximum value: 0.3075813, minimum value: 0.2087368) were higher than the prediction accuracy of the charge capacity of a lithium-ion battery using descriptor 1 (average value: 0.17975116, maximum value: 0.265202, minimum value: 0.1110717) and the prediction accuracy of the charge capacity of a lithium-ion battery using descriptor 3 (average value: 0.22153996, maximum value: 0.2498841, minimum value: 0.162449).

[0022] FIG. 8 is a flowchart for explaining an example of processing executed in the TEM image analysis apparatus 1 of the first embodiment. In the example shown in FIG. 8, in step S10, the image vectorization unit 131 vectorizes a TEM image of the negative electrode carbon (structure to be analyzed) of a lithium ion battery. In step S11, the line information generating unit 132 extracts lines included in the TEM image vectorized in step S10. The line information generating unit 132 also digitizes the extracted lines. In step S12, a plurality of explanatory variables (descriptors) are generated that represent the characteristics of the entire lines included in the TEM image digitized in step S11. Furthermore, the explanatory variable extraction unit 134 extracts explanatory variables that are (highly) related to the objective variable that represents the performance of the lithium-ion battery including the analysis target structure (negative electrode carbon) from the plurality of explanatory variables (descriptors) generated (specifically, for example, confirmation of the data distribution of the explanatory variables (descriptors), dimensionality reduction, screening of the explanatory variables (descriptors), etc. are performed).

[0023] In step S13, the image inverting unit 133 inverts the TEM image vectorized in step S10. In step S14, the line information generating unit 132 extracts lines included in the TEM image that has been inverted in step S13. The line information generating unit 132 also digitizes the extracted lines. In step S15, a plurality of explanatory variables (descriptors) are generated that represent the characteristics of the entire lines included in the TEM image digitized in step S14. Furthermore, the explanatory variable extraction unit 134 extracts explanatory variables that are (highly) related to the objective variable that represents the performance of the lithium-ion battery including the analysis target structure (negative electrode carbon) from the plurality of explanatory variables (descriptors) generated (specifically, for example, confirmation of the data distribution of the explanatory variables (descriptors), dimensionality reduction, screening of the explanatory variables (descriptors), etc. are performed).

[0024] <Second embodiment> A second embodiment of the TEM image analysis device, the TEM image analysis method, and the program of the present invention will be described below. The TEM image analysis device 1 of the second embodiment has the same configuration as the TEM image analysis device 1 of the first embodiment described above, except for the points described below.

[0025] As described above, the TEM image analyzer 1 of the first embodiment analyzes the negative electrode carbon of a lithium ion battery as the structure to be analyzed, whereas the TEM image analyzer 1 of the second embodiment analyzes the electrode material containing a carbon material of a fuel cell. 1 (an example of the TEM image analysis device 1 of the first embodiment), the image vectorization unit 131 vectorizes a TEM image of the negative electrode carbon (structure to be analyzed) of a lithium ion battery. Meanwhile, in an example of the TEM image analysis device 1 of the second embodiment, the image vectorization unit 131 vectorizes a TEM image of the carbon material (structure to be analyzed) of the electrode material of a fuel cell. 1, the explanatory variable extraction unit 134 extracts explanatory variables that are (highly) related to the objective variable indicating the performance of a lithium ion battery including the structure to be analyzed (negative electrode carbon). On the other hand, in one example of the TEM image analysis device 1 of the second embodiment, the explanatory variable extraction unit 134 extracts explanatory variables that are (highly) related to the objective variable indicating the performance of a fuel cell including the structure to be analyzed (carbon material of the electrode material). 1, the explanatory variable extraction unit 134 selects explanatory variables that contribute to the performance of the lithium ion battery before the performance prediction unit 135 predicts the performance of the lithium ion battery. On the other hand, in one example of the TEM image analysis device 1 of the second embodiment, the explanatory variable extraction unit 134 selects explanatory variables that contribute to the performance of the fuel cell before the performance prediction unit 135 predicts the performance of the fuel cell. 1, the line information generating unit 132 extracts and digitizes lines included in a TEM image of the structure to be analyzed (negative electrode carbon of a lithium ion battery) that has not been subjected to black-white inversion by the image inversion unit 133, thereby characterizing the surface of the carbon material that is the electron movement path in the structure to be analyzed. On the other hand, in one example of the TEM image analysis device 1 of the second embodiment, the line information generating unit 132 extracts and digitizes lines included in a TEM image of the structure to be analyzed (carbon material of an electrode material of a fuel cell) that has not been subjected to black-white inversion by the image inversion unit 133, thereby characterizing the surface of the carbon material that is the electron movement path in the structure to be analyzed. 1, the line information generating unit 132 extracts and digitizes lines included in the TEM image of the structure to be analyzed that has been subjected to black-white inversion by the image inversion unit 133, thereby characterizing the space between two adjacent carbon materials, which is a migration path of positive charges (lithium ions) in the structure to be analyzed that move in a direction generally opposite to that of electrons. Meanwhile, in one example of the TEM image analyzing device 1 of the second embodiment, the line information generating unit 132 extracts and digitizes lines included in the TEM image of the structure to be analyzed (carbon material of the electrode material of a fuel cell) that has been subjected to black-white inversion by the image inversion unit 133, thereby characterizing the space between two adjacent carbon materials, which is a migration path of positive charges (protons) in the structure to be analyzed (carbon material of the electrode material of a fuel cell).

[0026] <Third embodiment> Hereinafter, a third embodiment of the TEM image analysis device, the TEM image analysis method, and the program of the present invention will be described. The TEM image analysis device 1 of the third embodiment has the same configuration as the TEM image analysis device 1 of the first embodiment described above, except for the points described below.

[0027] The TEM image analyzer 1 of the third embodiment analyzes a transparent conductive film containing a carbon material. In one example of the TEM image analysis device 1 of the third embodiment, the image vectorization unit 131 vectorizes a TEM image of the carbon material (analysis target structure) of the transparent conductive film. The explanatory variable extraction unit 134 extracts explanatory variables that are (highly) related to the objective variable indicating the performance of the transparent conductive film including the analysis target structure (carbon material). The explanatory variable extraction unit 134 selects explanatory variables that contribute to the performance of the transparent conductive film. The line information generation unit 132 extracts and digitizes lines included in the TEM image of the analysis target structure (carbon material of the transparent conductive film) that has not been subjected to black-and-white inversion by the image inversion unit 133, thereby characterizing the surface of the carbon material, which is the migration path of electrons in the analysis target structure. In addition, in one example of the TEM image analysis device 1 of the third embodiment, the line information generating unit 132 extracts and digitizes lines contained in the TEM image of the structure to be analyzed (the carbon material of the transparent conductive film) that has been black-white inverted by the image inversion unit 133, thereby characterizing the space between two adjacent carbon materials, which is the movement path of positive charges (protons) in the structure to be analyzed (the carbon material of the transparent conductive film) that move in approximately the opposite direction to electrons.

[0028] <Fourth embodiment> A fourth embodiment of the TEM image analysis device, the TEM image analysis method, and the program of the present invention will be described below. The TEM image analysis device 1 of the fourth embodiment has the same configuration as the TEM image analysis device 1 of the first embodiment described above, except for the points described below.

[0029] The TEM image analyzer 1 of the fourth embodiment analyzes structures that are different from those analyzed in the first to third embodiments, such as a proton exchange membrane containing a carbon material of a water electrolysis device. In one example of the TEM image analysis device 1 of the fourth embodiment, the image vectorization unit 131 vectorizes a TEM image of the structure to be analyzed (e.g., a carbon material contained in a proton exchange membrane of a water electrolysis device). The explanatory variable extraction unit 134 extracts explanatory variables that are (highly) related to a target variable that indicates the performance of a transparent conductive film including the structure to be analyzed. The explanatory variable extraction unit 134 selects explanatory variables that contribute to the performance of the proton exchange membrane of a water electrolysis device, for example. The line information generation unit 132 extracts and digitizes lines included in the TEM image of the structure to be analyzed (e.g., a carbon material contained in a proton exchange membrane of a water electrolysis device) that has not been subjected to black-and-white inversion by the image inversion unit 133, thereby characterizing the surface of the carbon material, which is a path of electron movement in the structure to be analyzed. In addition, in one example of the TEM image analysis device 1 of the fourth embodiment, the line information generating unit 132 extracts and digitizes lines contained in a TEM image of a structure to be analyzed (such as a carbon material contained in a proton exchange membrane of a water electrolysis device) that has been black-white inverted by the image inversion unit 133, thereby characterizing the space between two adjacent carbon materials, which is the movement path of positive charges (protons) in the structure to be analyzed that move in approximately the opposite direction to electrons.

[0030] As described above, the embodiments of the TEM image analysis device, the TEM image analysis method, and the program of the present invention have been described with reference to the drawings, but the TEM image analysis device, the TEM image analysis method, and the program of the present invention are not limited to the above-mentioned embodiments, and appropriate modifications can be made without departing from the spirit of the present invention. The configurations of the examples of the above-mentioned embodiments may be appropriately combined. [Explanation of symbols]

[0031] 1...TEM image analysis device, 11...input / output interface, 12...memory, 13...processor, 131...image vectorization unit, 132...line information generation unit, 133...image inversion unit, 134...explanatory variable extraction unit, 135...performance prediction unit, 136...structure prediction unit

Claims

1. an image vectorization unit that vectorizes a transmission electron microscope (TEM) image of an analysis target structure, the structure including a carbon material and capable of moving electrons; a line information generating unit that extracts and digitizes lines included in the TEM image vectorized by the image vectorizing unit; an image inversion unit that inverts the TEM image vectorized by the image vectorization unit; an explanatory variable extraction unit that generates a plurality of explanatory variables that represent characteristics of an entire line included in the TEM image digitized by the line information generation unit, and extracts explanatory variables related to a target variable that represents a performance of the analysis target structure from the plurality of generated explanatory variables; The line information generating unit A line included in the TEM image that has not been subjected to black-and-white inversion by the image inversion unit is extracted and digitized, thereby characterizing the surface of the carbon material, which is the movement path of the electrons in the structure to be analyzed; A TEM image analysis device that extracts and digitizes lines contained in a TEM image that has been black-and-white inverted by the image inversion unit, thereby characterizing the space between two adjacent carbon materials, which is the path of movement of a positive charge in the structure to be analyzed that moves in approximately the opposite direction to the electron.

2. 2. The TEM image analysis apparatus according to claim 1, wherein the structure to be analyzed is a negative electrode carbon of a lithium ion battery.

3. a performance prediction unit that predicts the performance of the lithium ion battery including the negative electrode carbon as the analysis target structure; a structure prediction unit that predicts a structure of the negative electrode carbon as the analysis target structure, The performance prediction unit is a first model in which a first learning was performed using first teacher data that is a set of explanatory variables extracted from a plurality of explanatory variables that represent the overall line characteristics included in a TEM image of the negative electrode carbon of a first learning lithium ion battery and the performance of the first learning lithium ion battery; predicting performance of the lithium ion battery including the negative electrode carbon as the structure to be analyzed based on a TEM image of the negative electrode carbon as the structure to be analyzed captured by a transmission electron microscope; The structure prediction unit By using a second model in which a second learning is performed using second teacher data that is a pair of a structure of a second learning negative-electrode carbon whose surface has been decorated using solution plasma and the performance of a second learning lithium-ion battery including the second learning negative-electrode carbon whose surface has been decorated, 3. The TEM image analysis device according to claim 2, which predicts a structure of the negative electrode carbon as the structure to be analyzed after surface decoration required to improve the performance of the lithium ion battery including the negative electrode carbon as the structure to be analyzed has been performed, based on a TEM image of the negative electrode carbon as the structure to be analyzed taken by the transmission electron microscope.

4. an image vectorization step in which the TEM image analysis device vectorizes a TEM image of an analysis target structure, the analysis target structure being a structure that contains a carbon material and in which electrons can move; a line information generating step in which the TEM image analysis device extracts and digitizes lines included in the TEM image vectorized in the image vectorization step; an image inversion step in which the TEM image analysis device inverts the TEM image vectorized in the image vectorization step; the TEM image analysis device generates a plurality of explanatory variables representing characteristics of an entire line included in the TEM image digitized in the line information generating step, and extracts explanatory variables related to a target variable indicating a performance of the analysis target structure from the plurality of explanatory variables generated; In the line information generating step, A line included in a TEM image that has not been subjected to black-and-white inversion in the image inversion step is extracted and digitized, thereby characterizing a surface of the carbon material, which is a path of movement of the electrons in the structure to be analyzed; A TEM image analysis method in which lines contained in the TEM image that has been black-and-white inverted in the image inversion step are extracted and quantified, thereby characterizing the space between two adjacent carbon materials, which is the path of movement of a positive charge in the structure to be analyzed that moves in approximately the opposite direction to the electrons.

5. The processor: an image vectorization step of vectorizing a TEM image of an analysis target structure, the structure including a carbon material and capable of moving electrons; a line information generating step of extracting and digitizing lines included in the TEM image vectorized in the image vectorizing step; an image inversion step of inverting the black and white of the TEM image vectorized in the image vectorization step; a program for executing an explanatory variable extraction step of generating a plurality of explanatory variables representing characteristics of an entire line included in the TEM image digitized in the line information generating step, and extracting an explanatory variable related to a target variable indicating a performance of the analysis target structure from the plurality of generated explanatory variables, In the line information generating step, A line included in a TEM image that has not been subjected to black-and-white inversion in the image inversion step is extracted and digitized, thereby characterizing a surface of the carbon material, which is a path of movement of the electrons in the structure to be analyzed; A program for characterizing the space between two adjacent carbon materials, which is a path for the movement of a positive charge in the structure to be analyzed that moves in roughly the opposite direction to the electron, by extracting and digitizing lines contained in the TEM image that has been black-and-white inverted in the image inversion step.

Citation Information

Patent Citations

  • Device and method for image processing

    JP2001060247A

  • Device and method for inspecting metallic worked surface

    JP2002245436A

  • Electrode material, paste for electrodes and lithium ion battery

    JP2016012458A

  • Battery capacity estimation method, and battery capacity estimating device

    JP2020101400A

  • Electrode material, paste for electrodes, and lithium ion battery

    US20150380725A1