Analysis method, program, and computing device

Abstract

To provide a method which analyzes substances contained in a sample and which is suitable for analyzing a substance whose composition has fluctuation.SOLUTION: A method for analyzing substances contained in a sample includes steps of: acquiring elemental map data from SEM-EDX analysis relative to the sample; extracting a particle part from the elemental map data; deriving a representative composition on the basis of density-based clustering in the particle part; and classifying the substances contained in the particle by comparing the representative composition with a reference list.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present disclosure relates to an analytical method, a program, and a computing device. More specifically, the present disclosure relates to an analytical method, a program, and a computing device for analyzing a substance contained in a sample. [Background technology]

[0002] The Mineral Liberation Analyzer (MLA) is a device that analyzes the types, content ratios, particle sizes, and bonding states of minerals contained in ore. In addition to minerals, this device is also used to analyze substances contained in recycled materials such as electrical and electronic equipment.

[0003] Patent Document 1 discloses a method for evaluating a sample-embedding resin having a plurality of types of granular samples to be analyzed and a resin material in which the granular samples are embedded and fixed.

[0004] Patent Document 2 discloses a method for processing data obtained by an analytical device. Specifically, Patent Document 2 discloses the following means for data processing: discrimination means 12 that uses a predetermined function to discriminate whether the mineral particles are simple ores of useful ores or unused ores, or combined ores formed by combining the useful ores with the unused ores; data compilation means 13 that compiles the size distribution of the mineral particles in the ore sample into a list format in which the simple ores and combined ores can be distinguished; and data output means 14 that outputs data on the distribution in the list format. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Japanese Patent Application Publication No. 2020-190497 [Patent Document 2] Japanese Patent Application Publication No. 2016-017816 Summary of the Invention

Problems to be Solved by the Invention

[0006] Ideally, minerals and the like have a certain chemical composition. However, in reality, the compositions of some minerals have fluctuations. This is because solid solutions exist. And the compositional fluctuations have been an obstacle in the analysis of substances. Therefore, an object of the present disclosure is to provide a method suitable for analyzing substances with compositional fluctuations in a method for analyzing substances contained in a sample.

Means for Solving the Problems

[0007] The present disclosure includes the following inventions in one aspect. (Invention 1) A method for analyzing substances contained in a sample, · a step of obtaining elemental map data from SEM-EDX analysis of the sample; · a step of extracting particle portions from the elemental map data; · a step of deriving a representative composition based on density-based clustering in the particle portions; · a step of classifying substances contained in the particles by comparing the representative composition with a reference list; and a method including the above steps. (Invention 2) The method according to Invention 1, wherein the density-based clustering is any one or more of DBSCAN and OPTICS. (Invention 3) The method according to Invention 1 or 2, wherein the step of extracting the particle portions includes segmenting the particle portions, and the step of deriving the representative composition includes extracting a portion with high density as a first region corresponding to at least a part of the principal component region based on density-based clustering, and deriving a representative composition of the principal component region. A method. (Invention 4) The method of Invention 3, wherein the segment division is based on any one or more of the SLIC method and the Quick Shift method. (Invention 5) The method according to Invention 3 or 4, comprising: · a step of extracting a region other than the first region as a second region; · a step of classifying at least a part of the second region by comparing it with a second reference list; A method comprising the above steps. (Invention 6) The method of Invention 5, further comprising a step of deriving a weight ratio of the substances specified by the classification of the first region and the second region. (Invention 7) The method according to any one of Inventions 1 to 6, wherein the sample is a solid solution type ore. (Invention 8) A program for executing the method according to any one of Inventions 1 to 7. (Invention 9) A computing device installed with the program of Invention 8.

Advantages of the Invention

[0008] On one hand, the above invention has a step of deriving a representative composition based on density-based clustering. Thereby, even if there is a fluctuation in the composition, the substance can be specified using the representative composition.

Brief Description of the Drawings

[0009]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Figure 14

Figure 15

Figure 16

Figure 17

Figure 18

Best Mode for Carrying Out the Invention

[0010] Hereinafter, specific embodiments for carrying out the invention will be described. The following description is for promoting the understanding of the invention. That is, it is not intended to limit the scope of the present invention.

[0011] 1. Definition Explanation will be given for various terms used in this specification.

[0012] The term "SEM-EDX" used in this specification is an analytical method combining SEM (Scanning Electron Microscope) and EDX (Energy Dispersive X-ray Spectroscopy, also referred to as EDS). EDX is a method for performing elemental analysis and composition analysis by detecting characteristic X-rays generated by electron beam irradiation and spectroscopically analyzing them by energy. Since secondary electron images and backscattered electron images can be obtained by EDX, SEM images can also be obtained. Note that the terms "EDX" and "EDS" can be used interchangeably in this specification.

[0013] The term "element mapping" used in this specification refers to element mapping obtained by EDX. For example, element mapping by EDX can be obtained by the following procedure as an example. · Two-dimensionally scan the field-of-view area of the analysis target with an electron beam. · Acquire the generated characteristic X-ray spectra for each pixel in the field-of-view area. · Calculate the X-ray intensity and / or element concentration of the elements to be analyzed from the acquired spectra. · Express element mapping within each pixel depending on the calculated X-ray intensity and / or element concentration (for example, express it in different colors within each pixel depending on the calculated X-ray intensity and / or element concentration).

[0014] The term "clustering process" used in this specification is a processing method corresponding to unsupervised learning. Therefore, a set of similar data is created without prior provision of information regarding the method of classification (teacher data). In this regard, it is distinguished from the "classification" process, which is a process of determining to which group a data belongs while referring to correct data (or reference data). The term "clustering process" used in this specification includes non-hierarchical clustering and hierarchical clustering. In one embodiment, the term "clustering process" used in this specification is non-hierarchical clustering. Also, the term "clustering process" used in this specification includes hard clustering and soft clustering. In one embodiment, the term "clustering process" used in this specification is hard clustering.

[0015] In one embodiment, the term "clustering process" used in this specification may be executed on a plurality of elemental composition data. Elemental composition data is data that includes at least an element and the content of the element. An example of elemental composition data is shown in FIG. 1. The content may be expressed as weight % (WT%) as shown in FIG. 1, or may be expressed in another form (for example, atomic %).

[0016] The term "segmentation" used in this specification belongs to image processing. Specifically, the term "segmentation" refers to a process of determining that a plurality of pixels similar to each other belong to one group while comparing the pixels with each other.

[0017] The term "density-based clustering" used in this specification is a method of determining clusters by paying attention to the density around each data point. The term "density-based clustering" used in this specification is also referred to as "density-based spatial clustering of applications with noise (DBSCAN)".

[0018] The term "DBSCAN" (Density-Based Spatial Clustering of Applications with Noise) used in this specification is a type of density-based clustering. In DBSCAN, each data point is determined to be either a core point, a reachable point, or an outlier. If there are N or more data points within a pre-determined distance X from a certain data point, that data point is judged as a core point. If a certain data point is within the pre-determined distance X from a core point, that data point is judged as a reachable point. A point that does not correspond to either a core point or a reachable point is judged as an outlier. As initial parameters, the distance X and the number N are set.

[0019] The term "OPTICS" (Ordering Points To Identify the Clustering Structure) used in this specification is an improved version of the above-mentioned DBSCAN. The value of the above-mentioned distance X is relaxed from a single value to a range of values.

[0020] 2. Execution Environment In one embodiment, the present disclosure relates to a method for analyzing substances contained in a sample. And the method is executable by a computer program (computer software). In another embodiment, the present disclosure relates to the program, the medium storing the program, the device including the program, and the method using the program.

[0021] The environment for executing the program and the method is not particularly limited, and a typical information processing device (or also referred to as a computing device) can be used. The information processing device (100) can typically include a processor (110), a memory (120), a non-transitory storage medium (130), and a communication module (140) as shown in Figure 2.

[0022] Examples of the information processing device (100) include, but are not limited to, a server, a personal computer, a tablet terminal, a smartphone, a smart watch, smart glasses, etc.

[0023] The program is stored in a non-transitory storage medium (130, e.g., HDD, SSD, etc.), loaded into a memory (120, e.g., RAM, etc.) as needed, and executed by a processor (110, e.g., CPU, etc.). If necessary, the program can connect to a network through a communication module (140) to send and receive information.

[0024] In one embodiment, the program may be installed as application software on one information processing device (100) and executed by the information processing device (100).

[0025] In another embodiment, the number of information processing devices 100 is not limited to one, and multiple information processing devices 100 may be used as needed. In this case, the functions of the program may be distributed among the multiple information processing devices 100.

[0026] Alternatively, as shown in Fig. 3, a system (200) may be configured in which a server (210) and a terminal (220) are interconnected via a network. In the system (200), the terminal (220) may receive input from a user and transmit at least a portion of the received input to the server (210). The server (210) may receive input information transmitted from the terminal (220), process the information, and transmit a portion of the output to the terminal (220). The terminal (220) may then receive output information transmitted from the server (210) and display it on the terminal (220).

[0027] Accordingly, in another aspect, the present disclosure also relates to an information processing apparatus including the program of the present disclosure, and a system including the information processing apparatus. In yet another aspect, the present disclosure relates to a terminal and / or a server constituting the system of the present disclosure. The internal configurations of the terminal and the server may be the same as those of the information processing apparatus shown in FIG. 2. In yet another aspect, the present disclosure relates to a storage medium (e.g., a non-transitory storage medium, e.g., a computer-readable non-transitory storage medium, e.g., HDD, SSD, flash memory, optical disk, etc.) storing a program.

[0028] 3. Outline of the Method In one embodiment, the present disclosure relates to a method for analyzing substances contained in a sample. The method at least includes the following steps (FIG. 4). · A step of obtaining elemental map data from SEM-EDX analysis of the sample · A step of extracting particle portions from the elemental map data · A step of deriving a representative composition based on density-based clustering in the particle portions · A step of classifying the substances contained in the particles by comparing the representative composition with a reference list

[0029] In a further embodiment, the step of extracting particle portions may include segmenting the particle portions. Also, in a further embodiment, the step of deriving a representative composition may include extracting, based on density-based clustering, the high-density portions as a first region corresponding to at least a part of the principal component region, and deriving the representative composition of the principal component region.

[0030] And, in a further embodiment, the method shown in FIG. 4 may further include the following steps (FIG. 5). · A step of extracting the region other than the first region as a second region · A step of classifying at least a part of the second region by comparing it with a second reference list

[0031] In a further embodiment, the method shown in FIG. 5 may further include the following steps (FIG. 6). · Deriving the weight ratio of the substances identified by the classification of the first region and the second region

[0032] Hereinafter, each step will be described in detail.

[0033] 3-1. Step of obtaining elemental map data from SEM-EDX analysis of a sample In the step of obtaining elemental map data from SEM-EDX analysis of a sample, it can include performing SEM-EDX analysis on the prepared sample and obtaining elemental map data.

[0034] The method for preparing the sample is not particularly limited. Typically, the substance to be analyzed may be pulverized so as to be adjusted to particles of a size suitable for analysis. Thereafter, the substance processed into a particle shape may be embedded in a resin.

[0035] After preparing the sample, SEM-EDX analysis can be performed on the sample. Specifically, an electron beam can be irradiated, and then characteristic X-rays can be detected and an electron image can be obtained.

[0036] Thereafter, based on the detected characteristic X-rays, elemental map data is created. The elemental map data includes position information and elemental composition data (or data similar thereto). The position information is information for identifying a part of the observation field of view. Although not limited, for example, when the observation field of view is divided into 100×100, the position information may be represented by a combination of the value of the x coordinate and the value of the y coordinate. The elemental composition data is data that includes at least an element and the content of the element as described above. Examples of data similar to the elemental composition data include, for example, characteristic X-ray spectra.

[0037] 3-2. Step of extracting particle portions from the elemental map data For the elemental map data obtained in the above process, a process of extracting the particle portion is performed. In SEM-EDX analysis, since the substance to be analyzed is embedded in resin, within the observation field of view, there are regions corresponding to the resin and regions corresponding to the substance to be analyzed. The method of extracting the region corresponding to the particle portion can be implemented by known means. Also, in the software attached to ordinary SEM-EDX analysis equipment, a function for extracting the particle portion is often implemented.

[0038] In a preferred embodiment, the step of extracting the particle portion may include segmenting the particle portion. As an example of segmenting, any one or more of the following may be included: SLIC method, Watershed method, Quick shift method, and felzenszwalb method, etc. Further, in a preferred embodiment, the segmenting may be based on any one or more of the SLIC method and the Quick Shift method.

[0039] 3-3. Step of deriving a representative composition based on density-based clustering in the particle portions After extracting the particle portion, density-based clustering is performed on the extracted region. Performing density-based clustering on the extracted region may include performing density-based clustering on the elemental map data of the extracted region (Figure 7). Alternatively, as described above, it may include segmenting and performing density-based clustering on the regions corresponding to each segment (Figure 8). The elemental composition data of each segment after segmenting may be created by calculating the average of the elemental composition data within the regions belonging to each segment. For example, as shown in Figure 9, D(X1,Y3), D(X1,Y4), D(X2,Y2), D(X2,Y3), D(X3,Y2) belong to the same one segment by segmenting. By calculating the average of the elemental composition data of these respective regions, the elemental composition data of the segment may be created. Then, density-based clustering may be performed on the elemental composition data of each segment.

[0040] The density-based clustering may be one or more of DBSCAN and OPTICS.

[0041] After density-based clustering, one or more clusters are derived. For example, as shown in Figure 10, when density-based clustering is performed on the elemental composition data of each segment, it is determined to which cluster each segment belongs.

[0042] Next, a representative composition for each cluster is derived. For example, the average value of the elemental composition data of all segments determined to belong to a particular cluster may be used as the representative composition (for example, the average value of all data plots included in the cluster indicated by the dotted line in FIG. 11).

[0043] In another example, among a plurality of segments determined to belong to a specific cluster, only a plurality of segments belonging to a high-density region may be extracted, and the average value of the elemental composition data of the extracted plurality of segments may be used as the representative composition (for example, the average value of all data plots included in the cluster indicated by the solid line in FIG. 11). In this case, the method for extracting the high-density region is not particularly limited, and for example, a region having a density equal to or higher than a predetermined standard may be derived based on the radius of a circle and the number of elements in the region defined by the circle.

[0044] 3-4. Step of classifying substances contained in the particles by comparing the representative composition with a reference list After the representative composition is derived, it is compared with the elemental composition data contained in the reference list to classify the substances contained in the particles.

[0045] Specifically, a comparison is made between the representative composition of the cluster and the elemental composition data included in a previously prepared reference list (Figure 12). The reference list includes the substance name, chemical formula, types of elements, and elemental content. If necessary, the reference list may include density information. Then, based on the elemental content, each cluster is searched for which substance in the reference list it most closely matches. The search method is not particularly limited. It is not essential for the elemental content to match exactly. If the degree of similarity is higher than a predetermined reference value, it may be determined to match a specific substance.

[0046] Also, in the reference list of Figure 12, the elemental content is expressed as specific numerical values. However, the numerical values of the elemental content included in the reference list may be expressed as an upper limit, a lower limit, or a combination of these numerical values. And based on the determination of whether the numerical value of the elemental content falls within the range, a determination of which substance it matches may be made.

[0047] Also, when performing a correlation analysis with a specific substance included in the reference list and a correlation coefficient equal to or greater than a predetermined reference value is calculated, it may be determined to match the specific substance. If there are multiple candidate substances that match, the substance with the highest correlation coefficient may be selected.

[0048] 3-5. Step of extracting a region other than the first region as the second region In the above series of steps, by density-based clustering, there is a region determined to belong to a specific cluster, which is regarded as the first region, and the remaining region other than the first region can be set as the second region. More precisely, within the extracted particle region, the region other than the first region can be set as the second region (that is, the expression "the remaining region other than the first region" theoretically includes the region of the resin part).

[0049] For example, when density-based clustering is performed on elemental composition data for each pixel (e.g., each region of an EDX analysis unit), pixels that do not belong to any cluster may be extracted as the second region. Alternatively, when density-based clustering is performed on elemental composition data for each segment, segments that do not belong to any cluster may be extracted as the second region.

[0050] 3-6. Step of classifying at least a part of the second region by comparing it with a second reference list After extracting the second region, the entire second region or a portion thereof may be compared with the second reference list for classification. The elemental composition data used for the comparison may be elemental composition data for each pixel or may be elemental composition data for each segment. The second reference list may have the same configuration as the reference list described above.

[0051] In a further embodiment, the elemental composition data of the pixels or segments included in the second region may be further subjected to a correlation analysis with the elemental composition data of the pixels or segments included in the first region. If the correlation analysis results in a correlation coefficient equal to or greater than a predetermined standard, the elemental composition data of these pixels or segments may be determined to belong to the same cluster as the elemental composition data of the pixels or segments included in the first region.

[0052] Fig. 13 shows an overview of the flow. As described above, particle portions can be extracted from the observation field of view, and density-based clustering can be performed on the particle portions. Regions determined to belong to a cluster are extracted as the first region, and this region can be determined to correspond to the principal component region. On the other hand, regions determined not to belong to any cluster are extracted as the second region, and correlation analysis can be performed between this region and the first region. And regions determined to have a correlation (more precisely, the correlation coefficient is the same as or higher than a certain reference value) are determined as the principal component region. On the other hand, regions determined to have no correlation coefficient (more precisely, the correlation coefficient is the same as or lower than a certain reference value) are determined as the sub-component region. And for the sub-component region, as described above, it may be classified by comparing with the second reference list.

[0053] In density-based clustering, since there are outliers, portions that would originally belong to the principal component region may be determined as outliers. However, by performing correlation analysis, regions that could not be picked up in density-based clustering can be extracted as the principal component region.

[0054] 3-7. Step of deriving the weight ratio of the substances identified by the classification of the first region and the second region As described above, by comparing with the reference list and the second reference list, classification can be performed to identify the substance. And, although not essential, the weight ratio of the identified substance may be derived. [[ID=1,2]]

[0055] [[ID=]]The method for deriving the weight ratio is not particularly limited, but exemplary methods will be described below.

[0056] First, a list containing elemental composition data is created. This list may be created for each pixel, or for each segment, or for each cluster. The elemental composition data is obtained by EDX analysis. And in the case of a segment or a cluster, the average value of the elemental composition data may be adopted.

[0057] Preferably, the list may include an area ratio.

[0058] After creating the list, it is compared with a previously prepared second reference list (e.g., mineral data) (Fig. 12). The second reference list includes a substance name, a chemical formula, elements, an element content, and a density. Then, based on the element content, each cluster searches for which substance in the reference list it most matches. The search method is not particularly limited. It is not essential for the element contents to match exactly, and if the degree of similarity is higher than a predetermined reference value, it may be determined to match a specific substance.

[0059] For each unidentified substance included in the created list, after identifying it from the reference list, the weight ratio of the substance is derived. Specifically, based on the density included in the second reference list (more preferably, further based on the area ratio), the weight ratio can be derived. The data format is not particularly limited, but it may be a format including a substance name and a weight ratio corresponding to each substance name (see Fig.

[14] ).

[0060] By performing the above steps, the substances contained in the sample can be analyzed. And as an analysis result, data on the weight ratio of the substances contained in the sample can be obtained. 4. Application Field The method and the like of the present disclosure in the above-described embodiment can be applied to the analysis of various substances. For example, the method and the like of the present disclosure can be applied to the analysis of substances contained in minerals collected from mines, the analysis of substances contained in recycled raw materials derived from electronic devices, and the like. In particular, the method and the like of the present disclosure in the above-described embodiment are suitable for the analysis of minerals having a solid solution.

Example

[0061] Example 1 A simulated sample of Ta ore was prepared. Specifically, a sample having the composition shown in Fig. 15 was prepared. The sample was embedded in resin. SEM-EDX analysis was performed on the sample embedded in resin. Information on the analytical instrument is as follows: SEM device: SU-70 Manufacturer: Hitachi High-Tech Corporation EDX detector: AZtecLive UltimMax170 Manufacturer: Oxford Instruments Co., Ltd. Measurement conditions: Acceleration voltage 20 kV

[0062] The image data of the observation field of view was processed to extract particles, and the inside of the particles was segmented based on the Quick Shift method.

[0063] Density-based clustering processing was applied to each segment. Specifically, DBSCAN clustering processing was applied. As initial parameters, the radius was set to 3 wt%, and the minimum number was set to 3% of the number of segments (for example, when the number of segments is 1000, it is set to 30).

[0064] The results of the clustering processing are shown in Fig. 16. Clusters corresponding to Ta minerals, Li minerals, and gangue were detected. In particular, the portions with high density were extracted, averaged, and elemental composition data serving as representative compositions were created. Then, comparison was made with mineral list data (data including substance names, types of elements, and element contents), and classification was carried out.

[0065] After performing the clustering processing, based on the elements and contents of each cluster, a query was made to the mineral database to perform density estimation. Then, the weight ratio for each mineral was calculated. The results are shown in Fig. 17. Since a simulated sample was used, the ideal results were clear in advance, and the values are shown on the left side of the graph in Fig. 17. The results obtained by the above clustering, etc. are shown in the center. Also, the measurement results by the conventional method (mineral particle analyzer (MLA)) are shown on the right side. The results of the weight ratio obtained by the density-based clustering processing were in good agreement with the ideal values.

[0066] Example 2 A sample of copper concentrate was prepared. Analysis was carried out in the same manner as in Example 1. However, after performing density-based clustering, the particularly high-density portions were extracted, averaged, and elemental composition data representing the composition was created. Then, correlation analysis was performed between the elemental composition data representing the composition and the elemental composition data of the portions that became outliers in the clustering. Among the portions that became outliers, portions having a certain degree of correlation (specifically, a correlation coefficient of 0.95 or more) were incorporated as having the same composition as the cluster.

[0067] Then, for the remaining portions that became outliers, by comparing with the mineral list, it was determined which substances they corresponded to.

[0068] The results are shown in Fig. 18. Also, the measurement results by the conventional method (mineral particle analyzer (MLA)) are shown on the right side. In the example by the above method (the left side of the measurement results), it is shown that a small amount of minerals can be detected.

[0069] The specific embodiments of the invention have been described above. The above embodiments are merely specific examples, and the present invention is not limited to the above embodiments. For example, the technical features disclosed in one of the above embodiments can be applied to other embodiments. Also, unless otherwise specified, for a specific method, some steps can be interchanged with the order of other steps, and additional steps can be added between two specific steps. The scope of the present invention is defined by the scope of the claims.

Claims

1. A method for analyzing substances contained in a sample, comprising: - obtaining elemental map data from SEM-EDX analysis of the sample; - extracting particle portions from the elemental map data; - deriving a representative composition based on density-based clustering in the particle portions; - comparing the representative composition with a reference list to classify substances contained in the particles. A method including the above steps.

2. The method according to Claim 1, wherein the density-based clustering is any one or more of DBSCAN and OPTICS.

3. The method according to Claim 1, wherein: the step of extracting the particle portions includes segmenting the particle portions; the step of deriving the representative composition includes, based on density-based clustering, extracting a portion with high density as a first region corresponding to at least a part of the principal component region, and deriving the representative composition of the principal component region. A method.

4. The method according to Claim 3, wherein the segmentation is based on any one or more of the SLIC method and the Quick Shift method.

5. The method according to Claim 3, further comprising: - extracting a region other than the first region as a second region; - classifying at least a part of the second region by comparing it with a second reference list. A method including the above steps.

6. The method according to Claim 5, further comprising a step of deriving the weight ratio of the substances identified by the classification of the first region and the second region.

7. The method according to Claim 1, wherein the sample is a solid solution type ore.

8. A program for executing the method according to any one of Claims 1 to 7.

9. A computing device installed with the program according to Claim 8.

Images