Sample analysis device and method for creating a pyrolysis product library

Abstract

To analyze a resin sample by using a thermal decomposition analysis technique.SOLUTION: A thermal decomposition product library 24 made of a plurality of prediction mass spectrum groups corresponding to a plurality of resin candidates is created by using a prediction model 13. An information processing unit 16 generates a plurality of actually measured mass spectrums corresponding to a plurality of compounds generated by a thermal decomposition of a resin sample 22. After that, the information processing unit 16 searches for the thermal decomposition product library 24 on the basis of the plurality of actually measured mass spectrums and determines at least one contained resin contained in the resin sample 22 on the basis of the library.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a sample analysis device and a method for creating a pyrolysis product library, and in particular to a technique for analyzing a sample by mass spectrometry of pyrolysis products produced from the sample. [Background technology]

[0002] Various sample analysis systems are known, including a sample analysis system that includes a pyrolysis device, a gas chromatograph, and a mass spectrometer (Py-GC-MS system). Hereinafter, this system will be referred to as a pyrolysis analysis system.

[0003] In a pyrolysis analysis system, the sample to be analyzed is typically a polymer (generally a resin). A mixed gas consisting of multiple pyrolysis products is generated from the sample introduced into the pyrolyzer. This mixed gas is passed through a column in a gas chromatograph, which separates the multiple pyrolysis products. These multiple pyrolysis products are multiple components generated from the sample, i.e., multiple compounds. The multiple compounds are sequentially introduced into a mass spectrometer, which performs mass analysis on each compound. As a result, a mass spectral train (which can also be called Py-GC-MS data) consisting of multiple mass spectra arranged on the retention time axis is generated.

[0004] The pyrolysis analysis system generally includes an information processing device that identifies a sample based on a mass spectral sequence. In conventional information processing devices, for example, a composite mass spectrum representing the entire mass spectral sequence is created from all or a significant portion of the mass spectral sequence. The composite mass spectrum is then compared with multiple registered mass spectra. The sample is identified based on the comparison results (see, for example, Patent Document 1).

[0005] When using such a sample analysis method, if the sample contains impurities, the shape of the composite mass spectrum will change due to the influence of the impurities, resulting in a decrease in the accuracy of the sample analysis. Furthermore, it is difficult to apply such a sample analysis method to a mixed sample consisting of multiple samples. Furthermore, when using such a sample analysis method, changes in the pyrolysis conditions or component separation conditions make sample analysis difficult.

[0006] Some existing compound databases contain the mass spectra (specifically, EI mass spectra) of each compound. Such databases can be called mass spectral databases. However, existing mass spectral databases only contain a portion of the pyrolysis products (and the corresponding mass spectra) produced by the pyrolysis of individual polymers. In pyrolysis analysis, it is currently difficult to perform mass spectral matching using existing mass spectral databases as they are.

[0007] It should be noted that Patent Document 1 does not describe a sample analysis technique that utilizes an artificially generated mass spectrum. [Prior art documents] [Patent documents]

[0008] [Patent Document 1] Japanese Patent Application Laid-Open No. 2000-35422 Summary of the Invention [Problem to be solved by the invention]

[0009] An object of the present invention is to realize highly reliable sample analysis using pyrolysis, to provide a sample analyzer capable of analyzing samples even when pyrolysis conditions or component separation conditions change, or to provide a pyrolysis product library capable of identifying many pyrolysis products. [Means for solving the problem]

[0010] The sample analyzer according to the present invention includes a measurement unit that performs mass analysis on a plurality of compounds produced by thermal decomposition of a polymer sample; a generation unit that generates a measured mass spectrum set consisting of a plurality of measured mass spectra corresponding to the plurality of compounds based on data output from the measurement unit; and an analysis unit that analyzes the polymer sample by comparing the measured mass spectrum set with all or part of a plurality of predicted mass spectra corresponding to a plurality of polymer candidates, wherein the predicted mass spectra corresponding to each of the polymer candidates includes a plurality of predicted mass spectra corresponding to a plurality of pyrolysis products that may be produced from each of the polymer candidates. and each predicted mass spectrum is a mass spectrum generated by a machine-learned prediction model. It is characterized by:

[0011] A program according to the present invention is a program executed on an information processing device, and includes: a function for comparing a set of measured mass spectra, which is made up of a plurality of measured mass spectra corresponding to a plurality of compounds produced by thermal decomposition of a polymer sample, with all or a portion of a group of predicted mass spectra corresponding to a plurality of polymer candidates; and a function for analyzing the polymer sample based on a result of comparing the set of measured mass spectra with all or a portion of the group of predicted mass spectra, wherein the group of predicted mass spectra corresponding to each of the polymer candidates includes a plurality of predicted mass spectra corresponding to a plurality of thermal decomposition products that may be produced from each of the polymer candidates. and each predicted mass spectrum is a mass spectrum generated by a machine-learned prediction model. It is characterized by:

[0012] The method for creating a pyrolysis product library according to the present invention is characterized by comprising the steps of: creating, for each of a plurality of polymer candidates, a plurality of structural formulas representing a plurality of pyrolysis products theoretically derived from the polymer candidate; generating, for each of the polymer candidates, a group of predicted mass spectra consisting of a plurality of predicted mass spectra based on the plurality of structural formulas; and creating a pyrolysis product library including a group of predicted mass spectra corresponding to the plurality of polymer candidates. [Effects of the Invention]

[0013] According to the present invention, highly reliable sample analysis can be achieved using a pyrolysis method. Alternatively, according to the present invention, it is possible to analyze samples even when the pyrolysis conditions or component separation conditions change. Alternatively, according to the present invention, it is possible to provide a pyrolysis product library that can identify many pyrolysis products. [Brief explanation of the drawings]

[0014] [Figure 1] 1 is a block diagram showing a sample analysis system according to a first embodiment. [Figure 2] FIG. 1 is a block diagram illustrating a prediction model generation device. [Figure 3] FIG. 1 is a block diagram showing a library creation device. [Figure 4] FIG. 2 is a block diagram showing a measurement unit. [Figure 5] FIG. 2 is a block diagram showing an information processing unit. [Figure 6] FIG. 10 is a diagram showing an example of a resin candidate list. [Figure 7] FIG. 1 is a block diagram showing a method for generating a group of compounds. [Figure 8] FIG. 1 is a diagram showing a first example of a primitive compound. [Figure 9] FIG. 10 is a diagram showing a second example of a primitive compound. [Figure 10] FIG. 1 is a diagram showing a first example of cutting of a primitive compound. [Figure 11] FIG. 10 is a diagram showing a second example of cleavage of a primitive compound. [Figure 12] FIG. 10 is a diagram illustrating an example of a connection change. [Figure 13] FIG. 1 shows an example of a pyrolysis product library. [Figure 14] FIG. 10 is a diagram illustrating a management table. [Figure 15] 1 is a flowchart showing a method for creating a pyrolysis product library. [Figure 16] 1 is a flowchart showing a sample analysis method. [Figure 17] FIG. 10 is a block diagram showing a measurement unit according to a second embodiment. [Figure 18]FIG. 10 is a diagram showing a sample analysis method according to a second embodiment. [Figure 19] FIG. 10 is a diagram showing a management table according to the second embodiment. [Figure 20] FIG. 10 is a block diagram showing a sample analyzing system according to a third embodiment. [Figure 21] FIG. 10 is a diagram showing a sample analysis method according to a third embodiment. [Figure 22] FIG. 11 is a diagram showing a management table according to the third embodiment. [Figure 23] FIG. 11 is a diagram showing a display example according to the third embodiment. [Figure 24] FIG. 10 is a diagram showing a sample analysis method according to a fourth embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0015] Hereinafter, an embodiment will be described with reference to the drawings.

[0016] (1) Overview of the embodiment A sample analyzer according to an embodiment includes a measurement unit, a generation unit, and an analysis unit. The measurement unit performs mass analysis on a plurality of compounds produced by thermal decomposition of a polymer sample. The generation unit generates a measured mass spectrum set consisting of a plurality of measured mass spectra corresponding to the plurality of compounds based on data output from the measurement unit. The analysis unit analyzes the polymer sample by comparing the measured mass spectrum set with all or part of a plurality of predicted mass spectra corresponding to a plurality of candidate polymers. The predicted mass spectra corresponding to each candidate polymer include a plurality of predicted mass spectra corresponding to a plurality of pyrolysis products that may be produced from each candidate polymer.

[0017] The above configuration analyzes a polymer sample through the individual identification of multiple compounds (i.e., multiple pyrolysis products) produced by the thermal decomposition of the polymer sample. Therefore, with this configuration, even if the polymer sample contains impurities, the sample analysis can be performed without being significantly affected by the impurities. Furthermore, with this configuration, mass spectrum comparison results can be obtained for each predicted mass spectrum group (i.e., for each polymer candidate), making it possible to simultaneously identify multiple polymers contained in the polymer sample. Furthermore, with this configuration, polymer samples can be analyzed even when the thermal decomposition conditions or component separation conditions change.

[0018] In the measuring unit according to the embodiment, pyrolysis, component separation, and mass analysis are sequentially performed. In the embodiment, each measured mass spectrum is an integrated mass spectrum corresponding to a compound peak, as described below. However, each measured mass spectrum may be a mass spectrum corresponding to a compound peak.

[0019] In an embodiment, each predicted mass spectrum is an artificial mass spectrum generated based on each pyrolysis product theoretically derived from the polymer candidate. Specifically, each predicted mass spectrum is a mass spectrum generated by providing a structural formula representing each pyrolysis product to a machine-learned prediction model.

[0020] In general, it is extremely difficult to identify or extract multiple mass spectra corresponding to multiple pyrolysis products from a mass spectral sequence obtained through the pyrolysis of a polymer. However, it is possible to theoretically (and comprehensively) predict multiple pyrolysis products that may be generated from a polymer, and it is also possible to predict mass spectra from the structural formulas of each predicted pyrolysis product. From this perspective, in an embodiment, a group of multiple predicted mass spectra corresponding to multiple candidate polymers is created in advance, and a polymer sample is analyzed using these groups. Individual predicted mass spectra may be generated when performing mass spectral matching.

[0021] In an embodiment, the analysis unit calculates a score for each polymer candidate by performing mass spectrum matching between the set of measured mass spectra and all or part of the group of predicted mass spectra, and then determines one or more polymers contained in the polymer sample based on the score for each polymer candidate.

[0022] For example, if a polymer sample contains a specific polymer candidate, the mass spectral matching results between the measured mass spectrum set and the predicted mass spectrum group corresponding to the specific polymer candidate will be good, resulting in a high score. On the other hand, if the polymer sample does not contain the specific polymer candidate, the mass spectral matching results between the measured mass spectrum set and the predicted mass spectrum group corresponding to the specific polymer candidate will be poor, resulting in a low score. It is possible to determine the possibility of each polymer candidate being contained based on each score. Each score is an index showing the degree of spectral matching.

[0023] In an embodiment, the analysis unit calculates a similarity matrix for each polymer candidate as a result of the mass spectrum matching. The analysis unit then calculates a score based on the similarity matrix for each polymer candidate. In an embodiment, the analysis unit counts the number of similarities in the similarity matrix for each polymer candidate that are higher than a threshold. The analysis unit then calculates a score for each polymer candidate based on the number of similarities that are higher than the threshold.

[0024] For example, if a measured mass spectrum set is composed of i measured mass spectra, and j predicted mass spectra predicted from j pyrolysis products are associated with a certain polymer candidate, i × j similarities are calculated, and a similarity matrix is ​​constructed from these similarities. In this case, if the polymer candidate is contained in the polymer sample, a relatively large number of high similarities will occur. The greater the number of high similarities, the more likely the polymer candidate is contained in the polymer sample. Generally, j is different for each polymer candidate. When calculating the score for each polymer candidate, j may or may not be taken into consideration. If screening is performed prior to mass spectrum matching, the number of predicted mass spectra compared with each measured mass spectrum can be reduced. Screening may eliminate some of the multiple polymer candidates from identification targets.

[0025] In an embodiment, the analysis unit extracts a selected predicted mass spectral sequence from a group of predicted mass spectra based on the molecular mass information of each compound, to be compared with each measured mass spectrum. This configuration enables screening of pyrolysis products (or polymer candidates) based on the molecular mass information.

[0026] In the embodiment, the molecular mass information is the exact mass of each compound molecule produced by pyrolysis. Information other than the exact mass may also be used as the molecular mass information.

[0027] In an embodiment, the analysis unit estimates the compositional formula of each compound based on the molecular mass information of the compound. The analysis unit then compares the estimated compositional formula with compositional formulas corresponding to pyrolysis products of each polymer candidate, thereby extracting a selected predicted mass spectral sequence. This configuration improves the analytical accuracy of polymer samples by using both compositional formula matching and mass spectral matching (specifically, applying them in stages).

[0028] In an embodiment, the measurement unit has a first ion source and a second ion source that are different from each other, and the generation unit generates a first measured mass spectrum set based on first data output from the measurement unit when the first ion source is used. Generation part generates a second measured mass spectrum set based on second data output from the measurement unit when the second ion source is used, and the analysis unit identifies multiple molecular mass information for multiple compounds based on multiple molecular ion peaks included in the second measured mass spectrum set.

[0029] For example, the first ion source is an ion source that follows the hard ionization method, and the second ion source is an ion source that follows the soft ionization method. The hard ionization method is an ionization method that is more likely to produce fragment ions than the soft ionization method. The soft ionization method is more likely to produce molecular ions than the hard ionization method. N This is an easy-to-produce ionization method.

[0030] The sample analyzer according to the embodiment further includes a pyrolysis product library having a plurality of predicted mass spectra and a compound library having a set of mass spectra corresponding to a set of known compounds. The analysis unit searches the pyrolysis product library and the compound library based on the set of measured mass spectra. Then, for each measured mass spectrum, the analysis unit determines the attribute of the compound corresponding to the measured mass spectrum based on the results of searching the pyrolysis product library and the compound library. This configuration improves the accuracy of identifying compound attributes by using the two libraries in combination. For example, the compound attribute may be determined as a first attribute indicating that the compound is a pyrolysis product of a polymer candidate, or a second attribute indicating that the compound is not a pyrolysis product of a polymer candidate (i.e., is a different compound).

[0031] In an embodiment, the set of mass spectra includes a plurality of predicted mass spectra generated from a plurality of known structural formulas corresponding to a plurality of known compounds. For example, the set of mass spectra is a set of predicted mass spectra. This configuration makes it possible to easily create a compound library based on an existing compound database.

[0032] In an embodiment, the analysis unit identifies the attribute of the compound by identifying the library to which the mass spectrum that produced the highest similarity belongs. That is, if the mass spectrum that produced the highest similarity belongs to the pyrolysis product library, the first attribute is determined as the attribute of the compound. On the other hand, if the mass spectrum that produced the highest similarity belongs to the compound database, the second attribute is determined as the attribute of the compound.

[0033] The sample analyzer according to the embodiment includes a TICC generator and a reference image generator. The TICC generator generates a total ion image including a plurality of compound peaks corresponding to a plurality of actually measured mass spectra based on data output from a measurement unit. current A reference image generator generates a reference image to be displayed together with the TICC based on the attributes of the multiple compounds identified according to the multiple measured mass spectra. The reference image includes multiple markers that are displayed together with the multiple compound peaks and indicate the attributes of the multiple compounds. With this configuration, by observing the reference image, it is possible to easily determine whether the compound peaks included in the TICC are peaks derived from pyrolysis products.

[0034] A program according to an embodiment is a program executed on an information processing device, and has a first function and a second function. The first function is a function of comparing a measured mass spectrum set consisting of a plurality of measured mass spectra corresponding to a plurality of compounds produced by thermal decomposition of a polymer sample with all or a portion of a plurality of predicted mass spectra corresponding to a plurality of polymer candidates. The second function is a function of analyzing the polymer sample based on the results of comparing the measured mass spectrum set with all or a portion of the plurality of predicted mass spectra. The predicted mass spectra corresponding to each polymer candidate include a plurality of predicted mass spectra corresponding to a plurality of pyrolysis products that may be produced from each polymer candidate.

[0035] The information processing device is, for example, a computer, a mass spectrometer, or a sample analysis system. The program is installed in the information processing device via a network or a portable storage medium. The information processing device has a non-transitory storage medium for storing the program.

[0036] A method for creating a pyrolysis product library according to an embodiment includes a first step, a second step, and a third step. In the first step, for each of a plurality of polymer candidates, a plurality of structural formulas representing a plurality of pyrolysis products theoretically derived from the polymer candidate are created. In the second step, for each of the polymer candidates, a group of predicted mass spectra consisting of a plurality of predicted mass spectra is generated based on the plurality of structural formulas. In the third step, a pyrolysis product library including a group of predicted mass spectra corresponding to a plurality of polymer candidates is created.

[0037] The above-mentioned creation method involves theoretically creating multiple pyrolysis products and generating multiple artificial predicted mass spectra from those pyrolysis products. This method makes it possible to artificially create a pyrolysis product library, which is difficult to create experimentally.

[0038] In an embodiment, the step of creating a plurality of structural formulas includes a step of creating a primitive structural formula of an original pyrolysis product as a monomer linkage, and a step of generating a plurality of derived structural formulas from the primitive structural formula. The plurality of structural formulas includes the primitive structural formula and a plurality of derived structural formulas. The polymer has a degree of polymerization distribution ranging, for example, from hundreds to tens of thousands. Pyrolysis of such a polymer generates a variety of relatively small pyrolysis products, including pyrolysis products consisting of k repeating units, where k is typically 1, 2, 3, 4, or 5. Based on this, it is desirable to create a primitive pyrolysis product consisting of, for example, six or more repeating units, and then create a plurality of derived pyrolysis products from the primitive pyrolysis product.

[0039] (2) Details of the embodiment Figure 1 shows a sample analysis system according to a first embodiment. The sample analysis system is specifically a polymer analysis system, and more specifically a resin analysis system. The resin analysis system is a system that identifies one or more resins contained in a resin sample 22. Samples other than resins may also be analyzed.

[0040] The resin analysis system includes a predictive model generation device 10, a library creation device 12, and a resin analysis device 14. The predictive model generation device 10 and the library creation device 12 function prior to actual sample analysis, and it is the resin analysis device 14 that functions during actual sample analysis.

[0041] The predictive model generation device 10 is a device that generates a predictive model by machine learning based on information stored in an existing mass spectrum database 18. The mass spectrum database 18 stores, for example, a variety of compounds and their corresponding EI mass spectra. A specific example of this is the NIST database. The predictive model after machine learning is transferred to the library creation device 12. An example configuration of the predictive model generation device 10 will be described later with reference to FIG. 2.

[0042] Prior to analyzing the resin sample, a resin candidate list 20 consisting of multiple resin candidates is created in advance. The entity of each resin candidate is a resin identifier such as a resin name. The multiple resin candidates are specified, for example, by a user. The number of resin candidates may be, for example, several tens. A larger number of resin candidates may be specified, or a smaller number of resin candidates may be specified. Each resin candidate is a resin that may be contained in the resin sample 22, and is a candidate for identification.

[0043] The library creation device 12 is a device that creates a pyrolysis product library using the prediction model 13 generated by the prediction model generation device 10. The library creation device 12 creates a group of pyrolysis products (specifically, a group of structural formulas) for each resin candidate in the resin candidate list 20, and then generates a group of predicted mass spectra by providing the group of pyrolysis products to the prediction model 13. By repeating this process, multiple groups of predicted mass spectra corresponding to multiple resin candidates are generated. These multiple groups of predicted mass spectra constitute a pyrolysis product library. The pyrolysis product library is installed in the resin analysis device 14. An example configuration of the library creation device 12 will be described later using FIG. 3.

[0044] The resin analysis device 14 has a measurement unit 15 and an information processing unit 16. The measurement unit 15 is equipped with a pyrolysis device, a gas chromatograph, and a mass spectrometer. In the pyrolysis device, a resin sample 22, which is a polymer sample, is pyrolyzed to generate a mixed gas consisting of multiple pyrolysis products. The mixed gas is introduced into the gas chromatograph, and the multiple pyrolysis products are separated. The multiple separated pyrolysis products are multiple components generated from the resin sample 22, which are multiple compounds. The multiple compounds are sequentially introduced into the mass spectrometer. In the mass spectrometer, mass analysis is performed on the multiple compounds.

[0045] In the information processing unit 16, multiple mass spectra arranged on the retention time axis are generated based on the data output from the measurement unit 15. These mass spectra form a mass spectral train. Next, multiple mass spectra are accumulated for each pyrolysis product, i.e., for each compound, to generate an accumulated mass spectrum. This generates multiple accumulated mass spectra corresponding to multiple compounds. Each accumulated mass spectrum is treated as an actual measured mass spectrum for compound identification.

[0046] The information processing unit 16 searches the pyrolysis product library 24 based on a plurality of measured mass spectra corresponding to a plurality of compounds, thereby identifying one or more contained resins contained in the resin sample 22. A specific configuration example of the measurement unit 15 will be described later using FIG. 4. A specific configuration example of the information processing unit 16 will be described later using FIG. 5.

[0047] In the embodiment, the predictive model generation device 10, the library creation device 12, and the information processing device 16 are each configured as a computer. The predictive model generation device 10 and the library creation device 12 may be configured as a single computer. The predictive model generation device 10, the library creation device 12, and the information processing device 16 may be configured as a single computer. Some or all of the functions performed by the predictive model generation device 10, the library creation device 12, and the information processing device 16 may be executed by one or more computers on a network.

[0048] An example configuration of the prediction model generation device 10 will be described with reference to FIG. 2. In the illustrated example configuration, the prediction model generation device 10 includes a conversion unit 26 and a model generation unit 30. The conversion unit 26 is a module that generates graph structure data from the structural formula of a compound. The model generation unit 30 functions as a learning device. Specifically, in the embodiment, the model generation unit 30 is configured by a GCN (Graph Convolutional Networks). That is, the GCN is the entity of the prediction model 32. Instead of the GCN, another network or another model may be used.

[0049] The mass spectrum database 18 stores information on a large number of compounds. Specifically, it stores a large number of structural formulas and mass spectra corresponding to a large number of compounds. The entire set of structural formulas and mass spectra is used as a training data set for machine learning. Each training data is composed of a structural formula 34 and a corresponding mass spectrum 36. The structural formula 34 is converted into graph structure data 38 by the conversion unit 26. The mass spectrum 36 is used as correct answer data. From the perspective of the GCN, the pair of graph structure data 38 and mass spectrum 36 is the actual training data.

[0050] A large amount of training data is provided to the model generation unit 30. This gradually improves the predicted mass spectrum output from the prediction model 32. Through this machine learning process, a prediction model 32 that predicts a mass spectrum from an arbitrary structural formula is completed.

[0051] 3 shows an example configuration of the library creation device 12. In the illustrated configuration example, the library creation device 12 includes a pyrolysis product creation unit 40, a conversion unit 42, and a prediction unit 44. The pyrolysis product creation unit 40 comprehensively generates a large number of pyrolysis products for each resin candidate in the resin candidate list 20. Specifically, the pyrolysis product creation unit 40 generates a large number of structural formulas representing the large number of pyrolysis products for each resin candidate. These structural formulas constitute a group of structural formulas.

[0052] There are various known methods for expressing structural formulas. For example, the MOL-SDF file format and linear notation methods such as SMILES and SMARTS are known. An example of the resin candidate list 20 will be explained later using Figure 6. The method for generating the group of pyrolysis products (group of structural formulas) will be explained in detail later using Figures 7 to 12.

[0053] The conversion unit 42 converts the input structural formula into graph structure data. The converted graph structure data is provided to the prediction model 13 in the prediction unit 44. As a result, a predicted mass spectrum is generated for each structural formula, i.e., for each pyrolysis product. One group of predicted mass spectra is generated for each resin candidate. One group of predicted mass spectra constitutes one sub-library 48. A plurality of sub-libraries 48 corresponding to a plurality of resin candidates constitute the pyrolysis product library 24. Note that the conversion unit 42 may convert the input structural formula into a data format other than a graph structure, for example, a data format that conforms to a model such as a fingerprint.

[0054] The pyrolysis product library 24 is not generated experimentally, but is an artificially generated virtual library. It is extremely difficult to construct a sub-library by analyzing the mass spectral sequence obtained by mass spectrometry of multiple compounds produced by the pyrolysis of a resin sample. In contrast, according to this embodiment, such a sub-library can be easily created.

[0055] 4 shows an example of the configuration of the measurement unit 15 in the resin analysis device. As described above, the measurement unit 15 has a pyrolysis device 52, a gas chromatograph 54, and a mass spectrometer 56. The mass spectrometer 56 has an EI ion source 58, a mass analyzer 60, and a detector 62.

[0056] In the pyrolysis device 52, a mixed gas consisting of multiple pyrolysis products is generated by pyrolysis of the resin sample 22. The mixed gas is introduced into a column in a gas chromatograph 54. This allows the multiple pyrolysis products (i.e., multiple compounds) to be separated in time. These compounds are then sequentially introduced into a mass spectrometer 56.

[0057] In the mass spectrometer 56, the EI ion source 58 is an ion source that complies with the electron ionization method (EI method). The EI method is a type of hard ionization method. When a sample is ionized using the EI method, fragment ions are likely to be generated. The mass analyzer 60 is, for example, a quadrupole mass analyzer or a time-of-flight mass analyzer. Ions that pass through the mass analyzer 60 are detected by a detector 62. As a result, detection data (detection signal) is output from the mass spectrometer 56.

[0058] FIG. 5 shows an example of the configuration of the information processing unit 16 in the resin analyzing apparatus. The information processing unit 16 has a processor 64. The processor 64 is, for example, a CPU that executes a program. The information processing unit 16 has a memory. In addition to the program, a pyrolysis product library 24 is stored in the memory. A pyrolysis product library stored on a server on a network may be used. In an embodiment, a management table is stored in the memory together with or including the pyrolysis product library. The management table is a working table in which various information required in the process of performing sample analysis and various information generated in the process are registered.

[0059] 5, multiple functions performed by the processor 64 are represented by multiple blocks. The generation unit 65 has a mass spectrum generator 67, a TICC (total ion chromatogram) generator 68, a peak detector 70, and an integrator 72. The analysis unit 66 has a similarity calculator 74, a score calculator 78, and a determiner 80. The processor 64 also functions as a display processing unit 76.

[0060] The mass spectrum generator 67 receives the detection data output from the measurement unit 15. Based on the received detection data, the mass spectrum generator 67 generates a plurality of mass spectra aligned along the retention time axis, which together form a mass spectrum train.

[0061] The TICC generator 68 generates a total ion current chromatograph (TICC) based on the mass spectral sequence. Specifically, each mass spectrum is integrated to calculate the total ion current (TIC). The TICC is created by plotting the time change of the TIC on the retention time axis. This TICC is also called a pyrogram.

[0062] The peak detector 70 is a module that detects multiple compound peaks contained in the TICC. The multiple compound peaks are basically multiple pyrolysis product peaks.

[0063] The integrator 72 sets an integration interval on the retention time axis for each compound peak and integrates multiple mass spectra within the integration interval. This generates an integrated mass spectrum for each compound peak. In this embodiment, the integrated mass spectrum is treated as an actual mass spectrum corresponding to a specific compound (i.e., a specific pyrolysis product). Multiple actual mass spectra are generated based on multiple compound peaks. These constitute an actual mass spectrum set.

[0064] The similarity calculator 74 accesses the pyrolysis product library 24 and refers to a plurality of predicted mass spectra corresponding to a plurality of resin candidates. The similarity calculator 74 performs mass spectrum matching between the measured mass spectrum set and the plurality of predicted mass spectra, thereby obtaining a plurality of similarity matrices, which will be described later.

[0065] For example, the similarity is calculated using the following formula:

number

[0066] In the above formula, mi represents the i-th mass. Ai represents the i-th intensity in the measured mass spectrum, and Pi represents the i-th intensity in the predicted mass spectrum. The above formula is used to calculate cosine similarity. Other similarities may also be calculated.

[0067] The score calculator 78 calculates multiple scores based on multiple similarity matrices corresponding to multiple resin candidates. The score is an index indicating the likelihood that the resin candidate is included in the resin sample. Prior to calculating the score, each similarity is compared with a threshold th. Specifically, it is determined whether each similarity is higher than the threshold th. Based on the above calculation formula, the threshold th is, for example, 700. In this embodiment, the number of similarities exceeding the threshold th is the score. However, the score may be calculated using other methods. A threshold th may be set for each resin candidate, or a common threshold th may be set for multiple resin candidates.

[0068] The determiner 80 determines one or more candidate resins contained in the resin sample based on multiple scores corresponding to the multiple resin candidates. For example, if a score exceeds a determination value, the resin candidate corresponding to that score is determined to be a contained resin. The determination value is, for example, 10. A determination value may be set for each resin candidate, or a common determination value may be set for multiple resin candidates. The resin candidate corresponding to the highest score may be determined to be a contained resin.

[0069] The display processing unit 76 generates an image to be displayed on the display. In this embodiment, the display displays the TICC and also displays the determination result of the determiner 80. Furthermore, the display may display one or more measured mass spectra, or may display a score string together with the resin candidate list.

[0070] An example of a resin candidate list is shown in Figure 6. The illustrated resin candidate list 20A has a resin name column 84 consisting of a plurality of resin names and a monomer structural formula column 86 consisting of a plurality of monomer structural formulas.

[0071] FIG. 7 illustrates a method for creating a group of pyrolysis products. Each pyrolysis product, i.e., each compound, is represented by a structural formula. Based on the selected resin candidate 20a, a primitive compound sequence (primitive pyrolysis product sequence) 88 is created (S10). The primitive compound sequence 88 is composed of one or more primitive compounds 90. Each primitive compound 90 is, for example, a linked structure made up of k monomers. k is, for example, an integer of 5 or more, preferably an integer in the range of 6 to 10, and in this embodiment, k=8. The user may specify a value of 9 or more or 7 or less as k.

[0072] An example of a primitive compound is shown in Figure 8. The primitive compound 100 is composed of eight monomers (repeating units) 102. In the example shown, each terminal group is H.

[0073] When the resin candidate is a copolymer, a series of primitive compounds are created depending on the bonding style (e.g., random or block), as shown in Figure 9. (A) shows a primitive compound consisting of eight first monomers 104. (B) shows a primitive compound consisting of seven first monomers 104 and one second monomer 106. (C) shows a primitive compound consisting of six first monomers 104 and two second monomers 106. (D) shows a primitive compound consisting of four first monomers 104 and four second monomers 106. (E) shows a primitive compound consisting of eight second monomers 106. As described above, for each resin candidate, all theoretically possible primitive compounds are created according to the condition k = 8.

[0074] In FIG. 7, a plurality of precursor compounds created based on a plurality of resin candidates are registered in memory 92A (S12). Meanwhile, a cleavage process is applied to each of the precursor compounds created in S10 (S14). The cleavage process is an artificial operation that simulates cleavage during thermal decomposition. Specifically, in this embodiment, assuming k=8, 1, 2, 3, and 4 are sequentially selected as the number of cleavages m, and the cleavage of each precursor compound is repeated as follows: As a result, a plurality of derivative compounds are generated from one precursor compound.

[0075] 10 shows an example of cutting when m=1. In the original compound 108, a specified position 110 is cut. As a result, two compounds 108A and 108B are created from the original compound 108. All cleavable positions in the original compound 108 are sequentially selected as the specified positions 110, and the cutting of the original compound 108 is repeated.

[0076] FIG. 11 shows an example of cutting when m=2. In the original compound 112, two specified positions 114 and 116 are cut. As a result, three compounds 112A, 112B, and 112C are created from the original compound 112. Original Compound 112 In this case, all combinations of the two specified positions are selected in order, and the primitive compound 112 The disconnection is repeated.

[0077] In FIG. 7, a plurality of derivative compounds created by the cutting process are registered in memory 92A (S16). All of the compounds registered in memory 92A constitute a primary compound group 94. Subsequently, a bond modification process is applied to compounds sequentially selected from the primary compound group 94 (S18). Specifically, assuming that k=8, 1 and 2 are sequentially selected as the number of modification positions n, and bond modifications are repeatedly performed on the selected primary compounds as follows. As a result, one or more derivative compounds are generated from one primary compound. In the primary compound group 94, primary compounds that do not satisfy the application conditions are excluded from the bond modification process.

[0078] FIG. 12 shows an example of a bond modification when n=1. In primary compound 118, a designated single bond 120 is modified to an unsaturated bond 122, thereby creating another compound 118a. The bond modification takes into account chemical conditions, specifically valence. If possible, other single bonds in primary compound 118 can be modified to unsaturated bonds, and this is also performed. A similar process is performed when n=2.

[0079] 7, multiple derivative compounds created by the bond modification process are registered in memory 92B (S20). At the same time, a primary compound group 94 is also registered in memory 92B (S22). All compounds registered in memory 92B constitute a secondary compound group 96. The secondary compound group 96 is a group of pyrolysis products (compound group) corresponding to a specific resin candidate.

[0080] The above process is repeated for each candidate resin. This results in a plurality of groups of pyrolysis products corresponding to the plurality of candidate resins. As described above, the substance of each pyrolysis product is a structural formula. The above cutting process is a process for creating a plurality of new derived structural formulas. The above bond modification process is also a process for creating a plurality of new derived structural formulas.

[0081] The entire series of steps shown in Figure 7 is generally executed by a processor. However, some steps may be executed manually. For each candidate resin, for example, 100 to 1 million pyrolysis products are created. Note that, in the process of creating a group of pyrolysis products for each candidate resin, if a new pyrolysis product that is the same as a pyrolysis product that has already been created is generated, that pyrolysis product is discarded.

[0082] As explained above, for each resin candidate, multiple structural formulas representing multiple pyrolysis products are sequentially fed into the prediction model (see Figure 3). This sequentially generates multiple predicted mass spectra. As a result, a group of predicted mass spectra is generated for each resin candidate.

[0083] 13 shows an example of a pyrolysis product library. The pyrolysis product library 24A is composed of a plurality of sub-libraries 48A corresponding to a plurality of resin candidates. Each sub-library 48A is composed of a plurality of records 124. Each record 124 is composed of a structural formula 125 and a mass spectrum 126. From another perspective, each sub-library 48A is composed of a group of structural formulas 127 and a group of predicted mass spectra 128.

[0084] An example of a management table is shown in Fig. 14. The illustrated management table 130 is a working table including the pyrolysis product library 24. However, the pyrolysis product library 24 and the working memory area 131 may be separate entities.

[0085] As already explained, the pyrolysis product library 24 is composed of a plurality of sub-libraries 48 corresponding to a plurality of resin candidates. Each sub-library 48 has a group of pyrolysis products (actually a group of structural formulas) 132 and a group of predicted mass spectra 133. In FIG. 14, each predicted mass spectrum is expressed abstractly. The group of predicted mass spectra 133 is composed of j predicted mass spectra, where j is an integer equal to or greater than 2, for example, in the range of 1,000,000 to 1,000,000. Generally, j is different for each resin candidate.

[0086] The measured mass spectrum set 134 is composed of i measured mass spectra corresponding to i compounds produced by thermal decomposition of a resin sample. In FIG. 14, the first measured mass spectrum is labeled (#1), and the second measured mass spectrum is labeled (#2). i is an integer equal to or greater than 2, and is, for example, in the range of 10 to 50. All numerical values ​​described in this specification are merely examples.

[0087] For each resin candidate, mass spectrum matching is performed between the measured mass spectrum set 134 and the predicted mass spectrum set 133. Specifically, for each measured mass spectrum, the measured mass spectrum is compared with the predicted mass spectrum set 133 for each resin candidate, thereby calculating j similarities α. This process is applied to each of the i predicted mass spectra. As a result, a similarity matrix 135 is obtained. The similarity matrix 135 is made up of i × j similarities α. Reference numeral 135y denotes a similarity column made up of j similarities α, and reference numeral 135x denotes a similarity row made up of i similarities α.

[0088] In the embodiment, the number of similarities α that exceed the threshold th is counted for each similarity matrix 135. The number is used as the score.

[0089] Specifically, for each pyrolysis product, it is checked whether there is a similarity α in the similarity row 135x that exceeds the threshold th. If the maximum similarity is, for example, 1000, the threshold th is, for example, 700. If there is a similarity α in each similarity row 135x that exceeds the threshold th, a YES flag is registered; if there is not, a NO flag is registered (see reference numeral 136). The number of YES flags is counted to determine the score 138. This process is repeated for each resin candidate. The score 138 is simply a hit rate, and more specifically, indicates the likelihood that the resin candidate is contained in the resin sample. In the embodiment, the score 138 is a count value.

[0090] For each measured mass spectrum, it may be investigated whether there is a similarity α in the similarity column 135y that exceeds the threshold th. For example, if there is no similarity α that exceeds the threshold th, the second attribute (other than a pyrolysis product) may be determined as the attribute of the compound corresponding to the measured mass spectrum.

[0091] Among the multiple scores corresponding to the multiple resin candidates, a superior score exceeding a judgment value is identified, and the resin candidate corresponding to the superior score is determined to be a resin contained in the resin sample (i.e., a contained resin). The judgment value is, for example, 10. When multiple superior scores are identified, multiple contained resins are determined. The judgment value may be the same across the multiple resin candidates, or may be different for each individual resin candidate. For example, the judgment value may be adaptively set for each individual resin candidate based on the value of j. The resin candidate corresponding to the best score may be determined to be a contained resin. As a determination result, a score list sorted in descending order may be displayed together with the resin candidate list.

[0092] Figure 15 shows pyrolysis A method for generating a product library is shown as a flowchart. In S30, one resin candidate is selected from a list of resin candidates. In S31, an original compound (specifically, a structural formula) is created based on the selected resin candidate. In S31, multiple original compounds may be created based on the selected resin candidate. In S32, a compound group is created by cutting and modifying the bond of the original compound. The compound group also includes the original compound. In S34, a group of predicted mass spectra is generated based on the compound group. In S36, a sub-library is constructed from the compound group. In S38, it is determined whether or not there is a next resin candidate. If there is a next resin candidate, the processes in S30 to S36 above are applied to the next resin candidate. Through the above process, a pyrolysis product library consisting of multiple sub-libraries is created.

[0093] FIG. 16 shows a flowchart of a resin sample analysis method. In S40, one resin candidate is selected from a plurality of resin candidates. In S42, mass spectrum matching is performed between the measured mass spectrum set and the predicted mass spectrum group corresponding to the selected resin candidate. This results in a similarity matrix being calculated. In S44, a score is calculated based on the similarity matrix. In S46, it is determined whether a next resin candidate exists. If a next resin candidate exists, the processes of S40 to S44 are applied to the next resin candidate. In S48, one or more resins contained in the resin sample are determined based on the score sequence. The determination result is displayed. Note that the comparison of the selected measured mass spectrum with the plurality of mass spectrum groups may be repeatedly performed.

[0094] The resin analysis system described above analyzes a resin sample through the individual identification of multiple compounds (i.e., multiple pyrolysis products) produced from the resin sample. Therefore, even if the resin sample contains impurities, the analytical accuracy is unlikely to decrease. Furthermore, multiple resins contained in the resin sample can be simultaneously identified. Since retention time information is not used during resin analysis, analytical accuracy can be maintained even if the pyrolysis conditions or component separation conditions change.

[0095] Next, a resin analysis system according to a second embodiment will be described with reference to Figures 17 to 19. Like the resin analysis system according to the first embodiment, the resin analysis system according to the second embodiment is also composed of a prediction model generation device, a library creation device, and a resin analysis device (see Figure 1). Below, we will describe the parts of the resin analysis system according to the second embodiment that are different from the resin analysis system according to the first embodiment.

[0096] Fig. 17 shows the configuration of a measurement unit 144 according to the second embodiment. Among the components shown in Fig. 17, the same components as those shown in Fig. 4 are given the same reference numerals, and the description thereof will be omitted.

[0097] In the measurement unit 144, the mass spectrometer 146 has a composite ion source 148. The composite ion source 148 has an EI ion source according to electron ionization (EI) and an FI ion source according to chemical ionization (FI). These ion sources are selectively used. The EI is a type of hard ionization method, and the FI is a type of soft ionization method. When the hard ionization method is used, fragment ions are more likely to be observed. When the soft ionization method is used, molecular ions are more likely to be observed. A hard ionization method other than the EI method may be used, and a soft ionization method other than the FI method may be used.

[0098] In the second embodiment, a first measurement on a resin sample and a second measurement on the same resin sample are performed sequentially. For example, in the first measurement, an EI ion source is used, and first detection data is output from the mass spectrometer 146. In the second measurement, an FI ion source is used, and second detection data is output from the mass spectrometer 146.

[0099] A generator in an information processing unit according to the second embodiment generates a first mass spectral sequence based on the first detection data, generates a second mass spectral sequence based on the second detection data, and then generates a first TICC (first pyrogram) based on the first mass spectral sequence and generates a second TICC (second pyrogram) based on the second mass spectral sequence.

[0100] 18 shows a schematic diagram of the processes executed in the generation unit and analysis unit in the information processing unit. In the first TICC 150 and the second TICC 152, the horizontal axis represents retention time (RT), and the vertical axis represents total ion current (TIC).

[0101] The retention time axis of the first TICC 150 and the retention time axis of the second TICC 152 correspond to each other and are parallel in Fig. 18. The first TICC 150 includes a plurality of compound peaks (a first compound peak sequence). Similarly, the second TICC 152 includes a plurality of compound peaks (a second compound peak sequence).

[0102] The generation unit repeatedly performs peak pairing between the first compound peak sequence and the second compound peak sequence, and based on the results, generates a first measured mass spectral sequence based on the first mass spectral sequence, and also generates a second measured mass spectral sequence based on the second mass spectral sequence.

[0103] Specifically, for example, a peak search range 158 is determined based on a representative position (e.g., apex position, center of gravity position) 156 of a first compound peak 154, and a second compound peak 160 belonging to the peak search range 158 is searched for in the second TICC 152. This associates the first compound peak 154 with the second compound peak 160. Such peak pairing is performed sequentially along the retention time axis.

[0104] An integration interval is determined for each first compound peak, and multiple first mass spectra belonging to the integration interval are integrated. As a result, a first integrated mass spectrum, i.e., a first measured mass spectrum, is generated for each first compound peak. Figure 18 shows a first measured mass spectrum 164 corresponding to first compound peak 154.

[0105] Similarly, an integration interval is determined for each second compound peak, and multiple second mass spectra belonging to the integration interval are integrated. This generates a second integrated mass spectrum, or a second measured mass spectrum, for each second compound peak. Figure 18 shows a second measured mass spectrum 166 corresponding to a second compound peak 160. The first measured mass spectrum 164 and the second measured mass spectrum 166 correspond to each other; that is, they are mass spectra representing the same compound.

[0106] The first measured mass spectrum 164 contains multiple fragment peaks but no molecular ion peak, while the second measured mass spectrum 166 contains a molecular ion peak 167 and a small number of other fragment ion peaks.

[0107] As shown in FIG. 18 , the analysis unit identifies the exact mass (specifically, mass-to-charge ratio) w1 of each compound based on the molecular ion peak 167, and then estimates the compound's compositional formula based on the exact mass w1 (see reference numeral 168). In this process, for example, the type and number of elements constituting the compositional formula are searched for so that the sum of masses based on the compositional formula matches the exact mass within a certain error range. The exact mass is molecular mass information. Other molecular mass information (e.g., nominal mass or molecular weight) may be used instead of the exact mass. When identifying the molecular ion peak 167, the first measured mass spectrum 164 and the second measured mass spectrum 166 may be compared.

[0108] Next, the analysis unit performs formula matching between the estimated formula and the multiple registered formula groups for each compound, assuming that multiple formula groups corresponding to multiple pyrolysis products of the multiple resin candidates are registered. For each compound, the analysis unit identifies each registered formula that matches in the formula matching and extracts a predicted mass spectrum corresponding to each identified registered formula (see reference numeral 170). This results in one or more selected predicted mass spectra to be used in mass spectrum matching being extracted from the multiple predicted mass spectra for each compound. A selected predicted mass spectrum sequence 171 is constructed from the one or more selected predicted mass spectra. The analysis unit then performs mass spectrum matching between the measured mass spectrum and the selected predicted mass spectrum sequence 171 for each compound to calculate a similarity sequence (see reference numeral 172). Based on the multiple similarity sequences corresponding to the multiple compounds, multiple similarity matrices corresponding to the multiple resin candidates are constructed (see reference numeral 173). The analysis unit calculates multiple scores based on the multiple similarity matrices and identifies one or more contained resins based on the multiple scores.

[0109] FIG. 19 shows a management table 130A according to the second embodiment. Reference numeral 24B denotes a pyrolysis product library. The pyrolysis product library 24B is composed of multiple sub-libraries 48B corresponding to multiple resin candidates. Each sub-library 48B includes an identifier group 139 indicating a pyrolysis product group, a composition formula group (registered composition formula group) 140, and a predicted mass spectrum group 141. As described above, for each compound produced by pyrolysis, the estimated composition formula (estimated composition formula) is compared with the multiple registered composition formula groups 140 corresponding to the multiple resin candidates, and one or more registered composition formulas matching the estimated composition formula are identified. From the multiple predicted mass spectrum groups 141, one or more selected predicted mass spectra a1, a2,... (i.e., selected predicted mass spectral sequences) corresponding to one or more registered composition formulas matching the estimated composition formula are extracted. Reference numeral 143 denotes a similarity sequence calculated between the measured mass spectrum (#1) and the selected predicted mass spectral sequence.

[0110] According to the second embodiment, a primary determination based on molecular mass information and a secondary determination by mass spectrum matching are performed in stages, thereby improving the accuracy of sample analysis. In the second embodiment, molecular mass information matching may be performed instead of composition formula matching. In that case, multiple pieces of molecular mass information may be managed in the management table 130A shown in FIG. 19 instead of multiple composition formulas.

[0111] Next, a sample analysis system according to a third embodiment will be described with reference to Figures 20 to 23. Like the resin analysis system according to the first embodiment, the resin analysis system according to the third embodiment is also composed of a prediction model generation device, a library creation device, and a resin analysis device (see Figure 1). Below, the parts of the resin analysis system according to the third embodiment that are different from the resin analysis system according to the first embodiment will be described.

[0112] FIG. 20 shows a resin analysis system according to a third embodiment. In FIG. 20, components similar to those shown in FIG. 1 are assigned the same reference numerals, and their description will be omitted. In FIG. 20, the resin analysis system has a first library creation device 12A and a second library creation device 190. The first library creation device 12A corresponds to the library creation device according to the first embodiment, and has a prediction model 13A. The prediction model 13A generates a group of predicted mass spectra for each resin candidate in the resin candidate list 20.

[0113] The second library creation device 190 has a prediction model 13B that is the same as the prediction model 13A. The compound database 192 is an existing compound database in which a large number of compounds are registered, including their structural formulas. The compound database 192 is, for example, PubChem. Like the first library creation device, the second library creation device 190 has a prediction model 13B and generates a large number of predicted mass spectra from the structural formulas of a large number of compounds. These predicted mass spectra constitute a compound library. The compound library may contain, for example, 100 million or more predicted mass spectra.

[0114] The information processing unit 16A of the resin analyzing device 14A stores a pyrolysis product library 24, and also stores a compound library 194 created by a second library creating device 190. The information processing unit 16A has a generating unit and an analyzing unit, similar to the information processing unit according to the first embodiment. The processing contents of the analyzing unit are shown in FIG. 21.

[0115] 21, TICC 196 includes multiple compound peaks. Measured mass spectrum 200 is an accumulated mass spectrum corresponding to compound peak 198. The analysis unit according to the third embodiment has an attribute determination function that determines, for each compound peak, whether or not it corresponds to a thermal decomposition product of a resin sample.

[0116] Specifically, for each measured mass spectrum, the analysis unit performs a cross-sectional search of the pyrolysis product library and the compound library based on the measured mass spectrum. In FIG. 21, a first search 202 indicates a search of the pyrolysis product library, and a second search 204 indicates a search of the compound library. Next, as indicated by reference numeral 206, the analysis unit identifies the highest similarity between the results of the first search 202 and the results of the second search. Then, if the highest similarity occurs with any predicted mass spectrum in the pyrolysis product library, the analysis unit determines the first attribute (pyrolysis product) as the attribute. On the other hand, if the highest similarity occurs with any predicted mass spectrum in the compound library, the analysis unit determines the second attribute (other compound) as the attribute. An example of the other compound is an impurity.

[0117] As described above, in the third embodiment, the attribute of the corresponding compound is determined for each compound peak included in the TICC. In mass spectrum matching, the measured mass spectrum corresponding to the pyrolysis product is used, while the measured mass spectrum corresponding to other compounds is excluded. The attribute determination results for each compound peak are provided to the user via a reference image, which will be described later.

[0118] 22 shows a management table 130B according to the third embodiment. The management table 130B has a first portion 208 and a second portion 210. The first portion 208 corresponds to a pyrolysis product library, and the second portion 210 corresponds to a compound library 212. The pyrolysis product library is composed of a plurality of sub-libraries 48C corresponding to a plurality of resin candidates. For each resin candidate, mass spectrum matching is performed between the measured mass spectrum set 134 and the predicted mass spectrum group 133, thereby calculating a first similarity matrix 135A.

[0119] The compound library 212 includes a compound set 214 and a predicted mass spectrum set 216. Mass spectrum matching is performed between the measured mass spectrum set 134 and the predicted mass spectrum set 216, and a second similarity matrix 220 is calculated thereby.

[0120] Next, for each measured mass spectrum, the highest similarity is determined from among the multiple similarities. For example, the highest similarity occurs between the measured mass spectrum (#1) and the predicted mass spectrum corresponding to pyrolysis product b2 (see reference numeral 222-1). As a result, the first attribute (Pyrolysis) 224 is determined as the attribute of the compound. On the other hand, the highest similarity occurs between the measured mass spectrum (#2) and the predicted mass spectrum corresponding to compound x2 (see reference numeral 222-2). As a result, the second attribute (Other) 226 is determined as the attribute of the compound.

[0121] Among the multiple measured mass spectra, the measured mass spectrum related to the determination of the second attribute is excluded in the score calculation. Specifically, in the example shown in Fig. 22, the similarity string corresponding to the measured mass spectrum (#2) is not referenced in the score calculation. Ultimately, among the multiple measured mass spectra, the measured mass spectrum related to the determination of the first attribute is determined to be a valid measured mass spectrum, and only the mass spectrum matching results obtained from the valid measured mass spectrum are subject to the score calculation.

[0122] An example of a display is shown in FIG. 23. The display processing unit of the information processing unit has a function of generating a reference image. In FIG. 23, an image 230 has a TICC 232 and a reference image 234 superimposed thereon. The reference image 234 has a plurality of markers corresponding to a plurality of compound peaks. The plurality of markers includes a first marker 236a representing a first attribute (thermal decomposition products) and a second marker 236b representing a second attribute (other compounds). By observing the reference image 234, for example, the compound peak can be evaluated, or the content ratio of impurities can be recognized.

[0123] According to the third embodiment, it is possible to use a valid measured mass spectrum from among multiple measured mass spectra for mass spectrum matching, thereby improving the accuracy of sample analysis. Compounds that exist in both the compound library and the pyrolysis product library are preferably deleted from the compound library.

[0124] Figure 24 shows an algorithm executed in a sample analysis system according to the fourth embodiment. The fourth embodiment corresponds to a combination of the second and third embodiments. Like the resin analysis systems according to the first to third embodiments, the resin analysis system according to the fourth embodiment is also composed of a prediction model generation device, a library creation device, and a resin analysis device. The above algorithm is executed in an information processing unit.

[0125] 24, a first TICC 150 and a second TICC 152 are generated from a resin sample under hard ionization. The first TICC 150 includes a plurality of first compound peaks aligned on the retention time axis, and similarly, the second TICC 152 includes a plurality of second compound peaks aligned on the retention time axis. Peak pairing is performed for each compound between the plurality of first compound peaks and the plurality of second compound peaks (see FIG. 18).

[0126] 24, multiple steps in block 250 are performed for each compound (hereinafter referred to as a compound of interest). Subsequently, in S50 and S52, a first compound peak and a second compound peak corresponding to the compound of interest are identified. In S54, a first measured mass spectrum is generated based on the first compound peak, and in S56, a second measured mass spectrum is generated based on the second compound peak.

[0127] In S58, a molecular ion peak contained in the second measured mass spectrum is identified, and molecular mass information, i.e., accurate mass, of the compound of interest is determined based on the mass-to-charge ratio corresponding to the molecular ion peak. In S60, the compositional formula of the compound of interest is estimated based on the accurate mass.

[0128] The pyrolysis product library 24B includes a plurality of predicted mass spectra corresponding to a plurality of pyrolysis products, and also includes a plurality of compositional formulas corresponding to a plurality of pyrolysis products. In S60, the estimated compositional formula of the compound of interest is compared with a plurality of registered compositional formulas in the pyrolysis product library 24B, and pyrolysis products (selected pyrolysis products) corresponding to each registered compositional formula that matches the estimated compositional formula are identified. This primary screening extracts one or more selected pyrolysis products for each compound of interest, that is, one or more selected predicted mass spectra. The one or more selected predicted mass spectra constitute a selected predicted mass spectral sequence 252.

[0129] In S62, first mass spectral matching is performed between the first measured mass spectrum and the selected predicted mass spectral sequence 252, and a similarity is calculated for each selected predicted mass spectrum. Meanwhile, in S64, second mass spectral matching is performed between the first measured mass spectrum and the set of predicted mass spectra in the compound library 194, and a similarity is calculated for each predicted mass spectrum.

[0130] In S66, the maximum similarity is identified from the multiple similarities calculated in S62 and the multiple similarities calculated in S64. If the maximum similarity is derived from a predicted mass spectrum belonging to the pyrolysis product library 24B, the first attribute (pyrolysis product) is determined as the attribute of the compound of interest, and this maximum similarity is considered to be the valid similarity. In other words, the first maximum similarity is referenced in the score calculation. On the other hand, if the maximum similarity is derived from a predicted mass spectrum belonging to the compound library 194, the second attribute (another compound) is determined as the attribute of the compound of interest. In this way, in the fourth embodiment, secondary screening is performed based on the attribute of the compound of interest. The above process is applied sequentially to each of the multiple compounds generated by pyrolysis.

[0131] In S68, a score is calculated for each resin candidate based on one or more corresponding valid similarities. A score of 0 may be assigned to resin candidates for which no valid similarities have been calculated. Subsequently, in S68, each resin candidate that has produced a score exceeding the threshold is determined to be an included resin. Meanwhile, in S68, a reference image is generated based on the results of the attribute determination, and is displayed together with the first TICC (or, in some cases, the first TICC and the second TICC). Note that in the fourth embodiment, a similarity matrix is ​​also generated for each resin candidate, and a score is calculated for each similarity matrix. However, the number of valid similarities contained in each similarity matrix differs. In the fourth embodiment, each similarity matrix contains a relatively large number of invalid elements (similarity 0 or blank).

[0132] In the fourth embodiment, a primary screening using a composition formula and a secondary screening using a compound library are performed, thereby improving the accuracy of resin analysis. Other screening techniques may also be applied to the fourth embodiment. [Explanation of symbols]

[0133] 10 predictive model generation device, 12 library creation device, 13 predictive model, 14 resin analysis device, 15 measurement unit, 16 information processing unit, 24 pyrolysis product library, 65 generation unit, 66 analysis unit.

Claims

1. a measurement unit that performs mass spectrometry on a plurality of compounds generated by thermal decomposition of the polymer sample; a generation unit that generates a measured mass spectrum set including a plurality of measured mass spectra corresponding to the plurality of compounds based on the data output from the measurement unit; an analysis unit that analyzes the polymer sample by comparing the set of measured mass spectra with all or part of a group of predicted mass spectra corresponding to a plurality of candidate polymers; Including, the group of predicted mass spectra corresponding to each of the polymer candidates includes a plurality of predicted mass spectra corresponding to a plurality of pyrolysis products that may be produced from each of the polymer candidates; Each of the predicted mass spectra is a mass spectrum generated by a machine-learned prediction model. A sample analysis device characterized by:

2. 2. The sample analyzer according to claim 1, Each of the predicted mass spectra is a mass spectrum generated by providing a structural formula representing each of the pyrolysis products to a machine-learned prediction model. A sample analysis device characterized by:

3. A measurement unit that performs mass spectrometry on multiple compounds generated by thermal decomposition of a polymer sample; a generation unit that generates a measured mass spectrum set including a plurality of measured mass spectra corresponding to the plurality of compounds based on the data output from the measurement unit; an analysis unit that analyzes the polymer sample by comparing the set of measured mass spectra with all or part of a group of predicted mass spectra corresponding to a plurality of candidate polymers; Including, the group of predicted mass spectra corresponding to each of the polymer candidates includes a plurality of predicted mass spectra corresponding to a plurality of pyrolysis products that may be produced from each of the polymer candidates; The analysis unit performing mass spectrum matching between the set of measured mass spectra and all or a portion of the group of predicted mass spectra, and calculating a similarity matrix for each of the polymer candidates as a result of the mass spectrum matching; calculating a score for each of the polymer candidates based on a similarity matrix for each of the polymer candidates; determining one or more polymers contained in the polymer sample based on the score for each of the polymer candidates; A sample analysis device characterized by:

4. 4. The sample analyzer according to claim 3, The analysis unit counting the number of similarities in the similarity matrix that are higher than a threshold for each of the candidate polymers; calculating the score for each of the polymer candidates based on the number of similarities higher than the threshold; A sample analysis device characterized by:

5. A measurement unit that performs mass spectrometry on multiple compounds generated by thermal decomposition of a polymer sample; a generation unit that generates a measured mass spectrum set including a plurality of measured mass spectra corresponding to the plurality of compounds based on the data output from the measurement unit; an analysis unit that analyzes the polymer sample by comparing the set of measured mass spectra with all or part of a group of predicted mass spectra corresponding to a plurality of candidate polymers; Including, the group of predicted mass spectra corresponding to each of the polymer candidates includes a plurality of predicted mass spectra corresponding to a plurality of pyrolysis products that may be produced from each of the polymer candidates; the analysis unit extracts, from the group of predicted mass spectra, a selected predicted mass spectral sequence to be compared with each of the measured mass spectra, based on the molecular mass information of each of the compounds; A sample analysis device characterized by:

6. 6. The mass spectrometer according to claim 5, The analysis unit estimating the composition formula of each compound based on the molecular mass information of the compound; extracting the selected predicted mass spectral sequence by comparing the estimated composition formula with a group of composition formulas corresponding to a group of pyrolysis products of each of the polymer candidates; A sample analysis device characterized by:

7. 6. The sample analyzer according to claim 5, the measurement unit has a first ion source and a second ion source that are different from each other, The generation unit generating a first measured mass spectrum set as the measured mass spectrum set based on first data output from the measurement unit when the first ion source is used; generating a second measured mass spectrum set based on second data output from the measurement unit when the second ion source is used; the analysis unit identifies a plurality of molecular mass information of the plurality of compounds based on a plurality of molecular ion peaks included in the second measured mass spectrum set. A sample analysis device characterized by:

8. A measurement unit that performs mass spectrometry on multiple compounds generated by thermal decomposition of a polymer sample; a generation unit that generates a measured mass spectrum set including a plurality of measured mass spectra corresponding to the plurality of compounds based on the data output from the measurement unit; an analysis unit that analyzes the polymer sample by comparing the set of measured mass spectra with all or part of a group of predicted mass spectra corresponding to a plurality of candidate polymers; A sample analyzer comprising: the group of predicted mass spectra corresponding to each of the polymer candidates includes a plurality of predicted mass spectra corresponding to a plurality of pyrolysis products that may be produced from each of the polymer candidates; The sample analyzer includes: a pyrolysis product library having the plurality of predicted mass spectra; a compound library having a set of mass spectra corresponding to a set of known compounds; Including, The analysis unit searching the pyrolysis product library and the compound library based on the measured mass spectrum set; determining, for each of the measured mass spectra, the attributes of the compound corresponding to the measured mass spectrum based on the search results of the pyrolysis product library and the compound library; A sample analysis device characterized by:

9. 9. The sample analyzer according to claim 8, The mass spectrum collection includes a plurality of predicted mass spectra generated from a plurality of known structural formulas corresponding to a plurality of known compounds. A sample analysis device characterized by:

10. 9. The sample analyzer according to claim 8, the analysis unit determines the attribute of the compound by identifying the library to which the mass spectrum that produced the highest similarity belongs; A sample analysis device characterized by:

11. 9. The sample analyzer according to claim 8, a TICC generator that generates a total ion chromatogram (TICC) including a plurality of compound peaks corresponding to the plurality of measured mass spectra based on the data output from the measurement unit; and a reference image generator that generates a reference image to be displayed with the TICC based on attributes of a plurality of compounds identified according to the plurality of measured mass spectra; Including, the reference image includes a plurality of markers displayed together with the plurality of compound peaks, the markers indicating attributes of the plurality of compounds; A sample analysis device characterized by:

12. A program executed in an information processing device, a function of comparing a set of measured mass spectra, which is made up of a plurality of measured mass spectra corresponding to a plurality of compounds produced by thermal decomposition of a polymer sample, with all or a portion of a group of predicted mass spectra corresponding to a plurality of polymer candidates; a function of analyzing the polymer sample based on a result of comparing the set of measured mass spectra with all or a portion of the plurality of predicted mass spectra; Including, the group of predicted mass spectra corresponding to each of the polymer candidates includes a plurality of predicted mass spectra corresponding to a plurality of pyrolysis products that may be produced from each of the polymer candidates; Each of the predicted mass spectra is a mass spectrum generated by a machine-learned prediction model. A program characterized by:

13. A program executed on an information processing device, a function of comparing a set of measured mass spectra, which is made up of a plurality of measured mass spectra corresponding to a plurality of compounds produced by thermal decomposition of a polymer sample, with all or a portion of a group of predicted mass spectra corresponding to a plurality of polymer candidates; a function of analyzing the polymer sample based on a result of comparing the set of measured mass spectra with all or a portion of the plurality of predicted mass spectra; Including, the group of predicted mass spectra corresponding to each of the polymer candidates includes a plurality of predicted mass spectra corresponding to a plurality of pyrolysis products that may be produced from each of the polymer candidates; The function of analyzing the polymer sample comprises: a function of performing mass spectrum matching between the set of measured mass spectra and all or part of the plurality of predicted mass spectra, and calculating a similarity matrix for each of the polymer candidates as a result of the mass spectrum matching; a function of calculating a score for each of the polymer candidates based on a similarity matrix for each of the polymer candidates; a function of determining one or more polymers contained in the polymer sample based on the score for each of the polymer candidates; A program comprising:

14. creating, for each of the plurality of polymer candidates, a plurality of structural formulas representing a plurality of pyrolysis products theoretically derived from the polymer candidate; generating a group of predicted mass spectra, each of which is composed of a plurality of predicted mass spectra, based on the plurality of structural formulas for each of the polymer candidates; creating a pyrolysis product library including a plurality of predicted mass spectra corresponding to the plurality of polymer candidates; A method for creating a pyrolysis product library, comprising:

15. The method for creating a pyrolysis product library according to claim 14, The step of generating a plurality of structural formulas comprises: creating an original structural formula of the original pyrolysis product as a monomer linkage; generating a plurality of derived structural formulas from the primitive structural formula; Including, The plurality of structural formulas includes the original structural formula and the plurality of derived structural formulas. A method for creating a pyrolysis product library.

Images