Method for acquiring calibration curve

Abstract

To solve the problem in which: in determining the composition ratio of a composition according to an infrared absorption method, proximity of absorption peaks of a plurality of substances that form the composition makes it difficult to determine the composition ratio.MEANS FOR SOLVING THE PROBLEM: A measuring method acquires calibration curve candidates, for each of a plurality of combinations of wavelengths near an absorption wavelength and reference wavelengths, from an optical intensity obtained by measuring samples having existing composition ratios according to an infrared absorption method. The measuring method performs residual evaluation for the calibration curve candidates, and determines a calibration curve candidate in which the combination of the wavelengths is in an appropriate relationship as a calibration curve to be used, thereby determining a composition ratio even if a composition has proximate absorption wavelengths.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a method for obtaining a calibration curve for detecting a composition ratio from observation results obtained by a plurality of observation means in which observation physical quantities are changed by a multivariate analysis method in order to determine the composition ratio of a composition.

Background Art

[0002] When deriving a value B that has a correlation with a certain measured value A, when it is difficult or laborious to directly obtain the B from the measured value A, a calibration curve showing the relationship between the measured value A and the value B is obtained from the measured value A of a plurality of known value B samples measured in advance and the known value B of the measured samples and used at the time of actual measurement.

[0003] Patent Document 1 describes that, as a standard principle for concentration quantification, a calibration curve is determined so that the sum of squared residuals is minimized.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] In the invention described in Patent Document 1, when a calibration curve candidate is obtained for each combination of levels of the plurality of conditions, the calibration curve candidate with the minimum sum of squared residuals is determined as the calibration curve to be used in actual observation (hereinafter, the used calibration curve). Patent Document 1 does not describe whether the content of the combination of levels of a plurality of conditions is physically considered when determining the used calibration curve. Therefore, there is a problem that a calibration curve candidate obtained from a combination of factor levels that is not physically appropriate can be determined as the used calibration curve.

[0006] Therefore, in the present invention, observation results (e.g., light intensity) obtained by changing an observed physical quantity (e.g., solid content concentration) for samples with known composition ratios are acquired, and a plurality of calibration curve candidates are acquired for each sample from the observed physical quantity and the observation results. As a result of selection using a residual evaluation rule and a selection rule (including physical rules), a calibration curve to be used is determined as the calibration curve to be used from within the calibration curve candidates. A method for obtaining a calibration curve is provided. Further, in addition to the light intensity at the absorption wavelength, the contribution degrees of the light intensities at wavelengths shorter and longer than the absorption wavelength are also optimized to obtain calibration curve candidates, thereby providing calibration curve candidates that are even more excellent in the result of residual evaluation and determining a more excellent calibration curve candidate as the calibration curve to be used. A method for obtaining a calibration curve can be provided.

[0007] In particular, by using the present invention, in the case where the measurement target is a coating film using a black material with a low infrared reflectance, or in a coating film (e.g., the positive electrode of a lithium-ion battery) for which the solid content concentration is to be measured, when the material contains a material having an absorption peak close to the solvent to be measured, or when the above two cases overlap, a calibration curve to be used can be obtained in a short time by the infrared absorption method that can be performed non-contact and non-destructively, and a solid content concentration measurement result can be obtained. In the production line, on-site, all or sampled high-frequency quality control solid content concentration measurements can be performed, improving and maintaining the quality of parts, and suppressing the occurrence of defective products or second-class products that do not meet the desired performance as products incorporating the parts.

Means for Solving the Problems

[0008] In order to solve the problems related to the above calibration curve acquisition method, in the present application, as a first invention, A method for obtaining a calibration curve for detecting a composition ratio from observation results obtained by a plurality of observation means that change an observed physical quantity by a multivariate analysis method in order to determine the composition ratio of a composition, A sample preparation step of preparing a plurality of samples with known composition ratios, An observation result acquisition step of acquiring a value indicating an observation result obtained by changing an observed physical quantity for each prepared sample, From the known composition ratios for each sample and the observation results obtained from the sample, a calibration curve candidate acquisition step of acquiring a plurality of calibration curve candidates for each observed physical quantity for which the observation results were obtained; A residual evaluation step of performing residual evaluation using a residual evaluation rule that is a rule for evaluating the residual for each of the acquired calibration curve candidates; A used calibration curve determination step of determining, using a selection rule for selecting a calibration curve candidate for which the residual has been evaluated from the results of the residual evaluation, a calibration curve to be used that uses a predetermined calibration curve candidate; Provided is a method for obtaining a calibration curve comprising the above.

[0009] As a second invention, based on the first invention, Provided is a method for obtaining a calibration curve, wherein the plurality of observation means is an observation of the reflected light intensity (or the same applies to absorbance and transmitted light intensity. The same applies hereinafter) when light of a plurality of wavelengths is applied to a composition.

[0010] As a third invention, based on any one of the first or second inventions, The selection rule is such that when the composition is manufactured with a preconceived composition ratio, the wavelength of light for which a large change in the reflected light intensity is expected at the targeted composition ratio, and observations using light of wavelengths longer and shorter than that wavelength are included in at least a plurality of observation means. Provided is a method for obtaining a calibration curve.

[0011] As a fourth invention, based on any one of the first to third inventions, The selection rule includes a selection rule of selecting an observation at a wavelength at which the ratio affected by the change is low when the reflected light intensity of the composition changes due to factors other than the composition ratio. Provided is a method for obtaining a calibration curve.

Effects of the Invention

[0012] With the measuring apparatus of the present invention having the above configuration, observation results (e.g., light intensity) are obtained by changing the observed physical quantity (e.g., solid content concentration) for a sample with a known composition ratio, and from the observed physical quantity and the observation results, a plurality of calibration curve candidates are obtained for each sample. As a result of selection using a residual evaluation rule and a selection rule (including physical rules), a calibration curve to be used is determined as the calibration curve to be used from within the calibration curve candidates. A calibration curve determination method is provided. Further, in addition to the light intensity at the absorption wavelength, the contribution degrees of the light intensities at wavelengths shorter and longer than the absorption wavelength are also optimized to obtain calibration curve candidates, thereby providing calibration curve candidates that are even more excellent in the result of residual evaluation and determining a more excellent calibration curve candidate as the calibration curve to be used. A method for obtaining a calibration curve can be provided.

[0013] By using the present invention, particularly in the case where the measurement target is a coating film using a black material with a low infrared reflectance, or in a coating film (e.g., the positive electrode of a lithium-ion battery) for which the solid content concentration is to be measured and contains a material having an absorption peak close to the solvent to be measured, or in the case where the above two cases overlap, a calibration curve to be used can be obtained in a short time by the infrared absorption method that can be performed non-contact and non-destructively, and a solid content concentration measurement result can be obtained. In the production line, solid content concentration measurements for quality control can be performed frequently in-situ, either for all or for a sampled batch, improving and maintaining the quality of the components and suppressing the occurrence of defective or second-class products that do not meet the desired performance as products incorporating the components.

Brief Description of the Drawings

[0014]

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

Modes for Carrying Out the Invention

[0015] <Premise for the description of all embodiments> <Regarding the hardware that can constitute the present invention> The present invention is an invention that uses a computer in principle. However, at least a part of it is realized by software, also realized by hardware, and also realized by the cooperation of software and hardware. In this case, the software uses hardware resources to perform various calculations and realizes various functions through the required data and information. It can be said that the information processing by software is specifically realized using hardware resources.

[0016] In the hardware that realizes all or part of each component of the present invention, it is composed of an MPU including a CPU which is the basic configuration of a computer, a memory, a bus, input / output devices, various peripheral devices, a user interface, and the like. The various peripheral devices may include a storage device, an interface such as the Internet, a device such as the Internet, a LAN device, a Wifi (registered trademark) device, a display, a display interface, a keyboard, a mouse, a speaker, a microphone, a camera, a video, a television, a CD device, a DVD device, a Blu-ray device, a USB memory, a USB memory interface, a removable type hard disk, a general hard disk, a projector device, an SSD, a telephone, a fax, a copier, a printing device, a movie editing device, various sensor devices, and the like.

[0017] Also, this system does not necessarily have to be configured by a single housing, and it may be configured by communicatively coupling a plurality of housings. Also, the communication may be a LAN, a WAN, Wifi (registered trademark), Bluetooth (registered trademark), infrared communication, ultrasonic communication, near field communication (NFC), a mobile phone network, and furthermore, a part of it may be installed across national borders.

[0018] <Satisfaction of the availability of the natural law of the invention of the present application in all embodiments>

[0019] The present invention functions in cooperation with a computer and software. In the present invention, in order to determine the composition ratio of a composition, a calibration curve for detecting the composition ratio from observation results obtained by a plurality of observation means that change an observed physical quantity by a multivariate analysis method is obtained from the results of observing a sample with a known composition ratio by a plurality of observation means that change the observed physical quantity. A plurality of calibration curve candidates are obtained from the results of observation by a plurality of observation means, the residual is evaluated for each calibration curve candidate, and from the results of the residual evaluation, the calibration curve candidate for which the residual has been evaluated is selected and determined as a calibration curve using a predetermined calibration curve candidate, and the process is performed on a computer. Since the observation means for determining the composition ratio of the composition using the observation results obtained by changing the observed physical quantity is an observation means that conforms to physical laws (for example, infrared absorption method, etc.), the present invention is an invention that utilizes natural laws.

[0020] <Hardware Configuration>

[0021] FIG. 3 shows the overall conceptual configuration of the measuring device (0300) of the present invention and, as an example using the measuring device (0300), the case of in-line measuring the solid content concentration of a positive electrode active material slurry, which is a mixture of a positive electrode active material and a solvent containing a positive electrode active material coated on a positive electrode substrate, in the positive electrode manufacturing process of a lithium-ion battery. As the positive electrode active material, a positive electrode active material of a general lithium-ion battery can be used. The positive electrode active material slurry, which is a mixture of a positive electrode active material and a solvent containing a positive electrode active material, is, for example, a mixture of a positive electrode active material (e.g., NCM; lithium nickel cobalt manganese oxide), a binder (e.g., PVdF; polyvinylidene fluoride), a conductive assistant (e.g., acetylene black), and a solvent (e.g., NMP; N-methyl-2-pyrrolidone). As the positive electrode substrate for coating the positive electrode active material slurry, for example, Al foil is used. In FIG. 3, the positive electrode active material slurry is coated on the surface of the Al foil drawn from the raw roll (0311) on the right side using known means such as a syringe or a slit coater (positive electrode active material coater (0313)), and the solid content concentration of the coating film is measured by the measuring device (0300) before drying with the heater (0311). After measuring the solid content concentration of the coating film, the solvent is volatilized with the heater (0311) and the dried film is wound up by the winding roller (0312).

[0022] As an example of the conceptual configuration of the measuring apparatus of the present invention, as shown in FIG. 3 which shows an example of a measuring apparatus using the infrared absorption method, a light source (0301) that irradiates infrared rays, and a filter disk (0302) provided with a plurality of filters that limit the light from the light source (0301) to a specific wavelength, and an optical system that irradiates infrared rays to the measurement object composed of a mirror (a plane mirror, a concave mirror, a convex mirror in the figure, may include a lens) and guides the reflected light to the sensor, a sensor (0303) that receives the infrared rays reflected from the positive electrode active material coating film applied to the positive electrode substrate that is the measurement object, a data processing unit (0304) that calculates the composition ratio (in the example of FIG. 3, the solid content concentration of the coating film) based on the light intensity received by the sensor (0303), a display device (0354) that displays the composition ratio output from the data processing unit (0304), and a PC (0352) or a server device (0351) that acquires the composition ratio from the data processing unit (0304) via the Internet line (0350).

[0023] Among the substances constituting the positive electrode active material slurry, those other than NMP which is a solvent are solutes that are solid components, and they are substances that remain as a film on the metal foil when the solvent volatilizes after coating. If the positive electrode active material slurry is coated with the same film thickness but the solid content concentrations are different, the resulting film thickness after solvent volatilization will be different, which will affect the final battery performance. Therefore, in a manufacturing site where the slurry is being continuously applied to a flowing component, it is important to manage the solid content concentration of the coated film in order to maintain quality. In order to determine the composition ratio of the solute, it is necessary to determine the composition ratios of the three types of materials, namely the positive electrode active material, the binder, and the conductive assistant, and three measurements are required. In the positive electrode active material slurry in this example, acetylene black is used as the positive electrode active material, and there is also a problem that the reflectance of infrared rays is small and it is difficult to measure.

[0024] Therefore, if the composition ratio of NMP which is a solvent is determined, the solid content concentration can be determined as the ratio other than NMP. Since the number of measurements can also be reduced, it is effective. The calibration curve for measuring NMP by the infrared absorption method and determining the solid content concentration of the coating film will be a calibration curve in which the solid content concentration has a negative correlation with the measured value of NMP.

[0025] The acquisition of the composition ratio is carried out by utilizing the property that the functional groups of the solvent substance in the coating film absorb infrared light at specific wavelengths. Since the ratio of infrared light absorbed by the solvent changes according to the film thickness and the ratio of the solvent, coating films with a plurality of known compositions are measured in advance, and the ratio of the solvent in the coating film is obtained from the light intensity, and a calibration curve used to obtain the solid content concentration, which is the remaining ratio, is determined and held.

[0026] Although the method for obtaining the calibration curve will be described in detail later, a plurality of wavelengths near the absorption wavelength of the solvent and candidates for the reference wavelength for measuring the light intensity for reference are selected, and filters for transmitting light for each of the wavelengths are set on a filter disk (0302). A plurality of samples with known composition ratios are prepared, and measurements are performed for each filter (wavelength) prepared for each sample. Since manually exchanging the filters is laborious, it is preferable to configure such that when a sample is set, the filter is automatically set, and when the measurement is completed, it is switched to the next filter, and the measurement at all wavelengths can be automatically performed. More preferably, the samples can also be configured to be automatically exchanged. The filter disk (0302) in FIG. 3 fits filters for each wavelength into holes provided on the same circumference within the disk, and rotates it as necessary to insert the desired filter into the optical path from the light source. The filter disk (0302) may be provided with a through hole for use when not using the filter.

[0027] In addition, when the positive electrode substrate transmits infrared light, the same configuration as the above example can be achieved by installing a mirror (not shown) below the positive electrode substrate in FIG. 3 and guiding the reflected light to the sensor. Alternatively, the sensor may be arranged on the side opposite to the light source with the positive electrode active material thin film and its positive electrode substrate interposed therebetween, and configured to measure the transmitted light.

[0028] The calculated composition ratio is output to a user interface (UI) such as a display device (0354) via a USB, LAN terminal, etc., or to an Internet line (0350) (or it may be a LAN line), etc. It is sent to a server device (0351) or a PC (0352) via the Internet line (0350). It can also be configured to output to a PC or a server device by connecting to a LAN line without directly connecting to the Internet line. In that case, a gateway and a firewall can be further connected to the LAN line and configured to connect to the Internet line via the gateway and the firewall. Although only the Internet line is shown in the figure, the same effect can be obtained with only a LAN line or both a LAN line and an Internet line.

[0029] FIG. 4 is a diagram showing an example of the hardware configuration of a data processing unit which is a computer of the present invention shown in FIG. 3. Taking the case of a configuration according to an embedded system as an example, the hardware configuration of the data processing unit in the device of the present invention will be described with reference to FIG. 4. Since FIG. 4 is a diagram for explaining the hardware configuration, the description of individual programs and data is omitted. Note that the hardware of the computer part of the device of the present invention may have a configuration according to a known PC. When a server is used in the system including the device of the present invention, the server may have a configuration according to a known PC.

[0030] FIG. 4 is a conceptual diagram showing the hardware configuration of a computer part (hereinafter sometimes abbreviated as an embedded system) in a measuring device (hereinafter sometimes abbreviated as this device) according to each embodiment of the present invention. As shown in the figure, this device includes an MPU, a non-volatile memory (e.g., ROM, SSD, HDD, flash memory, etc.), a main memory (e.g., DRAM, SRAM, etc.), a LAN I / F (I / F: interface) for connection to a control PC, a recording meter, etc., and an interface for connection to a control module, etc., which is "USB, I 2 C, SPI, etc.", and a user I / F, and includes a system bus (thick line in the figure) for performing signal transmission and reception, etc. between them. Further, "USB, I 2It is also connected to the "infrared sensor, arithmetic board" via "C, SPI, etc."

[0031] <Hardware: MPU> "MPU" is an abbreviation for Microprocessor Unit, which refers to an integrated circuit in which a microprocessor is the center and a CPU, memory, input / output interface, timer, interrupt controller, etc. are integrated. The integrated circuit is often manufactured in a form integrated on a single chip. Examples of functions to be integrated include, in addition to the above examples, CPU, flash memory, memory I / F, AD converter, PLL (Phase-Locked Loop), LAN I / F, USB I / F, I 2 C I / F, SPI I / F, power control I / F, graphic function, floating-point calculation unit, etc. can be integrated. For example, if a LAN I / F is integrated, it is not necessary to prepare a control IC for the LAN I / F in addition to the MPU. The MPU in Figure 4 is an example in which a memory I / F and a graphic function are integrated with a CPU.

[0032] <Hardware: non-volatile memory> As the non-volatile memory mounted on the embedded system, EP-ROM, EEP-ROM, flash memory, or SSD is used as ROM. HDD can also be used, but there are restrictions on vibration resistance and shock resistance.

[0033] The non-volatile memory stores firmware, which is a basic program for operating the embedded system and controlling the hardware. The firmware is read into the main memory, expanded, and executed in the same way as other programs. Also, when a non-volatile memory is mounted on the MPU, the firmware may be stored in the non-volatile memory on the MPU side and operated on the non-volatile memory. The amount of main memory mounted and used can be reduced.

[0034] The "device driver" stored in the non-volatile memory is software used to operate hardware interfaces such as LAN and USB. At runtime, it is read into the main memory, expanded, and executed. Also, when the MPU has a non-volatile memory implemented, the device driver may be stored in the non-volatile memory on the MPU side and operated on the non-volatile memory, which can reduce the amount of main memory installed and used. The specifications and configuration of the firmware and the device driver are determined when the configuration of the computer part (including various I / F configurations) incorporated in this device is defined.

[0035] In addition, other software necessary for operating this device is stored in the non-volatile memory and read into the main memory, expanded, and executed at runtime.

[0036] <Hardware: Main Memory: Volatile Memory> The main memory is used as a work memory when executing various software during the operation of the embedded system. Therefore, DRAM or SRAM with high operating speeds such as reading and writing is used. In particular, SDRAM and DDR SDRAM in DRAM are widely used and mass-produced for PCs and mobile devices, making them easily available.

[0037] Each functional block described in the embodiments below can be implemented by either hardware, software, or a combination of hardware and software. Specifically, in the case of using an embedded system, hardware components such as an MPU, main memory, bus, non-volatile memory, input devices such as operation buttons used for information input, a mouse, touch panel, electronic pen used solely for touching the touch panel, joystick, or pointer position input device similar to a joystick and other external peripheral devices, as well as interfaces for those external peripheral devices, communication interfaces such as LAN terminals, image processing circuits, wireless communication interfaces, driver programs for controlling those hardware, and other application programs can be mentioned. In particular, plotters, data recording devices, personal computers for data acquisition or device control, server devices for data acquisition, wired and wireless networks, and interfaces are used.

[0038] Through the arithmetic processing of the MPU according to the program deployed on the main memory, data input from input devices and other interfaces and held in memory or hardware is processed and stored, or instructions for controlling the above-mentioned hardware and software are generated. Here, the above program may be implemented as a plurality of modularized programs, or may be implemented as one program by combining two or more programs.

[0039] Also, a part of the present invention can be configured as software. Furthermore, a storage medium on which such software is recorded is naturally included in the technical scope of the present invention (the same applies throughout this specification, not limited to this embodiment).

[0040] <Regarding the terms used in the present invention of the application>

[0041] In the present specification, the term "association" is used in the sense that it includes not only the case where two or more pieces of information are directly associated, but also the case where two or more pieces of information are indirectly associated through one or more other pieces of information. Indirect association is not necessarily limited to association within one device (a device with a single housing), and also includes cases where association is made across a plurality of devices.

[0042] "Based on" includes both the case where it is based on the object itself and the case where it is based on the object after some processing has been performed on the object. For example, "calculating the temperature based on the measured radiance" means the "temperature" obtained by substituting the "acquired radiance" obtained by measurement into a predetermined mathematical formula. It is also possible to substitute the "acquired radiance" itself obtained by measurement into a predetermined mathematical formula, or to substitute it into a predetermined formula after performing processing such as obtaining an average value by averaging the "acquired radiance" obtained by measurement several times.

[0043] <Embodiment 1 Outline> Mainly Claim 1 A calibration curve acquisition method configured to obtain observation results by changing an observed physical quantity for a plurality of samples having known composition ratios, obtain a plurality of calibration curve candidates for each sample from the values indicating the observation results and the observed physical quantity for which the observation results were obtained, and determine the calibration curve to be used by residual evaluation or the like.

[0044] <Configuration of the Calibration Curve Acquisition Method in Embodiment 1> FIG. 1 shows a flowchart of the calibration curve acquisition method according to Embodiment 1. The calibration curve acquisition method according to Embodiment 1 is composed of a sample preparation step (S0101), an observation result acquisition step (S0102), a calibration curve candidate acquisition step (S0103), a residual evaluation step (S0104), and a used calibration curve determination step (S0105). In order to obtain a predetermined number k of calibration curve candidates, the observation result acquisition step (S0102) and the calibration curve candidate acquisition step (S0103) are configured to be repeated k times.

[0045] Note that the configuration of the above acquisition method is an example for implementing the present invention, and the method may be appropriately omitted or new steps may be added within the range that does not conflict with the problems to be overcome by the present invention and its effects. The same applies to the descriptions after Embodiment 1.

[0046] <Explanation of Each Step of the Calibration Curve Acquisition Method in Embodiment 1> Regarding each step in the method for acquiring a calibration curve for detecting a composition ratio from observation results obtained by a plurality of observation means using a multivariate analysis method in order to determine the composition ratio of a composition, the following explanation will be given.

[0047] A composition is a substance composed of one or more substances mixed together, and alloys, aqueous solutions, colloidal solutions, gases, etc. are applicable. The composition ratio of a composition is the ratio of one or more substances constituting the composition to the whole composition. When the composition is solid, the composition ratio is the weight ratio or weight% of the constituent substances, and when it is liquid, it is the weight ratio or weight%, or the volume ratio or volume%. When the composition is gas, the composition ratio is the volume ratio or volume%.

[0048] The multivariate analysis method is a general term for statistical techniques used when a plurality of variables (or characteristics) affect each other. There are various methods for the multivariate analysis method, and common methods include multiple regression analysis, principal component analysis, independent component analysis, factor analysis, cluster analysis, etc.

[0049] An observation means is a means for obtaining an observation result, such as measurement, analysis, or observation. More specifically, it includes measurement of the light intensity of infrared rays with a predetermined wavelength by the infrared absorption method, measurement of a spectrum indicating the light intensity over the entire predetermined wavelength range, measurement of the radiation luminance of infrared rays emitted from an object, energy-dispersive X-ray analysis, observation of the tensile strength of a composition, etc. Multiple observation means refer to cases where a physical quantity to be observed (such as the wavelength of electromagnetic waves, frequency, amount of heat for heating an object, acceleration voltage of an electron beam, magnitude of tensile force, etc.) is changed and observed multiple times with one observation means to obtain multiple observation results, or cases where observations are made with multiple observation means (changing the measurement method (infrared absorption method, chromatography, mass spectrometry, etc.)) to obtain multiple observation results. In the present invention, the former case is referred to.

[0050] A calibration curve is, so to speak, a conversion function used to convert a value indicating an observation result into a value in a desired state. In the present invention, a calibration curve is used to determine the composition ratio of a composition from the value indicating the observation result.

[0051] <Embodiment 1 Sample Preparation Step (S0101)> The "Sample Preparation Step" (S0101) performs a process of preparing a plurality of samples with known composition ratios. For factors considered to affect the observation results described later under the sample creation conditions, a plurality of levels are set, and samples with a plurality of different composition ratios are prepared. For example, using a coating film obtained by applying a coating solution in which a substance for forming a target film is dispersed in a solvent as a sample, a plurality of levels are set with factors such as the amount of the target substance (or the solid content concentration indicating the ratio contained in the coating solution) and the film thickness of the coating solution, and corresponding samples are created.

[0052] The known composition ratio is the solid content concentration of the coating film in the example of the lithium-ion battery positive electrode manufacturing process shown in FIG. 3 above. Another example is the composition ratio of carbon and a binder that holds carbon on a substrate in a coating film obtained by applying and drying a solution containing the same.

[0053] <Embodiment 1 Observation Result Acquisition Step (S0102)> The "Observation Result Acquisition Step" (S0102) performs a process of acquiring a value indicating an observation result obtained by changing an observation physical quantity for each prepared sample.

[0054] The observation physical quantity is a physical quantity used in an observation means to observe a sample and obtain an observation result. "Observation" is broadly performed by electromagnetic waves, and the observation physical quantity means the frequency (or wavelength) of the electromagnetic waves used in the observation. For example, when using an observation means (infrared absorption method) of irradiating a sample with infrared rays and detecting the infrared rays emitted from the sample and measuring its light intensity, the observation physical quantity is the wavelength when measuring the light intensity, and the observation result is the light intensity. For infrared measurement, infrared sensors used in known infrared absorption methods such as pyroelectric type and thermoelectric power type can be used. In this example of the infrared absorption method, the observation result obtained by changing the observation physical quantity is the light intensity measured by changing the measurement wavelength. In order to select an absorption wavelength specific to a substance or a functional group contained in the substance and a reference wavelength for reference of the state of the light source and the sample, measurements are performed by changing the absorption wavelength and the reference wavelength.

[0055] As another example, when irradiating a sample with laser light and observing the degree of temperature rise on the surface of the irradiated sample, the observation means is temperature measurement, the observation physical quantity is the energy density of the laser light, the irradiation time, or the type (wavelength) of the laser light, and the observation result is the radiation luminance of infrared rays. From the obtained infrared radiation luminance, it is possible to convert to temperature or calculate the rate of temperature rise using a calibration curve. The reference wavelength is a wavelength that is far enough away from the absorption peak of the measurement target substance so as not to be affected by the decrease in light intensity due to the absorption peak, and it is preferable to select a wavelength that is not more than 10 times the half-value width of the absorption peak away from the absorption wavelength so as not to be too far away from the absorption wavelength. More than one reference wavelength can also be used.

[0056] The value indicating the observation result may be the value of the observation result itself obtained by the observation means by changing the observation physical quantity, or may be a value obtained based on the observation result obtained by changing the observation physical quantity. For example, it may be a value calculated using a plurality of observation results obtained by changing the observation physical quantity.

[0057] In the example where the infrared absorption method described above is used as the observation means, measuring the reference wavelength in addition to the absorption wavelength is to eliminate the influences other than the light absorption specific to the measurement target substance. For example, it is the influence caused by fluctuations such as external light, measurement distance, dust in the optical path, surface roughness of the composition being measured, and fluctuations in the absolute value of the light quantity due to external factors such as variations and fluctuations in the luminance of the irradiation light source, and variations and fluctuations in the sensitivity of the sensor. Since it is considered that the reference wavelength is also affected by the external factors in the same way as the absorption wavelength, the influence is canceled by the measurement result at the reference wavelength.

[0058] <Embodiment 1 Calibration Curve Candidate Acquisition Step (S0103)> The "calibration curve candidate acquisition step" (S0103) performs a process of acquiring a plurality of calibration curve candidates for each observation physical quantity from which the observation result is obtained, from the known composition ratio for each sample and the observation result obtained from the sample. Note that the plurality of calibration curve candidates may be processed so as to be acquired for each combination of a plurality of observation physical quantities instead of for each observation physical quantity.

[0059] In the infrared absorption method which is an example of the observation means, for samples with known composition ratios, the measurement wavelength (absorption wavelength, reference wavelength), which is the observation physical quantity, is changed, and the light intensity (observation result) at each measurement wavelength is measured. From the known composition ratio and the measurement results of a plurality of light intensities with the measurement wavelength changed, a plurality of calibration curve candidates are acquired for each combination of the measured absorption wavelength and reference wavelength. The acquired calibration curve candidates are calibration curves for obtaining the composition ratio from the measurement results of the light intensity for each combination of the measured absorption wavelength and reference wavelength, and are candidates for selecting the calibration curve to be actually used for measuring samples with unknown composition ratios. Examples of the method from the acquisition of candidates to the determination of the calibration curve to be used will be described later.

[0060] <Embodiment 1 Residual Evaluation Step (S0104)> The residual evaluation step (S0104) performs a process of evaluating the residual using a residual evaluation rule which is a rule for evaluating the residual for each of the acquired calibration curve candidates.

[0061] The evaluation of the residuals is performed using, for example, for each of the obtained calibration curve candidates, the difference between the known composition ratio of the measured (observed) sample and the predicted composition ratio, which is the composition ratio predicted using the calibration curve candidate from the observation results obtained from the sample, i.e., the residual.

[0062] Examples of residual evaluation rules include a rule of calculating the sum of squared residuals obtained by squaring and summing the residuals of each data used for obtaining the calibration curve candidate and raising the priority of the calibration curve candidate with a smaller sum of squared residuals, a rule of raising the priority of the calibration curve candidate with a smaller mean squared residual obtained by dividing the sum of squared residuals by the number of data used for obtaining the calibration curve candidate (especially used when the number of observation results is different for each changed observed physical quantity), and a residual evaluation rule of raising the priority of the calibration curve candidate with a smaller sum of squared residuals (or mean squared residual) obtained from the calibration curve candidate obtained by excluding data of some levels of the factors of the sample preparation conditions or / and the observed physical quantity and the data excluded. Using the sum of squared residuals (or mean squared residual) obtained from the calibration curve candidate obtained by excluding a part of the data and the excluded data is for preventing overfitting when obtaining the calibration curve candidate using all the data and for verifying the calibration curve candidate.

[0063] <Embodiment 1 Calibration Curve Determination Step (S0105)> The "Calibration Curve Determination Step" (S0105) performs a process of determining, as the calibration curve to use, a predetermined calibration curve candidate using a selection rule for selecting the calibration curve candidate whose residuals were evaluated from the results of the residual evaluation.

[0064] Examples of selection rules include a rule determined based on the observed physical quantity during observation, a rule determined based on the evaluation result of the bias of the residuals, and a rule combining the two types of rules. Although multiple selection rules are conceivable, in addition to using them individually, one or more selection rules may be combined and used. The combination method and combination order of the selection rules can also be selection rules. Some specific examples of the selection rules will be introduced in Examples 3 and 4 below.

[0065] <Embodiment 1 Use calibration curve determination step: Selection rule: Reference wavelength> As an example of the selection rule, prioritization is performed by the residual evaluation rule so that the priority of the calibration curve candidate with a smaller sum of squared residuals or mean squared residual from the result of the residual evaluation is higher, and from the calibration curve candidate with the higher priority, based on the observed physical quantity (for example, combination of wavelengths) for which the calibration curve candidate was obtained, a rule for determining the calibration curve to be used can be considered.

[0066] <Specific example up to the determination of the use calibration curve in Embodiment 1> <In the case of the positive electrode active material slurry in the lithium-ion battery positive electrode manufacturing process of Embodiment 1>

[0067] For measures against global warming, electric vehicles have attracted attention and their production volume has increased significantly in recent years. Since lithium-ion batteries are used in electric vehicles, the production volume of the batteries has also increased significantly accordingly. As one of the manufacturing methods of the positive electrode of a lithium-ion battery, a slurry which is a mixture of a positive electrode active material (for example, NCM: lithium nickel cobalt manganate), a binder (for example, PVdF; polyvinylidene fluoride), a conductive assistant (for example, acetylene black), and a solvent (for example, NMP; N-methyl-2-pyrrolidone) is applied to a strip-shaped metal foil to form it.

[0068] Among the substances constituting the positive electrode active material slurry, those other than the solvent NMP are solid components, and when the solvent volatilizes after coating, they remain as a film on the metal foil. Even if the slurry is coated with the same film thickness, if the solid content concentration is different, the resulting film thickness after solvent volatilization will be different, which will affect the final battery performance. Therefore, in a manufacturing site where the slurry is being applied to continuously flowing parts, managing the solid content concentration of the applied film is important for maintaining quality. The infrared absorption method that uses the light absorption wavelength specific to functional groups and the like contained in the materials constituting the parts for solid content concentration measurement can measure the infrared light intensity non-destructively in a short time and is a measurement method that satisfies the above conditions.

[0069] Taking the measurement of the solid content concentration of the coating film in the positive electrode manufacturing process of the lithium-ion battery shown in Fig. 3 as an example, the method for obtaining the calibration curve used in the present invention will be specifically described.

[0070] <Embodiment 1 Conventional Problems in the Lithium-Ion Battery Positive Electrode Manufacturing Process> As the positive electrode active material of the lithium-ion battery, a general positive electrode active material of a lithium-ion battery can be used. For example, as described above, a slurry obtained by mixing lithium nickel cobalt manganate as an active material, PVdF as a binder, acetylene black as a conductive assistant, and NMP as a solvent is applied onto an Al foil substrate to form a coating. In the manufacture of lithium-ion batteries, if the solid content concentration is different even when the positive electrode active material slurry is applied with the same film thickness, the resulting film thickness will be different, which will affect the final battery performance. However, for the reasons described below, it was impossible to measure the solid content concentration in the production line immediately after coating, and the determination was made based on the performance evaluation of the battery after the product was completed. Therefore, even positive electrodes that should originally be removed as insufficient quality had to be treated as good products and used in product assembly for commercialization, making it difficult to stably manufacture batteries that meet the desired performance.

[0071] To measure the solid content concentration of the coating film coated with the positive electrode active material slurry in the production line, the infrared absorption method can be a candidate for quality control inspection because it is non-contact and non-destructive. However, there were the following problems. (1) Insufficient transmitted light intensity and reflected light intensity Since acetylene black, which is one of the materials constituting the positive electrode active material slurry for manufacturing lithium-ion battery positive electrodes, is black, the coating film becomes black and the transmitted light intensity and reflected light intensity of infrared rays are very small. (2) Proximity of the absorption wavelengths of the solvent and the solute material When measuring the solid content concentration immediately after coating (before drying) by the infrared absorption method, it is difficult to directly measure the solid content concentration of a mixture of multiple substances. Therefore, the ratio of NMP, which is a solvent, is determined, and the remainder is calculated as the solid content concentration. However, as shown in Fig. 5, the infrared absorption wavelengths of NMP, which is a solvent, and PVdF, which is a binder, are close to around 3.4 μm and around 3.3 μm (the spectra of each alone. PVdF is that of the solid state. Source: NIST Chemistry WebBook. https: / / webbook.nist.gov / chemistry), making it difficult to measure only NMP.

[0072] Regarding the problem in (1) above, since the light intensity of infrared rays due to transmission and reflection from the coating film is small, a preamplifier for amplifying the signal from the infrared sensor is provided as a countermeasure. Regarding (2), as will be described below, it was solved by a method of determining the calibration curve to be used from among the calibration curve candidates after obtaining the calibration curve candidates.

[0073] <Embodiment 1 Method for obtaining the calibration curve to be used: Pre-measurement: Measurement wavelength> In the production line as shown in Fig. 3, a method for irradiating infrared rays onto the measurement target and obtaining the solid content concentration of the coating film (before drying) on the Al foil, which is the base material, will be described. As described above, since the solid content concentration cannot be directly obtained, the ratio of the solvent is determined, and the solid content concentration is calculated by subtracting that ratio from the whole.

[0074] In the measuring device shown in FIG. 3, instead of a spectroscope for wavelength switching, it is configured to be able to switch a plurality of types of filters on the short wavelength side and the long wavelength side of the absorption wavelength centered on the absorption wavelength of the substance to be measured. A configuration having a spectroscope is costly, and there is also a problem that the light intensity of the desired wavelength becomes small because the light is split in either case (or a combination of both) when splitting the light irradiated to the measurement object or when splitting the reflected light from the measurement object. When it is desired to increase the light intensity of the reflected light as much as possible, it is better to adopt a configuration that does not use a spectroscope. In the configuration example shown in FIG. 3, since the composition of the measurement object contains acetylene black and has a low reflectance, due to the above two problems, it is configured not to have a spectroscope.

[0075] In the measurement, infrared rays of a predetermined wavelength are irradiated onto the coating film using a filter, and the reflected light is detected by a sensor to obtain the light intensity of the reflected light. It is preferable to configure the device so that the filter of the wavelength to be used is automatically set during the measurement. In the case of pre-measurement, after setting the sample and pressing the measurement start button, it is preferable to configure it to automatically measure the reflected light intensity corresponding to all types of filters while rotating a disk equipped with a plurality of filters to switch the filters. The wavelength to be irradiated is the absorption wavelength of NMP, which is the solvent contained in the coating film to be measured, and light of a wavelength shorter than or / and longer than the absorption wavelength is also irradiated as a reference wavelength, and the light intensity of the reflected light is measured respectively. Measuring the light intensity at the latter reference wavelength is to determine whether the ratio of the solvent has decreased or increased, or whether it is due to other factors (such as the irradiated light from the light source, the sensitivity of the sensor, etc.) when the light intensity of the reflected light at the absorption wavelength increases or decreases each time of measurement, and to eliminate the influence of that factor.

[0076] If the approximate value of the absorption wavelength specific to the material contained in the measurement target in advance is known, two or three types near that wavelength are used as candidates for the absorption wavelength, and several types of wavelengths are selected on the shorter wavelength side and the longer wavelength side of the absorption wavelength as candidates, and the corresponding filters are set in the filters of FIG. 3. The best combination of wavelengths within the selected wavelengths is narrowed down by the method described later, and among the calibration curve candidates for calculating the solid content concentration from the light intensity measured using each wavelength, which one to use is determined. Along with the determination of the calibration curve to be used, the combination of the absorption wavelength candidates and the reference wavelength candidates used in the measurement is determined as the wavelength combination for actual measurement.

[0077] The filter disk of FIG. 3 is formed by making a plurality of circular holes (e.g., 9 holes) on concentric circles of a disk that rotates around the central axis and fitting filters with different transmission wavelengths. During measurement, the filter disk is rotated, and the filter with the required wavelength is appropriately selected and automatically inserted into the optical path from the light source. FIG. 6 shows 9 types of wavelengths as examples of the absorption wavelength and the reference wavelength. It has 3.39 μm and 3.43 μm, which are wavelengths near the absorption wavelength of 3.4 μm of NMP. For the 3.39 μm filter, two types of filters with different half-value widths are provided. The types of the filters to be set are not limited to 9.

[0078] <Embodiment 1 Method for obtaining calibration curve for use: Preliminary measurement: Measurement sample> In order to select the set of wavelengths to be used in the actual measurement, for samples with known specifications such as solid content concentration and coating film thickness, the light intensity of the reflected light is measured for each of the 9 types of wavelengths. For this purpose, samples with combinations of levels as shown in FIG. 7 were prepared. The levels are the levels of three factors: the level (3 levels) of changing the ratio of NMP as the solvent and PVdF close to the absorption wavelength in the solute, the level (5 levels) of changing the solid content concentration, and the level (3 levels) of changing the film thickness applied to the substrate, resulting in 45 types of level combinations of samples.

[0079] <Embodiment 1 Method for obtaining calibration curve for use: Preliminary measurement: Ratio calculation value> Since the absorption peak of NMP is around 3.4 μm as described above, among the nine wavelengths shown in Fig. 6, three wavelengths near 3.4 μm, namely 3.39 μm (two types with half-value widths of 87 nm and 138 nm) and 3.43 μm, were taken as candidate absorption wavelengths, and the light intensity was measured using two wavelengths from the remaining six wavelengths as reference wavelengths (Fig. 8). When one of the three candidate absorption wavelengths is selected as the absorption wavelength, the remaining two wavelengths may also be treated as candidates for reference wavelengths.

[0080] When it cannot be said that the light intensities of the reference wavelengths arranged on the short-wavelength side and the long-wavelength side with respect to the absorption wavelength are substantially equal, there is a concern that the error from the baseline of the spectrum will be large when obtaining the absorbance from the light intensity at the absorption wavelength of the measurement target substance NMP. When there are two reference wavelengths, if both wavelengths on the shorter wavelength side or the longer wavelength side with respect to the absorption wavelength are used, there is a similar concern that the error will be large. If the baseline is approximately constant in the region including the absorption peak, two wavelengths may be selected on the short-wavelength side or the long-wavelength side.

[0081] Furthermore, the contribution degrees of the two reference wavelengths are set.

[0082] Particularly, as shown in Fig. 5, when there is an absorption peak of PVdF near the short-wavelength side of the absorption peak of NMP to be measured, the light intensity measurement result will be a value according to the spectrum synthesized from the spectra of NMP and PVdF. The spectrum of NMP is almost horizontal on both sides of the absorption peak at an absorption wavelength of 3.4 μm, while the spectrum of PVdF has a slightly downward slope in transmittance (the light intensity decreases as going to the long-wavelength side) centered around the absorption peak near 3.3 μm. Therefore, one wavelength on each side of the absorption wavelength of 3.4 μm is used as the reference wavelength, but the two reference wavelengths are not treated equally, and weights are multiplied by the contribution degrees respectively. The contribution degrees of the two wavelengths are from 0 to 1 in steps of 0.1, and are configured such that the sum of the two contribution degrees is 1. Therefore, the contribution degree can take 11 values (note that it is not limited to steps of 0.1). Therefore, considering the contribution degree, the combination of the absorption wavelength and the reference wavelength is 3 wavelengths of absorption wavelength × reference wavelength 6 C 2The contribution degree of 11 types = 3×15×11 = 495 cases.

[0083] Absorption wavelength λ abs The light intensity measured using the light of is Y_λ abs The reference wavelength λ 1 The light intensity measured using the light of is Y_λ 1 The reference wavelength λ 2 The light intensity measured using the light of is Y_λ 2 Then, the relationship of the light intensities at the three wavelengths related to the ratio of the solvent NMP is expressed as the following formula 1 using the coefficient a 1 and a 2 as follows. R NMP = 1 - Y_λ abs / (a 1 Y_λ 1 + a 2 Y_λ 2 ) Formula 1 As described above, in the above formula 1, a 1 + a 2 = 1. Formula 1 is a value obtained by correcting the light intensity at the absorption wavelength using the light intensities at two reference wavelengths in order to exclude the influence of external factors (0 < R NMP < 1). And if the film thickness of the coating film is constant, R NMP correlates with the ratio containing NMP.

[0084] Since the ratio of the solvent NMP and the solid content concentration in the coating liquid is in a relationship where their sum is 1, the calibration curve candidate between R NMP in formula 1 that correlates with the ratio of NMP and the solid content concentration will be obtained hereafter. Since the correlation between R NMP and the solid content concentration is a negative correlation, the calibration curve candidate will be a downward-sloping straight line on the right shoulder as shown in FIG. 10 described later.

[0085] FIG. 9 shows, as an example, 30 out of 495 results obtained by calculating the ratio values for samples with a coating film thickness of 100 μm for each solid content concentration using the above formula 1 from the measurement results in FIG. 8.

[0086] <Embodiment 1 Method for obtaining calibration curve: Pre-measurement: Ratio calculation value and solid content concentration> For each ratio of PVdF and coating film thickness, for each combination of wavelengths, the ratio calculation value R expressed by Equation 1 NMP and the relationship with the solid content concentration of the measured sample are plotted as a graph with the horizontal axis: ratio calculation value R NMP , vertical axis: solid content concentration. Fig. 10 shows, as an example, a graph showing the relationship between the ratio calculation value and the solid content concentration at a film thickness of 100 μm with the wavelength combination described in the top row of Fig. 9. Such graphs can be obtained for each coating film thickness in 495 types of wavelength combinations as calculated above. For each graph, in the example of this lithium-ion battery positive electrode, an approximation line is drawn as a linear approximation. This approximation line is a calibration curve candidate for the wavelength combination. Depending on the system for obtaining the calibration curve candidate, an nth-order approximation such as a quadratic or cubic approximation may be used instead of a linear approximation.

[0087] <Embodiment 1 Method for obtaining calibration curve: Pre-measurement: Residual evaluation and selection of calibration curve to be used 1> Using the calibration curve candidate, the solid content concentration predicted for the ratio calculation value R NMP is obtained, and the sum of squares of the differences (residuals) from the measured values and / or the mean square error of the residuals is calculated. Fig. 11 shows the results corresponding to Fig. 9. In Fig. 11, the sum of squares of the residuals is shown in the rightmost column of the table. Considering the magnitude of the sum of squares of the residuals and / or the mean square error of the residuals and the magnitude relationship of the three selected wavelengths (absorption wavelength and two reference wavelengths), the calibration curve to be used is selected. The magnitude relationship of the wavelengths when two reference wavelengths are used means that one reference wavelength is located on the shorter wavelength side and one reference wavelength is located on the longer wavelength side with respect to the absorption wavelength, and the absorption wavelength is located between the two reference wavelengths.

[0088] <Embodiment 1 Method for obtaining calibration curve: Selection of calibration curve to be used 2: Cross-validation> <Embodiment 1 Method for obtaining calibration curve: Selection of calibration curve to be used 2: Excluding part of measurement data> When a calibration curve is obtained using all the measurement data, overfitting may occur due to the calibration curve candidates. When overfitting occurs, there is a possibility that the deviation becomes large and the residual becomes large for newly measured data. Cross-validation is performed to prevent overfitting. For each factor, calibration curve candidates are obtained by excluding a part of the measurement data for each level. The exclusion of the above data excludes about 20 to 40% of the number of levels of the measurement samples. For example, for the levels shown in FIG. 7, when there are three levels such as the ratio of PVdF and the film thickness, one level is excluded, and when there are five levels such as the solid content concentration, one or two levels are excluded.

[0089] Verification is performed by obtaining the mean square error of residuals (or the sum of squared residuals) between the excluded data and the predicted values for the excluded data obtained using the calibration curve candidates. Among the calibration curve candidates with a small mean square error of residuals for each combination of measurement wavelengths, those with an appropriate magnitude relationship of the combination of measurement wavelengths are selected. The appropriate magnitude relationship of the combination of measurement wavelengths means the magnitude relationship in which the reference wavelengths are located on the short wavelength side and the long wavelength side with respect to the absorption wavelength as described above.

[0090] <Embodiment 1 Acquisition method of calibration curve: Selection of calibration curve to be used 2: Residual evaluation and determination of calibration curve to be used> As described above, in order to avoid overfitting, cross-validation was performed in which a calibration curve was created by excluding a part of the levels and verified with the data of the excluded levels. FIG. 12 shows 30 results obtained by calculating the mean square error of residuals for each combination of wavelengths and arranging them in ascending order, and FIG. 13 shows the 30 results as a bar graph (the numbers on the horizontal axis correspond to the order at the left end of FIG. 12). The combination of wavelengths with the smallest mean square error of residuals is the absorption wavelength λ abs is 3.39 μm, the half-value width is 138 μm, the reference wavelength λ 1 is 3.70 μm (contribution coefficient 0.1), and the reference wavelength λ 2 is 2.10 μm (contribution coefficient 0.9). This combination has the reference wavelengths λ abs (3.70 μm) and λ 1 with respect to the absorption wavelength λ 2(2.10 μm) is a desirable combination of wavelengths located on the long-wavelength side and the short-wavelength side.

[0091] Thus, the calibration curve to be used is selected from among the calibration curve candidates. The above has been described by taking, as an example, the method for selecting a calibration curve when using the infrared absorption method for solid content concentration measurement, but it is also applicable to other systems. For example, X-ray diffraction data is acquired at different angles, diffraction data is acquired, and calibration curve candidates for estimating the parameters of the crystal structure are generated. Examples such as selecting the optimal calibration curve by residual evaluation and cross-validation are also conceivable.

[0092] <Embodiment 1 Flow of Processing> FIG. 1 shows a flowchart of the method for obtaining a calibration curve according to Embodiment 1. The predetermined number k in the figure is the number of calibration curve candidates to be obtained. In the above-described example, the predetermined number k is 495, which is the number of wavelength combinations. The acquisition of the calibration curve according to Embodiment 1 is performed along the flowchart of FIG. 1, The sample preparation step (S0101) performs the above-described processing, The observation result acquisition step (S0102) performs the above-described processing, The calibration curve candidate acquisition step (S0103) performs the above-described processing, Until the predetermined number k is reached, the process returns to the observation result acquisition step (S0102). After repeating the process k times to obtain k calibration curve candidates, The residual evaluation step (S0104) performs the above-described processing, The used calibration curve determination step (S0105) performs the above-described processing. Among these processes, at least the processes from the observation result acquisition step (S0102) to the used calibration curve determination step (S0105) are performed on a computer.

[0093] <Embodiment 1 Description of Hardware> The hardware of the computer that performs processes such as the acquisition of observation results, the acquisition of calibration curve candidates, the residual evaluation, and the determination of the used calibration curve as described above will be described. The hardware configuration of the computer part of the synchrotron radiation thermometer in this Embodiment 1 will be described with reference to FIG. 2.

[0094] Figure 2 is a diagram showing the hardware configuration of the computer part in Embodiment 1. As shown in this figure, an example of the computer part in this embodiment includes an "MPU" that performs various arithmetic processes, a "main memory", a "non-volatile memory" that holds various programs and data (information), a "user I / F", and "USB, I 2 C, SPI, etc.", and a "LAN I / F". For example, an "infrared sensor" is connected via "USB, I 2 C, SPI, etc.". And they are interconnected by a "system bus" shown as a thick line in the figure to transmit, receive, and process information. Note that this computer part can also have a configuration similar to a known PC as described above.

[0095] Various programs and data (information) stored in the non-volatile memory are expanded into the main memory when the system is started, and are configured such that the MPU sequentially performs operations using the data according to received execution instructions.

[0096] When the system is started, various programs and data (information) stored in the "non-volatile memory" are read out, expanded, and stored in the "main memory", and at the same time, a work area that is also the working area of the program is provided. The MPU sequentially performs operations using the data according to received execution instructions. Note that a plurality of addresses are assigned to each of this "main memory" and "non-volatile memory", and the programs executed by the MPU can exchange data with each other by specifying and accessing those addresses to perform processing.

[0097] The programs stored in the "main memory" in this embodiment are an observation result acquisition program, a calibration curve candidate acquisition program, a residual evaluation program, and a used calibration curve determination program. Also, observation results, calibration curve candidates, sum of squared residuals, used calibration curves, etc. are stored in the "main memory" and "non-volatile memory".

[0098] The "MPU" executes the observation result acquisition program stored in the "main memory" to obtain a value indicating the observation result obtained by changing the observation physical quantity for each sample of the prepared known composition ratio. Then, it executes the calibration curve candidate acquisition program stored in the "main memory" to obtain a plurality of calibration curve candidates for each observation physical quantity from which the observation result was obtained, based on the known composition ratio for each sample and the observation result obtained from the sample. The observation result acquisition program and the calibration curve candidate acquisition program are executed a predetermined number of times, k times. Then, it executes the residual evaluation program stored in the "main memory" and performs residual evaluation using the residual evaluation rule, which is a rule for evaluating the residual for each of the obtained calibration curve candidates. Then, it executes the used calibration curve determination program stored in the "main memory" and determines, using a selection rule for selecting a calibration curve candidate whose residual was evaluated from the result of the residual evaluation, a calibration curve to use a predetermined calibration curve candidate.

[0099] <Effect of Embodiment 1> The method for obtaining a calibration curve for detecting the composition ratio from the observation results obtained by a plurality of observation means in which the observation physical quantity is changed by the multivariate analysis method of the present Embodiment 1 can obtain and use a calibration curve candidate in a system where it is difficult to determine the composition ratio because it is difficult to obtain the observation result, and can determine the composition ratio of the composition.

[0100] <Outline of Embodiment 2> Mainly Claim 2 The method for obtaining the calibration curve of Embodiment 2 based on Embodiment 1 is configured such that the plurality of observation means is the observation of the reflected light intensity (or absorbance or transmitted light intensity, which is the same hereinafter) when light of a plurality of wavelengths is applied to the composition.

[0101] <Configuration of the Calibration Curve Acquisition Method of Embodiment 2> The method for obtaining the calibration curve of Embodiment 2 based on Embodiment 1 is the same as the flowchart of Embodiment 1 shown in FIG. 1. The difference lies in the observation of the reflected light intensity (or absorbance or transmitted light intensity, the same hereinafter) when a plurality of observation means irradiate light of a plurality of wavelengths onto the composition. The following describes the differences.

[0102] By measuring (observing) the reflection intensity when light of a plurality of wavelengths is applied to the composition and observing the phenomenon that the reflection intensity decreases at a specific wavelength, the presence of a functional group that absorbs light of the specific wavelength can be known (infrared absorption method). When the types of substances constituting the composition are known and the composition ratio is unknown, the composition ratio can be estimated by observing the degree of decrease in the reflection intensity at the absorption wavelength specific to each constituent substance. When the light irradiated on the composition is transmitted and not reflected, the light intensity of the transmitted light may be measured, or a mirror may be installed in the optical path after the composition is transmitted to measure the light intensity of the light transmitted through the composition again.

[0103] In Embodiment 2, when the infrared absorption method is used, the composition ratio of the composition can be obtained from a calibration curve based on the observation results using light with different wavelengths.

[0104] <Overview of Embodiment 3> Mainly Claim 3 The method for obtaining the calibration curve of Embodiment 3 based on either Embodiment 1 or 2 is the same as the flowchart of Embodiment 1 shown in FIG. 1 (or Embodiment 2 based on Embodiment 1). The method for obtaining the calibration curve of Embodiment 3 is configured such that when the composition is manufactured with a targeted composition ratio in advance, the wavelength of light for which a large change in the reflected light intensity is expected at the targeted composition ratio, and observations using light of wavelengths longer and shorter than that wavelength are included in at least a plurality of observation means.

[0105] <Configuration of the Calibration Curve Acquisition Method of Embodiment 3> In the calibration curve acquisition method of Embodiment 3 based on Embodiment 1 or Embodiment 2, when the composition is manufactured with a predetermined target composition ratio in advance, if the target composition ratio is met, a large change in the reflected light intensity is expected at the wavelength of light, and observations using light with wavelengths longer and shorter than that wavelength are included in at least a plurality of observation means.

[0106] The calibration curve acquisition method of Embodiment 3 is particularly used when detecting the composition ratio of a composition by measuring the light intensity at a predetermined wavelength by the infrared absorption method. When applying this Embodiment 3 to the infrared absorption method, the absorption wavelength specific to the substance for which the composition ratio is to be determined becomes the "wavelength of light at which a large change in the reflected light intensity is expected if the target composition ratio is met". The light with wavelengths longer and shorter than that wavelength is set to the wavelengths located on both sides of the absorption peak having the absorption wavelength at the bottom of the valley in the spectrum showing the light intensity in the entire wavelength range with the absorption wavelength as the approximate center. It is preferable that the light with short wavelength and long wavelength is light with a wavelength that does not cross the valley of the absorption peak (at least separated by more than the half-value width of the absorption peak) in the spectrum.

[0107] It is also possible to use not two but two or more of the above reference wavelengths. Even when using two or more, at least one reference wavelength is selected on either the short wavelength side or the long wavelength side with respect to the absorption wavelength.

[0108] The reason for preferably selecting reference wavelengths from both the short wavelength side and the long wavelength side with respect to the absorption wavelength will be explained below. The decrease in the light intensity other than the infrared absorption of the measurement target substance may vary depending on the wavelength. For example, due to the absorption of other substances such as moisture in the optical path, dust in the optical path, the surface roughness of the sample, and the difference in the content component ratio of the composition for each sample, the light intensity may change on the short wavelength side and the long wavelength side of the absorption wavelength λ abs An example of the case where the decrease in the light intensity other than the infrared absorption is not constant in the wavelength range near the absorption wavelength including the absorption wavelength of the measurement target substance is shown in the spectrum schematic diagram of FIG. 14. FIG. 14 is a diagram with the wavelength on the horizontal axis and the transmittance on the vertical axis. The transmittance and the change in the light intensity in the explanation so far are almost synonymous.

[0109] In FIG. 14, the transmittance is higher on the short-wavelength side. If the wavelengths λ a and λ a ' on the short-wavelength side are used as reference wavelengths, λ a , λ a ' the ratios of the decrease in light intensity other than the infrared absorption (decrease in transmittance) are respectively (1 - T a ), (1 - T a '). Similarly, for λ b , λ b ' on the long-wavelength side, they are (1 - T b ), (1 - T b '). In the example of FIG. 14, the decrease in light intensity other than the infrared absorption is greater on the long-wavelength side (1 - T a ) < (1 - T b ). Therefore, in the spectrum of FIG. 14, it is considered that the transmittance decreases as going to the long-wavelength side, and the influence other than the infrared absorption of the measurement target substance strengthens and the transmittance decreases. In an example where the baseline slopes downward to the right like this, if the decrease in transmittance other than the infrared absorption at the absorption wavelength of the measurement target substance is defined as U(λ abs ), it is considered that the relationship (1 - T a ) < 1 - U(λ abs ) < (1 - T b ) holds.

[0110] When obtaining R NMP using the ratio calculation formula of Equation 1, the light intensity in Equation 1 can be treated synonymously with the transmittance. The ratio calculation formulas described in FIG. 14 are expressed in terms of transmittance rather than light intensity. When the wavelengths λ 1 , T 2 on the short-wavelength side are substituted into T a , λ a ' in the ratio calculation formula in FIG. 14, the ratio of NMP is calculated excessively in order to underestimate the decrease in transmittance other than the infrared absorption. Conversely, when the wavelengths λ b , λ b ' on the long-wavelength side are substituted into T 1 , T 2 , the ratio of NMP is calculated insufficiently. For λ a (λ a ' is also acceptable) on the short-wavelength side, λ b (λb By using “also possible” as the reference wavelength, it is possible to consider the influence of the decrease in the light intensity other than the infrared absorption of the measurement target substance at both wavelengths. From Equation 1, which is used by multiplying the contribution degrees (values between 0 and 1, and the sum of the contribution degrees of both wavelengths is 1) to the light intensities at both wavelengths, the absorption peak height at the absorption wavelength of the measurement target can be obtained more accurately.

[0111] When selecting the reference wavelength, the spectrum of the measurement target should be measured using a spectroscope and determined in consideration of the position of the absorption wavelength. However, the configuration of the measurement device having a spectroscope has the problem of high cost, and when splitting the light irradiated to the measurement target or the reflected light from the measurement target (either case or a combination of both), the light intensity of the desired wavelength becomes small because the light is split. When it is desired to increase the light intensity of the reflected light from the measurement target as much as possible, it is better to adopt a configuration without using a spectroscope. For the above reasons, when the measurement system does not have a spectroscope, the shape of the spectrum of the composition of the measurement target becomes unknown. Therefore, one or more candidate wavelengths of the reference wavelength are set on the shorter wavelength side and the longer wavelength side with respect to the absorption wavelength known from documents or the like as described in the above description of the present invention. A sample with a known composition ratio is measured, and the calibration curve candidate to be used is determined from the residual evaluation of the calibration curve candidates. The selection rule at the time of determining the calibration curve used in the third embodiment is a rule for selecting a calibration curve in which at least one or more reference wavelengths are selected from the shorter wavelength side and the longer wavelength side, respectively.

[0112] According to the third embodiment, by including observations using light of longer and shorter wavelengths than the wavelength of light for which a large change is expected in the reflected light intensity for the target composition ratio, the result of the observation by the wavelength of light for which a large change is expected can be corrected, and a more accurate observation result can be obtained.

[0113] <Embodiment 4 Outline> Mainly in claim 4 In the method for obtaining a calibration curve according to the fourth embodiment based on any one of the first to third embodiments, when the reflected light intensity of the composition changes due to factors other than the composition ratio, the observation at a wavelength with a low rate of being affected by the change is selected.

[0114] <Embodiment 4 Configuration of Calibration Curve Acquisition Method> In the calibration curve acquisition method of Embodiment 4, in the used calibration curve determination step, when determining a calibration curve using a predetermined calibration curve candidate, among the selection rules for selecting a calibration curve candidate whose residual is evaluated from the result of residual evaluation, there is a selection rule that when the light reflection intensity of the composition changes due to factors other than the composition ratio, the observation of wavelengths with a low rate of being affected by the change is selected. When the light reflection intensity of the composition changes due to factors other than the composition ratio, by using the selection rule of selecting the observation of wavelengths with a low rate of being affected by the change, it is possible to narrow down as candidates for determining a calibration curve using a calibration curve candidate that is less affected by factors other than the composition ratio. Note that the calibration curve acquisition method of Embodiment 4 has the same flowchart as Embodiment 1 shown in FIG. 1 or Embodiment 2 or 3 based directly or indirectly on Embodiment 1.

[0115] As a specific example of the case where the light reflection intensity (or the same applies to absorbance or transmitted light intensity) of the composition changes due to factors other than the composition ratio, it is a case of obtaining a calibration curve for detecting the solid content concentration as the composition ratio of a composition that is a solution in which a plurality of substances are mixed, such as the positive electrode active material coating process of the lithium ion battery exemplified in Embodiment 1. The composition consists of one type of solvent and a solute containing a plurality of types of substances, and as a method for detecting the solid content concentration, the ratio of the solvent is obtained and the rest is calculated as the solid content concentration. In this example, when attempting to measure the light intensity of the absorption wavelength of the solvent by the infrared absorption method, at least one of the substances constituting the solute has an absorption wavelength close to the absorption wavelength of the solvent, so the light reflection intensity of the composition changes near the absorption wavelength. In such a case, a selection rule is used to determine as the calibration curve to be used a calibration curve candidate obtained by measurement using a combination of wavelengths (absorption wavelength and reference wavelength) that is less affected by the absorption peak of the solvent.

[0116] <Embodiment 4 Used Calibration Curve Determination Step: Selection Rule: Residual Evaluation + Reference Wavelength> As selection rules, for example, a rule for determining a calibration curve using a calibration curve candidate with a small sum of squared residuals or mean squared residuals by residual evaluation can be considered. Further, for example, priority ranking is performed according to a residual evaluation rule so that the priority of a calibration curve candidate with a smaller sum of squared residuals or mean squared residuals is higher from the results of residual evaluation, and from the calibration curve candidate with the higher priority, based on the observed physical quantity (for example, combination of wavelengths) from which the calibration curve candidate was obtained, a rule for determining the calibration curve to be used can be considered. As an example based on the observed physical quantity, not only is the sum of squared residuals or mean squared residuals small, but for calibration curve candidates within the smallest 10, a rule for selecting a calibration curve candidate in which the reference wavelengths are arranged separated on the short-wavelength side and the long-wavelength side with respect to the absorption wavelength at which the light intensity was measured when the calibration curve candidate was obtained can be considered. In the above rule, a selection rule can also be considered in which the reference wavelength is a wavelength that is 1 to 10 times, preferably 2 to 10 times, away from the absorption wavelength by one half-value width of the absorption peak.

[0117] <Embodiment 4 Calibration Curve Determination Step: Selection Rule: Reference Wavelength and Residual Bias> When there are a plurality of calibration curve candidates in which the reference wavelengths are separated into the short-wavelength side and the long-wavelength side with respect to the absorption wavelength, the one with the smaller sum of squared residuals or mean squared residuals is determined as the calibration curve to be used. Or as a selection rule, in addition to the criterion of a calibration curve candidate with a small sum of squared residuals or mean squared residuals, a yy plot created with the known composition ratio on the horizontal axis and the predicted composition ratio obtained from the calibration curve on the vertical axis, a residual plot created with the predicted composition ratio predicted from the calibration curve for the observation results on the horizontal axis and the residual on the vertical axis, or a qq plot created with the value of the quantile of the normal distribution corresponding to the predicted composition ratio predicted from the calibration curve for the observation results on the horizontal axis and the value obtained by dividing the residual for the observation results by the standard deviation of the residual on the vertical axis, a rule for excluding calibration curve candidates with a bias in the residuals using any one or more of them can be considered.

[0118] <Embodiment 4 Calibration Curve Determination Step: Selection Rule: Residual Bias: yy Plot> Regarding the yy plot, a straight line passing through the plotted data is obtained by the least squares method. The range of the known composition ratio is divided into three parts, and from the smaller known composition ratio, they are defined as regions A, B, and C. For each of the regions A, B, and C, the values of the predicted composition ratio for the known composition ratio are described in a box plot. For the straight line obtained by the least squares method, check for the presence of regions A, B, and C having a bias where more than half of the predicted composition ratios for which the box in the box plot (indicating the middle 50% of the predicted composition ratio) deviates in the same direction. For example, a selection rule can be considered where if only one of the regions has a bias, or if two of the regions and the remaining one region have biases on opposite sides with respect to the straight line obtained by the least squares method, it is determined that there is a bias. In the above description, the regions are divided into three, namely A, B, and C, but two or more are sufficient. Preferably, it is three or more. However, if it is divided too finely, it becomes difficult to judge the bias, so the number of regions to be divided is preferably five or less.

[0119] <Embodiment 4 Calibration Curve Determination Step: Selection Rule: Residual Bias: Residual Plot> Regarding the residual plot, the average value of the residuals is close to 0, the average of the residuals is half or less of the standard deviation of the residuals, preferably the average is 1 / 4 or less of the standard deviation of the residuals, and more preferably 1 / 10 or less of the standard deviation. Also, the residual plot may be corrected such that the vertical axis is the value obtained by dividing each residual by the standard deviation of the residuals with respect to the predicted composition ratio predicted from the calibration curve for the observation result. Then, it becomes easier to identify how much each residual deviates from the average based on the standard deviation σ centered around 0. For example, a rule can be considered where the condition for selection is that there are no residuals exceeding 3σ. Or, similar to the yy plot, when the predicted composition ratio on the horizontal axis is divided into two or more regions, for example, three regions, the average value of the residuals in each region is at the same distance from 0. The distance is at most the standard deviation of the residuals, preferably half or less, and more preferably 1 / 4 or less.

[0120] <Embodiment 4 Calibration Curve Determination Step: Selection Rule: Residual Bias: qq Plot> The qq-plot is created to determine whether two distributions to be compared are similar. When the two distributions arranged on the horizontal and vertical axes have similar distributions, the graph will be a straight line. If both the predicted composition ratio predicted from the calibration curve for the observation results and the residuals for the observation results are normally distributed, the graph will be a straight line; otherwise, if either is not normally distributed, the graph will not be a straight line but a curve. The qq-plot also preferably divides the region into two or more, preferably three or more, and observes the deviation and the magnitude of the scatter from the straight line where the values on both axes are equal. When the straight line is divided into three or more regions, a rule can be considered such that if the deviation from the straight line in any region exceeds the 10th percentile of the normal distribution, it is determined that the bias is large.

[0121] <Embodiment 4 Use calibration curve determination step: Selection rule: Determination coefficient when the calibration curve is a straight line> When the calibration curve for the observation results is obtained by the least squares method using a linear approximation formula, the closer the determination coefficient is to 1, the smaller the deviation of the predicted composition ratio from the calibration curve. The determination coefficient is 0.9 or more, preferably 0.92 or more, and more preferably 0.95 or more.

[0122] By using this Embodiment 4, when the light reflection intensity of a sample as a composition changes due to factors other than the composition ratio, it is possible to determine the calibration curve to be used as the calibration curve obtained by measurement using a wavelength with less influence from the change, and it is possible to use a calibration curve that can detect a more accurate composition ratio.

[0123] <5. Effects>

[0124] With the measuring device of the present invention having the above configuration, observation results (e.g., light intensity) in which the observed physical quantity (e.g., solid content concentration) is changed for a sample with a known composition ratio are obtained, and a plurality of calibration curve candidates are obtained for each sample from the observed physical quantity and the observation results. As a result of selection using a residual evaluation rule and a selection rule (including physical rules), a calibration curve to be used is determined as the calibration curve to be used from within the calibration curve candidates, thereby providing a calibration curve determination method. Further, in addition to the light intensity at the absorption wavelength, the contribution degrees of the light intensities at wavelengths shorter and longer than the absorption wavelength are also optimized to obtain calibration curve candidates, thereby providing calibration curve candidates that are even more excellent in the result of residual evaluation and providing a method for obtaining a calibration curve in which a more excellent calibration curve candidate is determined as the calibration curve to be used.

[0125] In particular, by the method for obtaining a calibration curve having the above configuration, in the case where the measurement target is a coating film using a black material with a low infrared reflectance, or in a coating film (e.g., the positive electrode of a lithium ion battery) for which the solid content concentration is to be measured, when a material having an absorption peak close to the solvent to be measured is included, or when the above two cases overlap, a calibration curve to be used can be obtained in a short time by the non-contact non-destructive infrared absorption method, and a solid content concentration measurement result can be obtained. In the manufacturing line, on-site, high-frequency quality control solid content concentration measurements can be performed either in full or by sampling, improving and maintaining the quality of components and suppressing the occurrence of defective products or second-class products that do not meet the desired performance as products incorporating the components.

Explanation of Signs

[0126] Sample preparation step... S0101 Observation result acquisition step... S0102 Calibration curve candidate acquisition step... S0103 Residual evaluation step... S0104 Calibration curve to be used determination step... S0105

Claims

1. A calibration curve acquisition method for detecting a composition ratio from observation results obtained by a plurality of observation means in which an observed physical quantity is changed by a multivariate analysis method to determine the composition ratio of a composition, comprising: a sample preparation step of preparing a plurality of samples with known composition ratios; an observation result acquisition step of obtaining a value indicating an observation result in which the observed physical quantity is changed for each of the prepared samples; a calibration curve candidate acquisition step of obtaining a plurality of calibration curve candidates for each observed physical quantity from which the observation result is obtained, from the known composition ratio for each sample and the observation result obtained from the sample; a residual evaluation step of performing a residual evaluation using a residual evaluation rule which is a rule for evaluating a residual for each of the obtained calibration curve candidates; a used calibration curve determination step of determining, using a selection rule for selecting a calibration curve candidate for which the residual has been evaluated from the results of the residual evaluation, a calibration curve to be used using a predetermined calibration curve candidate; A calibration curve acquisition method comprising the above steps.

2. The calibration curve acquisition method according to claim 1, wherein the plurality of observation means is an observation of the reflected light intensity (or the absorbance or the transmitted light intensity, which is the same hereinafter) when light of a plurality of wavelengths is applied to the composition.

3. The selection rule is such that when the composition is manufactured with a preconceived composition ratio, the wavelength of light for which a large change in the reflected light intensity is expected at the preconceived composition ratio, and observations using light of wavelengths longer and shorter than that wavelength are included in at least a plurality of observation means. The calibration curve acquisition method according to claim 2.

4. The selection rule according to claim 3 includes a selection rule of selecting an observation at a wavelength at which the ratio affected by the change is low when the reflected light intensity of the composition changes due to factors other than the composition ratio. Calibration curve acquisition method.

Images