Particle measuring instrument, particle measuring method, and storage medium

Abstract

The present application relates to a particle measuring instrument, a particle measuring method, and a storage medium, and provides a technique for reducing the measurable particle diameter of particles while improving counting efficiency. In a signal processing section, signal processing using stochastic resonance is performed on an input signal. The input signal from a data input section is added to k pieces of uncorrelated noise generated by a noise generation section in an adding section. Each signal after addition is binarized by a binarization section, and a simple average thereof is calculated by a calculation section. Components in a frequency band deviating from a target signal in an output signal corresponding to the calculation result are cut off by a filter section, and a determination is made in a determination section as to whether or not a particle threshold (presence of a particle) is exceeded. At the time of binarization, a binarization threshold automatically calculated by a binarization threshold calculation section in correspondence with the medium of the measurement sample is used, so that particles of a smaller particle diameter can be detected while reducing false detection.

Description

Technical Field

[0001] This invention relates to an apparatus and method for measuring particles present in a liquid sample, and more particularly to an apparatus and method for using random resonance phenomena when detecting minute particles. Background Technology

[0002] Stochastic resonance refers to the phenomenon where, by adding noise to a signal, the signal becomes stronger with a certain probability, thus increasing its response. It is a preferred method for detecting weak signals under the influence of noise. For example, methods are known for parallel processing of input signals containing weak signals to add uncorrelated noise to the signal (e.g., see Non-Patent Literature 1). In conventional signal processing utilizing this method, such as... Figure 9 As shown, in k parallel processing steps, the input signal is added to the noise to obtain each signal, which is then binarized using a fixed binarization threshold.

[0003] Furthermore, in the field of particle detection, signal processing methods that utilize random resonance to improve the detection accuracy (SN ratio) of small particles contained in the air are known (see, for example, Patent Document 1). In this signal processing, the SN ratio is optimized by using an electrical signal obtained by adding an analog signal (a weak signal with a known number of pulses and pulse waveform) to a noise signal.

[0004] In the field of liquid particle measurement, it is known that the minimum measurable particle size is set at 30 nm. However, with the further miniaturization of processes, such as semiconductors, it is required to measure particles with a diameter smaller than 30 nm. However, the minimum measurable particle size is approaching its limit in the extension of existing measurement methods. Furthermore, when researching new methods, the unique problem of measuring samples in liquids must be considered.

[0005] In other words, in the measurement of particles in liquid, the scattered light from the sample itself becomes a noise source. This scattered light is significantly larger than that from air, and its intensity varies depending on the sample and the measurement environment, such as temperature. Therefore, even when using stochastic resonance for particle measurement in liquid, it is difficult to achieve good particle measurement accuracy using only the conventional signal processing described above with a fixed binarization threshold. To ensure sufficient accuracy, the variation in the noise amplitude of the sample, which is mainly influenced by factors such as the sample's refractive index and isothermal compressibility, should be addressed, requiring adjustment of the binarization threshold.

[0006] Specifically, with a fixed binarization threshold, it needs to be set to a value larger than the assumed maximum noise level to eliminate false detections caused by noise from the sample's scattered light under all conditions. Furthermore, when the sample's scattered light is low, a smaller binarization threshold is needed to detect weak particle signals. Manually adjusting the binarization threshold for each measurement based on the sample is not only time-consuming but may also introduce errors during adjustment.

[0007] On the other hand, while existing technologies in the field of particle detection can reduce the influence of noise in the circuit system, it is difficult to reduce the influence of noise caused by scattered light generated by the sample. Furthermore, in existing technologies, adjustments must be made based on known pulse correlation information before measurement, which is time-consuming in preparation.

[0008] Furthermore, to utilize stochastic resonance, multiple uncorrelated noises must be prepared. It is known that in signal processing using stochastic resonance, the higher the parallel processing speed (k), the higher the detection accuracy of weak signals; therefore, k uncorrelated noises are required. However, the resources available for the measuring instrument are limited, making the preparation of these noises a challenge.

[0009] Existing technical documents

[0010] Patent documents

[0011] Patent Document 1: Japanese Patent No. 6531499

[0012] Non-patent literature

[0013] Non-patent literature 1: J.J. Collins et al., “Stochastic resonance without tuning”, Nature Vol. 376, pp. 236-238, July 20, 1995 Summary of the Invention

[0014] The problem that the invention aims to solve

[0015] Therefore, the technical problem of the present invention is to provide a technique for reducing the measurable particle size while improving counting efficiency.

[0016] Solution for solving the problem

[0017] To address the aforementioned problems, this invention employs the following liquid particle measuring instrument, liquid particle measuring method, and signal processing program. It should be noted that the text in parentheses below is merely illustrative, and this invention is not limited thereto.

[0018] That is, in the liquid particle measuring instrument, liquid particle measuring method, and signal processing program of the present invention, a predetermined number of parallel processing operations are performed on a signal of magnitude corresponding to the intensity of the scattered light to add a predetermined number of uncorrelated noises to the signal, and the summed signals are output. The scattered light is generated by the interaction between particles contained in the liquid sample and light incident on the liquid sample. Then, each signal is binarized using a binarization threshold set to match the liquid sample, and each binarized signal is output. A value based on each binarized signal is calculated and output. In the calculated output, a component of a predetermined frequency is first allowed to pass through; if the passed output exceeds a predetermined particle threshold, it is determined that particles are present.

[0019] According to this scheme, the level of the signal originating from the background light contained in the scattered light signal decreases, while the signal originating from particles can be highlighted relatively clearly, enabling reliable detection of weak signals originating from tiny particles. Furthermore, according to this scheme, since binarization is performed using a binarization threshold set to match the liquid sample, particles can be detected with good accuracy compared to using a fixed binarization threshold.

[0020] Preferably, in the aforementioned liquid particle measuring instrument, liquid particle measuring method, and signal processing program, a binarization threshold is calculated based on the characteristic quantities of the liquid sample, and the calculated binarization threshold is used for binarization. Furthermore, more preferably, the binarization threshold is calculated based on the RMS or DC value of the scattered light component of the liquid sample. The RMS or DC values ​​will be explained later.

[0021] According to this scheme, the binarization threshold is automatically calculated and set based on the characteristic quantities of the liquid sample, such as the RMS value or DC value of the scattered light component of the liquid sample. Therefore, compared with the case of manual setting, the workload required for measurement preparation can be reduced, and false detections can be further reduced while detecting signals with lower levels originating from particles.

[0022] More preferably, in the above-mentioned liquid particle measuring instrument, liquid particle measuring method, and signal processing program, a specified number of noises are generated by generating multiple unrelated noises by making the starting points of pre-prepared noise data of a specified length different, and these noises are added together by a specified number of parallel processing operations.

[0023] According to this scheme, the following method is used: using a certain amount of noise data (one, of a certain length), a number of uncorrelated noises corresponding to the number of parallel processing operations are generated. Furthermore, the generation method refers to generating multiple uncorrelated noises by making the starting points of the noise data different. Therefore, the types and volume of noise data required for noise generation can be suppressed, and the noise required for measurement can be generated efficiently.

[0024] Invention Effects

[0025] As described above, according to the present invention, the measurable particle size can be reduced while the counting efficiency is improved. Attached Figure Description

[0026] Figure 1 This is a block diagram showing the structure of a particle measuring instrument.

[0027] Figure 2 A is a three-dimensional diagram showing the structure of the particle detection unit.

[0028] Figure 2 B is a vertical sectional view showing the structure of the particle detection unit.

[0029] Figure 3 This is a block diagram showing the structure of the signal processing unit.

[0030] Figure 4 This is a diagram illustrating how noise is generated.

[0031] Figure 5 This is a diagram showing an example of the scattered light components and RMS values ​​after AC coupling, where the DC component is removed and only the AC component is set.

[0032] Figure 6 This is a flowchart illustrating an example of particle counting processing.

[0033] Figure 7 This is a diagram that briefly illustrates the signal processing method using stochastic resonance in the implementation.

[0034] Figure 8 Figure A is a comparison of the waveforms of signals in various processes used to detect particles with a diameter of less than 30 nm.

[0035] Figure 8 B is a graph showing a comparison of the waveforms of signals in each process of the treatment used to detect particles with a diameter of less than 30 nm.

[0036] Figure 8 C is a graph showing a comparison of the waveforms of signals in each process used to detect particles with a diameter of less than 30 nm.

[0037] Figure 9 This is a diagram that briefly illustrates the signal processing method using stochastic resonance in the comparative example.

[0038] Explanation of reference numerals in the attached figures

[0039] 1: Particle measuring instrument; 10: Particle detection unit; 14: Photoelectric conversion unit (detection unit, detection process); 20: A / D conversion unit (detection unit, detection process); 30: Signal processing unit; 31: Data input unit; 32: Noise generation unit; 33: Noise addition unit (addition unit, addition process); 34: Binarization processing unit (binarization unit, binarization process); 35: Averaging processing unit (calculation unit, calculation process); 36: LPF unit (filtering unit, filtering process); 37: Judgment unit (judgment process); 38: Binarization threshold calculation unit (binarization threshold calculation process); 39: Binarization threshold storage unit; 40: Reset unit; 41: Pulse amplitude analysis unit; 42: Particle counting unit; 50: Output unit. Detailed Implementation

[0040] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. It should be noted that the following embodiments are preferred examples, and the present invention is not limited to these examples.

[0041] [Particle measuring instrument]

[0042] Figure 1 This is a block diagram showing the structure of particle measuring instrument 1.

[0043] The particle measuring instrument 1 comprises, for example, a particle detection unit 10, an A / D converter 20, a signal processing unit 30, and an output unit 50. The particle detection unit 10 detects scattered light generated by the interaction between particles suspended in a liquid sample and the irradiated light, and outputs an electrical signal corresponding to the intensity of the scattered light. The A / D converter 20 converts the electrical signal output by the particle detection unit 10 into a digital signal. The signal processing unit 30 performs various signal processing on the digital signal output by the A / D converter 20, detects particles (determines the presence of particles), and counts them according to particle size ranges. The output unit 50 outputs the counting results based on the signal processing unit 30.

[0044] In this embodiment, to accurately detect particles with a diameter less than 30 nm, signal processing using stochastic resonance is performed in the signal processing unit 30. It should be noted that regarding the accuracy-related capabilities of the optical system used in this embodiment, it is assumed that it can measure particles as small as approximately 20 nm. If the accuracy of the optical system is further improved, it is also expected that it can accurately detect particles with diameters of 10 nm or less. It should be noted that the configuration of the particle detection unit 10 and the signal processing unit 30 will be further explained below using other accompanying drawings.

[0045] [Particle Detection Department]

[0046] Figure 2 A and Figure 2B is a perspective view and a cross-sectional view that briefly show the structure of the particle detection unit 10. Figure 2 The sectional view of B is along Figure 2 The vertical sectional view of the section line II-II shown in the perspective view of A.

[0047] The particle detection unit 10 comprises, for example, a flow cell 11, a light source 12, a focusing lens 13, and a photoelectric converter 14. The flow cell 11 is L-shaped, with a flow path 11a extending through its interior. A sample is introduced into the flow path 11a through a first opening 11b and discharged to the outside through a second opening 11c. Furthermore, the light source 12 irradiates a laser from a direction perpendicular to the direction of movement of the sample. In this embodiment, considering the detection of particles with a diameter less than 30 nm, a 532 nm green laser is used as the irradiation light. When the laser is irradiated, an irradiation area L and a detection area M are formed at predetermined positions within the flow path 11a, and scattered light is generated by particles P passing through the detection area M. The focusing lens 13 focuses the scattered light generated by particles P. The photoelectric converter 14 detects the focused scattered light, converts it into a voltage corresponding to the intensity of the scattered light, and outputs an electrical signal. It should be noted that the photoelectric converter 14 is part of the detection unit in this embodiment of the invention.

[0048] The condenser lens 13 and photoelectric converter 14 are positioned opposite the direction of sample movement. This allows for accurate detection of scattered light generated by particle P. Furthermore, a concave portion 11d is formed on the inner wall of the flow cell 11 located between the detection area M and the condenser lens 13, which suppresses refraction that may occur when scattered light generated by particle P passing through the detection area M strikes the inner wall of the flow cell 11. It should be noted that the arrangement of each device in the particle detection unit 10 is determined based on the condenser axis, and the incident position of the irradiation light is designed such that the center of the detection area M is located on the condenser axis.

[0049] [A / D converter]

[0050] The electrical signal (“detection signal”) output by the photoelectric converter 14 includes, in addition to the signal detected when particle P passes through the detection area M (“target signal”), noise caused by the scattered light from the sample (“scattering noise”), etc. Furthermore, the target signal is assumed to be a Gaussian-distributed signal in a frequency band cutoff to approximately 15 kHz.

[0051] The detection signal is sampled and quantized in the A / D conversion unit 20 to convert it into a digital signal. In this embodiment, to prevent aliasing, a second-order low-pass filter is used for the detection signal, with a cutoff frequency of 60kHz to ensure sufficient bandwidth for the target signal. Furthermore, to minimize quantization noise, oversampling is performed at approximately 20 times the detection signal, and sampling is performed at a rate of 5 Msps. The quantization bits are set to 16 bits. It should be noted that the A / D conversion unit 20 is part of the detection unit in this embodiment of the invention. It should also be noted that while the photoelectric converter 14 implements a part of the detection process in this embodiment of the invention, the A / D conversion unit also implements a part of the detection process.

[0052] [Signal Processing Department]

[0053] Figure 3 This is a block diagram showing the configuration of the signal processing unit 30.

[0054] The signal processing unit 30 comprises, for example, a data input unit 31, a noise generation unit 32, a noise addition unit 33, a binarization processing unit 34, an averaging processing unit 35, an LPF unit 36, a determination unit 37, a binarization threshold calculation unit 38, a binarization threshold storage unit 39, a reset unit 40, a pulse amplitude analysis unit 41, and a particle counting unit 42. Among these functional units, the noise generation unit 32 to the reset unit 40 facilitate the detection and counting of particles with a diameter less than 30 nm, while the pulse amplitude analysis unit 41 facilitates the detection and counting of particles with a diameter greater than 30 nm. The functions of each functional unit constituting the signal processing unit 30 will be explained below along the direction of signal movement.

[0055] [Data Input Department]

[0056] The digital signal (signal data) output from the A / D converter 20 is input to the data input unit 31. The data input unit 31 outputs the input signal unchanged. It should be noted that in this embodiment, the signal data is input sequentially in real time, but it is also possible to read the signal data stored in a storage area (not shown) such as an internal memory for input instead.

[0057] Furthermore, the input signal includes noise generated during the measurement process in addition to the target signal. In this embodiment, particles with a diameter of less than 30 nm are also used as the detection target, so it is necessary to suppress the noise that accompanies the measurement as low as possible.

[0058] Representative noises contained in the input signal, besides the aforementioned scattering noise, include quantization noise during A / D conversion and thermal noise originating from the circuit. Regarding quantization noise during A / D conversion, as described above, countermeasures are taken during processing in the A / D conversion unit 20. Furthermore, the circuit's thermal noise is on the order of 1mV, which is considered sufficiently small compared to the scattering noise, which is on the order of 100mV. Moreover, scattering noise is addressed in subsequent signal processing.

[0059] [Noise Generation Section]

[0060] The noise generation unit 32 generates a number of uncorrelated noises corresponding to the number of parallel processes in the noise addition unit 33 (described later) and outputs them. In this embodiment, a noise data of a certain data length stored in a storage area (storage unit) (not shown) is repeatedly linked, using a position that is successively offset from the starting point of the obtained linked data by a predetermined data length as the starting point, thereby generating multiple noises from a single noise data. It should be noted that, ideally, the noise data used for generation is Gaussian white noise.

[0061] [Methods of noise generation]

[0062] Figure 4 This is a diagram illustrating how noise is generated.

[0063] In the noise generation unit 32, for example, when the number of parallel operations k = 100, seven unrelated noise data A to G with a data length of 16kB are prepared. In the linked data obtained by repeatedly linking each noise data, a position that is staggered by 1kB from each of its starting points is used as the starting point to generate 100 noise data.

[0064] Specifically, noise data is generated from the linked data based on noise data A: noise 1 starting from its starting point, noise 2 starting from a position offset by 1 kB from the starting point, noise 3 starting from a position offset by 2 kB from the starting point, ..., noise 15 starting from a position offset by 14 kB from the starting point, and noise 16 starting from a position offset by 15 kB from the starting point. Noise data after noise data B is processed in the same way, and a single noise data can generate 16 noises.

[0065] Since the noise data A through G are uncorrelated, the noise generated based on them but with different starting points is also uncorrelated. By generating noise in this way, it is possible to generate multiple uncorrelated noises while suppressing the amount and volume of noise data required for generation.

[0066] It should be noted that the length of the noise data and the specified length for staggering the start points can be appropriately changed according to the situation. In any case, based on these values, the amount of noise that a single noise data point can generate is naturally determined. Therefore, the number of pre-prepared noise data points is the quotient of the parallel quantity k divided by the amount of noise that a single noise data point can generate, plus 1.

[0067] Furthermore, in this embodiment, noise is generated using noise data stored in the storage area, but it can also be generated through computation instead. For example, a pseudo-random number generation method utilizing the M-sequence can also be used to generate noise.

[0068] [Noise Addition Section]

[0069] The noise addition unit 33 performs parallel processing on the input signal (the output signal from the data input unit 31) with a parallel quantity of k. It adds k unrelated noises generated by the noise generation unit 32 in each processing step to the input signal with the same intensity (noise additive amount β) and outputs the result. In this embodiment, the parallel quantity k is set to 100. Furthermore, the noise additive amount β is related to the target signal, the intensity of the scattered noise originating from the background light, the measured medium and its temperature and concentration, etc. In this embodiment, a value determined based on prior experiments is used. It should be noted that the noise additive amount β can also be considered when generating noise in the noise generation unit 32. In this case, the noise addition unit 33 directly adds the input signal to each noise. It should be noted that the noise addition unit 33 is part of the addition unit in the embodiments of the present invention. It should be noted that the noise addition unit 33 implements part of the addition process in the embodiments of the present invention.

[0070] Increasing the number of parallel operations increases the computational load, but in the processing described later, it makes the target signal stand out more clearly relative to the noise signal, thus improving detection accuracy. However, when the number of parallel operations exceeds a certain limit, the improvement in detection accuracy becomes gradual. Therefore, ideally, the number of parallel operations should be selected by taking into account the target signal, noise levels, and the computational processing capability of the signal processing unit 30.

[0071] [Binarization Processing Department]

[0072] The binarization processing unit 34 compares each of the k outputs from the noise addition unit 33 with the binarization threshold m of the random resonance. If the output is greater than or equal to m, it outputs "1"; otherwise, it outputs "0". The binarization threshold m of the random resonance is calculated by the binarization threshold calculation unit 38 (described later) and stored in the binarization threshold storage unit 39 (described later). It should be noted that the binarization processing unit 34 is part of the binarization unit in the embodiments of the present invention. It should be noted that the binarization process in the embodiments of the present invention is implemented by the binarization processing unit 34.

[0073] [Averaging Processing Department]

[0074] The averaging processing unit 35 is an example of the calculation unit of the present invention. A simple average is calculated by adding the k outputs from the binarization processing unit 34 and dividing by k. The resolution is set to 10 bits, and the calculation result is converted to 1 to 1024 (=2^k). 10 The value of k is calculated and output. It should be noted that, considering the convenience of the latter half of the processing step, the division by k calculation may not be performed in the averaging processing unit 35. In this case, the calculation based on the parallel quantity k is performed in the preceding and following processing steps. Alternatively, a weighted average (weighted average) can be calculated instead of a simple average. It should be noted that the averaging processing unit 35 is part of the calculation unit in the embodiment of the present invention. It should be noted that the averaging processing unit 35 implements part of the calculation process in the embodiment of the present invention.

[0075] [LPF section]

[0076] The LPF unit 36 ​​uses a low-pass filter on the output signal from the averaging processing unit 35. In this embodiment, a third-order low-pass filter with a cutoff frequency of 15kHz is used to allow components of the output signal corresponding to the frequency band of the target signal (approximately 100Hz to 15kHz) to pass through. This eliminates false detections of particles because short-duration signals are cut off. It should be noted that the slope of the filter does not need to be very steep, so a second-order filter can be used instead of a third-order filter. Furthermore, it is not limited to a low-pass filter; it can be implemented using a combination of band-pass filters that allow the target signal's frequency band to pass through, high-pass filters, and band-stop filters that restrict the frequency band outside the target signal's frequency band. It should be noted that the LPF unit 36 ​​is part of the filtering unit in this embodiment of the invention. It should be noted that the LPF unit 36 ​​implements part of the filtering process in this embodiment of the invention.

[0077] [Judgment Department]

[0078] The determination unit 37 checks whether the output signal from the LPF unit 36 ​​exceeds a predetermined threshold ("particle threshold") determined according to the minimum measurable particle size of the particle measuring instrument 1. If the output signal exceeds the particle threshold, it determines that particles exist, that is, it determines that the signal originates from particles, and outputs a pulse indicating this situation. On the other hand, if the output signal is below the particle threshold, it determines that particles do not exist, that is, it determines that the signal does not originate from particles, and no pulse is output. In this embodiment, for example, the above determination is performed on an output with the minimum measurable particle size set to "20 nm" and a particle threshold of "850" predetermined to correspond to "20 nm" is set, taking a value from 1 to 1024. The detection threshold is set to a threshold for detecting particles above the minimum measurable particle size in a way that will not cause false detection, and is a value derived based on prior experiments. It should be noted that the determination unit 37 implements part of the determination process in the embodiment of the present invention.

[0079] [Binarization Threshold Calculation Unit]

[0080] When the binarization threshold calculation unit 38 receives a reset signal from the reset unit 40 (described later), it calculates the binarization threshold of the random resonance based on the measured sample at that time point. The binarization threshold is calculated using the following formula. It should be noted that the binarization threshold calculation unit 38 implements part of the binarization threshold calculation process in the embodiments of the present invention.

[0081] <Formula 1>

[0082] Binarization threshold = (binarization coefficient) × (noise baseline setting)

[0083] In the above calculation formula, the binarization coefficient is a predetermined value based on the design value of the optical equipment. Furthermore, the noise baseline setting is a value set based on the characteristic quantity of the scattered light component of the sample during measurement; specifically, it is equivalent to a value based on the RMS value, DC value, or both of the RMS and DC values ​​of the scattered light component of the sample. The RMS value refers to the effective value of the AC component in the measurement signal, calculated from the root mean square of the signal. The DC value refers to the DC component value in the measurement signal. That is, in this embodiment, the RMS value and DC value are used as characteristic quantities of the sample being measured.

[0084] It should be noted that the characteristic quantities of the scattered light components considered when setting the noise baseline settings are not limited to RMS values ​​or DC values. For example, they can also be: various average values ​​(summary average of amplitudes, weighted average, etc.) related to the scattered light components of the measured sample, the median value, maximum value, peak value representing the amplitude distribution, or any index of the characteristic quantities of the measured sample, or values ​​based on a formula that yields values ​​close to these indices, or combinations thereof. In any case, the binarization coefficients can be appropriately set according to the selection of the noise baseline settings.

[0085] For both RMS and DC values, the binarization coefficients, which take into account the characteristics of the sample such as medium, concentration, refractive index, and isothermal compressibility, are determined according to the design time of the optical equipment and are preset in the storage area before leaving the factory (not shown).

[0086] The RMS and DC values, which form the basis of the noise baseline settings, are set based on their respective measurements taken over a specified period (e.g., 5 seconds) after the reset signal is received. For example, in the case of a reset following a power-on operation, the RMS and DC values ​​are set based on the 5-second RMS and DC values ​​measured after the warm-up operation of the light source 12 has ended. The RMS value varies depending on the medium. For example, it is approximately 1.2 times that of pure water, 96% concentrated sulfuric acid, and approximately 1.3 to 1.4 times that of IPA. Furthermore, the DC value is approximately 60 to 70 mV when the medium is pure water.

[0087] In this embodiment, only the RMS value is used as the noise baseline setting. Therefore, only the binarization coefficient corresponding to the RMS value is preset, and the binarization threshold is automatically calculated based on the product of the binarization coefficient and the RMS value during reset.

[0088] Figure 5 A specific example is shown where the sample is measured in pure water, and the scattered light component after AC coupling and the RMS value set accordingly are given. In this example, the binarization coefficient is set to "3", the calculated RMS value of the scattered light component is "0.03Vrms", and the binarization threshold of "0.09" is calculated by multiplying the binarization coefficient and the RMS value.

[0089] It should be noted that although only the RMS value is used as the noise baseline setting value in this embodiment, if only the DC value is used instead, only the binarization coefficient corresponding to the DC value is preset. Furthermore, when using values ​​determined based on both the RMS and DC values, binarization coefficients corresponding to the RMS and DC values ​​are preset respectively. When employing a configuration that can utilize multiple noise baseline setting values, it can be configured to automatically set the optimal setting value based on the selection of the test sample name from a preset list of test sample names and the selection of the characteristic quantity of the scattered light component, or it can be configured to manually set any setting value.

[0090] When choosing from multiple noise baseline settings, for example, if selecting a DC value might affect the particle measurement results (such as changes in laser light intensity due to light source 12 degradation, resulting in variations in scattered light intensity), an RMS value is suitable. Furthermore, an RMS value is also suitable when using samples with high cleanliness (few suspended particles). Conversely, when using samples with high particle concentration, a DC value is more suitable because it is difficult to distinguish between the signal originating from particles (target signal) and the signal of scattered light originating from the sample (scattering noise). Alternatively, a setting can be used where a higher accuracy measurement result can be obtained based on both RMS and DC values.

[0091] [Binarization Threshold Storage Unit]

[0092] In addition to storing the aforementioned binarization coefficients that are predetermined and preset before leaving the factory, the binarization threshold storage unit 39 also stores the binarization threshold calculated by the binarization threshold calculation unit 38. When a reset signal is received from the reset unit 40 (described later), the binarization threshold storage unit 39 deletes the binarization threshold stored at that point in time while maintaining the binarization coefficients.

[0093] [Reset Department]

[0094] The reset unit 40 sends reset signals to each functional unit related to signal processing using random resonance in response to the reset operation of the particle measuring instrument 1. Here, "reset operation" refers to the operation that accompanies the user's operation of the particle measuring instrument 1, specifically, the operation of turning on the power, pressing the setting button due to changes in the measured sample, etc. After the user turns on the power and the warm-up operation of the light source 12 is completed, a reset signal is sent from the reset unit 40 at the timing when the user presses the setting button. Then, in conjunction with this, the binarization threshold calculation unit 38 automatically calculates the binarization threshold, and the processing in other functional units is initialized.

[0095] [Pulse Amplitude Analysis Department]

[0096] The functional units 32-40 described above are helpful for detecting and counting particles with a diameter less than 30 nm. In contrast, the pulse amplitude analysis unit 41 is helpful for detecting and counting particles with a diameter of 30 nm or larger. Specifically, the pulse amplitude analysis unit 41 uses, for example, the same method as the particle counting method in Japanese Patent No. 4960772 (hereinafter referred to as the "conventional method") to determine which of the multiple particle size ranges from 30 nm or larger to a few μm that the particle detected based on the input signal (the output signal from the data input unit 31) matches, and outputs a pulse corresponding to the matching particle size range. It should be noted that since the conventional method is known, a detailed description of the processing in the pulse amplitude analysis unit 41 is omitted.

[0097] [Particle Counting Section]

[0098] The particle counting unit 42 counts the number of detected particles according to particle size range based on the pulses output from the determination unit 37 and the pulse amplitude analysis unit 41. It should be noted that the noise addition unit 33 to the determination unit 37 and the pulse amplitude analysis unit 41 perform different processing on the same input signal simultaneously, and the pulses output by the determination unit 37 and the pulse amplitude analysis unit 41 and input to the particle counting unit 42 are synchronized.

[0099] [Particle counting processing]

[0100] Figure 6 This is a flowchart illustrating an example of the particle counting process performed by the particle counting unit 42. The particle counting unit 42 continuously performs particle counting processing from the start to the end of the measurement. The process example will be explained below.

[0101] Step S1: The particle counting unit 42 checks whether a pulse has been input from the pulse amplitude analysis unit 41. If the result of the check is that a pulse has been input from the pulse amplitude analysis unit 41 (Step S1: Yes), the particle counting unit 42 executes step S2. On the other hand, if the result is that no pulse has been input from the pulse amplitude analysis unit 41 (Step S1: No), the particle counting unit 42 executes step S3.

[0102] Step S2: The particle counting unit 42 counts the number of particles in the particle size range corresponding to the pulse from the pulse amplitude analysis unit 41. For example, if the pulse from the pulse amplitude analysis unit 41 corresponds to a particle size range of 40 to 50 nm, the number of particles in that particle size range is incremented by 1.

[0103] Step S3: The particle counting unit 42 checks whether a pulse has been input from the determination unit 37. If the result of the check is that a pulse has been input from the determination unit 37 (Step S3: Yes), the particle counting unit 42 executes step S4. On the other hand, if the result is that no pulse has been input from the determination unit 37 (Step S3: No), the particle counting unit 42 returns to step S1.

[0104] Step S4: The particle counting unit 42 counts the number of particles in the particle size range of less than 30 nm. As described above, the particle threshold that serves as the reference for the pulse output in the determination unit 37 is a threshold corresponding to 20 nm. Therefore, through this process, the number of particles in the particle size range of 20 nm and above but less than 30 nm is incremented by 1.

[0105] When the above process ends, the particle counting unit 42 returns to step S1 and repeats the process of steps S1 to S4.

[0106] As described above, the pulses from the pulse amplitude analysis unit 41 correspond to a certain particle size range of 30 nm or larger. In contrast, the pulses from the determination unit 37 are input when the particle size exceeds a threshold corresponding to 20 nm. Therefore, when processing signals originating from particles with a particle size of 30 nm or larger, although pulses may sometimes be input simultaneously from both the pulse amplitude analysis unit 41 and the determination unit 37, counting the particles according to the above process avoids counting a single particle twice, enabling accurate counting of the number of particles.

[0107] It should be noted that the particle counting result from the particle counting unit 42 is input to the output unit 50. Then, the output unit 50 is used to output the counting result to a monitor or printer, or to send it to other devices via the network.

[0108] The above describes the configuration of the signal processing unit 30 in this embodiment. It should be noted that the signal processing unit 30 can be installed in a DSP (Digital Signal Processor) or a device capable of processing digital signals, such as an FPGA (Field Programmable Gate Array). Alternatively, it can be installed as a program in a computer with ordinary functions, such as a PC (Personal Computer). It should also be noted that a computer-readable medium storing a program for executing the processing of this embodiment can also be used.

[0109] [Signal processing using stochastic resonance (implementation method)]

[0110] Figure 7This diagram briefly illustrates the signal processing method using the stochastic resonance described in this embodiment, specifically the processing for detecting particles with a diameter less than 30 nm. Furthermore, Figure 8 A~ Figure 8 C is a graph showing a comparison of the waveforms of the signals in each process of signal processing that uses stochastic resonance.

[0111] like Figure 7 As shown, in the signal processing of this embodiment, firstly, the input signal ( Figure 8 A) The signal is sent from the data input unit 31 to the noise addition unit 33 unchanged. Additionally, k unrelated noises (Gaussian noise 1 to k) are generated in the noise generation unit 32 and sent to the noise addition unit 33. Then, k parallel processing operations are performed in the noise addition unit 33, where each of the k noises is added to the input signal by the same noise addition amount β. The added signals are then sent to the binarization processing unit 34, where they are binarized using a binarization threshold automatically calculated based on the medium of the measured sample. Finally, a simple average of the binarized values ​​is calculated by the averaging processing unit 35, and the result is converted into a 10-bit signal and output. Figure 8 B). In this output signal, components that deviate from the frequency band of the target signal are removed by the LPF section 36. Figure 8 C). Then, the determination unit 37 determines the signal that has passed through the LPF unit 36, and outputs a pulse indicating this condition if the particle threshold is exceeded.

[0112] Here, if we try to focus on the waveforms in each process of signal processing, then as follows: Figure 8 A~ Figure 8 As shown in Figure C, it can be seen that as signal processing progresses, the signal waveform becomes simpler. Specifically, the input signal ( Figure 8 In signal A), the signal originating from particles (target signal) and the signal originating from background light (scattering noise) are mixed, resulting in a very complex signal waveform. In contrast, it can be seen that the signal after averaging ( Figure 8 In B), the signal peak is more clearly visible compared to the input signal. This is because, when the target signal is included, even if the signal after adding noise is averaged, the target signal remains intact. On the other hand, when only noise is included, the signal level decreases because the unrelated noise is averaged, thus making the target signal stand out relatively.

[0113] The determination of the presence of particles larger than 20 nm is based on the signal after LPF processing. Figure 8In step C), for example, when the particle threshold is set to 1.5 times the peak value of the scattering noise peak x (indicated by the arrow in the figure), two particles can be detected based on the signal in the interval shown in the figure. In contrast, even when processing the input signal in the same interval using conventional methods (processing by the pulse amplitude analysis unit 41), these particles are not easily detected. Therefore, it is clear that according to this embodiment, particles with smaller particle sizes can be detected compared to conventional methods.

[0114] [Signal processing using stochastic resonance (comparative example)]

[0115] Figure 9 The diagram briefly illustrates, as a comparative example, the conventional approach to signal processing using stochastic resonance.

[0116] In the comparative example, k parallel processes are performed, where the input signal and k pre-prepared, uncorrelated noises are each added to the input signal by the same amount of β. The summed signals are then binarized using a fixed binarization threshold, and a simple average of the binarized values ​​is calculated. The result is sent in the latter half of the next processing step.

[0117] Thus, in the comparative example, k noise samples need to be prepared in advance. Therefore, the larger the number of parallel samples (k), the more time-consuming the preparation becomes, and the larger the data capacity required to store the noise also becomes. Furthermore, since a fixed binarization threshold is used during binarization, it cannot handle variations in the measured samples. If the measurement threshold is manually reset for each measurement, it is not only time-consuming in its operation but may also introduce deviations during adjustment. Moreover, this situation can significantly affect the particle detection accuracy.

[0118] In contrast, in this embodiment, k noises are prepared by the noise generation unit 32 generating multiple noises based on a certain noise data. Therefore, the number of noise data to be prepared in advance is small, which can suppress the data capacity required for storing the noise. Furthermore, during binarization, a binarization threshold calculated by the binarization threshold calculation unit 38 based on the medium of the test sample is used. Therefore, compared with the comparative example, the workload required for measurement preparation can be reduced, while the particle detection accuracy can be improved.

[0119] [Advantages of the present invention]

[0120] As described above, the following effects can be obtained according to the above-described embodiments.

[0121] (1) Since the output signal is binarized using a binarization threshold calculated based on the RMS value and DC value of the scattered light component of the sample, false detection can be reduced while detecting signals with lower levels originating from particles, and measurement with a minimum measurable particle size of 20 nm can be achieved.

[0122] (2) Since it is possible to measure a smaller minimum measurable particle size than existing methods, it can improve the detection accuracy of particles contained in the liquid and the counting accuracy of particle count.

[0123] (3) Since the binarization threshold is automatically calculated in conjunction with the reset operation, the amount of work required to prepare the measurement based on the sample can be reduced.

[0124] (4) Since multiple noises are generated by using the noise generation unit 32 to generate multiple noises from a certain noise data with different starting points, the uncorrelated noises added to the input signal in each parallel processing can be generated efficiently while suppressing the amount and capacity of noise data required for noise generation.

[0125] The present invention is not limited to the above-described embodiments and can be implemented in various modifications.

[0126] In the above embodiment, the number of parallel processes k in the noise addition unit 33 is set to 100, but it is not limited to this. Depending on the target signal, the noise condition, the computing power of the signal processing unit 30, etc., a configuration in which only one process (set to k=1) is performed without performing multiple parallel processes may also be adopted.

[0127] Furthermore, by setting the parallel quantity k to a larger value, the signal data can be temporarily stored in a storage area beforehand instead of being processed in real time, and then processed at regular intervals (e.g., data from 1 minute). By adopting such a configuration, signal processing can be performed stably with limited computing power while reducing the noise level after averaging, resulting in further improvement in the detection accuracy of particles with a diameter of less than 30 nm.

[0128] In the above embodiment, the minimum measurable particle size of the particle measuring instrument 1 is set to 20 nm, and a corresponding particle threshold for determination by the determination unit 37 is determined. However, the minimum measurable particle size is not limited to this. Furthermore, the determination unit 37 outputs a single pulse based on the determination result, but it can also be configured to set multiple particle thresholds and output different pulses based on the particle threshold exceeded by the output signal. With this configuration, even particles smaller than 30 nm can be counted in multiple particle size ranges.

[0129] In the above embodiment, an L-shaped flow cell 11 was used, but the shape of the flow cell can be any shape with a part that is bent into an L-shape. For example, a ko-shaped or crank-shaped flow cell can be used instead of an L-shaped flow cell.

[0130] Furthermore, the materials, values, etc. listed as examples of the constituent components of the particle measuring instrument 1 are merely examples, and can of course be appropriately modified when implementing the present invention.

Claims

1. A liquid particle measuring instrument, characterized in that, have: The detection unit detects the scattered light generated by the interaction between particles contained in the liquid sample and light incident on the liquid sample, and converts it into a signal of corresponding intensity. The addition unit performs a predetermined number of parallel processing operations on the signal to add the predetermined number of uncorrelated noises to the signal, and outputs the summed signals. The binarization unit performs binarization on each of the summed signals using a binarization threshold set to match the liquid sample, and outputs each binarized signal. The computing unit calculates the value based on each binarized signal and outputs the value; The filtering unit allows components of a predetermined frequency in the output of the calculation unit to pass through; The determination unit determines that particles are present if the output of the filtering unit exceeds a predetermined particle threshold. as well as The binarization threshold calculation unit calculates the binarization threshold based on the characteristic quantities of the scattered light components of the liquid sample. The binarization unit performs binarization using the binarization threshold calculated by the binarization threshold calculation unit.

2. The liquid particle measuring instrument according to claim 1, characterized in that, The binarization threshold calculation unit calculates the binarization threshold based on the RMS or DC value of the scattered light component of the liquid sample.

3. The liquid particle measuring instrument according to claim 1 or 2, characterized in that, It also has: The storage unit stores noise data of a specified length; and The noise generation unit generates the predetermined number of noises by generating multiple unrelated noises from the noise data by using different starting points. The addition unit adds together a predetermined number of unrelated noises generated by the noise generation unit.

4. The liquid particle measuring instrument according to claim 1 or 2, characterized in that, The computing unit calculates the average value of each binarized signal and outputs it.

5. A method for measuring particles in a liquid, characterized in that, have: The detection process involves detecting the scattered light generated by the interaction between particles contained in the liquid sample and light incident on the liquid sample, and converting it into a signal of corresponding intensity. The addition process involves performing a specified number of parallel processing steps on the signal to add the specified number of unrelated noises to the signal, and then outputting the summed signals. The binarization process involves binarizing each of the summed signals using a binarization threshold set to match the liquid sample, and outputting each binarized signal. The calculation process involves calculating the value based on each binarized signal and outputting that value. The filtering process allows components of a specified frequency in the output of the calculation process to pass through. The determination process determines the presence of particles if the output of the filtering process exceeds a specified particle threshold. as well as The binarization threshold calculation process calculates the binarization threshold based on the characteristic quantities of the scattered light components of the liquid sample. The binarization process uses the binarization threshold calculated by the binarization threshold calculation process to perform binarization.

6. A storage medium storing a signal processing program for measuring particles contained in a liquid sample, characterized in that, The signal processing program enables the computer to function as at least the following components: The addition unit performs a predetermined number of parallel processing operations on a signal corresponding to the intensity of the scattered light to add the predetermined number of unrelated noises to the signal and outputs the summed signals, wherein the scattered light is generated by the interaction between the particles and light incident on the liquid sample. The binarization unit performs binarization on each of the summed signals using a binarization threshold set to match the liquid sample, and outputs each binarized signal. The computing unit calculates the value based on each binarized signal and outputs the value; The filtering unit allows components of a predetermined frequency in the output of the calculation unit to pass through; The determination unit determines that particles are present if the output of the filtering unit exceeds a predetermined particle threshold. as well as The binarization threshold calculation unit calculates the binarization threshold based on the characteristic quantities of the scattered light components of the liquid sample. The binarization unit performs binarization using the binarization threshold calculated by the binarization threshold calculation unit.

Images