Cold mirror type water vapor concentration and dew point detection system and method based on single pixel imaging

Abstract

The invention provides a cold mirror type water vapor concentration and dew point detection system and method based on single-pixel imaging, the cold mirror type water vapor concentration detection system based on single-pixel imaging comprises a mirror surface refrigeration module, a detection module and a control module, the detection system can distinguish phase change and pollutants in the mirror surface state detection process, the reliability is high, and the detection accuracy is high. According to the detection system, the low-cost single-pixel detector is combined with the spatial light modulator, an expensive and complex area-array camera and an optical lens group in a traditional scheme are replaced, the cost and the size of the system are remarkably reduced, and the reliability of water vapor concentration detection in the semiconductor low-dew-point environment is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of semiconductor dew point and water vapor concentration detection, and particularly relates to a cold mirror type water vapor concentration and dew point detection system and method based on single-pixel imaging. BACKGROUND

[0002] In the semiconductor manufacturing process, the humidity control of the environment and the reaction gas is one of the key factors affecting the chip yield. Especially with the continuous evolution of advanced process nodes, the wafer structure is becoming increasingly precise, and the sensitivity to trace moisture in the process environment is rising sharply. Even if there is only a few ppb level of water vapor exceeding the standard in the reaction gas, it is enough to cause irreversible defects on the wafer surface. For example, in the chemical vapor deposition or etching process, excessive water molecules will participate in unintended reactions, generate particulate pollutants or cause thin film structure abnormalities, and then cause chip interlayer short circuit, dielectric breakdown and other defects. Such defects are generally discovered only in the electrical testing stage, directly leading to wafer scrap and causing economic losses. Therefore, real-time monitoring of gas humidity with high precision and high reliability has become an indispensable quality control link in the semiconductor manufacturing process.

[0003] At present, the cold mirror dew point meter is widely recognized as the benchmark instrument for measuring gas humidity due to its direct measurement characteristics based on the thermodynamic phase equilibrium principle. Its core working mechanism is to cool the metal mirror surface below the dew point temperature of the measured gas by using a thermoelectric cooler, so that the mirror surface condenses water vapor to form a dew layer or frost layer. The system detects the change of the optical properties of the mirror surface to identify the moment of condensation, and uses a high-precision platinum resistance thermometer to measure the mirror surface temperature synchronously, so as to determine the dew point value of the gas. The traditional cold mirror instrument generally adopts a single-point photoelectric detection scheme, that is, the light source emits a light beam to the mirror surface. When the mirror surface is in a dry state, the light is almost totally reflected to the photoelectric detector, and the signal strength is high. Once the mirror surface dew condenses or frosts, the reflectivity of the mirror surface decreases and the scattering increases, resulting in a significant attenuation of the light intensity received by the detector. The system determines the dew point by monitoring the mutation point of the light intensity signal, and usually combines a PID control algorithm to realize closed-loop regulation of the mirror surface temperature to maintain the dynamic balance of condensation and evaporation. However, it still has obvious limitations when facing the high requirements of semiconductor manufacturing. First, the traditional single-point photoelectric detection method only relies on a single light intensity signal for judgment, and cannot effectively distinguish between dew, frost and mirror surface contamination. In low temperature or low humidity conditions, the mirror surface may alternately or mixedly form dew and frost, and the corresponding saturated water vapor pressures are different. If the identification is not accurate, it will introduce significant measurement error. In addition, the mirror surface is easily contaminated by volatile organic compounds or particulate matter during use, and the contamination layer will also cause the reflected light intensity to decrease, and the system is easy to misjudge it as a dew condensation signal, resulting in false alarms, which seriously affects the measurement reliability. The above prior art solution relies on high-resolution industrial cameras and precise lens groups, resulting in a large instrument size and high cost, which is not conducive to portability and low-cost popularization. The traditional threshold judgment method cannot distinguish between dew, frost and contamination, and the scheme based on camera imaging also fails to integrate intelligent diagnostic algorithms, resulting in low reliability and automation level under complex working conditions.

[0004] Therefore, there is a need in the art to provide a new cold mirror water vapor concentration and dew point detection system and method based on single-pixel imaging to better adapt to the requirements of the semiconductor manufacturing industry for gas humidity measurement. SUMMARY

[0005] The purpose of the present application is to provide a cold mirror water vapor concentration and dew point detection system and method based on single-pixel imaging, which reduces the cost and size of the detection system and improves the reliability of water vapor concentration detection.

[0006] The technical scheme adopted by the present application to solve its technical problems is: the present application provides a cold mirror water vapor concentration detection system based on single-pixel imaging, comprising: a mirror surface refrigeration module, including a mirror surface, a refrigeration device for refrigerating and heating the mirror surface, and a temperature sensor for real-time monitoring of the mirror surface temperature; The detection module comprises a light source, a spatial light modulator, and a single-pixel detector; the spatial light modulator is configured to modulate the light emitted by the light source into a series of preset coded patterns and project the coded patterns to the mirror; the single-pixel detector is configured to receive the light intensity signals carrying the mirror state information after the coded patterns are reflected by the mirror, and output light intensity measurement values corresponding to each coded pattern; The control module is configured to: acquire the light intensity measurement values output by the single-pixel detector and the temperature values measured by the temperature sensor; determine the state of the mirror based on the sequence of light intensity measurement values and the sequence of corresponding coded patterns; control the working power of the refrigerator based on the determined state of the mirror and the temperature value, so that the mirror temperature is dynamically maintained at the dew point temperature or the frost point temperature, and output the dew point temperature or the frost point temperature; obtain the water vapor concentration according to the dew point temperature and the ambient temperature.

[0007] In an embodiment, the control module comprises: an image reconstruction unit configured to reconstruct a two-dimensional image of the mirror based on the sequence of light intensity measurement values and the sequence of corresponding coded patterns by a single-pixel imaging algorithm; an intelligent recognition unit comprising a trained machine learning model, configured to analyze the reconstructed two-dimensional image and output a classification result representing the state of the mirror, the state of the mirror including at least one of clean, dew layer formation, frost layer formation, and pollution.

[0008] In an embodiment, the control module further comprises a cleaning unit configured to trigger an alarm signal and / or automatically clean the mirror when the state of the mirror is identified as "pollution".

[0009] In an embodiment, the machine learning model is a convolutional neural network or a support vector machine.

[0010] In an embodiment, the single-pixel imaging algorithm is an image reconstruction algorithm based on compressed sensing; and the coded pattern is a Hadamard basis pattern.

[0011] In an embodiment, the light source is an infrared waveband light source, and the single-pixel detector is an infrared waveband detector.

[0012] In an embodiment, the detection system further comprises a gas pretreatment module, the pretreatment module comprising a filter, a flow controller, and a pressure regulator.

[0013] The second aspect of the present application provides a cold mirror type water vapor concentration detection method based on single-pixel imaging, which adopts the cold mirror type water vapor concentration detection system based on single-pixel imaging as described above, and the detection method comprises the following steps: making the to-be-detected gas flow through the mirror surface, and controlling the refrigerator to cool the mirror surface; modulating the light emitted by the light source into a series of preset coded patterns by the spatial light modulator and projecting the coded patterns onto the mirror surface; receiving the light intensity signal reflected by the mirror surface by the single-pixel detector, and collecting the light intensity measurement value corresponding to each coded pattern; determining the state of the mirror surface according to the sequence of light intensity measurement values and the sequence of corresponding coded patterns; when it is determined that the state of the mirror surface is dew layer formation or frost layer formation, recording the current temperature of the mirror surface as the dew point temperature or the frost point temperature, and controlling the refrigerator to stabilize the temperature of the mirror surface at the dew point temperature or the frost point temperature; obtaining the water vapor concentration according to the dew point temperature and the ambient temperature.

[0014] In an embodiment, the step of determining the state of the mirror surface comprises: reconstructing a two-dimensional image of the mirror surface by a single-pixel imaging algorithm according to the sequence of light intensity measurement values and the sequence of corresponding coded patterns; inputting the reconstructed two-dimensional image into a trained machine learning model to obtain a classification result representing the state of the mirror surface, wherein the state of the mirror surface includes at least one of clean, dew layer formation, frost layer formation and pollution.

[0015] In an embodiment, the method further comprises: triggering an alarm and / or automatically cleaning the mirror surface when it is determined that the state of the mirror surface is pollution.

[0016] In an embodiment, the single-pixel imaging algorithm adopts an image reconstruction algorithm of compressed sensing; and the specific steps of reconstructing the two-dimensional image of the mirror surface by the single-pixel imaging algorithm comprise: constructing the coded pattern as a measurement matrix Φ; constructing the light intensity measurement value as a measurement vector Y; and reconstructing a vector X representing the two-dimensional image of the mirror surface by solving the least squares solution of the linear equation system Y = ΦX.

[0017] The third aspect of the present application provides a cold mirror type dew point detection system based on single-pixel imaging, which comprises: a mirror surface refrigeration module comprising a mirror surface, a refrigerator for cooling and heating the mirror surface, and a temperature sensor for monitoring the temperature of the mirror surface in real time; The detection module includes a light source, a spatial light modulator, and a single-pixel detector. The spatial light modulator is used to modulate the light emitted by the light source into a series of preset coded patterns and project them onto the mirror surface. The single-pixel detector is used to receive the light intensity signal carrying the mirror surface state information after the coded patterns are reflected by the mirror surface, and output the light intensity measurement value corresponding to each coded pattern. The control module is configured as follows: The light intensity measurement value output by the single-pixel detector and the temperature value measured by the temperature sensor are collected. The state of the mirror is determined based on the sequence of light intensity measurements and the corresponding coded pattern sequence. Based on the determined state of the mirror and the temperature value, the operating power of the cooler is controlled so that the mirror temperature is dynamically maintained at the dew point temperature or frost point temperature, and the dew point temperature or frost point temperature is output.

[0018] A fourth aspect of the present invention provides a cold mirror dew point detection method based on single-pixel imaging, employing the cold mirror dew point detection system based on single-pixel imaging as described above, the detection method comprising: The gas to be tested is allowed to flow through the mirror, and the cooler is controlled to cool the mirror. The light emitted by the light source is modulated into a series of preset coded patterns by a spatial light modulator and projected onto the mirror surface; The light intensity signal reflected by the mirror is received by a single-pixel detector, and the light intensity measurement value corresponding to each of the coded patterns is collected. The state of the mirror is determined based on the sequence of light intensity measurements and the corresponding coded pattern sequence; When the mirror surface is determined to be in a state of dew or frost formation, the current mirror surface temperature is recorded as the dew point temperature or frost point temperature, and the cooler is controlled to stabilize the mirror surface temperature at the dew point temperature or frost point temperature.

[0019] The cold mirror type water vapor concentration detection system based on single-pixel imaging of the application comprises a mirror surface refrigeration module, a detection module and a control module, the mirror surface refrigeration module comprises a mirror surface, a refrigerator for refrigerating and heating the mirror surface and a temperature sensor for monitoring the temperature of the mirror surface in real time, the detection module comprises a light source, a spatial light modulator and a single-pixel detector, the spatial light modulator is used for modulating the light emitted by the light source into a series of preset coded patterns and projecting the coded patterns to the mirror surface, the single-pixel detector is used for receiving the light intensity signal carrying the state information of the mirror surface after the coded patterns are reflected by the mirror surface, and outputting the light intensity measurement value corresponding to each coded pattern, the control module is configured to: collect the light intensity measurement value output by the single-pixel detector and the temperature value measured by the temperature sensor, determine the state of the mirror surface based on the sequence of light intensity measurement values and the sequence of corresponding coded patterns, control the working power of the refrigerator according to the determined state of the mirror surface and the temperature value, so that the temperature of the mirror surface is dynamically maintained at the dew point temperature or the frost point temperature, and the dew point temperature or the frost point temperature is output, and the water vapor concentration is obtained according to the dew point temperature and the ambient temperature, in the process of detecting the state of the mirror surface, the detection system can distinguish the phase change from the pollutants, and has high reliability, the detection system uses the low-cost single-pixel detector combined with the spatial light modulator, replaces the expensive and complex area array camera and optical lens group in the traditional scheme, significantly reduces the system cost and volume, and the single-pixel detector combined with the spatial light modulator improves the reliability of water vapor concentration detection in a low dew point environment. BRIEF DESCRIPTION OF DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the application, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments of the application, and other drawings can also be obtained by those skilled in the art without any creative effort based on these drawings.

[0021] Figure 1 The overall module diagram of the cold mirror type water vapor concentration detection system based on single-pixel imaging provided by the embodiments of the application is shown in the figure. Figure 2 The flowchart of the cold mirror type dew point detection method based on single-pixel imaging provided by the embodiments of the application is shown in the figure. Figure 3 The flowchart of the cold mirror type dew point detection method based on single-pixel imaging provided by the embodiments of the application is shown in the figure. DETAILED DESCRIPTION

[0022] In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of but not all of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts should fall into the scope of the present application.

[0023] In the description of the present application, it should be understood that the terms "include" and "have" and any variations thereof used herein are intended to cover a non-exclusive inclusion, for example, a process, method, system, product or device including a series of steps or units need not be limited to those clearly listed steps or units, but can include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0024] It should be understood that the terms "length", "width", "upper", "lower", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only used to facilitate the description of the present application and simplify the description, and do not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation of the present application.

[0025] In addition, the terms "first", "second" are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the technical features indicated. Therefore, the features defined with "first", "second" can explicitly or implicitly include one or more of the features. It should be understood that the term "and / or" used herein is only a description of the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B, which means that there are three cases of A alone, A and B together, and B alone. In the description of the present application, unless otherwise specified, the meaning of "multiple" is two or more.

[0026] The existing cold mirror dew point meter has significant defects, and the interference of mirror surface pollution is difficult to overcome: the attachment of any non-aqueous pollutants (such as oil stains and dust) will change the reflectivity of the mirror surface, causing the light intensity signal to decay before reaching the true dew point, resulting in a false high dew point measurement value, and the existing automatic cleaning technology has limited effect on non-volatile pollutants; secondly, the optical system is complex and costly: the use of high-resolution industrial cameras and precision lens groups results in a large instrument size and high cost, which is not conducive to portability and low-cost popularization. Thirdly, the recognition ability and intelligent level are insufficient: the traditional threshold judgment method cannot distinguish between dew, frost and local pollution, and the camera imaging-based scheme also fails to deeply integrate intelligent diagnosis algorithms, resulting in low reliability and automation level under complex working conditions. In addition, when measuring very low dew point, the signal-to-noise ratio of the traditional photoelectric detection system is low and the response is slow, making it difficult to capture the starting point of phase change, resulting in limited measurement lower limit and slow response speed. The present application provides a single-pixel imaging-based cold mirror water vapor concentration and dew point detection system and method based on the above problems.

[0027] The single-pixel imaging-based cold mirror water vapor concentration and dew point detection system and method provided by the present application will be described in detail below in conjunction with specific embodiments.

[0028] Figure 1 The overall module diagram of the single-pixel imaging-based cold mirror water vapor concentration detection system provided by the present application is shown in Figure 1 The first aspect of the present application provides a single-pixel imaging-based cold mirror water vapor concentration detection system, which comprises a mirror surface refrigeration module 1, a detection module 2 and a control module 3. The mirror surface refrigeration module 1 comprises a mirror surface 11, a refrigerator 12 for refrigerating and heating the mirror surface, and a temperature sensor 13 for monitoring the temperature of the mirror surface in real time. The detection module 2 comprises a light source 21, a spatial light modulator 22 and a single-pixel detector 23. The spatial light modulator 22 is used to modulate the light emitted by the light source 21 into a series of preset coded patterns and project them onto the mirror surface 11. The single-pixel detector 23 is used to receive the light intensity signal carrying the mirror surface state information after the coded patterns are reflected by the mirror surface 11, and output the light intensity measurement value corresponding to each coded pattern. The control module 3 is configured to: collect the light intensity measurement value output by the single-pixel detector 23 and the temperature value measured by the temperature sensor 13; determine the state of the mirror surface 11 based on the sequence of light intensity measurement values and the corresponding sequence of coded patterns; control the working power of the refrigerator 12 according to the determined state of the mirror surface 11 and the temperature value, so that the temperature of the mirror surface 11 is dynamically maintained at the dew point temperature or the frost point temperature, and the dew point temperature or the frost point temperature is output; and obtain the water vapor concentration based on the dew point temperature and the environmental temperature. The mirror 11 of the present embodiment is made of a metal (e.g. gold, copper) with high thermal conductivity and high reflectivity, and is precisely polished to serve as a base for water vapor condensation. The refrigerator 12 of the present embodiment is preferably a semiconductor refrigerator (Thermoelectric Cooler; TEC), which can achieve rapid cooling and heating by changing the direction and magnitude of the electric current, and has a fast response speed and accurate temperature control. The temperature sensor 13 of the present embodiment is used to monitor the temperature of the mirror 11 in real time, and the specific structure of the temperature sensor 13 is not particularly limited in the present embodiment.

[0029] The light source 21 of the embodiment preferably is an infrared LED or laser diode with a central wavelength of 850 nm or 940 nm. This wavelength band has weak water vapor absorption, which can reduce the interference of gas absorption on the light intensity signal, and avoids visible light to avoid affecting the operator. The spatial light modulator 22 of the embodiment uses a Digital Micromirror Device (DMD), such as a DLP LightCrafter 4500 module. The DMD is composed of hundreds of thousands to millions of micromirrors, each representing a pixel. By controlling the "on" and "off" states of the micromirrors, a binary random or structured coded pattern sequence (such as a Hadamard basis) is generated. The single-pixel detector 23 of the embodiment uses a photodiode with high sensitivity and high signal-to-noise ratio in the corresponding wavelength band of the light source 21. The core of the control module 3 of the embodiment is an embedded industrial computer, which is connected to each hardware unit through a general-purpose input / output port. Exemplarily, the control module 3 includes a data acquisition unit: periodically reading the voltage output of the single-pixel detector 23 and the resistance value of the temperature sensor 13 through an Analog to Digital Converter (ADC) chip and digitizing them; a state determination unit: determining the state of the mirror surface 11, such as one of dew layer formation, frost layer formation, and pollution state, based on the light intensity measurement value sequence and the corresponding coded pattern sequence; a power control unit: when the state of the mirror surface 11 is determined to be dew layer formation or frost layer formation, controlling the working power of the cooler 12 according to the determined state of the mirror surface 11 and the temperature value, so that the temperature of the mirror surface 11 is dynamically maintained at the dew point temperature or the frost point temperature, and outputting the dew point temperature or the frost point temperature. The power control unit of the embodiment generates a control signal through a PID control algorithm to drive the driving circuit of the TEC, accurately adjusts the current and power of the TEC; a calculation unit: obtaining the water vapor concentration according to the dew point temperature and the ambient temperature. The calculation unit of the embodiment converts the measured dew point temperature and ambient temperature into water vapor concentration by calling a pre-stored conversion formula. The calculation formula for obtaining the water vapor concentration according to the dew point temperature and the ambient temperature is a prior art, for example, the Goff-Gratch formula recommended by the World Meteorological Organization is used to first calculate the saturated water vapor pressure, and then the actual water vapor pressure at the dew point of the mirror surface 11 is combined to finally calculate the relative humidity, i.e., the water vapor concentration parameter. After the system of the embodiment is started, the measured gas flows through the mirror 11, the control module 3 cools the mirror 11, and simultaneously drives the DMD to project the coded pattern according to the preset sequence, and the modulated light is reflected by the mirror 11 and received by the single-pixel detector 23. When the mirror 11 is clean and dry, the light is highly reflected, and the signal of the single-pixel detector 23 is strong. When the temperature of the mirror 11 drops to the dew / frost point, the condensate causes the light to scatter, and the signal of the single-pixel detector 23 is weakened. The control module 3 determines the state of the mirror 11 according to the sequence of light intensity measurement values and the corresponding sequence of coded patterns. Once the “dew / frost layer formation” state is identified, the current temperature of the mirror 11 is immediately locked as the dew / frost point temperature, and the PID closed-loop control is entered to dynamically maintain the phase change balance. The final water vapor concentration value is calculated using the dew point temperature and the ambient temperature collected by the thermometer 14.

[0030] The water vapor concentration detection system of the embodiment can be applied in the etching process or the wafer transmission process. For example, in the etching process, the real-time dew point of the chlorine gas is monitored to ensure that the water vapor concentration is below a certain value, so as to prevent the water vapor from reacting with the chlorine gas to generate corrosive hydrogen chloride and thereby damage the fine circuit pattern on the wafer surface. In the wafer transmission process, the dew point of the nitrogen gas used for purging is strictly controlled below -70°C. If the water vapor concentration detection system erroneously reports due to mirror contamination, the nitrogen gas with excessive water content will directly contact the wafer, causing the oxide film to thicken and other electrical parameter drifts, which seriously affect the performance and yield of the chip. The cold mirror type water vapor concentration detection system of the embodiment uses single-pixel imaging technology, replaces the expensive and complex area array camera and optical lens group in the traditional scheme by using only one low-cost single-pixel detector 23 combined with a spatial light modulator 22, and significantly reduces the system cost and volume. The spatial light modulator 22 encodes and modulates the target light field, and the single-pixel detector 23 uses a single-point detector without spatial resolution capability, which has an advantage in detecting the weak signal of the extremely thin frost layer formed at a low dew point, expands the lower limit of measurement, and the control module determines the state of the mirror 11 based on the sequence of light intensity measurement values and the corresponding sequence of coded patterns, which can distinguish between phase change and contamination, and has strong reliability. The system has the advantages of simple structure, low cost, strong detection capability in the non-visible light band and weak light environment.

[0031] Further, the control module 3 comprises an image reconstruction unit and an intelligent recognition unit. The image reconstruction unit is configured to reconstruct a two-dimensional image of the mirror 11 by a single-pixel imaging algorithm based on the sequence of light intensity measurements and the corresponding sequence of encoding patterns. The intelligent recognition unit comprises a trained machine learning model configured to analyze the reconstructed two-dimensional image and output a classification result representing the mirror 11 state, which at least includes one of clean, dew layer formation, frost layer formation, and pollution. For example, after the detection system of the present embodiment is started, the DMD projects 1024 Hadamard base patterns in turn, and the single-pixel detector synchronously collects 1024 light intensity values, which are transmitted to the control module. The image reconstruction unit quickly reconstructs a two-dimensional image of the mirror with 128x128 pixels using the TVAL3 algorithm (total variation minimization algorithm based on augmented Lagrangian and alternating direction algorithm). The intelligent recognition unit uses a lightweight pre-trained CNN model to perform real-time analysis on the image and classify the mirror state. The intelligent recognition unit of the present embodiment integrates a pre-trained lightweight convolutional neural network, which can automatically extract features from the reconstructed image and accurately classify the mirror 11 state. This scheme upgrades the traditional single light intensity threshold judgment to an intelligent diagnosis based on image visual features, which can clearly distinguish between dew layer, frost layer, and point / patch pollution, and fundamentally solves the core defect of false judgment caused by pollution, greatly improving the reliability and intelligent level of the measurement.

[0032] Further, the control module 3 further comprises a cleaning unit configured to trigger an alarm signal and / or automatically clean the mirror 11 when the mirror 11 state is identified as "pollution". Any non-aqueous contaminants (such as dust, oil stains, organic volatile condensation, salt) on the mirror 11 will change the optical properties (such as reducing reflectivity) and water phase change properties (such as reducing saturated water vapor pressure) of the mirror 11. This will cause the light intensity signal received by the photodetector to decay prematurely before the actual dew point is reached, and the system will mistakenly judge dew formation, resulting in a false and high dew point that deviates from the true value. Existing detection systems mainly use periodic manual cleaning to remove contaminants from the mirror 11, which requires interrupting the measurement and opening the instrument, and is operated by professional personnel, with very low automation level. For example, the automatic cleaning of the mirror 11 of the present embodiment includes the processes of high-temperature burning of the mirror 11 and gas flushing of the mirror 11. The organic matter is ablated by briefly heating the mirror 11 at a high temperature, and the non-volatile contaminants (such as salt stains and dust) are removed by gas flushing. The present embodiment realizes real-time and automatic response to the pollution state of the mirror 11, changes from passive deviation discovery to active early warning and intervention, significantly reduces the measurement interruption time and manual maintenance frequency caused by pollution, and improves the automation level and long-term operation stability of the equipment. Optionally, the machine learning model is a convolutional neural network or a support vector machine. The convolutional neural network (CNN) of the embodiment can directly process the two-dimensional image of the reconstructed mirror. For example, the CNN receives the reconstructed image, extracts multi-level features (such as edges, textures, and spot patterns) in the image, integrates the extracted features, uses a Softmax activation function, and outputs the probability distribution of four categories (clean, dew layer, frost layer, and pollution). The CNN of the embodiment uses a constructed data set to train the selected model. The entire data set is randomly divided into a training set (for example, 70%), a validation set (for example, 15%), and a test set (for example, 15%). The training set is used for model parameter learning, the validation set is used for adjusting hyperparameters and selecting the best model, and the test set is used for final evaluation of the model performance. The CNN scheme of the embodiment has high recognition accuracy and good automation; the SVM scheme has fast training speed in a small sample scenario. Specifically, the single-pixel imaging algorithm is an image reconstruction algorithm based on compressed sensing, and the encoding pattern is a Hadamard basis pattern. The embodiment preferably uses the Hadamard basis pattern as the basis of the measurement matrix Φ. The Hadamard matrix elements are only +1 and -1, which correspond to the on and off states of the DMD, and are orthogonal bases, which are beneficial to improve the signal-to-noise ratio and reconstruction speed of the reconstructed image. The embodiment uses the Hadamard basis pattern, which is simple to implement in hardware and has high modulation efficiency. The single-pixel imaging algorithm of the embodiment is an image reconstruction algorithm based on compressed sensing, which, in combination with compressed sensing, enables the system to quickly perceive the macroscopic changes of the mirror 11 state with only partial data acquisition, and enables faster initial response speed than traditional methods, which is beneficial to the monitoring of dynamic processes. Further, the light source 21 is an infrared waveband light source 21, and the single-pixel detector 23 is an infrared waveband detector. The light source 21 and the single-pixel detector 23 of the embodiment work in the infrared waveband. On the one hand, this avoids the discomfort that visible light may cause to the operator, and is more suitable for industrial sites that require light avoidance; on the other hand, the detection sensitivity of a high-performance infrared single-point detector (such as a photomultiplier tube) in weak light is much higher than that of an ordinary area array CMOS camera, which makes the system have an advantage in detecting the weak scattering signals generated by the extremely thin and sparse frost layer formed at a low dew point, effectively expanding the lower limit of the instrument measurement. Preferably, the detection system of the present embodiment further comprises a gas pre-processing module, which includes a filter, a flow controller and a pressure regulator. Exemplarily, the pre-processing module of the present embodiment is connected at the gas inlet. The filter is used to remove solid particles in the gas to be measured, thereby reducing the pollution of the mirror 11 from the source. The flow controller ensures that the gas flows through the mirror 11 at a constant flow rate, thereby ensuring the stability and repeatability of the measurement. The pressure regulator stabilizes the gas pressure at a set value. The present embodiment provides clean, stable and pressure-controlled sample gas for the core detection unit through the pre-processing module, effectively reduces external interference, and ensures the accuracy and long-term reliability of the final water vapor concentration measurement result. The single-pixel imaging-based cold mirror water vapor concentration detection system of the present embodiment uses a spatial light modulator and a single-pixel detector to replace an expensive high-resolution industrial camera and a complex lens group, thereby significantly reducing the hardware cost and volume of the system and facilitating the miniaturization and portability of the device. The high light collection efficiency inherent in the single-pixel detector provides a natural advantage in detecting the weak signal of the extremely thin frost layer formed at a low dew point, thereby extending the lower limit of the measurement.

[0033] Figure 2 The flow chart of the single-pixel imaging-based cold mirror water vapor concentration detection method provided by the present embodiment is shown in Figure 2 The second aspect of the present embodiment provides a single-pixel imaging-based cold mirror water vapor concentration detection method, which uses the single-pixel imaging-based cold mirror water vapor concentration detection system as described in any one of the above embodiments. The detection method comprises the following steps: S101, flowing the gas to be measured through the mirror and controlling the refrigerator to cool the mirror; Specifically, the mirror 11 is installed in a detection chamber, and the detection chamber is provided with a gas inlet and a gas outlet, and the gas inlet is connected with a gas source. The structure inside the chamber is designed to guide the gas to flow smoothly and uniformly over the surface of the mirror 11. The refrigerator 12 is controlled to reduce the temperature of the mirror 11 to the dew point / frost point temperature of the gas.

[0034] S102, modulating the light emitted by the light source into a series of preset coded patterns by the spatial light modulator and projecting the coded patterns onto the mirror; Specifically, the DMD continuously projects N (for example, 1024) pre-stored Hadamard basis patterns at a frequency of up to kHz.

[0035] S103, receiving the light intensity signal reflected by the mirror by the single-pixel detector, and collecting the light intensity measurement value corresponding to each coded pattern; Specifically, the single-pixel detector 23 receives the light intensity signal carrying the state information of the mirror 11 after the coded pattern is reflected by the mirror 11, and outputs the light intensity measurement value corresponding to each coded pattern.

[0036] S104, determining the state of the mirror according to the sequence of light intensity measurement values and the corresponding sequence of encoded patterns; Specifically, the image reconstruction unit reconstructs a two-dimensional image of the mirror 11 quickly (delay <100 ms) according to the sequence of light intensity measurement values and the corresponding sequence of encoded patterns using the TVAL3 algorithm, and then the intelligent recognition unit (for example, using CNN) classifies the image to determine the state of the mirror, for example, one of clean, dew layer formation, frost layer formation, and pollution.

[0037] S105, when it is determined that the mirror state is dew layer formation or frost layer formation, recording the current mirror temperature as the dew point temperature or the frost point temperature, and controlling the refrigerator to stabilize the mirror temperature at the dew point temperature or the frost point temperature. Exemplarily, when the detection system identifies the dew layer formation target state, the working power of the refrigerator is controlled according to the determined dew layer formation state and temperature value, so that the mirror temperature is dynamically maintained at the dew point temperature, and the dew point temperature is output. The power control unit of the embodiment generates a control signal by a PID control algorithm to drive the driving circuit of the TEC, and accurately adjusts the power of the refrigerator.

[0038] S106, obtaining the water vapor concentration according to the dew point temperature and the ambient temperature.

[0039] Specifically, the first step is to calculate the actual water vapor pressure e from the dew point temperature Td

[0040] e: actual water vapor pressure (unit: hPa, 1 hPa = 1 mbar) T d : dew point temperature (unit: ℃) exp: natural exponential function When the ambient temperature and the dew point temperature are lower than 0℃ (for supercooled water), the formula for the ice surface is used:

[0041] The second step is to calculate the relative humidity RH from the actual water vapor pressure e and the ambient temperature T First, calculate the saturation water vapor pressure e s :

[0042] The water vapor concentration, i.e., the relative humidity (RH), is:

[0043] Further, the step of determining the state of the mirror 11 comprises: reconstructing a two-dimensional image of the mirror 11 by a single-pixel imaging algorithm according to the sequence of light intensity measurements and the corresponding sequence of coded patterns; inputting the reconstructed two-dimensional image into a trained machine learning model to obtain a classification result representing the state of the mirror 11, the state of the mirror 11 including at least one of clean, dew layer formation, frost layer formation, and contamination. Illustratively, after the detection system of the embodiment is started, the DMD projects 1024 Hadamard base patterns in turn, and the single-pixel detector synchronously collects 1024 light intensity values, which are transmitted to the control module. The image reconstruction unit quickly reconstructs a two-dimensional image of the mirror with 128x128 pixels using the TVAL3 algorithm (total variation minimization algorithm based on augmented Lagrangian and alternating direction algorithm). The intelligent recognition unit uses a lightweight pre-trained CNN model to analyze the image in real time and classify and recognize the state of the mirror. The intelligent recognition unit of the embodiment integrates a pre-trained lightweight convolutional neural network, which can automatically extract features from the reconstructed image and accurately classify the state of the mirror 11. This scheme upgrades the traditional single light intensity threshold judgment to an intelligent diagnosis based on image visual features, which can clearly distinguish between dew layer, frost layer, and point-shaped / pie-shaped contaminants, and fundamentally solves the core defect of false judgment caused by contaminants, greatly improving the reliability and intelligent level of the measurement. In an embodiment, the cold mirror type water vapor concentration detection method further comprises triggering an alarm and / or automatically cleaning the mirror 11 when it is determined that the state of the mirror 11 is “contamination”. Any non-aqueous contaminants (such as dust, oil stains, organic volatile condensation, and salt) on the mirror 11 will change the optical properties (such as reducing reflectivity) of the mirror 11 and the phase change properties (reducing the saturation water vapor pressure) of water. This will cause the light intensity signal received by the photodetector to decay prematurely before the actual dew point is reached, and the system will mistakenly judge that dewing has occurred, resulting in a false and high dew point that deviates from the true value. Existing detection systems mainly use periodic manual cleaning to remove contaminants from the mirror 11, which requires interrupting the measurement and opening the instrument, and is operated by professional personnel, with a very low degree of automation. Illustratively, the automatic cleaning of the mirror 11 of the embodiment includes the processes of high-temperature burning of the mirror 11 and gas flushing of the mirror 11. The organic matter is ablated by briefly heating the mirror 11 to a high temperature, and the non-volatile contaminants (such as salt stains and dust) are removed by gas flushing. The embodiment realizes real-time and automatic response to the contamination state of the mirror 11, changes from passive discovery of deviation to active early warning and intervention, significantly reduces the measurement interruption time and manual maintenance frequency caused by contamination, and improves the degree of automation and long-term operation stability of the equipment. Preferably, the single-pixel imaging algorithm adopts an image reconstruction algorithm of compressed sensing; and the specific steps of reconstructing the two-dimensional image of the mirror surface 11 by the single-pixel imaging algorithm include: constructing the coded pattern as a measurement matrix Φ; constructing the light intensity measurement value as a measurement vector Y; and reconstructing a vector X representing the image of the mirror surface 11 by solving the least square solution of the linear equation set Y = ΦX. Illustratively, the spatial light modulator 22 is configured to modulate the light emitted by the light source 21 into a preset coded pattern, vectorize the coded pattern, for example, vectorize the coded pattern as Φi=[ϕi1,ϕi2,...,ϕin], perform m measurements, and construct an m*n measurement matrix Φ=[Φ1 T ,Φ2 T ,...,Φm T ] T The single-pixel detector 23 receives the light intensity signal carrying the state information of the mirror surface 11 after the coded pattern is reflected by the mirror surface 11, and outputs a light intensity measurement value corresponding to each coded pattern, and constructs the light intensity measurement value as a measurement vector Y, Y=[ y 1, y 2,..., ym ] T The linear equation set Y = ΦX is solved to obtain a vector X representing the two-dimensional image of the mirror surface 11. In this embodiment, the least square method is used for solving, which is simple and efficient. The single-pixel imaging algorithm in this embodiment adopts an image reconstruction algorithm of compressed sensing, which can quickly perceive the trend of the state change of the mirror surface under the premise of incomplete sampling, and can achieve a faster dynamic response than the traditional "waiting-judging" mode.

[0044] A third aspect of the embodiment provides a cold mirror dew point detection system based on single-pixel imaging, comprising: A mirror surface refrigeration module, comprising a mirror surface, a refrigerator for refrigerating and heating the mirror surface, and a temperature sensor for monitoring the temperature of the mirror surface in real time; A detection module, comprising a light source, a spatial light modulator, and a single-pixel detector; the spatial light modulator is configured to modulate the light emitted by the light source into a series of preset coded patterns and project the coded patterns onto the mirror surface; and the single-pixel detector is configured to receive the light intensity signal carrying the state information of the mirror surface after the coded pattern is reflected by the mirror surface, and output a light intensity measurement value corresponding to each coded pattern; A control module configured to: Collect the light intensity measurement value output by the single-pixel detector and the temperature value measured by the temperature sensor; Determine the state of the mirror surface based on the sequence of light intensity measurement values and the sequence of corresponding coded patterns; According to the determined state of the mirror and the temperature value, the working power of the refrigerator is controlled to dynamically maintain the mirror temperature at the dew point temperature or the frost point temperature, and the dew point temperature or the frost point temperature is output.

[0045] The detection system can distinguish phase change and pollutants during detection of the mirror state, and has high reliability. The detection system uses a low-cost single-pixel detector combined with a spatial light modulator to replace an expensive and complex area array camera and an optical lens group in a traditional scheme, significantly reduces the system cost and volume, and improves the reliability of detection in a low dew point environment.

[0046] Figure 3 The flowchart of the cold mirror type dew point detection method based on single-pixel imaging provided by the embodiment is shown in Figure 3 The fourth aspect of the embodiment provides a cold mirror type dew point detection method based on single-pixel imaging, which uses the cold mirror type dew point detection system based on single-pixel imaging as described above. The detection method comprises the following steps. S201, flowing the gas to be detected through the mirror, and controlling the refrigerator to cool the mirror; S202, modulating light emitted by a light source into a series of preset coded patterns by a spatial light modulator and projecting the coded patterns onto the mirror; S203, receiving the light intensity signal reflected by the mirror by a single-pixel detector, and collecting the light intensity measurement value corresponding to each coded pattern; S204, determining the state of the mirror according to the sequence of light intensity measurement values and the sequence of corresponding coded patterns; S205, when it is determined that the mirror state is dew layer formation or frost layer formation, recording the current temperature of the mirror as the dew point temperature or the frost point temperature, and controlling the refrigerator to stabilize the mirror temperature at the dew point temperature or the frost point temperature.

[0047] The implementation process of S201-S205 is the same as that of S101-S105.

[0048] The cold mirror type dew point detection method can distinguish phase change and pollutants during detection of the mirror state, and has high reliability. The detection system uses a low-cost single-pixel detector combined with a spatial light modulator to replace an expensive and complex area array camera and an optical lens group in a traditional scheme, significantly reduces the system cost and volume, and improves the reliability of detection in a low dew point environment.

[0049] The cold mirror dew point detection method based on single-pixel imaging can be applied in the field of on-line monitoring of semiconductor manufacturing processes. For example, in the semiconductor front-end process, the inlet dew point monitoring of the chemical vapor deposition reaction chamber is the key to ensuring the quality of the thin film, and it is usually required that the dew point of the special gas be stabilized below-70 DEG C. In the application of lithium battery drying, dew point measurement is also the only objective standard for judging whether the drying process meets the standard, and it is usually required that the dew point of the internal environment of the battery be controlled below-40 DEG C to-60 DEG C.

[0050] In the above description, the description of the terms "one embodiment", "some embodiments", "an example", "a specific example" and the like means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In the present specification, the illustrative description of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any appropriate manner in any one or more embodiments or examples. In addition, the person skilled in the art can combine and combine the different embodiments or examples described in the specification and the features of the different embodiments or examples without contradiction.

[0051] The above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A cold mirror-based water vapor concentration detection system based on single-pixel imaging, characterized in that, include: A mirror cooling module includes a mirror, a cooler for cooling and heating the mirror, and a temperature sensor for real-time monitoring of the mirror temperature. The detection module includes a light source, a spatial light modulator, and a single-pixel detector; The spatial light modulator is used to modulate the light emitted by the light source into a series of preset coded patterns and project them onto the mirror surface; the single-pixel detector is used to receive the light intensity signal carrying the mirror surface state information after the coded pattern is reflected by the mirror surface, and output the light intensity measurement value corresponding to each coded pattern. The control module is configured as follows: The light intensity measurement value output by the single-pixel detector and the temperature value measured by the temperature sensor are collected. The state of the mirror is determined based on the sequence of light intensity measurements and the corresponding coded pattern sequence. Based on the determined state of the mirror and the temperature value, the operating power of the cooler is controlled so that the mirror temperature is dynamically maintained at the dew point temperature or frost point temperature, and the dew point temperature or frost point temperature is output. The water vapor concentration is obtained based on the dew point temperature and the ambient temperature.

2. The cold mirror-type water vapor concentration detection system based on single-pixel imaging according to claim 1, characterized in that, The control module includes: The image reconstruction unit is configured to reconstruct a two-dimensional image of the mirror surface based on the light intensity measurement value sequence and the corresponding coded pattern sequence using a single-pixel imaging algorithm. The intelligent recognition unit, including a trained machine learning model, is configured to analyze the reconstructed two-dimensional image and output a classification result representing the mirror state, which includes at least one of clean, dew-covered, frost-covered, and contaminated.

3. The cold mirror-type water vapor concentration detection system based on single-pixel imaging according to claim 2, characterized in that: The control module further includes a cleaning unit, which is configured to trigger an alarm signal and / or automatically clean the mirror when the mirror surface is identified as "contaminated".

4. The cold mirror-type water vapor concentration detection system based on single-pixel imaging according to claim 2, characterized in that: The machine learning model is a convolutional neural network or a support vector machine.

5. The cold mirror-type water vapor concentration detection system based on single-pixel imaging according to claim 2, characterized in that: The single-pixel imaging algorithm is an image reconstruction algorithm based on compressed sensing; the encoded pattern is a Hadamard pattern.

6. The cold mirror-type water vapor concentration detection system based on single-pixel imaging according to claim 1, characterized in that: The light source is an infrared light source, and the single-pixel detector is an infrared detector.

7. The cold mirror-type water vapor concentration detection system based on single-pixel imaging according to claim 1, characterized in that: The detection system also includes a gas pretreatment module, which includes a filter, a flow controller, and a pressure regulator.

8. A method for detecting water vapor concentration using a cold mirror based on single-pixel imaging, employing the cold mirror-based water vapor concentration detection system according to any one of claims 1-7, characterized in that, The detection method includes: The gas to be tested is allowed to flow through the mirror, and the cooler is controlled to cool the mirror. The light emitted by the light source is modulated into a series of preset coded patterns by a spatial light modulator and projected onto the mirror surface; The light intensity signal reflected by the mirror is received by a single-pixel detector, and the light intensity measurement value corresponding to each of the coded patterns is collected. The state of the mirror is determined based on the sequence of light intensity measurements and the corresponding coded pattern sequence; When the mirror surface is determined to be in the state of dew or frost formation, the current mirror surface temperature is recorded as the dew point temperature or frost point temperature, and the cooler is controlled to stabilize the mirror surface temperature at the dew point temperature or frost point temperature. The water vapor concentration is obtained based on the dew point temperature and the ambient temperature.

9. The cold mirror-type water vapor concentration detection system based on single-pixel imaging according to claim 8, characterized in that: The step of determining the state of the mirror includes: reconstructing a two-dimensional image of the mirror using a single-pixel imaging algorithm based on the light intensity measurement value sequence and the corresponding coded pattern sequence; inputting the reconstructed two-dimensional image into a trained machine learning model to obtain a classification result representing the state of the mirror, wherein the state of the mirror includes at least one of clean, dew formation, frost formation, and contamination.

10. The cold mirror-type water vapor concentration detection system based on single-pixel imaging according to claim 8, characterized in that, Also includes: When the mirror surface is determined to be "contaminated", an alarm is triggered and / or the mirror surface is automatically cleaned.

11. The cold mirror-type water vapor concentration detection system based on single-pixel imaging according to claim 9, characterized in that: The single-pixel imaging algorithm employs a compressed sensing image reconstruction algorithm. The specific steps for reconstructing the two-dimensional image of the mirror using the single-pixel imaging algorithm include: constructing the coded pattern into a measurement matrix Φ; constructing the light intensity measurement value into a measurement vector Y; and reconstructing the vector X representing the two-dimensional image of the mirror by solving the least-squares solution of the linear equation system Y = ΦX.

12. A cold mirror dew point detection system based on single-pixel imaging, characterized in that, include: A mirror cooling module includes a mirror, a cooler for cooling and heating the mirror, and a temperature sensor for real-time monitoring of the mirror temperature. The detection module includes a light source, a spatial light modulator, and a single-pixel detector; The spatial light modulator is used to modulate the light emitted by the light source into a series of preset coded patterns and project them onto the mirror surface; the single-pixel detector is used to receive the light intensity signal carrying the mirror surface state information after the coded pattern is reflected by the mirror surface, and output the light intensity measurement value corresponding to each coded pattern. The control module is configured as follows: The light intensity measurement value output by the single-pixel detector and the temperature value measured by the temperature sensor are collected. The state of the mirror is determined based on the sequence of light intensity measurements and the corresponding coded pattern sequence. Based on the determined state of the mirror and the temperature value, the operating power of the cooler is controlled so that the mirror temperature is dynamically maintained at the dew point temperature or frost point temperature, and the dew point temperature or frost point temperature is output.

13. A cold mirror dew point detection method based on single-pixel imaging, employing the cold mirror dew point detection system based on single-pixel imaging as described in claim 12, characterized in that, The detection method includes: The gas to be tested is allowed to flow through the mirror, and the cooler is controlled to cool the mirror. The light emitted by the light source is modulated into a series of preset coded patterns by a spatial light modulator and projected onto the mirror surface; The light intensity signal reflected by the mirror is received by a single-pixel detector, and the light intensity measurement value corresponding to each of the coded patterns is collected. The state of the mirror is determined based on the sequence of light intensity measurements and the corresponding coded pattern sequence; When the mirror surface is determined to be in a state of dew or frost formation, the current mirror surface temperature is recorded as the dew point temperature or frost point temperature, and the cooler is controlled to stabilize the mirror surface temperature at the dew point temperature or frost point temperature.

Images