Online detection device for trace impurities in electronic grade quartz crystal material

Abstract

The invention relates to the technical field of material detection, in particular to an online detection device for trace impurities in an electronic-grade quartz crystal material. Collecting material data of a plurality of target materials, then creating a detection model, inputting a plurality of pre-processed data into the detection model, creating a standard impurity spectrum database based on the pre-processed data of each target material through the detection model, and performing calibration by combining a dynamic correlation matching algorithm with a signal enhancement factor to obtain a standard impurity spectrum database; the method comprises the following steps of: training a target material to detect the content of impurities contained in the target material, and finally inputting real-time data into a trained detection model to detect the content of impurities contained in a real-time material. By fusing a bionic micro-nano enrichment technology and a light trace-cavity enhanced Raman technology, a signal enhancement factor reaches 10 orders of magnitude; the method can accurately detect trace impurities and micro elements in the electronic-grade quartz crystal material, and adopts multispectral combination and anti-interference design, so that the detection error is less than or equal to + / -2%, and the detection sensitivity is improved.

Description

Technical Field

[0001] This invention relates to the field of materials testing technology, specifically to an online detection device for trace impurities in electronic-grade quartz crystal materials. Background Technology

[0002] Electronic-grade quartz crystal materials (SiO2 purity ≥ 99.99%) have become core raw materials in cutting-edge fields such as semiconductor silicon melting, photovoltaic crucibles, and optical components due to their high hardness, high temperature resistance, low coefficient of thermal expansion, excellent light transmittance, and electrical insulation. Their performance stability and reliability are highly dependent on material purity. The presence of trace (ppm level), especially ultra-trace (ppb / ppt level) impurities and micro-elements, can significantly degrade material performance, such as reducing carrier lifetime, increasing optical loss, and affecting thermal stability, ultimately leading to the failure of downstream devices. Therefore, trace detection, micro-detection, and micro-element analysis are core aspects of the production and quality control of electronic-grade quartz crystal materials.

[0003] Currently, the mainstream technologies for detecting trace impurities and microelements in electronic-grade quartz crystal materials include inductively coupled plasma mass spectrometry (ICP-MS), glow discharge mass spectrometry (GDMS), and Raman spectroscopy. Traditional Raman spectroscopy is limited by low signal intensity and severe fluorescence interference, resulting in insufficient sensitivity for detecting trace impurities and microelements, making it difficult to meet the requirements for ppb-level detection. Furthermore, it is susceptible to the "coffee ring effect," leading to uneven sample distribution and poor signal reproducibility. Summary of the Invention

[0004] The purpose of this invention is to address the problems existing in the background technology by proposing an online detection device for trace impurities in electronic-grade quartz crystal materials.

[0005] The technical solution of this invention: On the one hand, this application provides an online detection device for trace impurities in electronic-grade quartz crystal materials, comprising: A conveying module; the conveying module conveys the target material, and the conveying includes a conveying track, an adaptive clamping mechanism, and a laser positioning unit; Enrichment module; The enrichment module enables rapid enrichment of trace impurities and micro-elements on the surface of the target material. The enrichment module includes a copper substrate and a silver nanowire heating unit. Excitation module; the excitation module provides multi-band stable laser light to adapt to the characteristic spectral response of different trace impurities and micro elements. The excitation module includes a laser source group, an optical path adjustment unit and a wavelength switching unit. Acquisition module; The acquisition module enables efficient enhancement and accurate acquisition of trace impurity Raman signals. The acquisition module includes a resonant cavity assembly, a signal filtering unit, and a high-sensitivity detector. The detection module is communicatively connected to the acquisition module. Through the detection module, real-time processing, qualitative and quantitative analysis, and result output of the detection data are realized. The detection module includes a data preprocessing unit, a characteristic spectrum matching unit, a quantitative calculation unit, and a result output unit.

[0006] Preferably, the adaptive clamping mechanism is disposed on both sides of the conveying track, and the adaptive clamping mechanism is used to fix different types of target materials. The positioning accuracy of the laser positioning unit is set to ±0.01mm.

[0007] Preferably, the surface of the copper substrate is etched into a fish-scale-like micro / nano structure, and the silver nanowire heating unit is integrated inside the copper substrate.

[0008] Preferably, the laser source group includes a multi-band laser generator, and the optical path adjustment unit includes a high-precision reflector and lens group to achieve collimation, focusing and calibration of the laser beam. The size of the focused spot is adjustable, and the switching response time of the wavelength switching unit is ≤1 second.

[0009] Preferably, the resonant cavity assembly is configured with a double-waisted folded resonant cavity design, so that the laser is reflected multiple times within the cavity. The high-sensitivity detector has a detection wavelength range of 200-4200cm⁻¹ and a resolution of ≤5cm⁻¹@585nm.

[0010] On the other hand, this application also provides an online detection method for trace impurities in electronic-grade quartz crystal materials, applied to the online detection device for trace impurities in electronic-grade quartz crystal materials described in any one of the foregoing, comprising: Material data of multiple target materials are collected, and all collected material data are preprocessed to obtain multiple preprocessed data. Create a detection model; Multiple preprocessed data are input into the detection model. The detection model creates a standard impurity spectral database based on the preprocessed data of each target material. The impurity content in the target material is detected by combining a dynamic correlation matching algorithm with a signal enhancement factor calibration, and the trained detection model is obtained. Real-time data of the material is collected and input into the trained detection model to detect the impurity content in the material.

[0011] Preferably, the step involves collecting material data from multiple target materials and preprocessing all collected material data to obtain multiple preprocessed data sets, including: Create a materials database; For each target material, corresponding acquisition parameters are set, and material data of multiple target materials are collected based on the acquisition parameters. All collected material data are put into the material database. The acquisition parameters include clamping angle and clamping force. Randomly select the material data of a target material from the material database; Determine if there are duplicate data in the material data of the target material; If duplicate data exists in the material data of the target material, delete the duplicate data; Returns the material data of a target material randomly selected from the material database, until all target materials have been selected, resulting in multiple preprocessed data sets.

[0012] Preferably, the step of inputting multiple preprocessed data into the detection model, creating a standard impurity spectral database based on the preprocessed data of each target material through the detection model, and detecting the impurity content in the target material through a dynamic correlation matching algorithm combined with signal enhancement factor calibration, yields a trained detection model, including: All preprocessed information is divided into training and test sets according to a random ratio; The training set is input into the detection model. The detection model collects the illumination data of each target material based on multispectral excitation technology. Then, the types of impurities are calculated by dynamic correlation matching algorithm, and the content of each type of impurity is calculated by standard curve method and internal standard method to obtain the trained detection model. Input the test set into the trained detection model to verify whether the training of the detection model is complete.

[0013] Preferably, the step of inputting the training set into the detection model, acquiring illumination data for each target material based on multispectral excitation technology using the detection model, calculating the types of impurities contained therein using a dynamic correlation matching algorithm, and calculating the content of each type of impurity using a standard curve method and an internal standard method, to obtain the trained detection model, includes: Randomly select a target material from the training set; The characteristic spectral signal obtained by multispectral excitation of the target material is acquired, and the type of impurities contained in the target material is obtained through the characteristic spectral signal; The content of each impurity type is calculated using the signal enhancement factor; Return to the point where a target material is randomly selected from the training set, and continue until all target materials in the training set have been selected, to obtain the type and content of impurities contained in each target material.

[0014] Preferably, the step of collecting real-time data of the material and inputting the real-time data into the trained detection model to detect the impurity content contained in the material includes: Collect real-time data from real-time materials; Real-time data is input into the trained detection model, and the types of impurities contained in the material and the impurity content of each type of impurity are obtained through the trained detection model. Set content thresholds for each type of impurity; Sequentially determine whether the impurity content of each type of impurity in the real-time material is greater than or equal to the content threshold; If the content of at least one type of impurity is greater than or equal to the content threshold, an early warning will be issued.

[0015] Compared with the prior art, the above-mentioned technical solution of the present invention has the following beneficial technical effects: By collecting material data from multiple target materials and preprocessing all collected material data to obtain multiple preprocessed data, a detection model is created. The multiple preprocessed data are then input into the detection model. The detection model creates a standard impurity spectral database based on the preprocessed data of each target material. Through dynamic correlation matching algorithm combined with signal enhancement factor calibration, the impurity content in the target material is detected, resulting in a trained detection model. Finally, real-time data of the material is collected and input into the trained detection model to detect the impurity content in the material. This application innovatively integrates biomimetic micro-nano enrichment technology with trace-cavity enhanced Raman technology, achieving a signal enhancement factor on the order of 10³ and a detection sensitivity on the order of ppt. It can accurately detect trace impurities and micro-elements in electronic-grade quartz crystal materials. At the same time, it adopts multispectral coupling and anti-interference design, with a detection error ≤ ±2%, excellent reproducibility, and improved detection sensitivity. Attached Figure Description

[0016] Figure 1 This is a schematic flowchart of an online detection method for trace impurities in electronic-grade quartz crystal materials proposed in this invention. Figure 2 This is a schematic diagram of the principle of an online detection device for trace impurities in electronic-grade quartz crystal materials proposed in this invention; Figure descriptions: 100, conveying module; 101, conveying track; 102, adaptive clamping mechanism; 103. Laser positioning unit; 200. Enrichment module; 201. Copper substrate; 202. Silver nanowire heating unit; 300. Excitation module; 301. Laser source group; 302. Optical path adjustment unit; 303. Wavelength switching unit; 400. Acquisition module; 401. Resonant cavity assembly; 402. Signal filtering unit; 403. High-sensitivity detector; 500. Detection module; 501. Data preprocessing unit; 502. Feature spectrum matching unit; 503. Quantitative calculation unit; 504. Result output unit. Detailed Implementation

[0017] Example 1, as Figure 1 As shown, this invention proposes an online detection device for trace impurities in electronic-grade quartz crystal materials, comprising a transport module 100, an enrichment module 200, an excitation module 300, a collection module 400, and a detection module 500. The transport module 100 transports the target material and includes a transport track 101, an adaptive clamping mechanism 102, and a laser positioning unit 103. The enrichment module 200 rapidly enriches trace impurities and micro-elements on the surface of the target material, including a copper substrate 201 and a silver nanowire heating unit 202. The excitation module 300 provides multi-band stable laser light, adaptable to different trace impurities and micro-elements. The excitation module 300 includes a laser source group 301, an optical path adjustment unit 302, and a wavelength switching unit 303. The acquisition module 400 achieves efficient enhancement and accurate acquisition of trace impurity Raman signals. The acquisition module 400 includes a resonant cavity assembly 401, a signal filtering unit 402, and a high-sensitivity detector 403. The detection module 500 is communicatively connected to the acquisition module 400. The detection module 500 enables real-time processing, qualitative and quantitative analysis, and result output of detection data. The detection module 500 includes a data preprocessing unit 501, a characteristic spectrum matching unit 502, a quantitative calculation unit 503, and a result output unit 504.

[0018] In this invention, the online, rapid, and accurate detection of trace impurities (such as Fe, Al, Na, K, etc., with a concentration ≤10ppb) in electronic-grade quartz crystals (purity ≥99.999%) is achieved through the synergy of various modules. The device integrates delivery, enrichment, excitation, acquisition, and detection, eliminating the need for offline sample preparation and improving detection efficiency by 5-10 times. It is suitable for continuous monitoring on production lines. The enrichment module combined with resonant cavity enhancement technology achieves a detection limit as low as 0.1ppb, far below the traditional detection limit, meeting the ultra-pure requirements of electronic-grade quartz crystals. Moreover, the multi-band laser and dynamic wavelength switching cover the entire ultraviolet-visible-near-infrared spectrum, enabling simultaneous detection of metallic / non-metallic / transition metal impurities. Finally, the data preprocessing unit 501 establishes an impurity characteristic spectral library based on machine learning through the characteristic spectral matching unit 502. The quantitative calculation unit 503 combines the standard curve method and internal standard method to eliminate the influence of instrument fluctuations. The result output unit 504 generates impurity type, content, and early warning reports in real time.

[0019] In an optional embodiment, the adaptive clamping mechanism 102 is disposed on both sides of the conveying track 101, and the adaptive clamping mechanism 102 is used to fix different types of target materials. The positioning accuracy of the laser positioning unit 103 is set to ±0.01mm.

[0020] It should be noted that the adaptive clamping mechanism 102 is symmetrically arranged on both sides of the conveying track 101. It has a built-in piezoelectric ceramic driven finger clamp, which adjusts the clamp opening (0-50mm) and force (0.1-0.5N) through force feedback closed-loop control. It is compatible with different shapes of quartz crystals. The laser positioning unit 103 adopts crosshair laser + machine vision, which identifies the material contour through edge detection algorithm and corrects the conveying position in real time to ensure that the center of the material coincides with the excitation spot, so that the detection deviation is ≤±0.01mm.

[0021] In an optional embodiment, the surface of the copper substrate 201 is etched into a fish-scale-like micro / nano structure, and the silver nanowire heating unit 202 is integrated inside the copper substrate 201.

[0022] It should be noted that the copper substrate 201 is etched by ultraviolet nanosecond laser to form a fish-scale-like micro-nano structure. The detection area has superhydrophobic and high adhesion properties, which can firmly lock the trace impurity droplets precipitated on the sample surface. The isolation area has superhydrophobic and low adhesion properties, which prevents the impurity liquid from overflowing. It supports multi-channel high-throughput detection. The fish-scale-like structure increases the specific surface area, and the copper substrate's affinity for impurities improves the enrichment efficiency.

[0023] The silver nanowire heating unit 202 is integrated inside the copper substrate 201. When an 8-15V DC voltage is applied, the temperature of the detection area can be rapidly raised to 50-100℃, shortening the natural evaporation time of impurity droplets from 30 minutes to 56 seconds, improving the detection efficiency by 38 times, achieving rapid enrichment of trace impurities, avoiding local overheating, preventing thermal stress cracking of quartz crystals, and precisely controlling the enrichment temperature.

[0024] By combining etched structures with magnetron sputtering, large Ag clusters are formed at the junctions of the grooves, while small Ag clusters are distributed in the flat areas of the fish scale pattern, artificially creating high-density electromagnetic field "hot spots" to provide a physical basis for signal enhancement.

[0025] In an optional embodiment, the laser source group 301 includes a multi-band laser generator, the optical path adjustment unit 302 includes a high-precision reflector and lens group to achieve collimation, focusing and calibration of the laser beam, the focused spot size is adjustable, and the switching response time of the wavelength switching unit 303 is ≤1 second.

[0026] It should be noted that the laser light source group 301 provides multi-band stable laser light, which can be adapted to the characteristic spectral response of different trace impurities and micro elements, thereby improving the applicability of detection.

[0027] The laser source group 301 includes multi-band laser generators of 266nm (ultraviolet), 532nm (visible light), and 775nm (near infrared). The 266nm ultraviolet laser can enhance the detection specificity of biomolecules and non-metallic impurities, the 532nm visible light is suitable for the detection of most metallic impurities, and the 775nm near infrared laser can reduce sample damage and meet the requirements of non-destructive detection.

[0028] The wavelength switching unit 303 adopts an FPGA-controlled AOM, which quickly switches the diffraction order through electrical signals. Different wavelength bands of laser cover the characteristic absorption peaks of impurities, avoiding missed detection of single wavelengths. The 1-second wavelength switching supports rapid switching of detection modes, meeting the switching needs of multiple product specifications on the production line.

[0029] In an optional embodiment, the resonant cavity assembly 401 is configured with a double-waisted folded resonant cavity design, so that the laser is reflected multiple times within the cavity, and the high-sensitivity detector 403 has a detection wavelength range of 200-4200cm⁻¹ and a resolution of ≤5cm⁻¹@585nm.

[0030] It should be noted that the acquisition module 400 achieves efficient enhancement and accurate acquisition of trace impurity Raman signals. The core technology used is optical trace-cavity enhanced Raman, which solves the problems of weak signals and large interference in traditional Raman spectra.

[0031] The resonant cavity assembly 401, designed with a double-waisted folded resonant cavity, enables an effective optical path of up to kilometers, exponentially increases the number of light-sample interactions, and achieves a signal enhancement factor on the order of 10³, thus enabling the effective capture of trace impurity signals at the ppb / ppt level.

[0032] The range of 200-4200 cm⁻¹ covers the entire Raman shift range, and the resolution of ≤5 cm⁻¹ can distinguish impurities with similar characteristic peaks.

[0033] like Figure 2 As shown, the present invention also provides an online detection method for trace impurities in electronic-grade quartz crystal materials, applied to the online detection device for trace impurities in electronic-grade quartz crystal materials described in any one of the preceding descriptions, comprising: S100: Collect material data of multiple target materials and preprocess all collected material data to obtain multiple preprocessed data. S200, Create a detection model; S300 inputs multiple preprocessed data into the detection model, and the detection model creates a standard impurity spectrum database based on the preprocessed data of each target material. The impurity content in the target material is detected by combining a dynamic correlation matching algorithm with a signal enhancement factor calibration, and the trained detection model is obtained. The S400 collects real-time data of the material and inputs the real-time data into the trained detection model to detect the impurity content contained in the material.

[0034] It should be noted that by collecting material data from multiple target materials and preprocessing all collected material data to obtain multiple preprocessed data, a detection model is created. The multiple preprocessed data are then input into the detection model. The detection model creates a standard impurity spectral database based on the preprocessed data of each target material. Through dynamic correlation matching algorithm combined with signal enhancement factor calibration, the impurity content in the target material is detected, resulting in a trained detection model. Finally, real-time data of the material is collected and input into the trained detection model to detect the impurity content in the material. This application innovatively integrates biomimetic micro-nano enrichment technology with trace-cavity enhanced Raman technology, achieving a signal enhancement factor on the order of 10³ and a detection sensitivity on the order of ppt. It can accurately detect trace impurities and micro-elements in electronic-grade quartz crystal materials. At the same time, it adopts multispectral coupling and anti-interference design, with a detection error ≤ ±2%, excellent reproducibility, and improved detection sensitivity.

[0035] In an optional embodiment, S100 includes: S110, Create a materials database; S120: For each target material, set corresponding acquisition parameters, and acquire material data of multiple target materials based on the acquisition parameters. Put all the acquired material data into the material database. The acquisition parameters include clamping angle and clamping force. S130, randomly select the material data of a target material from the material database; S140, Determine whether there is duplicate data in the material data of the target material; S150, If there are duplicate data in the material data of the target material, delete the duplicate data; S160 returns the material data of a target material randomly selected from the material database, until all target materials have been selected, resulting in multiple preprocessed data.

[0036] It should be noted that the material database is stored using MySQL, and each data entry contains fields such as target material ID, clamping angle, clamping force, Raman spectrum, and acquisition time.

[0037] The criteria for identifying duplicate data are spectral similarity ≥ 99% and all samples in the database are traversed to ensure no omissions. By removing duplicate data, model overfitting is avoided and the model's generalization ability is improved.

[0038] In an optional embodiment, S300 includes: S310, divide all preprocessed information into training and test sets according to a random ratio; S320: The training set is input into the detection model. The detection model collects the illumination data of each target material based on multispectral excitation technology. Then, the dynamic correlation matching algorithm is used to calculate the types of impurities contained therein. The content of each type of impurity is calculated by combining the standard curve method and the internal standard method to obtain the trained detection model. S330: Input the test set into the trained detection model to verify whether the training of the detection model is complete.

[0039] It should be noted that the training and test sets are randomly divided in a 7:3 ratio. Random partitioning combined with stratified sampling avoids data bias. Multimodal data fusion improves the comprehensiveness of impurity identification. Multispectral excitation sequentially emits 266nm, 532nm, and 785nm lasers through the excitation module, simultaneously acquiring three-channel Raman spectra. The core formula of the dynamic correlation matching algorithm is:

[0040] Among them, S 实测 For the measured spectrum, S 标准 For standard spectra, a Corr ≥ 0.95 is considered a match. Model validation metrics include accuracy (≥ 98%), recall (≥ 97%), and F1 score (≥ 97.5%).

[0041] In an optional embodiment, S320 includes: S321, randomly select a target material from the training set; S322, Obtain the characteristic spectral signal obtained by multispectral excitation of the target material, and obtain the type of impurities contained in the target material through the characteristic spectral signal; S323 calculates the content of each impurity type using the signal enhancement factor; S324 returns the impurity type and content of each target material by randomly selecting one target material from the training set until all target materials in the training set have been selected.

[0042] It should be noted that this application adopts multi-threaded parallel processing, randomly selects target materials from the training set, calls the excitation module to obtain its three-channel characteristic spectrum, and quickly retrieves the standard library based on the hash table to identify the impurity type through the characteristic spectrum matching unit. For example, if a 680cm⁻¹ peak is detected, it is determined to be Fe³⁺.

[0043] The signal enhancement factor is used to calibrate content calculations. The signal enhancement factor is calculated using G=I. 原始 / I 增强 I represents the Raman peak intensity, calculated using the formula, while the impurity content = peak intensity × calibration coefficient / G. By iteratively processing all training samples, the impurity type and content label for each sample are finally generated for model supervised learning. The fluctuations of the resonant cavity / enrichment module are compensated by the signal enhancement factor to ensure the long-term stability of the content calculation.

[0044] In an optional embodiment, S400 includes: S410, collects real-time data of real-time materials; S420 inputs real-time data into the trained detection model, and obtains the types of impurities contained in the material in real time and the impurity content of each type of impurity through the trained detection model; S430 sets content thresholds for each type of impurity; S440, sequentially determine whether the impurity content of each type of impurity in the real-time material is greater than or equal to the content threshold. S450: If the content of at least one type of impurity is greater than or equal to the content threshold, an early warning will be issued.

[0045] It should be noted that the real-time data acquisition frequency is ≥1Hz, the trained detection model is deployed on an edge computing terminal, the inference time is ≤0.5 seconds, and the content threshold is set according to the industry standard for electronic grade quartz crystals.

[0046] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited thereto. Various changes can be made within the scope of knowledge possessed by those skilled in the art without departing from the spirit of the present invention.

Claims

1. An online detection device for trace impurities in electronic-grade quartz crystal materials, characterized in that, include: A conveying module; the conveying module conveys the target material, and the conveying module includes a conveying track, an adaptive clamping mechanism, and a laser positioning unit; Enrichment module; The enrichment module enables rapid enrichment of trace impurities and micro-elements on the surface of the target material. The enrichment module includes a copper substrate and a silver nanowire heating unit. Excitation module; the excitation module provides multi-band stable laser light to adapt to the characteristic spectral response of different trace impurities and micro elements. The excitation module includes a laser source group, an optical path adjustment unit and a wavelength switching unit. Acquisition module; The acquisition module enables efficient enhancement and accurate acquisition of trace impurity Raman signals. The acquisition module includes a resonant cavity assembly, a signal filtering unit, and a high-sensitivity detector. The detection module is communicatively connected to the acquisition module. Through the detection module, real-time processing, qualitative and quantitative analysis, and result output of the detection data are realized. The detection module includes a data preprocessing unit, a characteristic spectrum matching unit, a quantitative calculation unit, and a result output unit.

2. The online detection device for trace impurities in electronic-grade quartz crystal materials according to claim 1, characterized in that, The adaptive clamping mechanism is located on both sides of the conveying track. The adaptive clamping mechanism is used to fix different types of target materials. The positioning accuracy of the laser positioning unit is set to ±0.01mm.

3. The online detection device for trace impurities in electronic-grade quartz crystal materials according to claim 2, characterized in that, The surface of the copper substrate is etched into a fish-scale-like micro / nano structure, and the silver nanowire heating unit is integrated inside the copper substrate.

4. The online detection device for trace impurities in electronic-grade quartz crystal materials according to claim 3, characterized in that, The laser source group includes a multi-band laser generator, and the optical path adjustment unit includes a high-precision reflector and lens group to achieve collimation, focusing and calibration of the laser beam. The focused spot size is adjustable, and the switching response time of the wavelength switching unit is ≤1 second.

5. The online detection device for trace impurities in electronic-grade quartz crystal materials according to claim 4, characterized in that, The resonant cavity assembly is configured with a double-waisted folded resonant cavity design, so that the laser is reflected multiple times within the cavity. The high-sensitivity detector has a detection wavelength range of 200-4200cm⁻¹ and a resolution of ≤5cm⁻¹@585nm.

6. A method for online detection of trace impurities in electronic-grade quartz crystal materials, applied to the online detection device for trace impurities in electronic-grade quartz crystal materials according to any one of claims 1 to 5, characterized in that, include: Material data of multiple target materials are collected, and all collected material data are preprocessed to obtain multiple preprocessed data. Create a detection model; Multiple preprocessed data are input into the detection model. The detection model creates a standard impurity spectral database based on the preprocessed data of each target material. The impurity content in the target material is detected by combining a dynamic correlation matching algorithm with a signal enhancement factor calibration, and the trained detection model is obtained. Real-time data of the material is collected and input into the trained detection model to detect the impurity content in the material.

7. The method for online detection of trace impurities in electronic-grade quartz crystal materials according to claim 6, characterized in that, The process involves collecting material data from multiple target materials and preprocessing all collected material data to obtain multiple preprocessed data sets, including: Create a materials database; For each target material, corresponding acquisition parameters are set, and material data of multiple target materials are collected based on the acquisition parameters. All collected material data are put into the material database. The acquisition parameters include clamping angle and clamping force. Randomly select the material data of a target material from the material database; Determine if there are duplicate data in the material data of the target material; If duplicate data exists in the material data of the target material, delete the duplicate data; Returns the material data of a target material randomly selected from the material database, until all target materials have been selected, resulting in multiple preprocessed data sets.

8. The method for online detection of trace impurities in electronic-grade quartz crystal materials according to claim 7, characterized in that, The process involves inputting multiple preprocessed data into a detection model, creating a standard impurity spectral database based on the preprocessed data for each target material, and then using a dynamic correlation matching algorithm combined with signal enhancement factor calibration to detect the impurity content in the target material, resulting in a trained detection model. This includes: All preprocessed information is divided into training and test sets according to a random ratio; The training set is input into the detection model. The detection model collects the illumination data of each target material based on multispectral excitation technology. Then, the types of impurities are calculated by dynamic correlation matching algorithm, and the content of each type of impurity is calculated by standard curve method and internal standard method to obtain the trained detection model. Input the test set into the trained detection model to verify whether the training of the detection model is complete.

9. The method for online detection of trace impurities in electronic-grade quartz crystal materials according to claim 8, characterized in that, The training set is input into the detection model, which then collects illumination data for each target material using multispectral excitation technology. The dynamic correlation matching algorithm is then used to calculate the types of impurities present, and the content of each type of impurity is calculated using the standard curve method and the internal standard method, resulting in the trained detection model. This includes: Randomly select a target material from the training set; The characteristic spectral signal obtained by multispectral excitation of the target material is acquired, and the type of impurities contained in the target material is obtained through the characteristic spectral signal; The content of each impurity type is calculated using the signal enhancement factor; Return to the point where a target material is randomly selected from the training set, and continue until all target materials in the training set have been selected, to obtain the type and content of impurities contained in each target material.

10. The method for online detection of trace impurities in electronic-grade quartz crystal materials according to claim 9, characterized in that, The process of collecting real-time data from the material and inputting this data into a trained detection model to detect the impurity content in the material includes: Collect real-time data from real-time materials; Real-time data is input into the trained detection model, and the types of impurities contained in the material and the impurity content of each type of impurity are obtained through the trained detection model. Set content thresholds for each type of impurity; Sequentially determine whether the impurity content of each type of impurity in the real-time material is greater than or equal to the content threshold; If the content of at least one type of impurity is greater than or equal to the content threshold, an early warning will be issued.

Images