Quantitative detection system and method for defects on quartz optical surface based on schlieren method

Abstract

The application provides a kind of quartz optical surface defect quantitative detection system and method based on schlieren method, belongs to quartz optical element detection technical field.The application adopts 450nm blue light source, Z-shaped light path, adjustable polarization system and electric control knife edge, combines adaptive image processing and multi-modal fusion algorithm, realizes the automatic identification, positioning and depth quantitative detection of quartz surface scratch, ripple, micro crack and pollution.The system can detect 100nm depth scratch and 50nm amplitude ripple, the positioning accuracy is less than or equal to 0.5μm, the depth measurement error is less than 5%, and the detection speed reaches 10fps.The application is non-contact, full-field coverage, small and portable, high precision and fast, effectively solves the problems of low efficiency, incomplete coverage and heavy equipment of traditional detection methods, and is suitable for online quality detection and high-precision screening of quartz elements in semiconductor, aerospace, optical manufacturing and other fields.

Description

Technical Field

[0001] This invention belongs to the field of quartz optical element inspection technology, specifically relating to a quantitative inspection system and method for quartz optical surface defects based on schlieren method. It is suitable for high-precision and high-efficiency quantitative screening, location determination and depth measurement of defects such as scratches, ripples, microcracks and contamination on the surface of quartz optical elements. It can be widely used in strategic fields that rely on high-precision quartz optical elements, such as semiconductors, aerospace, quantum computing, and optical manufacturing. Background Technology

[0002] As the core component of high-precision optical systems, the surface quality of quartz optical elements directly determines the imaging accuracy, light transmission performance, and service life of the optical system. Especially in high-end fields such as semiconductor lithography, aerospace remote sensing, and quantum computing, extremely high requirements are placed on the smoothness, surface accuracy, and defect control of quartz surfaces, requiring the accurate detection and quantitative analysis of nanoscale defects.

[0003] Currently, traditional methods for detecting defects on quartz optical surfaces mainly include white light interferometry, atomic force microscopy (AFM), ultrasonic testing, and laser scanning confocal microscopy. Each method has significant limitations: while white light interferometry and atomic force microscopy can achieve nanometer-level detection accuracy, their detection speed is slow and their detection range is limited. The point scanning mode cannot achieve real-time full-field coverage, making it difficult to meet the needs of large-scale, rapid quality control in industrial production lines. Ultrasonic testing can detect subsurface defects, but its spatial resolution is low, its ability to identify nanometer-level defects is insufficient, and it cannot achieve quantitative characterization of defects. Laser scanning confocal microscopy has high detection accuracy, but its equipment cost is high, its operation is complex, and its detection efficiency is low, which is not conducive to widespread industrial application.

[0004] Schlieren measurement technology is a non-contact optical inspection method based on light refraction and diffraction. Its core principle is to convert the density gradient or refractive index change of a medium into observable light intensity contrast. It boasts unique advantages such as non-contact operation, high sensitivity, fast detection speed, and full-field coverage, and has been applied in fields such as fluid mechanics and optical inspection. However, existing schlieren systems are mostly large-scale commercial devices with long optical path distances, bulky size, and poor portability. Furthermore, they are not optimized for the negative uniaxial birefringence characteristics of quartz optical elements, failing to fully capture the optical response differences caused by quartz surface defects (especially microcracks and stress defects). This makes it difficult to achieve accurate quantitative detection of quartz surface defects, particularly in the effective identification and deep quantification of nanoscale defects such as scratches and microcracks, and also fails to meet the automation and miniaturization requirements of industrial production lines.

[0005] Meanwhile, in the development of existing large-aperture optical systems, the manufacturing of ultra-smooth surfaces and high-precision surface shape control are core challenges. Quartz optical components must meet the "three lows and one high" requirements of "low roughness, low defects, high-precision surface shape, and low subsurface damage." Traditional inspection methods struggle to simultaneously address the needs for inspection accuracy, efficiency, and full-field coverage, failing to provide real-time and comprehensive data support for optimizing quartz component processing. Therefore, developing a high-precision, high-efficiency, and miniaturized quantitative inspection system and method specifically designed for quartz optical surface defects, and addressing the shortcomings of traditional inspection methods, has become an urgent technical challenge in the field of quartz optical component quality control. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of existing quartz optical surface defect detection methods, such as slow detection speed, insufficient full-field coverage, low quantitative accuracy, large equipment size, and poor adaptability. This invention provides a quantitative detection system and method for quartz optical surface defects based on schlieren. The optical path and algorithm are optimized for the birefringence characteristics of quartz to achieve high-sensitivity, non-contact, full-field coverage quantitative detection of quartz surface defects, balancing detection accuracy and efficiency, and meeting the needs of automated quality control in industrial production lines and high-precision detection in high-end fields.

[0007] To achieve the above objectives, the present invention adopts the following technical solution:

[0008] A quantitative detection system for quartz optical surface defects based on schlieren method, comprising:

[0009] The light source module provides uniform illumination; the optical path module, employing a Z-shaped optical path configuration (consisting of two concave mirrors with equal focal lengths, the optical path folds in a Z-shape to achieve parallel light collimation, sample transmission, and beam convergence), enhances the optical contrast of birefringence defects in quartz materials; the knife-edge module converts light deflection caused by defects into light intensity contrast; the imaging module acquires schlieren images; the control and processing module integrates light source intensity adjustment, knife-edge position control, and image processing algorithms to achieve automated defect screening and quantitative analysis; and the high-precision positioning module, on which the sample is mounted via a three-dimensional motion platform, enables sub-micron-level positioning and full-field scanning coverage of the sample.

[0010] The uniform parallel light output from the light source module is sequentially incident on the λ / 4 waveplate and H polarizer in the optical path module. It is transmitted through a concave mirror and passes through the quartz sample under test on the high-precision positioning module. The beam carrying defect information is transmitted through another concave mirror, sequentially passing through the V polarizer and λ / 4 waveplate before being incident on the knife edge module. The light signal modulated by the knife edge is acquired by the imaging module and converted into a digital image signal, which is then transmitted to the control and processing module. At the same time, the control and processing module is electrically connected to the light source module, knife edge module, and high-precision positioning module to realize closed-loop control of light intensity, knife edge position, sample scanning and positioning, and complete the detection, analysis and result output of quartz surface defects.

[0011] A quantitative detection method for quartz optical surface defects based on schlieren analysis, using the aforementioned detection system, includes the following steps:

[0012] Step 1, System Calibration: Prepare standard quartz samples containing stepped scratches, simulated ripples, and subsurface defects. Measure the same defect area using schlieren method, atomic force microscopy, and white light interferometer, respectively. Establish a quantitative relationship between light intensity contrast and defect depth and ripple amplitude, and determine the system calibration coefficient k.

[0013] Step 2, Sample Placement and Optical Path Adjustment: Fix the quartz sample to be tested on the T-slot platform, and adjust the concave reflector, polarization system and knife edge position of the Z-shaped optical path to ensure that the parallel beam passes perpendicularly through the sample surface.

[0014] Step 3, Image Acquisition: Start the light source module and imaging module to acquire the schlieren image of the quartz sample to be tested, and simultaneously acquire four sets of orthogonal polarization images for polarization difference enhancement processing;

[0015] Step 4, Image Preprocessing: The acquired schlieren image is sequentially processed with illumination compensation, polarization difference enhancement, background suppression, and texture removal to eliminate the effects of uneven illumination and background noise.

[0016] Step 5, Quantitative Defect Screening and Location: Analyze the preprocessed image, identify the defect type and calculate the light intensity contrast of the defect area, and obtain the defect depth by combining the system calibration coefficient; determine the specific location of the defect through the high-precision positioning module to achieve quantitative screening and location of defects throughout the field;

[0017] Step 6, Error Compensation and Result Output: Through multi-frame averaging, light intensity normalization, and real-time optical path calibration, the detection errors caused by light source fluctuations, optical path offsets, and environmental vibrations are compensated, and the defect type, depth, location, and detection data are finally output.

[0018] The beneficial effects of this invention are as follows:

[0019] 1. High detection accuracy: This invention optimizes the polarization system for the birefringence characteristics of quartz, and combines a 450nm blue light source and an electrically adjustable blade to detect surface scratches with a depth as low as 100nm, surface ripples with an amplitude as low as 50nm, and subsurface defects with a depth of up to 0.5μm. The defect positioning accuracy is ≤0.5μm, and the depth measurement error is <5%. The results are highly consistent with those of atomic force microscopy and white light interferometer, meeting the "three lows and one high" detection requirements for quartz optical components.

[0020] 2. High detection efficiency: Adopting a full-field scanning mode, the detection speed can reach more than 10fps, which is 20 times that of a white light interferometer and 100 times that of an atomic force microscope. It can realize rapid full-field coverage detection of quartz optical components, avoid the omission risk of traditional point scanning methods, and is suitable for real-time quality monitoring of industrial production lines, which can significantly improve detection efficiency and reduce production costs.

[0021] 3. Non-contact testing: No physical contact with the sample surface is required, avoiding secondary damage to the quartz optical components during the testing process. It is especially suitable for testing high-precision, easily damaged quartz components, protecting the integrity of the sample.

[0022] 4. High degree of automation: It integrates automatic control and image processing algorithms, which can realize full-process automation from optical path debugging, image acquisition, defect identification to result output, reduce manual operation costs, reduce human error, and improve the repeatability and stability of detection results.

[0023] 5. Miniaturized and highly portable system: Adopting a Z-shaped optical path design, the optical path length is only about 300mm, which significantly reduces the size compared to traditional large schlieren systems. The compact structure can be flexibly adapted to different detection scenarios, making it easy to integrate into industrial production lines and conduct on-site inspections, thus solving the problems of traditional schlieren equipment being bulky and inconvenient to move.

[0024] 6. Wide range of applications: It can quantitatively detect various quartz surface defects such as scratches, ripples, microcracks, and contamination. It can also be adapted to quartz optical components of different sizes and thicknesses. Its application scenarios cover multiple fields such as semiconductors, aerospace, optical manufacturing, and quantum computing, and it can meet the testing needs of quartz components in large-aperture optical systems. Attached Figure Description

[0025] Figure 1 This is a schematic diagram of a quantitative detection system for quartz optical surface defects based on schlieren method;

[0026] Figure 2 This is a flowchart of a quantitative detection method for quartz optical surface defects based on schlieren method. Detailed Implementation

[0027] The present invention will be further described in detail below with reference to specific embodiments to enable those skilled in the art to understand it. It should be noted that these embodiments are only for explaining the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made without departing from the principles of the present invention should be included within the scope of protection of the present invention.

[0028] Example 1: Quantitative Detection System for Quartz Optical Surface Defects Based on Schlieren Method

[0029] like Figure 1 As shown, this embodiment provides a quantitative detection system for quartz optical surface defects based on schlieren method, including:

[0030] The system comprises the following modules: a light source module to provide uniform illumination; an optical path module with a Z-shaped configuration to enhance the optical contrast of birefringence defects in quartz materials; a knife-edge module to convert light deflection caused by defects into light intensity contrast; an imaging module to acquire schlieren images; a control and processing module integrating light source intensity adjustment, knife-edge position control, and image processing algorithms for automated defect screening and quantitative analysis; and a high-precision positioning module, on which the sample is mounted via a three-dimensional motion platform to achieve sub-micron-level positioning and full-field scanning coverage.

[0031] The light source module employs a 450nm blue LED array (10W power) paired with a Thorlabs AC127-045-A-ML aspherical homogenizing lens to achieve uniform illumination. Testing showed a light intensity non-uniformity of 4.2%, meeting the requirement of ≤5%, ensuring illumination uniformity and reducing false defect signals. It provides a stable and uniform monochromatic parallel light input for the entire schlieren system, serving as the illumination source for subsequent defect imaging and representing the system's optical signal starting point.

[0032] The optical path module adopts a Z-shaped optical path configuration, including two concave mirrors with a focal length of 300mm and an adjustable polarization system. Both concave mirrors have a focal length of 300mm. The adjustable polarization system consists of a λ / 4 waveplate (with a fast axis adjustable in 0°-90° steps with an adjustment accuracy of 1°) and an H-polarizer, used to enhance the contrast of defects under anisotropic material conditions. The λ / 4 waveplate and H-polarizer are installed sequentially along the light source emission direction, and a V-polarizer and adjustable λ / 4 waveplate are sequentially arranged along the imaging path. The fast axis of the λ / 4 waveplate forms a 45° angle with the quartz optical axis, which can improve the contrast of birefringent defects by 18%, optimizing the defect identification effect. The concave mirrors receive the emitted light from the light source module, shaping it into a parallel beam and introducing it into the quartz sample under test. Through the polarization system and the concave mirrors, the light deflection and polarization state changes caused by sample defects are modulated and converged, providing a beam carrying defect information to the knife-edge module.

[0033] The cutting edge module uses an electrically adjustable cutting edge, with the cutting edge width adjustable within the range of 1-5μm. It is equipped with a micron-level displacement platform, achieving a cutting edge position adjustment accuracy of 0.1μm. The settings are optimized based on the Fresnel diffraction formula, which is: Where λ is the wavelength of light. and The lengths of the two segments of the Z-shaped optical path are given, and d is the depth of the defect on the quartz surface. Let n be the refractive index difference between the defective and normal regions, and n be the refractive index of quartz (n≈1.544 in the visible light band). This ensures that the light refraction caused by tiny defects can be effectively converted into light intensity contrast. According to the Fresnel diffraction formula, for a scratch with a depth of 100 nm... , , ), position of the blade When the aperture is set to 2μm, the light refraction caused by the scratch can be effectively converted into light intensity contrast, achieving a detection rate of over 95%. The specific conversion capability can be quantitatively expressed as follows:

[0034] ;

[0035] in, For changes in light intensity, For phase gradient, For the change in refractive index, Geometric thickness, Background light intensity, These are the system calibration coefficients. The wavelength of light. This is the phase. This part is used to partially block the converging beam emitted from the optical path module, converting the minute light deflection into a detectable difference in light intensity contrast. It is the core modulation unit for achieving defect visualization using schlieren.

[0036] The imaging module employs a high-resolution or high-speed camera with a resolution ≥1080×824 pixels, a dynamic range ≥120dB, a pixel size ≤4.65×4.65μm, and a frame rate ≥10fps. This allows for clear acquisition of defect images, meeting the needs of various inspection scenarios (static inspection, dynamic monitoring). The preferred camera is the Baslerace acA4000-22gc high-resolution camera or the Phantom VEO710 high-speed camera, balancing imaging accuracy and inspection efficiency. More specifically, the imaging module uses a Baslerace acA4000-22gc high-resolution camera with a resolution of 4000×3000 pixels, a pixel size of 4.65×4.65μm, a dynamic range of 120dB, and a frame rate ≥10fps. This allows for clear acquisition of images of quartz surface scratches up to 100nm deep, meeting high-precision inspection requirements. It receives the light signal modulated by the blade module, converting light intensity and contrast information into digital image data, providing the raw inspection image to the control and processing module.

[0037] The control and processing module employs an industrial computer, integrating light source intensity adjustment, blade position control, and image processing algorithms to achieve automated defect screening and quantitative analysis. The image processing algorithms include image preprocessing, defect quantitative screening, and defect depth quantitative algorithms. The image preprocessing algorithms include Adaptive Histogram Equalization (CLAHE) illumination compensation, polarization difference enhancement, Extremely Fast Empirical Mode Decomposition (EFEMD) background suppression, and Relative Total Variation (RTV) texture removal to eliminate interference from uneven illumination, background noise, and quartz surface grinding textures. The defect quantitative screening algorithm uses a multimodal fusion algorithm combining an improved U-Net with traditional feature extraction to accurately classify four types of defects: scratches, ripples, microcracks, and contamination. The defect depth quantitative algorithm is based on the linear relationship between light intensity contrast and defect depth. Where C is the light intensity contrast ratio. , I represents the light intensity in the normal region, k represents the light intensity in the defect region, and d represents the system calibration coefficient. More specifically, the control and processing module uses an industrial computer (CPU: Intel Core i7-12700H, 32GB memory), integrating a light source intensity adjustment module, a knife edge position control module, and dedicated image processing software; the image processing algorithms include adaptive histogram equalization (CLAHE), illumination compensation (CLAHE parameters: block size 8×8, contrast limit 0.02), polarization difference enhancement, extreme rapid empirical mode decomposition (EFEMD) background suppression, and relative total variation (RTV) texture removal (regularization parameter ε=0.001).

[0038] An improved U-Net multimodal fusion defect classification algorithm (loss functions: cross-entropy loss and Dice loss, weights 0.7:0.3) is applied to the preprocessed image. The specific calculation of the cross-entropy loss (CE) is as follows:

[0039] ;

[0040] in, For batch size, For real labels, To predict probabilities, the Dice loss is calculated using the following formula:

[0041] ;

[0042] in, Let be the smoothing factor. In summary, the total loss function is: Finally, a defect depth quantitative algorithm is used to invert the defect depth. First, a correlation is established between the grayscale changes in the schlieren image and the refractive index, which approximates a one-dimensional integral relationship, specifically expressed as:

[0043] ;

[0044] This is the phase change. This is the position of the maximum score. The position of the lower limit of integration. For integration variables, This represents the distribution of changes in light intensity. This is the system sensitivity factor. Subsequently, the refractive index n of the quartz substrate and the refractive index of the defect region are known. Determine the depth of defects The final defect classification accuracy reached 99.2%. This module is used for closed-loop control of the light source module, blade module, and high-precision positioning module to realize the automatic adjustment of the optical path and platform; on the other hand, it is used to process the image data acquired by the imaging module, identify and quantitatively analyze defects, and output the final detection results.

[0045] The high-precision positioning module uses a high-precision T-slot platform (flatness ≤0.01mm / m, reflectivity ≤5%) as a reference. The sample is placed on it, and standard crosshair positioning marks are affixed to the sample surface. Combined with a sub-pixel edge detection algorithm, sub-micron level defect positioning is achieved with a positioning accuracy ≤0.5μm. Simultaneously, a high-precision 3D motion platform (accuracy 0.1μm) is equipped, employing a line-by-line scanning strategy to achieve full-field coverage. The scanning step size is 80% of the image height, and the image stitching accuracy is ≤0.5 pixels, avoiding defect omissions. The algorithm includes ROI extraction, coarse edge detection (Sobel, Canny, etc.), grayscale curve generation, and Gaussian fitting. The specific form of Gaussian fitting is as follows:

[0046] ;

[0047] For the fitted function value, Spatial location coordinates, For amplitude, The mean, Standard deviation, This represents the baseline offset. The sub-pixel edge positions are obtained by fitting the above sampling points using the least squares method.

[0048] The 3D motion platform (0.1μm accuracy) employs a line-by-line scanning strategy with a scanning step size of 80% of the image height and an image stitching accuracy of ≤0.5 pixels, enabling full-field coverage inspection of a 100×100mm² quartz sample. It supports the quartz sample under test, achieving precise positioning and full-field scanning of the sample on the optical path. Working in conjunction with the control and processing module, it completes high-precision calibration of defect locations, ensuring the traceability of test results.

[0049] In summary, the uniform parallel light output from the light source module is sequentially incident on the λ / 4 waveplate and H polarizer in the optical path module. It is then transmitted through a concave mirror of a Z-shaped structure and passes through the quartz sample to be tested on the high-precision positioning module. The beam carrying defect information is transmitted through a concave mirror, through the V polarizer and adjustable λ / 4 waveplate of the optical path module, and then incident on the knife edge module. The light signal modulated by the knife edge is acquired by the imaging module and converted into a digital image signal, which is then transmitted to the control and processing module. At the same time, the control and processing module is electrically connected to the light source module, knife edge module, and high-precision positioning module to realize closed-loop control of light intensity, knife edge position, sample scanning and positioning, and complete the detection, analysis and result output of quartz surface defects.

[0050] like Figure 2 As shown, the present invention also provides a quantitative detection method for quartz optical surface defects based on schlieren method, which, using the above-mentioned detection system, includes the following steps:

[0051] Step 1, System Calibration: Prepare standard quartz samples containing stepped scratches, simulated ripples, and subsurface defects. Measure the same defect area using schlieren microscopy, atomic force microscopy (AFM), and white light interferometer, respectively. Establish a quantitative relationship between light intensity contrast and defect depth and ripple amplitude, and determine the system calibration coefficient k. The calibration experimental environment is strictly controlled at temperature 25±0.1℃, humidity 40%±1%RH, and vibration ≤0.01μm / rms to ensure the accuracy and repeatability of the calibration results. The calibration qualification standard is a correlation coefficient R² > 0.95 between the schlieren microscopy and AFM / WLI measurement results, and a root mean square error RMSE < 5%.

[0052] The standard samples were prepared as follows: stepped scratch samples were precisely etched using focused ion beam (FIB) or atomic force microscopy (AFM), with a depth gradient of 100 nm to 5 μm, uniformly distributed on the sample surface, and a spacing of ≥ 5 mm; simulated ripple samples were prepared using photolithography or precision machining, with an amplitude gradient of 50 nm to 500 nm and a period gradient of 1 μm to 10 μm, and at least 3 samples were set for each amplitude-period combination; subsurface defect samples were prepared using a combination of magnetorheological polishing and CCOS small grinding head technology, with a depth gradient of 0.5 μm to 5 μm, uniformly distributed on the sample surface, and a spacing of ≥ 5 mm.

[0053] Step 2, Sample Placement and Optical Path Adjustment: Fix the quartz sample to be tested on a high-precision T-slot platform, adjust the concave reflector, polarization system and knife edge position of the Z-shaped optical path to ensure that the parallel beam passes perpendicularly through the sample surface, and optimize the knife edge position according to the Fresnel diffraction formula based on the defect detection requirements; adjust the fast axis angle of the λ / 4 waveplate to enhance the contrast of birefringent defects and ensure optical path stability and clear imaging.

[0054] Step 3, Image Acquisition: Start the light source module and imaging module to acquire the schlieren image of the quartz sample to be tested, and simultaneously acquire four sets of orthogonal polarization images (0°, 45°, 90°, 135°) for subsequent polarization differential enhancement processing; during the acquisition process, control the light source intensity to keep it stable and avoid light intensity fluctuations affecting the detection results.

[0055] Step 4, Image Preprocessing: The acquired schlieren images are sequentially processed with illumination compensation, polarization difference enhancement, background suppression, and texture removal to eliminate the effects of uneven illumination, background noise, and quartz surface grinding texture, thereby improving the recognition of defect areas. Among them, polarization difference enhancement calculates the gradient magnitude difference map between adjacent angular polarization images and selects the channel with the largest difference as the main image for defect detection, further enhancing the contrast between defects and the background.

[0056] Step 5, Quantitative Defect Screening and Location: The preprocessed image is analyzed using a multimodal fusion defect classification algorithm to identify the defect type and calculate the light intensity contrast of the defect area. The defect depth is obtained by combining the system calibration coefficient k. The specific location of the defect is determined by a high-precision positioning module, realizing quantitative screening and location of defects throughout the field.

[0057] The criteria for classifying defects are as follows: scratches correspond to light intensity contrast C < 0.2 and aspect ratio > 5:1; ripples correspond to light intensity contrast C < 0.4 and aspect ratio < 2:1; microcracks correspond to light intensity contrast C < 0.3 and shape factor S < 0.8; and contamination corresponds to light intensity contrast C > 0.4 and random location.

[0058] Step 6, Error Compensation and Result Output: Through multi-frame averaging, light intensity normalization, and real-time optical path calibration, the detection errors caused by light source fluctuations, optical path offsets, tool edge position drift, and environmental vibrations are compensated. The final output includes defect type, depth, location, and detection accuracy data. Specifically, the error compensation methods are as follows: multi-frame averaging and light intensity normalization are used to compensate for light source fluctuation errors, with the error controlled within <1%; real-time optical path calibration and adaptive background suppression algorithms are used to compensate for optical path offset errors, with the error controlled within <0.5%; electrically adjustable tool edges and real-time position monitoring are used to compensate for tool edge position drift errors, with the error controlled within <0.2μm; and a high-precision vibration isolation platform and adaptive image fusion algorithm are used to compensate for environmental vibration errors, with the error controlled within <0.1μm.

[0059] Example 2: Implementation of a quantitative detection method for quartz optical surface defects based on schlieren method

[0060] This embodiment uses the detection system built in Example 1 to quantitatively detect surface defects in a 100×100×1mm high-purity quartz wafer (purity ≥99.99%). The specific steps are as follows:

[0061] Step 1: System Calibration

[0062] Standard quartz samples were prepared, including stepped scratch samples, simulated ripple samples, and subsurface defect samples. The stepped scratch samples were prepared using focused ion beam etching with depth gradients of 100 nm, 200 nm, 300 nm, 400 nm, 500 nm, 1 μm, 2 μm, 3 μm, 4 μm, and 5 μm, uniformly distributed on the sample surface with a spacing of 5 mm. The simulated ripple samples were prepared using photolithography with amplitude gradients of 50 nm, 100 nm, 200 nm, 300 nm, 400 nm, and 500 nm, and period gradients of 1 μm, 2 μm, 3 μm, 4 μm, 5 μm, and 10 μm, with three samples for each amplitude-period combination. The subsurface defect samples were prepared using a combination of magnetorheological polishing and CCOS small grinding head technology, with depth gradients of 0.5 μm, 1 μm, 2 μm, 3 μm, 4 μm, and 5 μm, uniformly distributed on the sample surface with a spacing of 5 mm.

[0063] The standard sample was placed on the T-slot platform of the detection system. The experimental environment was controlled at a temperature of 25±0.1℃, humidity of 40%±1%RH, and vibration ≤0.01μm / rms. The same defect area was measured using schlieren imaging, atomic force microscopy (AFM), and white light interferometry. The light intensity contrast of the defect area in the schlieren image was calculated as C=(I0-I) / I0. The system calibration coefficient k=0.0009 and the correlation coefficient R were obtained by least squares fitting. 2 =0.97, RMSE=3.2%, which meets the calibration qualification standard.

[0064] Step 2, Sample Placement and Optical Path Adjustment: Fix the quartz sample to be tested on the high-precision T-slot platform, adjust the concave reflector of the Z-shaped optical path to ensure that the parallel beam passes perpendicularly through the sample surface; adjust the fast axis of the λ / 4 waveplate to form a 45° angle with the quartz optical axis to enhance the contrast of birefringence defects; according to the type of defect to be tested, set the blade width to 3μm and the blade position Δx to 2μm to ensure optical path stability and clear imaging.

[0065] Step 3, Image Acquisition: Start the light source module and imaging module to acquire the schlieren image of the quartz sample to be tested. Set the acquisition frame rate to 10fps and simultaneously acquire four sets of orthogonal polarization images (0°, 45°, 90°, 135°) for polarization difference enhancement processing. Keep the light source intensity stable during the acquisition process to avoid light intensity fluctuations affecting the detection results.

[0066] Step 4, Image Preprocessing: Adaptive histogram equalization illumination compensation is sequentially applied to the acquired schlieren images to eliminate the influence of uneven illumination; through polarization difference enhancement, gradient magnitude difference maps of four sets of orthogonal polarization images are calculated, and the channel with the largest difference is selected as the main image for defect detection; the Extremely Fast Empirical Mode Decomposition (EFEMD) algorithm is used to suppress the inherent scattering noise of the quartz substrate, and the Relative Total Variation (RTV) texture removal algorithm is used to remove the grinding texture on the quartz surface, thereby improving the recognition of defect areas.

[0067] Step 5, Quantitative Defect Screening and Localization: The improved U-Net multimodal fusion defect classification algorithm was used to analyze the preprocessed image, identifying two scratches and one microcrack on the sample surface. The light intensity contrast of the scratch area was calculated to be 0.18 and 0.16, respectively. Combined with the calibration coefficient k=0.0009, the scratch depths were found to be 200nm and 178nm, respectively. The light intensity contrast of the microcrack area was calculated to be 0.27, and the shape factor S=0.72, resulting in a microcrack depth of 300nm. Through a high-precision positioning module, the location coordinates of the two scratches were determined to be (25.32mm, 48.15mm) and (67.89mm, 72.46mm), respectively, and the location coordinates of the microcrack were determined to be (49.63mm, 31.78mm), with a positioning accuracy of ≤0.5μm for all.

[0068] Step 6, Error Compensation and Result Output: Multi-frame averaging (averaging 10 frames of images) and light intensity normalization are used to compensate for light source fluctuation errors, with the error controlled within 0.8%; real-time optical path calibration and adaptive background suppression algorithms are used to compensate for optical path offset errors, with the error controlled within 0.4%; real-time monitoring of the electrically adjustable tool edge position is used to compensate for tool edge position drift errors, with the error controlled within 0.15μm; a high-precision vibration isolation platform is used to compensate for environmental vibration errors, with the error controlled within 0.08μm; finally, a defect detection report is output, including defect type, depth, location, and detection accuracy data. The error between the detection results and the AFM measurement results is 2.8%, meeting the high-precision detection requirements of industry.

[0069] Example 3: Comparative Experiment of Different Detection Methods

[0070] The detection system of this invention was compared with traditional detection methods (white light interferometer, AFM, and ultrasonic testing) to detect and compare the same batch of 100×100×1mm high-purity quartz wafers (containing scratches with a depth of 100nm-5μm, amplitude ripples of 50nm-500nm, and subsurface defects with a depth of 0.5μm-5μm). The experimental environment was the same as in Example 2. The detection results are shown in the table below:

[0071]

[0072] As can be seen from the above comparative experiments, the detection accuracy of the detection system of the present invention is close to that of white light interferometer and AFM, the detection speed is far superior to traditional detection methods, the lower limit of detection depth can meet the high-precision detection requirements of industry, and at the same time, it takes into account both detection accuracy and efficiency. Compared with traditional detection methods, it is more suitable for large-scale and rapid quality control in industrial production lines.

Claims

1. A quantitative detection system for quartz optical surface defects based on schlieren method, characterized in that, include: The light source module is used to provide uniform illumination. The optical path module, employing a Z-shaped optical path configuration, enhances the optical contrast of birefringence defects in quartz materials; the knife-edge module converts light deflection caused by defects into light intensity contrast; the imaging module acquires schlieren images; the control and processing module integrates light source intensity adjustment, knife-edge position control, and image processing algorithms to achieve automated defect screening and quantitative analysis; and the high-precision positioning module, on which the sample is mounted via a three-dimensional motion platform, enables sub-micron-level positioning and full-field scanning coverage of the sample. The uniform parallel light output from the light source module is sequentially incident on the λ / 4 waveplate and H polarizer in the optical path module. It is transmitted through a concave mirror and passes through the quartz sample to be tested on the high-precision positioning module. The beam carrying defect information is transmitted through another concave mirror, through the V polarizer and λ / 4 waveplate, and then incident on the knife edge module. The light signal modulated by the knife edge is acquired by the imaging module and converted into a digital image signal, which is then transmitted to the control and processing module. At the same time, the control and processing module is electrically connected to the light source module, knife edge module, and high-precision positioning module to realize closed-loop control of light intensity, knife edge position, sample scanning and positioning, and complete the detection, analysis and result output of quartz surface defects.

2. The quantitative detection system for quartz optical surface defects based on schlieren method according to claim 1, characterized in that, The light source module uses a 450nm blue LED array.

3. The quantitative detection system for quartz optical surface defects based on schlieren method according to claim 1, characterized in that, The polarization system in the optical path module includes a light source-side polarization component and an imaging-side polarization component; the light source-side polarization component consists of a λ / 4 waveplate and an H polarizer along the beam direction; the imaging-side polarization component includes a V polarizer and an adjustable λ / 4 waveplate.

4. The quantitative detection system for quartz optical surface defects based on schlieren method according to claim 1, characterized in that, The blade width and position of the blade module can be electrically adjusted to convert minute light deflections into detectable differences in light intensity contrast.

5. The quantitative detection system for quartz optical surface defects based on schlieren method according to claim 1, characterized in that, The imaging module uses a high-resolution or high-speed camera to acquire images.

6. The quantitative detection system for quartz optical surface defects based on schlieren method according to claim 1, characterized in that, The control and processing module uses an industrial computer and integrates light source intensity adjustment, blade position control, and image processing algorithms to achieve automated defect screening and quantitative analysis. The image processing algorithms include image preprocessing algorithms, defect quantitative screening algorithms, and defect depth quantitative algorithms. The image preprocessing algorithms include adaptive histogram equalization illumination compensation, polarization difference enhancement, rapid empirical mode decomposition background suppression, and relative total variation texture removal to eliminate interference from uneven illumination, background noise, and quartz surface grinding textures. The defect quantitative screening algorithm uses an improved U-Net combined with a traditional feature extraction multimodal fusion algorithm to accurately classify four types of defects: scratches, ripples, microcracks, and contamination. The defect depth quantitative algorithm establishes a linear relationship between light intensity contrast and defect depth.

7. The quantitative detection system for quartz optical surface defects based on schlieren method according to claim 1, characterized in that, The high-precision positioning module includes a reference platform and a three-dimensional motion platform, which are used to achieve sub-micron level positioning and full-field scanning stitching of samples.

8. A quantitative detection method for quartz optical surface defects based on schlieren method, characterized in that, The detection system according to any one of claims 1-7 includes the following steps: Step 1, System Calibration: Prepare standard quartz samples containing stepped scratches, simulated ripples, and subsurface defects. Measure the same defect area using schlieren method, atomic force microscopy, and white light interferometer, respectively. Establish a quantitative relationship between light intensity contrast and defect depth and ripple amplitude, and determine the system calibration coefficient k. Step 2, Sample Placement and Optical Path Adjustment: Fix the quartz sample to be tested on the T-slot platform, and adjust the concave reflector, polarization system and knife edge position of the Z-shaped optical path to ensure that the parallel beam passes perpendicularly through the sample surface. Step 3, Image Acquisition: Start the light source module and imaging module to acquire the schlieren image of the quartz sample to be tested, and simultaneously acquire four sets of orthogonal polarization images for polarization difference enhancement processing; Step 4, Image Preprocessing: The acquired schlieren image is sequentially processed with illumination compensation, polarization difference enhancement, background suppression, and texture removal to eliminate the effects of uneven illumination and background noise. Step 5, Quantitative Defect Screening and Location: Analyze the preprocessed image, identify the defect type and calculate the light intensity contrast of the defect area, and obtain the defect depth by combining the system calibration coefficient; determine the specific location of the defect through the high-precision positioning module to achieve quantitative screening and location of defects throughout the field; Step 6, Error Compensation and Result Output: Through multi-frame averaging, light intensity normalization, and real-time optical path calibration, the detection errors caused by light source fluctuations, optical path offsets, and environmental vibrations are compensated, and the defect type, depth, location, and detection data are finally output.

9. The quantitative detection method for quartz optical surface defects based on schlieren according to claim 8, characterized in that, The standard sample preparation method described in step 1 is as follows: stepped scratch samples are etched using focused ion beam or atomic force microscopy, simulated ripple samples are etched using photolithography or precision machining, and subsurface defect samples are prepared using a combination of magnetorheological polishing and CCOS small grinding head technology.

Images