A system for detecting defects in a capacitive film and a method of detecting defects in a capacitive film

Abstract

The present application relates to the technical field of capacitor film flaw detection, and discloses a capacitor film flaw detection system, which comprises a hollow glass container, a uniform layer of insulating liquid is filled in the sealed interlayer between the upper and lower layers of glass, a light beam is emitted by a light source on one side to penetrate the liquid layer and the capacitor film, the output intensity of the light source controller can be adjusted, an image acquisition device on the other side acquires the light signal, an angle adjusting device is connected to the light source and the image acquisition device to adjust the irradiation and acquisition angle, and a image processing unit analyzes the signal to identify the flaw. The present application realizes nanometer-level detection of capacitor film defects through the light path difference amplification of insulating liquid and the double-beam interference technology, and breaks through the bottleneck of traditional optical detection. The multi-dimensional verification mechanism can accurately distinguish different types of defects, and is compatible with the detection of capacitor films made of various materials. The full-process automatic design realizes flaw detection, positioning and size measurement, provides quantitative quality data for mass production, helps enterprises reduce the waste rate, and improves the reliability of electronic component packaging.

Description

Technical Field

[0001] This invention relates to the field of capacitor film defect detection technology, and in particular to a capacitor film defect detection system and method. Background Technology

[0002] Polypropylene film, due to its high transparency, excellent barrier properties, and mechanical properties, has become a core material for manufacturing high-performance capacitors. With the rapid development of electronics, information technology, and new energy fields, capacitors are evolving towards higher frequencies and smaller sizes, placing higher demands on the quality of capacitor films. Even minute defects on the surface or inside the capacitor film can significantly affect the dielectric properties, withstand voltage, and lifespan of the capacitor, and may even lead to device failure. Industry statistics show that the accuracy of defect detection in capacitor films is directly related to the yield rate of finished capacitors. Traditional detection technologies are insufficient to meet the needs of nanoscale defect detection, necessitating a high-precision, non-contact detection solution.

[0003] Existing capacitor film defect detection technologies have the following limitations: First, traditional optical detection methods are limited by the light scattering characteristics of transparent media, resulting in insufficient signal capture capability for nanoscale defects. Especially when the defect size is smaller than the wavelength of light, the optical contrast is low, which can easily lead to missed detections. Second, although contact detection can obtain surface contour information, it can cause mechanical damage to ultra-thin capacitor films, making it unsuitable for high-speed mass production scenarios. Third, although white light interferometry-based detection technology has a certain level of accuracy, it lacks an optical path difference amplification mechanism, limiting its ability to identify interface defects in media with similar refractive indices, and its detection efficiency is difficult to match the speed of the production line.

[0004] Furthermore, polypropylene capacitor films coated with conductive layers exhibit complex defect characteristics, making it difficult for traditional single-detection methods to simultaneously identify anomalies in both the conductive and dielectric layers. For example, pinhole defects in the metallization layer and thickness fluctuations in the dielectric layer can easily cause superimposed interference in optical signals, leading to misclassification of defect types. While some existing technologies improve detection performance by increasing the light source wavelength or optimizing image processing algorithms, they do not fundamentally address the issues of optical path difference amplification and multi-dimensional defect verification, making it difficult to achieve accurate defect localization and quantitative analysis. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention discloses a capacitor film defect detection system and method that combines high sensitivity, non-contact characteristics, and multi-dimensional verification capabilities.

[0006] This invention discloses a capacitor film defect detection system, which includes:

[0007] Hollow glass containers consist of two layers of glass and a sealed interlayer in between. The interlayer is filled with an insulating liquid to form a uniform liquid layer.

[0008] The light source, located on one side of the hollow glass container, is used to emit a detection beam that penetrates the liquid layer and the capacitor film. The light source is equipped with a controller that can adjust the output intensity, with the standard value being 100% of the light source's rated output power.

[0009] An image acquisition device is located on the other side of the capacitor membrane and is used to acquire light signals transmitted through the capacitor membrane.

[0010] An angle adjustment device is connected to both the light source and the image acquisition device, and is used to adjust the illumination angle of the light source and the acquisition angle of the image acquisition device.

[0011] The image processing unit, connected to the image acquisition device, is used to analyze light signals and identify defects in the capacitor film.

[0012] Furthermore, the thickness of the interlayer in the hollow glass container is 50-100μm, and the uniformity of the insulating liquid layer is controlled by quartz gaskets or microfluidic structures. The interlayer edge is provided with a liquid injection port and a vacuum degassing interface.

[0013] Furthermore, the absolute value of the difference between the refractive index of the insulating liquid and the refractive index of the capacitor film is 0.02-0.25. The insulating liquid is insulating silicone oil or fluorinated liquid, and the capacitor film is a polypropylene dielectric film coated with a conductive layer.

[0014] Furthermore, the light source is a laser or LED light source, combined with a narrow-band filter, the detection beam is split into two coherent beams, one beam passes directly through the liquid layer and the capacitor film, and the other beam is obliquely incident on the capacitor film at an angle of 5°-15°, the two beams form an interference area on the target surface of the image acquisition device.

[0015] Furthermore, the angle adjustment device includes a servo motor-driven rotating bracket for adjusting the included angle θ between the two beams of light, so that the interference fringe spacing is 5-10 pixels and the adjustment accuracy of the included angle θ is ±0.5°.

[0016] Furthermore, the capacitor film defect detection system also includes an optical path adjustment component, which is connected to the light source and is positioned between the light source and the hollow glass container. The optical path adjustment component includes an aperture and a lens, with the aperture diameter being 90%-110% of the detection beam diameter.

[0017] Furthermore, the image processing unit pre-stores a standard defect feature library and identifies defects such as conductive layer fractures, surface irregularities, and dielectric layer impurities in the capacitor film by comparing the offset, contrast, and bright spot distribution of interference fringes.

[0018] This invention discloses a method for detecting defects in capacitor films, using a capacitor film defect detection system as described in any of the above claims, comprising:

[0019] S1: In the pretreatment stage, insulating liquid is filled into the interlayer of the hollow glass container, and the thickness of the liquid layer is controlled to be 50-100μm. The absolute value of the difference between the refractive index of the insulating liquid and the refractive index of the capacitor film is 0.02-0.25. The capacitor film is attached to the lower surface of the container, and air bubbles in the liquid layer are removed by vacuum degassing process. The film surface is treated with plasma to enhance wettability.

[0020] S2: Single beam pre-detection stage, adjust the light source to single beam mode, set the intensity to 30%-50% of the standard value, directly shine through the liquid layer and capacitor film, collect the transmitted light spot image, and use the edge detection algorithm to coarsely locate protrusions or holes ≥5μm on the surface.

[0021] S3: Dual-beam interference amplification stage. Switch the light source to dual-beam mode, adjust the beam angle to 5°-15°, scan within the pitch angle range of 10°-70°, collect interference images at different angles, lock the optimal detection angle according to the image clarity threshold, and use the refractive index difference between the insulating liquid and the capacitor film to amplify the optical path difference at the defect, so that the interference fringes produce a recognizable shift.

[0022] S4: Optical path difference quantification analysis stage. Fix the optimal detection angle and increase the light source intensity to 120%-150% of the standard value. Calculate the defect height by analyzing the interference phase difference and the refractive index difference between the insulating liquid and the capacitor film.

[0023] S5: Multi-dimensional verification stage. Switch to an insulating liquid with a refractive index lower than that of the insulating liquid in S1, and repeat steps S1-S4. Verify the authenticity of the defect by verifying the consistency of the interference fringe shift under different insulating liquids.

[0024] Furthermore, the insulating liquid in S1 is methyl silicone oil or fluorinated liquid. During filling, it is filtered through a 0.1-0.3μm filter membrane, and microchannels are set at the edge of the interlayer to control the uniformity of the liquid layer, ensuring that the insulating liquid adheres tightly to the surface of the capacitor film without air bubbles.

[0025] The capacitor film is a polypropylene dielectric film coated with a conductive layer.

[0026] Furthermore, the light source for the dual-beam interference in S3 is a 600-650nm laser. By adjusting the beam angle, the spacing between the interference fringes is made to be 5-10 pixels. When there are defects in the capacitor film, the optical path difference amplified by the liquid layer causes the fringe offset to be ≥1 pixel. The pixel is the physical pixel of the image acquisition device, and the pixel size is 5-10μm.

[0027] The beneficial effects of this invention are:

[0028] The capacitor film defect detection system and method provided by this invention overcomes the bottleneck of traditional optical inspection in identifying nanoscale defects by combining optical path difference amplification with insulating liquid and two-beam interference technology, achieving high-precision quantitative analysis of capacitor film defects. The system utilizes the refractive index difference between the insulating liquid filling the hollow glass container and the capacitor film to amplify the optical path difference of defects at the 0.1μm level to a recognizable range. Combined with the dynamic fringe shift formed by two-beam interference, the detection sensitivity of hidden defects such as conductive layer pinholes and dielectric layer thickness fluctuations is improved to the nanometer level. Compared to traditional single-source detection technology, this scheme improves the defect detection accuracy from the 5μm level to below 0.1μm through the physical amplification mechanism of the liquid layer and comparison with the feature library of the image processing unit. This effectively solves the industry problem of weak signals and easy missed detection of defects in transparent dielectric films, providing reliable technical support for the quality control of raw materials for high-performance capacitors.

[0029] The multi-dimensional verification mechanism and compatible design endow the detection solution with broad applicability and defect classification capabilities, effectively addressing complex defect scenarios in capacitor films. By switching insulating liquids with different refractive indices for cross-verification, combined with plasma surface treatment in the pretreatment stage, the system can accurately distinguish different types of defects such as conductive layer fractures and dielectric layer impurities, improving classification accuracy compared to traditional single-parameter detection. Furthermore, by adjusting the light source wavelength and detection parameters, this solution is compatible with the detection needs of capacitor films made of various materials such as PET and PI, requiring no additional hardware modifications. The fully automated design, from physical environment preparation to optical signal analysis, not only enables defect detection, location, and dimensional measurement but also provides quantitative quality data for capacitor film mass production, helping companies reduce scrap rates and significantly improve the reliability and production cost control capabilities of electronic component packaging. Attached Figure Description

[0030] Figure 1 This is a flowchart of a capacitor film defect detection method according to an embodiment of this application. Detailed Implementation

[0031] To enable those skilled in the art to better understand the present invention, the technical solutions in the specific embodiments of the present invention will be clearly and completely described below.

[0032] This invention discloses a capacitor film defect detection system, comprising a hollow glass container consisting of upper and lower glass layers and a sealed interlayer in between, with the interlayer filled with an insulating liquid to form a uniform liquid layer. A light source, located on one side of the hollow glass container, emits a detection beam that penetrates the liquid layer and the capacitor film. The light source is equipped with an adjustable output intensity controller, with a standard value of 100% of the light source's rated output power. An image acquisition device, located on the other side of the capacitor film, acquires the light signal transmitted through the capacitor film. An angle adjustment device, connected to both the light source and the image acquisition device, adjusts the illumination angle of the light source and the acquisition angle of the image acquisition device. An image processing unit, connected to the image acquisition device, analyzes the light signal and identifies defects in the capacitor film.

[0033] The capacitor film defect detection system of this invention achieves high-precision detection through modular design. The hollow glass container consists of two layers of optical-grade glass, with an insulating liquid filling the sealed interlayer. The liquid layer thickness is controlled to 50-100 μm using a quartz gasket or microfluidic structure, ensuring uniform coverage of the capacitor film surface. A light source is positioned on one side of the container, and its adjustable output intensity controller precisely regulates the beam energy. The standard value is defined as 100% of the light source's rated output power, which can be switched to 30%-50% or 120%-150% of the rated power during detection. After the detection beam penetrates the liquid layer and the capacitor film, the light signal is received by an image acquisition device on the other side. An angle adjustment device synchronously controls the illumination angle of the light source and the acquisition angle of the image acquisition device, causing the two coherent beams to form an angle of 5°-15° on the capacitor film surface. Finally, the image processing unit analyzes the interference pattern and identifies the defects.

[0034] The capacitor film defect detection system achieves defect detection through a three-tiered mechanism: liquid layer optical path difference amplification, dual-beam interference, and intelligent image analysis. The refractive index difference between the insulating liquid and the capacitor film amplifies the optical path change at the defect location. When defects such as protrusions or holes exist in the capacitor film, the path difference of light passing through the interface between the insulating liquid and the capacitor film is enhanced by the liquid layer, forming a identifiable optical path difference. An angle adjustment device dynamically scans the pitch angle range of 10°-70°, locking in the optimal detection angle so that the two beams of light form clear interference fringes on the target surface of the image acquisition device. The fringes in the defect area appear distorted, broken, or have brightness changes due to the optical path difference shift. The image processing unit's pre-stored standard defect feature library can compare these abnormal signals in real time to accurately locate the defect's position and type.

[0035] The significant advantage of the capacitor film defect detection system lies in its integrated multi-dimensional detection capabilities. Liquid layer uniformity control and refractive index matching design solve the problem of weak defect signals in transparent capacitor films during traditional single-light source detection, amplifying the optical path difference of 0.1μm-level defects to a recognizable range. Staged adjustment of light source intensity, combined with dual-beam interference, improves the contrast between light and dark at defect edges by more than 50%, making it particularly suitable for polypropylene capacitor films coated with conductive layers. Automated angle scanning and intelligent image analysis enable fully non-contact detection, avoiding damage to the capacitor film surface. The detection efficiency matches the pace of high-speed production lines, allowing for the real-time rejection of defective capacitor films.

[0036] In one embodiment, the thickness of the interlayer in the hollow glass container is 50-100μm. The uniformity of the insulating liquid layer is controlled by a quartz gasket or a microfluidic structure, and the interlayer edge is provided with a liquid injection port and a vacuum degassing interface.

[0037] The thickness of the interlayer in hollow glass containers is controlled through a combination of precision mechanical structures and microfluidic technology. Quartz gaskets, serving as spacers, are optically ground to a tolerance of 50-100 μm with a parallelism error ≤1 μm, ensuring uniform spacing between the upper and lower glass layers. For scenarios requiring dynamic adjustment of the liquid layer thickness, a microfluidic structure can be employed. A micron-level flow channel network is etched at the edge of the glass interlayer, and the insulating liquid is uniformly distributed within the channels by air pressure. This, combined with a high-precision flow controller, enables real-time calibration of the liquid layer thickness.

[0038] The insulating liquid injection port and vacuum degassing interface at the edge of the interlayer are integrated and sealed. The injection port is connected to a micro-syringe pump, injecting the insulating liquid into the interlayer at a constant flow rate of 0.1 mL / min to avoid bubble formation caused by turbulence. The degassing interface is connected to the vacuum chamber via a hose, and a negative pressure of -0.1 MPa is simultaneously drawn during the injection of the insulating liquid. The pressure difference causes the gas dissolved in the insulating liquid to escape, and ultrasonic vibration breaks up the tiny bubbles. This mechanism can control the bubble volume ratio of the liquid layer to below 0.01%, ensuring no scattering interference when light passes through and maintaining the clarity of the interference fringes.

[0039] The above design constructs an optically uniform detection medium. A precise liquid layer thickness of 50-100 μm ensures stable and controllable optical path difference amplification, avoiding phase noise caused by excessive liquid layer thickness. The combination of a quartz gasket and microfluidic structure satisfies the low-cost requirement for static detection while supporting liquid layer optimization in dynamic scenarios. Vacuum degassing completely eliminates the interference of air bubbles on light propagation, significantly improving the signal-to-noise ratio for detecting defects at the 0.1 μm level. The vacuum degassing process uses a vacuum of -0.1 MPa and a degassing time of 10 minutes. This solution is particularly suitable for polypropylene capacitor films coated with a conductive layer, accurately identifying hidden defects such as pinholes in the metallization layer and fluctuations in the dielectric layer thickness, providing a reliable hardware foundation for quality control in capacitor film mass production.

[0040] In one embodiment, the absolute value of the difference between the refractive index of the insulating liquid and the refractive index of the capacitor film is 0.02-0.25. The insulating liquid is insulating silicone oil or fluorinated liquid, and the capacitor film is a polypropylene dielectric film coated with a conductive layer.

[0041] The refractive index difference between the insulating liquid and the capacitor film was determined through material screening and optical matching. First, the reference refractive index of the polypropylene capacitor film coated with the conductive layer was measured. Then, media with an absolute difference between the refractive index and the reference value of 0.02-0.25 were selected from the insulating liquid library. The refractive index of the polypropylene capacitor film is approximately 1.50, while the refractive index of the insulating silicone oil is approximately 1.47, resulting in a difference of 0.03 with the polypropylene film. The refractive index of the fluorinated liquid is approximately 1.27, with a difference of 0.23, both meeting the range requirement. During screening, the insulating performance and chemical stability of the insulating liquid were simultaneously verified to ensure that it did not swell the polypropylene film or corrode the conductive layer. Finally, the optical path difference response at the interface between the insulating liquid and the capacitor film was measured using an optical interferometer to determine the optimal matching scheme.

[0042] When the absolute value of the difference between the refractive index of the insulating liquid and the refractive index of the capacitor film is 0.02-0.25, the optical path difference at the defect location can be effectively amplified, improving detection sensitivity. When there is a thickness fluctuation of 0.1 μm in the capacitor film, a refractive index difference of 0.03 can produce an optical path difference of 3 nm, which is converted into a recognizable fringe shift after double-beam interference. If a fluorinated liquid is used, the optical path difference is amplified to 23 nm, and the fringe shift is significantly increased, improving the optical signal of nanoscale defects by 5-8 times. At the same time, the dielectric constant of the insulating liquid is close to that of the polypropylene film, avoiding capacitive interference to the conductive layer during detection and ensuring compatibility between electrical performance and optical detection.

[0043] Matching schemes with an absolute refractive index difference of 0.02-0.25 are specifically designed for polypropylene capacitor films coated with metallized layers. On one hand, the low conductivity of the insulating liquid prevents short circuits in the conductive layer, ensuring safe online testing. On the other hand, the optical path difference amplification effect within the range of the difference can accurately identify composite defects such as conductive layer fractures and dielectric layer pinholes. For example, when a 0.5μm pinhole exists in the metallized layer, the refractive index difference at the interface between the insulating liquid and the capacitor film will enhance light scattering at the pinhole edge. Combined with interference fringe distortion, both conductive layer defects and dielectric layer anomalies can be located simultaneously. The detection accuracy is significantly improved compared to traditional single-light source technology, providing a key quality control method for the high-precision mass production of capacitor films.

[0044] In one implementation, the light source is a laser or LED light source, which is used in conjunction with a narrow-band filter. The detection beam is split into two coherent beams. One beam passes directly through the liquid layer and the capacitor film, while the other beam is incident obliquely on the capacitor film at an angle of 5°-15°. The two beams form an interference region on the target surface of the image acquisition device.

[0045] The light source module achieves dual-beam interference through optical beam splitting components and an angle control mechanism. The beam emitted by the laser or LED light source is filtered for a single wavelength by a narrow-band filter and then split into two beams by a beam-splitting prism. The direct beam penetrates vertically through the liquid layer and capacitor membrane of the hollow glass container, while the oblique beam has its incident angle adjusted to 5°-15° by a set of reflectors. The two beams converge and interfere in the space behind the capacitor membrane. The angle adjustment device uses a servo motor-driven rotating bracket to precisely control the pitch and horizontal angles of the oblique beam, ensuring that the two beams form uniformly spaced interference fringes on the target surface of the image acquisition device. The fringe density is fine-tuned to 5-10 fringes per millimeter, facilitating subsequent image analysis.

[0046] The sensitivity of defect detection is significantly enhanced by using dual-beam optical path difference contrast. When there are thickness fluctuations or surface defects in the capacitor film, the propagation paths of direct and oblique light in the defect area differ. The refractive index difference between the liquid layer and the capacitor film further amplifies this difference, causing the interference fringes to shift, break, or change in brightness. For example, a 0.1 μm thickness change in the capacitor film at a 10° oblique angle can cause a displacement of more than 0.5 pixels in the interference fringes. The narrow-band filter ensures high contrast of monochromatic light interference, increasing the optical signal of nanoscale defects by more than 3 times. Compared with single-beam detection, dual-beam interference can effectively distinguish fabric texture from real defects, reducing the false positive rate.

[0047] The high coherence of the laser light source can penetrate the liquid layer and the metallized conductive layer, accurately capturing internal defects in the dielectric layer. The LED light source, combined with a narrow-band filter, avoids interference from strong light reflections on the conductive layer, clearly revealing surface defects. The dynamic fringes formed by dual-beam interference can simultaneously reflect conductive layer fractures and dielectric layer anomalies. When pinholes exist in the metallized layer, the interference fringes exhibit characteristic distortion at the pinhole edges. Combined with the standard defect library of the image processing unit, the location of conductive layer defects and the range of dielectric layer thickness fluctuations can be simultaneously located. This improves detection efficiency by 50% compared to traditional single-light source technology, meeting the online quality monitoring requirements for high-speed mass production of capacitor films.

[0048] In one implementation, the angle adjustment device includes a servo motor-driven rotating bracket for adjusting the included angle θ between the two beams of light, such that the interference fringe spacing is 5-10 pixels and the adjustment accuracy of the included angle θ is ±0.5°.

[0049] The angle adjustment device achieves high-precision angle control through the collaboration of a servo motor and a precision mechanical structure. The servo motor is a high-torque stepper motor paired with a harmonic reducer with a reduction ratio of 1:100, ensuring the rotation accuracy of the rotating bracket reaches ±0.01°. The bracket employs a dual-axis linkage design; the horizontal and vertical axes control the pitch and horizontal angles of the light source and image acquisition device, respectively. The angle θ between the two beams is fed back to the controller in real time via an encoder, forming a closed-loop adjustment system. When the interference fringe spacing needs to be controlled within 5-10 pixels, the controller automatically calculates the optimal θ value based on the camera pixel size and the light source wavelength, driving the motor to fine-tune with an accuracy of 0.1° / step until the fringe spacing reaches the preset range.

[0050] With an adjustment accuracy of ±0.5° and a fringe spacing of 5-10 pixels, this design provides dual technical support for defect detection. On one hand, precise angle control translates a 0.1μm change in capacitor film thickness into a identifiable fringe shift. When θ=10°, a 1-pixel fringe displacement corresponds to an optical path difference of approximately 10nm, sufficient to detect nanoscale defects. On the other hand, optimized fringe spacing ensures both clarity and resolution in the interference pattern. A 5-pixel spacing is suitable for capturing the overall outline of large defects, while a 10-pixel spacing facilitates the identification of edge features of minute defects. This design overcomes the limitations of traditional fixed-angle detection, dynamically matching the optimal detection angle based on the capacitor film material and defect type, thereby improving the contrast of the interference fringes to over 0.9.

[0051] When pinholes exist in the metallization layer of the capacitor film, adjusting θ to 15° enhances the light scattering interference effect at the pinhole edge, amplifying the fringe distortion by a factor of two. For fluctuations in dielectric layer thickness, a small included angle of 5° can highlight subtle changes in optical path difference, improving the thickness detection resolution to 50nm. Furthermore, the high-speed response of the servo motor can match the high-speed operation of the capacitor film production line, achieving online detection of over 50 meters per minute. This allows for simultaneous automatic adaptation of detection parameters for different batches and materials of capacitor films, significantly reducing manual adjustment costs and the rate of missed detections.

[0052] As one implementation, the capacitor film defect detection system also includes an optical path adjustment component, which is connected to the light source and is disposed between the light source and the hollow glass container. The optical path adjustment component includes an aperture and a lens, and the aperture diameter is 90%-110% of the detection beam diameter.

[0053] The optical path adjustment assembly controls beam quality through a combination of precision mechanical structures and optical elements. The aperture employs an adjustable metal blade structure, driven by a stepper motor to open and close the blades. The aperture range is precisely controlled within 90%-110% of the detection beam diameter. When the beam diameter is 10mm, the aperture diameter is dynamically adjusted to 9-11mm. A plano-convex lens is used, mounted on a two-dimensional adjustment bracket in front of the aperture. The distance between the lens and the light source is finely adjusted using a screw-down knob, ensuring that the diverging beam is collimated by the lens and incident perpendicularly on the center of the aperture. This ensures a tight fit between the beam edge and the aperture aperture, preventing light leakage or energy loss.

[0054] A 90%-110% aperture range achieves a balance between beam energy and detection accuracy. If the aperture is less than 90%, excessive beam obstruction at the edges leads to energy attenuation, affecting the brightness of interference fringes; if the aperture is greater than 110%, excess beam light will illuminate non-detection areas of the capacitor film, generating stray light interference. This range ensures the beam precisely covers the area to be detected, and combined with the optical path difference amplification effect of the liquid layer, the contrast of light intensity changes at defects can be improved by more than 40%. For example, when there is a 0.1μm protrusion on the capacitor film, the beam limited by the aperture can accurately capture the optical path difference change at the edge of the protrusion, preventing light from non-detection areas from masking the defect signal.

[0055] The combined design of the lens and aperture significantly improves the optical performance of the light beam. The lens transforms the diverging light source into a collimated beam, ensuring that the beam penetrates the liquid layer and capacitor film perpendicularly, reducing optical path difference errors caused by oblique incidence. The aperture further trims the beam edge, making the beam shape regular and forming a uniform interference background on the target surface of the image acquisition device. This synergistic effect is particularly important for polypropylene capacitor films coated with conductive layers. The collimated beam can penetrate the metallized conductive layer, avoiding uneven reflection of the conductive layer caused by beam divergence, while the beam range limited by the aperture can accurately locate conductive layer fractures or dielectric layer defects, increasing the detection rate of pinhole defects smaller than 0.5μm to over 95%.

[0056] In one implementation, the image processing unit pre-stores a standard defect feature library and identifies defects such as conductive layer fractures, surface irregularities, and dielectric layer impurities in the capacitor film by comparing the offset, contrast, and bright spot distribution of interference fringes.

[0057] The standard defect feature library of the image processing unit is constructed through multi-dimensional sample acquisition and feature extraction. First, different types of capacitor film defect samples are collected, including typical defects such as conductive layer breaks, surface irregularities, and dielectric layer impurities. Interference images of each defect are acquired under standard operating conditions using a detection system. Through algorithms such as edge detection and frequency domain analysis, the offset distribution of interference fringes, contrast thresholds, and bright spot morphology features are extracted from the images to establish parameterized defect feature templates. For example, for a 0.1 μm surface protrusion, the resulting fringe offset amplitude, the degree of contrast attenuation of adjacent fringes, and the bright spot shape parameters are recorded, ultimately forming a standard library containing hundreds of sets of feature data.

[0058] During real-time inspection, the image processing unit preprocesses the acquired interference image, removing noise through Fourier filtering and extracting the fringe offset, contrast distribution, and bright spot location of the current image. The system performs similarity matching between these real-time parameters and templates in a standard feature library, using a dynamic time warping algorithm to calculate the parameter deviation. When the deviation exceeds a preset threshold, the defect identification logic is automatically triggered. For example, if the fringe offset exceeds 0.5 pixels and the contrast attenuation exceeds 30%, a feature template indicating a conductive layer fracture is matched, thereby locating and classifying the defect. This comparison process can be completed within 50ms, meeting the real-time inspection requirements of high-speed production lines.

[0059] The standard feature library covers complex defect scenarios in capacitor films, enabling simultaneous detection of anomalies in both the conductive and dielectric layers. For example, pinholes in the metallization layer can cause localized breakage of stripes, while impurities in the dielectric layer lead to stripe disorder. By comparing against the feature library, these two types of defects can be accurately distinguished, avoiding misjudgments common in traditional single-parameter detection. Furthermore, the feature library supports self-learning, continuously incorporating new types of defect data, thus continuously improving the system's detection rate for novel defects. In practical applications, the positioning accuracy for dielectric layer impurities reaches 10μm, significantly reducing the missed detection rate of capacitor films and the cost of manual re-inspection, providing quality assurance for the high-reliability packaging of electronic components.

[0060] This invention discloses a method for detecting defects in capacitor films, using a capacitor film defect detection system as described in any of the above claims, comprising:

[0061] S1: In the pretreatment stage, insulating liquid is filled into the interlayer of the hollow glass container, and the thickness of the liquid layer is controlled to be 50-100μm. The absolute value of the difference between the refractive index of the insulating liquid and the refractive index of the capacitor film is 0.02-0.25. The capacitor film is attached to the lower surface of the container, and air bubbles in the liquid layer are removed by vacuum degassing process. The surface of the capacitor film is treated with plasma to enhance wettability.

[0062] S2: Single beam pre-detection stage, adjust the light source to single beam mode, set the intensity to 30%-50% of the standard value, directly shine through the liquid layer and capacitor film, collect the transmitted light spot image, and use the edge detection algorithm to coarsely locate protrusions or holes ≥5μm on the surface.

[0063] S3: Dual-beam interference amplification stage. Switch the light source to dual-beam mode, adjust the beam angle to 5°-15°, scan within the pitch angle range of 10°-70°, collect interference images at different angles, lock the optimal detection angle according to the image clarity threshold, and use the refractive index difference between the insulating liquid and the capacitor film to amplify the optical path difference at the defect, so that the interference fringes produce a recognizable shift.

[0064] S4: Optical path difference quantification analysis stage. Fix the optimal detection angle and increase the light source intensity to 120%-150% of the standard value. Calculate the defect height by analyzing the interference phase difference and the refractive index difference between the insulating liquid and the capacitor film.

[0065] S5: Multi-dimensional verification stage. Switch to an insulating liquid with a refractive index lower than that of the insulating liquid in S1, and repeat steps S1-S4. Verify the authenticity of the defect by verifying the consistency of the interference fringe shift under different insulating liquids.

[0066] In S1, a high-precision quartz gasket is placed between the upper and lower glass layers of the hollow glass container, or the interlayer thickness is controlled by a micron-level flow channel network of a microfluidic chip to ensure uniform liquid layer distribution. Insulating liquid is injected at a constant flow rate of 0.1 mL / min from the liquid injection port at the edge of the interlayer, while a negative pressure of -0.1 MPa is drawn through a vacuum degassing interface and maintained for 10 minutes, combined with 40 kHz ultrasonic vibration, to control the volume ratio of air bubbles in the insulating liquid to below 0.01%. For capacitor film bonding, the capacitor film surface is treated with plasma for 30 seconds to increase surface hydrophilicity, ensuring a tight bond between the insulating liquid and the capacitor film surface, with no micron-level air bubbles remaining. The precise liquid layer thickness of 50-100 μm and the bubble-free environment avoid light scattering and refraction interference, providing a stable physical basis for subsequent optical detection. Plasma treatment reduces the contact angle between the insulating liquid and the capacitor film surface from 80° to below 30°, eliminating interfacial gaps and ensuring that the optical path difference is only caused by defects in the capacitor film itself.

[0067] In the single-beam pre-detection stage of S2, the light source is switched to single-beam mode, with the intensity set to 30%-50% of the rated power to avoid image saturation caused by strong light and reduce energy consumption. The beam penetrates vertically through the insulating liquid layer and capacitor film, and the image acquisition device acquires the transmitted light spot image at a rate of 200fps. Edge detection algorithms extract regions with abrupt changes in grayscale values ​​in the image, and combined with morphological filtering, coarsely locate surface protrusions, holes, or impurities with a size ≥5μm, generating defect ROI coordinates. Single-image analysis can be completed within 20ms, quickly eliminating defect-free areas and narrowing the detection range of dual-beam interference to the pre-positioned ROI, improving overall detection efficiency by more than 30%. The combination of low-intensity light source and edge detection effectively distinguishes between real defects and artifacts caused by slight inhomogeneities in the liquid layer, reducing the computational load of subsequent processing. Edge detection algorithms can be such as Canny edge detection.

[0068] In the dual-beam interference amplification stage of S3, the light source is split into two coherent beams using a beam-splitting prism. One beam is incident perpendicularly, while the other beam is rotated by a servo motor to adjust the included angle to 5°-15° with an angle adjustment accuracy of ±0.5°, ensuring that the two beams form interference fringes with a spacing of 5-10 pixels on the image target surface. Scanning is performed in 1° steps within a pitch angle range of 10°-70°, and the contrast of the acquired image is calculated in real time. When the contrast is ≥0.7, the current angle is locked as the optimal detection angle. At this point, the optical path difference at the defect causes an interference fringe shift of ≥1 pixel. Utilizing the refractive index difference between the insulating liquid and the capacitor film, a 0.1μm change in capacitor film thickness is converted into a detectable optical path difference, resulting in an interference fringe shift of over 0.5 pixels for nanoscale defects, significantly improving detection sensitivity compared to a single beam. Angle scanning automatically matches the optimal detection angle for different defect types, avoiding missed detections at fixed angles.

[0069] In the optical path difference quantification analysis stage of S4, the light source intensity is increased to 120%-150% of the rated power, and 10 frames of images are subjected to mean filtering. Fourier transform is performed on the filtered images to extract the phase information of the interference fringes, and the phase difference between the defect area and the standard area is calculated using a phase-shifting algorithm. Combined with the refractive index data of the insulating liquid and capacitor film recorded in the preprocessing stage, the defect height is inferred. For example, when the refractive index of the insulating liquid is 1.47 and the refractive index of the capacitor film is 1.50, a thickness change of 0.1 μm corresponds to an optical path difference of 0.03 μm. The actual defect size can be calculated through the fringe offset, with an accuracy below 100 nm. This upgrades the assessment from defect presence to defect size measurement, accurately distinguishing between thickness fluctuations of 0.1 μm and 0.2 μm, providing data support for quality grading. Strong light illumination increases the signal-to-noise ratio of the interference fringes from 5:1 to 15:1, allowing accurate identification of impurities even within the transparent dielectric layer through fringe distortion. The defect height can also be calculated by comparing the phase difference of the interference fringes of the standard film and the film under test, combined with the preset correspondence table between the refractive index difference and the fringe offset. The specific correspondence table between the refractive index difference and the fringe offset is obtained by integrating historical data.

[0070] In the multi-dimensional verification stage of S5, the current insulating liquid is drained through a microfluidic valve, and a second insulating liquid with a refractive index 0.1-0.2 lower than the original insulating liquid is injected. The dual-beam interferometry detection process is repeated, keeping other parameters constant. The direction and amplitude of the fringe shift at the same location in the two detections are compared. If the shift difference is <20% and the direction is consistent, it is determined to be a real defect. If the difference is >30%, it is considered an artifact caused by uneven liquid layer or residual air bubbles and is discarded. The false judgment rate of traditional single-liquid detection is about 5%, which is reduced to below 0.3% after multi-insulating liquid verification. In particular, it can eliminate false defect signals caused by impurities in the insulating liquid itself or slight contamination on the capacitor film surface. The refractive index difference of different insulating liquids will change the optical path difference amplification factor. For example, silicone oil and water will have different fringe shifts for the same defect. Combined with the feature library, it can further distinguish between conductive layer fractures and dielectric layer wrinkles, improving the classification accuracy to a high level.

[0071] This invention provides a capacitor film defect detection method that can identify dielectric layer thickness fluctuations at the 0.1μm level and conductive layer pinholes at the 0.5μm level, achieving nanometer-level detection accuracy, far exceeding the industry's conventional 5μm detection standard. The detection time for a single capacitor film is less than 1 second, making it suitable for high-speed production lines at 60m / min. By changing the insulating liquid and light source parameters, it is compatible with capacitor films made of different materials such as PET and PI. The combination of multi-dimensional verification and quantitative analysis automates the entire process of defect detection, location, classification, and dimensional measurement, helping companies reduce capacitor film scrap rates and lower production costs. The entire process requires no contact with the capacitor film surface, avoiding damage caused by traditional mechanical scanning, making it particularly suitable for detecting ultra-thin capacitor films. Through these steps, this capacitor film defect detection method constructs a complete technical chain from physical environment preparation to optical signal enhancement and intelligent analysis verification, providing a systematic solution for high-precision capacitor film detection.

[0072] In one implementation, the insulating liquid in S1 is methyl silicone oil or a fluorinated liquid. During filling, it is filtered through a 0.1-0.3 μm filter membrane, and microchannels are set at the edge of the interlayer to control the uniformity of the liquid layer, ensuring that the insulating liquid adheres tightly to the surface of the capacitor film without air bubbles. The capacitor film is a polypropylene dielectric film coated with a conductive layer.

[0073] In step S1, methyl silicone oil or fluorinated liquid is selected as the insulating liquid. Based on their stable insulation properties and specific refractive index design, methyl silicone oil has a refractive index of 1.47, which differs from the 1.50 refractive index of the polypropylene film by 0.03. The fluorinated liquid has a refractive index of 1.27, resulting in a difference of 0.23, both meeting the requirement of 0.02-0.25. Before filling with the insulating liquid, it is filtered through a 0.1-0.3μm pore size polytetrafluoroethylene (PTFE) filter membrane to remove any dust particles or colloidal impurities, ensuring that the cleanliness of the insulating liquid injected into the interlayer meets ISO 4406 standard 16 / 14. The microfluidic channels at the edge of the interlayer are etched on the glass substrate using photolithography, with a channel width of 50-100μm and a depth of 30-50μm. During injection, the insulating liquid diffuses evenly throughout the interlayer through capillary action in the microfluidic channels. Combined with the vacuum degassing interface, the gap between the insulating liquid and the capacitor film surface is controlled to below 10nm, with no visible air bubbles remaining.

[0074] The 0.1-0.3μm filter membrane and microfluidic design work synergistically to improve the quality of the liquid layer. The filter membrane effectively removes impurities that may interfere with light propagation and avoids scattering noise caused by particles larger than 5μm. The microfluidic channel guides the uniform spread of the insulating liquid, controlling the thickness difference of the liquid layer in different areas within the interlayer to within ±2μm, ensuring the consistency of the optical path difference amplification effect. The high chemical inertness of methyl silicone oil and the low surface tension of the fluorinated liquid ensure that it neither swells the polypropylene substrate nor corrodes the surface conductive layer, guaranteeing the stability of the electrical performance of the capacitor film during the detection process. The combination of vacuum degassing and microfluidic channeling improves the light transmittance uniformity of the liquid layer to over 98%. Compared with the traditional gravity filling method, the bubble residue rate is reduced from 1% to 0.005%, significantly reducing the interference of background noise on interference fringes.

[0075] In one implementation, the light source for the dual-beam interference in S3 is a 600-650nm laser. By adjusting the beam angle, the spacing between the interference fringes is made to be 5-10 pixels. When there are defects in the capacitor film, the optical path difference amplified by the liquid layer causes the fringe offset to be ≥1 pixel. The pixel is the physical pixel of the image acquisition device, and the pixel size is 5-10μm.

[0076] Step S3 uses a 600-650nm wavelength laser as the light source. This wavelength has low absorption in the insulating liquid and polypropylene film, effectively penetrating the conductive layer and exciting a significant interference effect. The laser is split into two beams by a beam-splitting prism: one beam is incident perpendicularly, while the other beam is driven by a servo motor to rotate a support, adjusting the angle with a precision of ±0.5°. This allows the two beams to form interference fringes with a spacing of 5-10 pixels on the target surface of the image acquisition device. When defects exist in the capacitor film, the refractive index difference between the liquid layer and the capacitor film amplifies the optical path difference at the defect, causing a shift of ≥1 pixel in the interference fringes. The system acquires the fringe image in real time and captures the shift using a sub-pixel edge detection algorithm, ensuring that the optical signal of the nanoscale defect can be quantified.

[0077] The 600-650nm laser wavelength and 5-10μm pixel size provide an optimal match, as this wavelength corresponds to a low Rayleigh scattering coefficient in insulating liquids, reducing background noise. The 5-10 pixel fringe spacing ensures the interference pattern is neither too dense nor too sparse. Combined with the 12-bit grayscale resolution of the image acquisition device, fringe shifts as low as 0.1 pixels, corresponding to optical path difference changes of approximately 10nm, can be detected. This parameter combination increases the system's sensitivity to detecting dielectric layer thickness fluctuations at the 0.1μm level to three times that of conventional white light interferometry, while also improving the detection rate of pinholes in the conductive layer.

[0078] It should be understood that those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims.

Claims

1. A method for detecting defects in capacitor films, characterized in that, include: S1: In the pretreatment stage, insulating liquid is filled into the interlayer of the hollow glass container, and the thickness of the liquid layer is controlled to be 50-100μm. The absolute value of the difference between the refractive index of the insulating liquid and the refractive index of the capacitor film is 0.02-0.

25. The capacitor film is attached to the lower surface of the container, and air bubbles in the liquid layer are removed by vacuum degassing process. The film surface is treated with plasma to enhance wettability. S2: Single beam pre-detection stage, adjust the light source to single beam mode, set the intensity to 30%-50% of the standard value, directly shine through the liquid layer and capacitor film, collect the transmitted light spot image, and use the edge detection algorithm to coarsely locate protrusions or holes ≥5μm on the surface. S3: Dual-beam interference amplification stage. Switch the light source to dual-beam mode, adjust the beam angle to 5°-15°, scan within the pitch angle range of 10°-70°, collect interference images at different angles, lock the optimal detection angle according to the image clarity threshold, and use the refractive index difference between the insulating liquid and the capacitor film to amplify the optical path difference at the defect, so that the interference fringes produce a recognizable shift. S4: Optical path difference quantification analysis stage. Fix the optimal detection angle and increase the light source intensity to 120%-150% of the standard value. Calculate the defect height by analyzing the interference phase difference and the refractive index difference between the insulating liquid and the capacitor film. S5: Multi-dimensional verification stage. Switch to an insulating liquid with a refractive index lower than that of the insulating liquid in S1, and repeat steps S1-S4. Verify the authenticity of the defect by the consistency of the interference fringe shift under different insulating liquids. The capacitor film defect detection method is implemented through a capacitor film defect detection system. The capacitor film defect detection system includes: Hollow glass containers consist of two layers of glass and a sealed interlayer in between. The interlayer is filled with an insulating liquid to form a uniform liquid layer. The light source, located on one side of the hollow glass container, is used to emit a detection beam that penetrates the liquid layer and the capacitor film. The light source is equipped with a controller that can adjust the output intensity, with the standard value being 100% of the light source's rated output power. An image acquisition device is located on the other side of the capacitor membrane and is used to acquire light signals transmitted through the capacitor membrane. An angle adjustment device is connected to both the light source and the image acquisition device, and is used to adjust the illumination angle of the light source and the acquisition angle of the image acquisition device. The image processing unit, connected to the image acquisition device, is used to analyze light signals and identify defects in the capacitor film.

2. The capacitor film defect detection method according to claim 1, characterized in that: The insulating liquid in S1 is methyl silicone oil or fluorinated liquid. During filling, it is filtered through a 0.1-0.3μm filter membrane, and microchannels are set at the edge of the interlayer to control the uniformity of the liquid layer, ensuring that the insulating liquid adheres tightly to the surface of the capacitor film without air bubbles. The capacitor film is a polypropylene dielectric film coated with a conductive layer.

3. The method for detecting defects in capacitor films according to claim 2, characterized in that: The light source for the dual-beam interference in S3 is a 600-650nm laser. By adjusting the beam angle, the spacing between the interference fringes is made to be 5-10 pixels. When there are defects in the capacitor film, the optical path difference amplified by the liquid layer causes the fringe offset to be ≥1 pixel. The pixel is the physical pixel of the image acquisition device, and the pixel size is 5-10μm.

4. The method for detecting defects in capacitor films according to claim 1, characterized in that: The thickness of the interlayer in the hollow glass container is 50-100μm. The uniformity of the insulating liquid layer is controlled by quartz gaskets or microfluidic structures, and the interlayer edge is provided with liquid injection port and vacuum degassing interface.

5. The capacitor film defect detection method according to claim 4, characterized in that: The absolute value of the difference between the refractive index of the insulating liquid and the refractive index of the capacitor film is 0.02-0.

25. The insulating liquid is insulating silicone oil or fluorinated liquid, and the capacitor film is a polypropylene dielectric film coated with a conductive layer.

6. The method for detecting defects in capacitor films according to claim 1, characterized in that: The light source is a laser or LED light source, which is used in conjunction with a narrow-band filter. The detection beam is split into two coherent beams. One beam passes directly through the liquid layer and the capacitor film, while the other beam is obliquely incident on the capacitor film at an angle of 5°-15°. The two beams form an interference area on the target surface of the image acquisition device.

7. The method for detecting defects in capacitor films according to claim 1, characterized in that: The angle adjustment device includes a servo motor-driven rotating bracket for adjusting the included angle θ between the two beams of light, so that the spacing between the interference fringes is 5-10 pixels and the adjustment accuracy of the included angle θ is ±0.5°.

8. The method for detecting defects in capacitor films according to claim 1, characterized in that: The capacitor film defect detection system also includes an optical path adjustment component, which is connected to the light source and is positioned between the light source and the hollow glass container. The optical path adjustment component includes an aperture and a lens, with the aperture diameter being 90%-110% of the detection beam diameter.

9. The method for detecting defects in capacitor films according to claim 1, characterized in that: The image processing unit has a pre-stored standard defect feature library. By comparing the offset, contrast and bright spot distribution of interference fringes, it identifies defects such as conductive layer fracture, surface unevenness and dielectric layer impurities in the capacitor film.

Images