A copper-embedded multilayer board defect detection method and system based on eddy current thermography and impedance spectrum fusion

Abstract

The application discloses a copper-embedded multilayer board defect detection method and system based on eddy current thermal imaging and impedance spectrum fusion, relates to the technical field of multilayer board defect detection, and comprises the following steps: applying a high-frequency alternating pulse magnetic field to a copper-embedded multilayer board to be detected; continuously collecting thermal field cooling sequence images after pulse excitation, then extracting thermal diffusion coefficients of copper-embedded regions a and thermal field isotheral line distribution characteristics; using test probes of an automatic machine to contact copper-embedded blocks and corresponding network test points on the copper-embedded multilayer board to be detected, injecting a wideband sweep signal, obtaining impedance spectrum and extracting characteristic parameters; and a , thermal field isotheral line asymmetry, Z LF , R HF and Δ θ are constructed into a multi-dimensional feature vector, input into a fusion judgment model, and a judgment result of the copper-embedded multilayer board defect is output according to a preset multiple threshold. The application realizes non-destructive defect detection, and the micro-crack detection rate is about 97%, and the copper-embedded offset and interlayer short-circuit detection rate is 100%.

Description

Technical Field

[0001] This invention relates to the field of multilayer board defect detection technology, and in particular to a method and system for detecting defects in copper-embedded multilayer boards based on eddy current thermal imaging and impedance spectroscopy fusion. Background Technology

[0002] With the development of high-frequency, high-speed, and high-power electronic devices, copper-embedded multilayer boards are increasingly widely used. However, due to the difference in the coefficient of thermal expansion between the copper-embedded blocks, the resin dielectric, and the inner layer circuitry, latent defects such as microcracks, delamination, or poor connections are easily generated at the interface during lamination, soldering, or thermal cycling.

[0003] Existing detection methods have the following limitations: Flying probe electrical testing: Its principle is to measure the continuity of a contact network (usually with a threshold of a few ohms), which cannot reflect the physical bonding state of the interface. For early microcracks or poor connections where there is still local contact, the slight change in contact resistance (usually <10mΩ) is far below the resolution of flying probe testing, which can easily lead to missed detection.

[0004] Automated Optical Inspection (AOI): Its principle is based on optical reflection to visually inspect the outer layer circuitry, and the inspection depth is limited to the surface of the board. The interface of the copper-embedded blocks is buried deep under the resin dielectric layer, and AOI cannot penetrate the non-transparent resin, so it cannot detect hidden delamination or micro-cracks at the interface.

[0005] X-ray inspection: Its imaging is based on a two-dimensional projection of material density differences. Although it can detect macroscopic physical offsets of copper-embedded blocks, it is difficult to quantify defects in the Z-axis direction of the bonding interface. For tightly bonded microcracks or delamination gaps, because the thickness difference in the transmission direction is extremely small, X-rays cannot effectively identify and determine their bonding state.

[0006] Metallographic sectioning: As a destructive physical analysis method, the plates need to be cut, mounted, and ground before the interface morphology can be observed under a microscope. This method has an extremely low sampling rate (usually ≤0.5%), a long sample preparation cycle, and is a post-production sampling inspection. It not only damages the product but also has a very high risk of missed detection, making it completely unsuitable for the needs of full inspection in mass production.

[0007] CN119310185B discloses a computer vision-based printed circuit board defect detection system that uses the difference in ultrasonic reflection in a medium to identify defects. However, the gaps between defects such as microcracks and delamination at copper-embedded interfaces are usually on the micrometer scale, and the wavelength and resolution of conventional ultrasonic waves are insufficient to effectively distinguish such tiny structures. For weakly bonded areas that have not yet formed obvious cavities, the acoustic impedance difference is extremely weak and cannot be quantitatively evaluated. At the same time, the system requires multiple sets of probes to alternately acquire data from both the front and back sides and complex 3D modeling, making it difficult to adapt to rapid full inspection in mass production.

[0008] CN104730078A discloses a visual inspection system for micro-line defects on circuit boards based on AOI technology. This system determines defects by heating the circuit board and acquiring surface thermal images. However, the detection depth of infrared thermal imaging is limited to the surface of the object and cannot penetrate the resin layer to directly reflect the bonding state of the inner copper-embedded interface. Even if interface defects indirectly cause changes in surface temperature distribution, the signal is diluted by the complex heat conduction network of the multilayer board, making it impossible to locate the defect depth and type, resulting in extremely low sensitivity to minute interface defects. Summary of the Invention

[0009] To address the shortcomings of existing technologies, this invention provides a method and system for detecting defects in copper-embedded multilayer boards based on eddy current thermal imaging and impedance spectroscopy fusion. This method enables non-destructive defect detection and can detect latent defects such as microcracks, delamination, or poor connections at the bonding interface.

[0010] In order to achieve the objective of this invention, the following solution is proposed: A defect detection method for copper-embedded multilayer boards based on eddy current thermal imaging and impedance spectroscopy fusion includes the following steps: S1. Perform eddy current pulse thermal imaging scan: A high-frequency alternating pulsed magnetic field is applied to the copper-embedded multilayer board under test using an array of excitation coils, inducing eddy currents in the copper-embedded blocks and inner network layers. Simultaneously, an infrared thermal imager continuously acquires images of the thermal field cooling sequence after pulse excitation. Then, the thermal diffusivity of the copper-embedded region is extracted based on a one-dimensional heat conduction model. a And the distribution characteristics of thermal isotherms; S2. Perform multi-frequency impedance spectroscopy analysis: The test probes of an automated testing machine are used to contact the copper-embedded blocks and their corresponding network test points on the copper-embedded multilayer board under test. A wideband sweep signal is injected to obtain the impedance spectrum and extract characteristic parameters, including low-frequency impedance amplitude. Z LF Real part of high-frequency impedance R HF And the phase angle offset Δ at a specific high frequency i ; S3. Defect determination based on a multi-dimensional feature fusion judgment model: thermal diffusivity a Thermal field isotherm asymmetry, low-frequency impedance amplitude Z LF Real part of high-frequency impedance R HF and phase angle offset △ iThe feature vector is constructed as a multidimensional feature vector and input into the fusion judgment model embedded in the machine tool control system. Based on the preset multiple thresholds, the judgment results of interface defects and positional offset of the copper-embedded multilayer board are output.

[0011] Furthermore, in step S1, the array-type excitation coil adopts a planar spiral structure; the frequency of the high-frequency alternating pulse magnetic field is 100kHz to 1MHz, and the single pulse width is 10ms to 50ms.

[0012] Furthermore, in step S1, the thermal diffusivity is extracted based on a one-dimensional heat conduction model. a The method is as follows: Extract the surface temperature change curve of the center pixel of the copper-embedded region from the thermal cooling sequence image. T ( t ); surface temperature T Corresponding time t Taking the natural logarithm of each factor, a polynomial fitting was used to obtain the logarithmic cooling function: In the formula, a 0 , a 1 , a 2 , a n This represents the polynomial fitting coefficients. n This represents the order of the polynomial; Find the logarithmic cooling function with respect to ln( t The second derivative of the curve is used to extract the characteristic time corresponding to the peak value of the curve. t peak ; Based on the designed embedment depth or known thickness of the copper insert L The thermal diffusivity of the region can be calculated using the above formula. a .

[0013] Furthermore, the test conditions for step S2 are as follows: Four-wire probes are used to contact network test points; The frequency range of the wideband sweep signal is set to 1kHz to 10MHz; Low frequency impedance amplitude Z LF The corresponding frequency is 1kHz, and the real part of the high-frequency impedance is... R HF and phase angle offset △ i The frequency is 10MHz.

[0014] Furthermore, in step S3, based on the preset multiple thresholds, the corresponding judgment result is: like a ≥80 mm 2 / s, and R HF ≤0.5mΩ, and Δ i If the angle is ≤2°, the copper-embedded multilayer board is considered to have good interface bonding. If satisfied a ≤65mm 2 / s or R HF ≥1.2mΩ△ i If the angle is ≥5°, it is determined that there are interfacial microcracks or delamination defects in the corresponding area; like a ≥80 mm 2 / s, and R HF If the resistance is ≥1.2 mΩ, it is determined that there is a false binding caused by thin-layer contamination in the corresponding area.

[0015] Furthermore, in step S3, based on the preset multiple thresholds, the corresponding judgment result is: If the asymmetry of the thermal field isotherm is greater than 15%, and the low-frequency impedance amplitude is extracted simultaneously. Z LF If a sudden drop occurs, it is determined that there is copper embedding offset or interlayer short circuit in the corresponding area.

[0016] A defect detection system for copper-embedded multilayer boards based on eddy current thermal imaging and impedance spectroscopy fusion is provided to implement the aforementioned detection method. The detection system includes: Thermal detection module: including arrayed excitation coils and infrared thermal imager, used to apply high-frequency alternating pulse magnetic field to copper-embedded multilayer boards and acquire thermal field cooling sequence images; Impedance spectrum testing module: includes a wideband LCR digital bridge and a four-wire probe connected to it, used to inject a wideband sweep signal to obtain the impedance spectrum; The machine tool control system is communicatively connected to the thermal detection module and the impedance spectroscopy testing module, respectively, and is used to control the movement and positioning of the thermal detection module and the test probe. The control system includes a computation module with an embedded multi-dimensional feature fusion judgment model for extracting the thermal diffusivity. a Thermal field isotherm asymmetry, low-frequency impedance amplitude Z LF Real part of high-frequency impedance R HF and phase angle offset △ i Then output the final judgment result.

[0017] like a ≥80 mm 2 / s, and R HF ≤0.5mΩ, and Δ i If the angle is ≤2°, the copper-embedded multilayer board is considered to have good interface bonding. If satisfied a ≤65mm 2 / s or R HF ≥1.2mΩ or Δ i If the angle is ≥5°, it is determined that there are interfacial microcracks or delamination defects in the corresponding area; like a ≥80 mm 2 / s, and R HF If the impedance is ≥1.2 mΩ, then the corresponding region is determined to have pseudo-binding caused by thin-layer contamination; If the asymmetry of the thermal field isotherm is greater than 15%, and the low-frequency impedance amplitude is extracted simultaneously. Z LF If a sudden drop occurs, it is determined that there is copper embedding offset or interlayer short circuit in the corresponding area.

[0018] The beneficial effects of this invention are as follows: it is the first time that eddy current thermal imaging has been applied to the inner layer inspection of copper-embedded boards, upgrading the detection of physical interfaces from destructive sampling to non-destructive full inspection; this invention integrates impedance spectrum and thermal diffusivity coefficient, increasing the microcrack detection rate to about 97%; this invention links thermal field morphology distortion with low-frequency impedance anomaly for judgment, accurately detecting copper embedding offset and interlayer short circuit defects with 100% accuracy. Attached Figure Description

[0019] Figure 1 This is a flowchart of the detection method; Figure 2 This is a schematic diagram of normal thermal diffusion in a copper-embedded multilayer board. Figure 3 This is a schematic diagram of the thermal diffusion of microcrack defects in a copper-embedded multilayer board. Detailed Implementation

[0020] Example 1 like Figure 1 As shown, this embodiment provides a defect detection method for copper-embedded multilayer boards based on eddy current thermal imaging and impedance spectroscopy fusion, including the following steps: S1. Perform eddy current pulse thermal imaging scan: A high-frequency alternating pulsed magnetic field is applied to the copper-embedded multilayer board under test using an array of excitation coils, inducing eddy currents in the copper-embedded blocks and inner layer networks. Figure 2 This is a schematic diagram of normal thermal diffusion in a copper-embedded multilayer board. Figure 3This diagram illustrates the thermal diffusion of microcracks in a copper-embedded multilayer board. The arrayed excitation coils, employing a planar spiral structure, are positioned 2mm to 3mm above the copper-embedded multilayer board. The applied high-frequency alternating pulsed magnetic field has a frequency of 100kHz to 1MHz, with a single pulse width of 10ms to 50ms. Simultaneously, an infrared thermal imager continuously acquires images of the thermal field cooling sequence after pulse excitation. Then, the thermal diffusivity of the copper-embedded region is extracted based on a one-dimensional heat conduction model. a And the distribution characteristics of thermal field isotherms.

[0021] Thermal diffusivity extracted based on a one-dimensional heat conduction model a The specific method is as follows: Extract the surface temperature change curve of the center pixel of the copper-embedded region from the thermal cooling sequence image. T ( t ); surface temperature T Corresponding time t Taking the natural logarithm of each factor, a polynomial fitting was used to obtain the logarithmic cooling function: In the formula, a 0 , a 1 , a 2 , a n These represent the polynomial fitting coefficients, which can be calculated using the least squares method. n This represents the order of the polynomial; Find the logarithmic cooling function with respect to ln( t The second derivative of the curve is used to extract the characteristic time corresponding to the peak value of the curve. t peak ; Based on the designed embedment depth or known thickness of the copper insert L The thermal diffusivity of the region can be calculated using the above formula. a .

[0022] S2. Perform multi-frequency impedance spectroscopy analysis: The test probes of an automated testing machine are used to contact the copper-embedded blocks and their corresponding network test points on the copper-embedded multilayer board under test. A wideband sweep signal is injected to obtain the impedance spectrum and extract characteristic parameters, including low-frequency impedance amplitude. Z LF Real part of high-frequency impedance R HF And the phase angle offset Δ at a specific high frequency i .

[0023] The test conditions were as follows: a four-wire probe contact network was used for the test points; the frequency range of the wideband sweep signal was set to 1kHz to 10MHz; and the low-frequency impedance amplitude was... Z LF The corresponding frequency is 1kHz, and the real part of the high-frequency impedance is... R HF and phase angle offset △ i The frequency is 10MHz.

[0024] Low-frequency extraction: Extract the low-frequency impedance amplitude at 1kHz. Z LF Low-frequency signals have extremely strong penetrating power and are mainly used to assess the macroscopic physical continuity of copper-embedded networks (such as severe breaks or open circuits).

[0025] High-frequency impedance extraction: Extract the real part of the high-frequency impedance at 10MHz. R HF Utilizing the skin effect of high-frequency alternating current, the current is mainly concentrated on the surface of the conductor and contact interface. When there is an extremely thin organic contamination or oxide layer at the copper intercalation interface, its low-frequency impedance may still show conduction, but at a high frequency of 10MHz, the contamination layer will exhibit significantly high contact resistance characteristics.

[0026] High-frequency phase angle offset extraction: Microcracks or delamination at the copper interface manifest as parallel parasitic microcapacitors in the equivalent circuit. As the sweep frequency enters the high-frequency region, the capacitive reactance of this capacitor gradually decreases, causing the phase angle of the overall impedance to shift from inductive to capacitive. Due to the uncertainty of crack size, the frequency point at which this capacitive effect reaches significant abrupt resonance is floating. Therefore, this invention does not use a fixed frequency point for phase extraction, but instead continuously compares and measures the phase angle with the standard good phase angle within a high-frequency sweep range of 1MHz to 10MHz. The system automatically captures the floating frequency point corresponding to the maximum extreme value of the difference between the two (i.e., the most significant capacitive effect) as the "specific high frequency," and uses the maximum deviation at this point as the phase angle offset Δ. i Input judgment model. This dynamic peak-finding extraction mechanism greatly improves the system's compatibility and detection rate for microcracks of various sizes.

[0027] S3. Defect determination based on a multi-dimensional feature fusion judgment model: thermal diffusivity a Thermal field isotherm asymmetry, low-frequency impedance amplitude Z LF Real part of high-frequency impedance R HF and phase angle offset △ iThe feature vector is constructed as a multidimensional feature vector and input into the fusion judgment model embedded in the machine tool control system. Based on the preset multiple thresholds, the judgment results of interface defects and positional offset of the copper-embedded multilayer board are output.

[0028] The method for establishing a multi-dimensional feature fusion judgment model is as follows: A sample set of copper-embedded multilayer boards containing known good products and known defect types was collected. The thermal diffusivity, thermal field isotherm asymmetry, low-frequency impedance amplitude, high-frequency impedance real part, and phase angle offset of each sample were extracted as training feature vectors. Combined with the results of metallographic section physical analysis as real labels, the boundary thresholds for distinguishing different defect states were determined by statistical calibration or classification algorithms, and a multi-dimensional feature fusion judgment model was established.

[0029] (1) Sample set preparation and feature extraction: At least 500 copper-embedded multilayer board samples were extracted, covering good products, artificially introduced thin-layer contamination defects, artificially induced interface microcrack defects, and copper embedding offsets of varying degrees. The feature vector matrix of each sample was extracted: [ a Asymmetry Z LF , R HF , △ i ].

[0030] (2) Obtaining the true state: Metallographic section analysis was performed on all samples, and the true physical microstructure of the copper-embedded interface was observed using a scanning electron microscope (SEM). The specific width of the microcracks, the thickness of the oxide layer, and the actual offset distance were confirmed, and these true physical states were used as the true labels of the corresponding feature vectors.

[0031] (3) Threshold determination: Statistical correlation analysis was performed on the feature vectors and the true labels. Findings: For samples with completely dense interfaces as shown in metallographic sections, the test features were highly concentrated. a ≥80 mm 2 / s and R HF Within a space ≤0.5mΩ. Sectioning revealed that samples with microcrack gaps greater than 2 micrometers exhibited a high-frequency phase angle shift Δ. i All showed capacitive mutations greater than 5°, and a Generally less than 65mm 2 / s. The physical offset of the copper-embedded block causes the heat source distribution generated by eddy current excitation to no longer be centered, and the calculated asymmetry exceeds 15%.

[0032] Based on the preset multiple thresholds, the corresponding judgment result is: like a ≥80 mm 2 / s, and R HF ≤0.5mΩ, and Δ i If the angle is ≤2°, the copper-embedded multilayer board is considered to have good interface bonding. If satisfied a ≤65mm 2 / s or R HF ≥1.2mΩ or Δ i If the angle is ≥5°, it is determined that there are interfacial microcracks or delamination defects in the corresponding area; like a ≥80 mm 2 / s, and R HF If the resistance is ≥1.2 mΩ, it is determined that there is a false binding caused by thin-layer contamination in the corresponding area.

[0033] If the asymmetry of the thermal field isotherm is greater than 15%, and the low-frequency impedance amplitude is extracted simultaneously. Z LF If a sudden drop occurs, it is determined that there is copper embedding offset or interlayer short circuit in the corresponding area.

[0034] To verify the accuracy and reliability of the detection method of this invention, a comparative verification experiment was designed, consisting of 4 groups of samples (a total of 130 panelized 8-layer copper-embedded PCBs), as detailed below: (1) Preparation of experimental samples: Group A (Normal quality, 30 pieces): Manufactured using standard processes; Group B (contamination sham bonding, 40 blocks): Before lamination, a very small amount of residual organosilicon was applied to the surface of the copper block to simulate contamination defects; Group C (interfacial microcracks, 30 pieces): After pressing, extreme thermal cycling (-65℃ to 150℃) was performed to induce microcracks; Group D (Offset and interlayer short circuit, 30 blocks): During the lamination process, human intervention caused the copper-embedded blocks to be offset laterally by 0.2mm to 0.3mm. Some samples were short-circuited due to contact with the adjacent inner layer network caused by the offset.

[0035] (2) Comparison of schemes: Traditional approach: 130 boards were subjected to conventional electrical probing, AOI inspection, and X-ray imaging in sequence. Five boards were then randomly selected for metallographic sectioning.

[0036] The present invention features the following scheme: excitation frequency 500kHz, infrared frame rate 200Hz, and sweep frequency range 1kHz-10MHz.

[0037] (3) Results: Table 1. Test Result Analysis and Comparison Table In group B (sham contamination), due to the lack of a significant air gap, the thermal diffusivity is... aThe change is not obvious, but the real part of the high-frequency impedance... R HF The impedance increases dramatically to over 2.5 mΩ. Traditional methods miss all detections because the DC resistance does not exceed the threshold, while this invention, relying on multi-frequency impedance spectroscopy, achieves a detection rate of 97.50%.

[0038] In group C (interfacial microcracks), microcracks impede heat conduction and reduce the thermal diffusivity. a The average thickness has dropped to 50mm. 2 / s or less, accompanied by a high-frequency phase angle offset Δ i Capacitive offset exceeding 6°.

[0039] In Group D (offset and interlayer short circuit), this invention links thermal field morphology distortion with low-frequency impedance anomaly for judgment, accurately detecting copper offset and interlayer short circuit defects with 100% accuracy, thus making up for the short circuit diagnosis limitation of simple X-ray detection.

[0040] Example 2 A defect detection system for copper-embedded multilayer boards based on eddy current thermal imaging and impedance spectroscopy fusion is provided to implement the detection method described in Example 1. The detection system includes a thermal detection module, an impedance spectroscopy testing module, and a machine control system.

[0041] I. The thermal detection module is a physical interface scanning execution unit, mainly used for non-contact acquisition of the physical thermal conduction characteristics of copper-embedded multilayer boards, specifically including: High-frequency pulse generator and array-type excitation coil: The excitation coil adopts a planar spiral array design to ensure the uniformity of the local magnetic field. The high-frequency pulse generator is electrically connected to the excitation coil. Under the trigger of the machine control system, it generates a high-frequency alternating pulse current with an adjustable frequency of 100kHz to 1MHz and a pulse width of 10ms to 50ms, thereby forming a high-frequency alternating magnetic field 2mm to 3mm above the copper-embedded multilayer board under test.

[0042] High-sensitivity infrared thermal imager: The data output terminal of the infrared thermal imager is connected to the machine tool control system. It is used to continuously acquire the thermal field cooling sequence image of the copper-embedded area in real time after pulse excitation, and convert it into a digital matrix to send to the machine tool control system.

[0043] II. The impedance spectroscopy testing module is an electrical connection quality detection unit, mainly used to quantify the microscopic electrical parameters of the copper interfacing interface. Specifically, it includes: Wideband LCR digital bridge: Features wideband sweep capability, covering a test frequency range of 1kHz to 10MHz, and can output high-precision impedance amplitude and phase angle data. A four-wire probe connects to the wideband LCR digital bridge; the four-wire design effectively eliminates interference from the probe's own contact resistance and the parasitic inductance of the wiring harness, ensuring accurate extraction of micro-ohm level contact resistance.

[0044] III. The machine tool control system serves as the central scheduling and data processing center, acting as the brain of the entire equipment. It connects to the drive end of the thermal detection module, the communication end of the impedance spectroscopy testing module, and the underlying servo motors of the machine tool. The machine tool control system integrates control units and computing modules. Details are as follows: Motion unit: used to move and press the thermal detection module to a designated position to complete the thermal scan; also used to drive the test probe to accurately drop down and contact the corresponding network test point to perform electrical frequency sweep.

[0045] The computation module receives the thermal field cooling sequence images transmitted from the infrared thermal imager, performs double logarithmic transformation, polynomial fitting, and second derivative peak finding on the temperature-time curve of the center pixel, and calculates the thermal diffusivity of the local area. a Simultaneously, the contour of the highest temperature isotherm is extracted, the distance deviation between its geometric center and the design center is calculated, and the asymmetry parameter is output. This is used to receive the frequency response curve returned by the LCR digital bridge and extract the low-frequency impedance amplitude. Z LF Real part of high-frequency impedance R HF and phase angle offset △ i The computing module has a built-in multi-dimensional feature fusion judgment model. The system inputs the extracted physical and electrical parameters into the model, compares them with preset multiple boundary thresholds, and finally outputs on the display terminal whether the test point is a good product or has specific defect diagnosis results such as microcracks, thin-layer contamination, copper embedding misalignment, and short circuit.

[0046] The machine tool control system is communicatively connected to the thermal detection module and the impedance spectroscopy testing module, respectively, and is used to control the movement and downward positioning of the thermal detection module and the test probe. The machine tool control system is equipped with a computing module, which embeds a multi-dimensional feature fusion judgment model for extracting the thermal diffusivity. a Thermal field isotherm asymmetry, low-frequency impedance amplitude Z LF Real part of high-frequency impedance R HF and phase angle offset △ i Then output the final judgment result.

[0047] The above embodiments are only used to illustrate the technical ideas and features of the present invention, and are not intended to be unique or to limit the present invention. Those skilled in the art should understand that various changes or equivalent substitutions made to the present invention without departing from its scope are all within the scope of protection of the present invention.

Claims

1. A defect detection method for copper-embedded multilayer boards based on eddy current thermal imaging and impedance spectroscopy fusion, characterized in that, Includes the following steps: S1. Perform eddy current pulse thermal imaging scan: A high-frequency alternating pulsed magnetic field is applied to the copper-embedded multilayer board under test using an array of excitation coils, inducing eddy currents in the copper-embedded blocks and inner network layers. Simultaneously, an infrared thermal imager continuously acquires images of the thermal field cooling sequence after pulse excitation. Then, the thermal diffusivity of the copper-embedded region is extracted based on a one-dimensional heat conduction model. a And the distribution characteristics of thermal isotherms; S2. Perform multi-frequency impedance spectroscopy analysis: The test probes of an automated testing machine are used to contact the copper-embedded blocks and their corresponding network test points on the copper-embedded multilayer board under test. A wideband sweep signal is injected to obtain the impedance spectrum and extract characteristic parameters, including low-frequency impedance amplitude. Z LF Real part of high-frequency impedance R HF And the phase angle offset Δ at a specific high frequency θ ; S3. Defect determination based on a multi-dimensional feature fusion judgment model: thermal diffusivity a Thermal field isotherm asymmetry, low-frequency impedance amplitude Z LF Real part of high-frequency impedance R HF and phase angle offset △ θ The feature vector is constructed as a multidimensional feature vector and input into the fusion judgment model embedded in the machine control system. Based on the preset multiple thresholds, the judgment result of the copper-embedded multilayer board defect is output.

2. The defect detection method for copper-embedded multilayer boards based on eddy current thermal imaging and impedance spectroscopy fusion according to claim 1, characterized in that, In step S1, the array-type excitation coil adopts a planar spiral structure; the frequency of the high-frequency alternating pulse magnetic field is 100kHz to 1MHz, and the single pulse width is 10ms to 50ms.

3. The defect detection method for copper-embedded multilayer boards based on eddy current thermal imaging and impedance spectroscopy fusion according to claim 1, characterized in that, In step S1, the thermal diffusivity is extracted based on a one-dimensional heat conduction model. a The method is as follows: Extract the surface temperature change curve of the center pixel of the copper-embedded region from the thermal cooling sequence image. T ( t ); surface temperature T Corresponding time t Taking the natural logarithm of each factor, a polynomial fitting was used to obtain the logarithmic cooling function: In the formula, a 0 , a 1 , a 2 , a n This represents the polynomial fitting coefficients. n This represents the order of the polynomial; Find the logarithmic cooling function with respect to ln( t The second derivative of the curve is used to extract the characteristic time corresponding to the peak value of the curve. t peak ; Based on the designed embedment depth or known thickness of the copper insert L The thermal diffusivity of the region can be calculated using the above formula. a .

4. The defect detection method for copper-embedded multilayer boards based on eddy current thermal imaging and impedance spectroscopy fusion according to claim 1, characterized in that, The test conditions for step S2 are: Four-wire probes were used to contact the network test points. The frequency range of the wideband sweep signal is set to 1kHz to 10MHz; Low frequency impedance amplitude Z LF The corresponding frequency is 1kHz, and the real part of the high-frequency impedance is... R HF and phase angle offset △ θ The frequency is 10MHz.

5. The defect detection method for copper-embedded multilayer boards based on eddy current thermal imaging and impedance spectroscopy fusion according to claim 1, characterized in that, In step S3, based on the preset multiple thresholds, the corresponding judgment result is: like a ≥80 mm 2 / s, and R HF ≤0.5mΩ, and Δ θ If the angle is ≤2°, the copper-embedded multilayer board is considered to have good interface bonding. If satisfied a ≤65mm 2 / s or R HF ≥1.2mΩ or Δ θ If the angle is ≥5°, it is determined that there are interfacial microcracks or delamination defects in the corresponding area; like a ≥80 mm 2 / s, and R HF If the resistance is ≥1.2 mΩ, it is determined that there is a false binding caused by thin-layer contamination in the corresponding area.

6. The defect detection method for copper-embedded multilayer boards based on eddy current thermal imaging and impedance spectroscopy fusion according to claim 5, characterized in that, In step S3, based on the preset multiple thresholds, the corresponding judgment result is: If the asymmetry of the thermal field isotherm is greater than 15%, and the low-frequency impedance amplitude is extracted simultaneously. Z LF If a sudden drop occurs, it is determined that there is copper embedding offset or interlayer short circuit in the corresponding area.

7. A defect detection system for copper-embedded multilayer boards based on eddy current thermal imaging and impedance spectroscopy fusion, used to implement the detection method according to any one of claims 1 to 6, characterized in that, The detection system includes: Thermal detection module: including arrayed excitation coils and infrared thermal imager, used to apply high-frequency alternating pulse magnetic field to copper-embedded multilayer boards and acquire thermal field cooling sequence images; Impedance spectrum testing module: includes a wideband LCR digital bridge and a four-wire probe connected to it, used to inject a wideband sweep signal to obtain the impedance spectrum; The machine tool control system is communicatively connected to the thermal detection module and the impedance spectroscopy testing module, respectively, and is used to control the movement and positioning of the thermal detection module and the test probe. The control system includes a computation module with an embedded multi-dimensional feature fusion judgment model for extracting the thermal diffusivity. a Thermal field isotherm asymmetry, low-frequency impedance amplitude Z LF Real part of high-frequency impedance R HF and phase angle offset △ θ Then output the final judgment result.

Images