An online quality evaluation method for high-reliability ceramic package hermetic sealing caps

Abstract

The application provides an online quality evaluation method for a high-reliability ceramic packaging airtight sealing cap, belongs to the technical field of semiconductor packaging and airtightness detection, and realizes dynamic monitoring on the whole process of sealing cap solder solidification, overcomes the limitation that a single sensor is easy to be disturbed, and significantly improves the robustness and accuracy of evaluation; ensures the physical consistency of vibration and capacitance signals on a time axis, lays a reliable foundation for subsequent feature extraction, and adaptively fits the decay time constant based on the vibration signal, and combines the sliding window integral dynamic determination of the end point of solidification shrinkage based on the capacitance signal, avoids the deviation caused by a fixed time window, and makes the calculation of solidification shrinkage completion more physically meaningful. The application integrates multiple feature criteria, realizes rapid, non-destructive online full detection on the airtightness of the sealing cap, effectively replaces the sampling inspection mode of traditional helium mass spectrometry leak detection, and greatly improves the production efficiency and quality control level.

Description

Technical Field

[0001] This invention belongs to the field of semiconductor packaging and hermeticity testing technology, specifically relating to an online quality assessment method for the hermeticity seal of a high-reliability ceramic package. Background Technology

[0002] High-reliability ceramic packaging is widely used in aerospace, military electronics, medical implants, and high-end industrial control. Its airtightness is a core indicator for ensuring the long-term stable operation of internal chips. In the capping process, solder (such as eutectic alloys like AuSn and AuGe) is typically used to fuse the cover plate to the metallization layer of the ceramic substrate. The dynamic behavior of the solder solidification process (including vibration release, capacitance changes, and electromagnetic coupling characteristics) directly determines the final airtightness quality. Currently, the mainstream airtightness testing method in the industry is helium mass spectrometry leak detection. Although this method has high sensitivity, it is a destructive sampling inspection and requires a long testing cycle, making it impossible to achieve full online evaluation. Furthermore, while non-destructive testing methods such as optical inspection, X-ray tomography, and ultrasonic microscopy can detect macroscopic defects such as voids and cracks, they struggle to capture transient physical information during solder solidification. Moreover, the testing equipment is costly and inefficient, making it unsuitable for real-time quality assessment on the production line. In recent years, some studies have attempted to use acoustic emission sensors or piezoelectric films to monitor vibration characteristics during the capping process, or to sense the solder flow state through capacitive sensors. However, these methods mostly monitor a single physical quantity and fail to integrate multi-source sensor information. Furthermore, they lack dynamic modeling and quantitative feature extraction methods for the entire solder solidification process, resulting in insufficient accuracy and robustness in the evaluation.

[0003] Existing technologies for monitoring the quality of sealing caps based on vibration or capacitance have the following main shortcomings: First, single sensor signals are easily affected by fluctuations in process parameters (such as changes in heating temperature and pressure) and environmental noise, resulting in poor stability of feature extraction and difficulty in establishing a reliable mapping relationship with airtightness. Second, existing methods mostly use fixed time windows or empirical thresholds for feature extraction, failing to adaptively adjust the analysis interval according to the actual solidification kinetics of the solder, leading to deviations in the positioning of key physical moments such as the shrinkage completion point. Third, there is a lack of effective processing for time synchronization and response delay compensation of multiple physical quantities. The temporal misalignment between vibration and capacitance signals caused by differences in sensor physical response time is not corrected, directly affecting the physical consistency of subsequent feature fusion. In addition, there are no reports in the existing technology of using the uniformity of phase angle distribution of magnetic core coil arrays to evaluate the coplanarity of sealing caps, and there is a lack of online airtightness detection schemes that fuse multi-dimensional (vibration, capacitance, electromagnetic) information. Summary of the Invention

[0004] The purpose of this section is to outline some aspects of the embodiments of the present invention and to briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section, as well as in the abstract and title of the present application, to avoid obscuring the purpose of this section, the abstract and title of the invention. Such simplifications or omissions shall not be used to limit the scope of the present invention.

[0005] In view of the aforementioned existing problems, the present invention is proposed.

[0006] Therefore, the technical problem solved by this invention is: how to simultaneously collect multi-dimensional physical information such as vibration, capacitance and electromagnetic phase during the solidification process of solder, and achieve accurate online determination of airtightness through adaptive signal processing and multi-feature fusion.

[0007] To address the aforementioned technical problems, the present invention provides the following technical solution: An online quality assessment method for the hermeticity of a high-reliability ceramic encapsulation cap includes: S1: synchronously acquiring the vibration signal output by the piezoelectric film during the cap solidification process, the relative rate of change of capacitance between the cover plate and the metallization layer of the substrate, and the phase angle of each measuring point of the magnetic core coil array below the substrate; S2: performing exponential decay fitting on the vibration signal and extracting the decay time constant as the first feature; S3: performing sliding window integration on the relative rate of change of capacitance to obtain the solidification shrinkage completion degree as the second feature; S4: constructing a phase angle matrix from the phase angles and calculating the ratio of the maximum eigenvalue to the minimum eigenvalue of the phase angle matrix as the phase angle consistency index; S5: determining whether the hermeticity of the cap is qualified based on the decay time constant, the solidification shrinkage completion degree, and the phase angle consistency index.

[0008] As a preferred embodiment of the present invention, wherein: S1 includes: attaching a piezoelectric thin film to the back of a ceramic substrate; connecting the cover plate and the substrate metallization layer to the two input terminals of a high-frequency capacitance measurement circuit respectively; placing a magnetic core coil array below the substrate; simultaneously starting continuous sampling of voltage signals, relative capacitance change rate, and phase angles at each measurement point after the capping heating is completed, with the sampling duration covering the entire solder solidification process; and calculating the relative capacitance change rate based on the collected real-time capacitance.

[0009] In a preferred embodiment of the present invention, after locating the vibration initiation point, signal alignment is further included: extracting waveform segments from the piezoelectric film output voltage signal, extracting waveform segments within the corresponding time interval from the relative rate of change of capacitance, and performing a cross-correlation algorithm to obtain a time delay estimate; and shifting and aligning the time series of the relative rate of change of capacitance to the piezoelectric film output voltage signal according to the time delay estimate to obtain the time-synchronized piezoelectric film output voltage signal and the relative rate of change of capacitance.

[0010] In a preferred embodiment of the present invention, the exponential decay fitting of the vibration signal includes: peak detection of the piezoelectric film output voltage signal after time synchronization, and positioning the sampling point with the largest absolute value of the signal amplitude as the vibration start point; extracting a signal segment within a fixed time window from the vibration start point as the signal to be fitted; taking the absolute value of the signal to be fitted point by point to form an absolute value sequence, performing a natural logarithmic transformation on the absolute value sequence to obtain a logarithmic domain sequence; fitting a straight line to the logarithmic domain sequence using the least squares method, the negative reciprocal of the slope of the straight line being the initial estimate of the decay time constant; using the initial estimate of the decay time constant as the initial value for iteration, performing a nonlinear least squares fitting of the single exponential decay model on the original signal to be fitted using the Levenberg-Marquardt algorithm, and extracting the result after convergence. The value is used as the decay time constant and is denoted as the first characteristic.

[0011] In a preferred embodiment of the present invention, the step of performing a sliding window integration on the relative rate of change of the capacitance includes: setting a length of [value missing] for the relative rate of change of the capacitance after time synchronization. A sliding window is used; starting from the vibration initiation point, the window slides backward point by point; at each window position, the area under the relative rate of change of capacitance within the window is calculated to obtain a sequence of integral values; when the increment of the integral value of multiple consecutive windows is less than the increment threshold, the termination time of the current window is recorded as the end point of solidification shrinkage; within the time interval from the vibration initiation point to the end point of solidification shrinkage, the relative rate of change of capacitance after time synchronization is calculated by definite integral, and the result of the definite integral is recorded as the solidification shrinkage completion degree as the second feature.

[0012] As a preferred embodiment of the present invention, wherein: the length Equal to the decay time constant The number of milliseconds after rounding the value down.

[0013] In a preferred embodiment of the present invention, the setting of the incremental threshold includes: based on the decay time constant. Taking the vibration initiation point as the time zero point, a tail signal segment of the relative change rate of capacitance after time synchronization is extracted from the tail interval on the time axis; the median of the absolute value of the first-order difference of the tail signal segment is calculated and denoted as the noise reference; the increment threshold is set to E times the noise reference and the length is... =The product of the sampling interval and the empirical amplification factor; where E is a constant.

[0014] As a preferred embodiment of the present invention, the construction of the phase angle matrix includes: extracting the phase angle values ​​of each of the four measuring points of the coil array at the time of the end of solidification shrinkage, and arranging them into a phase angle matrix according to the spatial row and column positions of each measuring point below the substrate.

[0015] In a preferred embodiment of the present invention, the generation of the phase angle consistency index includes: representing the four phase angle values ​​in the phase angle matrix in radians, and constructing a real matrix with each element of the matrix as a real value. For the real matrix After symmetry transformation, a symmetric matrix is ​​obtained. The symmetric matrix is ​​solved using the Jacobian eigenvalue decomposition algorithm. Two eigenvalues, arranged in descending order of numerical value. Obtain the maximum and minimum values; if the minimum value is less than 0, replace the minimum value with a preset regularization threshold; calculate the ratio of the maximum value to the minimum value, and record the ratio as the phase angle consistency index.

[0016] In a preferred embodiment of the present invention, the step of determining whether the sealing cap's airtightness is qualified includes: collecting multiple sealing cap samples that have been confirmed as airtight by helium mass spectrometry leak detection, and calculating the decay time constant, solidification shrinkage completion degree, and phase angle consistency index for each sample; and calculating the mean value of the decay time constant. with standard deviation Calculate the mean value of the solidification shrinkage completion rate. with standard deviation Calculate the mean of the phase angle consistency index. with standard deviation Set the upper limit threshold for the decay time constant to be... Set the lower limit threshold for solidification shrinkage completion as follows: Set the upper limit threshold for the phase angle consistency index to be The decay time constant of the current test cap is compared with the upper threshold of the decay time constant; the solidification shrinkage completion degree of the current test cap is compared with the lower threshold of the solidification shrinkage completion degree; the phase angle consistency index of the current test cap is compared with the upper threshold of the phase angle consistency index; if the decay time constant is less than or equal to the upper threshold of the decay time constant, and the solidification shrinkage completion degree is greater than or equal to the lower threshold of the solidification shrinkage completion degree, and the phase angle consistency index is less than or equal to the upper threshold of the phase angle consistency index, then the airtightness of the current cap is determined to be qualified; otherwise, it is determined to be unqualified.

[0017] The beneficial effects of this invention are as follows: Compared with the prior art, the technical effects of this invention are as follows: By synchronously acquiring multi-dimensional physical information such as vibration, capacitance, and phase angle, this invention achieves dynamic monitoring of the entire solidification process of the sealing cap solder, overcoming the limitations of single sensors being susceptible to interference, and significantly improving the robustness and accuracy of the evaluation; by using a cross-correlation algorithm to accurately correct the sensor response delay, it ensures the physical consistency of vibration and capacitance signals on the time axis, laying a reliable foundation for subsequent feature extraction; and by adaptively fitting the decay time constant of the vibration signal and combining it with the dynamic determination of the solidification shrinkage end point through the sliding window integration of the capacitance signal, it avoids the deviation caused by a fixed time window, making the calculation of the solidification shrinkage completion degree more physically meaningful. This invention integrates multiple feature criteria to achieve rapid, non-destructive online full inspection of the sealing cap's airtightness, effectively replacing the traditional helium mass spectrometry leak detection sampling mode, and significantly improving production efficiency and quality control level. Attached Figure Description

[0018] Figure 1 This is a flowchart of an online quality assessment method for a high-reliability ceramic encapsulation hermeticity seal, as described in this invention.

[0019] Figure 2 This is a flowchart for extracting the decay time constant and solidification shrinkage completion.

[0020] Figure 3 This is a flowchart for calculating the phase angle consistency index and determining its pass / failability. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of this invention. The embodiments described in this application are merely some embodiments of this invention, and not all embodiments. Based on the spirit of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of this invention.

[0022] like Figures 1-3 As shown, the online quality assessment method for a high-reliability ceramic encapsulation hermeticity seal of the present invention includes: S1: Synchronously acquire the vibration signal output by the piezoelectric film during the sealing process, the relative change rate of capacitance between the cover plate and the substrate metallization layer, and the phase angle of each measuring point of the magnetic core coil array below the substrate.

[0023] S1.1: Attach the piezoelectric film to the back of the ceramic substrate, connect the cover plate and the substrate metallization layer to the two input terminals of the high-frequency capacitance measurement circuit respectively, and place the 2×2 magnetic core coil array under the substrate.

[0024] Specifically, a polyvinylidene fluoride piezoelectric film is centrally attached to the back of a ceramic substrate using cyanoacrylate adhesive. Its two electrode surfaces are connected to the positive and negative input terminals of a charge amplifier via shielded wires. The cover metallization layer and the substrate metallization layer are used as the two poles of a capacitor, respectively, and connected to the two input terminals of a high-frequency capacitance measurement circuit via a coaxial cable. The high-frequency capacitance measurement circuit uses a 1MHz sinusoidal excitation signal with an excitation amplitude of 5Vpp.

[0025] A 2×2 magnetic core coil array is placed 0.3mm to 1.0mm directly below the substrate in an adjustable position (the specific distance is determined based on the wiring on the back of the substrate, the heat sink, and the fixture space, ensuring no interference with the heating block and clamping mechanism). The magnetic core is made of manganese-zinc ferrite material, with 20 turns per coil and a center-to-center distance of 2mm between each coil. The two ends of each coil are connected to the differential input of a multi-channel phase detection module. The measurement principle is as follows: the magnetic core coil is driven by a built-in high-frequency sinusoidal excitation source (frequency 1MHz, amplitude 5Vpp), generating an alternating magnetic field. Both the cover plate and the substrate metallization layer are conductive, generating eddy currents in the alternating magnetic field. These eddy currents change the equivalent impedance of the coil, especially causing a phase shift in the voltage across the coil relative to the excitation current. The phase shift is related to the cover plate-substrate spacing, solder filling state, and the integrity of the metallization layer. This invention uses a lock-in amplifier to extract the phase difference between the voltage and excitation current of each coil, obtaining the phase angle (unit: degrees) at each measurement point.

[0026] The parameters mentioned above can be selected by those skilled in the art without affecting the feasibility of the solution.

[0027] S1.2: After the cap heating is completed, continuous sampling of voltage signal, relative rate of change of capacitance and phase angle of each measuring point is started simultaneously, and the sampling time covers the entire process of solder solidification.

[0028] At the moment the cap heating ends, a TTL trigger signal is sent to the data acquisition card. The data acquisition card simultaneously starts continuous sampling of three channel groups: the first channel group acquires the output voltage signal of the piezoelectric film at a sampling rate of 50kHz; the second channel group acquires the real-time capacitance output by the capacitance measurement circuit at a sampling rate of 50kHz and processes it to obtain the relative rate of change of capacitance; the third channel group polls the phase angle of the four coil measurement points in turn at a sampling rate of 10kHz (each measurement point has an equivalent sampling rate of 2.5kHz, which is sufficient to reflect the mechanical changes during the solidification process of the solder).

[0029] The relative rate of change of capacitance is calculated by dividing the difference between the real-time capacitance and the initial capacitance (the initial capacitance value at the moment the cap heating ends) by the initial capacitance.

[0030] The total sampling duration is set to 20 times the typical value of the decay time constant (e.g., 2ms) pre-determined based on historical data or offline experiments, and not less than 100ms, to ensure coverage of the entire solder solidification process. This typical value can be obtained by averaging the results of offline testing on at least five qualified samples of the same package type and solder model. If the actual fitted value... If the sampling time exceeds three times the preset typical value, an alarm will be triggered, indicating that the sampling time may be insufficient and the test needs to be repeated after extending the sampling time.

[0031] It should be noted that by using a unified trigger signal to achieve synchronous sampling of the three physical quantities, timing deviations caused by inconsistent start times are avoided.

[0032] S2: Perform exponential decay fitting on the vibration signal and extract the decay time constant as the first feature.

[0033] S2.1: Output voltage signal of the piezoelectric film after time synchronization Perform a global scan, calculate the absolute value of the amplitude at each sampling point, and record the maximum absolute value. ;by 50% is the voltage threshold. Searching backwards from the signal start point, the sampling moment when the absolute value of the amplitude of the first three consecutive sampling points all exceeds the voltage threshold is defined as the vibration start point, denoted as... .

[0034] S2.1a: Perform a cross-correlation algorithm on the piezoelectric film output voltage signal and the relative rate of change of capacitance to calculate the estimated time delay of both.

[0035] Preferably, after determining the vibration initiation point, a waveform segment from 1 ms before to 5 ms after the vibration initiation point is extracted from the acquired piezoelectric film output voltage signal, and a waveform segment within the corresponding absolute time interval is extracted from the relative rate of change of capacitance. Cross-correlation is then performed on the two waveform segments. ; in, It is a piezoelectric signal sequence. It is a sequence of capacitance signals. For sequence length, For delay points; search makes Delay points of the maximum value , This represents the index of the sampling point position in the sequence.

[0036] Delay estimate The calculation is as follows: ; in, The sampling frequency is 50kHz; if A positive value indicates that the capacitive signal lags behind the piezoelectric signal, while a negative value indicates that it leads.

[0037] S2.1b: If the estimated time delay value If the value is not equal to 0, then the time series of the relative rate of change of capacitance will be shifted as a whole. :when When > 0, before deleting the capacitor sequence Add a dot and pad with zeros at the end; when When < 0, pad the capacitor sequence with zeros and delete the corresponding point at the end. If If the value is 0, then no translation is needed; the original relative rate of change of capacitance can be directly used as the relative rate of change of capacitance after time synchronization.

[0038] The relative rate of change of the translated capacitance corresponds point-by-point on the time axis to the original piezoelectric film output voltage signal, and is denoted as the time-synchronized piezoelectric film output voltage signal. Relative rate of change of capacitance after time synchronization .

[0039] It can be seen that by using the above explicit translation and zero-padding operations to eliminate the influence of sensor response delay on subsequent feature extraction (such as decay time constant fitting and sliding window integration), the consistency of the two features in the temporal physical sense can be guaranteed.

[0040] S2.2: Starting point of vibration Starting from the beginning, extract a fixed duration. The signal segment, in which the fixed duration The time constant is set to 5 times the typical value of the expected decay time constant (obtained from offline calibration, 2ms in this embodiment), i.e., 10ms. If the actual fitted value... If the value is significantly larger than the typical value (e.g., more than twice the typical value), it is recommended to retest and extend the sampling time; if the total signal length is insufficient... The actual signal endpoint is taken as the termination point; the intercepted signal segment is denoted as... , .

[0041] S2.3: The signal segment to be fitted By taking the absolute value of each sample point, a non-negative sequence is obtained. For non-negative sequences Perform a natural logarithmic transformation on each non-zero element in the sequence to obtain a logarithmic field sequence. For sampling points with an amplitude of zero, assign them a very small positive number (e.g., 1×10). -6 Take the logarithm after multiplying the maximum absolute value of the signal.

[0042] It should be noted that the above operations convert the oscillation waveform into a one-sided decaying envelope by taking the absolute value, which conforms to the physical form of the exponential decay model. Furthermore, the logarithmic transformation converts the exponential relationship into a linear relationship, assigns a very small positive number to the zero point to avoid logarithmic overflow, and ensures the numerical stability of the subsequent linear fitting.

[0043] S2.4: Using logarithmic field sequences As the dependent variable, corresponding to time Using the least squares method to fit the linear equation as the independent variable. ,in, The slope Intercept; calculate slope The negative reciprocal of, i.e. ,Will Let be the initial estimate of the decay time constant; if the fitted value is... ≥ 0 (indicating no signal attenuation), then Set to a preset upper limit value (100ms).

[0044] It should be noted that this invention obtains the analytical solution of the exponential decay parameter quickly through linear regression, without the need for iteration, and has high computational efficiency.

[0045] S2.5: Using the initial estimate of the decay time constant as the initial value for iteration, the Levenberg-Marquardt algorithm is used to perform nonlinear least-squares fitting of the original signal to be fitted with a single exponential decay model. After convergence, the data is extracted. The value is taken as the decay time constant and denoted as the first characteristic.

[0046] A better option is the single exponential decay model, which is expressed as: ; in, For amplitude parameters, This is the baseline offset parameter. , , All are parameters to be fitted. The initial value is set to (Originated from S2.1) The initial value is set to 0. The initial values ​​are obtained from S2.4. The Levenberg-Marquardt algorithm is used to simultaneously fit the three parameters, and the results are extracted after convergence. value.

[0047] The iteration termination condition is set as follows: the time between two consecutive iterations... The relative change is less than 0.1% or the maximum number of iterations (50) is reached; extraction is performed after convergence. The value of is denoted as the first feature, and its symbol is . .

[0048] It should be noted that this invention combines the advantages of the Levenberg-Marquardt algorithm with the gradient descent method and the Gauss-Newton method, exhibiting good convergence in exponential fitting under noise interference. Furthermore, during solder solidification, internal stress release and solid-liquid interface movement cause micro-vibrations in the substrate. The vibration decay time constant reflects the phase transition kinetics of the solder from liquid to solid and the interfacial bonding strength. Experiments show that a too-small decay time constant usually corresponds to rapid solder solidification but with internal microcracks, while a too-large constant may lead to slow solidification and poor wetting due to solder oxidation or insufficient temperature, both of which reduce airtightness. Therefore, [the following is a continuation of the previous sentence, but the context is unclear: "…"] Controlling the airtightness within the range of ±3σ of the mean of qualified samples is one of the criteria for airtightness compliance.

[0049] S3: Perform a sliding window integral on the relative change rate of the capacitance to obtain the solidification shrinkage completion degree as the second feature.

[0050] S3.1: Read the decay time constant The value is in milliseconds, representing the calculation window length (number of sampling points). ,in, = 50kHz is the sampling frequency. This is the floor function; if If it is less than 3, then Assign a value of 3 (to ensure that the integration window contains at least 3 sampling points).

[0051] It should be noted that when the decay time constant When less than 0.06ms (corresponding to) <3) If the solder solidifies too quickly, the dynamic characteristics of the capacitance change signal are not obvious, and the evaluation accuracy of the present invention may decrease. In this case, it is recommended to use other supplementary detection methods or adjust the process parameters to slow down the solidification process.

[0052] by The number of sampling points contained in the sliding window, and the relative rate of change of the capacitance after time synchronization. Set a sliding window with a length of [value missing]. A series of sampling points; starting from the vibration point The corresponding sampling index is the starting index of the first window. The window moves one sampling point at a time along the increasing time direction.

[0053] It can be seen that the window length Adaptive to decay time constant , ( The larger the window, the slower the solidification (and the longer the window), and it is correctly converted into the number of sampling points.

[0054] S3.2: At the k-th window position, extract the contents of the window. The relative rate of change of capacitance corresponding to each sampling point .

[0055] The integral area within the window is calculated using the compound trapezoidal quadrature formula: ; in, The sampling interval is 1 / 50kHz = 20μs. The integration area of ​​all windows is calculated sequentially to obtain the sequence of integration values. , This represents the total number of windows.

[0056] S3.3: When the increment of the integral value of multiple consecutive windows (3 in this embodiment) is less than the increment threshold, the termination time of the current window is recorded as the end point of solidification shrinkage.

[0057] Based on the decay time constant Taking the vibration initiation point as the zero point of time, in the tail section of the time axis, such as ( The relative rate of change of capacitance after time synchronization is measured within the interval (unit: milliseconds, with the endpoints of the interval measured in milliseconds). The tail signal segment is denoted as .right Calculate the first-order forward difference point by point: Take the median of the first-order difference absolute value sequence, and denote it as the noise benchmark. (The dimensions are the same as the relative rate of change of capacitance); Incremental threshold The noise reference and length are set to E times (preferably 3 times in this embodiment of the invention). The product of the sampling interval and the empirical amplification factor, for example... .in, The empirical amplification factor (e.g., 2~10, taken in this embodiment) =5), the specific number can be determined based on actual needs.

[0058] S3.4: Starting from the first window, calculate the increment of the integral area of ​​each adjacent window in turn. Set the number of consecutive judgment windows to 3. When there exists a certain k such that... Stop sliding when the time is right.

[0059] Record the termination time of the current k-th window. ,in, The window number that meets the condition (counting from 1). If the window number starts counting from 0, then the formula is: This invention uses a counting method starting from 1, so it is corrected to subtract 2. This is recorded as the point at which solidification shrinkage ends.

[0060] If the condition is not met after iterating through all windows, then the termination time of the last window is taken as the end time. .

[0061] S3.5: From the starting point of vibration To the end of solidification shrinkage The relative rate of change of the capacitor after time synchronization within the time interval. To improve accuracy when performing definite integral calculations, the compound Simpson quadrature formula is used: ; in, This represents the number of sampling points within the interval (must be even; if it is odd, discard the last point). , , For the time coordinates of odd-indexed sampling points, Use even-numbered indexed sampling points for their time coordinates (excluding the last index). Calculate the results. This is denoted as the degree of solidification shrinkage completion, and is used as the second characteristic.

[0062] It should be noted that in the operation of this invention, the definite integral interval is defined by the adaptively determined end point of solidification shrinkage, rather than a fixed duration, so that... Physically, it represents the cumulative effect of the relative rate of change of capacitance from the start of solder solidification to the point where shrinkage is basically complete. It is directly related to the amount of solder filling and uses Simpson's integral, which is more accurate than trapezoidal integral, and is especially suitable for applications with smooth capacitance change curves.

[0063] Furthermore, during solder solidification and shrinkage, the distance between the cover plate and the metallized layer of the substrate decreases, resulting in an increase in capacitance. (Integral) This represents the cumulative change in capacitance from the start of vibration to the end of solidification shrinkage, physically equivalent to the total displacement during solder solidification shrinkage, and is positively correlated with the solder filling volume. Insufficient filling will lead to... Low (below) Excessive filler can cause solder to overflow, forming burrs or voids, which also compromises airtightness. In actual production, Too high (e.g., greater than) It should also be judged as unqualified, but the lower limit threshold of the present invention is mainly used to detect insufficient filling defects, and the upper limit can be indirectly guaranteed by process parameters (such as solder amount control).

[0064] To simplify online determination, this embodiment only uses the lower threshold. Those skilled in the art can symmetrically increase the upper threshold as needed.

[0065] S4: Calculate the ratio of the maximum eigenvalue to the minimum eigenvalue of the phase angle matrix, as a phase angle consistency index.

[0066] S4.1: Determine the end point of solidification shrinkage Then, in the phase angle data stream acquired by the third channel group, the distance was searched respectively. The phase angle values ​​corresponding to the four most recent sampling points are denoted as The first subscript represents the row coordinate, and the second subscript represents the column coordinate. The row coordinate corresponds to the X-direction position of the coil array, and the column coordinate corresponds to the Y-direction position.

[0067] It should be noted that, since the third channel group uses a polling method to sequentially collect the phase angles of the four coils, the sampling times at each measuring point are different. Therefore, the search should be performed independently for each measuring point: if the end time falls exactly between two sampling points of that measuring point, then the phase angle at the end time should be estimated using linear interpolation; if the deviation between the end time and a certain sampling time of that measuring point is less than half of the sampling interval, then the value of that sampling point should be directly taken.

[0068] Arrange the four phase angle values ​​in row-major order into a 2×2 real matrix: ; matrix It is denoted as the phase angle matrix.

[0069] It should be noted that fixing the phase angle extraction time to the end point of solidification shrinkage can avoid phase angle fluctuations caused by unstable solder flow due to selecting other times (such as the starting point or any intermediate point).

[0070] S4.2: Read the phase angle matrix The four elements in the equation, each representing an angle value in degrees, are denoted as follows: , , , Multiply each angle value by Convert to radians to get Construct a 2x2 real matrix using four radian values ​​as real matrix elements, arranged in row-major order. : ; It should be noted that the angle unit is uniformly converted to radians here to avoid numerical distortion caused by the angle dimension in subsequent characteristic value calculations (for example, 1° is about 57 times different from 1 rad, and characteristic values ​​have rotation invariance in radians).

[0071] S4.3: Due to the non-ideal nature of coil coupling and phase detection circuits in actual measurements, the original phase angle matrix may be asymmetric. Symmetry processing can ensure that subsequent eigenvalue decomposition yields real eigenvalues, avoiding the occurrence of conjugate complex eigenvalues ​​that lead to unclear physical meaning. Specifically, as follows: For real matrix Perform symmetry transformation and calculate the real matrix. transpose matrix ,in The element in the i-th row and j-th column is equal to The element in the j-th row and i-th column; calculate the symmetric matrix. ,Right now Each element in is and The arithmetic mean of the elements at corresponding positions; if the original Since it is already symmetrical, then .

[0072] It should be noted that in actual measurements, due to the spatial layout of the coil array and the non-ideal nature of the phase detection circuit, the original phase angle matrix... It may be asymmetric. Directly performing eigenvalue decomposition on an asymmetric matrix may yield complex eigenvalues, losing its physical interpretability. Therefore, we should adopt... Symmetry ensures that all eigenvalues ​​are real numbers, and the symmetric matrix retains the main information of the phase angles at each measurement point. The ratio of eigenvalues ​​still reflects the degree of dispersion among the four phase angles. In engineering terms, this symmetry operation is equivalent to performing a least-squares orthogonal projection on the phase angles of each measurement point, without introducing substantial deviations.

[0073] S4.4: Solving symmetric matrices using the Jacobian eigenvalue decomposition algorithm Two eigenvalues and The specific steps are as follows: Set convergence threshold Maximum number of iterations = 100; when The maximum absolute value of the off-diagonal elements is greater than And the number of iterations is less than Repeat the following steps: (1) Select the off-diagonal element with the largest absolute value. ,in ; (2) Calculate the rotation angle: ; (3) Construct the Jacobian rotation matrix Its diagonal element is 1: The rest are 0; (4) Update And multiply and record the rotation matrix; After the iteration ends, The diagonal elements are the two eigenvalues, denoted as . and .

[0074] Furthermore, the eigenvalues and Sort by numerical value in descending order: Let ;like ≤0, then Replace with regularization threshold or The larger of one ten-thousandth, the regularization threshold The value is 1×10 -6 To avoid when When the value is very small, excessive regularization can negatively impact the ratio R; if If the value is greater than 0, the original value remains unchanged.

[0075] Finally, calculation and ratio ; the ratio It is denoted as the phase angle consistency index and is dimensionless.

[0076] It should be noted that the ratio The closer a value is to 1, the more equal the two eigenvalues ​​are, corresponding to similar values ​​for each element of the phase angle matrix. Physically, this reflects a uniform phase angle distribution at the four measurement points, indicating good electromagnetic coupling consistency between the cover plate and the substrate, and indirectly characterizing good coplanarity of the cap. If... A value significantly greater than 1 indicates a large difference in phase angles at each measuring point, suggesting that the cover plate may be tilted or that there may be an abnormality in the local gap. It can serve as an effective criterion for airtightness. It should be noted that... This mainly reflects coplanarity and the uniformity of solder distribution. For cases where the solder is oxidized but coplanarity is good, It may still be acceptable, in which case a comprehensive judgment needs to be made based on the decay time constant and the degree of solidification shrinkage completion. The multi-feature fusion of this invention is precisely to cover the blind spots of a single indicator.

[0077] S5: Determine whether the airtightness of the cap is qualified based on the decay time constant, the solidification shrinkage completion degree and the phase angle consistency index.

[0078] S5.1: Collect multiple (30-50) sealing cap samples that have been confirmed to be airtight by helium mass spectrometry leak detection, and record the number of samples as . For each qualified sample, the decay time constant, solidification shrinkage completion rate, and phase angle consistency index are calculated according to steps S2, S3, and S4 respectively, resulting in three sets of data sequences: .

[0079] S5.2: Calculate the arithmetic mean of the decay time constant sequence: ; Calculate the sample standard deviation : ; Similarly, calculate the mean value of solidification shrinkage completion. with standard deviation Mean of phase angle consistency index with standard deviation .

[0080] Here, the sample standard deviation (denominator) is used. -1) instead of the population standard deviation (denominator) This allows for unbiased estimation of statistics for small samples, improving the reliability of threshold calculation.

[0081] S5.3: Set the upper limit threshold for the decay time constant as follows: Set the lower limit threshold for solidification shrinkage completion as follows: Set the upper limit threshold for the phase angle consistency index to be If the calculated lower limit threshold for solidification shrinkage completion is less than 0, then the lower limit threshold for solidification shrinkage completion is set to 0; if the upper limit threshold for phase angle consistency index is less than 1, then it is set to 1.

[0082] It should be noted that 3 is used The principle (i.e., the control limits of the Shewhart control chart) sets a threshold. When the process is stable, the probability of qualified samples falling within the threshold range is 99.73%, which allows customers to effectively balance missed detections and false detections.

[0083] S5.4: Compare the decay time constant of the current test cap with the upper limit threshold of the decay time constant; compare the solidification shrinkage completion degree of the current test cap with the lower limit threshold of the solidification shrinkage completion degree; compare the phase angle consistency index of the current test cap with the upper limit threshold of the phase angle consistency index. If the decay time constant is less than or equal to the upper threshold of the decay time constant, and the solidification shrinkage completion degree is greater than or equal to the lower threshold of the solidification shrinkage completion degree, and the phase angle consistency index is less than or equal to the upper threshold of the phase angle consistency index, if all three criteria are met, the current sealing cap airtightness is deemed qualified; otherwise, it is deemed unqualified, and the index name corresponding to the unqualified item is recorded.

[0084] The method also includes one or more processors and memory.

[0085] The memory is used to store operable instructions that, when executed by the one or more processors, cause the one or more processors to perform operations, including the flow of the online quality assessment method for the high-reliability ceramic package hermetic caps of the foregoing embodiments, especially... Figure 1 The flowchart of the method is shown.

[0086] Other aspects disclosed in the embodiments of the present invention also propose a computer-readable medium for storing software including instructions executable by one or more computers, which, upon execution, cause the one or more computers to perform operations including the flow of the online quality assessment method for the high-reliability ceramic package hermetic seal of the foregoing embodiments, particularly... Figure 1 The flowchart of the method is shown.

[0087] It should be recognized that embodiments of the present invention may be implemented or carried out by computer hardware, a combination of hardware and software, or by computer instructions stored in a non-transitory computer-readable storage medium.

[0088] The method can be implemented using standard programming techniques, including a non-transitory computer-readable storage medium configured with a computer program in the computer program, wherein the storage medium is configured such that the computer operates in a specific and predefined manner.

[0089] Each program can be implemented in a high-level procedural or object-oriented programming language to communicate with the computer system; however, if required, the program can be implemented in assembly or machine language.

[0090] In any case, the language can be either compiled or interpreted.

[0091] Furthermore, for this purpose, the program can run on programmed application-specific integrated circuits.

[0092] The processes described herein (or variations and / or combinations thereof) can be executed under the control of one or more computer systems configured with executable instructions, and can be implemented by hardware or a combination thereof as code (e.g., executable instructions, one or more computer programs, or one or more applications) that commonly executes on one or more processors. The computer program includes a plurality of instructions executable by one or more processors.

[0093] Furthermore, the method can be implemented in any suitable computing platform, including but not limited to personal computers, minicomputers, mainframes, workstations, networked or distributed computing environments, standalone or integrated computer platforms, or in communication with charged particle tools or other imaging devices.

[0094] Various aspects of the present invention can be implemented in machine-readable code stored on a non-transitory storage medium or device, whether portable or integrated into a computing platform, such as a hard disk, optical read and / or write storage medium, RAM, ROM, etc., such that it can be read by a programmable computer, and when the storage medium or device is read by the computer, it can be used to configure and operate the computer to perform the processes described herein.

[0095] Furthermore, machine-readable code, or parts thereof, can be transmitted via wired or wireless networks.

[0096] When such media includes instructions or programs that combine with a microprocessor or other data processor to implement the steps described above, the invention described herein includes these and other different types of non-transitory computer-readable storage media.

[0097] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. An online quality assessment method for a high-reliability ceramic encapsulation hermeticity seal, characterized in that, include: S1: Simultaneously acquire the vibration signal output by the piezoelectric film during the sealing solidification process, the relative change rate of capacitance between the cover plate and the substrate metallization layer, and the phase angle of each measuring point of the magnetic core coil array below the substrate. S2: Perform exponential decay fitting on the vibration signal and extract the decay time constant as the first feature; S3: Perform a sliding window integral on the relative rate of change of the capacitance to obtain the solidification shrinkage completion degree as the second feature; S4: Construct a phase angle matrix from the phase angles, and calculate the ratio of the maximum eigenvalue to the minimum eigenvalue of the phase angle matrix as a phase angle consistency index; S5: Determine whether the airtightness of the cap is qualified based on the decay time constant, the solidification shrinkage completion degree and the phase angle consistency index.

2. The online quality assessment method for the high-reliability ceramic encapsulation hermeticity sealing cap according to claim 1, characterized in that, S1 includes: A piezoelectric film is attached to the back of a ceramic substrate, and the cover plate and the substrate metallization layer are respectively connected to the two input terminals of the high-frequency capacitance measurement circuit. The magnetic core coil array is placed under the substrate. After the cap heating is completed, continuous sampling of voltage signal, relative rate of change of capacitance and phase angle of each measuring point is started simultaneously, and the sampling time covers the entire process of solder solidification. The relative rate of change of capacitance is calculated based on the collected real-time capacitance data.

3. The online quality assessment method for the high-reliability ceramic encapsulation hermeticity sealing cap according to claim 2, characterized in that, This also includes signal alignment: Waveform segments are extracted from the output voltage signal of the piezoelectric film, and waveform segments within the corresponding time interval are extracted from the relative rate of change of capacitance. A cross-correlation algorithm is then performed to obtain the time delay estimate. Based on the time delay estimate, the time series of the relative change rate of the capacitor is shifted and aligned with the piezoelectric film output voltage signal to obtain the time-synchronized piezoelectric film output voltage signal and the relative change rate of the capacitor.

4. The online quality assessment method for the high-reliability ceramic encapsulation hermeticity sealing cap according to claim 3, characterized in that, Performing exponential decay fitting on the vibration signal includes: Peak detection is performed on the output voltage signal of the piezoelectric film after time synchronization, and the sampling point with the largest absolute value of the location signal amplitude is taken as the vibration start point; The signal segment within a fixed-duration time window, starting from the vibration initiation point, is used as the signal to be fitted. After taking the absolute value point by point of the signal to be fitted, an absolute value sequence is formed. The natural logarithmic transformation is then performed on the absolute value sequence to obtain a logarithmic domain sequence. A straight line is fitted to the logarithmic domain sequence using the least squares method, and the negative reciprocal of the slope of the straight line is the initial estimate of the decay time constant. Using the initial estimate of the decay time constant as the initial value for iteration, the Levenberg-Marquardt algorithm is used to perform nonlinear least-squares fitting of the single exponential decay model on the original signal to be fitted. After convergence, the data is extracted. The value is used as the decay time constant and is denoted as the first characteristic.

5. The online quality assessment method for the high-reliability ceramic encapsulation hermeticity sealing cap according to claim 4, characterized in that, The sliding window integration of the relative rate of change of capacitance includes: A length is set for the relative rate of change of the capacitance after time synchronization. A sliding window; starting from the vibration initiation point, the window slides backward point by point. Calculate the area under the relative rate of change of capacitance within the window at each window location to obtain a sequence of integral values. When the increment of the integral value of multiple consecutive windows is less than the increment threshold, the termination time of the current window is recorded as the end point of solidification shrinkage. Within the time interval from the vibration start point to the solidification shrinkage end point, the relative change rate of capacitance after time synchronization is calculated by definite integral, and the result of the definite integral is recorded as the solidification shrinkage completion degree, as the second feature.

6. The online quality assessment method for the high-reliability ceramic encapsulation hermeticity sealing cap according to claim 5, characterized in that, The length Equal to the decay time constant The number of milliseconds after rounding the value down.

7. The online quality assessment method for the high-reliability ceramic encapsulation hermeticity sealing cap according to claim 5, characterized in that, The setting of the incremental threshold includes: According to the decay time constant Taking the vibration start point as the time zero point, the tail signal segment of the relative change rate of the capacitance after time synchronization is extracted from the tail interval on the time axis. Calculate the median of the absolute value of the first-order difference of the tail signal segment, and denot it as the noise reference; set the incremental threshold to E times the noise reference and length. =The product of the sampling interval and the empirical amplification factor; where E is a constant.

8. The online quality assessment method for the high-reliability ceramic encapsulation hermeticity seal according to claim 5, characterized in that, The construction of the phase angle matrix includes: The phase angle values ​​of each of the four measuring points of the coil array at the end of solidification and shrinkage are extracted and arranged into a phase angle matrix according to the row and column positions of each measuring point in the space below the substrate.

9. The online quality assessment method for the high-reliability ceramic encapsulation hermeticity sealing cap according to claim 8, characterized in that, The generation of the phase angle consistency index includes: Represent the four phase angle values ​​in the phase angle matrix in radians, and construct a real matrix with each element of the matrix as a real value. ; For the real matrix After symmetry transformation, a symmetric matrix is ​​obtained. ; The symmetric matrix is ​​solved using the Jacobian eigenvalue decomposition algorithm. Given two eigenvalues, sort them in descending order of their numerical values ​​to obtain the maximum and minimum values; If the minimum value is less than 0, then the minimum value will be replaced with a preset regularization threshold. Calculate the ratio of the maximum value to the minimum value, and record the ratio as the phase angle consistency index.

10. The online quality assessment method for the high-reliability ceramic encapsulation hermeticity seal according to claim 9, characterized in that, The determination of whether the airtightness of the sealing cap is qualified includes: Multiple sealing cap samples that were confirmed to be airtight by helium mass spectrometry leak detection were collected consecutively, and the decay time constant, solidification shrinkage completion degree and phase angle consistency index of each sample were calculated. Calculate the mean of the decay time constant. with standard deviation Calculate the mean value of the solidification shrinkage completion rate. with standard deviation Calculate the mean of the phase angle consistency index. with standard deviation ; Set the upper limit threshold of the decay time constant as Set the lower limit threshold for solidification shrinkage completion as follows: Set the upper limit threshold for the phase angle consistency index to be ; Compare the decay time constant of the current cap under test with the upper limit threshold of the decay time constant; Compare the solidification shrinkage completion rate of the current test cap with the lower threshold of the solidification shrinkage completion rate; compare the phase angle consistency index of the current test cap with the upper threshold of the phase angle consistency index; If the decay time constant is less than or equal to the upper threshold of the decay time constant, and the solidification shrinkage completion degree is greater than or equal to the lower threshold of the solidification shrinkage completion degree, and the phase angle consistency index is less than or equal to the upper threshold of the phase angle consistency index, then the current sealing cap airtightness is determined to be qualified; otherwise, it is determined to be unqualified.

Images