CN-121834624-B - High-reliability packaging quality assessment method and system based on multi-source process data

CN121834624BCN 121834624 BCN121834624 BCN 121834624BCN-121834624-B

Abstract

The invention discloses a high-reliability packaging quality assessment method and a high-reliability packaging quality assessment system based on multi-source process data, which relate to the technical field of quality assessment, and realize online perception of orientation consistency of ceramic particles by adopting terahertz spectrum to perform nondestructive scanning on unsintered green ceramic tapes, and avoid material waste and process interruption caused by destructive detection; the method comprises the steps of carrying out a process of dividing the cutting edge into a plurality of pieces, synchronously collecting the shape of the cutting edge in a scribing process and carrying out geometric feature quantification, improving the identification of the broken edge state from subjective qualitative judgment to objective quantitative characterization, obviously improving the detection consistency and precision, carrying out point-by-point correlation on spectrum coding data reflecting the internal quality and shape coding data reflecting the external defects, and constructing a two-dimensional mapping relation of internal process parameters and external processing results, so that whether the broken edge is generated by potential abnormality in a casting stage in a cutting process can be accurately traced, and a data closed loop is provided for the tracing optimization of the process parameters.

Inventors

AO GUOJUN
CHEN YEMING
GAO HUI
LI ZONGYA

Assignees

南京睿芯峰电子科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260311

Claims (10)

1. A high-reliability packaging quality assessment method based on multi-source process data is characterized by comprising the following steps: Carrying out terahertz spectrum scanning on the unsintered green ceramic tape, carrying out space slicing according to the preset chip position, extracting terahertz frequency domain characteristics in each slicing, and generating spectrum coding data representing the casting orientation consistency of ceramic particles; synchronously collecting the morphology of the cutting edge corresponding to each chip position of the sintered ceramic substrate in a scribing process, and carrying out space slicing and geometric feature extraction on the morphology of scratches at each position to generate morphology coding data representing the cutting edge breakage state; Taking the chip positions as indexes, correlating the spectrum coding data with the morphology coding data point by point, and constructing a multi-source process data pair of each chip position; Calculating casting consistency indexes of the spectrum coding data of each chip position, calculating edge collapse severity indexes of the morphology coding data, obtaining comprehensive quality scores of the corresponding chips through weighted fusion, and marking the chips with the comprehensive quality scores lower than a preset score threshold as abnormal areas of the process parameters.
2. The method for evaluating the packaging quality based on the multi-source process data according to claim 1, wherein the method is characterized in that the unsintered green ceramic tape is subjected to terahertz spectrum scanning to obtain a three-dimensional terahertz data cube covering the whole surface of the green ceramic tape; three dimensions of the three-dimensional terahertz data cube are a transverse space coordinate X, a longitudinal space coordinate Y and terahertz time-domain sampling points respectively, and each space coordinate point (X, Y) corresponds to a complete time-domain spectrum curve; Dividing the three-dimensional terahertz data cube into a plurality of space fragments according to a preset chip layout position Each space slice Corresponds to a chip position and contains time domain spectrum data of all sampling points in the area.
3. The method for highly reliable package quality assessment based on multi-source process data as set forth in claim 2 wherein said generating of said spectrally encoded data comprises: For each space slice Performing Fourier transform on all the time domain spectrum curves in the sampling points to obtain frequency domain spectrums of the sampling points; Calculating an amplitude average value on a preset characteristic frequency band as a frequency domain characteristic amplitude value of a corresponding sampling point for the frequency domain spectrum of each sampling point; calculating standard deviation of frequency domain characteristic amplitude values of all sampling points in the space slicing as casting consistency characteristic values of the space slicing; Slicing each space The standard deviation of the frequency domain characteristic amplitude is arranged according to the chip position sequence to form a frequency spectrum coding data sequence for representing the casting orientation consistency of ceramic particles.
4. The method for evaluating the quality of a package based on multi-source process data as recited in claim 3, wherein said spatially slicing and geometrically extracting the scratch morphology of each location comprises: in the process of scribing a ceramic substrate, synchronously acquiring the shape of a cutting edge of each chip position through a line scanning white light interferometer to obtain three-dimensional surface profile data covering all cutting channels; Dividing the three-dimensional surface profile data into a plurality of spatial slices Each space slice A complete cut boundary region corresponding to one chip; For each space slice And extracting the maximum value of the edge breaking depth and the area of the edge breaking area of the cutting edge from the profile data in the chip as the edge breaking characteristic parameters of the chip.
5. The method for evaluating the packaging quality based on the multi-source process data according to claim 4, wherein the maximum value of the edge collapse depth and the area of the edge collapse area of each chip are formed into a two-dimensional feature vector, the two-dimensional feature vectors are arranged according to the chip position sequence, and a morphology coding data sequence representing the cutting edge collapse state is generated, wherein each element in the morphology coding data sequence corresponds to one two-dimensional feature vector of the chip position.
6. The method for high reliability package quality assessment based on multi-source process data according to claim 5 wherein said point-by-point correlation comprises: and matching the standard deviation of the frequency domain characteristic amplitude of the corresponding position in the spectrum coding data sequence with the two-dimensional characteristic vector of the same position in the morphology coding data sequence by taking the coordinate of each chip position as an index to generate a multi-source process data pair of the corresponding chip position.
7. The method for evaluating the packaging quality based on the multi-source process data according to claim 6, wherein for each chip position, the standard deviation of the frequency domain characteristic amplitude is extracted from the frequency spectrum coding data sequence, the Euclidean distance between the standard deviation of the frequency domain characteristic amplitude and a preset standard casting characteristic value is calculated, and the normalized reciprocal of the Euclidean distance is used as a casting consistency index value of the corresponding chip; extracting a maximum value of the edge collapse depth and the area of the edge collapse area from the morphology coding data sequence aiming at each chip position, and carrying out weighted summation and normalization on the maximum value and the area of the edge collapse area to obtain an index value of the edge collapse severity of the corresponding chip; And taking the 1-edge collapse severity index value as an edge collapse quality index, and carrying out weighted fusion on the edge collapse severity index value and a casting consistency index to obtain a comprehensive quality score, wherein the weight sum of the two indexes is 1.
8. A high-reliability packaging quality evaluation system based on multi-source process data, which is based on the high-reliability packaging quality evaluation method based on multi-source process data according to any one of claims 1 to 7, and is characterized by further comprising: the terahertz spectrum coding module is used for carrying out terahertz spectrum scanning on the unsintered green ceramic tape, carrying out space slicing according to the preset chip position, extracting terahertz frequency domain characteristics in each slicing and generating spectrum coding data representing the casting orientation consistency of ceramic particles; the cutting morphology coding module synchronously collects the morphology of the cutting edge corresponding to each chip position of the sintered ceramic substrate in the dicing process, performs space slicing and geometric feature extraction on the scratch morphology of each position, and generates morphology coding data representing the cutting edge breakage state; The multi-source data association module is used for associating the spectrum coding data with the morphology coding data point by taking the chip position as an index, and constructing a multi-source process data pair of each chip position; The comprehensive quality scoring module is used for calculating casting consistency indexes for the spectrum coding data of each chip position respectively, calculating edge collapse severity indexes for the morphology coding data, obtaining comprehensive quality scores of the corresponding chips through weighted fusion, and marking the chips with the comprehensive quality scores lower than a preset scoring threshold as abnormal areas of the process parameters.
9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that the processor implements the steps of the multi-source process data based high reliability package quality assessment method according to any one of claims 1 to 7 when executing the computer program.
10. A computer readable storage medium having stored thereon a computer program, wherein the computer program when executed by a processor implements the steps of the multi-source process data based high reliability package quality assessment method of any of claims 1 to 7.

Description

High-reliability packaging quality assessment method and system based on multi-source process data Technical Field The invention relates to the technical field of quality evaluation, in particular to a high-reliability packaging quality evaluation method and system based on multi-source process data. Background The low temperature co-fired ceramic (Low Temperature Co-FIRED CERAMIC, LTCC) technology is used as a key technology for realizing integrated packaging of microwave components, millimeter wave circuits and high-density systems, and has been widely applied in the fields of communication, radar, aerospace and the like in recent years. The LTCC process generally involves multiple processes of green tape casting, slicing, punching, hole filling, conductor printing, lamination, sintering, and subsequent dicing, wherein the green tape casting process determines the directional arrangement state of ceramic particles, directly affects the dielectric uniformity and mechanical strength of the substrate after sintering, and the dicing process after sintering determines the geometric integrity and edge quality of the chip or circuit unit. Currently, for quality control of LTCC substrate manufacturing processes, offline spot inspection is mostly used in the industry, for example, by observing the profile of a cross section with a scanning electron microscope to evaluate the casting orientation, or by detecting the edge breakage of a cut edge with an optical microscope. However, the method has inherent limitations such as low detection efficiency, limited sample representativeness, difficult realization of full-flow online feedback, and the like. In the prior art, consistency evaluation of casting process mainly depends on dielectric property test or destructive section microscopic analysis of the substrate after sintering. The former can only reflect macroscopic average characteristics, and can not locate the local casting abnormal region, while the latter is time-consuming and labor-consuming and can not be applied to real-time monitoring of the production line. For quality control of the dicing process, the conventional means is manual visual inspection or traditional machine vision inspection based on two-dimensional images, and the methods can only identify obvious edge breakage, and lack quantitative characterization capability for micron-scale edge breakage morphology and crack propagation trend, and the detection standard is easily influenced by subjective factors. More importantly, in the prior art, the casting process control and the cutting quality control are often treated as two mutually independent links, and the inherent association between the two links cannot be established, so that whether the problem of cutting edge breakage is caused by the potential defect of a casting stage cannot be traced, and the closed loop capacity of process optimization is limited. Disclosure of Invention The present invention has been made in view of the above-described problems occurring in the background art. Therefore, the problem to be solved by the present invention is how to realize efficient, nondestructive and online detection of the internal quality feature of casting consistency and the external geometric feature of cutting edge breakage in the LTCC substrate manufacturing process, and perform correlation analysis on the two to locate the abnormal process region. In order to solve the technical problems, the invention provides the following technical scheme: The invention provides a high-reliability packaging quality assessment method based on multi-source process data, which comprises the steps of carrying out terahertz spectrum scanning on unsintered green ceramic tapes, carrying out space slicing according to preset chip positions, extracting terahertz frequency domain characteristics in each slice, generating spectrum coding data representing casting orientation consistency of ceramic particles, synchronously collecting cutting edge morphology corresponding to each chip position of a sintered ceramic substrate in a slicing process, carrying out space slicing and geometric characteristic extraction on scratch morphology of each position, generating morphology coding data representing cutting edge collapse state, carrying out point-by-point correlation on the spectrum coding data and the morphology coding data by taking the chip position as an index, constructing multi-source process data pairs of each chip position, respectively calculating casting consistency indexes for the spectrum coding data of each chip position, calculating edge collapse severity indexes for the morphology coding data, obtaining comprehensive quality scores of corresponding chips through weighting fusion, and marking chips with the comprehensive quality scores lower than a preset scoring threshold as process parameter abnormal areas. The high-reliability packaging quality assessment method based on multi-source process data is char