A process for controlling the surface flatness of LTCC substrate

Abstract

The present invention discloses a method for controlling the surface flatness of an LTCC substrate, which relates to the technical field of LTCC substrate processing. The method comprises: obtaining structural information of a signal layer, a power layer, and a ground layer superimposed within a target LTCC substrate to be sintered; determining K initial channel flow rates and K initial airflow directions; obtaining an array sequence of multi-point temperature timing monitoring values; obtaining K adjustment channel flow rates and K adjustment airflow directions; and after the target LTCC substrate cools, obtaining the target LTCC substrate and surface flatness detection results, performing feedback adjustment based on the surface flatness detection results, and sintering the target LTCC substrate again. The present invention solves the technical problem in the prior art of lacking in-depth analysis of the operating conditions within the sintering furnace, resulting in an inability to adaptively adjust the airflow, which reduces the reliability of the substrate surface flatness. The method achieves the technical effect of dynamically adjusting the airflow within the sintering furnace and improving the substrate surface flatness.

Description

Technical Field

[0001] The present invention relates to the technical field of LTCC substrate processing, and in particular to a process method for controlling the surface flatness of an LTCC substrate. Background Art

[0002] Low-temperature co-fired ceramic (LTCC) substrates are an important electronic packaging material, widely used in the manufacture of high-frequency, high-temperature, and complex circuits. During the sintering process, the substrate's quality and performance are significantly impacted. During this process, various substrate layers (such as the signal, power, and ground layers) are subjected to high temperatures. Inconsistent densification temperatures and uneven temperature distribution can lead to surface defects such as warping and protrusions, compromising surface flatness.

[0003] The existing technology lacks an in-depth analysis of the operating conditions in the sintering furnace, resulting in an inability to adaptively adjust the airflow, which reduces the reliability of the substrate surface flatness. Summary of the Invention

[0004] The present application provides a method for controlling the surface flatness of an LTCC substrate, which is used to solve the technical problem in the prior art that there is a lack of in-depth analysis of the operating conditions in a sintering furnace, resulting in an inability to adaptively adjust the airflow and reducing the reliability of the substrate surface flatness.

[0005] In view of the above problems, the present application provides a method for controlling the surface flatness of an LTCC substrate, the method comprising:

[0006] Acquire structural information of the signal layer, power layer, and ground layer stacked in the target LTCC substrate to be sintered, and obtain signal line structure information, power grid structure information, and ground grid structure information;

[0007] Identifying channel flow rates and airflow directions of K independent airflow channels in the sintering furnace based on the signal line structure information, the power grid structure information, and the ground grid structure information, and determining K initial channel flow rates and K initial airflow directions, where K is a positive integer;

[0008] The target LTCC substrate is placed in a sintering furnace for debinding and sintering, and a temperature monitoring array is used to perform multi-point temperature time series monitoring on the target LTCC substrate according to a preset monitoring window to obtain a multi-point temperature time series monitoring value array sequence;

[0009] Performing temperature balance analysis on the array sequence of the multi-point temperature time series monitoring values, and adjusting the K initial channel flow rates and the K initial airflow directions according to the analysis results to obtain K adjusted channel flow rates and K adjusted airflow directions;

[0010] The K independent airflow channels are adjusted according to the K adjustment channel flow rates and the K adjustment airflow directions until sintering is completed, and the airflow rates of the K independent airflow channels are reduced according to a preset airflow rate reduction gradient. After the target LTCC substrate is cooled, a sintered target LTCC substrate is obtained, and a surface flatness test is performed to obtain a surface flatness test result. If the surface flatness test result does not meet the requirements, the K independent airflow channels are adjusted according to the surface flatness test result, and the target LTCC substrate is sintered again.

[0011] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0012] The present application obtains the structural information of the signal layer, power layer and ground layer superimposed in the target LTCC substrate to be sintered, and obtains the signal line structure information, the power grid structure information and the ground grid structure information; based on the signal line structure information, the power grid structure information and the ground grid structure information, the channel flow and airflow direction of K independent airflow channels in the sintering furnace are identified, and K initial channel flow rates and K initial airflow directions are determined, where K is a positive integer; the target LTCC substrate is sent into the sintering furnace for debinding and sintering, and the target LTCC substrate is subjected to multi-point temperature time series monitoring according to a preset monitoring window using a temperature monitoring array to obtain a multi-point temperature time series monitoring value array sequence; the multi-point temperature time series monitoring value array sequence is analyzed. A temperature balance analysis is performed, and based on the analysis results, K initial channel flow rates and K initial airflow directions are adaptively adjusted to obtain K adjusted channel flow rates and K adjusted airflow directions. K independent airflow channels are then adjusted based on the K adjusted channel flow rates and K adjusted airflow directions until sintering is complete. The airflow in the K independent airflow channels is then reduced according to a preset airflow reduction gradient. After the target LTCC substrate has cooled, a sintered target LTCC substrate is obtained and subjected to surface flatness testing to obtain a surface flatness test result. If the surface flatness test result does not meet the requirements, the K independent airflow channels are adjusted based on the surface flatness test result, and the target LTCC substrate is sintered again. This achieves real-time adjustment and dynamic feedback control of the airflow rate and airflow direction, making the temperature distribution during sintering more uniform, thereby effectively improving the surface flatness of the LTCC substrate. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Attachment Figure 1 The present invention provides a schematic flow chart of a method for controlling the surface flatness of an LTCC substrate.

[0014] Attachment Figure 2The present invention provides a schematic flow chart of determining K initial channel flow rates and K initial airflow directions in a method for controlling the surface flatness of an LTCC substrate provided by an embodiment of the present invention.

[0015] Attachment Figure 3 The present invention provides a schematic flow chart of obtaining K flow rates of adjustment channels and K airflow directions in a method for controlling the surface flatness of an LTCC substrate provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0016] Below in conjunction with specific embodiment, further set forth the present invention.Should be understood that these embodiments are only used to illustrate the present invention and are not used in limiting the scope of the present invention.In addition, should be understood that after reading the content taught by the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms fall equally within the scope limited by the appended claims of the application.

[0017] Reference Figure 1-Figure 3 , as attached Figure 1 As shown, the present application provides a method for controlling the surface flatness of an LTCC substrate, wherein the method comprises:

[0018] S1: Acquire structural information of the signal layer, power layer, and ground layer stacked in the target LTCC substrate to be sintered, and obtain signal line structure information, power grid structure information, and ground grid structure information;

[0019] In one possible embodiment, the screen printer and drying equipment are initially adjusted, specifically by adjusting the pressure of the screen printer to between 0.3±0.02MPa and adjusting the drying equipment to a gradient temperature control range of 70-90°C. A polyimide support sheet is placed under the signal layer, power layer, and bottom layer of the green ceramic sheet (i.e., the unsintered LTCC substrate) to provide stable support for the green ceramic sheet. A diagonal film peeling method is used, that is, the film material is peeled off at a diagonal position to ensure uniform film removal. Silver-palladium slurry is prepared to ensure that its viscosity is between 400,000±50,000 lipos. Optionally, a viscometer can be used to confirm this. A screen is placed on the green ceramic sheet, and the silver-palladium slurry is evenly applied using a screen printer to ensure uniform wiring of the signal layer, ground layer, and power layer, and that the conductive material is printed. This results in a signal layer for transmitting signals, a power layer for forming a power grid, and a ground layer for forming a ground structure.

[0020] Preferably, in the initial stage, the temperature of the drying equipment is adjusted to 70-75°C, and the signal layer after the transmission signal is formed, the power layer forming the power grid, and the ground layer forming the ground structure are preliminarily evaporated to remove the solvent in the slurry and reduce the surface stress of the substrate. Furthermore, in the intermediate stage, the temperature of the drying equipment is adjusted to 75-85°C to ensure further solidification of the slurry and prevent the slurry from cracking or the substrate from warping due to excessive temperature rise. In the final stage, the temperature of the drying equipment is adjusted to 85-90°C to complete the complete drying process. Through screen printing and drying, the processed signal layer, power layer, and ground layer are obtained. The technical effect of laying the material foundation for subsequent low-temperature sintering is achieved.

[0021] That is to say, the target LTCC substrate is a low-temperature co-fired ceramic substrate that has completed silver-palladium slurry printing and drying and is ready for sintering, and is composed of a signal layer, a power layer and a ground layer. The signal layer, the power layer and the ground layer are different functional layers of the target LTCC substrate respectively. The signal layer, the power layer and the ground layer are placed in layers and combined together through a lamination process to facilitate subsequent sintering. Among them, the signal line structure information: refers to the wiring structure information in the signal layer, including the width, spacing and shape of the signal transmission line. The power grid structure information: refers to the power distribution structure of the power layer, including the layout, size and shape of the power grid. The ground grid structure information: refers to the electrical grounding structure of the ground layer, including the distribution, shape and spacing of the ground grid.

[0022] By obtaining structural information about the signal, power, and ground layers, we can better understand the electrical requirements and thermal expansion characteristics of each layer. Based on this structural information, the airflow paths, airflow volume, and airflow direction within the sintering furnace are precisely adjusted to ensure uniform temperature distribution on the substrate during sintering, avoiding warping, cracking, or other defects caused by uneven temperature differences, thereby ensuring the surface flatness of the final product.

[0023] S2: Identifying channel flow rates and airflow directions of K independent airflow channels in the sintering furnace based on the signal line structure information, the power grid structure information, and the ground grid structure information, and determining K initial channel flow rates and K initial airflow directions, where K is a positive integer;

[0024] Further, such as Figure 2 As shown, based on the signal line structure information, the power grid structure information, and the ground grid structure information, the channel flow and airflow direction of K independent airflow channels in the sintering furnace are identified to determine K initial channel flow rates and K initial airflow directions. In this embodiment of the application, step S2 further includes:

[0025] Acquiring an airflow regulation memory library, wherein the airflow regulation memory library includes a set of substrate stacking layer structure prototypes and a set of independent airflow channel regulation schemes;

[0026] Inputting the signal line structure information, the power grid structure information, and the ground grid structure information into the airflow adjustment memory bank, performing similarity matching with the substrate stacking layer structure prototype set respectively, and taking the substrate stacking layer structure prototype corresponding to the maximum matching similarity as the matching substrate stacking layer structure prototype;

[0027] Based on the matching substrate stacking layer structure prototype, the independent airflow channel adjustment scheme set is mapped and extracted to obtain a matching independent airflow channel adjustment scheme, and K initial channel flow rates and K initial airflow directions are determined according to the matching independent airflow channel adjustment scheme.

[0028] In one possible embodiment, the airflow channels within the sintering furnace are used to transmit and regulate airflow. Each airflow channel is independent of the others, and each can adjust the flow rate and direction of the airflow. The operation of each independent airflow channel influences the temperature distribution in various areas within the furnace. Where K represents the number of channels and is a positive integer, indicating that the furnace is equipped with K independent airflow channels.

[0029] Preferably, the channel flow rate refers to the amount of gas flowing in the airflow channel, which is usually measured by the volume of gas passing through per unit time. The regulation of the flow rate directly affects the temperature distribution in the sintering furnace. The airflow direction is the flow direction of the gas in the channel. By controlling the airflow direction during the sintering process, the temperature can be evenly distributed to various areas of the substrate, reducing deformation of the substrate due to local overheating or overcooling. The airflow regulation memory is for storing typical data related to the substrate structure and the airflow regulation scheme, including a set of substrate stacking layer structure prototypes and a set of independent airflow channel regulation schemes. Each substrate stacking layer structure prototype corresponds to an independent airflow channel regulation scheme. In other words, the substrate stacking layer structure prototype set is a set of common situations for different substrate stacking layer structures. The independent airflow channel regulation scheme set is an airflow channel regulation scheme for different substrate stacking layer structures, including settings for channel flow and airflow direction.

[0030] The structure of the target LTCC substrate is determined based on the signal line structure information, the power grid structure information, and the ground grid structure information, and then a comparison and matching is performed in the airflow regulation memory bank to determine the most suitable substrate stacking layer structure prototype, and the channel flow rates and airflow directions of K independent airflow channels when sintering the target LTCC substrate are determined based on the corresponding independent airflow channel regulation scheme, thereby obtaining the K initial channel flow rates and K initial airflow directions.

[0031] Preferably, a cosine similarity function is used to calculate the similarity between the signal line structure information, power grid structure information, and ground line grid structure information and substrate stacking layer structure prototypes in a set of substrate stacking layer structure prototypes. The substrate stacking layer structure prototype corresponding to the maximum matching similarity in the calculation result is used as the matching substrate stacking layer structure prototype. Furthermore, a solution corresponding to the matching substrate stacking layer structure prototype in a set of independent airflow channel adjustment solutions is obtained to obtain the matching independent airflow channel adjustment solution. The channel flow rates and airflow directions of the K independent airflow channels in the matching independent airflow channel adjustment solution are used as the K initial channel flow rates and K initial airflow directions.

[0032] By accurately analyzing the structural information of the target LTCC substrate and combining it with the airflow adjustment scheme in historical data, the initial settings of the airflow channel are dynamically determined, making the airflow adjustment more intelligent and adaptive, achieving the technical effect of avoiding the fixed settings in traditional sintering furnaces and improving the reliability of surface flatness.

[0033] Furthermore, an airflow adjustment memory library is obtained, wherein the airflow adjustment memory library includes a set of substrate stacking layer structure prototypes and a set of independent airflow channel adjustment schemes. In the embodiment of the present application, step S2 further includes:

[0034] Acquire multiple historical substrate stacking layer structure information and corresponding multiple historical substrate surface flatness defect features;

[0035] Performing similarity identification on the multiple historical substrate stacking layer structure information from three dimensions of signal line structure similarity, power grid structure similarity, and ground grid structure similarity, and performing mean calculation on the similarity identification results to determine mean information of the multiple similar historical substrate stacking layer structures, and using the mean information of the multiple similar historical substrate stacking layer structures as substrate stacking layer structure prototypes, and adding them to the substrate stacking layer structure prototype set;

[0036] Mapping and identifying the plurality of historical substrate surface flatness defect features based on the similarity recognition result to determine a plurality of similar historical substrate surface flatness defect feature sets;

[0037] Identifying defect positions and defect degrees for the plurality of similar historical substrate surface flatness defect feature sets to obtain a plurality of similar defect position sets and a plurality of similar defect degree sets;

[0038] Based on the K coverage areas of the K independent airflow channels, the multiple similar defect location sets and the multiple similar defect degree sets are divided respectively, and the airflow direction and channel airflow volume are identified on the division results to determine multiple independent airflow channel adjustment schemes, wherein each independent airflow channel adjustment scheme corresponds to K channel flow prototypes and K airflow direction prototypes;

[0039] The plurality of independent airflow channel adjustment schemes are aggregated into an independent airflow channel adjustment scheme set, and the airflow adjustment memory library is constructed in combination with the substrate stacking layer structure prototype set.

[0040] In one possible embodiment, historical substrate stacking layer structure information refers to the stacking layer structure design information of LTCC substrates sintered in a sintering furnace during historical time periods. This information includes the design layout, thickness, and material of signal layers, power layers, and ground layers. Historical substrate surface flatness defect characteristics refer to surface defect information that occurred during the sintering process of LTCC substrates during historical time periods, including characteristic data of defects such as surface warpage, cracks, and protrusions.

[0041] A predetermined number of historical substrate stacking layer structure information pieces, determined by a person skilled in the art, are randomly extracted from a plurality of historical substrate stacking layer structure information pieces as a plurality of clustered historical substrate stacking layer structure information pieces. The average of the three similarities between any two of the plurality of clustered historical substrate stacking layer structure information pieces, namely, signal line structure similarity, power grid structure similarity, and ground grid structure similarity, is less than or equal to a predetermined similarity threshold. This ensures that the similarity between any two of the plurality of clustered historical substrate stacking layer structure information pieces is low, thereby improving the reliability of subsequent similarity clustering.

[0042] In one embodiment, the similarity averages of any one piece of historical substrate stacking layer structure information among the plurality of historical substrate stacking layer structure information and the plurality of clustered historical substrate stacking layer structure information are calculated based on the three dimensions of signal line structure similarity, power grid structure similarity, and ground grid structure similarity. The historical substrate stacking layer structure information is then added to the clustered historical substrate stacking layer structure information set corresponding to the maximum similarity average, thereby obtaining a similarity identification result. The similarity identification result includes the plurality of clustered historical substrate stacking layer structure information sets.

[0043] Furthermore, the mean values ​​of the multiple clustered historical substrate stacking layer structure information sets are calculated to obtain the mean values ​​of multiple similar historical substrate stacking layer structure information sets, each of which is used as a substrate stacking layer structure prototype and added to the substrate stacking layer structure prototype set. Furthermore, based on the multiple clustered historical substrate stacking layer structure information sets, the multiple historical substrate surface flatness defect features are mapped and clustered to obtain the multiple similar historical substrate surface flatness defect feature sets.

[0044] Data extraction is performed on the plurality of similar historical substrate surface flatness defect feature sets from the two dimensions of defect location and defect degree to obtain the plurality of similar defect location sets and the plurality of similar defect degree sets. Furthermore, K coverage areas of the K independent airflow channels in the sintering furnace are obtained, wherein the K coverage areas include the locations of the coverable areas. Based on the locations of the K coverage areas, the defect locations and defect degrees of the different coverage areas in the plurality of similar defect location sets and the plurality of similar defect degree sets are identified, and the corresponding K channel flow prototypes and K airflow direction prototypes, that is, the corresponding plurality of independent airflow channel adjustment schemes, are determined. After summarizing, the independent airflow channel adjustment scheme set is obtained.

[0045] The independent airflow channel adjustment scheme set and the corresponding substrate stacking layer structure prototypes in the substrate stacking layer structure prototype set are mapped one to one, stored in a database, and the constructed airflow adjustment memory library is obtained.

[0046] Optimization based on historical data can determine the flow rate and airflow direction of the airflow channel according to the design structure and surface defect distribution of different substrates, providing a basis for precise temperature control during the sintering process, achieving the technical effect of improving the surface flatness of the substrate and reducing sintering defects caused by uneven airflow.

[0047] Furthermore, step S2 of the embodiment of the present application further includes:

[0048] Extracting a first similar defect position set and a corresponding first similar defect degree set from the multiple similar defect position sets and the multiple similar defect degree sets;

[0049] Dividing the first similar defect position set and the first similar defect degree set according to the K coverage areas to obtain K regional first similar defect position sets and K regional first similar defect degree sets;

[0050] Determining corresponding K first channel flow prototypes and K first airflow direction prototypes based on the K regions' first similar defect position sets and the K regions' first similar defect degree sets, and constructing a first independent airflow channel adjustment scheme;

[0051] The multiple similar defect position sets and the multiple similar defect degree sets are traversed, and airflow channel adjustment identification is performed in combination with the K coverage areas to determine multiple independent airflow channel adjustment schemes.

[0052] Furthermore, based on the K regions' first similar defect position sets and the K regions' first similar defect degree sets, corresponding K first channel flow prototypes and K first airflow direction prototypes are determined to construct a first independent airflow channel adjustment scheme. In this embodiment, step S2 further includes:

[0053] Traversing the K regions' first similar defect location sets and performing an intra-set union to obtain the K regions' first similar merged defect location sets;

[0054] Using the first similar defect degree sets of the K regions, the first similar merged defect positions of the K regions are marked with defect degrees superimposed to obtain a first similar merged defect position-similar superimposed defect degree set of the K regions;

[0055] The channel flow and airflow direction are identified according to the first similar merged defect position-similar superimposed defect degree sets of the K regions, and K first channel flow prototypes and K first airflow direction prototypes are determined. After summarizing, the first independent airflow channel adjustment scheme is obtained.

[0056] Furthermore, channel flow and airflow direction are identified based on the K regions' first similar merged defect position-similar superimposed defect degree sets, to determine K first channel flow prototypes and K first airflow direction prototypes. In this embodiment of the application, step S2 further includes:

[0057] Respectively extracting the K extracted regions' first similar merged defect position-similar superimposed defect degree sets whose similar superimposed defect degrees are in the top m positions from the K regions' first similar merged defect position-similar superimposed defect degree sets, where m is a positive integer;

[0058] The centers of the defect positions in the first similar merged defect position-similar superimposed defect degree set of the K extraction regions are respectively used as K first airflow direction prototypes;

[0059] K first channel flow prototypes are determined respectively according to the average values ​​of the similar superimposed defect degrees in the first similar merged defect position-similar superimposed defect degree set of the K extraction areas.

[0060] In one possible embodiment, a first similar defect location set and a corresponding first similar defect degree set are extracted from multiple similar defect location sets and multiple similar defect degree sets. To precisely adjust the flow rate and flow direction of each airflow channel, based on the regional positions of K coverage areas and in combination with the defect positions in the first similar defect location set, similar defect locations belonging to each coverage area are divided into corresponding coverage areas, thereby obtaining K regional first similar defect location sets and K regional first similar defect degree sets.

[0061] Preferably, the channel flow and airflow direction of the independent airflow channel in each area are determined based on the distribution of the defect positions and the defect degree in each area. Specifically, the same defect positions in each set of the K area first similar defect position sets are combined, and the repeated defect positions are eliminated to obtain K area first similar merged defect position sets. Then, the first similar defect degree of each area in the K area first similar defect degree sets is superimposed on the corresponding area first similar merged defect position, thereby obtaining K area first similar merged defect position-similar superimposed defect degree sets. Among them, the K area first similar merged defect position-similar superimposed defect degree sets reflect the defect positions and defect degrees within the coverage area of ​​the K independent airflow channels.

[0062] Furthermore, the K regions' first similar merged defect positions - similar superposition defect degree sets are sorted in descending order of similar superposition defect degree, and the regions' first similar merged defect positions - similar superposition defect degrees in the sorting results are added to the K extracted regions' first similar merged defect positions - similar superposition defect degree sets. The centers of the defect positions in the K extracted regions' first similar merged defect positions - similar superposition defect degree sets are preferably obtained by obtaining the defect position coordinates of the K extracted regions' first similar merged defect positions - similar superposition defect degree sets, and then the K defect position coordinate sets are averaged to obtain the defect position center of each region. Furthermore, the directions of the K independent airflow channels to the corresponding defect position centers are taken as the airflow directions to obtain K first airflow direction prototypes.

[0063] The mean values ​​of the similar superimposed defect degrees in the first similar merged defect location-similar superimposed defect degree set of the K extracted regions are calculated to obtain the mean values ​​of the similar superimposed defect degrees of the K regions. A pre-built flow identifier is obtained and used to identify the mean values ​​of the similar superimposed defect degrees of the K regions to determine K channel flow rates. These K channel flow rates are used as the K first channel flow prototypes.

[0064] Preferably, the mean values ​​of similar superimposed defect levels in multiple sample areas and the corresponding channel flow rates of multiple samples are obtained as training data. The framework constructed based on the feedforward neural network is supervised and trained using the training data until the training reaches convergence, thereby obtaining the trained flow identifier. By using the flow identifier, the corresponding channel flow rate can be quickly determined based on the defect level.

[0065] Based on the same principle as the above-mentioned first independent airflow channel adjustment scheme, the multiple similar defect position sets and the multiple similar defect degree sets are combined with the K coverage areas to divide them into regions, and the airflow channel adjustment conditions are identified separately according to the division results to obtain corresponding multiple independent airflow channel adjustment schemes.

[0066] S3: sending the target LTCC substrate into a sintering furnace for debinding and sintering, and using a temperature monitoring array to perform multi-point temperature time series monitoring on the target LTCC substrate according to a preset monitoring window to obtain a multi-point temperature time series monitoring value array sequence;

[0067] In one possible embodiment, a debinding phase is first performed. During this phase, the sintering furnace temperature is kept relatively low (approximately 500°C) to remove organic matter from the substrate. Next, the target LTCC substrate enters a sintering phase within the furnace, where the temperature is gradually increased until it reaches the sintering temperature (typically between 700°C and 850°C). During this process, the ceramic material within the substrate begins to crystallize and densify.

[0068] At the same time, the temperature monitoring array begins performing multi-point temperature time-series monitoring at various locations on the target LTCC substrate. These monitoring points are distributed across different regions of the substrate, collecting real-time temperature data at each location. This temperature data is recorded according to a pre-set monitoring window, generating an array of multi-point temperature time-series monitoring values. This data array provides the basis for subsequent temperature uniformity analysis, accurately reflecting temperature variations across the substrate during the sintering process.

[0069] S4: performing temperature balance analysis on the array sequence of the multi-point temperature time series monitoring values, and adaptively adjusting the K initial channel flow rates and the K initial airflow directions according to the analysis results to obtain K adjusted channel flow rates and K adjusted airflow directions;

[0070] Further, such as Figure 3 As shown, the temperature balance analysis is performed on the array sequence of the multi-point temperature time series monitoring values, and the K initial channel flow rates and the K initial airflow directions are adaptively adjusted according to the analysis results to obtain K adjusted channel flow rates and K adjusted airflow directions. In this embodiment of the application, step S4 further includes:

[0071] Dividing the multi-point temperature time series monitoring value array sequence based on the K coverage areas of the K independent airflow channels to obtain K regional multi-point temperature time series monitoring value array sequences;

[0072] Traversing the K-region multi-point temperature time series monitoring value array sequence to perform feature recognition in two dimensions: temperature distribution balance and temperature difference deviation, and obtaining the K-region temperature distribution balance feature-temperature difference deviation feature sequence;

[0073] Performing feature interaction on the K regional temperature distribution balance feature-temperature difference deviation feature sequences in a time-ordered order to determine K regional temperature distribution balance interaction features-temperature difference deviation interaction features;

[0074] Based on the K regional temperature distribution balance interaction characteristics-temperature difference deviation interaction characteristics, the K initial channel flow rates and the K initial airflow directions are adaptively adjusted to obtain the K adjusted channel flow rates and the K adjusted airflow directions.

[0075] In one possible embodiment, a multi-point temperature time-series monitoring value array sequence refers to temperature data collected at different locations by multiple temperature sensors. This data is arranged in chronological order and records the temperature changes of the substrate during the sintering process. The multi-point temperature time-series monitoring value array sequence is divided according to the positions of K coverage areas. The multi-point temperature time-series monitoring value sequences belonging to the same coverage area are divided into one array, thereby obtaining K-area multi-point temperature time-series monitoring value array sequences.

[0076] Obtain the target temperature within the sintering furnace for each stage, calculate the deviation between it and the K regional multi-point temperature time-series monitoring value array sequences, and compare the calculated result with the target temperature to obtain K regional temperature difference deviation characteristic sequences. Simultaneously, calculate the fluctuation variance of each regional multi-point temperature time-series monitoring value array within the K regional multi-point temperature time-series monitoring value array sequences to obtain K regional temperature distribution balance characteristic sequences. Map and associate the K regional temperature difference deviation characteristic sequences with the K regional temperature distribution balance characteristic sequences to obtain the K regional temperature distribution balance characteristic-temperature difference deviation characteristic sequences.

[0077] Among them, the K-region temperature distribution balance feature-temperature difference deviation feature sequence implies the temperature distribution balance and temperature difference deviation of different regions. It is necessary to capture the correlation between different features between the previous and next times in order from front to back, and gradually superimpose the correlation into the latter feature, so as to obtain the K-region temperature distribution balance interaction feature-temperature difference deviation interaction feature that can reflect the actual situation of each region.

[0078] Acquire multiple sample area temperature distribution balance interaction characteristics - temperature difference deviation interaction characteristics and multiple sample initial channel flow rates and multiple sample initial airflow directions, as well as corresponding multiple sample adjustment channel flow rates and multiple sample adjustment airflow directions as training data, use the training data to supervise the training of the framework constructed based on the feedforward neural network, and adjust the parameters of the framework according to the output deviation during training to obtain a trained airflow adjuster. Use the airflow adjuster to identify the K area temperature distribution balance interaction characteristics - temperature difference deviation interaction characteristics, the K initial channel flow rates and the K initial airflow directions, and obtain the K adjustment channel flow rates and K adjustment airflow directions. By obtaining the K adjustment channel flow rates and K adjustment airflow directions, a technical effect is achieved to pave the way for subsequent ensuring that the temperature of each area can be more uniform and reducing surface defects caused by excessive local temperature differences.

[0079] Furthermore, in a time-ordered manner, feature interaction is performed on the K regional temperature distribution balance feature-temperature difference deviation feature sequences to determine K regional temperature distribution balance interaction features-temperature difference deviation interaction features. In this embodiment of the present application, step S4 further includes:

[0080] Extracting K first-region temperature distribution balance characteristics-temperature difference deviation characteristics and K second-region temperature distribution balance characteristics-temperature difference deviation characteristics from the K-region temperature distribution balance characteristics-temperature difference deviation characteristics sequence in chronological order;

[0081] Using the K first-region temperature distribution balance characteristics-temperature difference deviation characteristics, the K second-region temperature distribution balance characteristics-temperature difference deviation characteristics are subjected to feature interaction to obtain K first-stage regional temperature distribution balance characteristics-temperature difference deviation interaction characteristics;

[0082] By analogy, the K first-stage regional temperature distribution balance characteristics-temperature difference deviation interaction characteristics are used to perform feature interaction on the K regional temperature distribution balance characteristics-temperature difference deviation feature sequences until the end of the sequence is reached, thereby obtaining the K regional temperature distribution balance interaction characteristics-temperature difference deviation interaction characteristics.

[0083] Furthermore, the K first-region temperature distribution balance characteristics-temperature difference deviation characteristics are used to perform feature interaction on the K second-region temperature distribution balance characteristics-temperature difference deviation characteristics to obtain K first-stage regional temperature distribution balance characteristics-temperature difference deviation interaction characteristics. In this embodiment of the application, step S4 further includes:

[0084] Calculating K balance feature similarity sets and K temperature deviation feature similarity sets of the K first region temperature distribution balance features-temperature difference deviation features and the K second region temperature distribution balance features-temperature difference deviation features;

[0085] Normalizing the K balance feature similarity sets and the corresponding K temperature deviation feature similarity sets respectively to construct K feature interaction matrices;

[0086] The K feature interaction matrices are used to enhance the interaction of the K second-region temperature distribution balance features-temperature difference deviation features to obtain the K first-stage regional temperature distribution balance features-temperature difference deviation interaction features.

[0087] In one possible embodiment, K first-region temperature distribution balance characteristics-temperature difference deviation characteristics and K second-region temperature distribution balance characteristics-temperature difference deviation characteristics are extracted in chronological order, and then the similarity of the balance characteristics and the similarity of the temperature deviation characteristics in the K first-region temperature distribution balance characteristics-temperature difference deviation characteristics and the K second-region temperature distribution balance characteristics-temperature difference deviation characteristics are calculated using a cosine similarity calculation formula to obtain K balance feature similarity sets and K temperature deviation feature similarity sets. Furthermore, the K balance feature similarity sets and the corresponding K temperature deviation feature similarity sets are normalized using a softmax function to obtain K balance feature similarity normalized values ​​and corresponding K temperature deviation feature similarity normalized values.

[0088] The K normalized similarity values ​​of the balance features and the corresponding K normalized similarity values ​​of the temperature deviation features are mapped and filled into an initially empty two-dimensional upper triangular matrix to obtain K feature interaction matrices. The K feature interaction matrices reflect the implicit trend correlation between the K first-region temperature distribution balance features and temperature difference deviation features and the K second-region temperature distribution balance features and temperature difference deviation features. Multiple sample feature interaction matrices, multiple sample first-region temperature distribution balance features and temperature difference deviation features, multiple sample second-region temperature distribution balance features and temperature difference deviation features, and corresponding multiple sample stage-region temperature distribution balance features and temperature difference deviation interaction features are obtained as convolution training data. The convolution training data is used to supervise the graph neural network training until the training converges, thereby obtaining the trained graph convolution enhancer. The graph convolution enhancer is used to enhance the interaction of the K feature interaction matrices and the K second-region temperature distribution balance features and temperature difference deviation features to obtain the K first-stage region temperature distribution balance features and temperature difference deviation interaction features.

[0089] Furthermore, based on the above process, the K first-stage regional temperature distribution balance characteristics and temperature difference deviation interaction characteristics are again used to perform feature interaction on the K regional temperature distribution balance characteristics and temperature difference deviation characteristics sequence until the end of the sequence is reached, thereby obtaining the K regional temperature distribution balance interaction characteristics and temperature difference deviation interaction characteristics. By weighting and similarity analysis of temperature characteristics at different times, the airflow volume and airflow direction in each region are optimized, achieving the technical effect of achieving uniform temperature distribution of the target LTCC substrate during the sintering process, improving substrate surface flatness, and reducing defects such as warping and cracking caused by excessive temperature differences.

[0090] S5: The K independent airflow channels are adjusted according to the K adjustment channel flow rates and the K adjustment airflow directions until sintering is completed, and the airflow rates of the K independent airflow channels are reduced according to a preset airflow rate reduction gradient. After the target LTCC substrate is cooled, a sintered target LTCC substrate is obtained, and a surface flatness test is performed on the substrate to obtain a surface flatness test result. If the surface flatness test result does not meet the requirements, the K independent airflow channels are adjusted according to the surface flatness test result, and the target LTCC substrate is sintered again.

[0091] Furthermore, step S5 of the embodiment of the present application further includes:

[0092] Obtaining a defect position set and a defect degree set in the surface flatness inspection result, matching the defect position set with the K coverage areas of the K independent airflow channels to obtain K matching defect position sets, and performing mapping matching on the defect degree set to obtain K matching defect degree sets;

[0093] performing airflow channel adjustment identification based on the K matching defect position sets and the K matching defect degree sets, and determining K independent airflow channel feedback adjustment schemes;

[0094] Parameters of the K independent airflow channels are adjusted according to the feedback adjustment scheme for the K independent airflow channels, and the target LTCC substrate is sintered again in the sintering furnace after the adjustment.

[0095] In one embodiment, first, the K independent airflow channels in the sintering furnace are adjusted according to the K adjustment channel flow rates and K adjustment airflow directions obtained in the previous step. According to the adjusted airflow rate and airflow direction, the airflow distribution in the sintering furnace will more accurately meet the requirements of uniform temperature distribution in each area, ensuring that the substrate will not produce defects such as warping and cracks due to uneven airflow during the sintering process. When the sintering is completed, in order to avoid thermal stress caused by a sudden drop in temperature during the cooling process, it is necessary to gradually reduce the airflow according to the preset airflow reduction gradient. Specifically, the airflow will gradually decrease according to the temperature changes during the sintering process to avoid the impact of rapid cooling on the substrate. During the cooling process, the gradual reduction in airflow helps to reduce surface defects caused by uneven thermal stress.

[0096] After the substrate cools, the surface flatness of the target LTCC substrate is inspected using tools such as a 3D scanner. The tool records the flatness data at each point on the substrate surface and generates a surface flatness test result, including the defect location and degree at each test point. If the surface flatness test result meets the requirements, the production process ends.

[0097] If the surface flatness test result does not meet the requirements, such as if the defectivity exceeds a preset defectivity, for example, if warping occurs at the test point, the warping height is 0.5 mm, and the preset warping height threshold is 0.3 mm, then the surface flatness test result does not meet the requirements. The K coverage areas are matched with the defect locations in the defect location set to determine the defect coverage of each independent airflow channel, thereby obtaining the K matching defect location sets and K matching defectivity sets.

[0098] Preferably, multiple sample matching defect position sets, multiple sample matching defect degree sets and corresponding identified multiple sample independent airflow channel feedback adjustment schemes are obtained as regulator training data, and the regulator training data is used to train a framework constructed based on a feedforward neural network. Multiple sample matching defect position sets and multiple sample matching defect degree sets are input into the framework to obtain corresponding multiple output independent airflow channel feedback adjustment schemes, and the difference between the identified multiple sample independent airflow channel feedback adjustment schemes and the multiple output independent airflow channel feedback adjustment schemes is compared to determine whether the difference is less than the preset difference. If so, the verification is passed and a defect feedback regulator is obtained.

[0099] A defect feedback regulator is used to identify the K matching defect position sets and the K matching defect degree sets, respectively, and determine K independent airflow channel feedback adjustment schemes. Each independent airflow channel feedback adjustment scheme includes feedback adjustment of the channel flow rate and feedback adjustment of the airflow direction. The parameters of the corresponding independent airflow channel are adjusted based on the feedback adjustment of the channel flow rate and the feedback adjustment of the airflow direction, thereby completing the airflow re-routing. The target LTCC substrate is then sintered again in the adjusted sintering furnace. During the sintering process, temperature monitoring and airflow adjustment are performed according to steps S3 and S4 until the surface flatness test results of the target LTCC substrate meet the requirements.

[0100] By precisely controlling the flow rate and direction of the airflow channels, the substrate's temperature distribution is uniform during the sintering process, avoiding defects such as warping or bulging caused by large temperature differences. Furthermore, by gradually reducing the airflow after sintering, thermal stress caused by rapid temperature changes is avoided, further optimizing the substrate's surface flatness. By implementing surface flatness detection and early warning mechanisms, the sintering process can be monitored in real time, allowing for timely adjustments to the control process to ensure that the final LTCC substrate meets the requirements.

[0101] In summary, the embodiments of the present application have at least the following technical effects:

[0102] 1. This application identifies the flow rates and airflow directions of K independent airflow channels within a sintering furnace based on signal line structure information, power grid structure information, and ground grid structure information. This allows precise adjustment of the flow rates and directions of the airflow channels based on the design characteristics and defect distribution of different substrates. This process ensures uniform temperature distribution across all areas during sintering, avoiding defects such as substrate surface warping and cracking caused by uneven airflow or large temperature differences.

[0103] 2. This application uses a temperature monitoring array to monitor multi-point temperature time series, collecting real-time temperature data from each region of the substrate and analyzing this data for temperature uniformity. Based on the analysis results, adaptive adjustments are made to the temperature distribution of different regions to ensure uniform temperature distribution across the substrate, avoiding thermal stress caused by excessive or uneven temperature differences, thereby further ensuring substrate surface flatness.

[0104] It should be noted that the order in which the embodiments of the present application are presented is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. Furthermore, the foregoing descriptions of specific embodiments of this specification are provided. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order or sequential sequence shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0105] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should be included in the scope of protection of the present application.

[0106] This specification and drawings are merely illustrative of the present application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Obviously, those skilled in the art may make various modifications and variations to this application without departing from the scope of this application. Thus, this application is intended to include such modifications and variations as fall within the scope of this application and its equivalents.

Claims

1. A method for controlling the surface flatness of an LTCC substrate, characterized in that: The method comprises: Acquire structural information of the signal layer, power layer, and ground layer stacked in the target LTCC substrate to be sintered, and obtain signal line structure information, power grid structure information, and ground grid structure information; Based on the signal line structure information, the power grid structure information, and the ground grid structure information, identifying channel flow rates and airflow directions of K independent airflow channels in the sintering furnace, and determining K initial channel flow rates and K initial airflow directions, where K is a positive integer; The target LTCC substrate is placed in a sintering furnace for debinding and sintering, and a temperature monitoring array is used to perform multi-point temperature time series monitoring on the target LTCC substrate according to a preset monitoring window to obtain a multi-point temperature time series monitoring value array sequence; Performing temperature balance analysis on the array sequence of the multi-point temperature time series monitoring values, and adjusting the K initial channel flow rates and the K initial airflow directions according to the analysis results to obtain K adjusted channel flow rates and K adjusted airflow directions; The K independent airflow channels are adjusted according to the K adjustment channel flow rates and the K adjustment airflow directions until sintering is completed, and the airflow rates of the K independent airflow channels are reduced according to a preset airflow rate reduction gradient. After the target LTCC substrate is cooled, a sintered target LTCC substrate is obtained, and a surface flatness test is performed to obtain a surface flatness test result. If the surface flatness test result does not meet the requirements, the K independent airflow channels are adjusted according to the surface flatness test result, and the target LTCC substrate is sintered again.

2. A method for controlling the surface flatness of an LTCC substrate according to claim 1, characterized in that: Based on the signal line structure information, the power grid structure information, and the ground grid structure information, K independent airflow channels in the sintering furnace are identified for channel flow rates and airflow directions, and K initial channel flow rates and K initial airflow directions are determined, including: Acquiring an airflow regulation memory library, wherein the airflow regulation memory library includes a set of substrate stacking layer structure prototypes and a set of independent airflow channel regulation schemes; Inputting the signal line structure information, the power grid structure information, and the ground grid structure information into the airflow adjustment memory bank, performing similarity matching with the substrate stacking layer structure prototype set respectively, and taking the substrate stacking layer structure prototype corresponding to the maximum matching similarity as the matching substrate stacking layer structure prototype; Based on the matching substrate stacking layer structure prototype, the independent airflow channel adjustment scheme set is mapped and extracted to obtain the matching independent airflow channel adjustment scheme, and K initial channel flow rates and K initial airflow directions are determined according to the matching independent airflow channel adjustment scheme.

3. The method for controlling the surface flatness of an LTCC substrate according to claim 2, wherein: Acquire an airflow regulation memory library, wherein the airflow regulation memory library includes a set of substrate stacking layer structure prototypes and a set of independent airflow channel regulation schemes, including: Acquire multiple historical substrate stacking layer structure information and corresponding multiple historical substrate surface flatness defect features; Performing similarity identification on the multiple historical substrate stacking layer structure information from three dimensions of signal line structure similarity, power grid structure similarity, and ground grid structure similarity, and performing mean calculation on the similarity identification results to determine mean information of the multiple similar historical substrate stacking layer structures, using the mean information of the multiple similar historical substrate stacking layer structures as substrate stacking layer structure prototypes, and adding them to generate the substrate stacking layer structure prototype set; Based on the similarity recognition result, mapping and recognizing the plurality of historical substrate surface flatness defect features to determine a plurality of similar historical substrate surface flatness defect feature sets; Identifying defect locations and defect levels for the plurality of similar historical substrate surface flatness defect feature sets to obtain a plurality of similar defect location sets and a plurality of similar defect level sets; Based on the K coverage areas of the K independent airflow channels, the multiple similar defect location sets and the multiple similar defect degree sets are divided respectively, and the airflow direction and channel airflow volume are identified on the division results to determine multiple independent airflow channel adjustment schemes, wherein each independent airflow channel adjustment scheme corresponds to K channel flow prototypes and K airflow direction prototypes; The plurality of independent airflow channel adjustment schemes are aggregated into an independent airflow channel adjustment scheme set, and combined with the substrate stacking layer structure prototype set to construct the airflow adjustment memory library.

4. A method for controlling the surface flatness of an LTCC substrate according to claim 3, characterized in that: include: Extracting a first similar defect position set and a corresponding first similar defect degree set from the multiple similar defect position sets and the multiple similar defect degree sets; Dividing the first similar defect position set and the first similar defect degree set according to the K coverage areas to obtain K regional first similar defect position sets and K regional first similar defect degree sets; Based on the K regions' first similar defect position sets and the K regions' first similar defect degree sets, determining corresponding K first channel flow prototypes and K first airflow direction prototypes, and constructing a first independent airflow channel adjustment scheme; The multiple similar defect position sets and the multiple similar defect degree sets are traversed, and airflow channel adjustment identification is performed in combination with the K coverage areas to determine multiple independent airflow channel adjustment schemes.

5. The method for controlling the surface flatness of an LTCC substrate according to claim 4, wherein: Determining corresponding K first channel flow prototypes and K first airflow direction prototypes based on the K regions' first similar defect position sets and the K regions' first similar defect degree sets, and constructing a first independent airflow channel adjustment scheme, including: Traversing the K regions' first similar defect location sets and performing an intra-set union to obtain the K regions' first similar merged defect location sets; Using the first similar defect degree sets of the K regions, superimposing defect degree marks on the first similar merged defect positions of the K regions to obtain a first similar merged defect position-similar superimposed defect degree set of the K regions; The channel flow and airflow direction are identified based on the first similar merged defect position-similar superimposed defect degree sets of the K regions, and K first channel flow prototypes and K first airflow direction prototypes are determined. After summarizing, the first independent airflow channel adjustment scheme is obtained.

6. A method for controlling the surface flatness of an LTCC substrate according to claim 5, characterized in that: According to the K regions' first similar merged defect position-similar superimposed defect degree sets, channel flow and airflow direction identification are performed to determine K first channel flow prototypes and K first airflow direction prototypes, including: Extracting the first similar merged defect position-similar superimposed defect degree sets of the K extracted regions whose similar superimposed defect degrees are in the top m positions, respectively, from the K regions' first similar merged defect position-similar superimposed defect degree sets, where m is a positive integer; The centers of the defect positions in the first similar merged defect positions-similar superimposed defect degree sets of the K extraction regions are respectively used as K first airflow direction prototypes; K first channel flow prototypes are determined respectively according to the average values ​​of the similar superimposed defect degrees in the first similar merged defect position-similar superimposed defect degree set of the K extraction areas.

7. The method for controlling the surface flatness of an LTCC substrate according to claim 1, wherein: Performing temperature balance analysis on the multi-point temperature time series monitoring value array sequence, and adaptively adjusting the K initial channel flow rates and the K initial airflow directions according to the analysis results to obtain K adjusted channel flow rates and K adjusted airflow directions, including: Dividing the multi-point temperature time series monitoring value array sequence based on the K coverage areas of the K independent airflow channels to obtain K regional multi-point temperature time series monitoring value array sequences; Traversing the K-region multi-point temperature time series monitoring value array sequence, performing feature recognition in two dimensions of temperature distribution balance and temperature difference deviation, and obtaining the K-region temperature distribution balance feature-temperature difference deviation feature sequence; Performing feature interaction on the K regional temperature distribution balance feature-temperature difference deviation feature sequences in a time-ordered order to determine K regional temperature distribution balance interaction features-temperature difference deviation interaction features; Based on the K regional temperature distribution balance interaction characteristics-temperature difference deviation interaction characteristics, the K initial channel flow rates and the K initial airflow directions are adjusted to obtain the K adjusted channel flow rates and the K adjusted airflow directions.

8. The method for controlling the surface flatness of an LTCC substrate according to claim 7, wherein: Performing feature interaction on the K regional temperature distribution balance feature-temperature difference deviation feature sequences in a time-ordered manner to determine K regional temperature distribution balance interaction features-temperature difference deviation interaction features, including: Extracting K first-region temperature distribution balance characteristics-temperature difference deviation characteristics and K second-region temperature distribution balance characteristics-temperature difference deviation characteristics from the K-region temperature distribution balance characteristics-temperature difference deviation characteristics sequence in chronological order; Using the K first-region temperature distribution balance characteristics-temperature difference deviation characteristics, the K second-region temperature distribution balance characteristics-temperature difference deviation characteristics are subjected to feature interaction to obtain K first-stage regional temperature distribution balance characteristics-temperature difference deviation interaction characteristics; By analogy, the K first-stage regional temperature distribution balance characteristics-temperature difference deviation interaction characteristics are used to perform feature interaction on the K regional temperature distribution balance characteristics-temperature difference deviation feature sequences until the end of the sequence is reached, thereby obtaining the K regional temperature distribution balance interaction characteristics-temperature difference deviation interaction characteristics.

9. A method for controlling the surface flatness of an LTCC substrate according to claim 8, characterized in that: The K first-region temperature distribution balance characteristics-temperature difference deviation characteristics are used to perform feature interaction on the K second-region temperature distribution balance characteristics-temperature difference deviation characteristics to obtain K first-stage regional temperature distribution balance characteristics-temperature difference deviation interaction features, including: Calculating K balance feature similarity sets and K temperature deviation feature similarity sets of the K first region temperature distribution balance features-temperature difference deviation features and the K second region temperature distribution balance features-temperature difference deviation features; Normalizing the K balance feature similarity sets and the corresponding K temperature deviation feature similarity sets respectively to construct K feature interaction matrices; The K feature interaction matrices are used to enhance the interaction of the K second-region temperature distribution balance features-temperature difference deviation features to obtain the K first-stage regional temperature distribution balance features-temperature difference deviation interaction features.

10. The method for controlling the surface flatness of an LTCC substrate according to claim 1, wherein: include: Obtaining a defect position set and a defect degree set in the surface flatness inspection result, matching the defect position set with the K coverage areas of the K independent airflow channels to obtain K matching defect position sets, and performing mapping matching on the defect degree set to obtain K matching defect degree sets; Performing airflow channel adjustment identification based on the K matching defect position sets and the K matching defect degree sets, and determining K independent airflow channel feedback adjustment schemes; Parameters of the K independent airflow channels are adjusted according to the feedback adjustment scheme for the K independent airflow channels, and the target LTCC substrate is sintered again in the sintering furnace after the adjustment.

Images