Defect detection method and system for evaporation OLED (Organic Light Emitting Diode)

Abstract

The invention discloses a defect detection method and system for evaporation OLED, and relates to the technical field of evaporation OLED defect detection.The defect detection method comprises the following steps that a clean substrate is selected as an evaporation foundation, a mask with a hollow pattern is installed on the substrate, and the mask and the substrate are installed in an aligned mode; and preparing a piezoelectric sensing film on the surface of the mask through a physical vapor deposition method. According to the invention, the piezoelectric sensing film is arranged between the OLED substrate and the mask, the defect of the mask is identified by detecting the pressure change of the film, the defect of the mask can be found and repaired before evaporation, adverse effects on the quality of an OLED panel are avoided, the product yield is improved, the whole surface of the mask can be covered, no-missing detection is realized, and the detection efficiency is improved. And a defect detection model is constructed by utilizing a machine learning algorithm, defects with different severity degrees are identified, high-accuracy defect identification is realized through accurate feature extraction and intelligent analysis of the machine learning model, and the quality reliability of OLED products is remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of vapor-deposited OLED defect detection, and in particular to a defect detection method and system for vapor-deposited OLED. Background Art

[0002] In the manufacturing process of OLED display panels, the evaporation process is one of the key steps. The evaporation process involves evaporating organic materials or metal materials at high temperatures and depositing them on the surface of the substrate to form the required thin film layer. With the continuous development of OLED display technology, consumers have higher and higher requirements for the quality of display panels. Therefore, it is necessary to ensure that each OLED display panel meets the quality requirements during production to enhance the overall competitiveness of the product.

[0003] For example, Chinese patent publication number CN106350767A discloses an OLED substrate evaporation structure and an OLED mask defect detection method, which comprises: a substrate; a mask attached to the surface of the substrate; and a piezoelectric sensing film disposed between the substrate and the mask. The OLED substrate evaporation structure of the present invention can detect defects in the mask before evaporating the substrate. This early detection method can avoid discovering problems with the mask after the finished product is produced, thereby improving the stability and yield of the product.

[0004] In the prior art, a piezoelectric sensing film is arranged between the substrate and the mask, and a piezoelectric signal conversion device is used to detect the pressure signal to identify whether the mask has defects. However, there is a coupling relationship between the piezoelectric signal and multiple factors such as mask defects, substrate characteristics, and environmental factors, which will cause the sensitivity and stability of the piezoelectric signal to be affected to a certain extent, and it is difficult to perform coverage analysis on the severity of OLED defects, thereby affecting the efficiency of detection. Therefore, a defect detection method and system for vapor-deposited OLEDs are proposed. Summary of the invention

[0005] The present invention aims to provide a defect detection method and system for vapor-deposited OLEDs to solve the problems raised in the above-mentioned background technology.

[0006] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0007] In a first aspect, a defect detection method for an evaporation-deposited OLED comprises the following steps:

[0008] Step 1, select a clean substrate as a vapor deposition base to ensure that there is no dust or contaminants, and install a mask with a hollow pattern on the substrate so that the mask and the substrate are aligned and installed;

[0009] Step 2, preparing a piezoelectric sensing film on the surface of the mask by a physical vapor deposition method, the material of the piezoelectric sensing film is aluminum nitride to ensure that it has good piezoelectric properties, and connecting a piezoelectric signal conversion device to the piezoelectric sensing film, debugging the signal conversion device to ensure that the piezoelectric signal can be accurately converted into an electrical signal for subsequent analysis, so as to convert the pressure signal detected by the piezoelectric sensing film into an electrical signal for subsequent analysis;

[0010] Step 3, collecting the piezoelectric signal of the piezoelectric sensing film based on the piezoelectric signal conversion device, filtering and amplifying the piezoelectric signal to improve the signal-to-noise ratio and sensitivity, comprehensively analyzing the influence of substrate characteristics and environmental factors on the piezoelectric signal, and correcting and compensating the piezoelectric signal;

[0011] Step 4: Apply machine learning to analyze the piezoelectric signal data, analyze the signal changes caused by mask defects, substrate characteristic abnormalities, and environmental factors, obtain the OLED defect detection coefficient, and distinguish the severity and defect level of the OLED defect;

[0012] Step 5: Based on the test results, generate a test report for the detected defects and feed it back to the upstream process to optimize the evaporation process parameters, reduce defects and improve product yield.

[0013] A further improvement of the technical solution of the present invention is that in step 1, the process of aligning and installing the mask and the substrate is as follows:

[0014] Step 101, cleaning and drying the substrate, and using a laser scanner to check the surface of the substrate to ensure that the substrate surface is flat, without scratches or depressions, and using ultrapure water or other cleaning agents to thoroughly clean the substrate surface to remove grease, dust and other pollutants. After cleaning, the substrate is placed in a dust-free environment for drying, and is dried in a drying oven using nitrogen to completely remove moisture from the substrate surface, ensuring that the substrate is dry and free of water stains to prevent dust from being adsorbed again. Before installing the mask, use a laser scanner to check the substrate surface in a clean environment;

[0015] Step 102, selecting a metal mask with a precise hollow pattern to match the size of the substrate, and cleaning the mask to ensure that there are no pollutants on the mask surface to avoid contamination of the substrate during the evaporation process;

[0016] Step 103, using a high-precision positioning fixture to place the mask above the substrate, align it with the substrate, and adjust the position of the fixture so that the hollow pattern on the mask completely corresponds to the predetermined evaporation area on the substrate without offset or misalignment. After the mask is installed, check the alignment accuracy of the mask and the substrate;

[0017] Step 104, applying pre-pressure to the mask and the substrate to fix the mask on the substrate so that the mask and the substrate are closely fitted to reduce gas or material penetration during the evaporation process;

[0018] Step 105 , recording the installation parameters of the mask and substrate, including alignment accuracy and pre-pressure, and placing the mask and substrate in a vapor deposition chamber to ensure that the cleanliness and vacuum of the chamber meet the requirements of the vapor deposition process.

[0019] A further improvement of the technical solution of the present invention is that in step 2, the process of connecting the piezoelectric signal conversion device to the piezoelectric sensing film is as follows:

[0020] Step 201, preparing the aluminum nitride piezoelectric sensing film, selecting high-quality aluminum nitride powder as a deposition source, and preparing PVD equipment;

[0021] Step 202, in a PVD device, heating the aluminum nitride powder to its evaporation temperature, so that the aluminum nitride powder is converted into a gaseous state by evaporation, and using the gas flow of the vacuum system of the PVD device to transport the aluminum nitride molecules into the evaporation chamber, and the vaporized aluminum nitride molecules are deposited on the surface of the substrate to form a piezoelectric sensing film;

[0022] Step 203, connecting a piezoelectric signal conversion device to a piezoelectric sensing film made of aluminum nitride, and calibrating the piezoelectric signal conversion device to ensure that it can accurately respond to the piezoelectric signal generated by the piezoelectric film, adjusting the sensitivity of the piezoelectric signal conversion device, and capturing weak piezoelectric signals, wherein the piezoelectric signal conversion device is a piezoelectric sensor;

[0023] Step 204, applying different pressures to the piezoelectric sensing film, observing the change of the electrical signal output by the piezoelectric signal conversion device, ensuring that the film can accurately convert the pressure signal into the electrical signal, analyzing the converted electrical signal, and evaluating the stability of the piezoelectric sensing film.

[0024] A further improvement of the technical solution of the present invention is that the process of evaluating the stability of the piezoelectric sensing film is:

[0025] Step 2041, applying different pressures to the piezoelectric sensing film, and recording the value of each pressure value;

[0026] Step 2042, corresponding to each applied pressure, measuring and recording the electrical signal output by the piezoelectric sensor, pairing and marking the acquired pressure value with the corresponding electrical signal to obtain a corresponding data pair;

[0027] Step 2043, comprehensively analyze all the data pairs of pressure values ​​and electrical signals, obtain the piezoelectric evaluation index, check the linear relationship between the pressure value and the electrical signal within a certain pressure range, evaluate the response of the piezoelectric sensing film to pressure changes, and then analyze the stability of the piezoelectric sensing film, check whether it shows a good linear relationship within a certain pressure range, and high linearity indicates that the film responds consistently to pressure changes. The expression of the piezoelectric evaluation index is:

[0028]

[0029] Where E(P) is the piezoelectric evaluation index, which represents the electrical signal output by the piezoelectric sensor under pressure P. i is the pressure value applied for the ith time, P r is the reference pressure value, which is the minimum pressure value at which the film starts to respond, n is the total number of piezoelectric signal data points, a, b, c are model parameters, P max is the maximum value of the applied pressure, It means that when x approaches P max The limit of time.

[0030] A further improvement of the technical solution of the present invention is that in step 3, the correction and compensation process of the piezoelectric signal is:

[0031] Step 301, using a piezoelectric signal conversion device to collect the original piezoelectric signal generated by the piezoelectric sensing film due to the force, filtering the collected original piezoelectric signal, and amplifying the filtered signal through an amplifier to improve the signal-to-noise ratio of the signal and facilitate subsequent measurement and analysis. The filtering process is used to remove noise and unnecessary frequency components;

[0032] Step 302, analyzing the amplified signal to identify the characteristics of the signal, including amplitude and frequency;

[0033] Step 303, analyzing the influence of environmental factors such as humidity and temperature on the piezoelectric signal, obtaining an environmental influence function, and performing environmental factor correction, then compensating the piezoelectric signal according to the material and thickness characteristics of the substrate, establishing a relationship model between the substrate characteristics and the piezoelectric signal, obtaining a substrate influence function, and eliminating the influence of the substrate characteristics on the measurement result;

[0034] Step 304, comprehensively analyzing the analysis results of substrate characteristics and environmental factors, building a signal correction and compensation model in combination with the environmental influence function and the substrate influence function, obtaining a signal correction and compensation index, and correcting and compensating the piezoelectric signal to obtain a more accurate measurement result.

[0035] A further improvement of the technical solution of the present invention is that the calculation formula of the signal correction and compensation index is:

[0036]

[0037] E env (T, H) = e qT -sin(wH);

[0038]

[0039] Among them, E co is the signal correction and compensation index, which is used to calculate the corrected and compensated piezoelectric signal, E a is the amplified piezoelectric signal, E env (T, H) is the environmental influence function, which indicates the influence of environmental factors (temperature T and humidity H) on the signal. B(λ, μ) is the substrate influence function, which indicates the influence of substrate characteristics. λ and μ represent the material parameters and thickness of the substrate, respectively. n is the total number of piezoelectric signal data points, which is used to calculate the volatility of the signal. E j is the piezoelectric signal value measured at the jth time, is the average value of all piezoelectric signals, q and w are environmental parameters.

[0040] A further improvement of the technical solution of the present invention is that in step 4, the process of obtaining the OLED defect detection coefficient is:

[0041] Step 401, collecting piezoelectric signal data of different types of OLED defects, including data of normal and various defect conditions, analyzing the severity caused by each defect condition, and performing pre-processing operations such as cleaning, denoising, and standardization on the collected data to ensure data quality;

[0042] Step 402, extracting features from the preprocessed piezoelectric signal data, including amplitude, frequency, energy, skewness, and kurtosis, and dividing the data set containing the preprocessed piezoelectric signal data into a training set, a validation set, and a test set;

[0043] Step 403, using the training set data to train the neural network model, constructing a defect detection model, and using the validation set and the test set to evaluate the performance of the defect detection model, comprehensively analyzing the effects of mask defects, substrate characteristic abnormalities, and environmental factors on signal changes, and obtaining the OLED defect recognition coefficient to perform defect recognition on the piezoelectric signal data;

[0044] Step 404, based on the severity caused by the defect situation and the OLED defect recognition coefficient, distinguish the severity of the OLED defect detection, divide the defect levels, and set warning thresholds corresponding to the defect levels, the defect levels are respectively a slight defect level, a medium defect level, and a severe defect level;

[0045] Step 405 , deploy the trained defect detection model to the production environment, perform real-time analysis on the new piezoelectric signal data, calculate the OLED defect recognition coefficient and use it to identify OLED defects and determine the defect level.

[0046] A further improvement of the technical solution of the present invention is that the calculation formula of the OLED defect recognition coefficient is:

[0047]

[0048] Where D is the OLED defect recognition coefficient, which is used to comprehensively evaluate the abnormality of the piezoelectric signal data to identify OLED defects, n is the total number of piezoelectric signal data points, and E is the total number of piezoelectric signal data points. k is the piezoelectric signal value of the kth data point, is the sum of the squares of all piezoelectric signal values, max(E k ) is the maximum value of all piezoelectric signal values, min(E k ) is the minimum value of all piezoelectric signal values, σ is the standard deviation of the piezoelectric signal data, indicating the dispersion of the signal value, is the mean error of the piezoelectric signal data, S is the skewness of the signal, which indicates the degree of skewness of the signal distribution, and K is the kurtosis of the signal, which indicates the degree of sharpness or flatness of the signal distribution;

[0049] A plurality of the defect levels correspond to a plurality of the warning thresholds, wherein the warning thresholds include an upper threshold and a lower threshold;

[0050] The multiple defect levels and the multiple warning thresholds satisfy the following relationship:

[0051] Slight defect level D <D T1 ;

[0052] Medium defect level D T1 ≤D <D T2 ;

[0053] Severe defect level D≥D T2 ;

[0054] Where D is the OLED defect recognition coefficient, D T1 is the lower threshold corresponding to the medium defect level and the upper threshold corresponding to the slight defect level, D T2 It is the lower threshold corresponding to the severe defect level and the upper threshold corresponding to the medium defect level.

[0055] A further improvement of the technical solution of the present invention is that in step 5, the process of optimizing the evaporation process parameters is:

[0056] Step 501, using the deployed defect detection model to analyze the piezoelectric signal data collected in real time, calculate the OLED defect recognition coefficient, and determine the level and severity of each defect according to a preset threshold;

[0057] Step 502: Generate a test report based on the defect detection results, and perform statistics and analysis on the data in the test report to identify the pattern and trend of defect generation. The test report contains basic information such as defect ID, level, severity, and detection time.

[0058] Step 503, feeding back the analysis results to the upstream evaporation process link, using the information in the test report to trace the process steps and conditions that caused the defects, find the root cause of the problem, and evaluate and adjust the evaporation process parameters based on the feedback information;

[0059] Step 504, applying the verified effective improvement measures to actual production, continuously monitoring the improved production data, and optimizing the process parameters based on the monitoring results and new feedback information to form a closed loop of continuous improvement.

[0060] In a second aspect, a defect detection system for vapor-deposited OLED is used to implement a defect detection method for vapor-deposited OLED, including a defect detection platform, wherein the defect detection platform is communicatively connected to a data acquisition module, a feature extraction module, a data management module, a model training module, a defect detection module, and a report generation module, wherein electrical signals are connected between the modules;

[0061] The data acquisition module is used to collect the original piezoelectric signal data detected by the piezoelectric sensor, provide basic data for subsequent data processing and analysis, and perform pre-processing operations such as denoising, filtering, and standardization on the original piezoelectric signal data to improve data quality, reduce the impact of noise and outliers, and provide accurate data for feature extraction;

[0062] The feature extraction module is used to extract key features from the preprocessed piezoelectric signal data, including amplitude, frequency, energy, skewness, and kurtosis, and convert the original signal into numerical features that can be used in the machine learning model;

[0063] The data management module is used to manage the division, storage and access of data sets to ensure the correct use of data during training, verification and testing stages;

[0064] The model training module is used to train the defect detection model using the training set data and evaluate the defect detection model;

[0065] The defect detection module comprehensively analyzes the influence of mask defects, substrate characteristic abnormalities and environmental factors on signal changes based on the defect detection model, obtains the OLED defect recognition coefficient, performs defect recognition on the piezoelectric signal data, and recognizes and classifies the defects, distinguishes defects of different severity, and determines the defect level;

[0066] The report generation module is used to generate a test report containing defect ID, level, severity, and detection time, provide detailed defect information to facilitate process adjustment and quality control, and feed back the test results to the upstream process links to guide the optimization of process parameters and reduce the occurrence of defects.

[0067] Due to the adoption of the above technical solution, the present invention has the following technical advances compared with the prior art:

[0068] 1. The present invention provides a defect detection method and system for vapor-deposited OLEDs. A piezoelectric sensing film is arranged between the OLED substrate and the mask. The defects of the mask are identified by detecting the pressure changes on the film. The mask defects can be found and repaired before vapor deposition, thereby avoiding adverse effects on the quality of the OLED panel and improving the product yield. The entire mask surface can be covered to achieve zero-omission detection. A defect detection model is constructed using a machine learning algorithm to identify defects of different severity. High-accuracy defect identification is achieved through precise feature extraction and intelligent analysis of the machine learning model, thereby significantly improving the quality reliability of OLED products.

[0069] 2. The present invention provides a defect detection method and system for vapor-deposited OLEDs. The method and system combine piezoelectric signal data and temperature and humidity data during the production process, conduct comprehensive analysis and judgment through multi-source data, comprehensively evaluate the quality of OLED panels, improve the comprehensiveness and accuracy of detection, and then adjust production parameters and process flow according to the detection results, optimize the production process, improve product quality and production efficiency. On the basis of defect detection, the report generation and feedback function are integrated to automatically generate a detailed detection report, including defect ID, level, severity, and detection time information, and automatically send it to relevant personnel, thereby realizing the rapid transmission and sharing of detection results, improving the efficiency and accuracy of information communication, and relevant personnel can quickly understand the defect situation and take corresponding treatment measures to reduce production delays and losses. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0071] Figure 1 is a flow chart of the method of the present invention;

[0072] Figure 2 The figure is a flow chart of the correction and compensation of the piezoelectric signal of the present invention;

[0073] Figure 3 This is a flow chart for obtaining OLED defect detection coefficients of the present invention;

[0074] Figure 4 It is a module diagram of the present invention. DETAILED DESCRIPTION

[0075] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0076] Embodiment 1, as Figure 1 , Figure 2 As shown, the present invention provides a defect detection method for evaporated OLED, comprising the following steps:

[0077] Step 1: Select a clean substrate as the basis for evaporation to ensure that there is no dust or contaminants, and install a mask with a hollow pattern on the substrate, align the mask with the substrate, clean and dry the substrate, and use a laser scanner to check the surface of the substrate to ensure that the surface of the substrate is flat, without scratches or depressions. Use ultrapure water or other detergents to thoroughly clean the surface of the substrate to remove grease, dust and other contaminants. After cleaning, place the substrate in a dust-free environment to dry, and use nitrogen to blow dry in a drying oven to completely remove moisture from the surface of the substrate to ensure that the substrate is dry and free of water stains to prevent dust from being adsorbed again. Before installing the mask, use a laser scanner to check the surface of the substrate in a clean environment and select a metal mask with a precise hollow pattern. Make it fit the size of the substrate, clean the mask, ensure that there are no pollutants on the mask surface, avoid contamination of the substrate during the evaporation process, use a high-precision positioning fixture to place the mask above the substrate, align it with the substrate, adjust the position of the fixture, make the hollow pattern on the mask completely correspond to the predetermined evaporation area on the substrate, without offset or misalignment, after the mask is installed, check the alignment accuracy of the mask and the substrate, apply pre-pressure on the mask and the substrate, fix the mask on the substrate, make the mask and the substrate fit tightly, reduce the gas or material penetration during the evaporation process, record the installation parameters of the mask and the substrate, including the alignment accuracy and pre-pressure, and place the mask and the substrate in the evaporation chamber to ensure that the cleanliness and vacuum of the chamber meet the requirements of the evaporation process;

[0078] Step 2, prepare a piezoelectric sensing film on the surface of the mask by a physical vapor deposition method. The material of the piezoelectric sensing film is aluminum nitride to ensure that it has good piezoelectric properties, and connect the piezoelectric signal conversion device to the piezoelectric sensing film, debug the signal conversion device to ensure that the piezoelectric signal can be accurately converted into an electrical signal for subsequent analysis, so as to convert the pressure signal detected by the piezoelectric sensing film into an electrical signal for subsequent analysis, prepare the aluminum nitride piezoelectric sensing film, select high-quality aluminum nitride powder as a deposition source, and prepare PVD equipment. In the PVD equipment, heat the aluminum nitride powder to its evaporation temperature so that the aluminum nitride powder is converted into a gaseous state by evaporation, and use the vacuum system of the PVD equipment to The gas flow transports the aluminum nitride molecules into the evaporation chamber, and the vaporized aluminum nitride molecules are deposited on the surface of the substrate to form a piezoelectric sensing film. The piezoelectric signal conversion device is connected to the piezoelectric sensing film made of aluminum nitride, and the piezoelectric signal conversion device is calibrated to ensure that it can accurately respond to the piezoelectric signal generated by the piezoelectric film, adjust the sensitivity of the piezoelectric signal conversion device, and capture weak piezoelectric signals. The piezoelectric signal conversion device is a piezoelectric sensor. Different pressures are applied to the piezoelectric sensing film, and the changes in the electrical signals output by the piezoelectric signal conversion device are observed to ensure that the film can accurately convert the pressure signal into the electrical signal. The converted electrical signal is analyzed to evaluate the stability of the piezoelectric sensing film.

[0079] Further, the process of evaluating the stability of the piezoelectric sensing film is:

[0080] Apply different pressures to the piezoelectric sensing film and record the value of each pressure value. For each applied pressure, measure and record the electrical signal output by the piezoelectric sensor. Pair the obtained pressure value with its corresponding electrical signal to obtain the corresponding data pair. Comprehensively analyze all the data pairs of pressure values ​​and electrical signals to obtain the piezoelectric evaluation index, check the linear relationship between the pressure value and the electrical signal within a certain pressure range, evaluate the response of the piezoelectric sensing film to pressure changes, and then analyze the stability of the piezoelectric sensing film. Check whether it shows a good linear relationship within a certain pressure range. High linearity indicates that the film responds consistently to pressure changes. The expression of the piezoelectric evaluation index is:

[0081]

[0082] Where E(P) is the piezoelectric evaluation index, which represents the electrical signal output by the piezoelectric sensor under pressure P. i is the pressure value applied for the ith time, P r is the reference pressure value, which is the minimum pressure value at which the film starts to respond, n is the total number of piezoelectric signal data points, a, b, c are model parameters, P max is the maximum value of the applied pressure, It means that when x approaches P max It should be noted that a is a proportional constant, which indicates the contribution of pressure change to the change of electrical signal, and b is a constant related to the limit function, which is used to describe the change of electrical signal when the pressure approaches the maximum value. As the pressure P approaches P max , the rate of change of E(P) will decrease, c is a constant term, which represents the basic electrical signal level when no pressure is applied, and its value is positive or zero, depending on the zero point calibration of the sensor;

[0083] Step 3, based on the piezoelectric signal conversion device, the piezoelectric signal is collected, and the piezoelectric signal is filtered and amplified to improve the signal-to-noise ratio and sensitivity, and the substrate characteristics and environmental factors are comprehensively analyzed to correct and compensate the piezoelectric signal. The original piezoelectric signal generated by the piezoelectric sensing film due to the force is collected by the piezoelectric signal conversion device, and the collected original piezoelectric signal is filtered and amplified by the amplifier to improve the signal-to-noise ratio of the signal and facilitate subsequent measurement and analysis. The filtering process is used to remove noise and unnecessary frequency components, and the amplified signal is filtered. Perform analysis to identify the characteristics of the signal, including amplitude and frequency, analyze the impact of environmental factors such as humidity and temperature on the piezoelectric signal, obtain the environmental impact function, and perform environmental factor correction. Then, compensate the piezoelectric signal according to the material and thickness characteristics of the substrate, establish a relationship model between substrate characteristics and piezoelectric signals, obtain the substrate impact function, eliminate the impact of substrate characteristics on the measurement results, comprehensively analyze the analysis results of substrate characteristics and environmental factors, and build a signal correction and compensation model based on the environmental impact function and the substrate impact function to obtain the signal correction and compensation index, correct and compensate the piezoelectric signal to obtain more accurate measurement results;

[0084] Furthermore, the calculation formula of the signal correction and compensation index is:

[0085]

[0086] E env (T, H) = e qT -sin(wH);

[0087]

[0088] Among them, E co is the signal correction and compensation index, which is used to calculate the corrected and compensated piezoelectric signal, E a is the amplified piezoelectric signal, E env (T, H) is the environmental influence function, which indicates the influence of environmental factors (temperature T and humidity H) on the signal. B(λ, μ) is the substrate influence function, which indicates the influence of substrate characteristics. λ and μ represent the material parameters and thickness of the substrate, respectively. n is the total number of piezoelectric signal data points, which is used to calculate the volatility of the signal. E j is the piezoelectric signal value measured at the jth time, is the average value of all piezoelectric signals, q and w are environmental parameters. It should be noted that Calculate the sum of the squares of the deviations of all measurements from the mean, which reflects the volatility of the signal, e qTIt increases exponentially with the increase of temperature, sin(wH) changes periodically with the change of humidity, the substrate material parameter λ is usually a positive number, and the influence function increases with the increase of λ, μ 2 It means that as the thickness increases, the influence of the substrate decreases, and E co It will be adjusted according to the environment and substrate characteristics, and its value range and trend depend on the original signal, environmental parameters, substrate parameters and measurement volatility;

[0089] Step 4: Apply machine learning to analyze the piezoelectric signal data, analyze the signal changes caused by mask defects, substrate characteristic abnormalities, and environmental factors, obtain the OLED defect detection coefficient, and distinguish the severity and defect level of the OLED defect;

[0090] Step 5: Based on the test results, generate a test report for the detected defects and feed it back to the upstream process to optimize the evaporation process parameters, reduce defects and improve product yield.

[0091] Embodiment 2, as Figure 3 As shown, based on Example 1, the present invention provides a technical solution: Preferably, in step 4, the process of obtaining the OLED defect detection coefficient is:

[0092] Collect piezoelectric signal data of different types of OLED defects, including normal and various defect conditions, analyze the severity caused by each defect condition, and perform preprocessing operations such as cleaning, denoising, and standardization on the collected data to ensure data quality. Extract features from the preprocessed piezoelectric signal data, including amplitude, frequency, energy, skewness, and kurtosis, and divide the data set containing the preprocessed piezoelectric signal data into a training set, a validation set, and a test set. Use the training set data to train a neural network model, build a defect detection model, and use the validation set and the test set to evaluate the performance of the defect detection model. Comprehensively analyze the effects of mask defects, substrate characteristic abnormalities, and environmental factors on signal changes, obtain the OLED defect recognition coefficient, and perform defect recognition on the piezoelectric signal data. Based on the severity caused by the defect condition and the OLED defect recognition coefficient, distinguish the severity of OLED defect detection, and divide the defect level. Set the warning threshold corresponding to the defect level. The defect levels are respectively a slight defect level, a medium defect level, and a severe defect level. Deploy the trained defect detection model to the production environment, perform real-time analysis on new piezoelectric signal data, calculate the OLED defect recognition coefficient, and use it to identify OLED defects and determine the defect level.

[0093] Furthermore, multiple defect levels correspond to multiple warning thresholds, wherein the warning threshold includes an upper threshold and a lower threshold;

[0094] Multiple defect levels and multiple warning thresholds satisfy the following relationship:

[0095] Slight defect level D <D T1 ;

[0096] Medium defect level D T1 ≤D <D T2 ;

[0097] Severe defect level D≥D T2 ;

[0098] Where D is the OLED defect recognition coefficient, D T1 is the lower threshold corresponding to the medium defect level and the upper threshold corresponding to the slight defect level, D T2 The lower threshold corresponding to the severe defect level and the upper threshold corresponding to the medium defect level;

[0099] In step 5, the process of optimizing the evaporation process parameters is as follows:

[0100] Use the deployed defect detection model to analyze the piezoelectric signal data collected in real time, calculate the OLED defect recognition coefficient, and determine the level and severity of each defect based on the preset threshold. Generate a test report based on the defect detection results, and perform statistics and analysis on the data in the test report to identify the patterns and trends of defect generation. The test report contains basic information such as defect ID, level, severity, and detection time. The analysis results are fed back to the upstream evaporation process. Use the information in the test report to trace the process steps and conditions that caused the defects, find the root causes of the problems, and evaluate and adjust the evaporation process parameters based on the feedback information. Apply the proven effective improvement measures to actual production, continuously monitor the improved production data, and optimize the process parameters based on the monitoring results and new feedback information to form a closed loop of continuous improvement.

[0101] Embodiment 3, as Figure 4 As shown, on the basis of Embodiment 1-2, the present invention further provides a defect detection system for vapor-deposited OLED, which is used to implement a defect detection method for vapor-deposited OLED, including a defect detection platform, the defect detection platform is communicatively connected with a data acquisition module, a feature extraction module, a data management module, a model training module, a defect detection module and a report generation module, wherein the modules are electrically connected with each other;

[0102] The data acquisition module is used to collect the original piezoelectric signal data detected by the piezoelectric sensor, provide basic data for subsequent data processing and analysis, and perform pre-processing operations such as denoising, filtering, and standardization on the original piezoelectric signal data to improve data quality, reduce the impact of noise and outliers, and provide accurate data for feature extraction;

[0103] The feature extraction module is used to extract key features from the preprocessed piezoelectric signal data, including amplitude, frequency, energy, skewness, and kurtosis, and convert the original signal into numerical features that can be used in the machine learning model;

[0104] Data management module, which is used to manage the division, storage and access of data sets to ensure the correct use of data in the training, validation and testing stages;

[0105] A model training module is used to train a defect detection model using training set data and evaluate the defect detection model;

[0106] The defect detection module comprehensively analyzes the effects of mask defects, substrate characteristic anomalies, and environmental factors on signal changes based on the defect detection model, obtains the OLED defect recognition coefficient, performs defect recognition on the piezoelectric signal data, and recognizes and classifies the defects, distinguishes defects of different severity, and determines the defect level;

[0107] The report generation module is used to generate a test report containing defect ID, level, severity, and detection time, providing detailed defect information to facilitate process adjustment and quality control, and feeding back the test results to the upstream process links to guide the optimization of process parameters and reduce the occurrence of defects.

[0108] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A defect detection method for vapor deposition OLED, characterized in that: The following steps are involved: Step 1, selecting a clean substrate as a vapor deposition base, and installing a mask with a hollow pattern on the substrate, so that the mask and the substrate are aligned and installed; Step 2, preparing a piezoelectric sensing film on the surface of the mask by a physical vapor deposition method, and connecting a piezoelectric signal conversion device to the piezoelectric sensing film; Step 3, collecting the piezoelectric signal of the piezoelectric sensing film based on the piezoelectric signal conversion device, filtering and amplifying the piezoelectric signal, comprehensively analyzing the influence of substrate characteristics and environmental factors on the piezoelectric signal, and correcting and compensating the piezoelectric signal; Step 4: Apply machine learning to analyze the piezoelectric signal data, analyze the signal changes caused by mask defects, substrate characteristic abnormalities, and environmental factors, obtain the OLED defect detection coefficient, and distinguish the severity and defect level of the OLED defect; Step 5: Based on the test results, generate a test report for the detected defects and feed it back to the upstream process to optimize the evaporation process parameters.

2. The defect detection method for vapor-deposited OLED according to claim 1, characterized in that: In step 1, the process of aligning and installing the mask and the substrate is as follows: Step 101, cleaning and drying the substrate, and using a laser scanner to inspect the surface of the substrate to ensure that the substrate surface is flat, scratch-free, and free of dents. Ultrapure water or other cleaning agents are used to thoroughly clean the substrate surface to remove grease, dust, and other contaminants. After cleaning, the substrate is placed in a dust-free environment for drying, and is dried in a drying oven using nitrogen to completely remove moisture from the substrate surface. Before installing the mask, a laser scanner is used to inspect the substrate surface in a clean environment; Step 102, selecting a metal mask with a precise hollow pattern, making it match the size of the substrate, and cleaning the mask; Step 103, using a positioning fixture to place the mask above the substrate, align it with the substrate, and adjust the position of the fixture so that the hollow pattern on the mask completely corresponds to the predetermined evaporation area on the substrate. After the mask is installed, check the alignment accuracy of the mask and the substrate; Step 104, applying pre-pressure to the mask and the substrate to fix the mask on the substrate so that the mask and the substrate are closely fitted; Step 105 , recording installation parameters of the mask and substrate, including alignment accuracy and pre-pressure, and placing the mask and substrate in a vapor deposition chamber.

3. A defect detection method for vapor-deposited OLED according to claim 2, characterized in that: In step 2, the process of connecting the piezoelectric signal conversion device to the piezoelectric sensing film is as follows: Step 201, preparing the aluminum nitride piezoelectric sensing film, selecting high-quality aluminum nitride powder as a deposition source, and preparing PVD equipment; Step 202, in a PVD device, heating the aluminum nitride powder to its evaporation temperature, so that the aluminum nitride powder is converted into a gaseous state by evaporation, and using the gas flow of the vacuum system of the PVD device to transport the aluminum nitride molecules into the evaporation chamber, and the vaporized aluminum nitride molecules are deposited on the surface of the substrate to form a piezoelectric sensing film; Step 203, connecting a piezoelectric signal conversion device to a piezoelectric sensing film made of aluminum nitride, and calibrating the piezoelectric signal conversion device to adjust the sensitivity of the piezoelectric signal conversion device to capture weak piezoelectric signals, wherein the piezoelectric signal conversion device is a piezoelectric sensor; Step 204 , applying pressures of different magnitudes to the piezoelectric sensing film, observing changes in the electrical signal output by the piezoelectric signal conversion device, analyzing the converted electrical signal, and evaluating the stability of the piezoelectric sensing film.

4. The defect detection method for vapor-deposited OLED according to claim 3, characterized in that: The process of evaluating the stability of the piezoelectric sensing film is: Step 2041, applying different pressures to the piezoelectric sensing film, and recording the value of each pressure value; Step 2042, corresponding to each applied pressure, measuring and recording the electrical signal output by the piezoelectric sensor, pairing and marking the acquired pressure value with the corresponding electrical signal to obtain a corresponding data pair; Step 2043, comprehensively analyze all the data pairs of pressure values ​​and electrical signals, obtain the piezoelectric evaluation index, check the linear relationship between the pressure value and the electrical signal within a certain pressure range, evaluate the response of the piezoelectric sensing film to the pressure change, and then analyze the stability of the piezoelectric sensing film. The expression of the piezoelectric evaluation index is: Where E(P) is the piezoelectric evaluation index, P i is the pressure value applied for the ith time, P r is the reference pressure value, which is the minimum pressure value at which the film starts to respond, n is the total number of piezoelectric signal data points, a, b, c are model parameters, P max is the maximum value of the applied pressure.

5. The defect detection method for vapor-deposited OLED according to claim 4, characterized in that: In step 3, the correction and compensation process of the piezoelectric signal is as follows: Step 301, using a piezoelectric signal conversion device to collect the original piezoelectric signal generated by the piezoelectric sensing film due to the force, filtering the collected original piezoelectric signal, and amplifying the filtered signal through an amplifier; Step 302, analyzing the amplified signal to identify the characteristics of the signal, including amplitude and frequency; Step 303, analyzing the influence of environmental factors such as humidity and temperature on the piezoelectric signal, obtaining an environmental influence function, and performing environmental factor correction, then compensating the piezoelectric signal according to the material and thickness characteristics of the substrate, establishing a relationship model between the substrate characteristics and the piezoelectric signal, obtaining a substrate influence function, and eliminating the influence of the substrate characteristics on the measurement result; Step 304 , comprehensively analyzing the analysis results of substrate characteristics and environmental factors, building a signal correction and compensation model in combination with the environmental influence function and the substrate influence function, obtaining a signal correction and compensation index, and correcting and compensating the piezoelectric signal.

6. The defect detection method for vapor-deposited OLED according to claim 5, characterized in that: The calculation formula of the signal correction and compensation index is: Yes env (T, H)=e qT -sin(wH); Among them, E co is the signal correction and compensation index, E a is the amplified piezoelectric signal, E env (T, H) is the environmental influence function, B(λ, μ) is the substrate influence function, λ and μ represent the material parameters and thickness of the substrate respectively, n is the total number of piezoelectric signal data points, E j is the piezoelectric signal value measured at the jth time, is the average value of all piezoelectric signals, q and w are environmental parameters.

7. The defect detection method for vapor-deposited OLED according to claim 6, characterized in that: In step 4, the process of obtaining the OLED defect detection coefficient is as follows: Step 401, collecting piezoelectric signal data of different types of OLED defects, including normal and various defect conditions, analyzing the severity of each defect condition, and performing pre-processing operations such as cleaning, denoising, and standardization on the collected data; Step 402, extracting features from the preprocessed piezoelectric signal data, including amplitude, frequency, energy, skewness, and kurtosis, and dividing the data set containing the preprocessed piezoelectric signal data into a training set, a validation set, and a test set; Step 403, using the training set data to train the machine learning model, constructing a defect detection model, and using the validation set and the test set to evaluate the performance of the defect detection model, comprehensively analyzing the effects of mask defects, substrate characteristic abnormalities, and environmental factors on signal changes, and obtaining the OLED defect recognition coefficient to perform defect recognition on the piezoelectric signal data; Step 404, based on the severity caused by the defect situation and the OLED defect recognition coefficient, distinguish the severity of the OLED defect detection, divide the defect levels, and set warning thresholds corresponding to the defect levels, the defect levels are respectively a slight defect level, a medium defect level, and a severe defect level; Step 405 , deploy the trained defect detection model to the production environment, perform real-time analysis on the new piezoelectric signal data, calculate the OLED defect recognition coefficient and use it to identify OLED defects and determine the defect level.

8. The defect detection method for vapor-deposited OLED according to claim 7, characterized in that: The calculation formula of the OLED defect recognition coefficient is: Where D is the OLED defect recognition coefficient, n is the total number of piezoelectric signal data points, E k is the piezoelectric signal value of the kth data point, max(E k ) is the maximum value of all piezoelectric signal values, min(E k ) is the minimum value of all piezoelectric signal values, σ is the standard deviation of the piezoelectric signal data, indicating the dispersion of the signal value, is the mean error of the piezoelectric signal data, S is the skewness of the signal, and K is the kurtosis of the signal.

9. The defect detection method for vapor-deposited OLED according to claim 8, characterized in that: In step 5, the process of optimizing the evaporation process parameters is as follows: Step 501, using the deployed defect detection model to analyze the piezoelectric signal data collected in real time, calculate the OLED defect recognition coefficient, and determine the level and severity of each defect according to a preset threshold; Step 502: Generate a test report based on the defect detection results, and perform statistics and analysis on the data in the test report to identify the pattern and trend of defect generation. The test report contains basic information such as defect ID, level, severity, and detection time. Step 503, feeding back the analysis results to the upstream evaporation process link, using the information in the test report to trace the process steps and conditions that caused the defects, find the root cause of the problem, and evaluate and adjust the evaporation process parameters based on the feedback information; Step 504, applying the verified effective improvement measures to actual production, continuously monitoring the improved production data, and optimizing the process parameters based on the monitoring results and new feedback information to form a closed loop of continuous improvement.

10. A defect detection system for vapor deposition OLED, used to implement the defect detection method for vapor deposition OLED as claimed in any one of claims 1 to 9, comprising a defect detection platform, characterized in that: The defect detection platform is communicatively connected with a data acquisition module, a feature extraction module, a data management module, a model training module, a defect detection module and a report generation module, wherein the modules are electrically connected with each other; The data acquisition module is used to collect the original piezoelectric signal data detected by the piezoelectric sensor, and perform pre-processing operations such as denoising, filtering, and standardization on the original piezoelectric signal data; The feature extraction module is used to extract key features from the preprocessed piezoelectric signal data, including amplitude, frequency, energy, skewness, and kurtosis, and convert the original signal into numerical features that can be used in the machine learning model; The data management module is used to manage the division, storage and access of data sets; The model training module is used to train the defect detection model using the training set data and evaluate the defect detection model; The defect detection module comprehensively analyzes the influence of mask defects, substrate characteristic abnormalities and environmental factors on signal changes based on the defect detection model, obtains the OLED defect recognition coefficient, performs defect recognition on the piezoelectric signal data, and recognizes and classifies the defects, distinguishes defects of different severity, and determines the defect level; The report generation module is used to generate a test report containing defect ID, level, severity, and detection time, provide detailed defect information, and feed back the test results to the upstream process link to guide the optimization of process parameters.

Images