Sorting and die bonding precision compensation method

Abstract

The invention discloses a sorting and die bonding precision compensation method. The method comprises the following steps: sorting crystal grains by using sorting equipment; the method comprises the following steps: collecting historical data of different models of substrates during last processing, classifying the collected historical data according to the models of the substrates, and establishing a database to store the data; a substrate to be processed is matched with data in the database, processing parameters of historical data are obtained, a compensation relation model between the suction nozzle and the crystal grains is established according to deviation values in the historical data, initial processing parameters in the historical data are used as preset processing parameters, and before die bonding, the initial processing parameters in the historical data are used as preset processing parameters; and the compensation model is used for adjusting the preset machining parameters, and accurate initial machining parameters are obtained. By collecting the historical data and compensating and adjusting the preset parameters according to the historical deviation data, the accuracy of the initial set parameters can be improved, meanwhile, the function of adjusting the die bonding position in real time is achieved, and the accuracy of die bonding each time is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing and analysis, and in particular to a method for compensating precision of crystal sorting and bonding. Background Art

[0002] Die bonding is a crucial link in the semiconductor manufacturing process. The accuracy of die bonding directly affects the performance and quality of semiconductor devices. At present, in the die bonding process, due to the influence of various factors, such as equipment accuracy, process parameters, material properties, etc., there are often certain accuracy deviations. These deviations may cause problems such as inaccurate position and angle deviation of the grains during sorting and die bonding, thereby affecting the yield and performance of the product;

[0003] In order to solve these problems, traditional die bonding operations usually require manual data search and calculation to determine the best processing parameters, and do not determine the parameters based on historical data. The manual parameter determination method is not only inefficient, but also reduces the accuracy of the first parameter setting. In addition, since deviations will occur each time a die bonding operation is performed, traditional die bonding operations do not perform image capture and calculation of deviations for each die bonding operation, so that problems such as position deviation and angle offset in the die bonding process cannot be detected in time, affecting the die bonding accuracy.

[0004] Therefore, there is an urgent need for a new sorting and bonding precision compensation method that can automatically and accurately compensate for the precision deviation in the bonding process, improve the yield and performance of the product, and meet the high-precision requirements of modern semiconductor manufacturing.

[0005] Currently, no effective solution has been proposed for the problems in the related technologies. Summary of the invention

[0006] In view of the problems in the related art, the present invention proposes a method for compensating the precision of sorting and bonding crystals to overcome the above technical problems existing in the existing related art.

[0007] To this end, the specific technical solution adopted by the present invention is as follows:

[0008] A method for compensating precision of crystal sorting and bonding, the method comprising the following steps:

[0009] S1. Use sorting equipment to sort the grains;

[0010] S2. Collect the historical data of the latest processing of different types of substrates, classify the collected historical data according to the substrate models, and establish a database to store the data;

[0011] S3, matching the substrate to be processed with the data in the database, obtaining the processing parameters of the historical data, establishing a compensation relationship model between the nozzle and the die according to the deviation value in the historical data, using the initial processing parameters in the historical data as the preset processing parameters, and adjusting the preset processing parameters using the compensation model before die bonding to obtain accurate initial processing parameters;

[0012] S4, performing a die bonding operation using the obtained initial parameters, taking a photo of the captured die after each die grabbing, recording an image, and performing defect detection on the outer surface of the die based on the captured image;

[0013] S5, comparing the position of the current captured grain with the position of the previous grain, calculating a relative deviation value, and calculating a compensation value based on the deviation value;

[0014] S6, outputting the compensation value, and performing the die bonding operation according to the compensated target position to ensure that the die is accurately solidified on the substrate;

[0015] S7. After the processing is completed, the processing data is integrated and input into the database, and the historical data of the processing of the current category of substrates in the database is replaced.

[0016] As a preferred embodiment, the collecting of historical data of the most recent processing of different types of substrates, classifying the collected historical data according to the substrate models, and establishing a database to store the data includes the following steps:

[0017] S21. Collect processing parameters, quality parameters, equipment status, and environmental parameters. Processing parameters include die bonding pressure, temperature, and time. Quality parameters include error parameters and compensation. Equipment status includes equipment number and calibration records. Environmental parameters include temperature, humidity, and vibration level.

[0018] S22, classifying the collected data according to the substrate model;

[0019] S23, cleaning and formatting the data to remove erroneous records;

[0020] S24. Create a database and import the sorted data.

[0021] The most recent processing data can most accurately reflect the status of the current production equipment, process and environment. The decisions made based on these data are more targeted and timely, and can adjust production parameters and optimize process flow in a timely manner to adapt to changes in actual production conditions.

[0022] As a preferred embodiment, the substrate to be processed is matched with the data in the database, the processing parameters of the historical data are obtained, a compensation relationship model between the nozzle and the die is established according to the deviation value in the historical data, the initial processing parameters in the historical data are used as the preset processing parameters, and before the die is bonded, the preset processing parameters are adjusted using the compensation model to obtain accurate initial processing parameters, which includes the following steps:

[0023] S31, input the model of the substrate to be processed, and match the historical data of the processing of the substrate with the same model from the database;

[0024] S32, obtaining deviation values ​​in historical data, and establishing a compensation relationship model between the nozzle and the grain;

[0025] S33, obtaining initial processing parameters from historical data, and using the initial processing parameters as preset processing parameters, and finally adjusting the preset processing parameters using a compensation relationship model.

[0026] The selected historical processing parameters are the most recent historical data, which are closest to the current processing situation in time. Factors such as production environment and equipment status usually do not change significantly in a short period of time, so these parameters can better reflect the current possible best processing status and have higher applicability.

[0027] As a preferred embodiment, the step of obtaining the deviation value in the historical data and establishing a compensation relationship model between the nozzle and the die includes the following steps:

[0028] S321, select a linear regression model to describe the compensation relationship between the nozzle and the grain, the formula is:

[0029] W x =a x p+b x ·m+c x ·t+d x ;

[0030] W y =a y p+b y ·m+c y ·t+d y ;

[0031] Among them, W x , W y Respectively represent the deviation values ​​of the X-axis and Y-axis, a x 、b x 、c x d x and a y 、b y 、c y dy are the linear regression coefficients for the X-axis and Y-axis, respectively, p is the die bonding pressure, m is the die bonding temperature, and t is the die bonding time;

[0032] The compensation relationship formula is:

[0033] p′=p+Δp;

[0034] m′=m+Δm;

[0035] t′=t+Δt;

[0036] Where p″ is the compensation pressure, m′ is the compensation temperature and t″ is the compensation time;

[0037] S322, outputting the calculated compensation parameters as actually determined processing parameters.

[0038] As a preferred embodiment, the method of performing a die bonding operation using the obtained initial parameters, taking a photo of the captured die after each die grabbing, recording an image, and performing defect detection on the outer surface of the die according to the captured image comprises the following steps:

[0039] S41, input the determined processing parameters and execute the grabbing action;

[0040] S42, the photographing device automatically executes a photographing function to photograph the captured die, and the photographed content includes an external image of the die and a position image of the die;

[0041] S43, preprocessing the captured external image of the grain, including denoising, contrast enhancement, brightness adjustment, and amplification;

[0042] S44, using image processing algorithms to process the images and automatically identify defects on the surface of the grains, including cracks, scratches, foreign matter, and uneven coatings;

[0043] S45. When defects are identified on the outside of the crystal grain, the defective crystal grain is captured to a special processing channel so that the crystal grain is reworked or scrapped. When defects are not identified on the outside of the crystal grain, the crystal bonding operation continues.

[0044] The photo taking mode is continuous shooting mode, which can obtain images at multiple moments, increasing the observation angle and amount of information on the state of the grain, and can also capture the performance of the grain under different postures and lighting conditions.

[0045] As a preferred embodiment, the step of comparing the position of the currently captured grain with the position of the previous grain, calculating the relative deviation value, and calculating the compensation value according to the deviation value comprises the following steps:

[0046] S51, preprocessing the position image of the currently captured grain and the position image of the last captured grain;

[0047] S52, using an image processing algorithm to locate the position of the grain, and determining the centroid position of the currently captured grain and the centroid position of the previous grain;

[0048] S53, calculating the difference between the current grain position and the previous grain position to obtain a relative deviation value, and using the obtained deviation value to calculate a corresponding compensation value;

[0049] S54, before performing the next die bonding operation, adjusting the processing parameters using the calculated compensation value.

[0050] Since errors will occur every time a grasping operation is performed, an image will be captured and the compensation amount for the specific deviation in the image will be calculated. This method can adapt to different working environments and working conditions and improve the grasping accuracy.

[0051] As a preferred embodiment, after the processing is completed, the processing data is integrated, input into the database, and the historical data of the processing of the current type of substrate in the database is replaced, which includes the following steps:

[0052] S71. Collect all relevant data during the processing;

[0053] S72. Screening the collected raw data, removing invalid or abnormal data, and organizing the data into a unified format;

[0054] S73, integrating all data into a complete data package and connecting it to the database through software;

[0055] S74, adding the currently processed data to the relevant table of the database, querying the database, filtering out the historical data associated with the currently processed category substrate, and performing a deletion operation.

[0056] When database storage is limited, replacing data can prevent excessive expansion of the database, save storage space, and make database management more efficient.

[0057] The beneficial effects of the present invention are:

[0058] 1. The present invention collects historical data of the most recent processing of different types of substrates, and classifies and stores the data, so that the initial processing parameters of subsequent different types of substrates can be quickly determined. In addition, it also effectively simplifies the operation process, so that there is no need to manually search for a large number of parameters for parameter determination during subsequent die bonding.

[0059] 2. The present invention obtains historical deviation values ​​and several processing parameters from historical data, and uses the historical deviation values ​​to establish a compensation relationship model between the suction nozzle and the grain. It can accurately adjust the difference between the suction nozzle and the grain, and then use the model to adjust the preset parameters before solidifying the crystal to obtain accurate initial processing parameters, thereby greatly improving the accuracy of the initial processing parameters.

[0060] 3. The present invention first detects defects on the outside of the grains in the image by taking an image each time the grains are grabbed. When defects are detected, the defective grains are grabbed to a special channel to facilitate the next step of rework or scrapping of the grains, thereby improving the control level of product quality, timely identifying and separating defective grains, and avoiding defective products from flowing into subsequent processes, reducing unnecessary processing and waste of resources;

[0061] If there are no defects on the outside of the grain, the grain position image is processed, the deviation value and compensation value are calculated, and then the compensation value is output and the die bonding operation is performed according to the compensated target position. This method can dynamically adjust the die bonding position in real time to ensure that each grain can be solidified onto the substrate as accurately as possible, greatly improving the accuracy and consistency of die bonding and effectively reducing the product defective rate caused by inaccurate position. Secondly, this method can also adapt to various changes and uncertainties in the production process, making the production process more stable and reliable.

[0062] 4. The present invention can ensure that the database always stores the latest processing information by integrating the processed data and inputting it into the database to replace the relevant historical data, providing the most timely reference for the subsequent processing of the same type of substrates. In addition, replacing the data can also avoid excessive expansion of the database and save storage space.

[0063] 5. The present invention uses historical data to obtain historical deviation data and historical initial setting data, so that the staff does not need to perform a large amount of data search and calculation to determine the initial processing parameters. In addition, the preset initial parameters are adjusted according to the deviation value of the most recent historical data, so that the parameters during the first crystal bonding operation are more accurate. In addition, when performing the crystal bonding operation, it also has the function of dynamically adjusting the crystal bonding position in real time according to the captured grain image, thereby improving the accuracy of each crystal bonding. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] In order to more clearly illustrate the embodiments of the present invention 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 of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0065] Figure 1 The present invention is a flowchart of a method for compensating precision of sorting and bonding die according to an embodiment of the present invention. DETAILED DESCRIPTION

[0066] To further illustrate each embodiment, the present invention provides drawings, which are part of the disclosure of the present invention and are mainly used to illustrate the embodiments and can be used in conjunction with the relevant descriptions in the specification to explain the operating principles of the embodiments. With reference to these contents, ordinary technicians in the field should be able to understand other possible implementations and advantages of the present invention. The components in the figures are not drawn to scale, and similar component symbols are generally used to represent similar components.

[0067] According to an embodiment of the present invention, a method for compensating precision of sorting and bonding crystals is provided.

[0068] The present invention is further described with reference to the accompanying drawings and specific embodiments. Figure 1 As shown, a method for compensating the precision of crystal sorting and bonding according to an embodiment of the present invention comprises the following steps:

[0069] S1. Use sorting equipment to sort the grains;

[0070] S2. Collect the historical data of the latest processing of different types of substrates, classify the collected historical data according to the substrate models, and establish a database to store the data;

[0071] Furthermore, the collecting of historical data of the most recent processing of substrates of different models, classifying the collected historical data according to the substrate models, and establishing a database to store the data includes the following steps:

[0072] S21. Collect processing parameters, quality parameters, equipment status, and environmental parameters. Processing parameters include die bonding pressure, temperature, and time. Quality parameters include error parameters and compensation. Equipment status includes equipment number and calibration records. Environmental parameters include temperature, humidity, and vibration level.

[0073] S22, classifying the collected data according to the substrate model;

[0074] S23, cleaning and formatting the data to remove erroneous records;

[0075] S24. Create a database and import the sorted data.

[0076] It should be noted that classifying historical data according to substrate models in advance can facilitate parameter selection and determination of subsequent substrates of the same model. When subsequent substrates of the same model are processed, historical data can be quickly matched by substrate model name.

[0077] S3, matching the substrate to be processed with the data in the database, obtaining the processing parameters of the historical data, establishing a compensation relationship model between the nozzle and the die according to the deviation value in the historical data, using the initial processing parameters in the historical data as the preset processing parameters, and adjusting the preset processing parameters using the compensation model before die bonding to obtain accurate initial processing parameters;

[0078] Furthermore, the substrate to be processed is matched with the data in the database, the processing parameters of the historical data are obtained, a compensation relationship model between the nozzle and the crystal grain is established according to the deviation value in the historical data, the initial processing parameters in the historical data are used as the preset processing parameters, and before the crystal is fixed, the preset processing parameters are adjusted by using the compensation model to obtain accurate initial processing parameters, which includes the following steps:

[0079] S31, input the model of the substrate to be processed, and match the historical data of the processing of the substrate with the same model from the database;

[0080] S32, obtaining deviation values ​​in historical data, and establishing a compensation relationship model between the nozzle and the grain;

[0081] S33, obtaining initial processing parameters from historical data, and using the initial processing parameters as preset processing parameters, and finally adjusting the preset processing parameters using a compensation relationship model.

[0082] It should be noted that processing parameters are usually preset before performing the die bonding operation, and when the substrate type is the same, the preset processing parameters are usually the same value. The parameters have been set during historical processing. Therefore, directly using the initial processing parameters in the historical data as the preset parameters can save a lot of parameter debugging time, avoid the process of repeated testing of different parameter combinations, and improve processing efficiency.

[0083] Furthermore, the step of obtaining the deviation value in the historical data and establishing a compensation relationship model between the nozzle and the die includes the following steps:

[0084] S321, select a linear regression model to describe the compensation relationship between the nozzle and the grain, the formula is:

[0085] W x =a x p+b x ·m+c x ·t+d x ;

[0086] W y =a y p+b y ·m+c y ·t+d y ;

[0087] Among them, W x , W y Respectively represent the deviation values ​​of the X-axis and Y-axis, a x , b x 、c x d x and a y , b y 、c y d y are the linear regression coefficients for the X-axis and Y-axis, respectively, p is the die bonding pressure, m is the die bonding temperature, and t is the die bonding time;

[0088] The compensation relationship formula is:

[0089] p′=p+Δp;

[0090] m′=m+Δm;

[0091] t′=t+Δt;

[0092] Where p″ is the compensation pressure, m′ is the compensation temperature and t″ is the compensation time;

[0093] S322, outputting the calculated compensation parameters as actually determined processing parameters.

[0094] It should be noted that by establishing a compensation relationship model based on the deviation values ​​in the historical data, precise adjustments can be made for different processing situations. Parameter adjustment before die bonding can effectively reduce the production of defective products caused by the mismatch between the nozzle and the die, and improve the product yield. In addition, the compensation model can be used to adjust the preset processing parameters without a long trial and error process, and accurate initial processing parameters can be quickly obtained.

[0095] S4, performing a die bonding operation using the obtained initial parameters, taking a photo of the captured die after each die grabbing, recording an image, and performing defect detection on the outer surface of the die based on the captured image;

[0096] Further, the obtained initial parameters are used to perform a die bonding operation, and after each grain is grabbed, the grabbed grain is photographed, an image is recorded, and defect detection is performed on the outer surface of the grain according to the photographed image, including the following steps:

[0097] S41, input the determined processing parameters and execute the grabbing action;

[0098] S42, the photographing device automatically executes a photographing function to photograph the captured die, and the photographed content includes an external image of the die and a position image of the die;

[0099] S43, preprocessing the captured external image of the grain, including denoising, contrast enhancement, brightness adjustment, and amplification;

[0100] S44, using image processing algorithms to process the images and automatically identify defects on the surface of the grains, including cracks, scratches, foreign matter, and uneven coatings;

[0101] S45. When defects are identified on the outside of the crystal grain, the defective crystal grain is captured to a special processing channel so that the crystal grain is reworked or scrapped. When defects are not identified on the outside of the crystal grain, the crystal bonding operation continues.

[0102] It should be noted that when a defect appears on the outside of a captured grain, the next time the grain is captured for bonding, the last position information of the defective grain will be used for bonding, to ensure the continuity and stability of the bonding process.

[0103] S5, comparing the position of the current captured grain with the position of the previous grain, calculating a relative deviation value, and calculating a compensation value based on the deviation value;

[0104] Furthermore, the step of comparing the position of the currently captured grain with the position of the last grain, calculating the relative deviation value, and calculating the compensation value according to the deviation value includes the following steps:

[0105] S51, preprocessing the position image of the currently captured grain and the position image of the last captured grain;

[0106] S52, using an image processing algorithm to locate the position of the grain, and determining the centroid position of the currently captured grain and the centroid position of the previous grain;

[0107] S53, calculating the difference between the current grain position and the previous grain position to obtain a relative deviation value, and using the obtained deviation value to calculate a corresponding compensation value;

[0108] S54, before performing the next die bonding operation, adjusting the processing parameters using the calculated compensation value.

[0109] It should be noted that this method can monitor the position changes during the grasping process in real time, detect possible deviations in time, and improve the accuracy and stability of grasping. Secondly, the calculated compensation value can accurately adjust the grasping position, effectively reduce the position error caused by various factors, and ensure that each grain can be accurately grasped and placed, thereby improving production efficiency and product quality.

[0110] In the application of multi-nozzle structure, in order to improve the relative accuracy of the overall system, the following adjustment steps can be taken: First, for each nozzle in different grasping situations, the system will obtain the specific deviation value of each nozzle when grasping the die. These deviation values ​​reflect the difference between the nozzle in actual operation and the ideal position, which is the basis for subsequent compensation calculations;

[0111] Next, the compensation amount is calculated based on the obtained deviation value, and through these calculations, the specific compensation amount required for each nozzle is obtained;

[0112] Finally, these calculated compensations are applied precisely to the corresponding nozzles.

[0113] This compensation includes not only fine-tuning of position, but also adjustment of force and speed to ensure that each nozzle can grab and place the die with the highest accuracy under various operating conditions. In this way, the overall relative accuracy of the multi-nozzle structure is significantly improved, further ensuring the efficient and stable operation of the entire system.

[0114] S6, outputting the compensation value, and performing the die bonding operation according to the compensated target position to ensure that the die is accurately solidified on the substrate;

[0115] S7. After the processing is completed, the processing data is integrated and input into the database, and the historical data of the processing of the current category of substrates in the database is replaced.

[0116] Furthermore, after the processing is completed, the processing data is integrated, input into the database, and the historical data of the processing of the current type of substrate in the database is replaced, including the following steps:

[0117] S71. Collect all relevant data during the processing;

[0118] S72. Screening the collected raw data, removing invalid or abnormal data, and organizing the data into a unified format;

[0119] S73, integrating all data into a complete data package and connecting it to the database through software;

[0120] S74, adding the currently processed data to the relevant table of the database, querying the database, filtering out the historical data associated with the currently processed category substrate, and performing a deletion operation.

[0121] It should be noted that after the processing is completed, the processing data will be integrated and entered into the database, and the historical data of the current category of substrate processing will be replaced to ensure that the data in the database is always kept up to date, providing the most realistic reference for subsequent processing of the same category of substrates, thereby greatly improving the accuracy and efficiency of parameter settings.

[0122] In summary, the present invention collects historical data of the most recent processing of different types of substrates, and classifies and stores the data, so that the initial processing parameters of subsequent different types of substrates can be quickly determined. In addition, the operation process is effectively simplified, so that there is no need to manually search for a large number of parameters to determine the parameters during the subsequent solidification. The present invention obtains historical deviation values ​​and several processing parameters from historical data, and uses historical deviation values ​​to establish a compensation relationship model between the suction nozzle and the grain, so that the difference between the suction nozzle and the grain can be accurately adjusted, so that the preset parameters are adjusted by the model before solidification to obtain accurate initial processing parameters, thereby greatly improving the accuracy of the initial processing parameters. The present invention first detects defects on the outside of the grain in the image by taking an image each time the grain is grabbed. When defects are detected, the defective grain will be grabbed to a special channel to facilitate the next step of rework or scrapping of the grain, thereby improving the control level of product quality, timely identifying and separating defective grains, and avoiding defective products from flowing into subsequent processes, reducing unnecessary processing and resource waste. If there is no defect on the outside of the grain, the grain position image is processed, the deviation value and the compensation value are calculated, and then After that, the compensation value is output and the solid crystal operation is performed according to the compensated target position. This method can adjust the solid crystal position in real time and dynamically, ensuring that each crystal grain can be solidified onto the substrate as accurately as possible, greatly improving the accuracy and consistency of solid crystal, and effectively reducing the product defect rate caused by inaccurate position. Secondly, this method can also adapt to various changes and uncertainties in the production process, making the production process more stable and reliable; the present invention can ensure that the database always stores the latest processing information by integrating the processed data and inputting the database to replace the relevant historical data, providing the most timely reference basis for the subsequent processing of the same type of substrates. In addition, replacing the data can also avoid excessive expansion of the database and save storage space; the present invention obtains historical deviation data and historical initial setting data by using historical data, so that the staff does not need to perform a large amount of data search and calculation to determine the initial processing parameters. In addition, the preset initial parameters are adjusted by the deviation value of the most recent historical data, so that the parameters during the first solid crystal operation are more accurate. In addition, when performing the solid crystal operation, it also has the function of adjusting the solid crystal position in real time and dynamically according to the captured crystal grain image, thereby improving the accuracy of each solid crystal.

[0123] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for compensating the precision of crystal sorting and bonding, characterized in that: The method comprises the following steps: S1. Use sorting equipment to sort the grains; S2. Collect the historical data of the latest processing of different types of substrates, classify the collected historical data according to the substrate models, and establish a database to store the data; S3, matching the substrate to be processed with the data in the database, obtaining the processing parameters of the historical data, establishing a compensation relationship model between the nozzle and the die according to the deviation value in the historical data, using the initial processing parameters in the historical data as the preset processing parameters, and adjusting the preset processing parameters using the compensation model before die bonding to obtain accurate initial processing parameters; S4, performing a die bonding operation using the obtained initial parameters, taking a photo of the captured die after each die grabbing, recording an image, and performing defect detection on the outer surface of the die based on the captured image; S5, comparing the position of the current captured grain with the position of the previous grain, calculating a relative deviation value, and calculating a compensation value based on the deviation value; S6, outputting the compensation value, and performing the die bonding operation according to the compensated target position to ensure that the die is accurately solidified on the substrate; S7. After the processing is completed, the processing data is integrated and input into the database, and the historical data of the processing of the current category of substrates in the database is replaced.

2. A method for compensating the precision of crystal sorting and bonding according to claim 1, characterized in that: The collecting of historical data of the most recent processing of different types of substrates, classifying the collected historical data according to the substrate models, and establishing a database to store the data includes the following steps: S21. Collect processing parameters, quality parameters, equipment status, and environmental parameters. Processing parameters include die bonding pressure, temperature, and time. Quality parameters include error parameters and compensation. Equipment status includes equipment number and calibration records. Environmental parameters include temperature, humidity, and vibration level. S22, classifying the collected data according to the substrate model; S23, cleaning and formatting the data to remove erroneous records; S24. Create a database and import the sorted data.

3. A method for compensating the precision of crystal sorting and bonding according to claim 1, characterized in that: The method of matching the substrate to be processed with the data in the database, obtaining the processing parameters of the historical data, establishing a compensation relationship model between the nozzle and the crystal grain according to the deviation value in the historical data, using the initial processing parameters in the historical data as the preset processing parameters, and adjusting the preset processing parameters by using the compensation model before crystal bonding to obtain accurate initial processing parameters includes the following steps: S31, input the model of the substrate to be processed, and match the historical data of the processing of the substrate with the same model from the database; S32, obtaining deviation values ​​in historical data, and establishing a compensation relationship model between the nozzle and the grain; S33, obtaining initial processing parameters from historical data, and using the initial processing parameters as preset processing parameters, and finally adjusting the preset processing parameters using a compensation relationship model.

4. A method for compensating the precision of crystal bonding during separation according to claim 3, characterized in that: The method of obtaining the deviation value in the historical data and establishing the compensation relationship model between the nozzle and the die comprises the following steps: S321, select a linear regression model to describe the compensation relationship between the nozzle and the grain, the formula is: W x =a x ·p+b x ·m+c x ·t+d x ; W y =a y ·p+b y ·m+c y ·t+d y ; Among them, W x , W y Respectively represent the deviation values ​​of the X-axis and Y-axis, a x 、b x 、c x ,d x and a y 、b y 、c y ,d y are the linear regression coefficients for the X-axis and Y-axis, respectively, p is the die bonding pressure, m is the die bonding temperature, and t is the die bonding time; The compensation relationship formula is: p′=p+Δp; m′=m+Δm; t′=t+Δt; Where p″ is the compensation pressure, m′ is the compensation temperature and t″ is the compensation time; S322, outputting the calculated compensation parameters as actually determined processing parameters.

5. The method for compensating the precision of crystal bonding during separation according to claim 1, characterized in that: The method of performing a die bonding operation using the obtained initial parameters, taking a photo of the captured die after each die grabbing, recording an image, and performing defect detection on the outer surface of the die according to the captured image comprises the following steps: S41, input the determined processing parameters and execute the grabbing action; S42, the photographing device automatically executes a photographing function to photograph the captured die, and the photographed content includes an external image of the die and a position image of the die; S43, preprocessing the captured external image of the grain, including denoising, contrast enhancement, brightness adjustment, and amplification; S44, using image processing algorithms to process the images and automatically identify defects on the surface of the grains, including cracks, scratches, foreign matter, and uneven coatings; S45. When defects are identified on the outside of the crystal grain, the defective crystal grain is captured to a special processing channel so that the crystal grain is reworked or scrapped. When defects are not identified on the outside of the crystal grain, the crystal bonding operation continues.

6. The method for compensating the precision of crystal bonding during separation according to claim 1, characterized in that: The step of comparing the position of the current captured grain with the position of the previous grain, calculating the relative deviation value, and calculating the compensation value according to the deviation value comprises the following steps: S51, preprocessing the position image of the currently captured grain and the position image of the last captured grain; S52, using an image processing algorithm to locate the position of the grain, and determining the centroid position of the currently captured grain and the centroid position of the previous grain; S53, calculating the difference between the current grain position and the previous grain position to obtain a relative deviation value, and using the obtained deviation value to calculate a corresponding compensation value; S54, before performing the next die bonding operation, adjusting the processing parameters using the calculated compensation value.

7. The method for compensating the precision of crystal bonding during separation according to claim 1, characterized in that: After the processing is completed, the processing data is integrated, input into the database, and the historical data of the current type of substrate processing in the database is replaced, including the following steps: S71. Collect all relevant data during the processing; S72. Screening the collected raw data, removing invalid or abnormal data, and organizing the data into a unified format; S73, integrating all data into a complete data package and connecting it to the database through software; S74, adding the currently processed data to the relevant table of the database, querying the database, filtering out the historical data associated with the currently processed category substrate, and performing a deletion operation.

Images