Chip transmission and packaging system and method based on electrostatic protection

Abstract

The invention discloses a chip transmission and packaging system and method based on electrostatic protection, and relates to the field of semiconductors. Chip packaging technological processes are divided, historical electrostatic data are collected through a sensor, and a charge quantity risk threshold value is set according to the historical electrostatic data of damaged chips; constructing a charge quantity prediction model by using a polynomial algorithm, predicting the charge quantity generated by each process flow by using the charge quantity prediction model according to the voltage value acquired by the sensor, and judging according to the sum of the charge quantities predicted by the two vertically associated process flows and a charge quantity risk threshold value; the method comprises the steps of judging whether a chip is packaged or not, performing different types of operations according to a judgment result, obtaining specific parameters through traceability analysis when a prediction result does not accord with an actual result, and dynamically updating a prediction model according to the parameters, thereby achieving the precise static monitoring, early warning and protection of the whole chip packaging process, and improving the yield and efficiency of chip packaging.

Description

Technical Field

[0001] The present invention relates to the field of semiconductors, and in particular to a chip transmission and packaging system and method based on electrostatic protection. Background Art

[0002] In the chip packaging field, traditional chip packaging technology relies solely on requiring technicians to wear anti-static rings or performing static elimination operations step by step during the chip packaging process for ESD protection. This lacks detailed monitoring of ESD throughout the entire process. Traditional technologies are mostly passive protection, typically only addressing ESD damage after it has already caused it. They lack the ability to predict ESD risks in advance and are unable to adjust protection strategies based on dynamic changes in the process. This results in inadequate and inefficient ESD protection, which can easily lead to low chip yields. The present invention proposes a chip transmission and packaging technology based on electrostatic protection. Through the systematic division of the process flow and sensor deployment, it realizes real-time monitoring of electrostatic data in the entire chip packaging process flow. According to the prediction model constructed based on historical data, it can predict whether the next process flow will cause chip damage due to static electricity after the current process flow is completed, and predict the electrostatic risk of the next process flow in advance. The prediction model is updated in real time through traceability analysis, which adapts to the dynamic adjustment of the process flow, effectively solves the passive electrostatic protection defects in traditional chip packaging technology, and improves the efficiency and yield of chip packaging. Summary of the Invention

[0003] The purpose of the present invention is to provide a chip transmission and packaging system and method based on electrostatic protection to solve the problems raised in the prior art.

[0004] To achieve the above-mentioned object, the present invention provides the following technical solution: a chip transmission and packaging method based on electrostatic protection, the chip transmission and packaging method comprising the following steps: Step S1: Divide the chip packaging process flow, set sensors in each process flow, use the sensors to collect electrostatic data of chip packaging in each process flow, extract electrostatic data of chip damage caused by static electricity, extract the charge amount, and set the charge amount risk threshold; Step S1-1, dividing the chip packaging process into wafer dicing, chip mounting, wire bonding, and plastic encapsulation; Step S1-2: Setting a sensor in each process flow and using the sensor to collect historical chip packaging electrostatic data of each process flow; Step S1-3: Based on the historical electrostatic data of chip packaging, extract the electrostatic data of chip damage caused by static electricity, extract the charge amount of each process flow, and set the charge amount risk threshold for each process flow, specifically: Step S1-3-1: Construct a charge data set for each process flow ; Charge data set 、 、 The amount of charge carried by the chips processed by the 1st, 2nd, and nth process flows; Step S1-3-2: Use the box plot method to eliminate the outliers in the charge data set constructed for each process flow. The outlier determination is specifically as follows: After the charge data set is sorted in ascending order, the value at the 25% position is the first quartile; After the charge data set is sorted in ascending order, the value at the 75th percentile is the third quartile; ; Where L is the lower threshold of the outlier, Q1 is the first quartile, and IQR is the interquartile range; ; Where U is the upper threshold of the outlier, Q3 is the third quartile, and IQR is the interquartile range; When the charge amount is less than the lower threshold or the charge amount is greater than the upper threshold, the charge amount data is determined to be an abnormal value; Step S1-3-3: Use the median filling method to fill in the missing values ​​of the charge data set constructed for each process flow; Step S1-3-4, setting the minimum charge value in the charge data set of chip damage caused by static electricity as the risk threshold; This step divides the chip packaging process, collects and processes electrostatic data, and sets risk thresholds, providing an accurate and quantifiable benchmark for subsequent electrostatic risk monitoring. It clarifies the electrostatic safety critical values ​​of each process and effectively identifies potential electrostatic risks.

[0005] Step S2: Based on the collected historical chip packaging electrostatic data, a scatter plot of the charge is drawn and then a polynomial algorithm is used to construct a charge prediction model for each process flow; Step S2-1: For each independent process flow of chip packaging, mark all processed historical sample data as scattered points in the coordinate system with charge as the vertical axis and voltage as the horizontal axis, and draw a scatter plot for each process flow; Step S2-2, set the process flow to P, for each process flow P i , the charge prediction model is constructed using the polynomial algorithm, specifically: ; Where, is the i-th process P i The predicted charge, 、 、 ,..., Process P i The polynomial coefficients are: V is the given voltage; Step S2-3: Solve the polynomial coefficients using the least squares method, specifically: ; In the formula, min is the minimum value of the sum of squared errors, is the cumulative error of all historical samples of the i-th process, is the error term, is the actual measured charge of the jth chip in the i-th process flow, To predict the charge; Step S2-4, calculating the polynomial coefficients of the model using the least squares method, and then inputting the polynomial coefficients into the algorithm model to construct a charge prediction model; A scatter plot is drawn based on the processed data, and a charge prediction model is constructed using a polynomial algorithm. A quantitative relationship between charge and voltage is established, and the function of predicting charge through voltage is realized. This also enables the ability to predict changes in electrostatic data in various chip packaging process flows, providing model support for real-time monitoring.

[0006] Step S3: During the chip packaging process, use a sensor to collect the voltage values ​​generated on the chip during each process flow; Step S3-1: number the sensors for each process of chip packaging and deploy sensor monitoring in each process; Step S3-2: using sensors to collect the voltage of each chip in each process flow and the amount of charge generated when the process flow performs chip packaging operations on the chip.

[0007] By numbering the sensors and deploying them in various process flows, the voltage and charge of each chip are collected in real time, ensuring real-time data acquisition, providing data support for the prediction model, and ensuring the timeliness of the prediction.

[0008] Step S4: using a charge prediction model to predict the charge generated by each process flow based on the voltage of each process flow collected by the sensor, and performing a judgment based on the sum of the predicted charge amounts of the two upper and lower related process flows and comparing it with the charge risk threshold, and selecting different operations based on the judgment result; Step S4-1: When the chip starts the chip packaging operation, the voltage value of the first process flow is collected by the sensor in the first process flow, and the charge amount generated by the first process flow is predicted according to the voltage value using the charge amount prediction model; Step S4-2: After each process is completed, at the beginning of the next process, the voltage value of the chip at the beginning of the next process is collected, and the charge value at the beginning of the process is predicted using the charge value prediction model based on the voltage value. The charge value predicted at the beginning of the process is added to the charge value predicted by the previous process, and the sum is compared with the charge value risk threshold for determination; Step S4-3: Repeat the process in the order of the process flow to obtain the accumulated charge of each process flow and make a judgment. When the accumulated charge of the corresponding process flow is greater than the charge risk threshold, perform the static elimination operation. When it is less than the risk threshold, continue to execute the current process flow; By comparing the accumulated charge with the risk threshold, a dynamic assessment of electrostatic risk is achieved, thereby improving chip packaging efficiency and yield.

[0009] Step S5: When the charge quantity prediction value does not conform to the actual charge quantity collected value, a source analysis is performed on the charge quantity anomaly, and then the charge quantity prediction model is updated according to the source analysis result.

[0010] Step S5-1: Professionals set a deviation threshold range [a, b] between the predicted charge amount and the actual charge amount according to the business scenario; Step S5-2: When the deviation between the predicted charge value and the actual collected charge value does not meet the threshold range, it is determined that the charge prediction model is abnormal, and the traceability analysis process is started; Step S5-3: Collect the charge and the voltage corresponding to the charge, determine the process flow to which the abnormality belongs by the sensor number, and then store the collected abnormal charge value and the corresponding voltage; Step S5-4. Professionals set a number threshold x based on the business scenario. When the number of times that the charge prediction value and the actual collected charge value do not meet the threshold interval exceeds x, it is determined to be a process adjustment. The charge and voltage values ​​are input into the charge prediction model constructed by the polynomial algorithm, and the polynomial coefficients of the charge prediction model of the process are recalculated. Finally, the polynomial coefficients are substituted into the charge prediction model constructed by the polynomial algorithm to obtain a new charge prediction model.

[0011] By setting a prediction deviation threshold, tracing analysis is performed when the deviation between the predicted value and the actual value is too large, and the model is updated when the number of abnormalities exceeds the limit, ensuring that the prediction model can adapt to process adjustments, continuously ensuring accuracy, and improving the long-term effectiveness and adaptability of the system.

[0012] The chip transmission and packaging system includes a risk threshold setting module, a prediction model construction module, a real-time data acquisition module, a charge risk determination module, and a model update optimization module; The risk threshold setting module is used to divide the chip packaging process flow, collect historical electrostatic data and set the charge risk threshold; The prediction model building module is used to build a charge prediction model based on historical electrostatic data; The real-time data acquisition module is used to number and deploy sensors for each process flow during the chip packaging process, and collect the voltage of each chip in each process flow and the charge generated by the packaging operation through the sensors; The charge risk determination module is responsible for predicting the charge generated by each process flow using a charge prediction model, comparing the sum of the predicted charge amounts of the upper and lower related process flows with the risk threshold, and performing static elimination when the predicted charge amounts of the upper and lower related process flows are greater than the risk threshold; otherwise, the current process flow continues to execute; The model update optimization module is used to start tracing analysis to determine the process flow to which the anomaly belongs and store relevant data when the deviation between the charge quantity prediction value and the actual collected value of the sensor exceeds a threshold range. When the number of anomalies exceeds a threshold, the charge quantity prediction model of the process flow is updated with relevant data.

[0013] The risk threshold setting module includes a data processing unit and a threshold setting unit; The data processing unit is used to divide the chip packaging process flow, set sensors in each process to collect historical electrostatic data, and process the charge data set; The threshold setting unit is used to set the minimum charge value as the risk threshold of each process flow from the charge amount data of the chip damaged by static electricity.

[0014] The prediction model building module includes a scatter plot drawing unit and a polynomial modeling unit; The scatter plot drawing unit is used for marking historical sample data as scatter points and drawing a scatter plot for each process flow, with charge as the vertical axis and voltage as the horizontal axis; The polynomial modeling unit is used to construct a charge quantity prediction model through a polynomial algorithm and solve the polynomial coefficients through a least squares method; The real-time data acquisition module includes a sensor deployment unit and a voltage acquisition unit; The sensor deployment unit is used to number and deploy sensors for each process flow to monitor the chip packaging process; The voltage acquisition unit is used to use a sensor to acquire the voltage generated by each chip in each process flow.

[0015] The charge risk determination module includes a charge prediction and accumulation unit and a risk response unit; The charge prediction and accumulation unit is used to obtain the charge amount prediction value of each process flow according to the voltage collected by the sensor using the prediction model, and add the charge amounts predicted by the two process flows that are connected vertically; The risk response unit is used to perform an electrostatic elimination operation when the sum of the charge amounts predicted by the two process flows associated with each other exceeds a charge amount risk threshold, and to continue to perform the current process flow when the sum is less than the charge amount risk threshold; The model update optimization module includes an anomaly tracing unit and a model update unit; The abnormality tracing unit is used to set a deviation threshold interval, and when it is determined that the charge amount prediction value is inconsistent with the actual collected value, the charge amount is stored and traced and analyzed; The model updating unit is used to input relevant data into the model to recalculate the polynomial and update the charge quantity prediction model when the number of times that the charge quantity pre-test and the actual collected value are inconsistent exceeds a threshold value.

[0016] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention divides the chip packaging process into standardized sections, systematically collects and processes electrostatic data, and sets thresholds based on the aggregated data, providing accurate and unified judgment criteria for electrostatic prevention and control in each process, thereby ensuring the scientificity and reliability of risk identification.

[0017] 2. The present invention constructs a charge prediction model through a polynomial algorithm. Combined with the voltage data collected by real-time sensors, it can predict the charge of each process, dynamically assess risks and initiate static elimination operations in a timely manner, thereby improving the yield rate of chip packaging.

[0018] 3. The present invention sets a deviation threshold and a number threshold between the predicted charge value and the actual collected value, performs a traceability analysis operation and updates the model when a prediction deviation occurs in the prediction model, so that the prediction model can adapt to changes in the process flow, continuously maintain a high prediction accuracy, and ensure that the system can dynamically adapt to adjustments to the process flow. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 This is a flow chart of a chip transmission and packaging method based on electrostatic protection according to the present invention; Figure 2 The figure is a schematic structural diagram of a chip transmission and packaging system based on electrostatic protection according to the present invention. DETAILED DESCRIPTION

[0020] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only 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 making creative efforts are within the scope of protection of the present invention.

[0021] Example: Figure 1-Figure 2 As shown, the present invention provides a technical solution, a chip transmission and packaging method based on electrostatic protection, the chip transmission and packaging method comprises the following steps: Step S1: Divide the chip packaging process flow, set sensors in each process flow, use the sensors to collect electrostatic data of chip packaging in each process flow, extract electrostatic data of chip damage caused by static electricity, extract the charge amount, and set the charge amount risk threshold; Step S1-1, dividing the chip packaging process into wafer dicing, chip mounting, wire bonding, and plastic encapsulation; Step S1-2: Setting a sensor in each process flow and using the sensor to collect historical chip packaging electrostatic data of each process flow; Step S1-3: Based on the historical electrostatic data of chip packaging, extract the electrostatic data of chip damage caused by static electricity, extract the charge amount of each process flow, and set the charge amount risk threshold for each process flow, specifically: Step S1-3-1: Construct a charge data set for each process flow ; Charge data set 、 、 The amount of charge carried by the chips processed by the 1st, 2nd, and nth process flows; Step S1-3-2: Use the box plot method to eliminate the outliers in the charge data set constructed for each process flow. The outlier determination is specifically as follows: After the charge data set is sorted in ascending order, the value at the 25% position is the first quartile; After the charge data set is sorted in ascending order, the value at the 75th percentile is the third quartile; ; Where L is the lower threshold of the outlier, Q1 is the first quartile, and IQR is the interquartile range; ; Where U is the upper threshold of the outlier, Q3 is the third quartile, and IQR is the interquartile range; When the charge amount is less than the lower threshold or the charge amount is greater than the upper threshold, the charge amount data is determined to be an abnormal value; Step S1-3-3: Use the median filling method to fill in the missing values ​​of the charge data set constructed for each process flow; Step S1-3-4, setting the minimum charge value in the charge data set of chip damage caused by static electricity as the risk threshold; This step divides the chip packaging process, collects and processes electrostatic data, and sets risk thresholds, providing an accurate and quantifiable benchmark for subsequent electrostatic risk monitoring. It clarifies the electrostatic safety critical values ​​of each process and effectively identifies potential electrostatic risks.

[0022] Step S2: Based on the collected historical chip packaging electrostatic data, a scatter plot of the charge is drawn and then a polynomial algorithm is used to construct a charge prediction model for each process flow; Step S2-1: For each independent process flow of chip packaging, mark all processed historical sample data as scattered points in the coordinate system with charge as the vertical axis and voltage as the horizontal axis, and draw a scatter plot for each process flow; Step S2-2, set the process flow to P, for each process flow P i , the charge prediction model is constructed using the polynomial algorithm, specifically: ; Where, is the i-th process P i The predicted charge, 、 、 ,..., Process P i The polynomial coefficients are: V is the given voltage; Step S2-3: Solve the polynomial coefficients using the least squares method, specifically: ; In the formula, min is the minimum value of the sum of squared errors, is the cumulative error of all historical samples of the i-th process, is the error term, is the actual measured charge of the jth chip in the i-th process flow, To predict the charge; Step S2-4, calculating the polynomial coefficients of the model using the least squares method, and then inputting the polynomial coefficients into the algorithm model to construct a charge prediction model; A scatter plot is drawn based on the processed data, and a charge prediction model is constructed using a polynomial algorithm. A quantitative relationship between charge and voltage is established, and the function of predicting charge through voltage is realized. This also enables the ability to predict changes in electrostatic data in various chip packaging process flows, providing model support for real-time monitoring.

[0023] Step S3: During the chip packaging process, use a sensor to collect the voltage values ​​generated on the chip during each process flow; Step S3-1: number the sensors for each process of chip packaging and deploy sensor monitoring in each process; Step S3-2: using sensors to collect the voltage of each chip in each process flow and the amount of charge generated when the process flow performs chip packaging operations on the chip.

[0024] By numbering the sensors and deploying them in various process flows, the voltage and charge of each chip are collected in real time, ensuring real-time data acquisition, providing data support for the prediction model, and ensuring the timeliness of the prediction.

[0025] Step S4: using a charge prediction model to predict the charge generated by each process flow based on the voltage of each process flow collected by the sensor, and performing a judgment based on the sum of the predicted charge amounts of the two upper and lower related process flows and comparing it with the charge risk threshold, and selecting different operations based on the judgment result; Step S4-1: When the chip starts the chip packaging operation, the voltage value of the first process flow is collected by the sensor in the first process flow, and the charge amount generated by the first process flow is predicted according to the voltage value using the charge amount prediction model; Step S4-2: After each process is completed, at the beginning of the next process, the voltage value of the chip at the beginning of the next process is collected, and the charge value at the beginning of the process is predicted using the charge value prediction model based on the voltage value. The charge value predicted at the beginning of the process is added to the charge value predicted by the previous process, and the sum is compared with the charge value risk threshold for determination; Step S4-3: Repeat the process in the order of the process flow to obtain the accumulated charge of each process flow and make a judgment. When the accumulated charge of the corresponding process flow is greater than the charge risk threshold, perform the static elimination operation. When it is less than the risk threshold, continue to execute the current process flow; By comparing the accumulated charge with the risk threshold, a dynamic assessment of electrostatic risk is achieved, thereby improving chip packaging efficiency and yield.

[0026] Step S5: When the charge quantity prediction value does not conform to the actual charge quantity collected value, a source analysis is performed on the charge quantity anomaly, and then the charge quantity prediction model is updated according to the source analysis result.

[0027] Step S5-1: Professionals set a deviation threshold range [a, b] between the predicted charge amount and the actual charge amount according to the business scenario; Step S5-2: When the deviation between the predicted charge value and the actual collected charge value does not meet the threshold range, it is determined that the charge prediction model is abnormal, and the traceability analysis process is started; Step S5-3: Collect the charge and the voltage corresponding to the charge, determine the process flow to which the abnormality belongs by the sensor number, and then store the collected abnormal charge value and the corresponding voltage; Step S5-4. Professionals set a number threshold x based on the business scenario. When the number of times that the charge prediction value and the actual collected charge value do not meet the threshold interval exceeds x, it is determined to be a process adjustment. The charge and voltage values ​​are input into the charge prediction model constructed by the polynomial algorithm, and the polynomial coefficients of the charge prediction model of the process are recalculated. Finally, the polynomial coefficients are substituted into the charge prediction model constructed by the polynomial algorithm to obtain a new charge prediction model.

[0028] By setting a prediction deviation threshold, tracing analysis is performed when the deviation between the predicted value and the actual value is too large, and the model is updated when the number of abnormalities exceeds the limit, ensuring that the prediction model can adapt to process adjustments, continuously ensuring accuracy, and improving the long-term effectiveness and adaptability of the system.

[0029] The chip transmission and packaging system includes a risk threshold setting module, a prediction model construction module, a real-time data acquisition module, a charge risk determination module, and a model update optimization module; The risk threshold setting module is used to divide the chip packaging process flow, collect historical electrostatic data and set the charge risk threshold; The prediction model building module is used to build a charge prediction model based on historical electrostatic data; The real-time data acquisition module is used to number and deploy sensors for each process flow during the chip packaging process, and collect the voltage of each chip in each process flow and the charge generated by the packaging operation through the sensors; The charge risk determination module is responsible for predicting the charge generated by each process flow using a charge prediction model, comparing the sum of the predicted charge amounts of the upper and lower related process flows with the risk threshold, and performing static elimination when the predicted charge amounts of the upper and lower related process flows are greater than the risk threshold; otherwise, the current process flow continues to execute; The model update optimization module is used to start tracing analysis to determine the process flow to which the anomaly belongs and store relevant data when the deviation between the charge quantity prediction value and the actual collected value of the sensor exceeds a threshold range. When the number of anomalies exceeds a threshold, the charge quantity prediction model of the process flow is updated with relevant data.

[0030] The risk threshold setting module includes a data processing unit and a threshold setting unit; The data processing unit is used to divide the chip packaging process flow, set sensors in each process to collect historical electrostatic data, and process the charge data set; The threshold setting unit is used to set the minimum charge value as the risk threshold of each process flow from the charge amount data of the chip damaged by static electricity.

[0031] The prediction model building module includes a scatter plot drawing unit and a polynomial modeling unit; The scatter plot drawing unit is used for marking historical sample data as scatter points and drawing a scatter plot for each process flow, with charge as the vertical axis and voltage as the horizontal axis; The polynomial modeling unit is used to construct a charge quantity prediction model through a polynomial algorithm and solve the polynomial coefficients through a least squares method; The real-time data acquisition module includes a sensor deployment unit and a voltage acquisition unit; The sensor deployment unit is used to number and deploy sensors for each process flow to monitor the chip packaging process; The voltage acquisition unit is used to use a sensor to acquire the voltage generated by each chip in each process flow.

[0032] The charge risk determination module includes a charge prediction and accumulation unit and a risk response unit; The charge prediction and accumulation unit is used to obtain the charge amount prediction value of each process flow according to the voltage collected by the sensor using the prediction model, and add the charge amounts predicted by the two process flows that are connected vertically; The risk response unit is used to perform an electrostatic elimination operation when the sum of the charge amounts predicted by the two process flows associated with each other exceeds a charge amount risk threshold, and to continue to perform the current process flow when the sum is less than the charge amount risk threshold; The model update optimization module includes an anomaly tracing unit and a model update unit; The abnormality tracing unit is used to set a deviation threshold interval, and when it is determined that the charge amount prediction value is inconsistent with the actual collected value, the charge amount is stored and traced and analyzed; The model updating unit is used to input relevant data into the model to recalculate the polynomial and update the charge quantity prediction model when the number of times that the charge quantity pre-test and the actual collected value are inconsistent exceeds a threshold value.

[0033] Example 1: Divide chip packaging into four process flows: wafer dicing, chip mounting, wire bonding, and plastic encapsulation; The historical electrostatic data of each process is collected through sensors, and the charge data set of a certain process is: P1[5,8,12,10,15,50,18,25,30,(missing values),16,13,40,(outliers)] P2[3,6,9,12,15,18,21,24,27,30,33,36,39] The data set of charge amount caused by static electricity to damage the chip is: [28,32,35,40] The sorted data set P1 is [5,8,10,12,13,15,16,18,20,25,30,40] According to the box plot method, Q1 is 10, Q3 is 20, and IQR is 10; The lower limit L=5, the upper limit U=35, 40>35, the charge is judged as an abnormal value, and the P1 data set after elimination is [5,8,10,12,13,15,16,18,20,25,30] The missing value filling method is used to fill the P1 data set, and the median is calculated to be 15. The final P1 data set is: [5,8,10,12,13,15,15,16,18,20,25,30] Based on the minimum value, the charge risk threshold is 28.

[0034] Draw a scatter plot. The sample data of the scatter plot are: (2,5), (3,8), (4,10), (5,12), (6,13), (7,15), (8,16), (9,18), (10,20), (11,25), (12,30).

[0035] For P1, the coefficients of the quadratic polynomial model are solved by the least squares method, and the P1 model is obtained as follows: ; The P2 model is: ; The chip voltage data is collected using a sensor. P1 is 5V and P2 is 7V. Substituting this into the model, the charge generated by P1 is 7.7 and the charge generated by P2 is 9.02. The sum does not exceed the risk threshold, so the process continues. Prediction and judgment are then performed in sequence. When the cumulative sum exceeds the risk threshold, static elimination is performed. Set the threshold deviation interval [a, b] = [-1.2, 1.2], the abnormal number threshold x = 3, the actual charge of P1 is 8.9, the predicted value is 7.7, the deviation is 1.2, and it is judged as abnormal; When the number of abnormalities accumulates to 4 times, the process flow is adjusted according to the threshold, and the new data is substituted into the polynomial algorithm to recalculate the coefficients and update the model.

[0036] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims, not the foregoing description, and all variations within the meaning and range of equivalents of the claims are intended to be included therein. Any reference sign in a claim should not be construed as limiting the claim to which it relates.

Claims

1. A chip transmission and packaging method based on electrostatic protection, characterized by: The chip transmission and packaging method comprises the following steps: Step S1: Divide the chip packaging process flow, set sensors in each process flow, use the sensors to collect electrostatic data of chip packaging in each process flow, extract electrostatic data of chip damage caused by static electricity, extract the charge amount, and set the charge amount risk threshold; Step S2: Based on the collected historical chip packaging electrostatic data, a scatter plot of the charge is drawn and then a polynomial algorithm is used to construct a charge prediction model for each process flow; Step S3: During the chip packaging process, use a sensor to collect the voltage values ​​generated on the chip during each process flow; Step S4: using a charge prediction model to predict the charge generated by each process flow based on the voltage of each process flow collected by the sensor, and performing a judgment based on the sum of the predicted charge amounts of the two upper and lower related process flows and comparing it with the charge risk threshold, and selecting different operations based on the judgment result; Step S5: When the charge quantity prediction value does not conform to the actual charge quantity collected value, a source analysis is performed on the charge quantity anomaly, and then the charge quantity prediction model is updated according to the source analysis result.

2. The chip transmission and packaging method based on electrostatic protection according to claim 1, characterized in that: The specific steps of step S1 are as follows: Step S1-1, dividing the chip packaging process into wafer dicing, chip mounting, wire bonding, and plastic encapsulation; Step S1-2: Setting a sensor in each process flow and using the sensor to collect historical chip packaging electrostatic data of each process flow; Step S1-3: Based on the historical electrostatic data of chip packaging, extract the electrostatic data of chip damage caused by static electricity, extract the charge amount of each process flow, and set the charge amount risk threshold for each process flow, specifically: Step S1-3-1: Construct a charge data set for each process flow ; Charge data set 、 、 The amount of charge carried by the chips processed by the 1st, 2nd, and nth process flows; Step S1-3-2: Use the box plot method to eliminate the outliers in the charge data set constructed for each process flow. The outlier determination is specifically as follows: After the charge data set is sorted in ascending order, the value at the 25% position is the first quartile; After the charge data set is sorted in ascending order, the value at the 75th percentile position is the third quartile; ； Where L is the lower threshold of the outlier, Q1 is the first quartile, and IQR is the interquartile range; ； Where U is the upper threshold of the outlier, Q3 is the third quartile, and IQR is the interquartile range; When the charge amount is less than the lower threshold or the charge amount is greater than the upper threshold, the charge amount data is determined to be an abnormal value; Step S1-3-3: using the median filling method to fill in the missing values ​​of the charge data set constructed for each process flow; Step S1-3-4: setting the minimum charge value in the chip damage charge data set caused by static electricity as the risk threshold.

3. The chip transmission and packaging method based on electrostatic protection according to claim 2, characterized in that: The specific steps of step S2 are as follows: Step S2-1: For each independent process flow of chip packaging, mark all processed historical sample data as scattered points in the coordinate system with charge as the vertical axis and voltage as the horizontal axis, and draw a scatter plot for each process flow; Step S2-2, set the process flow to P, for each process flow P i , the charge prediction model is constructed using the polynomial algorithm, specifically: ； Where, is the i-th process P i The predicted charge, 、 、 ,..., Process P i The polynomial coefficients are: V is the given voltage; Step S2-3: Solve the polynomial coefficients using the least squares method, specifically: ； In the formula, min is the minimum value of the sum of squared errors, is the cumulative error of all historical samples of the i-th process, is the error term, is the actual measured charge of the jth chip in the i-th process flow, To predict the charge; Step S2-4: Calculate the polynomial coefficients of the model using the least squares method, and then input the polynomial coefficients into the algorithm model to construct a charge prediction model.

4. The chip transmission and packaging method based on electrostatic protection according to claim 3, characterized in that: The specific steps of step S3 are as follows: Step S3-1: number the sensors for each process of chip packaging and deploy sensor monitoring in each process; Step S3-2: using sensors to collect the voltage of each chip in each process flow and the amount of charge generated when the process flow performs chip packaging operations on the chip.

5. The chip transmission and packaging method based on electrostatic protection according to claim 4, characterized in that: The specific steps of step S4 are as follows: Step S4-1: When the chip starts the chip packaging operation, the voltage value of the first process flow is collected by the sensor in the first process flow, and the charge amount generated by the first process flow is predicted according to the voltage value using the charge amount prediction model; Step S4-2: After each process is completed, at the beginning of the next process, the voltage value of the chip at the beginning of the next process is collected, and the charge value at the beginning of the process is predicted using the charge value prediction model based on the voltage value. The charge value predicted at the beginning of the process is added to the charge value predicted by the previous process, and the sum is compared with the charge value risk threshold for determination; Step S4-3, repeat the operation in the order of the process flow to obtain the amount of charge accumulated in each process flow and make a judgment. When the amount of charge accumulated in the corresponding process flow is greater than the charge risk threshold, perform the static elimination operation. When it is less than the risk threshold, continue to execute the current process flow.

6. The chip transmission and packaging method based on electrostatic protection according to claim 5, characterized in that: The specific steps of step S5 are as follows: Step S5-1: Professionals set a deviation threshold range [a, b] between the predicted charge amount and the actual charge amount according to the business scenario; Step S5-2: When the deviation between the predicted charge value and the actual collected charge value does not meet the threshold range, it is determined that the charge prediction model is abnormal, and the traceability analysis process is started; Step S5-3: Collect the charge and the voltage corresponding to the charge, determine the process flow to which the abnormality belongs by the sensor number, and then store the collected abnormal charge value and the corresponding voltage; Step S5-4. Professionals set a number threshold x based on the business scenario. When the number of times that the charge prediction value and the actual collected charge value do not meet the threshold interval exceeds x, it is determined to be a process adjustment. The charge and voltage values ​​are input into the charge prediction model constructed by the polynomial algorithm, and the polynomial coefficients of the charge prediction model of the process are recalculated. Finally, the polynomial coefficients are substituted into the charge prediction model constructed by the polynomial algorithm to obtain a new charge prediction model.

7. A chip transmission and packaging system based on electrostatic protection, characterized by: The chip transmission and packaging system includes a risk threshold setting module, a prediction model building module, a real-time data acquisition module, a charge risk determination module, and a model update optimization module; The risk threshold setting module is used to divide the chip packaging process flow, collect historical electrostatic data and set the charge risk threshold; The prediction model building module is used to build a charge prediction model based on historical electrostatic data; The real-time data acquisition module is used to number and deploy sensors for each process flow during the chip packaging process, and collect the voltage of each chip in each process flow and the charge generated by the packaging operation through the sensors; The charge risk determination module is responsible for predicting the charge generated by each process flow using a charge prediction model, comparing the sum of the predicted charge amounts of the upper and lower related process flows with the risk threshold, and performing static elimination when the predicted charge amounts of the upper and lower related process flows are greater than the risk threshold; otherwise, the current process flow continues to execute; The model update optimization module is used to start tracing analysis to determine the process flow to which the anomaly belongs and store relevant data when the deviation between the charge quantity prediction value and the actual collected value of the sensor exceeds a threshold range. When the number of anomalies exceeds a threshold, the charge quantity prediction model of the process flow is updated with relevant data.

8. The chip transmission and packaging system based on electrostatic protection according to claim 7, characterized in that: The risk threshold setting module includes a data processing unit and a threshold setting unit; The data processing unit is used to divide the chip packaging process flow, set sensors in each process to collect historical electrostatic data, and process the charge data set; The threshold setting unit is used to set the minimum charge value as the risk threshold of each process flow from the charge amount data of the chip damaged by static electricity.

9. The chip transmission and packaging system based on electrostatic protection according to claim 7, characterized in that: The prediction model building module includes a scatter plot drawing unit and a polynomial modeling unit; The scatter plot drawing unit is used for marking historical sample data as scatter points and drawing a scatter plot for each process flow, with charge as the vertical axis and voltage as the horizontal axis; The polynomial modeling unit is used to construct a charge quantity prediction model through a polynomial algorithm and solve the polynomial coefficients through a least squares method; The real-time data acquisition module includes a sensor deployment unit and a voltage acquisition unit; The sensor deployment unit is used to number and deploy sensors for each process flow to monitor the chip packaging process; The voltage acquisition unit is used to use a sensor to acquire the voltage generated by each chip in each process flow.

10. The chip transmission and packaging system based on electrostatic protection according to claim 7, characterized in that: The charge risk determination module includes a charge prediction and accumulation unit and a risk response unit; The charge prediction and accumulation unit is used to obtain the charge amount prediction value of each process flow according to the voltage collected by the sensor using the prediction model, and add the charge amounts predicted by the two process flows that are connected vertically; The risk response unit is used to perform an electrostatic elimination operation when the sum of the charge amounts predicted by the two process flows associated with each other exceeds a charge amount risk threshold, and to continue to perform the current process flow when the sum is less than the charge amount risk threshold; The model update optimization module includes an anomaly tracing unit and a model update unit; The abnormality tracing unit is used to set a deviation threshold interval, and when it is determined that the charge amount prediction value is inconsistent with the actual collected value, the charge amount is stored and traced and analyzed; The model updating unit is used to input relevant data into the model to recalculate the polynomial and update the charge quantity prediction model when the number of times that the charge quantity pre-test and the actual collected value are inconsistent exceeds a threshold value.

Images