Simulation parameter optimization method for photon de-bonding of wafer level packaging

Abstract

The invention relates to the field of temperature control, in particular to a photon de-bonding simulation parameter optimization method for wafer level packaging, which comprises the following steps: acquiring a value range of photon de-bonding simulation parameters, performing grid division according to the value range to obtain a grid disc, and enabling each node of the grid disc to correspond to a parameter group; constructing an objective function of the ant colony algorithm, for each iteration in the ant colony algorithm, calculating a pheromone dynamic volatilization coefficient of a node where each ant is located to control the pheromone attenuation rate of the next iteration, screening all the parameter groups in the iteration process to obtain a global optimal parameter group, and completing parameter optimization. And the global optimal parameter group is used for temperature regulation and control of photon de-bonding. According to the method, the optimal simulation parameters can be obtained through the ant colony algorithm, so that the accuracy of regulating and controlling the photon de-bonding temperature is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of temperature control. More particularly, the present application relates to a simulation parameter optimization method for photonic debonding of wafer level packaging. BACKGROUND

[0002] In the wafer level packaging process, photonic debonding refers to using ultraviolet light or laser irradiation to make the temporary bonding glue layer lose adhesion through photochemical or photothermal reaction, so as to separate the wafer from the carrier board. In this process, the temperature needs to be accurately controlled to ensure safe and uniform photonic debonding. The prior art usually uses a control algorithm to calculate the temperature control value, and the algorithm parameters in the control algorithm are very important to the performance of the control algorithm. Suitable parameters can enable the algorithm to quickly and accurately track the set value and avoid overshoot and oscillation. However, improper parameters may cause slow system response, large steady-state error or even instability.

[0003] The traditional technology uses an ant colony algorithm to obtain optimal simulation parameters to achieve temperature control of photonic debonding. However, the pheromone evaporation coefficient of the traditional ant colony algorithm is a fixed value. Since the ant colony algorithm is an iterative dynamic process, the fixed pheromone evaporation coefficient cannot adapt to different iteration stages, which causes the ant colony algorithm to easily fall into a local optimal solution or have insufficient global search capability, and thus the optimal simulation parameters cannot be obtained, affecting the temperature control of photonic debonding. SUMMARY

[0004] The main purpose of the embodiments of the present application is to propose a simulation parameter optimization method for photonic debonding of wafer level packaging, which aims to obtain optimal simulation parameters through an ant colony algorithm to improve the accuracy of controlling the temperature of photonic debonding.

[0005] To achieve the above object, the application provides a simulation parameter optimization method for wafer-level packaged photonic debonding, which comprises the following steps: obtaining a value range of simulation parameters of photonic debonding, performing grid division according to the value range to obtain a grid disc, each node of the grid disc corresponding to a parameter group; constructing an objective function of an ant colony algorithm, for each iteration of the ant colony algorithm, calculating a pheromone dynamic evaporation coefficient of a node where each ant is located to control a pheromone decay rate of the next iteration, screening all the parameter groups during the iteration process to obtain a globally optimal parameter group, completing parameter optimization, and the globally optimal parameter group is used for temperature regulation of photonic debonding by a PID control algorithm; wherein the calculation of the pheromone dynamic evaporation coefficient of the node where each ant is located comprises: calculating a target function value of the parameter group corresponding to the node where each ant is located, taking any ant as a target ant, calculating a good-bad degree of the node where the target ant is located according to a ratio of a minimum value of the target function value of the current iteration to the target function value of the target ant, calculating a convergence degree of the current iteration, taking a product of the good-bad degree, the convergence degree and an initial pheromone evaporation coefficient of the target ant as the pheromone dynamic evaporation coefficient of the target ant, and traversing to obtain the pheromone dynamic evaporation coefficient of the node where each ant is located.

[0006] In some embodiments, the simulation parameters comprise a proportional gain parameter, an integral gain parameter and a differential gain parameter of a PID control algorithm.

[0007] In some embodiments, the construction of the objective function of the ant colony algorithm comprises: obtaining a real-time temperature of photonic debonding, taking a difference between the real-time temperature and a preset temperature as a temperature deviation, and calculating a temperature adjustment amount by using a PID control algorithm; and taking a sum of a square integral of the temperature deviation and a square integral of the temperature adjustment amount as the objective function.

[0008] In some embodiments, the calculation of the good-bad degree of the node where the target ant is located comprises: performing a negative correlation mapping on the minimum value of the target function value and the target function value of the target ant, and taking a normalized value of the mapping result as the good-bad degree.

[0009] In some embodiments, the convergence degree comprises: calculating a convergence distance of the current iteration; obtaining a ratio of a preset maximum iteration number to a current iteration number; and taking a normalized result of a product of the ratio and the convergence distance as the convergence degree.

[0010] In some embodiments, the convergence distance of the current iteration is calculated by: taking the parameter group corresponding to the minimum value of the objective function value of the current iteration as a local optimal group of the current iteration, taking the parameter group corresponding to the second minimum value of the objective function value of the current iteration as a local suboptimal group of the current iteration, taking the Euclidean distance between the local optimal group of the current iteration and the local suboptimal group of the current iteration as a first distance; obtaining the local optimal group of the last iteration adjacent to the current iteration, wherein the local optimal groups of the two adjacent iterations are obtained in the same way, taking the Euclidean distance between the local optimal group of the current iteration and the local optimal group of the last iteration as a second distance; taking the parameter group corresponding to the node where the target ant is located as a target group, and taking the distance between the target group and the local optimal group of the current iteration as a third distance; and taking the product of the first distance, the second distance and the third distance as the convergence distance of the current iteration.

[0011] In some embodiments, the initial pheromone evaporation coefficient of each ant is the same.

[0012] In some embodiments, the iteration convergence condition is that the difference between the minimum values of the objective function values of the two adjacent iterations is less than a preset threshold, or the preset iteration number is reached.

[0013] Advantages of the present application: The present application constructs the objective function of the ant colony algorithm, searches for the global optimal parameter group of the objective function in the value range of the simulation parameters by using the ant colony algorithm, updates the initial pheromone evaporation coefficient of each ant according to the ratio of the minimum value of the objective function value of the current iteration to the objective function value corresponding to the node where each ant is located and the convergence degree of the current iteration, and obtains the dynamic pheromone evaporation coefficient of each ant. Therefore, the dynamic pheromone evaporation coefficient of each ant changes with the convergence degree of the iteration and the size of the objective function value of the node where each ant is located relative to the minimum value of the objective function value. Therefore, the present application can dynamically change the pheromone evaporation coefficient according to the iteration stage and the advantages and disadvantages of the node where each ant is located. Compared with the fixed pheromone evaporation coefficient of the traditional ant colony algorithm, the dynamic pheromone evaporation coefficient of the present application is more adaptable, can reduce the probability of the ant colony algorithm falling into a local optimal solution or being insufficient in global search ability, and is more likely to obtain optimal simulation parameters to improve the accuracy of the photonic debonding temperature. BRIEF DESCRIPTION OF DRAWINGS

[0014] Figure 1 is a flowchart of steps S1-S2 in the simulation parameter optimization method for wafer-level packaging photonic debonding according to an embodiment of the present application. DETAILED DESCRIPTION

[0015] Clearly, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments.

[0016] The specific embodiments of the present application will be described in detail below with reference to the accompanying drawings.

[0017] Referring to Figure 1 The simulation parameter optimization method for photonic debonding of wafer level packaging includes steps S1-S2, and specifically as follows: Step S1: Obtain the value range of the simulation parameters of photonic debonding, and perform grid division to obtain a grid disk according to the value range, each node of the grid disk corresponding to a parameter group.

[0018] In some embodiments, the simulation parameters include proportional gain parameters, integral gain parameters and differential gain parameters of the PID control algorithm.

[0019] It should be noted that the value range of the simulation parameters of photonic debonding can be selected according to actual conditions.

[0020] Step S2: Construct the objective function of the ant colony algorithm, for each iteration in the ant colony algorithm, calculate the pheromone dynamic evaporation coefficient of the node where each ant is located to control the pheromone decay rate of the next iteration, filter all parameter groups during the iteration process to obtain a globally optimal parameter group, complete parameter optimization, and the globally optimal parameter group is used for temperature regulation of the photonic debonding by the PID control algorithm.

[0021] Wherein, calculating the pheromone dynamic evaporation coefficient of the node where each ant is located includes: calculating the objective function value of the parameter group corresponding to the node where each ant is located, taking any ant as a target ant, calculating the degree of superiority and inferiority of the node where the target ant is located according to the ratio of the minimum value of the objective function value of the current iteration to the objective function value of the target ant, calculating the convergence degree of the current iteration, and taking the product of the degree of superiority and inferiority, the convergence degree and the initial pheromone evaporation coefficient of the target ant as the pheromone dynamic evaporation coefficient of the target ant. The pheromone dynamic evaporation coefficient of the node where each ant is located is obtained by traversing.

[0022] In some embodiments, constructing the objective function of the ant colony algorithm includes: obtaining the real-time temperature of the photonic debonding, taking the difference between the real-time temperature and the preset temperature as the temperature deviation, calculating the temperature adjustment amount using the PID control algorithm; and taking the sum of the square integral of the temperature deviation and the square integral of the temperature adjustment amount as the objective function.

[0023] It should be further explained that multiple temperature sensors are used to collect real-time temperature data of the heating zone in the photonic debonding process. These sensors are distributed at different locations within the heating zone, and the average of these temperature data is used as the real-time temperature for photonic debonding. The photonic debonding process demands extremely high uniformity and stability of the heating zone temperature. Localized overheating can lead to wafer damage, while temperature unevenness can cause differences in debonding efficiency or process failure. Using multiple temperature sensors to collect the heating zone temperature and averaging the data as the real-time temperature improves the representativeness of temperature monitoring.

[0024] Among them, real-time temperature is used Indicates that the preset temperature is used This indicates that the preset temperature is the stable temperature that needs to be maintained in the photonic debonding process, and the temperature deviation... Temperature adjustment amount For example, the proportional gain parameter Integral gain parameter Differential gain parameter .

[0025] For example, the objective function is as follows: In the formula middle, For the first The objective function value of the parameter set corresponding to the node where the ant is located. For the first Only ants Temperature deviation at any time for Time of the first The temperature adjustment amount for the parameter group corresponding to the node where the ant is located. This is the integral of the square of the temperature deviation. This is the square integral of the temperature adjustment amount.

[0026] For the formula It should be added that, The smaller the value, the smaller the temperature tracking error; The smaller the value, the less energy is required to adjust the temperature error. This is taken after the iteration ends. The parameter set corresponding to the minimum value is taken as the globally optimal parameter set.

[0027] In some embodiments, calculating the quality of the node where the target ant is located includes: performing a negative correlation mapping between the minimum value of the objective function and the objective function value of the target ant, and using the normalized value of the mapping result as the quality.

[0028] In some embodiments, the convergence degree comprises: calculating a convergence distance of the current iteration; obtaining a ratio of the preset maximum iteration number and the current iteration number; and taking a normalized result of the product of the ratio and the convergence distance as the convergence degree.

[0029] In some embodiments, the calculating of the convergence distance of the current iteration comprises: taking a parameter group corresponding to a minimum value of the objective function value of the current iteration as a local optimal group of the current iteration, taking a parameter group corresponding to a second minimum value of the objective function value of the current iteration as a local suboptimal group of the current iteration, and taking a Euclidean distance between the local optimal group of the current iteration and the local suboptimal group of the current iteration as a first distance; obtaining a local optimal group of a last iteration adjacent to the current iteration, wherein the local optimal groups of the adjacent two iterations are obtained in the same way, and taking a Euclidean distance between the local optimal group of the current iteration and the local optimal group of the last iteration as a second distance; taking a parameter group corresponding to a node where the target ant is located as a target group, and taking a distance between the target group and the local optimal group of the current iteration as a third distance; and taking a product of the first distance, the second distance and the third distance as the convergence distance of the current iteration.

[0030] Exemplarily, the calculation formula of the pheromone dynamic evaporation coefficient is as follows: Exemplarily, the calculation formula of the convergence distance is as follows: In the formula , is the current iteration number, is the th iteration number, is the pheromone dynamic evaporation coefficient of the node where the th ant is located in the th iteration, is the initial pheromone evaporation coefficient, is the normalization function, is the exponential function, is the minimum value of the objective function value in the th iteration, is the objective function value of the node where the th ant is located in the th iteration, is the normalization function, is the preset maximum iteration number, is the convergence distance of the

[0031] th iteration. In the formula , is the mapping value, is the degree of superiority and inferiority, wherein 1 in the formula is a hyperparameter and can be adjusted as needed.

[0032] In the formula In addition, it should be noted that the ratio Indicates the first Only ants in the first The ratio of the objective function value represented in the nth iteration to the minimum objective function value in the current iteration reflects the quality of that node. The closer this ratio is to 1, the better the node is. Only ants in the first The smaller the difference between the objective function value represented in the next iteration and the local optimum function value in the current iteration, the smaller the pheromone evaporation coefficient should be at that node to preserve high-quality path information; conversely, the closer the ratio is to 0, the smaller the difference should be at the next iteration. Only ants in the first The greater the difference between the objective function value represented in the next iteration and the local optimum function value in the current iteration, the greater the pheromone evaporation coefficient at that node should be increased to promote exploration.

[0033] In the formula In addition, it should be noted that To preset the ratio of the maximum number of iterations to the current number of iterations, in the early stages of iteration, Larger, correspondingly Larger pheromones evaporate faster, which can promote exploration; conversely, smaller pheromones evaporate faster in later iterations. Smaller, correspondingly Smaller pheromones evaporate more slowly, which can accelerate the elimination of inferior solutions and improve the convergence speed.

[0034] In the formula In addition, it should be noted that... Only less than or equal to To prevent the pheromone evaporation coefficient from being adjusted too low, the first method is adopted. The initial pheromone evaporation coefficient of the node where the ant is located during the initialization of the ant colony algorithm parameters.

[0035] In the formula middle, For the first The convergence distance of the next iteration For the first The local optimum group in the next iteration. For the first The local suboptimal group in the next iteration The first distance, No. The local optimum group in the next iteration. The second distance, For the first The node where the ant is located is at the th The parameter set for the next iteration. This is the third distance.

[0036] In the formula In the middle, it should be noted that the first distance The larger the value, the more likely it is to be the first. The greater the distance between the local optimal group and the local suboptimal group in the next iteration, the less likely it is to converge. Therefore, the pheromone evaporation coefficient is increased to promote exploration. Conversely, the smaller the first distance, the closer the local optimal group is to the global optimal parameter group in the next iteration. In this case, the pheromone evaporation coefficient is decreased to retain high-quality node information and speed up the algorithm convergence.

[0037] In the formula In addition, it should be noted that the second distance... The larger the value, the higher the value. The second iteration and the first The larger the distance between the local optima in the next iteration, the less likely it is to converge, and the lower the probability of finding the globally optimal parameter set in these two iterations. Therefore, the pheromone evaporation coefficient is increased at this point to promote exploration; the first distance The smaller the value, the higher the value. The second iteration and the first The smaller the distance between the local optima in each iteration, the closer to convergence, and the greater the probability of finding the global optimal parameter set in these two iterations. Therefore, reducing the pheromone evaporation coefficient at this point can accelerate convergence.

[0038] In the formula In addition, it should be noted that the third distance... The larger the value, the more similar the parameter set of that node is to the first node. In the next iteration, the greater the distance to the local optimum, the greater the pheromone evaporation coefficient of that node, thus promoting exploration; conversely, the smaller the third distance, the closer the node is to the optimum, and the less the pheromone evaporation coefficient is, thus preserving the information of high-quality nodes.

[0039] It should be further explained that by using the distance between the optimal solutions in two adjacent iterations, and the distance between the optimal and suboptimal solutions in the current iteration, it is determined whether the current iteration is close to convergence. Then, the Euclidean distance between the current node and the optimal solution is used to adjust the pheromone evaporation coefficient. The closer the node is to the optimal solution, the more pheromone is retained, which accelerates the convergence of the algorithm to the global optimal solution.

[0040] In some embodiments, each ant has the same initial pheromone evaporation coefficient.

[0041] It should be added that, in the first During the nth iteration, the 1st The parameter set represented by a single ant is: The corresponding objective function value is The initial pheromone evaporation coefficient of the node is improved by using the number of iterations at this time and the objective function value of the node.

[0042] For example, after the ant colony algorithm is initialized, the initial pheromone evaporation coefficient of each ant is 0.8.

[0043] In some embodiments, the iteration convergence condition is that the difference between the minimum values of the objective function values of two adjacent iterations is less than a preset threshold.

[0044] It should be noted that the iteration convergence condition can also be that the preset number of iterations is reached, for example, the preset number of iterations is 10 and the preset threshold is 0.1.

[0045] Based on the improved pheromone evaporation coefficient of the ant colony algorithm, the PID parameter selection is optimized, and then the temperature in the photonic unbonding simulation process is controlled in real time. First, the pheromone evaporation coefficient is adjusted in combination with the number of iterations and the advantages and disadvantages of the node, the exploration and development are balanced, the premature convergence is avoided, the global search ability is improved, and the convergence speed is accelerated; in combination with the distance of the optimal solution of adjacent iterations and the distance of the optimal and suboptimal solutions of the current iteration, the evaporation coefficient is further adjusted, the convergence is dynamically optimized according to the convergence condition, and the convergence speed and accuracy are improved.

[0046] It should be pointed out that the above description with reference to the drawings illustrates the preferred embodiments of the application, and does not limit the scope of the application. For those skilled in the art, without departing from the concept of the application, a number of modifications and improvements can be made, which are all within the scope of the application. Therefore, the protection scope of the patent of the application should be subject to the appended claims.

Claims

1. A method for simulation parameter optimization of photonic debonding for wafer level packaging, characterized in that, The method comprises the following steps: obtaining a value range of simulation parameters of photonic debonding, performing grid division on the value range to obtain a grid disk, each node of the grid disk corresponding to a parameter group; constructing an objective function of an ant colony algorithm, for each iteration of the ant colony algorithm, calculating a pheromone dynamic evaporation coefficient of a node where each ant is located to control a pheromone decay rate of the next iteration, screening all the parameter groups during the iteration process to obtain a globally optimal parameter group, completing parameter optimization, and the globally optimal parameter group is used for temperature regulation of photonic debonding by a PID control algorithm; wherein, the pheromone dynamic evaporation coefficient of the node where each ant is located comprises: calculating the objective function value of the parameter group corresponding to the node where each ant is located, taking any ant as a target ant, calculating the degree of superiority and inferiority of the node where the target ant is located according to the ratio of the minimum value of the objective function value of the current iteration to the objective function value of the target ant, calculating the convergence degree of the current iteration, taking the product of the degree of superiority and inferiority, the convergence degree and the initial pheromone evaporation coefficient of the target ant as the pheromone dynamic evaporation coefficient of the target ant, and traversing to obtain the pheromone dynamic evaporation coefficient of the node where each ant is located.

2. The simulation parameter optimization method for photonic de-bonding of wafer level packaging according to claim 1, wherein, The simulation parameters include proportional gain parameters, integral gain parameters and differential gain parameters of the PID control algorithm.

3. The simulation parameter optimization method for photonic de-bonding of wafer level packaging according to claim 1, wherein, The construction of the objective function of the ant colony algorithm comprises: obtaining a real-time temperature of photonic debonding, taking the difference between the real-time temperature and a preset temperature as a temperature deviation, and calculating a temperature adjustment amount by using a PID control algorithm; taking the sum of the square integral of the temperature deviation and the square integral of the temperature adjustment amount as an objective function.

4. The simulation parameter optimization method for photonic de-bonding of wafer level packaging according to claim 1, wherein, The calculation of the degree of superiority and inferiority of the node where the target ant is located comprises: performing a negative correlation mapping on the minimum value of the objective function value and the objective function value of the target ant, and taking the normalized value of the mapping result as the degree of superiority and inferiority.

5. The simulation parameter optimization method for photonic de-bonding of wafer level packaging according to claim 1, wherein, The convergence degree comprises: calculating a convergence distance of the current iteration; obtaining the ratio of a preset maximum iteration number to a current iteration number; taking the normalized result of the product of the ratio and the convergence distance as the convergence degree.

6. The simulation parameter optimization method for photonic de-bonding of wafer level packaging according to claim 5, wherein, The calculation of the convergence distance of the current iteration comprises: taking a parameter group corresponding to the minimum value of the objective function value of the current iteration as a local optimal group of the current iteration, taking a parameter group corresponding to the second minimum value of the objective function value of the current iteration as a local suboptimal group of the current iteration, and taking the Euclidean distance between the local optimal group of the current iteration and the local suboptimal group of the current iteration as a first distance; obtaining a local optimal group of a last iteration adjacent to the current iteration, wherein the local optimal groups of the adjacent two iterations are obtained in the same way, taking the Euclidean distance between the local optimal group of the current iteration and the local optimal group of the last iteration as a second distance; taking a parameter group corresponding to the node where the target ant is located as a target group, and taking the distance between the target group and the local optimal group of the current iteration as a third distance; taking the product of the first distance, the second distance and the third distance as the convergence distance of the current iteration.

7. The simulation parameter optimization method for photonic de-bonding of wafer level packaging according to claim 1, wherein, The initial pheromone evaporation coefficients of each ant are the same.

8. The simulation parameter optimization method for photonic de-bonding of wafer level packaging according to claim 1, wherein, The iteration convergence condition is: the difference between the minimum values of the objective function values of the adjacent two iterations is less than a preset threshold value; or a preset iteration number is reached.

Images