Preparation process optimization method and system of gallium nitride avalanche detector

Abstract

The invention provides a preparation process optimization method and system for a gallium nitride avalanche detector, and relates to the technical field of semiconductors, and the method comprises the steps: obtaining technological parameter information of gallium nitride epitaxial growth, carrying out the growth process prediction according to the technological parameter information, obtaining a thickness prediction result of a first epitaxial layer, and obtaining a growth control parameter set of epitaxial equipment; and according to the growth control parameter set and the first epitaxial layer thickness prediction result, a process control scheme is obtained, and epitaxial equipment is controlled. The technical problems that in the prior art, a gallium nitride avalanche detector manufacturing process is insufficient in epitaxial layer thickness control precision and poor in uniformity are solved. The technical effects of improving the control precision and uniformity of the thickness of the epitaxial layer and further improving the overall performance and reliability of the gallium nitride avalanche detector are achieved.

Description

Technical Field

[0001] The present invention relates to the field of semiconductor technology, and in particular to a method and system for optimizing the preparation process of a gallium nitride avalanche detector. Background Art

[0002] An avalanche detector is a photoelectric device that uses the avalanche multiplication effect produced by semiconductor materials under strong electric fields to amplify weak light signals, thereby achieving high-sensitivity detection. Its performance mainly depends on the quality of the material, the design of the device structure, and the precise control of various key parameters during the preparation process. In the preparation process of gallium nitride avalanche detectors, the thickness control of the epitaxial layer is one of the key factors that determine the performance of the device. The thickness of the epitaxial layer directly affects the location where the avalanche multiplication effect occurs and the breakdown voltage of the device, which has an important impact on the gain, response speed and signal-to-noise ratio of the detector. However, due to the complex growth conditions of gallium nitride materials and the difficulty in controlling the thickness, this poses a severe challenge to the preparation of avalanche detectors.

[0003] In the prior art, there are technical problems in the preparation process of gallium nitride avalanche detectors, such as insufficient control accuracy and poor uniformity of the epitaxial layer thickness, which leads to a decrease in the performance of the gallium nitride avalanche detectors. Summary of the invention

[0004] The present application provides a method and system for optimizing the preparation process of a gallium nitride avalanche detector, which is used to solve the technical problem in the prior art that the gallium nitride avalanche detector preparation process has insufficient control accuracy and poor uniformity in the thickness of the epitaxial layer, resulting in a decrease in the performance of the gallium nitride avalanche detector.

[0005] In view of the above problems, the present application provides a method and system for optimizing the preparation process of a gallium nitride avalanche detector.

[0006] In a first aspect of the present application, a method for optimizing the preparation process of a gallium nitride avalanche detector is provided, the method comprising: obtaining process parameter information of gallium nitride epitaxial growth, the process parameter information comprising gas flow, temperature and pressure; predicting the growth process according to the process parameter information, and obtaining a prediction result of the thickness of a first epitaxial layer; obtaining a growth control parameter set of an epitaxial device, the growth control parameter set comprising a control strategy, response speed and adjustment accuracy of the epitaxial device; obtaining a process control scheme according to the growth control parameter set and the prediction result of the thickness of the first epitaxial layer; and inputting the process control scheme into a control panel of the epitaxial device to control the epitaxial device.

[0007] According to a second aspect of the present application, a system for optimizing the preparation process of a gallium nitride avalanche detector is provided, the system comprising: a process parameter information acquisition module, the process parameter information acquisition module is used to acquire process parameter information of gallium nitride epitaxial growth, the process parameter information including gas flow, temperature and pressure; a growth process prediction module, the growth process prediction module is used to predict the growth process according to the process parameter information, and acquire a prediction result of the thickness of a first epitaxial layer; a growth control parameter acquisition module, the growth control parameter acquisition module is used to acquire a growth control parameter set of an epitaxial device, the growth control parameter set including a control strategy, response speed and adjustment accuracy of the epitaxial device; a process control scheme acquisition module, the process control scheme acquisition module is used to acquire a process control scheme according to the growth control parameter set and the prediction result of the thickness of the first epitaxial layer; and an equipment control module, the equipment control module is used to input the process control scheme into a control panel of the epitaxial device to control the epitaxial device.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0009] By obtaining the process parameter information of GaN epitaxial growth, the process parameter information includes gas flow, temperature and pressure; predicting the growth process according to the process parameter information, obtaining the prediction result of the thickness of the first epitaxial layer; obtaining the growth control parameter set of the epitaxial device, the growth control parameter set includes the control strategy, response speed and adjustment accuracy of the epitaxial device; obtaining the process control scheme according to the growth control parameter set and the prediction result of the thickness of the first epitaxial layer; inputting the process control scheme into the control panel of the epitaxial device to control the epitaxial device. The technical effect of improving the control accuracy and uniformity of the epitaxial layer thickness, thereby improving the overall performance and reliability of the GaN avalanche detector is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. 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 creative work.

[0011] Figure 1 A schematic diagram of a process flow of a gallium nitride avalanche detector preparation process optimization method provided in this application;

[0012] Figure 2 A schematic diagram of the structure of a gallium nitride avalanche detector manufacturing process optimization system provided in this application.

[0013] Explanation of the reference numerals: process parameter information acquisition module 11 , growth process prediction module 12 , growth control parameter acquisition module 13 , process control scheme acquisition module 14 , equipment control module 15 . DETAILED DESCRIPTION

[0014] The present application provides a method and system for optimizing the manufacturing process of a gallium nitride avalanche detector, which is used to solve the technical problem in the prior art that the manufacturing process of a gallium nitride avalanche detector has insufficient control precision and poor uniformity in the thickness of the epitaxial layer, resulting in reduced performance of the gallium nitride avalanche detector. The technical effect of improving the control precision and uniformity of the epitaxial layer thickness, thereby improving the overall performance and reliability of the gallium nitride avalanche detector, is achieved.

[0015] Below, the technical solutions in the present invention will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments of the present invention. It should be understood that the present invention is not limited to the example embodiments described herein. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. It should also be noted that, for the convenience of description, only the parts related to the present invention are shown in the accompanying drawings, rather than all of them.

[0016] Embodiment 1, as Figure 1 As shown, the present application provides a method for optimizing the preparation process of a gallium nitride avalanche detector, the method comprising:

[0017] The process parameter information of GaN epitaxial growth is obtained, wherein the process parameter information includes gas flow, temperature and pressure.

[0018] Specifically, GaN epitaxial growth refers to the process of growing a layer of GaN thin film with a high-quality crystal structure on the surface of a substrate (usually silicon, sapphire or silicon carbide) by physical or chemical deposition. For GaN avalanche detectors, the quality of epitaxial growth directly affects the performance and stability of the device. In the epitaxial growth process, the precise control of process parameters is the key factor in determining the GaN layer. Therefore, obtaining GaN process parameter information is the first step in optimizing the preparation process of GaN avalanche detectors. The process parameter information of GaN epitaxial growth is collected in real time through a sensor network integrated in the epitaxial equipment. The process parameter information includes gas flow, temperature and pressure, where gas flow refers to the supply rate of the reaction gas during the epitaxial growth process. The control of gas flow directly affects the growth rate and surface morphology of GaN materials. For example, too high a gas flow may cause the epitaxial layer to grow too fast, resulting in a decrease in crystal quality or uneven layer thickness, while too low a flow may result in insufficient growth rate and failure to achieve the expected thickness. Temperature is used to control the chemical reaction rate and determines the quality and defect density of the crystal. Pressure is used to adjust the reaction atmosphere and control the reaction rate, which determines the thickness and uniformity of the GaN epitaxial layer.

[0019] The growth process is predicted according to the process parameter information to obtain a prediction result of the thickness of the first epitaxial layer.

[0020] Furthermore, a growth process prediction is performed based on the process parameter information to obtain a first epitaxial layer thickness prediction result, including: collecting historical process parameter information, analyzing the historical process parameter information to form a historical process parameter sequence data set, each data in the historical process parameter sequence data set contains the values ​​of each process parameter within a corresponding time window; extracting the epitaxial layer thickness record of the corresponding time period from the historical growth data log of the epitaxial equipment; constructing a growth prediction model through the historical process parameter sequence data set and the epitaxial layer thickness record; inputting the process parameter information into the growth prediction model, and outputting the first epitaxial layer thickness prediction result.

[0021] Specifically, first, historical process parameter information of GaN epitaxial growth process is collected from multiple data sources such as GaN avalanche detector production records and equipment logs, industry reports, experimental data, etc., including gas flow, growth temperature, pressure and time, etc. The historical process parameter information covers records under various process conditions. Then, the collected historical process parameter information is analyzed to form a historical process parameter sequence data set, which means that the historical process parameter information data is sorted and the process parameter values ​​in each time window are arranged in chronological order, so that each data point in the process parameter sequence data set not only contains the specific process parameter value at a certain moment, but also reflects the sequence of these parameters changing over time.

[0022] Next, the epitaxial layer thickness record of the time period corresponding to the process parameter value is extracted from the historical growth data log of the epitaxial device. The epitaxial device refers to a special device for the preparation of gallium nitride avalanche detectors, which is equipped with a growth control parameter set such as control strategy, response speed, and adjustment accuracy, which is used to optimize the thickness and quality of the epitaxial layer. The device also has real-time monitoring and feedback adjustment functions. The epitaxial layer thickness record reflects the actual thickness of the epitaxial layer grown under specific process parameter conditions, providing the necessary labels (i.e., epitaxial layer thickness) for model training. The epitaxial layer thickness record corresponds to the time window of each process parameter sequence, which can ensure the consistency and accuracy of the data. Using the historical process parameter sequence data set and the epitaxial layer thickness record, a growth prediction model is constructed through machine learning to predict the growth results of the epitaxial layer under given process parameter conditions. During the construction process, the collected historical process data and thickness records are divided into a training set and a validation set, and the model is trained using the training set, and the accuracy and stability of the model are tested through the validation set. Through multiple iterations and model optimization, a prediction model that can accurately reflect the growth process is finally formed.

[0023] Finally, the process parameter information is input into the constructed growth prediction model, and the growth prediction model automatically calculates and outputs the first epitaxial layer thickness prediction result according to the input parameter information, and the first epitaxial layer thickness prediction result is the thickness of the first epitaxial film formed during the epitaxial growth process. By constructing a model to accurately and reliably predict the thickness of the first epitaxial layer, it can effectively guide the subsequent epitaxial growth process, solve the problem of insufficient thickness control accuracy during the epitaxial growth process, and ensure that the thickness of the epitaxial layer can be accurately achieved within a controllable range, thereby improving the performance level of the gallium nitride avalanche detector.

[0024] A growth control parameter set of an epitaxial device is obtained, wherein the growth control parameter set includes a control strategy, a response speed, and an adjustment accuracy of the epitaxial device.

[0025] Specifically, by communicating with the control interface of the epitaxial device, the control parameters of the epitaxial device are obtained in real time to form a growth control parameter set, which includes parameter data such as the control strategy, response speed and adjustment accuracy of the epitaxial device, wherein the control strategy refers to the operating procedures and methods adopted by the epitaxial device when performing the growth task, and the effectiveness of the control strategy directly affects the stability of the epitaxial process and the precise execution capability of the device for the process parameters. The response speed refers to the time required for the epitaxial device to perform the corresponding operation after receiving the control instruction. By obtaining the response speed parameter of the device, it can help determine the reaction sensitivity of the device during the growth process. The adjustment accuracy refers to the minimum adjustment unit or error range that the epitaxial device can achieve when adjusting parameters such as gas flow, temperature and pressure. The acquisition and analysis of the adjustment accuracy can help optimize the control parameters, making the epitaxial growth process more refined and avoiding thickness deviation or crystal quality degradation caused by inaccurate adjustment. Through the comprehensive collection and integration of the control strategy, response speed and adjustment accuracy, a reliable control basis can be provided for the subsequent epitaxial growth process, ensuring that the growth process of the gallium nitride epitaxial layer can be carried out under optimal conditions, thereby improving the performance and consistency of the final gallium nitride avalanche detector.

[0026] A process control scheme is obtained according to the growth control parameter set and the first epitaxial layer thickness prediction result.

[0027] Furthermore, based on the growth control parameter set and the first epitaxial layer thickness prediction result, a process control scheme is obtained, including: traversing the growth control parameter set to construct a multidimensional control parameter coordinate system; based on the multidimensional control parameter coordinate system, a target optimization space is constructed in combination with the constraints of the production process, wherein the constraints include the process parameter information and the value range of the epitaxial equipment control parameters; according to the first epitaxial layer thickness prediction result as the optimization requirement, iterative optimization is performed in the target optimization space to obtain the process control scheme.

[0028] Specifically, first, the growth control parameter set is traversed to comprehensively analyze the control strategy, response speed, and adjustment accuracy of the epitaxial equipment to understand the value range of each parameter and its impact on the epitaxial growth process. Based on the parameter information obtained through the traversal, a multidimensional control parameter coordinate system is constructed. The multidimensional control parameter coordinate system is a multidimensional space containing all growth control parameters. In this space, each dimension corresponds to a control parameter, and the value range of the parameter is the boundary of the dimension. In the multidimensional control parameter coordinate system, different combinations of parameters represent different control strategies and operation schemes.

[0029] Then, in combination with the actual production process constraints, a target optimization space is constructed in a multidimensional control parameter coordinate system. The constraints refer to the technical and operational restrictions that need to be met in the actual growth process, including the process parameter information and the value range of the epitaxial equipment control parameters, such as the actual operable range of temperature and gas flow, the adjustment capacity of the epitaxial equipment, etc. The constraints limit the size of the target optimization space and ensure that the optimization process is carried out within a feasible parameter range.

[0030] Finally, in the target optimization space, the first epitaxial layer thickness prediction result is used as the optimization target, and iterative optimization is performed, and the control parameter combination is adjusted by continuous iteration to approach or reach the predicted epitaxial layer thickness target. Iterative optimization is a process of gradually approaching the optimal solution by continuously trying different parameter combinations. In each iteration, the control parameters are adjusted and optimized by calculating the difference between the predicted result and the target thickness. Each adjustment requires finding a new parameter combination in the target optimization space, and evaluating its effect based on the first epitaxial layer thickness prediction result, until the optimal parameter combination that meets or approaches the first epitaxial layer thickness prediction result is found or the maximum number of iterations is reached, and the optimal parameter combination, i.e., the process control scheme, is obtained. The process control scheme can maximize the satisfaction of the parameter conditions required by the first epitaxial layer thickness prediction result, ensure that the thickness control during the epitaxial growth process reaches an ideal state, and improve the performance of the gallium nitride avalanche detector.

[0031] The process control scheme is input into a control panel of the epitaxial device to control the epitaxial device.

[0032] Furthermore, the process control scheme is input into the control panel of the epitaxial device to control the epitaxial device, including: converting the control parameters in the process control scheme to obtain process control parameters; performing parameter verification on the process control parameters; configuring the operating state of the epitaxial device according to the verified process control parameters, and performing preparation process optimization of the gallium nitride avalanche detector.

[0033] Specifically, first, the control parameters in the process control scheme are converted through standardized mapping to obtain process control parameters. The parameter conversion refers to converting the parameter values ​​in the process control scheme into a process control parameter format that can be directly recognized and executed by the epitaxial equipment, ensuring that the control scheme can be correctly understood and executed by the epitaxial equipment. Then, the converted process control parameters are checked using an automated verification tool using a predefined set of verification rules. If the verification tool detects that the parameters are out of range or in an incorrect format, a warning or error report will be automatically generated, and the process flow will be prevented from continuing, ensuring that during the conversion process, there are no deviations or errors in the accuracy and range of the control parameters. That is, through verification, any potential inconsistencies or anomalies can be detected to avoid process problems caused by parameter errors during actual operation.

[0034] After the parameter verification is completed, the epitaxial equipment is automatically configured according to the verified process control parameters, and all parameters are set to the operating state of the epitaxial equipment, so that the equipment can perform precise production operations according to the process control plan. For example, using existing driver programs or control interfaces, these interfaces can receive the verified process control parameters and automatically configure the operating state of the equipment, or read the verified process control parameters through automated scripts, programs, etc., and automatically configure the operating state of the epitaxial equipment through the device driver or interface. The configured operating state refers to adjusting the specific operations such as the operating mode, temperature setting, gas flow control and pressure regulation of the equipment according to the verified control parameters. Through this configuration, the preparation process optimization of the gallium nitride avalanche detector is performed, so that the epitaxial equipment can prepare the gallium nitride avalanche detector in the optimal state, minimize the errors and uncertainties in the growth process, improve the control accuracy and uniformity of the epitaxial layer thickness, and thus improve the overall performance and reliability of the gallium nitride avalanche detector.

[0035] Furthermore, configuring the operating state of the epitaxial equipment according to the verified process control parameters includes: monitoring the operating state of the equipment in real time; comparing the monitored operating state with the process control parameters, and generating a feedback signal if a deviation is detected; and dynamically adjusting the operating parameters of the epitaxial equipment according to the feedback signal.

[0036] Specifically, various state data of the epitaxial equipment during operation are collected in real time and continuously through sensors and data acquisition devices, including key process parameters such as temperature, gas flow, pressure, and electrical parameters inside the equipment. Through real-time monitoring, it can be ensured that the operating state of the equipment is within the control range at any time during the entire growth process. Then, the central control system of the equipment automatically performs a comparison of the operating data with the pre-set process control parameters item by item, and analyzes the difference between the standard value of each process parameter and the current operating data. If a deviation is found in the comparison, a feedback signal is immediately generated, in which any difference exceeding the set tolerance range will be marked as a deviation. According to the generated feedback signal, the operating parameters of the epitaxial equipment are automatically and dynamically adjusted. The adjustment process may include operations such as resetting the temperature, adjusting the gas flow, and modifying the pressure setting value to ensure that the operating state of the equipment returns to the control range. The real-time feedback and dynamic adjustment mechanism not only improves the accuracy of process control, but also significantly reduces the risk of quality problems in the production process, ensures the thickness uniformity of the epitaxial layer, and improves the performance stability of the gallium nitride avalanche detector.

[0037] Furthermore, a growth prediction model is constructed through the historical process parameter sequence data set and the epitaxial layer thickness records, including: randomly dividing the historical process parameter sequence data set and the epitaxial layer thickness records into a training set and a verification set; selecting multiple different prediction models, training the multiple prediction models through the training set, and verifying the matching relationship of the multiple prediction models through the verification set; based on the matching relationship of the multiple prediction models, extracting the difference features of the multiple prediction models; using the difference features to optimize the multiple prediction models; when the matching relationship includes multiple prediction models, allocating parameter weights of the multiple prediction models according to the matching ratio relationship, fusing the parameters of the multiple prediction models based on the parameter weights, and generating the growth prediction model.

[0038] Specifically, the collected historical process sequence data set and the corresponding epitaxial layer thickness records are divided into a training set and a validation set by random sampling. Optionally, the ratio of training set to validation set is 8:2. Next, multiple different prediction models are selected, such as linear regression model, support vector machine, random forest, neural network, etc., and the number of models is greater than or equal to 2. Each model has its own unique advantages, such as linear regression is simple and intuitive, and neural network is good at processing nonlinear relationships. These models are trained through the training set so that they can learn the mapping relationship from process parameters to epitaxial layer thickness. After the model training is completed, the matching relationship of multiple prediction models is verified through the validation set. The matching relationship refers to the degree of closeness between the prediction results of different models and the true value, as well as the stability of the model under different conditions. Then, according to the matching relationship of multiple prediction models, the difference features of these models are extracted. The difference features refer to the uniqueness of different models when processing data, such as some models perform better when processing linear data, while other models perform stronger when processing nonlinear data. By extracting these difference features, the strengths and weaknesses of the model in different scenarios can be identified. Then, the obtained difference features are used to further optimize these prediction models, including adjusting model parameters, introducing regularization to prevent overfitting, etc., to improve the overall prediction ability and stability of the model. When it is found that multiple models show good prediction performance on the validation set, model fusion technology is used to further improve the prediction effect. According to the matching ratio of each model on the validation set (i.e., the relative proportion of prediction accuracy), different parameter weights are assigned to the models. These weights reflect the contribution of each model in the overall prediction. By assigning higher weights to models with better performance, the final fusion model can perform more stably and accurately in different data scenarios. Based on the assigned parameter weights, multiple prediction models are fused to generate the final growth prediction model. This fusion model combines the advantages of multiple single models, enabling it to more accurately and stably predict the thickness of the gallium nitride epitaxial layer, significantly improving the controllability of the growth process and the consistency of the performance quality of the gallium nitride avalanche detector.

[0039] Furthermore, according to the first epitaxial layer thickness prediction result as the optimization requirement, iterative optimization is performed in the target optimization space to obtain the process control scheme, including: obtaining a first growth control parameter; using the first epitaxial layer thickness prediction result, evaluating the first growth control parameter set to obtain a first optimization score; randomly selecting a growth control parameter set in the target optimization space as the second growth control parameter set; using the first epitaxial layer thickness prediction result, evaluating the second growth control parameter set to obtain a second optimization score; when the second optimization score is greater than the first optimization score, replacing the second growth control parameter set as a temporary process control parameter set; when the second optimization score is less than or equal to the first optimization score, replacing the second growth control parameter set as a temporary process control parameter set according to probability; repeating the above-mentioned random selection, evaluation and replacement process until the preset number of iterations is reached, outputting the final temporary process control parameter set to obtain the process control scheme.

[0040] Furthermore, obtaining the first growth control parameter includes: randomly selecting a growth control parameter set in the target optimization space as the first growth control parameter set; adding the first growth control parameter set to a taboo table, and using the first growth control parameter set as a temporary process control parameter set, wherein the taboo table includes the number of taboo iterations.

[0041] Specifically, an initial first growth control parameter set is randomly selected in the target optimization space. Randomly selecting the initial parameter set can ensure that the starting point of the optimization process will not be affected by any specific assumptions or biases, and increase the possibility of global optimization. Once a growth control parameter set is selected, it is used as the first growth control parameter set, and the parameter set is added to the taboo table to avoid selecting the same parameter set again in subsequent iterations. The taboo table is a structure for recording parameter combinations that have been tried and their results to avoid loops and repetitions and improve optimization efficiency. The initial parameter set, that is, the first growth control parameter set, is set as a temporary process control parameter set as the starting point for iterative optimization. The first epitaxial layer thickness prediction result is used to evaluate the selected first production control parameter, and the first optimization score is calculated and obtained. The optimization score is determined by comparing the deviation between the predicted epitaxial layer thickness and the target thickness. The higher the score, the closer the parameter set is to the ideal state.

[0042] In the target optimization space, a new second growth control parameter set is randomly selected and evaluated using the same method to obtain a second optimization score. The second optimization score is compared with the first optimization score. If the second optimization score is greater than the first optimization score, it means that the second growth control parameter set is closer to the target state. At this time, the second growth control parameter set is replaced with a new temporary process control parameter set. If the second optimization score is less than or equal to the first optimization score, the second growth control parameter set is replaced as a temporary process control parameter set according to a certain probability. This probabilistic replacement mechanism can help avoid falling into a local optimum and increase the chance of finding a global optimal solution. For example, define an acceptance probability function that calculates the probability of accepting a differential solution based on the current optimization progress (such as the number of iterations, the gap between the current optimal solution and the predicted target, etc.) and the difference between the second optimization score and the first optimization score. This function is monotonically decreasing. As the number of iterations increases or the quality of the solution gradually approaches the optimal, the probability of accepting a poor solution should gradually decrease. In each iteration, when the second optimization score is less than or equal to the first optimization score, the defined acceptance probability function is used to calculate the probability of accepting the second growth control parameter set as the temporary process control parameter set in the current situation. Generate a random number between 0 and 1 and compare it with the calculated acceptance probability. If the random number is less than or equal to the acceptance probability, accept the second growth control parameter set as the new temporary process control parameter set; otherwise, keep the original temporary process control parameter set unchanged. In addition, in the early iterations, a higher acceptance probability can be set to encourage a wider search, and in the later iterations, the acceptance probability is gradually reduced to ensure convergence to a high-quality solution.

[0043] Repeat the above process, and in each round of iteration, randomly select a new parameter set for scoring comparison and replacement until the preset number of iterations or optimization conditions are reached. After the preset conditions are met, the current temporary process control parameter set is output as the final process control solution. Through the above iterative optimization steps, the process control parameters can be gradually optimized, and the optimal process control solution can be finally found, which not only improves the efficiency of optimization, but also effectively avoids the problem of local optimality. By making full use of randomness and taboo search, the optimization process has better global exploration capabilities. Ultimately, it ensures that the process parameters in the gallium nitride epitaxial growth process can be precisely controlled, thereby greatly improving the thickness uniformity and quality stability of the epitaxial layer, and improving the performance and production consistency of the gallium nitride avalanche detector.

[0044] Embodiment 2, based on the same inventive concept as the method for optimizing the preparation process of a gallium nitride avalanche detector in the above embodiment, Figure 2 As shown, the present application provides a system for optimizing the preparation process of a gallium nitride avalanche detector, wherein the system comprises:

[0045] A process parameter information acquisition module 11, the process parameter information acquisition module 11 is used to obtain process parameter information of gallium nitride epitaxial growth, and the process parameter information includes gas flow, temperature and pressure; a growth process prediction module 12, the growth process prediction module 12 is used to predict the growth process according to the process parameter information and obtain a prediction result of the thickness of the first epitaxial layer; a growth control parameter acquisition module 13, the growth control parameter acquisition module 13 is used to obtain a growth control parameter set of the epitaxial equipment, and the growth control parameter set includes a control strategy, response speed and adjustment accuracy of the epitaxial equipment; a process control scheme acquisition module 14, the process control scheme acquisition module 14 is used to obtain a process control scheme according to the growth control parameter set and the prediction result of the thickness of the first epitaxial layer; an equipment control module 15, the equipment control module 15 is used to input the process control scheme into the control panel of the epitaxial equipment to control the epitaxial equipment.

[0046] Furthermore, the growth process prediction module 12 is used to perform the following steps: collect historical process parameter information, analyze the historical process parameter information to form a historical process parameter sequence data set, each data in the historical process parameter sequence data set contains the value of each process parameter within a corresponding time window; extract the epitaxial layer thickness record of the corresponding time period from the historical growth data log of the epitaxial equipment; construct a growth prediction model through the historical process parameter sequence data set and the epitaxial layer thickness record; input the process parameter information into the growth prediction model, and output the first epitaxial layer thickness prediction result.

[0047] Furthermore, the growth process prediction module 12 is also used to perform the following steps: randomly divide the historical process parameter sequence data set and the epitaxial layer thickness records into a training set and a verification set; select multiple different prediction models, train the multiple prediction models through the training set, and verify the matching relationship of the multiple prediction models through the verification set; based on the matching relationship of the multiple prediction models, extract the difference features of the multiple prediction models; use the difference features to optimize the multiple prediction models; when the matching relationship includes multiple prediction models, allocate parameter weights of the multiple prediction models according to the matching ratio relationship, fuse the parameters of the multiple prediction models based on the parameter weights, and generate the growth prediction model.

[0048] Furthermore, the process control scheme acquisition module 14 is used to perform the following steps: traverse the growth control parameter set to construct a multidimensional control parameter coordinate system; based on the multidimensional control parameter coordinate system, a target optimization space is constructed in combination with the constraints of the production process, wherein the constraints include the process parameter information and the value range of the epitaxial equipment control parameters; according to the first epitaxial layer thickness prediction result as the optimization requirement, iterative optimization is performed in the target optimization space to obtain the process control scheme.

[0049] Furthermore, the process control scheme acquisition module 14 is also used to perform the following steps: obtain a first growth control parameter set; use the first epitaxial layer thickness prediction result to evaluate the first growth control parameter set to obtain a first optimization score; randomly select a growth control parameter set in the target optimization space as a second growth control parameter set; use the first epitaxial layer thickness prediction result to evaluate the second growth control parameter set to obtain a second optimization score; when the second optimization score is greater than the first optimization score, replace the second growth control parameter set as a temporary process control parameter set; when the second optimization score is less than or equal to the first optimization score, replace the second growth control parameter set as a temporary process control parameter set according to probability; repeat the above-mentioned random selection, evaluation and replacement process until a preset number of iterations is reached, output the final temporary process control parameter set, and obtain the process control scheme.

[0050] Furthermore, the process control scheme acquisition module 14 is also used to perform the following steps: randomly selecting a growth control parameter set in the target optimization space as the first growth control parameter set; adding the first growth control parameter set to a taboo table, and using the first growth control parameter set as a temporary process control parameter set, wherein the taboo table includes the number of taboo iterations.

[0051] Furthermore, the equipment control module 15 is used to perform the following steps: convert the control parameters in the process control scheme to obtain process control parameters; perform parameter verification on the process control parameters; configure the operating state of the epitaxial equipment according to the verified process control parameters, and perform preparation process optimization of the gallium nitride avalanche detector.

[0052] Furthermore, the equipment control module 15 is also used to perform the following steps: monitor the operating status of the equipment in real time; compare the monitored operating status with the process control parameters, and generate a feedback signal if a deviation is detected; and dynamically adjust the operating parameters of the epitaxial equipment according to the feedback signal.

[0053] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

[0054] This specification and the drawings are merely exemplary illustrations of the present application and are deemed to cover any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, a person skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application intends to include these modifications and variations.

Claims

1. A method for optimizing the preparation process of a gallium nitride avalanche detector, characterized in that: The method comprises: Acquiring process parameter information of GaN epitaxial growth, wherein the process parameter information includes gas flow, temperature and pressure; Predicting the growth process according to the process parameter information to obtain a prediction result of the thickness of the first epitaxial layer; Acquiring a growth control parameter set of an epitaxial device, wherein the growth control parameter set includes a control strategy, a response speed, and an adjustment accuracy of the epitaxial device; Obtaining a process control scheme according to the growth control parameter set and the first epitaxial layer thickness prediction result; The process control scheme is input into a control panel of the epitaxial device to control the epitaxial device.

2. The method for optimizing the preparation process of a gallium nitride avalanche detector according to claim 1, characterized in that: The growth process is predicted according to the process parameter information to obtain a prediction result of the thickness of the first epitaxial layer, including: Collecting historical process parameter information, analyzing the historical process parameter information to form a historical process parameter sequence data set, wherein each data in the historical process parameter sequence data set contains the value of each process parameter within a corresponding time window; Extracting the epitaxial layer thickness record of the corresponding time period from the historical growth data log of the epitaxial device; Constructing a growth prediction model through the historical process parameter sequence data set and the epitaxial layer thickness record; The process parameter information is input into the growth prediction model, and the first epitaxial layer thickness prediction result is output.

3. The method for optimizing the preparation process of a gallium nitride avalanche detector according to claim 2, characterized in that: A growth prediction model is constructed by using the historical process parameter sequence data set and the epitaxial layer thickness record, including: Randomly dividing the historical process parameter sequence data set and the epitaxial layer thickness record into a training set and a validation set; Selecting a plurality of different prediction models, training the plurality of prediction models through the training set, and verifying the matching relationship of the plurality of prediction models through the verification set; Extracting difference features of the multiple prediction models based on the matching relationship between the multiple prediction models; Optimizing the multiple prediction models using the difference features; When the matching relationship includes multiple prediction models, parameter weights of the multiple prediction models are allocated according to the matching proportion relationship, and the parameters of the multiple prediction models are fused based on the parameter weights to generate the growth prediction model.

4. The method for optimizing the preparation process of a gallium nitride avalanche detector according to claim 1, characterized in that: According to the growth control parameter set and the first epitaxial layer thickness prediction result, a process control scheme is obtained, including: Traversing the growth control parameter set to construct a multi-dimensional control parameter coordinate system; Based on the multi-dimensional control parameter coordinate system, a target optimization space is constructed in combination with the constraints of the production process, wherein the constraints include the process parameter information and the value range of the epitaxial device control parameter; According to the first epitaxial layer thickness prediction result as the optimization requirement, iterative optimization is performed in the target optimization space to obtain the process control solution.

5. The method for optimizing the preparation process of a gallium nitride avalanche detector according to claim 4, characterized in that: According to the first epitaxial layer thickness prediction result as the optimization requirement, iterative optimization is performed in the target optimization space to obtain the process control solution, including: obtaining a first growth control parameter set; Using the first epitaxial layer thickness prediction result, evaluating the first growth control parameter set to obtain a first optimization score; Randomly selecting a growth control parameter set in the target optimization space as a second growth control parameter set; Using the first epitaxial layer thickness prediction result, evaluating the second growth control parameter set to obtain a second optimization score; When the second optimization score is greater than the first optimization score, replacing the second growth control parameter set as a temporary process control parameter set; When the second optimization score is less than or equal to the first optimization score, replacing the second growth control parameter set as a temporary process control parameter set according to probability; The above-mentioned random selection, evaluation and replacement process is repeated until a preset number of iterations is reached, and the final temporary process control parameter set is output to obtain the process control solution.

6. The method for optimizing the preparation process of a gallium nitride avalanche detector according to claim 5, characterized in that: A first growth control parameter set is obtained, including: Randomly selecting a growth control parameter set in the target optimization space as the first growth control parameter set; The first growth control parameter set is added to a taboo table, and the first growth control parameter set is used as a temporary process control parameter set, wherein the taboo table includes taboo iteration numbers.

7. The method for optimizing the preparation process of a gallium nitride avalanche detector according to claim 1, characterized in that: Inputting the process control scheme into a control panel of the epitaxial device to control the epitaxial device includes: Converting the control parameters in the process control scheme to obtain process control parameters; Performing parameter verification on the process control parameters; The operating state of the epitaxial equipment is configured according to the verified process control parameters, and the preparation process optimization of the gallium nitride avalanche detector is performed.

8. The method for optimizing the preparation process of a gallium nitride avalanche detector according to claim 7, characterized in that: Configuring the operating state of the epitaxial device according to the verified process control parameters includes: Real-time monitoring of the equipment's operating status; comparing the monitored operating state with the process control parameters and generating a feedback signal if a deviation is detected; According to the feedback signal, the operating parameters of the epitaxial equipment are dynamically adjusted.

9. A gallium nitride avalanche detector manufacturing process optimization system, characterized in that: The steps for implementing the method according to any one of claims 1 to 7 include: A process parameter information acquisition module, wherein the process parameter information acquisition module is used to acquire process parameter information of GaN epitaxial growth, wherein the process parameter information includes gas flow, temperature and pressure; A growth process prediction module, the growth process prediction module is used to predict the growth process according to the process parameter information and obtain a prediction result of the thickness of the first epitaxial layer; A growth control parameter acquisition module, the growth control parameter acquisition module is used to acquire a growth control parameter set of the epitaxial device, the growth control parameter set includes a control strategy, a response speed, and an adjustment accuracy of the epitaxial device; A process control scheme obtaining module, wherein the process control scheme obtaining module is used to obtain a process control scheme according to the growth control parameter set and the first epitaxial layer thickness prediction result; An equipment control module, wherein the equipment control module is used to input the process control scheme into a control panel of the epitaxial equipment to control the epitaxial equipment.

Images