Wet oxidation aperture optimization method for VCSEL chips with dynamic adjustment

Abstract

The present invention discloses a method for optimizing the wet oxidation aperture of a VCSEL chip by dynamic adjustment, which relates to the technical field of semiconductor manufacturing. The method includes: fitting and observing the oxidation process, dividing key stages, and setting the optimization priorities and characteristics of control parameters. Collecting data on the temperature, humidity, aperture change, and oxidation time of the VCSEL chip during the wet oxidation process to generate a global decision. Transmitting the global decision to the parameter correlation control layer for analysis, establishing a global target, optimizing the global decision, and calling the local control layer to execute the optimized decision. It solves the technical problems in the prior art of inaccurate control of the wet oxidation process of VCSEL chips and insufficient consideration of the linkage relationship between parameters, and achieves the technical effect of precise control of the wet oxidation process of VCSEL chips.

Description

Technical Field

[0001] This application relates to the field of semiconductor manufacturing technology, and particularly to a method for optimizing the wet oxidation aperture of a VCSEL chip with dynamic adjustment. Background Art

[0002] In the scenario of VCSEL chip manufacturing, the wet oxidation process is crucial for chip performance, but there are many problems in this process. The control of the traditional wet oxidation process of VCSEL chips is not precise enough. There is a lack of detailed division and targeted optimization for different stages of the oxidation process. Fixed parameters are often used for control, making it difficult to adapt to complex oxidation environments and the requirements of different stages. This may lead to uneven oxidation apertures, affecting chip performance and consistency. At the same time, there is a lack of effective real-time monitoring means and dynamic adjustment mechanisms, making it difficult to obtain key parameters in real time and unable to adjust control parameters in a timely manner, resulting in insufficient stability and reliability in the production process. In addition, the interaction relationship between various control parameters during the oxidation process is not adequately considered. Traditional methods adjust parameters in isolation and do not fully consider the synergistic effect between parameters, limiting the optimization space of the wet oxidation process of VCSEL chips.

[0003] In the current related technologies, there are technical problems such as inaccurate control of the wet oxidation process of VCSEL chips, insufficient consideration of the interaction relationship between parameters, and lack of dynamic adjustment. Summary of the Invention

[0004] This application provides a method for optimizing the wet oxidation aperture of a VCSEL chip with dynamic adjustment. It adopts fitting observation of the oxidation process, divides key stages, and sets the optimization priority and characteristics of control parameters for each stage. Monitoring sensors are used to collect data on the temperature, humidity, aperture change, and oxidation time of the VCSEL chip during the wet oxidation process in real time. Through the global control layer, based on the stage characteristics and real-time data set, the key stage is matched, the optimization priority is activated, and a global decision is generated. The global decision is passed to the parameter correlation control layer for analysis, a global goal is established, the global decision is optimized through parameter linkage prediction and synchronous adjustment strategies, and the local control layer is called to execute the optimized decision. It realizes the dynamic adjustment of control parameters according to the characteristics of different stages, ensures that the oxidation process is always in the optimal state, and achieves the technical effect of precise control of the wet oxidation process of VCSEL chips.

[0005] This application provides a method for optimizing the wet oxidation aperture of a VCSEL chip with dynamic adjustment, including:

[0006] Perform a fitting observation on the oxidation process, divide the oxidation process into stages according to the fitting observation results, and establish a key stage division result. Among them, each key stage division result is set with an optimization priority of control parameters, and the key stage division result is configured with stage division characteristics; deploy monitoring sensors to monitor the wet oxidation process of the VCSEL chip with the monitoring sensors, and establish a real-time data set. The real-time data set includes oxidation environment temperature data, humidity data, aperture change data, and oxidation time data; configure a global control layer, perform key stage matching through the stage division characteristics and the real-time data set, and activate the optimization priority of the corresponding key stage division result with the matching result. Initialize the global control layer with the activated optimization priority, and process the real-time data set through the initialized global control layer to generate a global decision; after sending the global decision to the parameter association control layer, parse the global decision with the parameter association control layer to establish a global goal; predict the linkage relationship between parameters with the global goal, configure a synchronous adjustment strategy based on the linkage relationship prediction result, optimize the global decision based on the synchronous adjustment strategy, and then call the local control layer to execute the optimized global decision.

[0007] It is intended to adopt the method for optimizing the wet oxidation aperture of the VCSEL chip with dynamic adjustment proposed in this application. First, perform a fitting observation on the oxidation process, divide key stages, and set the optimization priority and characteristics of control parameters for each stage. Use monitoring sensors to collect in real time the temperature, humidity, aperture change, and oxidation time data of the VCSEL chip during the wet oxidation process. Through the global control layer, match the key stages based on the stage characteristics and the real-time data set, activate the optimization priority, and generate a global decision. Transmit the global decision to the parameter association control layer for parsing, establish a global goal, optimize the global decision through parameter linkage prediction and synchronous adjustment strategy, call the local control layer to execute the optimized decision, and achieve the technical effect of dynamically adjusting control parameters according to the characteristics of different stages to ensure that the oxidation process is always carried out in the optimal state and achieve precise control of the wet oxidation process of the VCSEL chip. Brief Description of the Drawings

[0008] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings of the embodiments of the present invention will be briefly introduced below. Flowcharts are used in this application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the operations in the front or below do not necessarily need to be executed precisely in sequence. On the contrary, according to the need, they can be executed in reverse order or simultaneously. At the same time, other operations can also be added to these processes, or one or several steps of operations can be removed from these processes.

[0009] Figure 1Schematic flowchart of the wet oxidation aperture optimization method for VCSEL chips with dynamic adjustment provided by the embodiments of the present application;

[0010] Figure 2 Schematic flowchart of the establishment of the key stage division result of the wet oxidation aperture optimization method for VCSEL chips with dynamic adjustment provided by the embodiments of the present application. Detailed implementation manners

[0011] The above description is only an overview of the technical solution of the present application. In order to understand the technical means of the present application more clearly, it can be implemented according to the content of the specification. And in order to make the above and other objects, features and advantages of the present application more obvious and understandable, the following specifically gives the detailed implementation manners of the present application.

[0012] In order to make the purpose, technical solution and advantages of the present application clearer, the present application will be further described in detail below with reference to the accompanying drawings. The described embodiments should not be regarded as limitations of the present application. All other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.

[0013] In the following description, "some embodiments" are involved, which describe subsets of all possible embodiments. However, it can be understood that "some embodiments" can be the same subset or different subsets of all possible embodiments, and can be combined with each other without conflict. The terms "first\second" involved are only used to distinguish similar objects and do not represent a specific order for the objects. The terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or server that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or are inherent to these processes, methods, products or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which the present application belongs. The terms used herein are only for the purpose of describing the embodiments of the present application.

[0014] The embodiments of the present application provide a wet oxidation aperture optimization method for VCSEL chips with dynamic adjustment, as Figure 1 shown, the method includes:

[0015] Step S100: Conduct a fitting observation on the oxidation process, divide the oxidation process into stages based on the fitting observation results, and establish the key stage division results. Each key stage division result is set with an optimization priority for control parameters, and the key stage division results are configured with stage division characteristics. Specifically, first, conduct a fitting observation on the wet oxidation process of the VCSEL chip, collect data such as oxidation time, temperature change, humidity change, and aperture size change, and use mathematical models and data analysis techniques to simulate the changing trends of various parameters over time. By collecting a large amount of relevant data during the wet oxidation process of the VCSEL chip, including oxidation time, temperature change, humidity change, and aperture size change, use the curve fitting method to simulate the changing trends of various parameters over time during the oxidation process. By continuously adjusting the parameters of the fitting model, make the fitting results as close as possible to the actual observation data. Then, divide the oxidation process into stages based on the fitting observation results, and comprehensively consider the changes of multiple parameters to determine the key stage division results. Then, set the optimization priority of the control parameters for each key stage because the influence degrees of different control parameters on the oxidation process are different in different stages. Finally, configure stage division characteristics for each key stage division result. These characteristics can be specific parameter value ranges, parameter change trends, time intervals, etc., so as to quickly and accurately identify the current stage of the oxidation process and provide a clear basis for the subsequent control of the oxidation process.

[0016] In a possible implementation, as Figure 2 shown, conduct a fitting observation on the oxidation process, divide the oxidation process into stages based on the fitting observation results, and establish the key stage division results. Each key stage division result is set with an optimization priority for control parameters, and the key stage division results are configured with stage division characteristics. Step S100 further includes step S110: Conduct a timing fitting of the control parameters according to the fitting observation results and establish a timing fitting curve. Specifically, after conducting a fitting observation on the wet oxidation process of the VCSEL chip, further conduct a timing fitting for the control parameters. First, determine the control parameters that need to be fitted. The parameters may include temperature, humidity, aperture change, and oxidation time, etc. Then, use polynomial fitting to fit according to the observed data of the changes of different parameters over time. By continuously adjusting the parameters of the fitting model, make the fitting curve as close as possible to the actual observation data and establish a timing fitting curve of the control parameters. This curve can intuitively display the changing trends of each control parameter in the time dimension.

[0017] Step S120: Extract feature points from the timing fitting curve and establish a clustering result in the time dimension based on the feature point extraction results. Specifically, after obtaining the timing fitting curve, perform feature point extraction on it. Feature points can be turning points, extreme points, or points with obvious change trends in the curve. For example, rapid rising points or falling points in the temperature fitting curve, mutation points in the pore size change curve, etc. can all be used as feature points. By extracting feature points, the oxidation process can be divided in the time dimension. According to the distribution and time sequence of feature points, time periods with similar features are classified into one category, thereby establishing a clustering result in the time dimension. The clustering result can initially reflect the characteristics and change rules of the oxidation process in different time periods.

[0018] Step S130: Use the fitting observation results to perform multi-dimensional clustering of control parameters and establish a multi-dimensional clustering result. The features extracted by multi-dimensional clustering include the temperature change rate, humidity volatility, pore size change rate, pore size stability, and cumulative effect of oxidation time. Specifically, in addition to analyzing in the time dimension, the relationships between multiple control parameters need to be considered. Using the previous fitting observation results, perform multi-dimensional clustering on the control parameters. During the multi-dimensional clustering process, the extracted features include the temperature change rate, humidity volatility, pore size change rate, pore size stability, and cumulative effect of oxidation time, etc. For the temperature change rate, it can be determined by calculating the ratio of the temperature difference between adjacent time points to the time interval; humidity volatility can be measured by the standard deviation of humidity; the pore size change rate and stability can be calculated similarly through the change amount and change trend of the pore size; the cumulative effect of oxidation time can consider the influence of oxidation time on other parameters. According to the features, use clustering algorithms such as K-Means clustering and hierarchical clustering to divide different oxidation states into different categories, thereby establishing a multi-dimensional clustering result. The multi-dimensional clustering result can more comprehensively reflect the mutual relationships and change patterns among various parameters during the oxidation process.

[0019] Step S140: Perform clustering fusion on the clustering result in the time dimension and the multi-dimensional clustering result to establish a key stage division result. Specifically, perform clustering fusion on the clustering result in the time dimension and the multi-dimensional clustering result. First, analyze the characteristics and similarities of each category in the clustering result in the time dimension and the multi-dimensional clustering result. If certain categories in the time dimension are similar to the categories in the multi-dimensional clustering result in terms of parameter characteristics and change trends, they can be merged. For example, a certain rapid oxidation time period in the time dimension matches the category with a high temperature change rate and a fast pore size change rate in the multi-dimensional clustering result, and they can be fused into a key stage. By continuously comparing and merging, establish a key stage division result. The result combines the information in the time dimension and multiple control parameters and can more accurately divide the key stages of the oxidation process, providing a more targeted basis for subsequent optimization control.

[0020] In a possible implementation, clustering fusion is performed on the clustering results in the time dimension and the multi-dimensional clustering results to establish the key stage division results. Step S140 further includes step S141 of establishing a time series window according to the clustering results in the time dimension. Specifically, when analyzing the wet oxidation process of the VCSEL chip, first, a time series window is established according to the clustering results in the time dimension. Since the clustering results in the time dimension have initially divided the oxidation process in terms of time, each cluster represents a time period with specific characteristics. Through the clustering results, some key time nodes can be determined, and a time series window is established with these time nodes as boundaries. For example, if the clustering results in the time dimension show that the temperature changes significantly during a certain time period, then a time series window can be established with the start and end times of this time period as boundaries, specifically for analyzing the characteristics of the oxidation process during this time period.

[0021] Step S142 is to perform trend consistency matching on the multi-dimensional clustering results within the time series window using the time series window. Specifically, after establishing the time series window, then use this window to perform trend consistency matching on the multi-dimensional clustering results within the window. The multi-dimensional clustering results classify the oxidation process from the perspectives of multiple control parameters, including characteristics such as the temperature change rate, humidity volatility, aperture change rate, aperture stability, and cumulative effect of oxidation time. Within the time series window, analyze whether the characteristic change trend of the multi-dimensional clustering results is consistent with the trend represented by the clustering results in the time dimension. For example, if the clustering results in the time dimension indicate that a certain time period is the rapid oxidation stage, and the multi-dimensional clustering results within the time series window also show characteristics such as a fast temperature change rate and a high aperture change rate, then it can be considered that the two trends are consistent.

[0022] Step S143 is to retain the stage division if the characteristic change trend of the multi-dimensional clustering results within the time series window is consistent with the clustering results in the time dimension. Specifically, if within the time series window, the characteristic change trend of the multi-dimensional clustering results is consistent with the clustering results in the time dimension, it indicates that the division of this time period is reasonable, and this stage division can be retained, meaning that within this time period, the characteristics of the oxidation process can be jointly described by the time dimension and the multi-dimensional clustering results, providing a clear basis for subsequent optimization control. For example, if the multi-dimensional clustering results within a time series window show that the temperature and aperture changes both conform to the characteristics of rapid oxidation, which is consistent with the time period being divided into the rapid oxidation stage in the time dimension, then it can be determined that this time period is indeed the rapid oxidation stage, and this stage division is retained.

[0023] Step S144. If the characteristic change trend of the multi-dimensional clustering result within the time series window is inconsistent with the clustering result in the time dimension, then perform window segmentation on the time series window and continue the matching iteration. Specifically, if the characteristic change trend of the multi-dimensional clustering result within the time series window is inconsistent with the clustering result in the time dimension, it indicates that there may be a problem with the division of this time period, and further adjustment of the time series window is required. The time series window can be segmented into smaller time periods for a more refined analysis of the characteristics of the oxidation process. Then, continue the iterative process of trend consistency matching until a time division consistent with the multi-dimensional clustering result is found. For example, if the multi-dimensional clustering result within a time series window shows that the characteristics of some time points do not match the clustering result in the time dimension, then this window can be segmented into two or more smaller windows, and trend consistency matching is performed separately. Continuously adjust the window size and position until the most suitable division is found.

[0024] Step S145. Complete clustering fusion according to the matching iteration result and the retained stage division result. Specifically, finally, complete clustering fusion according to the result of the matching iteration and the retained stage division result. By continuously adjusting the time series window and performing trend consistency matching, a set of stage divisions that not only conform to the clustering result in the time dimension but also are consistent with the multi-dimensional clustering result can be obtained ultimately. Fusing the stage divisions can obtain a more accurate and comprehensive key stage division result. The result synthesizes the information of the time dimension and multiple control parameters and can provide more powerful support for the optimal control of the wet oxidation process of VCSEL chips. For example, after multiple iterations and adjustments, several key oxidation stages are determined, each stage has a clear time range and characteristic description. Fusing the stages can obtain a complete key stage division result for guiding the subsequent oxidation process control.

[0025] In a possible implementation manner, if the characteristic change trend of the multi-dimensional clustering result within the time series window is inconsistent with the clustering result in the time dimension, then perform window segmentation on the time series window and continue the matching iteration. Step S144 further includes step S1441, evaluate the characteristic change density within the window for the time series window as follows: ; where represents the local probability density corresponding to the feature vector at each time point , is the total number of multi-dimensional clustering results, represents the bandwidth function, which is adaptively adjusted by temperature , humidity , aperture , represents any one multi-dimensional clustering result, is the weight coefficient, representing the influence degree of the th multi-dimensional clustering result on density estimation, is the kernel function, is the time point at the feature vector, is the th feature vector of the multi-dimensional clustering result. Specifically, for a given time series window, it is necessary to evaluate the feature change density within the window, and the formula is used for calculation. First, represents the local probability density corresponding to the feature vector at each time point . The result calculated by the formula can reflect the comprehensive change of each feature during the oxidation process at a specific time point. Then, is the total number of multi-dimensional clustering results. Multi-dimensional clustering is the classification of control parameters such as temperature, humidity, pore size, etc. The total number represents all possible clustering situations. represents the bandwidth function, which is adaptively adjusted by temperature , humidity , pore size . This indicates that the bandwidth function is not fixed, but dynamically adjusted according to the changes of key parameters such as temperature, humidity, and pore size during the oxidation process to better adapt to different oxidation states. represents any multi-dimensional clustering result, is the weight coefficient, representing the influence degree of the th multi-dimensional clustering result on density estimation. Different clustering results play different roles in the evaluation of feature change density. The weight coefficient can be adjusted according to the actual situation to highlight the contribution of important clustering results to density estimation. is the kernel function, is the time point at the feature vector, is the th feature vector of the multi-dimensional clustering result. The role of the kernel function is to map the feature vector to a high-dimensional space for better density estimation.

[0026] Step S1442: Perform window segmentation based on the evaluation result of feature change density to continue the matching iteration. Specifically, once the evaluation result of feature change density is obtained, window segmentation can be carried out according to this result. If the feature change density is uneven within a certain time series window or does not conform to the expected feature change trend, this window can be considered for segmentation. For example, if the feature change density in a certain area is high while that in another area is low, it means that there are differences in the oxidation processes of these two areas, and the window needs to be segmented and analyzed separately. Through window segmentation, the complex oxidation process can be further refined to more accurately perform trend consistency matching. After segmentation, continue the matching iteration process, that is, re-evaluate the feature change density and perform trend consistency matching on the new window until the best stage division result is found. By continuously iterating, the stage division of the wet oxidation process of the VCSEL chip can be gradually optimized, improving the accuracy and reliability of the division.

[0027] Step S200: Install monitoring sensors to monitor the wet oxidation process of the VCSEL chip with the monitoring sensors and establish a real-time data set. The real-time data set includes oxidation environment temperature data, humidity data, aperture change data, and oxidation time data. Specifically, during the wet oxidation process of the VCSEL chip, first select a suitable type of monitoring sensor according to its characteristics and requirements and install it reasonably. For example, use a temperature sensor to monitor the oxidation environment temperature, a humidity sensor to measure the environmental humidity, a special device to monitor the aperture change, and a timer to record the oxidation time. During the oxidation process, the sensors work continuously, transmit the collected data to the data processing system at a certain frequency, and then integrate the data to establish a real-time data set, including oxidation environment temperature data, humidity data, aperture change data, and oxidation time data, providing important basic data support for subsequent oxidation process analysis, stage division, and control parameter optimization.

[0028] Step S300: Configure the global control layer, perform key stage matching through the stage division features and the real-time data set, and activate the optimization priority corresponding to the key stage division result with the matching result. Initialize the global control layer with the activated optimization priority and process the real-time data set through the initialized global control layer to generate a global decision. Specifically, first configure the global control layer. This layer uses the stage division features and the real-time data set to perform key stage matching. After determining the key stage where the current oxidation process is located, activate the optimization priority corresponding to the key stage and initialize the global control layer with this. The initialized global control layer processes the real-time data set, comprehensively considering the oxidation state, optimization priority, and various parameters, and generates a global decision through analysis and calculation to achieve the overall optimization of the wet oxidation process of the VCSEL chip and improve its quality and efficiency.

[0029] In a possible implementation, a global control layer is configured to perform key stage matching through the stage division features and the real-time data set, and activate the optimization priority corresponding to the key stage division result with the matching result. The global control layer is initialized with the activated optimization priority, and the real-time data set is processed through the initialized global control layer to generate a global decision. Step S300 further includes step S310. The global control layer includes a decision module and a feedback module. After the global control layer is initialized, the decision module is called to process the real-time data set to generate a global decision. After any global decision is generated, the global decision is sent to the feedback module to establish a feedback window, and decision optimization is performed with the feedback window during the execution of the global decision. Specifically, the global control layer is composed of a decision module and a feedback module. After the global control layer is initialized, the decision module processes the real-time data set, comprehensively considers various factors to generate a global decision. After the global decision is generated, it is sent to the feedback module. The feedback module establishes a feedback window and optimizes the global decision according to the information collected by the feedback window during the execution of the global decision. Through the collaborative action of the decision module and the feedback module, the control accuracy and efficiency of the wet oxidation process of the VCSEL chip are continuously improved.

[0030] Step S400, after the global decision is sent to the parameter association control layer, the parameter association control layer analyzes the global decision to establish a global goal. Specifically, after the global control layer generates a global decision, it sends it to the parameter association control layer. After receiving it, the parameter association control layer carefully analyzes the global decision, extracts key points and integrates them, so as to establish a clear and refined global goal. This goal makes the adjustment of subsequent control parameters more targeted and purposeful, so as to better achieve the optimized control of the wet oxidation process of the VCSEL chip. For example, if the global decision proposes to adjust the temperature and humidity parameters in a specific stage to optimize the oxidation effect, after analysis by the parameter association control layer, it is converted into specific temperature and humidity target values, and achieving these target values is taken as the global goal. Subsequent operations can be carried out around this global goal, making the adjustment of each control parameter more targeted and purposeful, so as to better achieve the optimized control of the wet oxidation process of the VCSEL chip.

[0031] Step S500: Predict the linkage relationship between parameters based on the global objective, configure a synchronous adjustment strategy based on the prediction result of the linkage relationship, optimize the global decision based on the synchronous adjustment strategy, and then call the local control layer to execute the optimized global decision. Specifically, based on the global objective, through the analysis of historical data and the application of mathematical models and algorithms, predict the linkage relationship between parameters. Configure a synchronous adjustment strategy based on this prediction result and integrate it into the global decision for optimization, making the global decision more comprehensive and reasonable. Finally, call the local control layer to execute the optimized global decision. The local control layer interacts with the oxidation equipment and sensors to precisely adjust each control parameter, ensuring that the oxidation process proceeds according to the optimized decision. At the same time, continuously monitor the actual oxidation state and feedback information for further optimization of the decision.

[0032] In a possible implementation manner, for step S500 of predicting the linkage relationship between parameters based on the global objective, configuring a synchronous adjustment strategy based on the prediction result of the linkage relationship, optimizing the global decision based on the synchronous adjustment strategy, and then calling the local control layer to execute the optimized global decision, step S500 further includes step S510: Extract multi-dimensional characteristic parameters and define the multi-dimensional characteristic parameters as parameter nodes. The parameter nodes include a temperature node, a humidity node, a pore size change node, and a time node. Specifically, first, extract multi-dimensional characteristic parameters from the wet oxidation process of the VCSEL chip. These parameters reflect different aspects of the oxidation process and have an important impact on the oxidation effect. Define the multi-dimensional characteristic parameters as parameter nodes to establish a basic structure for subsequent analysis and prediction. Among them, the parameter nodes include a temperature node, a humidity node, a pore size change node, and a time node. The temperature node represents the change in the ambient temperature during the oxidation process; the humidity node reflects the humidity level of the oxidation environment; the pore size change node reflects the dynamic change of the chip pore size during the oxidation process; and the time node records the time experienced by the oxidation process. By clarifying the parameter nodes, the relationship between each parameter can be analyzed more systematically.

[0033] Step S520: Call historical data based on the global objective to establish an initial causal relationship between parameter nodes. Specifically, after determining the parameter nodes, call historical data based on the global objective. The historical data includes the values of each parameter and the corresponding oxidation results during past oxidation processes. Through the analysis of historical data, attempt to establish an initial causal relationship between parameter nodes. For example, observe the situation of pore size change and oxidation time under different temperature and humidity conditions, and preliminarily judge whether temperature and humidity have a causal impact on pore size change and oxidation time. If it is found in the historical data that when the temperature increases, the pore size change often accelerates, then an initial causal relationship can be preliminarily established between the temperature node and the pore size change node.

[0034] Step S530: Construct the conditional probability table between each parameter node and its parent node according to the initial causal relationship, and perform the iteration of the causal relationship graph. Specifically, according to the established initial causal relationship, construct the conditional probability table between each parameter node and its parent node. For example, if it is determined that the temperature node is the parent node of the aperture change node, then the probability of aperture change under different temperature conditions can be calculated based on historical data to form the conditional probability table between the temperature node and the aperture change node. By constructing the conditional probability table, the relationship between parameter nodes can be described more quantitatively. Then, perform the iteration of the causal relationship graph. In each iteration, according to the new observation data and the existing conditional probability table, continuously adjust and optimize the causal relationship between parameter nodes. For example, if the new observation data shows that in some cases, humidity also has an important impact on aperture change, then the relationship between the humidity node and the aperture change node needs to be adjusted, and the corresponding conditional probability table needs to be updated.

[0035] Step S540: Complete the prediction of the linkage relationship between parameters according to the iteration result of the causal relationship graph. Specifically, according to the iteration result of the causal relationship graph, complete the prediction of the linkage relationship between parameters. After multiple iterations, the causal relationship graph will gradually tend to be stable. At this time, the linkage relationship between each parameter node can be predicted more accurately. For example, the trend of aperture change and the possible range of oxidation time can be predicted under specific temperature and humidity conditions. The prediction result of the linkage relationship can provide an important basis for the optimization control of the oxidation process, helping to formulate a more reasonable control strategy to achieve the high efficiency and stability of the wet oxidation process of VCSEL chips.

[0036] In a possible implementation, the linkage relationship between parameters is predicted based on the global objective, a synchronization adjustment strategy is configured based on the prediction result of the linkage relationship, and after optimizing the global decision based on the synchronization adjustment strategy, the local control layer is called to execute the optimized global decision. Step S500 further includes step S550 of establishing an anomaly observation layer, which is used to monitor the execution result and accumulate anomaly monitoring. The anomaly monitoring accumulation includes frequency accumulation and outlier accumulation. Specifically, first, an anomaly observation layer is established. The anomaly observation layer is used to closely monitor the execution result to ensure that the oxidation process proceeds according to the expected global decision and optimization strategy. One of the main functions of the anomaly observation layer is to monitor the execution result. During the oxidation process, continuously pay attention to the actual changes of each key parameter and the execution effect of the global decision. For example, in real-time monitor whether parameters such as temperature, humidity, pore size change, and oxidation time are within a reasonable range, and at the same time check whether the local control layer accurately executes the optimized global decision. By continuously collecting and analyzing these data, the anomaly observation layer can promptly detect any situation that does not conform to the expected result. The anomaly observation layer performs anomaly monitoring accumulation, including frequency accumulation. Frequency accumulation refers to counting the frequency of the occurrence of anomaly situations. For example, if the temperature exceeds the normal range a relatively large number of times within a certain period, the occurrence frequency of this anomaly situation will be recorded. Through frequency accumulation, it can be understood which anomaly situations occur more frequently, so as to focus on these problem areas and take corresponding measures for adjustment and optimization. In addition to frequency accumulation, the anomaly monitoring accumulation also includes outlier accumulation. Outlier accumulation mainly records and accumulates the parameter values that exceed the normal range. For example, if a large outlier suddenly appears in the pore size change, the anomaly observation layer will record the outlier and add it to the outlier accumulation. Through outlier accumulation, it can be understood the severity of the anomaly situation and which parameters' outliers have a greater impact on the oxidation process.

[0037] Step S560, perform anomaly reporting according to the triggering result of the anomaly monitoring accumulation result. Specifically, finally, perform anomaly reporting according to the triggering result of the anomaly monitoring accumulation result. When the anomaly monitoring accumulation reaches a certain threshold or meets specific triggering conditions, the anomaly observation layer will issue an anomaly report. The triggering result can be set according to the actual situation. For example, when the anomaly frequency exceeds a certain number of times or the outlier accumulation exceeds a certain specific value, the anomaly reporting will be triggered. The anomaly reporting can be carried out in various ways, such as emitting an alarm sound, sending an email or a text message to notify relevant personnel, etc., so as to promptly take measures for processing and avoid the anomaly situation having a greater impact on the oxidation process.

[0038] In a possible implementation, anomaly reporting is performed according to the triggering result of the anomaly monitoring cumulative result. Step S560 further includes step S561 of establishing an anomaly triggering threshold, and verifying the anomaly monitoring cumulative result through the anomaly triggering threshold to generate a triggering result. Specifically, first, an anomaly triggering threshold is established. The threshold is used to determine whether the anomaly monitoring cumulative result reaches the standard for anomaly reporting. The setting of the anomaly triggering threshold needs to comprehensively consider factors such as the characteristics of the wet oxidation process of the VCSEL chip, historical data, and the tolerance for abnormal situations. For example, the appropriate threshold can be determined according to the frequency and severity of anomalies occurring in previous oxidation processes, as well as the impact on chip quality. Then, the established anomaly triggering threshold is used to verify the anomaly monitoring cumulative result. The anomaly monitoring cumulative result includes information such as frequency accumulation and anomaly value accumulation. The cumulative result is compared with the anomaly triggering threshold to determine whether it exceeds the threshold. If the cumulative result exceeds the threshold, it indicates that the abnormal situation is relatively serious and corresponding measures may need to be taken for processing. By comparing the anomaly monitoring cumulative result and the anomaly triggering threshold, a triggering result is generated. There are two possible triggering results: verification passed result and verification failed result. If the anomaly monitoring cumulative result does not exceed the anomaly triggering threshold, the generated triggering result is the verification failed result, indicating that the current abnormal situation is still within the acceptable range and no anomaly reporting is required. If the anomaly monitoring cumulative result exceeds the anomaly triggering threshold, the generated triggering result is the verification passed result.

[0039] Step S562, if the triggering result is a verification passed result, then anomaly reporting is performed. Specifically, when the triggering result is a verification passed result, anomaly reporting is performed. The method of anomaly reporting can be selected according to the actual situation. For example, an alarm sound can be emitted, abnormal information can be displayed on the monitoring interface, an email or text message can be sent to notify relevant personnel, etc. The purpose of anomaly reporting is to timely remind relevant personnel to pay attention to the abnormal situation so that corresponding measures can be taken for processing to avoid greater impact of the abnormal situation on the oxidation process and chip quality.

[0040] The embodiments of the present application adopt fitting observations of the oxidation process, divide key stages, and set the optimization priorities and characteristics of control parameters for each stage. Monitoring sensors are used to collect data on the temperature, humidity, aperture change, and oxidation time of the VCSEL chip during the wet oxidation process in real time. Through the global control layer, key stages are matched based on stage characteristics and the real-time data set, the optimization priorities are activated, and a global decision is generated. The global decision is passed to the parameter correlation control layer for parsing, a global goal is established, the global decision is optimized through parameter linkage prediction and synchronous adjustment strategies, the local control layer is called to execute the optimized decision, and the control parameters are dynamically adjusted according to the characteristics of different stages to ensure that the oxidation process is always carried out in an optimal state, achieving the technical effect of precise control of the wet oxidation process of the VCSEL chip.

[0041] The above specific embodiments do not constitute a limitation on the protection scope of the present application. Those skilled in the art should understand that various modifications, combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principle of the present application shall be included within the protection scope of the present application. In some cases, the actions or steps recorded in the present application can be executed in a different order from that in the embodiments and still achieve the desired results. Additionally, the processes depicted in the drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. A method for optimizing the wet oxidation aperture of a VCSEL chip by dynamic adjustment, characterized in that The method includes: Performing a fitting observation on the oxidation process, dividing the oxidation process into stages according to the fitting observation results, and establishing a key stage division result. Wherein, each key stage division result is set with an optimization priority of control parameters, and the key stage division result is configured with stage division characteristics; Deploying monitoring sensors to monitor the wet oxidation process of the VCSEL chip with the monitoring sensors, and establishing a real-time data set. The real-time data set includes oxidation environment temperature data, humidity data, aperture change data, and oxidation time data; Configuring a global control layer, performing key stage matching through the stage division characteristics and the real-time data set, activating the optimization priority of the corresponding key stage division result with the matching result, initializing the global control layer with the activated optimization priority, and processing the real-time data set through the initialized global control layer to generate a global decision; After sending the global decision to the parameter correlation control layer, parsing the global decision with the parameter correlation control layer to establish a global target; Predicting the linkage relationship between parameters with the global target, configuring a synchronous adjustment strategy based on the linkage relationship prediction result, optimizing the global decision based on the synchronous adjustment strategy, and then calling the local control layer to execute the optimized global decision; The step of dividing the oxidation process into stages according to the fitting observation results and establishing a key stage division result further includes: Performing a timing fitting of control parameters according to the fitting observation results to establish a timing fitting curve; Extracting feature points from the timing fitting curve, and establishing a clustering result in the time dimension according to the feature point extraction results; Performing multi-dimensional clustering of control parameters using the fitting observation results to establish a multi-dimensional clustering result. The features extracted by the multi-dimensional clustering include temperature change rate, humidity volatility, aperture change rate, aperture stability, and cumulative oxidation time effect; Performing clustering fusion on the clustering results in the time dimension and the multi-dimensional clustering results to establish a key stage division result.

2. The wet oxidation aperture optimization method of the VCSEL chip using dynamic adjustment according to claim 1, characterized in that, The step of performing clustering fusion on the clustering results in the time dimension and the multi-dimensional clustering results further includes: Establishing a timing window according to the clustering result in the time dimension; Using the timing window to match the trend consistency of the multi-dimensional clustering results within the timing window; If the feature change trend of the multi-dimensional clustering results within the timing window is consistent with the clustering result in the time dimension, then retain the stage division; If the feature change trend of the multi-dimensional clustering results within the timing window is inconsistent with the clustering result in the time dimension, then perform window segmentation on the timing window and continue the matching iteration; Completing the clustering fusion according to the matching iteration results and the retained stage division results.

3. The method for optimizing the wet oxidation aperture of a VCSEL chip using dynamic adjustment as described in claim 2, wherein, The step of performing window segmentation on the timing window and continuing the matching iteration further includes: Evaluating the feature change density within the timing window as follows: ； Among them, characterizes the feature vector at each time point corresponding local probability density, is the total number of multi-dimensional clustering results, characterizes the bandwidth function, which is adaptively adjusted by temperature , humidity , aperture ; represents any multi-dimensional clustering result, is the weight coefficient, which characterizes the influence degree of the th multi-dimensional clustering result on density estimation, is the kernel function, is the time point at which the feature vector is located, is the th feature vector of the multi-dimensional clustering result; Performing window segmentation based on the feature change density evaluation result to continue the matching iteration.

4. The wet oxidation aperture optimization method of the VCSEL chip using dynamic adjustment as described in claim 1, wherein The global control layer includes a decision-making module and a feedback module. After the global control layer is initialized, the decision-making module is called to process the real-time data set to generate a global decision. When any global decision is generated, the global decision is sent to the feedback module to establish a feedback window, and the decision is optimized with the feedback window during the execution of the global decision.

5. The wet oxidation aperture optimization method for VCSEL chips using dynamic adjustment as described in claim 1, characterized in that, The prediction of the linkage relationship between parameters with the global target further includes: Extract multi-dimensional feature parameters and define the multi-dimensional feature parameters as parameter nodes. The parameter nodes include a temperature node, a humidity node, a pore size change node, and a time node; Based on the global target, call historical data to establish an initial causal relationship between parameter nodes; Construct a conditional probability table between each parameter node and its parent node according to the initial causal relationship, and perform causal graph iteration; Complete the prediction of the linkage relationship between parameters according to the iteration result of the causal graph.

6. The wet oxidation aperture optimization method of the VCSEL chip using dynamic adjustment according to claim 1, characterized in that, The method further includes: Establish an abnormal observation layer, which is used to monitor the execution result and perform abnormal monitoring accumulation. The abnormal monitoring accumulation includes frequency accumulation and abnormal value accumulation; Report an abnormality according to the trigger result of the abnormal monitoring accumulation result.

7. The wet oxidation aperture optimization method for VCSEL chips using dynamic adjustment as described in claim 6, characterized in that, The method further includes: Establish an abnormal trigger threshold, and verify the abnormal monitoring accumulation result through the abnormal trigger threshold to generate a trigger result; If the trigger result is a verification passed result, report an abnormality.

Images