Parameter ecological niche regulation and control method in semiconductor wafer manufacturing

Abstract

The invention discloses a parameter ecological niche regulation and control method in semiconductor wafer manufacturing, and relates to the technical field of semiconductor processing, and the method comprises the steps: collecting etching parameter data, obtaining corresponding etching quality evaluation information, carrying out abnormal quality parameter extraction, and determining an optimization quality target; analyzing an influence relationship between the parameter data and the etching quality evaluation information, and carrying out same parameter data item superposition to obtain an abnormal parameter ratio of each parameter data; screening optimization parameters according to the abnormal parameter ratio; and according to the optimization quality target, the optimization parameters and the influence relationship, constructing an optimization space for optimization, determining the optimization configuration of the optimization parameters, and performing process control. The technical problem that in the prior art, due to continuous reduction of the device size and continuous improvement of the integration degree, a traditional etching technology cannot meet the semiconductor wafer machining precision requirement is solved, and the technical effect that the machining precision of the semiconductor wafer etching process is improved by analyzing and optimizing key parameters of the etching process is achieved.

Description

Technical Field

[0001] This invention relates to the field of semiconductor processing technology, and more specifically to a method for parameter niche control in semiconductor wafer manufacturing. Background Technology

[0002] With the rapid development of the semiconductor industry, the requirements for wafer fabrication technology are becoming increasingly stringent. Improvements in wafer fabrication technology directly impact the performance, quality, and market competitiveness of semiconductor chips. Among these processes, etching is an indispensable part of semiconductor manufacturing. Its main purpose is to remove unwanted material layers from the wafer surface using physical or chemical methods to form precise circuit patterns and structures.

[0003] Furthermore, the etching precision requirements are extremely high, necessitating consistent etching results across every wafer to ensure device performance and reliability. However, due to the continuous miniaturization of device dimensions and the increasing integration density, traditional etching techniques can no longer meet the processing precision requirements. Summary of the Invention

[0004] This application provides a method for parametric niche control in semiconductor wafer manufacturing, which addresses the technical problem that traditional etching techniques can no longer meet the precision requirements of semiconductor wafer processing due to the continuous shrinkage of device size and the continuous increase in integration.

[0005] This application provides a parameter niche control method in semiconductor wafer manufacturing. The method includes: collecting etching parameter data and constructing a parameter dataset, wherein the parameter data includes etching rate, selectivity, uniformity, temperature, and gas flow rate, and the parameter data has equipment batch labels; obtaining etching quality assessment information corresponding to each parameter data, extracting abnormal quality parameters from the etching quality assessment information, and determining the optimization quality target; normalizing the parameter dataset, analyzing the influence relationship between the parameter data and the etching quality assessment information based on the abnormal quality parameters; aligning the parameter data with the target based on the influence relationship, superimposing the same parameter data items based on the influence coefficient of the influence relationship, and obtaining the abnormal parameter ratio of each parameter data; screening optimization parameters based on the abnormal parameter ratio; constructing an optimization space based on the optimization quality target, the optimization parameters, and the influence relationship, optimizing to maximize the optimization quality target, determining the optimal configuration of the optimization parameters, and performing process control based on the optimized configuration.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages: The parameter niche control method in semiconductor wafer manufacturing provided in this application relates to the field of semiconductor processing technology. It involves collecting etching parameter data, extracting abnormal quality parameters, determining optimized quality targets, analyzing the influence relationship between parameter data and etching quality assessment information, obtaining the abnormal parameter ratio of each parameter data, screening optimized parameters, constructing an optimization space based on the optimized quality targets and influence relationships, determining the optimal configuration of optimized parameters, and controlling the process. This solves the technical problem that traditional etching technology can no longer meet the precision requirements of semiconductor wafer processing due to the continuous shrinking of device size and the continuous increase in integration. It achieves the technical effect of improving the processing precision of semiconductor wafer etching process and thus improving product quality by analyzing and optimizing key parameters of the etching process. Attached Figure Description

[0007] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0008] Figure 1 This is a schematic diagram of a parameter niche control method in semiconductor wafer manufacturing provided in an embodiment of this application; Figure 2 A flowchart illustrating the process of obtaining the abnormal parameter ratio of each parameter data in the parameter niche control method in semiconductor wafer manufacturing provided in this application embodiment; Figure 3 This is a schematic diagram illustrating the process of obtaining optimized configuration of parameters in the parameter niche control method for semiconductor wafer manufacturing provided in this application embodiment. Detailed Implementation

[0009] This application provides a method for parametric niche control in semiconductor wafer manufacturing, which addresses the technical problem that traditional etching techniques can no longer meet the precision requirements of semiconductor wafer processing due to the continuous shrinkage of device size and the continuous increase in integration.

[0010] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0011] It should be noted that the terms "first," "second," etc., in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, 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 is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules not explicitly listed or inherent to such processes, methods, products, or devices.

[0012] Example 1 like Figure 1 As shown, this application provides a method for parameter niche control in semiconductor wafer manufacturing, the method comprising: Etching parameter data is collected and a parameter dataset is constructed. The parameter data has equipment batch labels.

[0013] Furthermore, the steps in the embodiments of this application also include: The distribution number of etching equipment is obtained, and parallel acquisition channels are constructed. The parallel acquisition channels are used to establish IoT data communication with the etching equipment respectively. The change rates of etching rate, selectivity, uniformity, temperature, and gas flow rate are obtained respectively, and the acquisition frequency is set. Data acquisition equipment is set based on the acquisition frequency, and the transmission frequency of the parallel acquisition channels is set using the acquisition frequency to obtain the parameter dataset.

[0014] It should be understood that, for the etching process in semiconductor wafer fabrication, key parameter data during the etching process, such as etching rate, selectivity, uniformity, temperature, and gas flow rate, are collected and compiled into a parameter dataset. Furthermore, the collected parameter data includes equipment batch labels, which helps identify performance differences between different batches of equipment.

[0015] Specifically, the distribution of etching equipment is determined through on-site surveys or system database queries, i.e., the total number of etching equipment and their locations. Based on the number and location of the equipment, an IoT technology-based data acquisition network is constructed. Each etching device connects to a data acquisition node via a sensor or interface. These data acquisition nodes are connected to a central server via wired or wireless means, forming parallel data acquisition channels. These channels establish IoT data connections with the etching equipment, ensuring that data can be transmitted from the equipment to the data center in real time and accurately.

[0016] Furthermore, the rates of change of the etching rate, selectivity, uniformity, temperature, and gas flow rate are acquired using sensors or measuring devices, and an appropriate acquisition frequency is set for each parameter based on its importance and rate of change. For example, a higher acquisition frequency can be set for parameters that change rapidly (such as temperature), while a lower acquisition frequency can be set for parameters that change slowly (such as etching rate).

[0017] Furthermore, based on the acquisition frequency, a corresponding model of data acquisition equipment is configured to ensure that data can be collected from the etching equipment at the specified frequency. The transmission frequency of the parallel acquisition channels is set using the acquisition frequency, for example, by adjusting network bandwidth and optimizing data transmission protocols, to ensure that the transmission frequency of the parallel acquisition channels matches the acquisition frequency of the data acquisition equipment, thereby ensuring the real-time performance and accuracy of the data. Through the data acquisition equipment and the parallel acquisition channels, parameter datasets from all etching equipment are acquired and integrated, serving as the basis for subsequent analysis and optimization.

[0018] Furthermore, the steps in the embodiments of this application also include: A standard experimental time is set, and the etching rate, selectivity, uniformity, temperature, and gas flow rate are collected respectively to obtain a standard time dataset; the change curves of each parameter are plotted according to the standard time dataset to obtain the change rate of each parameter; the maximum change rate is selected, and the sampling frequency is set to twice the maximum change rate based on the Nyquist sampling theorem.

[0019] Specifically, the process of setting the sampling frequency can be as follows: First, in order to obtain the variation characteristics of etching parameters under stable operating conditions, a standard experimental time needs to be set. This time should be long enough to cover all significant changes in parameters that may occur during the etching process. Within the standard experimental time, data on key parameters such as etching rate, selectivity, uniformity, temperature, and gas flow rate are collected to obtain a standard time dataset.

[0020] Furthermore, using the data from the aforementioned standard time dataset, curves showing the variation of each parameter over time are plotted. These curves are then analyzed, and by employing methods such as adjusting the slope of the curves or using other mathematical techniques, the average rate of change of each parameter over the standard experimental time is calculated.

[0021] Furthermore, the maximum rate of change is selected from all the rate of change of all parameters, and the sampling frequency is set to twice the maximum rate of change based on the Nyquist sampling theorem. Specifically, the Nyquist sampling theorem states that to avoid aliasing, the sampling frequency must be greater than or equal to twice the highest frequency in the signal. Therefore, this application considers the maximum rate of change as the highest frequency in the signal and sets twice the maximum rate of change as the sampling frequency. This ensures that during the etching process, the data acquisition device can capture all significant changes in all parameters, thereby providing accurate and reliable data support for subsequent data analysis and process optimization.

[0022] Obtain etching quality assessment information corresponding to each parameter data, extract abnormal quality parameters from the etching quality assessment information, and determine the optimization quality target.

[0023] Optionally, in the semiconductor wafer etching process, in addition to collecting etching parameter data, it is also necessary to perform a quality assessment on the etched wafer to obtain etching quality assessment information corresponding to each parameter data. This quality assessment information typically includes key indicators such as wafer surface flatness, etching depth consistency, and edge roughness. By using the wafer quality assessment information corresponding to the etching parameter data, a correlation between parameters and quality can be established.

[0024] Furthermore, after obtaining the etching quality assessment information, this information needs to be analyzed to extract abnormal or unexpected quality parameters. These abnormal quality parameters may be caused by improper etching parameter settings, equipment performance fluctuations, or other factors. Based on these abnormal quality parameters, the quality parameters that need optimization are determined, and clear optimization quality objectives are set. For example, optimization objectives can be set to reduce wafer surface roughness, improve the consistency of etching depth, or increase wafer yield. By determining the optimization quality objectives, direction and basis are provided for subsequent optimization work.

[0025] The parameter dataset is normalized, and the influence relationship between the parameter data and the etching quality assessment information is analyzed based on the abnormal quality parameters. The parameter data is aligned with the target, and the same parameter data items are superimposed based on the influence coefficient of the influence relationship to obtain the abnormal parameter ratio of each parameter data.

[0026] It should be understood that direct comparison or analysis of different parameters may lead to distorted results due to potential differences in their physical meaning and numerical range. Therefore, it is necessary to normalize the parameter dataset. An appropriate normalization method can be selected based on the specific circumstances, such as the min-max normalization method. This method converts the data to values ​​between 0 and 1 by subtracting the minimum value and dividing by the range (maximum value - minimum value). This transforms the values ​​of different parameters to the same scale for subsequent analysis.

[0027] Furthermore, after identifying the abnormal quality parameters, it is necessary to further analyze the relationship between these parameters and the etching quality assessment information to identify parameters that have a significant impact on etching quality, thereby determining the focus of optimization. Specifically, statistical methods (such as Pearson correlation coefficient, Spearman rank correlation coefficient, etc.) can be used to calculate the correlation between parameter data and etching quality assessment information. If there is a linear or nonlinear relationship between the parameters and etching quality, regression analysis can be used to build a predictive model, identify parameters that have a significant impact on etching quality, and estimate the degree of influence of these parameters. For complex nonlinear relationships, machine learning models (such as decision trees, random forests, neural networks, etc.) can be used to build predictive models. These models can automatically learn the relationship between parameters and etching quality and provide more accurate prediction results.

[0028] Analysis reveals the degree and direction of each parameter's influence on etching quality, i.e., the relationship between parameter data and etching quality assessment information, providing an important basis for subsequent optimization work.

[0029] Furthermore, such as Figure 2 As shown, the steps in the embodiments of this application further include: Based on the alignment parameter data, a parameter matrix is ​​set, with each horizontal column representing a complete abnormal quality parameter item and each vertical column representing the parameter data. The parameter matrix is ​​weighted using the influence coefficient as a weighting factor to obtain a weighted parameter matrix, which represents the weighted data of the parameters in each vertical column. The mean vector and standard deviation vector for each parameter data point are calculated based on the weighted parameter matrix. The abnormal parameter ratio is then calculated on the weighted parameter matrix using the mean and standard deviation vectors to obtain the abnormal parameter ratio for each parameter data point.

[0030] The abnormal parameter ratio for obtaining each parameter data includes: ,in, , , For mean vector, For the standard deviation vector, For the weighted parameter matrix, Let be the i-th parameter data in the weighted parameter matrix, and n be the number of parameter data.

[0031] Optionally, to more accurately analyze the impact of each parameter on etching quality, the parameter data needs to be aligned according to their influence relationships. This can be achieved by sorting or grouping the parameter datasets based on these relationships, ensuring that parameter data items with similar influence relationships correspond to each other. Furthermore, the data items with the same parameter are superimposed; that is, each parameter data item is multiplied by its corresponding influence coefficient, and the results are summed. The superimposed result reflects the comprehensive contribution of that parameter data to etching quality anomalies. This is achieved by calculating the comprehensive contribution ratio of each parameter data and the ratio of abnormal parameters for each parameter data.

[0032] Specifically, firstly, based on the alignment parameter data, a parameter matrix is ​​set, where each row represents a complete abnormal quality parameter item, i.e., the set of all parameter data under that abnormal quality parameter, and each column represents different parameter data. Furthermore, to consider the different degrees of influence of different parameters on etching quality, the parameter matrix is ​​weighted. The aforementioned influence coefficients can be used as weighting coefficients to weight each column of the parameter matrix, obtaining a weighted parameter matrix.

[0033] Furthermore, to quantify the distribution of each parameter data point in the weighted parameter matrix, the mean vector and standard deviation vector for each parameter data point are calculated based on the weighted parameter matrix. This includes calculating the mean for each column of the weighted parameter matrix to obtain a mean vector, and calculating the standard deviation for each column of the weighted parameter matrix to obtain a standard deviation vector.

[0034] Furthermore, based on the mean vector and standard deviation vector, and in conjunction with the outlier ratio calculation formula, the outlier ratio is calculated on the weighted parameter matrix, wherein the outlier ratio calculation formula is: ,in, , , The ratio of the outlier parameters to the i-th parameter data. For mean vector, For the standard deviation vector, For the weighted parameter matrix, Let be the i-th parameter data in the weighted parameter matrix, and n be the number of parameter data.

[0035] The ratio of abnormal parameters is obtained by calculation. The ratio of abnormal parameters can reflect the contribution of each parameter to the etching quality anomaly, which helps to identify the key factors affecting the etching quality.

[0036] Based on the abnormal parameter ratio, optimization parameters are screened, and an optimization space is constructed using the optimization quality target, the optimization parameters, and the influence relationship. Optimization is performed with the optimization quality target being maximized to determine the optimal configuration of the optimization parameters, and process control is carried out based on the optimal configuration.

[0037] Specifically, optimized parameters are obtained by screening based on the outlier parameter ratio. Before screening optimized parameters, one or more screening criteria are set. These criteria can be based on thresholds, percentages, or other statistical indicators of the outlier parameter ratio. For example, parameters with outlier parameter ratios higher than a certain threshold are selected as potential optimized parameters. Further, optimized parameters that meet the conditions are selected from the parameter dataset according to the set screening criteria. These optimized parameters are key parameters that have a significant impact on etching quality and require close attention and adjustment.

[0038] At the same time, when selecting and optimizing parameters, it is also necessary to consider actual process limitations and feasibility. For example, some parameters may be limited by equipment performance, material properties, or process conditions, and cannot be directly adjusted. Therefore, these factors need to be comprehensively considered during the selection process to ensure that the selected parameters have practical operability and optimization potential.

[0039] After selecting the optimization parameters, the optimization order and priority can be determined based on factors such as the degree of influence of each parameter on etching quality, optimization difficulty, and potential benefits. This ensures that, within limited resources and time, priority is given to those parameters that have the greatest impact on etching quality and the most obvious optimization effect, thereby improving the operability and efficiency of the optimization process.

[0040] Furthermore, such as Figure 3 As shown, the steps in the embodiments of this application further include: When the optimization quality objective is singular, an evaluation function is fitted based on the optimization quality objective, the optimization parameters, and the influence relationship; constraint information of each parameter data is obtained; based on the constraint information and the evaluation function, the optimization space is constructed, and optimization rules are configured; parameter optimization is performed through the optimization space, including: constructing an initial solution, which is the historical best solution; setting the optimization direction and optimization step size with the initial solution as the starting point; obtaining an updated solution; evaluating the updated solution using the evaluation function; replacing the updated solution when its evaluation is higher than that of the initial solution; iterating in this way until the optimization objective or the number of iterations is reached, thereby obtaining the optimized configuration of the optimization parameters.

[0041] It should be understood that an optimization space is constructed based on the optimization quality objective, the optimization parameters, the influencing relationships, and the influencing relationships between them. The optimization space is a multi-dimensional parameter space containing all possible parameter configuration combinations. Within this optimization space, parameter optimization is performed with the criterion of maximizing the optimization quality objective, to find a set of parameter configurations that optimizes the etching process performance, i.e., satisfies or exceeds the predetermined optimization quality objective.

[0042] Specifically, when the optimization quality objective is a single objective, such as etching depth or uniformity, an evaluation function is fitted based on the optimization quality objective, optimization parameters, and their interrelationships. This evaluation function quantifies the performance of the etching process under different parameter configurations, providing an evaluation basis for the subsequent optimization process.

[0043] Furthermore, constraint information for each parameter data is obtained based on equipment performance, material properties, or process conditions. A specific optimization space is then constructed based on this constraint information and the evaluation function. Simultaneously, optimization rules, including optimization algorithms, are configured for this optimization space to ensure the effectiveness and efficiency of the optimization process.

[0044] Furthermore, parameter optimization is performed through the optimization space. First, an initial solution is constructed based on historical best solutions. Then, starting from the initial solution, an optimization direction and step size are set according to the parameter characteristics of the initial solution, and a search is performed according to the optimization direction and step size to obtain a new parameter configuration (i.e., an updated solution). The updated solution is evaluated using the evaluation function. If its performance is better than the initial solution, the initial solution is replaced with the updated solution. This process is iterated continuously until the predetermined optimization objective is reached or the maximum number of iterations is reached. The final parameter configuration obtained is the optimized configuration of the optimized parameters.

[0045] Furthermore, the steps in the embodiments of this application also include: When there are multiple optimization quality objectives, the influence direction of the optimization parameters on the optimization quality objectives is obtained according to the optimization quality objectives; based on the influence direction of the optimization parameters on each optimization quality objective, game optimization quality objectives are extracted; optimization parameters are optimized with the optimal game optimization quality objective as the goal to obtain game parameter configuration; weight coefficients of the game optimization quality objectives are configured, and interest balance is performed on the game parameter configuration based on the weight coefficients using an interest balance algorithm to obtain the optimal configuration.

[0046] Optionally, when the optimized quality objective is no longer singular but comprises multiple optimized quality objectives, it is first necessary to determine the direction of influence of each optimized parameter on each optimized quality objective. This can be achieved by analyzing the correlation, sensitivity, or causal relationship between the optimized parameters and different quality objectives. For example, based on multiple optimized parameters and multiple optimized quality objectives, a mathematical model can be established to perform sensitivity analysis on each optimized parameter. By changing the value of one optimized parameter in the model and observing its impact on the objective, the influence of each optimized parameter on each optimized quality objective can be understood. This can be accomplished. Based on the results of the sensitivity analysis, the direction of influence of each optimized parameter on each optimized quality objective is determined. The direction of influence can be positive (i.e., an increase in the parameter leads to an increase in the target value) or negative (i.e., an increase in the parameter leads to a decrease in the target value).

[0047] Furthermore, based on the direction of influence of the optimization parameters on each optimization quality objective, game-theoretic optimization quality objectives are extracted, i.e., optimization quality objectives with conflicting or competitive relationships, where optimizing one objective may lead to a decrease in the performance of another objective. For each game-theoretic optimization quality objective, optimization parameters are optimized separately. By searching different parameter configurations within the optimization space and using an evaluation function to assess the impact of each configuration on each objective, through multiple iterations and trade-offs, the parameter configuration that achieves the optimal state for each objective is found, thus obtaining the game-theoretic parameter configuration.

[0048] Furthermore, after obtaining the optimal parameter configurations corresponding to multiple game-theoretic optimization quality objectives, a benefit-balancing algorithm is needed to balance these configurations to obtain the final optimized configuration. This can be achieved by configuring weight coefficients for each objective to reflect its relative importance and priority in the overall optimization. Based on these weight coefficients, the game-theoretic parameter configurations are balanced—that is, weight adjustments are made—to balance the conflicts and competition between different objectives, ultimately resulting in a parameter configuration that achieves good performance across multiple objectives. The final optimized configuration will serve as guiding parameters for process control, used to adjust and optimize the etching process in actual production. This ensures good performance across multiple key quality objectives, thereby improving the overall quality and competitiveness of the product.

[0049] Furthermore, embodiments of this application also include the following steps: Based on the deviation between the game parameter configuration and the optimized configuration, a compensation tracking control module is constructed; process tracking is performed for the optimized quality target to obtain process evaluation information; based on the process evaluation information, the position of the deviation wafer is located, and parameter compensation control is performed using the compensation tracking control module based on the position of the deviation wafer.

[0050] Specifically, after determining the optimized configuration of the optimization parameters, this application embodiment also includes a compensation tracking control and parameter compensation control method to ensure that the etching process can operate stably in the optimized state during actual production.

[0051] Optionally, the deviation between the game parameter configuration and the optimized configuration is first calculated, and a compensation tracking control module is constructed based on the deviation. The main function of this module is to monitor the operating status of the etching process in real time, and automatically calculate and output the corresponding compensation control signal according to the deviation between the actual operating parameters and the optimized configuration.

[0052] Furthermore, during actual production, the etching process is continuously monitored. This includes collecting various data related to the etching process, such as temperature, pressure, and flow rate, as well as the final etching quality data. Based on the collected data, the etching process is evaluated. For example, by comparing it with preset optimized quality targets, actual process performance indicators, such as etching depth and uniformity, are calculated to obtain process evaluation information.

[0053] Furthermore, based on the process evaluation information, the locations of the deviating wafers are identified. That is, during the process evaluation, if the process performance of certain wafers (or batches) deviates from expectations—meaning their process performance is below the optimized quality target—then the locations of these deviating wafers need to be identified. Based on the identified locations of the deviating wafers, the compensation tracking control module performs parameter compensation control. Specifically, the compensation tracking control module automatically calculates the parameter values ​​that need adjustment based on the deviation between the actual operating parameters and the optimized configuration, and sends these parameter values ​​to the etching equipment through the control system to achieve parameter compensation control of the etching process. This ensures that the etching process can operate stably in an optimized state during actual production, thereby improving product quality and consistency.

[0054] In summary, the embodiments of this application have at least the following technical effects: This application collects etching parameter data, extracts abnormal quality parameters, determines optimization quality targets, analyzes the influence relationship between parameter data and etching quality assessment information, obtains the abnormal parameter ratio of each parameter data, screens optimization parameters, constructs an optimization space based on optimization quality targets and influence relationships, determines the optimal configuration of optimization parameters, and performs process control.

[0055] This technology achieves the goal of improving the processing accuracy of semiconductor wafer etching by analyzing and optimizing key parameters of the etching process, thereby improving product quality.

[0056] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0057] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0058] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application intends to include such modifications and variations.

Claims

1. A method of parametric niche regulation in semiconductor wafer manufacturing, characterized by, The parameter niche regulation method in the semiconductor wafer manufacturing comprises: Collecting etching parameter data, constructing parameter data set, parameter data has equipment batch label; Obtaining etching quality evaluation information corresponding to each parameter data, extracting abnormal quality parameter of etching quality evaluation information, and determining optimization quality target; The parameter data set is normalized, the influence relationship between the parameter data and the etching quality evaluation information is analyzed based on the abnormal quality parameter, the parameter data is aligned as the target, the same parameter data item is superimposed based on the influence coefficient of the influence relationship, and the abnormal parameter ratio of each parameter data is obtained; According to the optimization parameter ratio, the optimization quality target, the optimization parameter, and the influence relationship, the optimization space is constructed, the optimization is carried out with the maximum optimization quality target, the optimization configuration of the optimization parameter is determined, and the process control is carried out based on the optimization configuration.

2. The method of claim 1, wherein the parameter niche is a parameter of a process used in the manufacturing of the semiconductor wafer. The collecting etching parameter data, constructing parameter data set comprises: Obtaining the distribution quantity of etching equipment, constructing parallel collection channel, and the parallel collection channel is used for respectively establishing material connection data communication with etching equipment; Respectively obtaining the change rate of etching rate, selection ratio, uniformity, temperature and gas flow, and setting collection frequency; Based on the collection frequency setting data collection equipment, and setting transmission frequency of the parallel collection channel by using the collection frequency, the parameter data set is obtained.

3. The method of claim 2, wherein the parameter niche is a parameter of a process used in the manufacturing of the semiconductor wafer. The respective acquisition of the etching rate, selection ratio, uniformity, temperature and gas flow change rate, and the setting of the collection frequency, comprises: Setting standard experiment time, respectively collecting the etching rate, selection ratio, uniformity, temperature and gas flow, and obtaining standard time data set; According to the standard time data set, respectively draw each parameter change curve, and obtain the change rate of each parameter; Screening the maximum change rate, and setting the double of the maximum change rate as the collection frequency based on Nyquist sampling theorem.

4. The method of claim 1, wherein the parameter niche is a parameter of a semiconductor wafer manufacturing process. The same parameter data item is superimposed based on the influence coefficient of the influence relationship, and the abnormal parameter ratio of each parameter data is obtained, comprising: According to the aligned parameter data, setting parameter matrix, horizontal column is a complete abnormal quality parameter item, and vertical column is each parameter data; Taking the influence coefficient as the weighting coefficient, the parameter matrix is weighted to obtain the weighted parameter matrix, and the weighted parameter matrix is the data after weighting of the vertical column parameter; Based on the weighted parameter matrix, the mean vector and the standard deviation vector of each parameter data point are calculated; According to the mean vector and the standard deviation vector, the abnormal parameter ratio of the weighted parameter matrix is calculated, and the abnormal parameter ratio of each parameter data is obtained.

5. The method of claim 4, wherein the parameter niche is a parameter of a process used in the manufacture of the semiconductor wafer. The abnormal parameter ratio of each parameter data is obtained, comprising: wherein , , is the abnormal parameter ratio of the i-th parameter data, is the mean vector, is the standard deviation vector, is the weighted parameter matrix, is the i-th parameter data in the weighted parameter matrix, and n is the number of parameter data.

6. The method of claim 1, wherein the parameter niche regulation is performed in a semiconductor wafer manufacturing process. According to the optimization quality target, the optimization parameter, and the influence relationship, the optimization space is constructed, the optimization is carried out with the maximum optimization quality target, comprising: When the optimization quality target is single, the evaluation function is fitted according to the optimization quality target, the optimization parameter and the influence relationship; Obtaining the constraint information of each parameter data, constructing the optimization space based on the constraint information and the evaluation function, and configuring the optimization rule; The parameter optimization is performed in the optimization space, including: constructing an initial solution, the initial solution being a historical optimal solution, setting an optimization direction and an optimization step based on the initial solution, obtaining an updated solution, evaluating the updated solution by using the evaluation function, replacing the initial solution when the evaluation of the updated solution is higher than that of the initial solution, and iterating until an optimization target is reached or the number of iterations reaches a preset value, to obtain an optimized configuration of the optimization parameter.

7. The method of claim 6, wherein the parameter niche is a parameter of a process used in the manufacture of the semiconductor wafer. The optimization space is constructed according to the optimization quality target, the optimization parameter and the influence relationship, the optimization is performed with the maximum optimization quality target, and the optimized configuration of the optimization parameter is determined, including: When the optimization quality target is multiple, the influence direction of the optimization parameter on the optimization quality target is obtained according to the optimization quality target; Based on the influence direction of the optimization parameter on each optimization quality target, the game optimization quality target is extracted; The optimization parameter is optimized respectively with the best game optimization quality target as the target, to obtain a game parameter configuration; The weight coefficient of the game optimization quality target is configured, and the game parameter configuration is balanced by using a benefit balance algorithm based on the weight coefficient, to obtain the optimized configuration.

8. The method of claim 7, wherein the parameter niche is a parameter of a process used in the manufacturing of the semiconductor wafer. The process control based on the optimized configuration further includes: Based on the deviation value of the game parameter configuration and the optimized configuration, a compensation tracking control module is constructed; Process tracking is performed for the optimization quality target, to obtain process evaluation information; Based on the process evaluation information, a deviation wafer position is located, and parameter compensation control is performed based on the deviation wafer position by using the compensation tracking control module.

Images