Wafer processing process machine abnormality explainability method and related device

Abstract

The application provides a wafer processing process machine abnormality explainability method and related equipment, the method comprises the following steps: acquiring multi-dimensional process parameter data in a wafer processing process, and constructing an analysis data set; sequentially removing each parameter subset in the analysis data set to obtain a first information entropy difference value set; sorting the first information entropy difference value set to extract the first K process parameters; generating a plurality of parameter combination items based on the K process parameters; sequentially removing each parameter combination item in the analysis data set to generate a second information entropy difference value set; merging and sorting the second information entropy difference value set and the first information entropy difference value set, and extracting the process parameters or parameter combination items corresponding to the first N difference values. The application quantitatively analyzes the information entropy difference value, automatically identifies the key parameters or parameter combinations that have the greatest impact on the stability of the machine from the massive multi-dimensional process parameters, and realizes the quantification, automation and explainability of the abnormality evaluation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of semiconductor technology, in particular to a wafer processing process machine abnormality explainability method and related equipment. BACKGROUND

[0002] In the field of semiconductor manufacturing, wafer processing technology is highly complex, usually involving multiple equipment, multiple process steps and a large number of process parameters. In order to improve yield and product consistency, existing manufacturing systems are generally equipped with fault detection and classification systems (FDC) for real-time monitoring of machine status and process stability.

[0003] The existing FDC system mainly realizes the detection and classification of abnormal signals by collecting and recording a large amount of process parameter data. When the wafer has quality abnormalities during the processing process, engineers often need to rely on the FDC system to analyze the relevant parameters to locate the root cause of the abnormality.

[0004] However, the existing method still has the following problems in practical application:

[0005] 1. The amount of data generated during the wafer processing process is extremely large, containing hundreds or even thousands of parameter dimensions, which is difficult for engineers to check one by one;

[0006] 2. The cause of the abnormality is often a deep parameter disturbance with a nonlinear coupling relationship, and it is difficult to accurately trace the cause by a single parameter;

[0007] 3. Multiple factors may interact, and the traditional analysis method is difficult to systematically identify the key parameters that have the greatest impact on the stability of the machine.

[0008] Due to the lack of quantitative modeling of abnormal causes and parameter influence degree evaluation mechanism, the auxiliary decision-making ability of the existing FDC system has certain limitations, and engineers still highly rely on experience when analyzing problems, and the analysis efficiency and accuracy need to be improved. Therefore, an explainable technology means combining big data processing and information theory method is needed to realize the quantitative analysis of the machine abnormality, help engineers quickly locate the key parameters causing quality problems, and improve the efficiency and accuracy of abnormal analysis. SUMMARY

[0009] In view of the deficiencies of the prior art, the present application provides a wafer processing process machine abnormality explainability method and related equipment to at least solve the problem that the prior art lacks quantitative identification and explainability analysis means of key parameters affecting the machine abnormality, and engineers cannot efficiently locate the abnormal root cause.

[0010] To achieve the above object and other advantages, the present application adopts the following technical solutions:

[0011] In a first aspect, the present application provides a wafer processing process machine abnormality explainability method, comprising:

[0012] Obtaining multi-dimensional process parameter data in a wafer processing process, constructing a structured analysis data set, the analysis data set comprising a plurality of parameter subsets, each parameter subset corresponding to a set of parameter values collected by the same sensor at multiple time points;

[0013] Sequentially excluding each parameter subset in the analysis data set to obtain a first information entropy difference value set;

[0014] Sorting the first information entropy difference value set, extracting the top M process parameters according to the sorting result, and generating a plurality of parameter combination items based on the M process parameters;

[0015] Sequentially excluding each parameter combination item in the analysis data set to obtain a second information entropy difference value set;

[0016] Merging and sorting the second information entropy difference value set and the first information entropy difference value set, extracting the process parameters or parameter combination items corresponding to the top N difference values according to the sorting result, to represent the key process factors that have the greatest impact on the stability of the machine in the wafer processing process.

[0017] According to the wafer processing process machine abnormality explainability method provided by the present application, the step of sequentially excluding each parameter subset in the analysis data set to obtain a first information entropy difference value set comprises:

[0018] Calculating the original information entropy value of the analysis data set;

[0019] Sequentially excluding the parameter values corresponding to each parameter subset from the analysis data set, and recalculating the first information entropy value based on the remaining data set after exclusion;

[0020] Differencing the first information entropy value from the original information entropy value to obtain the first information entropy difference value corresponding to each parameter subset;

[0021] The first information entropy difference values corresponding to all parameter subsets form the first information entropy difference value set.

[0022] According to the wafer processing process machine abnormality explainability method provided by the present application, before the step of sequentially excluding each parameter subset in the analysis data set to obtain a first information entropy difference value set, the analysis data set is further normalized to eliminate the dimensional differences between different process parameters.

[0023] According to the wafer processing process machine abnormality explainability method provided in the application, the step of generating a plurality of parameter combination items based on the M process parameters comprises:

[0024] The M process parameters are combined to generate all non-empty subsets with a length less than M as parameter combination items, wherein each parameter combination item includes one or more process parameters.

[0025] According to the wafer processing process machine abnormality explainability method provided in the application, the step of sequentially removing each parameter combination item from the analysis data set to generate a second information entropy difference value set comprises:

[0026] Calculate the original information entropy value of the analysis data set;

[0027] Sequentially remove the parameter values corresponding to each parameter combination item from the analysis data set, and recalculate the second information entropy value based on the remaining data set after removal;

[0028] Subtract the second information entropy value from the original information entropy value to obtain the second information entropy difference value corresponding to each parameter combination item;

[0029] The second information entropy difference values corresponding to all parameter combination items form the second information entropy difference value set.

[0030] According to the wafer processing process machine abnormality explainability method provided in the application, the step of calculating the original information entropy value of the analysis data set comprises:

[0031] Based on the multi-dimensional process parameter data obtained in the wafer processing process, a full-dimensional high-dimensional data matrix is constructed, each row of the full-dimensional high-dimensional data matrix represents a sample point, each sample point corresponds to a group of parameter values collected by different sensors at the same time point, and has a plurality of dimensional characteristic information;

[0032] The full-dimensional high-dimensional data matrix is normalized and preprocessed to eliminate the dimensional differences between parameters;

[0033] Traverse the full-dimensional high-dimensional data matrix to determine K nearest neighbors between each sample point and the remaining sample points;

[0034] Based on the distribution of the K nearest neighbors, the information entropy value of the full-dimensional high-dimensional data matrix is calculated by using a K nearest neighbor estimation method, and is taken as the original information entropy value.

[0035] According to a wafer processing process machine abnormality explainability method provided in the present application, K and N are respectively positive integers, K is used to adjust the granularity of the process parameter preliminary screening, that is, the number of process parameters participating in the parameter combination analysis; N is used to control the depth of the final output parameter set, that is, the number of key parameters or parameter combinations used to represent the machine running stability influence factor.

[0036] In a second aspect, the present application provides an electronic device, which comprises:

[0037] one or more processors; and a memory storing computer program instructions which, when executed, cause the processor to perform the wafer processing process machine abnormality explainability method as described in any of the above.

[0038] In a third aspect, the present application provides a computer readable storage medium, which stores computer programs and / or instructions, and the computer programs and / or instructions are executed by a processor to implement the wafer processing process machine abnormality explainability method as described in any of the above.

[0039] In a fourth aspect, the present application provides a computer program product, which comprises computer programs and / or instructions, and the computer programs / instructions are executed by a processor to implement the wafer processing process machine abnormality explainability method as described in any of the above.

[0040] A wafer processing process machine abnormality explainability method and related equipment are provided. Multi-dimensional process parameter data in a wafer processing process is obtained, and a structured analysis data set is constructed. The analysis data set includes a plurality of parameter subsets, and each parameter subset corresponds to a parameter value set collected by the same sensor at a plurality of time points. Each parameter subset is sequentially removed in the analysis data set to obtain a first information entropy difference value set. The first information entropy difference value set is sorted, and the top M process parameters are extracted according to the sorting result. Based on the M process parameters, a plurality of parameter combination items are generated. Each parameter combination item is sequentially removed in the analysis data set to generate a second information entropy difference value set. The second information entropy difference value set and the first information entropy difference value set are merged and sorted, and the process parameters or parameter combination items corresponding to the top N difference values are extracted according to the sorting result to represent the key process factors that have the greatest impact on the stability of the machine in the wafer processing process. Based on the quantitative analysis method of the information entropy difference value, the key parameters or parameter combinations that have the greatest impact on the stability of the machine are automatically identified from the massive multi-dimensional process parameters, and the quantification, automation and explainability of the abnormality evaluation are realized. Compared with the way of relying on manual experience to backtrack parameters in the traditional FDC system, the method can automatically calculate the importance ranking of the process parameters, and further mine the multi-factor interaction by parameter combination removal, effectively solving the problems of complex abnormal causes and unexplainability. Engineers can directly locate and trace the cause according to the output parameter list, greatly reducing the manual analysis time, enhancing the explainability and objectivity of the abnormality evaluation, and improving the overall process quality management level of the wafer processing. The method is suitable for different types of machines, has good universality and expandability, and has significant engineering application value. BRIEF DESCRIPTION OF DRAWINGS

[0041] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other embodiments can be obtained by those skilled in the art without creating laborious work.

[0042] Figure 1 is a flowchart of a wafer processing process machine abnormality explainability method provided by the embodiments of the present application;

[0043] Figure 2 is a legend of part of the desensitized process parameter data provided by the embodiments of the present application;

[0044] Figure 3 is a legend of part of the data matrix for constructing K nearest neighbor samples provided by the embodiments of the present application;

[0045] Figure 4is a user interface schematic diagram of operation data screening and task setting provided by an embodiment of the present application;

[0046] Figure 5 is an interface schematic diagram of important parameter analysis results provided by an embodiment of the present application;

[0047] Figure 6 is a visual interface schematic diagram of important parameter analysis results provided by an embodiment of the present application;

[0048] Figure 7 is a structural schematic diagram of an electronic device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0049] The above description is only a summary of the technical solutions of the present application. In order to enable the technical means of the present application to be more clearly understood, and to be implemented according to the content of the description, and in order to enable the above and other purposes, characteristics and advantages of the present application to be more apparent and easy to understand, the following preferred embodiments are described in detail below, and the accompanying drawings are described as follows.

[0050] It should be noted that those of ordinary skill in the art explicitly and implicitly understand that the embodiments described in the present application can be combined with other embodiments without conflict. Unless otherwise defined, the technical terms or scientific terms involved in the present application should be understood as the usual meaning by those skilled in the art in the technical field to which the present application belongs. The terms "one", "a", "an", "the", and similar words involved in the present application do not represent quantity limitation, but can represent singular or plural. The terms "include", "contain", "have", and any variations thereof involved in the present application are intended to cover non-exclusive inclusion; the terms "first", "second", "third", and the like involved in the present application are only to distinguish similar objects, and do not represent a specific order of the objects.

[0051] The Fault Detection and Classification (FDC) system is an important system for real-time monitoring, detecting and classifying process faults in semiconductor manufacturing processes. The FDC system collects process parameters, equipment states and environmental data in real time through various sensors installed on production equipment, thereby continuously monitoring the production process.

[0052] Information Entropy is a core statistical quantity in information theory, which is used to quantify the degree of uncertainty of probability distribution. In the wafer processing process of semiconductors, various sensors on the machine continuously collect a large amount of process-related parameter data. For example, temperature, pressure, gas flow, current, voltage, etc. The amount of data generated by each wafer after processing is extremely large, forming a typical high-dimensional and diverse process data set.

[0053] Under the condition of stable running state of the machine and well-controlled process parameters, the parameter data shows strong regularity, and the distribution structure is relatively concentrated, so the corresponding information entropy value is low, indicating small uncertainty. On the contrary, when the machine runs abnormally or there is potential deviation (such as temperature mutation, air pressure fluctuation, etc.), the volatility of the process parameters is enhanced, and the data distribution tends to be discrete or irregular, thereby causing the information entropy value to increase significantly.

[0054] However, in the actual wafer manufacturing site, due to the lack of quantitative modeling ability of the cause of the abnormality and the systematic evaluation mechanism of the influence degree of the process parameters in the existing FDC system, the auxiliary decision function still has certain limitations. When facing abnormal conditions, engineers often need to rely on personal experience to screen parameters and judge trends, and the analysis process is highly subjective, low in efficiency and unstable in accuracy, which is difficult to adapt to the needs of large-scale and high-complexity production environment.

[0055] Therefore, the present application provides a wafer processing process machine abnormality explainability method and related equipment, which combines big data processing capability and explainability analysis technology of information theory model, uses information entropy difference to evaluate the influence degree of different parameters or parameter combinations on system stability, and can effectively identify the key process factors causing quality problems. The method can significantly improve the abnormal positioning efficiency and analysis accuracy, provide a clear troubleshooting direction for engineers to quickly locate the problem, and thus improve the production yield and quality control level.

[0056] Referring to Figure 1 The embodiment of the present application provides a wafer processing process machine abnormality explainability method, which comprises:

[0057] Step S1: acquiring multi-dimensional process parameter data in a wafer processing process, constructing a structured analysis data set, and the analysis data set comprising a plurality of parameter subsets, each parameter subset corresponding to a parameter value set collected by a same sensor at a plurality of time points.

[0058] It should be noted that in the wafer processing process, the FDC system is used to monitor the equipment running state and process in real time. The FDC system collects temperature, air pressure, flow, voltage, current, radio frequency power, cavity vacuum degree, reaction time and other process parameters through multiple sensors during processing, and stores the collected process parameter data in a structured manner in the database according to the wafer batch or processing time period. Therefore, the existing data architecture and interface of the FDC system can be directly called, so as to efficiently acquire process parameter records under different processing batches, different time periods, different machines or chambers, and construct the structured data set required for analysis.

[0059] The analysis data set comprises a plurality of parameter subsets, each parameter subset corresponding to a set of values of a process parameter collected by the same sensor at a plurality of time points. For example, the continuous parameter values recorded by a temperature sensor constitute a temperature parameter subset, and the corresponding parameter subset of a pressure sensor is a pressure parameter subset.

[0060] The analysis data set is defined as: wherein, represents a parameter subset, and n is the number of process parameters. That is, the multi-dimensional process parameter data in the wafer processing process is structured and arranged to form a two-dimensional data matrix, the column vectors of the data matrix form parameter subsets, and all parameter subsets are combined to constitute the entire analysis data set D.

[0061] Before calculating the information entropy difference value, the analysis data set needs to be uniformly scaled in value range, and the original sampling value is mapped to the same scale interval, usually [0, 1] or [-1, 1]. In this embodiment, the minimum-maximum normalization method is used to achieve comparability of different dimensional parameters in information entropy calculation and to avoid bias caused by value scale difference. The normalization formula is as follows:

[0062]

[0063] wherein, is the normalized value of the data, X is the original parameter value, and are the minimum value and the maximum value appearing in the analysis data set, respectively.

[0064] After normalization processing, the process parameters with different value sizes have a unified scale, eliminating the dimensional difference between different parameters.

[0065] Step S2: sequentially removing each parameter subset in the analysis data set to obtain a first information entropy difference value set.

[0066] In this embodiment, step S2 specifically comprises:

[0067] Step S201: calculating the original information entropy value of the analysis data set, specifically comprising:

[0068] Step S2011: based on the acquired multi-dimensional process parameter data in the wafer processing process, constructing a full-amount high-dimensional data matrix, each row of the full-amount high-dimensional data matrix representing a sample point, each sample point corresponding to a group of parameter values collected by different sensors at the same time point, and having a plurality of dimensional characteristic information;

[0069] Step S2012: normalizing the full-amount high-dimensional data matrix to eliminate the dimensional difference between parameters;

[0070] Step S2013: traversing the full high-dimensional matrix, determining K nearest neighbor samples between each sample point and the rest of the sample points;

[0071] Step S2014: based on the distribution of the K nearest neighbor samples, using K nearest neighbor estimation method to calculate the information entropy value of the full high-dimensional data matrix as the original information entropy value.

[0072] Specifically, the entire full high-dimensional data matrix can be defined as: , wherein, , represents the sensor data set of the i th wafer or the i th processing cycle, N is the total amount of data samples, and n is the number of process parameters. Each sample point is an n-dimensional vector, represents the j th process parameter value of the i th sample point.

[0073] For example, in a certain processing process, if 589 sensor parameters are collected in the wafer processing process, and the process section involves 1568 wafers, then the final data set contains 1568 data points, each point has 589 dimensions, corresponding to a complete structured process parameter data matrix (matrix size is 1568x589). An example of the desensitized part of the process parameter data is shown in Figure 2 Each row in the table corresponds to a data point, representing the process state of a wafer or a time; each column corresponds to a specific process parameter variable value, such as temperature, voltage, flow, radio frequency power, air pressure, reaction time, etc. Take such structured process data as a starting point, and perform normalization preprocessing, neighbor relationship construction and information entropy estimation analysis in subsequent steps to realize quantitative evaluation of machine running stability.

[0074] The implementation process of the normalization preprocessing of the full high-dimensional data matrix is specifically described in the normalization processing process in the foregoing, which will not be described here.

[0075] The current sample point is taken as a target data point, and the distance between the target data point and all other sample points (j≠i) is calculated in turn. The first K sample points with the smallest distance from the target data point are selected, denoted as the neighbor samples of the target data point.

[0076] The parameter K is a preset constant, which is used to specify K nearest neighbor samples of each sample point. The value of K is selected according to experience or data set characteristics. In actual application, a smaller K value can better reflect the sensitivity of local structure, and is suitable for capturing small abnormal changes; while a larger K value has stronger anti-noise ability and statistical stability, and is suitable for overall trend evaluation. In this embodiment, the value of K is set to 5, and an example of the generated K nearest neighbor sample data matrix is shown inFigure 3 as shown.

[0077] In this embodiment, the distance between data points is calculated in any of the Euclidean distance, Manhattan distance or Chebyshev distance.

[0078] According to the K nearest neighbor samples that have been constructed, the distance values are sorted from small to large. The distance corresponding to the Kth nearest neighbor point in the sorted result is selected as the nearest neighbor distance ε of the target data point, which is used to represent the local density of the target data point in the high-dimensional space. The nearest neighbor distance is not only the local boundary of the target data point, but also a key variable for constructing a high-dimensional hypersphere and estimating density, which directly affects the accuracy and stability of the entropy value.

[0079] After obtaining the nearest neighbor distance ε, a high-dimensional hypersphere is constructed in the d-dimensional data space with the target data point as the center. The high-dimensional hypersphere represents the local neighborhood range containing the K nearest neighbor sample data points with the target data point as the center.

[0080] The volume of the high-dimensional hypersphere is calculated by the following formula:

[0081]

[0082] wherein, is the volume of the high-dimensional hypersphere with a unit radius, d is the data space dimension, is the Gamma function.

[0083] The final information entropy estimation model is shown as follows:

[0084]

[0085] wherein, is the information entropy of the full high-dimensional data set, N is the total number of data samples, d is the data space dimension, is the distance between the ith data point and the Kth nearest neighbor data point, is the value of the Digamma function at the total number of samples N, is the value of the Digamma function at the number of neighbors k, is the deviation correction term between the number of samples and the number of neighbors, is a constant term for correcting the estimation deviation, is the volume of the high-dimensional hypersphere with a unit radius.

[0086] The information entropy value is used to measure the overall distribution uncertainty of the data set. The larger the value, the more dispersed and unstable the data is; the smaller the value, the more concentrated the data distribution is, and the more stable the machine operation is.

[0087] The information entropy estimation model uses the spatial neighborhood structure of sample data points to measure the overall uncertainty of high-dimensional data sets by means of a non-distribution assumption. Since it does not depend on labels or yield labels, it is suitable for online machine stability detection. Moreover, it does not require data to conform to specific distributions such as Gaussian, and has good versatility, suitable for multiple types of wafer processing equipment. The information entropy estimation model can be deployed in existing FDC systems as an anomaly detection module for quantitative evaluation of machine stability, and can assist in early warning, root cause analysis, and yield prediction

[0088] Step S202: sequentially remove the parameter values corresponding to each parameter subset from the analysis data set, and recompute the first information entropy value based on the remaining data set after removal;

[0089] Step S203: difference the first information entropy value from the original information entropy value to obtain the first information entropy difference value corresponding to each parameter subset;

[0090] Step S204: form a first information entropy difference value set by combining the first information entropy difference values corresponding to all parameter subsets.

[0091] Specifically, the information entropy estimation model is used to calculate the original information entropy value of the analysis data set D, denoted as H(D). Without changing other parameters, a parameter subset is sequentially removed from the original analysis data set D (i.e., the corresponding column vector is deleted) to obtain a new reduced data set. The information entropy is recalculated for the data set after removing the parameter subset to obtain the first information entropy value under the removal operation .

[0092] The first information entropy value corresponding to each removed data set is subtracted from the original information entropy value in step S201 to obtain the first information entropy difference value , which is calculated as follows:

[0093]

[0094] The difference value reflects the influence of the parameter subset on the overall entropy value, i.e., its sensitivity to machine abnormality. The larger the difference value, the greater the disturbance of the parameter fluctuation on the system stability.

[0095] Therefore, an information entropy difference value set is obtained.

[0096] Next, a positive integer parameter M is set to represent the number of important parameters to be extracted. This parameter is set by the user according to the analysis requirements, system resources, or process complexity, and is used to adjust the granularity of process parameter preliminary screening, i.e., the number of process parameters involved in parameter combination analysis. Subsequently, the information entropy difference values The parameter subsets corresponding to the first M information entropy difference values are extracted according to the sorting result, and a parameter set is obtained: .

[0097] By setting the parameter M and sorting and screening the information entropy difference value set, a high-quality input set is provided for the parameter combination analysis in the next stage, so that the synergistic influence relationship between multiple parameters can be more effectively mined.

[0098] Step S3: The first information entropy difference value set is sorted, the first M process parameters are extracted according to the sorting result, and a plurality of parameter combination items are generated based on the M process parameters.

[0099] In this embodiment, step S3 specifically includes: performing combination processing on the M process parameters to generate all non-empty subsets with a length less than M as parameter combination items, wherein each parameter combination item includes one or more process parameters.

[0100] The M process parameters are arranged and combined to generate all possible parameter combination items, and the length of each combination is required to be less than M, that is, all non-empty subsets with a length of 1 to are included in the analysis range. In this way, the combination strategy can effectively avoid the calculation complexity caused by high-dimensional combination, while retaining sufficient multi-factor interaction potential.

[0101] For example, when M=5, the process parameter set participating in the combination is , and all non-empty subsets with a length less than 5 include:

[0102] Binary combination:

[0103]

[0104] Ternary combination:

[0105]

[0106] Quaternary combination:

[0107]

[0108] Step S4: A second information entropy difference value set is generated by sequentially eliminating each parameter combination item in the analysis data set.

[0109] In this embodiment, step S4 specifically includes:

[0110] Step S401: Calculate the original information entropy value of the analysis data set;

[0111] Step S402: sequentially remove the parameter value corresponding to each parameter combination item from the analysis data set, and re-calculate the second information entropy value based on the remaining data set after removal;

[0112] Step S403: difference the second information entropy value from the original information entropy value to obtain the second information entropy difference value corresponding to each parameter combination item;

[0113] Step S404: form a second information entropy difference value set by the second information entropy difference values corresponding to all parameter combination items.

[0114] Specifically, the calculation method of the original information entropy value of the analysis data set in step S401 can refer to the aforementioned step S201, which will not be described here. The original information entropy value of the analysis data set D is calculated using the information entropy estimation model constructed as .

[0115] For example, assuming that a parameter combination item is , the first column and the second column are removed from the original analysis data set, and only the data corresponding to other parameters are retained. Based on the data set after removal, the information entropy value of the remaining data set is recalculated and denoted as the second information entropy value .

[0116] Difference the second information entropy value from the original information entropy value in step S401 to obtain the second information entropy difference value , and the calculation method is as follows:

[0117]

[0118] Each parameter combination item to be analyzed is sequentially traversed, and for each combination, all parameter subsets involved in the combination are removed from the original analysis data set (i.e., the corresponding column vector combination in the data matrix is deleted), and the information entropy value of the remaining data set is calculated. Difference the information entropy value obtained after removing each parameter combination item from the original information entropy value obtained in step S401, and the difference values are summarized to form a second information entropy difference value set.

[0119] By generating and analyzing multiple parameter combination items, complex abnormal patterns caused by non-single parameters can be mined, which is especially suitable for capturing the influence law of multi-parameter coordinated fluctuation on system stability, especially for those scenes that are not sensitive to single parameter changes but significantly fluctuate in combination, thereby further improving the depth and accuracy of abnormal root cause analysis.

[0120] Step S5: merge and sort the second information entropy difference value set and the first information entropy difference value set, extract the process parameters or parameter combination items corresponding to the top N difference values according to the sorting result, and represent the key process factors that have the greatest impact on the stability of the machine running in the wafer processing process.

[0121] Specifically, the first set of information entropy difference values formed in step S2 and the second set of information entropy difference values formed in step S4 are merged to form a unified set of entropy difference values, each of which represents the degree of change in system information entropy caused by the elimination of a certain parameter or parameter combination. Then, the merged set of entropy difference values is sorted in descending order according to the difference values, obtaining an ordered parameter list. The larger the difference value, the more significant the impact of the corresponding parameter or parameter combination on the stability of the machine.

[0122] A parameter N can be set to represent the number of key factors expected to be extracted. According to the sorting result, the system selects the parameter items (which can be single parameters or parameter combinations) corresponding to the top N information entropy difference values as the final set of abnormality influencing factors, denoted as:

[0123]

[0124] where each represents a parameter or a group of parameters as the most significant process factor affecting the abnormality degree change in the wafer processing process.

[0125] N is used to control the depth of the final output parameter set, i.e., the number of key parameters or parameter combinations representing the machine stability influencing factors. When N = 3 is set, the top 3 parameters or parameter combinations are taken out, for example , which are the most important key process factors affecting the wafer processing.

[0126] This method is integrated into the existing FDC system framework and the information entropy calculation module for evaluating the machine abnormality degree is optimized and upgraded. By analyzing the information entropy difference values of the normalized multi-dimensional process parameters, this method can automatically filter out the key parameters and their combinations affecting the abnormality degree change, forming a quantitative, visual, and interpretable analysis process to help engineers quickly locate the abnormality causes.

[0127] Taking the PVD (Physical Vapor Deposition) process as an example, the wafer processing process machine abnormality degree interpretability method provided in this embodiment can be integrated into the FDC system to support engineers to complete abnormality degree analysis configuration and parameter filtering through a graphical interface. As shown in Figure 4 , the user can select the process data range to be analyzed in the condition configuration area on the left side of the interface, such as setting the time range (setting the start time and end time of the analysis period), equipment and chamber (selecting the target equipment EQP number and its corresponding chamber Chamber), wafer number and batch information (inputting specific Lot ID or Wafer number for accurate positioning), and process condition filtering (optional parameters include Recipe, Operation, Step, Product, etc.).

[0128] After configuring the conditions, click the "Analyze" button, and the system will load all Run batch records under the filtering conditions on the right side of the interface, and support users to select the target Run for analysis through the single selection box. During the analysis process, the system performs normalization processing, information entropy difference calculation, parameter importance sorting and combination analysis according to the set parameters, and finally outputs the key process parameters or parameter combinations that most significantly affect the abnormality degree of the current Run. The user interface supports exporting the results as Excel, PPT or PDF format files, which is convenient for archiving analysis and conference reports.

[0129] Users can set the number of key parameters to be output, for example, set to extract the top N most important parameters, and after clicking the analysis button, the system calls the analysis module in the background, and the results are displayed on the front-end page, as shown in the important parameter analysis result interface diagram Figure 5 . The interface shows the key process parameters identified in the wafer processing process tool abnormality analysis and their impact degree sorting. The interface provides a "Draw Chart" chart generation button for drawing the corresponding visualization bar chart in the right area. Click the "Draw Chart" button to display the column chart, as shown in Figure 6 . Each column corresponds to a parameter, and its height represents the contribution proportion of the parameter to the system abnormality degree change. The figure shows that "sv3" is the most critical parameter (contribution proportion: 24%), followed by "sv1" (22%), "sv4" (21%), "sv2" (20%), and "RecipeStep_ID" (12%).

[0130] To sum up, the wafer processing process machine abnormality explainability method provided by the application comprises the following steps: acquiring multi-dimensional process parameter data in a wafer processing process, constructing a structured analysis data set, the analysis data set comprising a plurality of parameter subsets, each parameter subset corresponding to a parameter value set collected by the same sensor at a plurality of time points; sequentially removing each parameter subset in the analysis data set to obtain a first information entropy difference value set; sorting the first information entropy difference value set, extracting the first M process parameters according to the sorting result, and generating a plurality of parameter combination items based on the M process parameters; sequentially removing each parameter combination item in the analysis data set to obtain a second information entropy difference value set; merging and sorting the second information entropy difference value set and the first information entropy difference value set, and extracting the process parameters or parameter combination items corresponding to the first N difference values according to the sorting result to represent the key process factors that have the greatest impact on the machine running stability in the wafer processing process. The quantitative analysis method based on the information entropy difference value can automatically identify the key parameters or parameter combinations that have the greatest impact on the machine stability from the massive multi-dimensional process parameters, and realize the quantification, automation and explainability of the abnormality evaluation. Compared with the parameter backtracking method relying on manual experience in the traditional FDC system, the method can automatically calculate the importance ranking of the process parameters, and further mine the multi-factor interaction by parameter combination removal, effectively solving the problems of complex abnormal causes and unexplainability. Engineers can directly locate and trace the causes according to the output parameter list, greatly reducing the manual analysis time, enhancing the explainability and objectivity of the abnormality evaluation, and improving the overall process quality management level of the wafer processing. The method is suitable for different types of machines, has good universality and expandability, and has significant engineering application value.

[0131] Those skilled in the art can understand that, in the above method of the specific implementation, the writing order of each step does not mean a strict execution order and does not constitute any limitation on the implementation process. The specific execution order of each step should be determined by its function and possible internal logic.

[0132] In addition, some embodiments of the application also provide an electronic device. The electronic device can be various forms of digital computers, such as laptops, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, etc. The electronic device can also be various forms of mobile devices, such as personal digital assistants, cellular phones, smart phones, wearable devices, and other similar computing devices.

[0133] The electronic device comprises one or more processors, and a memory storing computer program instructions, which, when executed, cause the processor to perform a wafer processing process machine abnormality explainability method provided by any one or more of the above embodiments. Figure 7An exemplary structural diagram of the electronic device is disclosed. As shown in Figure 7 The electronic device includes one or more processors 1101, a memory 1102, and an interface for connecting the components, including a high-speed interface and a low-speed interface. The components are interconnected using different buses, and can be mounted on a common motherboard or otherwise, as desired. The processor can process instructions for execution within the electronic device, including instructions stored in the memory or on the memory to display graphical information for a GUI on an external input / output device, such as a display device coupled to the interface. In some other embodiments, multiple processors and / or multiple buses can be employed as desired, along with multiple memories and types of memory. Also, multiple electronic devices can be connected, with each device providing portions of the necessary operations (e.g., as a server array, a group of blade servers, or a multi-processor system). In all cases, the components shown herein can be implemented with various computer hardware and software layers, with the components relating to the present application being implemented in the higher software layers in some embodiments. The components, their connections and relationships, and their functions, as described herein, are meant to be examples only, and are not intended to limit the present application as described and / or claimed herein.

[0134] The electronic device can also include input devices 1103 and output devices 1104. The processor 1101, the memory 1102, the input devices 1103, and the output devices 1104 can be connected through a bus or other means, Figure 7 as an example of a connection through a bus.

[0135] The input devices 1103 can receive input digital or character information, and generate key signal input with respect to user settings and function controls of the electronic device, such as touch screens, keypads, mice, trackpads, touchpads, pointing sticks, one or more mouse buttons, trackballs, joysticks, and the like. The output devices 1104 can include display devices, auxiliary lighting devices (e.g., LEDs), and tactile feedback devices (e.g., vibration motors), among others. The display devices can include, but are not limited to, liquid crystal displays (LCDs), light emitting diode (LED) displays, and plasma displays. In some embodiments, the display devices can be touch screens.

[0136] To provide for interaction with a user, the electronic device can be a computer. The computer has a display device (e.g., a cathode ray tube (CRT) or liquid crystal display (LCD) monitor) for displaying information to the user and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the computer. Other kinds of devices can be used to provide for interaction with a user as well; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form, including acoustic, speech, or tactile input.

[0137] In the embodiments of the present application, the computer program / instruction is stored on the computer readable medium, and the computer program / instruction is executed by the processor to implement the wafer processing process machine abnormality explainability method provided by any one or more of the above embodiments. The computer readable medium can be included in the electronic device described in the above embodiments, or can exist separately and not be assembled into the device. The computer readable medium carries one or more computer readable instructions.

[0138] The memory 1102 can be used as a non-transitory computer readable storage medium to store non-transitory software programs, non-transitory computer executable programs and modules. The processor 1101 executes various functions and data processing of the server by running the non-transitory software programs, instructions and modules stored in the memory 1102, so as to implement the program instructions / modules corresponding to the method provided by any one or more of the above embodiments in the embodiments of the present application.

[0139] The memory 1102 can include a program storage area and a data storage area. The program storage area can store an operating system, at least one application required by a function, and the data storage area can store data created according to the use of the electronic device. In addition, the memory 1102 can include a high-speed random access memory, and can also include a non-transitory memory, such as at least one magnetic disk storage device, a flash memory device, or other non-transitory solid-state memory device. In some embodiments, the memory 1102 can optionally include a memory disposed remotely with respect to the processor 1101, and these remote memories can be connected to the electronic device through a network. Examples of the above network include but are not limited to the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof.

[0140] Note that more specific examples of the computer-readable storage medium can include but are not limited to an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the present application, the computer-readable storage medium can be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.

[0141] Computer-readable storage media include permanent and non-permanent, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD), or other optical storage, magnetic cassette, magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information accessible to a computing device.

[0142] Computer program code for carrying out operations of the present application can be written in any combination of one or more programming languages, including an object oriented programming language such as Java, Smalltalk, C++ or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code can execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computer (for example, through the Internet using an Internet Service Provider).

[0143] In the above-described embodiments, all or part of the embodiments can be implemented by software, hardware, firmware, or any combination thereof. For example, application specific integrated circuits (ASICs), general purpose computers or any other similar devices can be used. In some embodiments, software programs of the present application can be executed by a processor to implement the above steps or functions. Also, software programs of the present application (including related data structures) can be stored in a computer-readable recording medium, such as a RAM memory, a magnetic or optical drive or diskette, and the like. In addition, some steps or functions of the present application can be implemented by hardware, such as a circuit that cooperates with a processor to perform the respective steps or functions.

[0144] The computer program product provided by the embodiments of the present application includes one or more computer programs / instructions, which, when executed by a processor, generate all or part of the processes or functions described in the embodiments of the present application. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center through wired (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (such as infrared, wireless, microwave, etc.) mode. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. integrated with one or more available media. The available media can be magnetic media (such as floppy disk, hard disk, magnetic tape), optical media (such as DVD), or semiconductor media (such as solid state disk, SSD, solid state disk), etc.

[0145] The flowcharts or block diagrams in the drawings illustrate the architecture, functionality, and operation of possible implementations of devices, methods, and computer program products according to various embodiments of the present application. In this regard, each block in the flowcharts or block diagrams can represent a module, a segment, or a portion of code, which contains one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions noted in the blocks can occur in a different order than that shown in the figures. For example, two blocks noted in succession can actually be executed substantially concurrently, or they can sometimes be executed in reverse order, depending on the functionality involved. It should also be noted that each block in the block diagrams and / or flowcharts, as well as combinations of blocks in the block diagrams and / or flowcharts, can be implemented by dedicated hardware-based systems that perform the specified functions or operations, or can be implemented by a combination of dedicated hardware and computer instructions.

[0146] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any skilled person in the art can easily make changes or replacements within the technical range disclosed in the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be limited by the protection scope of the claims, and the above-mentioned embodiments should be regarded as exemplary and non-limiting.

Claims

1. A method for interpreting the abnormality of equipment in a wafer fabrication process, characterized in that, The application relates to a method for identifying key process factors in a wafer processing process. The method comprises the following steps: acquiring multi-dimensional process parameter data in a wafer processing process, and constructing a structured analysis data set, wherein the analysis data set comprises a plurality of parameter subsets, and each parameter subset corresponds to a set of parameter values collected by the same sensor at a plurality of time points; sequentially removing each parameter subset from the analysis data set to obtain a first information entropy difference value set; sorting the first information entropy difference value set, extracting the first M process parameters according to the sorting result, and generating a plurality of parameter combination items based on the M process parameters; sequentially removing each parameter combination item from the analysis data set to obtain a second information entropy difference value set; 2. The wafer processing tool abnormality explainability method of claim 1, wherein, merging and sorting the second information entropy difference value set and the first information entropy difference value set, and extracting the first N process parameters or parameter combination items corresponding to the first N difference values according to the sorting result, so as to represent the key process factors that have the greatest influence on the running stability of a machine table in the wafer processing process. The step of sequentially removing each parameter subset from the analysis data set to obtain a first information entropy difference value set comprises the following steps: calculating an original information entropy value of the analysis data set; sequentially removing the parameter values corresponding to each parameter subset from the analysis data set, recalculating a first information entropy value based on the remaining data set after removal, and calculating a first information entropy difference value corresponding to each parameter subset by subtracting the first information entropy value from the original information entropy value; combining the first information entropy difference values corresponding to all parameter subsets to obtain the first information entropy difference value set.

3. The wafer processing tool abnormality explainability method according to claim 1 or 2, characterized in that, Before the step of sequentially removing each parameter subset from the analysis data set to obtain a first information entropy difference value set, the analysis data set is subjected to normalization processing to eliminate the dimensional differences between different process parameters.

4. The wafer processing tool abnormality explainability method of claim 1, wherein, The step of generating a plurality of parameter combination items based on the M process parameters comprises the following steps: combining the M process parameters to generate all non-empty subsets with a length less than M as parameter combination items, wherein each parameter combination item comprises one or more process parameters.

5. The wafer processing tool abnormality explainability method of claim 1, wherein, The step of sequentially removing each parameter combination item from the analysis data set to obtain a second information entropy difference value set comprises the following steps: calculating an original information entropy value of the analysis data set; sequentially removing the parameter values corresponding to each parameter combination item from the analysis data set, recalculating a second information entropy value based on the remaining data set after removal, and calculating a second information entropy difference value corresponding to each parameter combination item by subtracting the second information entropy value from the original information entropy value; combining the second information entropy difference values corresponding to all parameter combination items to obtain the second information entropy difference value set. The step of calculating an original information entropy value of the analysis data set comprises the following steps:

6. The wafer processing tool abnormality explainability method according to claim 2 or 5, wherein, based on the acquired multi-dimensional process parameter data in the wafer processing process, constructing a full-quantity high-dimensional data matrix, wherein each row of the full-quantity high-dimensional data matrix represents a sample point, each sample point corresponds to a set of parameter values collected by different sensors at the same time point, and has a plurality of dimensional characteristic information; performing normalization preprocessing on the full-quantity high-dimensional data matrix to eliminate the dimensional differences between parameters; and ​ Traverse the full quantity high-dimensional matrix, determine K nearest neighbor samples between each sample point and the rest of the sample points; Based on the distribution of the K nearest neighbor samples, the information entropy value of the full quantity high-dimensional data matrix is calculated by using K nearest neighbor estimation method, to serve as the original information entropy value.

7. The wafer processing tool abnormality explainability method of claim 1, wherein, M and N are respectively set positive integers, M is used to adjust the particle size of the process parameter preliminary screening, that is, the number of process parameters participating in parameter combination analysis; N is used to control the depth of the final output parameter set, that is, the number of key parameters or parameter combinations used to represent the influence factors of machine running stability.

8. An electronic device, comprising: The electronic device comprises: One or more processors; and memory having stored computer program instructions that, when executed, cause the processors to perform the wafer processing tool abnormality explainability method of any one of claims 1-7.

9. A computer readable storage medium having stored thereon a computer program and / or instructions, characterized in that, The computer program and / or instructions, when executed by a processor, implement the wafer processing tool abnormality explainability method of any one of claims 1-7.

10. A computer program product comprising computer programs and / or instructions, characterized in that, The computer and / or when executed by a processor implement the wafer processing tool abnormality explainability method of any one of claims 1-7.

Images