Wafer processing data interpretable method and device, medium and product

Abstract

The embodiment of the invention relates to the technical field of semiconductor manufacturing processes, and discloses an interpretable method and device for wafer processing data, a medium and a product. Obtaining wafer processing data collected from a machine table by an FDC system to form an input data set; processing the input data set by using an anomaly detection model constructed by deep learning to obtain an output data set; the input data set comprises input features of different combinations, and the output data set is a reconstruction error corresponding to the input features; calculating the marginal influence of each input feature on the output result of the anomaly detection model through a feature importance degree algorithm, and calculating the feature importance degree based on the marginal change of all feature combinations; and providing visual explanation of the key process parameters according to the feature importance. The method can be at least used for solving the analysis problem of deep learning model output.

Description

Technical Field

[0001] The present application relates to the field of semiconductor manufacturing process technology, and in particular to a method, device, medium and product for interpreting wafer processing data. Background Art

[0002] In the semiconductor manufacturing process, wafer processing quality directly impacts the yield and performance of the final product. With the continuous advancement of manufacturing processes, the amount of data generated during semiconductor production has increased dramatically. Traditional anomaly detection methods based on rules or statistical methods are unable to meet the high-precision and high-complexity process requirements. Therefore, modern semiconductor manufacturing widely adopts anomaly detection models based on deep learning to improve the accuracy and efficiency of defect detection.

[0003] Fault Detection and Classification (FDC) systems are common data acquisition and anomaly monitoring systems in semiconductor manufacturing. They collect various process parameters from various machines in real time and use them to analyze potential anomalies during production. Currently, many advanced FDC systems incorporate deep learning models, such as encoder-decoder (autoencoder) architectures, to achieve efficient anomaly detection. These models learn the distribution patterns of normal process parameters and, in actual production, identify large reconstruction errors in input data as possible anomalies.

[0004] However, existing deep learning anomaly detection models often exhibit "black box" characteristics, meaning the model's internal reasoning process is complex and difficult to explain, making it difficult for users to intuitively understand the basis for the model's judgments. Specifically, when an anomaly is detected, engineers struggle to determine which process parameters contribute most to the anomaly, nor can they directly determine which parameters should be adjusted to optimize the process. This lack of explainability limits the application of deep learning models in actual production environments, forcing engineers to rely on experience or additional experiments to analyze and adjust when faced with anomalies, increasing the cost and time of troubleshooting. Summary of the Invention

[0005] One purpose of this application is to provide an interpretable method, device, medium and product for wafer processing data, at least to solve the problem of parsing the output of deep learning models.

[0006] To achieve the above objectives, some embodiments of the present application provide the following aspects: In the first aspect, some embodiments of the present application also provide an interpretable method for wafer processing data, including obtaining wafer processing data collected by an FDC system from a machine to form an input data set; processing the input data set using an anomaly detection model constructed by deep learning to obtain an output data set; the input data set includes input features of different combinations, and the output data set is a reconstruction error corresponding to the input features; through a feature importance algorithm, the marginal impact of each input feature on the output result of the anomaly detection model is calculated, and the feature importance is calculated based on the marginal changes of all feature combinations; based on the feature importance, a visual explanation of key process parameters is provided.

[0007] In a second aspect, some embodiments of the present application further provide an electronic device comprising: one or more processors; and a memory storing computer program instructions, wherein the computer program instructions, when executed, cause the processor to perform the steps of the method described above.

[0008] In a third aspect, some embodiments of the present application further provide a computer-readable medium having computer program instructions stored thereon, wherein the computer program instructions can be executed by a processor to implement the method described above.

[0009] In a fourth aspect, some embodiments of the present application further provide a computer program product, comprising a computer program / instruction, which implements the steps of the above-described method when executed by a processor.

[0010] Compared to related technologies, the solution provided in this embodiment builds on the FDC system and combines it with a feature importance analysis algorithm to quantitatively analyze the input features of the anomaly detection model, thereby calculating the degree of influence of each process parameter on the model output. This method allows users to clearly identify the key process parameters that cause anomalies and make targeted optimization adjustments accordingly, improving troubleshooting efficiency, optimizing production processes, and ultimately improving the stability and yield of the wafer manufacturing process. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] One or more embodiments are exemplarily illustrated by pictures in the corresponding drawings. These exemplifications do not constitute limitations on the embodiments. Elements with the same reference numerals in the drawings are represented as similar elements. Unless otherwise stated, the figures in the drawings do not constitute proportional limitations.

[0012] Figure 1 A schematic diagram of a flow chart of a method for interpreting wafer processing data according to an embodiment of the present application; Figure 2 The figure is a schematic diagram of the structure of an electronic device provided according to an embodiment of the present application. DETAILED DESCRIPTION

[0013] To make the purpose, technical solutions, and advantages of the embodiments of this application more clear, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0014] First embodiment The first embodiment of the present application relates to a method for interpreting wafer processing data. Figure 1 As shown, the method may include the following steps: S101, obtaining wafer processing data collected by the FDC system from the tool to form an input data set; S102, processing the input data set using an anomaly detection model built using deep learning to obtain an output data set; S103, the input data set includes input features of different combinations, and the output data set is the reconstruction error corresponding to the input features; S104, calculating the marginal impact of each input feature on the output result of the anomaly detection model using a feature importance algorithm, and calculating the feature importance based on the marginal changes of all feature combinations; S105: Provide a visual explanation of the key process parameters based on the feature importance.

[0015] The following is a detailed description of the above steps: In step S101, the FDC system monitors the operating status of equipment in real time during the semiconductor manufacturing process. This system continuously collects multi-dimensional process parameters through a sensor network deployed on key process equipment, including lithography machines, etchers, and deposition equipment. These parameters include physical environmental parameters (such as temperature, pressure, and humidity), equipment operating parameters (speed, voltage, and current), and process control parameters (such as lithography exposure dose, etchant gas flow rate, and deposited film thickness). This data is stored in a structured format in CSV files, SQL databases, or HDF5 storage systems. Each record includes a precise timestamp, unique equipment identifier, process batch number, and corresponding parameter value.

[0016] To ensure data quality, the system implements a three-level preprocessing mechanism: first, statistical analysis is used to identify and eliminate extreme values ​​caused by sensor anomalies; second, dynamic time warping is used to align asynchronously sampled data across different devices; and finally, Z-score or Min-Max normalization methods are used to unify parameter dimensions. This processed dataset ensures traceability of the production process, standardization enhances comparability of data across multiple devices, and integrity provides reliable support for subsequent analysis. This standardized dataset lays the foundation for process optimization and quality control. By establishing a correlation model between parameter fluctuations and product yield, intelligent monitoring of the production process can be achieved.

[0017] In step S102, the anomaly detection model employs an intelligent analysis model based on a deep neural network. Its core structure consists of two components: an encoder and a decoder. Using a self-supervised learning mechanism, the model first compresses high-dimensional process parameters into low-dimensional feature vectors and then reconstructs these features back into their original data form. By calculating the difference between the original data and the reconstructed data (i.e., the reconstruction error), the model can quantitatively assess whether the current production status is within the normal range.

[0018] During the model training phase, only normal process data is input, and network parameters are optimized through unsupervised learning. An error minimization strategy is employed during training, enabling the model to deeply understand the variations in normal process parameters. This training approach avoids the limitations of traditional methods that rely on anomaly sample labeling, significantly improving the model's adaptability to diverse process scenarios.

[0019] After processing, the system outputs three key pieces of information: the original process parameter combination, the corresponding reconstruction error value, and anomaly determination results based on thresholds. For example, in a photolithography process, if a significant deviation in exposure dose is detected from the standard value, the system will issue an anomaly alert based on the significant increase in reconstruction error.

[0020] Through the adaptive learning capabilities of deep neural networks, the model can accurately capture subtle changes in parameter fluctuations; the dynamic judgment mechanism based on reconstruction error effectively reduces the misjudgment rate of traditional fixed threshold methods; the unsupervised training mode enables it to quickly adapt to new equipment or new process parameters; the output error data also provides a basis for subsequent analysis of the impact weights of key process parameters, providing data support for production process optimization.

[0021] For step S103, the input dataset consists of key process parameters collected by the FDC system, covering core indicators across multiple steps, including lithography, etching, and deposition. Examples include light source wavelength, exposure dose, and depth of focus in lithography, etch rate and gas flow in etching, and film thickness in deposition. These parameters are input into the detection model as single, dual, or multivariate combinations, forming diverse feature combination analysis scenarios.

[0022] The detection model analyzes the changing patterns of parameter combinations and generates corresponding reconstruction errors. This error reflects the degree to which the current parameter combination deviates from the normal process pattern. For example, in a photolithography process scenario, when the exposure dose parameter is input alone, the reconstruction error output by the model increases significantly to 4.8, indicating that abnormal fluctuations in this parameter are the primary factor causing process deviation. However, when both the light source wavelength and exposure dose are input simultaneously, the error drops to 0.39, indicating that the synergistic effect of these two parameters can more accurately reflect the process status.

[0023] By comparing the reconstruction errors of different feature combinations, engineers can intuitively identify the impact of key parameters. For example, the example data shows that the error is highest when exposure dose is entered alone, but significantly decreases when combined with other parameters, indicating that this parameter has the most significant impact on process stability. This analysis model not only reveals potential correlations between parameters but also provides a quantitative basis for process optimization. When the error value of a parameter combination exceeds the threshold, the system triggers an abnormality alarm and generates a detailed error analysis report, helping engineers quickly identify the root cause of the problem.

[0024] Through multi-dimensional feature combination analysis, the core parameters that cause anomalies can be accurately identified; the quantitative output of reconstruction errors improves the interpretability of detection results, allowing engineers to intuitively understand the model decision logic; the analysis results based on parameter influence weights can guide the targeted adjustment of process parameters and effectively reduce the occurrence rate of anomalies.

[0025] In step S104, the impact of process parameters on anomaly detection results is analyzed using feature importance analysis. This technique quantifies the marginal contribution of each parameter to the model output, revealing the differences in the weights of different process variables in anomaly determination. Its core principle is that when a parameter is introduced or removed, the change in the model reconstruction error directly reflects the parameter's impact on anomaly detection results.

[0026] A game-theoretic Shapley value algorithm is employed to comprehensively consider all possible parameter combinations. This approach iterates through permutations and combinations of different parameter subsets, calculating each parameter's marginal contribution to the reconstruction error under different collaborative environments. A weighted average is then used to derive a global importance metric. This process, similar to analyzing individual contributions in teamwork, ensures that the importance of each parameter is assessed independently while also taking into account synergistic effects.

[0027] This analysis enables precise identification of critical process parameters. For example, in the photolithography process, exposure dose is often significantly more important than other parameters, indicating that even small fluctuations can directly lead to process anomalies. Engineers can adjust monitoring strategies accordingly, focusing resources on optimizing these critical parameters. Furthermore, the analysis results provide a quantitative basis for tracing the root causes of anomalies. When an anomaly is detected, the system can quickly identify the parameter combination with the greatest impact, shortening the troubleshooting cycle.

[0028] For step S105, based on the feature importance calculated previously, the key process parameters are visualized, allowing engineers to intuitively grasp which parameters have the greatest impact on the anomaly detection results, thereby optimizing the process flow and improving the wafer processing yield.

[0029] Visual explanations have several core goals. First, they clearly demonstrate the importance of key process parameters, allowing engineers to clearly understand the impact of each parameter on anomaly detection results. Second, they help engineers quickly locate problems and identify the key factors causing anomalies, reducing troubleshooting time. Third, they provide a decision-making basis for optimizing process parameters and improving the efficiency of parameter adjustments.

[0030] There are multiple visualization options available. Feature importance ranking lists can be used to present the importance of each parameter in a tabular format, allowing engineers to prioritize which parameters for optimization. Bar charts allow engineers to intuitively compare the impact of different process parameters. Radar charts are suitable for comparing multiple process parameters, clearly demonstrating the relative importance of each. Heat maps, on the other hand, use color depth to intuitively display the correlation between features and anomaly detection results.

[0031] Visualized results can provide decision support for process optimization. Parameters with significant impact are optimized through a combination of production experience and data analysis. The optimization results are monitored in real time to see if the reconstruction error of the anomaly detection model decreases. A feedback mechanism for process adjustments is also established. If the anomaly rate remains high after optimization, further analysis of other important parameters is performed. If the impact of certain parameters varies significantly across batches, the feature importance calculation weights can be dynamically adjusted.

[0032] This improves model interpretability, allowing engineers to intuitively understand analysis results. It accurately identifies key influencing factors, optimizes processes, and improves wafer manufacturing yield. It provides data support, speeds up optimization decisions, and improves decision-making efficiency. Furthermore, this approach is applicable not only to photolithography but also to other wafer processing steps such as etching and deposition, improving overall semiconductor manufacturing efficiency. Feature importance calculation and visual analysis help engineers better understand the impact of process parameters on anomaly detection.

[0033] Furthermore, the anomaly detection model is a deep learning anomaly detection model based on an encoder-decoder structure.

[0034] Furthermore, the feature importance algorithm includes: The anomaly detection model is denoted as ; The characteristics of the input data set are recorded as , Indicates the number of input features; To not include The set of all input features; is a feature subset The output value after inputting the anomaly detection model; the importance of each feature is recorded as ;No. The importance of a feature is calculated as follows: ; in, express The features have combinations.

[0035] After determining the subset, Features have a specific order Combination case, the feature set is: , subset Sequential combinations, remaining features Combinations, determine subsets Later combination of situations, so For subset The proportion of feature combinations, all possible subsets The sum of the proportions of feature combinations is equal to 1.

[0036] Furthermore, the FDC system is used to collect processing parameters of the lithography machine, the etching machine and the deposition machine.

[0037] Furthermore, the reconstruction error of the anomaly detection model is used to evaluate whether the combination of input features causes anomalies in wafer processing. The larger the reconstruction error, the higher the anomaly rate of the feature combination.

[0038] Furthermore, the method further comprises: According to the feature importance, the process parameters corresponding to the features with high importance are adjusted; The reconstruction error of the adjusted anomaly detection model is monitored to determine whether a processing anomaly occurs.

[0039] In the semiconductor wafer manufacturing process, feature importance analysis identifies the degree of influence of each process parameter on anomaly detection results. Based on this analysis, parameters with the highest degree of influence are prioritized for adjustment, parameters with the second highest degree of influence are considered secondary, and parameters with the lowest degree of influence are not considered for adjustment.

[0040] For critical process parameters, we developed adjustment strategies based on past production experience and data analysis. By fine-tuning the ranges of these key parameters, we achieved greater stability. Furthermore, we employed a closed-loop control approach, utilizing the FDC system to monitor actual parameter values ​​in real time and compare them with set values. If deviations were detected, the system automatically adjusted the equipment parameters to ensure they remained within the new set ranges.

[0041] After adjusting process parameters, monitor the anomaly detection model's reconstruction error in real time. Using visualization tools, present the reconstruction error as a time series graph, visually demonstrating its changing trend over time. Compare the post-adjustment reconstruction error to the pre-adjustment error to quickly determine the effectiveness of the adjustment strategy.

[0042] Changes in the reconstruction error are used to determine whether a processing anomaly has occurred. If the reconstruction error decreases significantly and stabilizes at a low level, the parameter adjustment strategy is effective, reducing anomalies and improving wafer processing stability. If the reconstruction error does not decrease significantly or even increases, further analysis is required. This could be due to inappropriate parameter adjustments or other important parameters significantly affecting anomaly detection results. In this case, engineers will consider adjusting other important parameters and continuously monitor the reconstruction error.

[0043] When adjusting process parameters, it's important to monitor the impact of certain parameters across batches. If you notice significant fluctuations in the impact of certain parameters, this suggests their importance may be changing with production conditions. In this case, we dynamically adjust the weights used in the feature importance calculation to enable the model to more accurately reflect the impact of each parameter on anomaly detection results, enabling more effective identification and resolution of anomalies.

[0044] By continuously adjusting process parameters based on feature importance, monitoring reconstruction errors in real time, and identifying any processing anomalies, we achieve continuous process optimization. With accumulated experience and continuous parameter optimization, process parameters gradually approach optimal levels, significantly reducing the occurrence of processing anomalies. Ultimately, this effectively improves wafer manufacturing yield and production efficiency.

[0045] Furthermore, the visual explanation includes: A ranking list of process parameters is output according to the feature importance, and the importance of each parameter is presented in a visual manner.

[0046] Visual explanations are provided to help understand the impact of each process parameter on anomaly detection. This process creates a ranked list of process parameters based on feature importance, clearly displaying the order of importance of each parameter. Furthermore, parameter importance is presented using visualizations such as bar charts, radar plots, and heat maps, allowing engineers to intuitively compare the impact of different parameters and quickly identify key parameters. This helps engineers focus on key parameters, providing clear guidance for subsequent process parameter optimization and adjustment, thereby improving wafer manufacturing yield and production efficiency.

[0047] Second embodiment The second embodiment of the present application relates to a method for interpreting wafer processing data. The second embodiment is an improvement on the first embodiment, and the specific improvements are: The actual wafer processing process is used as an example to explain the operation of this method. In the wafer processing process, lithography (LITHO) is a very critical link. For the sake of simplicity, the input features are set as Wavelength (light source wavelength), Exposure Dose (exposure dose), DOF (depth of focus), and the anomaly detection model is used. Its output is represented by out.

[0048] Model baseline: When the number of features = 0, the feature set is , out=0.5 for the anomaly detection model Under different feature combinations, the output matrix of out is shown in the following table:

[0049] When the number of features = 1 and the feature is Wavelength, the marginal change is:

[0050] Calculate the marginal change of Wavelength in all scenarios:

[0051] Calculate the marginal change of Exposure Dose in all scenes:

[0052] Calculate the marginal change in DOF across all scenes:

[0053] Calculate the importance of feature Wavelength:

[0054] Calculate the importance of feature exposed:

[0055] Calculate the importance of feature DOF:

[0056] From the above calculations, the most important feature is Exposure Dose.

[0057] The FDC system continuously collects key process parameters of the lithography machine. After cleaning and standardization, this data is divided into a training set and a validation set. Most of this data is used to train a deep learning model based on an encoder-decoder architecture. This model uses reconstruction errors to identify abnormal conditions in the production process.

[0058] By simulating changes in model output under different parameter combinations, the importance of each parameter was quantified. The results showed that one core parameter contributed significantly more to anomaly detection results than the others, while another parameter exhibited the characteristic of suppressing abnormal signals. Based on these analysis results, targeted adjustments were made to the monitoring strategy for key parameters. By optimizing the fluctuation control range of this parameter and appropriately adjusting the threshold setting of another parameter, the stability of the production process was significantly improved. The anomaly detection system achieved quantifiable improvements in both accuracy and production yield, while significantly reducing the time cost for engineers to troubleshoot anomalies.

[0059] The implementation process of this solution can be summarized as: data acquisition and preprocessing → model training and validation → feature importance analysis → process parameter optimization. This entire process requires no architectural changes to the existing FDC system. By integrating an interpretable analysis module, it seamlessly integrates deep data value mining with process optimization. This approach has proven effective in enhancing the intelligent analysis capabilities of the semiconductor manufacturing process, providing technical support for continuous improvement in production quality.

[0060] The step division of the above various methods is only for the purpose of clear description. During implementation, they can be combined into one step or some steps can be split and decomposed into multiple steps. As long as they include the same logical relationship, they are all within the scope of protection of this application; adding insignificant modifications or introducing insignificant designs to the algorithm or process without changing the core design of the algorithm and process are all within the scope of protection of this application.

[0061] In addition, some embodiments of the present application further provide an electronic device. The electronic device may be various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, etc. The electronic device may also be various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices, and other similar computing devices.

[0062] The electronic device includes: one or more processors; and a memory storing computer program instructions, wherein the computer program instructions, when executed, enable the processor to perform the steps of the method provided in any one or more of the above embodiments. Figure 2 An exemplary structural diagram of the electronic device is disclosed. Figure 2 As shown, the electronic device includes: one or more processors 1101, memory 1102, and interfaces for connecting various components, including high-speed and low-speed interfaces. The various components are interconnected using different buses and can be mounted on a common motherboard or in other ways as needed. The processor can process instructions executed within the electronic device, including instructions stored in or on the memory for displaying graphical information of a GUI on an external input / output device (such as a display device coupled to the interface). In some other embodiments, if desired, multiple processors and / or multiple buses can be used with multiple memories and multiple storage devices. Similarly, multiple electronic devices can be connected, with each device providing some of the necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present application described and / or claimed herein.

[0063] The electronic device may further include: an input device 1103 and an output device 1104. The processor 1101, the memory 1102, the input device 1103 and the output device 1104 may be connected via a bus or other means. Figure 2 The bus connection is taken as an example.

[0064] Input device 1103 can receive input digital or character information and generate key signal input related to user settings and function control of the electronic device. Examples include a touch screen, keypad, mouse, trackpad, touchpad, pointing stick, one or more mouse buttons, trackball, joystick, and other input devices. Output device 1104 may include a display device, auxiliary lighting devices (e.g., LEDs), and tactile feedback devices (e.g., vibration motors). The display device may include, but is not limited to, a liquid crystal display (LCD), a light-emitting diode (LED) display, and a plasma display. In some embodiments, the display device may be a touch screen.

[0065] To provide user interaction, the electronic device may be a computer. The computer includes a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user, as well as a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices may also be used to provide user interaction; for example, feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback), and input from the user may be received in any form, including acoustic input, voice input, or tactile input.

[0066] In the embodiments of the present application, a computer program / instruction is stored on a computer-readable medium. When executed by a processor, the computer program / instruction implements the steps of the method provided in any one or more of the above embodiments. The computer-readable medium may be included in the electronic device described in the above embodiments, or it may exist independently and not be incorporated into the device. The computer-readable medium carries one or more computer-readable instructions.

[0067] 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 the non-transitory software programs, instructions, and modules stored in the memory 1102 to execute various functional applications and data processing of the server, thereby implementing the program instructions / modules corresponding to the method provided in any one or more of the above embodiments of the present application.

[0068] The memory 1102 may include a program storage area and a data storage area, wherein the program storage area may store an operating system and applications required for at least one function; the data storage area may store data created based on the use of the electronic device, etc. In addition, the memory 1102 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory 1102 may optionally include a memory remotely located relative to the processor 1101, and these remote memories may be connected to the electronic device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0069] It should be noted that the computer-readable medium described in this application may be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. Computer-readable media may include, for example, but not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or components, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to, an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this application, a computer-readable medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, device, or component.

[0070] Computer-readable media includes both permanent and non-permanent, removable and non-removable media, and can be implemented using any method or technology for information storage. Information can be 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 technologies, compact disc-read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape, disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information that can be accessed by a computing device.

[0071] Computer program code for performing the operations of the present application can be written in one or more programming languages, or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0072] In the above embodiments, all or part of the embodiments may be implemented using software, hardware, firmware, or any combination thereof. For example, implementation may be achieved using an application-specific integrated circuit (ASIC), a general-purpose computer, or any other similar hardware device. In some embodiments, the software program of the present application may be executed by a processor to implement the above steps or functions. Similarly, the software program of the present application (including related data structures) may be stored in a computer-readable recording medium, such as a RAM memory, a magnetic or optical drive, a floppy disk, or the like. In addition, some steps or functions of the present application may be implemented using hardware, for example, as a circuit that cooperates with a processor to perform the various steps or functions.

[0073] The computer program product provided in the embodiments of the present application includes one or more computer programs / instructions, which, when executed by a processor, generate, in whole or in part, the processes or functions described in the embodiments of the present application. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via a wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) method. The computer-readable storage medium may be any available medium that can be accessed by a computer or a data storage device such as a server or data center that includes one or more available media. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a magnetic tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state disk (SSD)).

[0074] The flowcharts or block diagrams in the accompanying drawings illustrate the possible architectures, functions and operations of the devices, methods and computer program products according to various embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, program segment or part of code, and the module, program segment or part of code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, as well as the combination of boxes in the block diagram and / or flowchart, can be implemented with a dedicated hardware-specific system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0075] The scope of this application is defined by the appended claims rather than the foregoing description and is therefore intended to encompass within this application all changes that come within the meaning and range of equivalents of the claims. Any reference signs in the claims should not be construed as limiting the claims to which they relate. In addition, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices stated in a device claim may also be implemented by one unit or device through software or hardware. Words such as "first" and "second" are only used to distinguish the description and do not indicate any particular order, nor should they be understood as indicating or implying relative importance.

[0076] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any person skilled in the art may easily propose variations or substitutions within the technical scope disclosed in the present application, and such variations or substitutions shall be encompassed within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be subject to the scope of protection of the claims, and the above embodiments shall be regarded as exemplary and non-limiting.

Claims

1. A method for interpreting wafer processing data, characterized in that: The method comprises: Obtain wafer processing data collected by the FDC system from the tool to form an input data set; Processing the input data set using an anomaly detection model built using deep learning to obtain an output data set; The input data set includes input features of different combinations, and the output data set is the reconstruction error corresponding to the input features; The marginal impact of each input feature on the output of the anomaly detection model is calculated using a feature importance algorithm, and the feature importance is calculated based on the marginal changes of all feature combinations. Based on the feature importance, a visual explanation of key process parameters is provided.

2. The method according to claim 1, characterized in that The anomaly detection model is a deep learning anomaly detection model based on an encoder-decoder structure.

3. The method according to claim 1, characterized in that The feature importance algorithm includes: The anomaly detection model is denoted as ; The characteristics of the input data set are recorded as , Indicates the number of input features; To not include The set of all input features; is a feature subset The output value after inputting the anomaly detection model; the importance of each feature is recorded as ;No. The importance of a feature is calculated as follows: ； in, express The features have combinations.

4. The method according to any one of claims 1 to 3, characterized in that The FDC system is used to collect processing parameters of photolithography tools, etching tools, and deposition tools.

5. The method according to claim 4, characterized in that The reconstruction error of the anomaly detection model is used to evaluate whether the combination of input features causes anomalies in wafer processing. The larger the reconstruction error, the higher the anomaly rate of the feature combination.

6. The method according to claim 5, characterized in that The method further comprises: According to the feature importance, the process parameters corresponding to the features with high importance are adjusted; The reconstruction error of the adjusted anomaly detection model is monitored to determine whether a processing anomaly occurs.

7. The method according to claim 6, characterized in that The visual explanation includes: A ranking list of process parameters is output according to the feature importance, and the importance of each parameter is presented in a visual manner.

8. An electronic device, characterized in that: The electronic device comprises: one or more processors; and A memory storing computer program instructions, which, when executed, cause the processor to perform the steps of the method according to any one of claims 1 to 7.

9. A computer-readable medium having a computer program / instruction stored thereon, characterized in that: When the computer program / instructions are executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images