Anomaly detection device and anomaly detection method

Abstract

Provided is an anomaly detection device for assessing whether an anomaly is present in processing results obtained through processing of a sample by a processing device, the anomaly detection device comprising: a machined shape prediction unit 41 that, by using a processing results prediction model that takes a control parameter value of the processing device and an observation parameter value obtained by observation of a phenomenon occurring within the processing device during processing by the processing device as an explanatory variable and that takes an evaluation value for processing by the processing device as an objective variable, predicts an evaluation value for processing by the processing device; and a first anomaly detection unit 43 that detects anomaly in the processing results produced by the processing device, on the basis of a difference between an evaluation value for processing to be assessed and a predicted evaluation value for processing to be assessed which is predicted by inputting, into the processing results prediction model, the control parameter value used in the processing to be assessed and the observation parameter value observed in the processing to be assessed.

Description

[Technical field]

[0001] The present invention relates to an abnormality detection device and an abnormality detection method. [Background technology]

[0002] In semiconductor processing, the desired semiconductor processing can be achieved by processing semiconductor samples under appropriate processing conditions. In recent years, new materials that constitute devices have been introduced and device structures have become more complex, expanding the control range of semiconductor processing equipment and adding many control parameters. Processes have become multi-step, enabling fine and complex processing. In order to produce high-performance devices using semiconductor processing equipment, it is necessary to develop a process that derives appropriate processing conditions that will achieve the desired processed shape of the semiconductor sample.

[0003] Patent document 1 discloses generating a predictive model showing the relationship between the processing conditions applied to a semiconductor processing device and the processing results by the semiconductor processing device, and using the predictive model to estimate the conditions for outputting a target value of the processing result. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2019-40984 A Summary of the Invention [Problem to be solved by the invention]

[0005] There are cases where a processing shape estimated for certain processing conditions using a prediction model such as that shown in Patent Document 1 deviates from the processing shape actually obtained by processing a semiconductor sample using a semiconductor processing device under the processing conditions. In this case, two causes are considered. The first is when the accuracy of the prediction model is insufficient. In this case, it is necessary to add learning data and improve the accuracy of the prediction model. The second is when some abnormality occurs in the processing step by the semiconductor processing device, and as a result, the desired processing result cannot be obtained.

[0006] Even if learning data including the latter experimental results are used to train a predictive model, there is no problem because the irregular learning data will eventually be ignored as exceptional events during training. However, this will require training the predictive model using a larger amount of learning data.

[0007] Repeated processing tests using semiconductor processing equipment have a significant impact on the cost and time required for process development, so it is desirable to exclude experimental results in which processing abnormalities have occurred in the semiconductor processing equipment from the learning data. [Means for solving the problem]

[0008] An anomaly detection device according to one embodiment of the present invention is an anomaly detection device that determines whether or not there is an anomaly in a processing result obtained by processing a sample using a processing device, and includes a processing shape prediction unit that predicts an evaluation value of the processing by the processing device using a processing result prediction model in which the control parameter values ​​of the processing device and observed parameter values ​​obtained by observing phenomena occurring within the processing device during processing by the processing device are used as explanatory variables, and the evaluation value of the processing by the processing device is used as a target variable, and a first anomaly detection unit that detects an anomaly in the processing result by the processing device based on the difference between the evaluation value of the processing to be determined and a predicted evaluation value of the processing to be determined, which is predicted by inputting the control parameter values ​​used in the processing to be determined and the observed parameter values ​​observed in the processing to be determined into the processing result prediction model. Effect of the Invention

[0009] It is possible to train a prediction model that predicts the processing of a processing device with a small amount of learning data. Other objects and novel features will become apparent from the description of this specification and the accompanying drawings. [Brief description of the drawings]

[0010] [Figure 1A] FIG. 1 is a system configuration diagram of an anomaly detection system. [Figure 1B] FIG. 2 is a hardware configuration diagram of the abnormality detection device. [Figure 1C] FIG. 2 is a diagram showing data and programs stored in a storage device. [Figure 2A] FIG. 2 is a functional block diagram of the anomaly detection device in a learning (training) process. [Figure 2B] FIG. 4 is a functional block diagram of an abnormality detection device in an abnormality detection process. [Diagram 3] 4 is an example of a data structure of processing recipe data. [Figure 4] 1 is an example of a data structure of observation data. [Diagram 5] 13 is an example of a data structure of experiment result data. [Figure 6] 13 is an example of a data structure of prediction result data. [Figure 7] 13 is an example of a data structure of normal contribution data. [Figure 8] 1 is an overall flowchart of anomaly detection. [Figure 9] 13 is a flow chart showing a calculation flow of a shape anomaly score of verification data. [Figure 10] 13 is a flow chart showing a calculation flow of the contribution anomaly score of the verification data. [Figure 11] 1 is an example of a data structure of knowledge data. [Figure 12] This is an example of a GUI. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0011] Hereinafter, preferred embodiments of the present invention will be described with reference to the drawings.

[0012] Figure 1A shows a system configuration diagram of the anomaly detection system. The following describes an example of using this system in process development for semiconductors or semiconductor devices that contain semiconductors. In process development, appropriate processing conditions are derived for semiconductor processing equipment that processes semiconductor samples to achieve a target, for example, a desired processed shape.

[0013] The processing device 2 is a device that processes semiconductor samples. The contents of the processing performed by the processing device 2 are not limited. For example, the device includes a lithography device, a film forming device, a pattern processing device, an ion implantation device, and a cleaning device. The lithography device includes an exposure device, an electron beam lithography device, and an X-ray lithography device. The film forming device includes a CVD (Chemical Vapor Deposition), a PVD (Physical Vapor Deposition), an evaporation device, a sputtering device, and a thermal oxidation device. The pattern processing device includes a wet etching device, a dry etching device, an electron beam processing device, and a laser processing device. The ion implantation device includes a plasma doping device and an ion beam doping device. The cleaning device includes a liquid cleaning device and an ultrasonic cleaning device.

[0014] In the following, a plasma processing apparatus for etching a semiconductor sample will be described as an example of the processing apparatus 2. In the plasma processing apparatus, a high-frequency alternating electromagnetic field acts on a processing gas in a reactor 2a to generate plasma, which is used to etch a sample 3. The etching process in the reactor 2a is controlled according to a processing recipe set in a control unit 2b.

[0015] The evaluation device 5 is a device that evaluates the processing performed by the processing device 2 on the sample 3. For example, it is a processed dimension measuring device using an electron microscope, and measures the processed dimensions of the sample 3 by the processing device 2.

[0016] The observation device 4 is a device that observes phenomena occurring in the reactor 2a while the processing device 2 is processing the sample 3. The phenomenon to be observed is not limited, and can be appropriately selected depending on the phenomenon acting on the sample 3 during processing by the processing device 2. Here, an example will be described in which a spectrophotometer that observes the light emission of plasma in the reactor 2a is used as the observation device 4.

[0017] The control unit 2b of the processing device 2 processes the sample 3 (here, etching process) according to the processing recipe data. The observation device 4 observes the emission state of plasma in the reactor 2a during the processing period of the processing device 2 and acquires observation data. After the processing by the processing device 2 is completed, the evaluation device 5 measures the processing dimensions of the sample 3 and acquires experimental result data. The processing recipe data and the acquired observation data and experimental result data are accessible from the abnormality detection device 1, and are used to create a processing result prediction model that predicts the processing result by the processing device and to detect and judge abnormalities in the processing result, which will be described later.

[0018] A user accesses the abnormality detection device 1 from a terminal 7 via the network 6 or directly from an input / output device of the abnormality detection device 1, and executes the abnormality detection process.

[0019] FIG. 1B shows the hardware configuration of the anomaly detection device 1. The anomaly detection device 1 is an information processing device (computer) and has the following configuration. The anomaly detection device 1 comprises a processor (CPU) 11, a memory 12, a storage device 13, an input device 14, an output device 15, and a communication device 16, which are connected by a bus 17. A GUI (Graphical User Interface) is implemented by the input device 14, which is a keyboard and a pointing device, and the display, which is the output device 15, and a user can use the device interactively via the GUI. The communication device 16 is an interface for connecting to a network 6. It is also possible to display the GUI implemented by the device on a terminal 7 via the network 6.

[0020] The storage device 13 is usually composed of a hard disk drive (HDD) or a solid state drive (SSD), and stores the programs executed by the anomaly detection device 1, data to be processed by the programs, or data resulting from processing by the programs. The memory 12 is composed of a random access memory (RAM), and temporarily stores the programs, data required to execute the programs, etc., in response to instructions from the processor 11. The processor 11 executes the programs loaded from the storage device 13 to the memory 12, thereby functioning as a functional unit (functional block) that provides a predetermined function.

[0021] The anomaly detection device 1 does not have to be realized by one information processing device, but may be realized by multiple information processing devices. Also, some or all of the functions of the anomaly detection device 1 may be realized as an application on a cloud.

[0022] 1C shows data and programs stored in the storage device 13. Details of all will be described later, but the data includes processing recipe data 21, observation data 22, experiment result data 23, prediction result data 24, normal contribution data 25, and knowledge data 26, and the programs include a processing shape prediction program 31, a prediction explanation program 32, a shape abnormality detection program 33, a contribution abnormality detection program 34, an integrated judgment program 35, and a knowledge linking program 36.

[0023] FIG. 8 shows the overall flow of anomaly detection performed by the anomaly detection device 1. Steps S01 to S03 are a learning (training) process for the processing result prediction model. In this embodiment, the processing by the processing device is described using an example of etching processing by a plasma processing device, so that the processing result prediction model is hereinafter referred to as a processing shape prediction model to match the example. In this process, normal case data is used to generate a processing shape prediction model and calculate a normal contribution degree. Here, the contribution degree refers to the magnitude of the contribution of each explanatory variable to the predicted result (objective variable) in the processing shape prediction model. Steps S04 to S07 are an anomaly detection process for the processing result by the processing device. In this process, an anomaly judgment is performed on the processing result obtained according to an arbitrary processing recipe. Each process will be described below.

[0024] (1) Learning (training) process The processing of steps S01 to S03 in the overall flow (see FIG. 8) will be described. A functional block diagram of the abnormality detection device 1 in this process is shown in FIG. 2A. The machining shape prediction unit 41 is a functional unit that functions as a result of the processor 11 executing the machining shape prediction program 31, and the prediction explanation unit 42 is a functional unit that functions as a result of the processor 11 executing the prediction explanation program 32.

[0025] The abnormality detection device 1 reads the processing recipe data 21, the observation data 22, and the experimental result data 23 (S01). The data used in the learning process is data on cases where the processing of the sample by the processing device 2 has been performed normally. Here, normal processing of the sample means that an evaluation value obtained as experimental result data, which will be described later, can be obtained from the processed sample. In this embodiment, the evaluation value is the processing dimension of the sample.

[0026] 3 shows an example of the data structure of the processing recipe data 21. The experiment number is a number that uniquely identifies a process (experiment) of the processing device 2, the feature name is a control parameter of the processing device 2, and the value is a value set for the control parameter in the experiment.

[0027] 4 shows an example of the data structure of the observation data 22. The same experiment number is used as the processing recipe data 21. The feature name is the observation parameter of the observation data acquired by the observation device 4 during the experiment with the experiment number, and the value is the value of the observation parameter observed in the experiment. In this embodiment, an example is shown in which the observation parameter value is the emission intensity in a specified band of plasma generated in the processing device 2.

[0028] 5 shows an example of the data structure of the experimental result data 23. The same experiment number is used as in the processing recipe data 21. The experimental result indicates an evaluation value acquired by the evaluation device 5 for the processing result obtained by the experiment with the corresponding experiment number. In this example, the evaluation value is the value of the shape parameter of the sample 3 processed by the processing device 2, specifically, the processing depth.

[0029] Next, the machining shape prediction unit 41 learns (trains) the machining shape prediction model using the read machining recipe data 21, observation data 22, and experiment result data 23 (S02). The machining shape prediction model learned in step S02 is a model in which the objective variables are shape parameter values ​​acquired as the experiment result data 23, and the explanatory variables are control parameter values ​​acquired as the machining recipe data 21 and observation parameter values ​​acquired as the observation data 22. The machining shape prediction unit 41 executes supervised learning using the read normal case data as learning data. The machining shape prediction model in this step includes the observation parameter values ​​as its explanatory variables, thereby making it possible to reflect the processing state of the processing device 2 during the experiment in the inference of the objective variables.

[0030] Furthermore, the values ​​of the processing recipe data 21 and the observation data 22 are inputted into the trained processing shape prediction model to obtain the predicted result data 24. FIG. 6 shows an example of the data structure of the predicted result data 24. The same number as the processing recipe data 21 is used as the experiment number. The feature name includes the control parameter of the processing recipe data 21 and the observation parameter of the observation data 22, and the value indicates the control parameter value of the processing recipe data 21 and the observation parameter value of the observation data 22 in the experiment. The predicted result is registered as a shape parameter value obtained by substituting the control parameter value and the observation parameter value of the experiment number into the trained processing shape prediction model. The shape parameter value calculated by the processing shape prediction model is a predicted value (predicted evaluation value) of the evaluation value defined in the experiment result data 23, and in this example, it is the processing depth.

[0031] Next, the contribution of each explanatory variable in the trained machining shape prediction model is calculated (S03). The prediction explanation unit 42 interprets on what basis the trained machining shape prediction model made the prediction. Since the contents of the machining shape prediction model, which is an AI model, are black boxes, it is not possible to know the reason why the prediction was obtained as it is. For this reason, the prediction explanation unit 42 utilizes an XAI (Explainable AI) technology that interprets on what basis the AI ​​model made the prediction, calculates the contribution indicating the contribution of each explanatory variable to the prediction result (objective variable), and accumulates it as normal contribution data 25. Since it is the contribution of the explanatory variable in normal case data, it is called normal contribution. As a tool for performing such calculation, SHAP (Shapley Additive explanations) is known. FIG. 7 shows an example of the data structure of the normal contribution data 25. The experiment number is the same as that of the machining recipe data 21. The feature name includes the control parameters and observation parameters, which are explanatory variables of the machining shape prediction model, and the contribution degree registers the contribution degree of each explanatory variable to the prediction result in the experiment (prediction result data 24).

[0032] In addition, when the machining shape prediction model is updated, for example, by performing additional learning, the value of the normal contribution data 25 calculated by the prediction explanation unit 42 is also changed. Therefore, when the machining shape prediction model is updated by the machining shape prediction unit 41, the prediction explanation unit 42 recalculates the contribution of the explanatory variables in each experiment and updates the normal contribution data 25.

[0033] (2) Anomaly detection process for processing results The processing of steps S04 to S07 of the overall flow (see FIG. 8) will be described. A functional block diagram of the abnormality detection device 1 in this process is shown in FIG. 2B. The shape abnormality detection unit 43 is a functional unit that functions as a result of the processor 11 executing the shape abnormality detection program 33, the contribution abnormality detection unit 44 is a functional unit that functions as a result of the processor 11 executing the contribution abnormality detection program 34, the integrated judgment unit 45 is a functional unit that functions as a result of the processor 11 executing the integrated judgment program 35, and the knowledge linking unit 46 is a functional unit that functions as a result of the processor 11 executing the knowledge linking program 36.

[0034] The shape abnormality detection unit 43 calculates a shape abnormality score of the verification data (S04). Details of step S04 are shown in FIG.

[0035] The abnormality detection device 1 reads the processing recipe data 51, the observation data 52, and the experiment result data 53, which are verification data (S11). These correspond to data for one experiment, including the processing recipe data 21 (see FIG. 3), the observation data 22 (see FIG. 4), and the experiment result data 23 (see FIG. 5).

[0036] Next, the machining shape prediction unit 41 inputs the read machining recipe data 51 and the observation data 52 into the learned machining shape prediction model to obtain prediction result data 54 (S12). The prediction result data 54 corresponds to one experimental portion of the prediction result data 24 (see FIG. 6).

[0037] Next, the shape abnormality detection unit 43 calculates the difference between the experimental result data 53 and the predicted result data 54 (S13), and calculates a shape abnormality score from the degree of difference (S14). There are no limitations on the method of calculating the shape abnormality score, but for example, it is defined that the larger the difference between the experimental result data 53 and the predicted result data 54, the larger the shape abnormality score.

[0038] Returning to the explanation of the overall flow (FIG. 8), the contribution degree anomaly detection unit 44 then calculates the contribution degree anomaly score of the verification data (S05). Details of step S05 are shown in FIG.

[0039] First, the prediction explanation unit 42 calculates the contribution of each explanatory variable in the trained processing shape prediction model to obtain the contribution data 55 (S21). The contribution data 55 corresponds to one experiment's worth of the normal contribution data 25 (see FIG. 7).

[0040] Next, the contribution abnormality detection unit 44 compares the contribution data of the normal case data stored in the normal contribution data 25 with the contribution data 55 (S22), and calculates a contribution abnormality score from the degree of difference between the normal contribution data pattern and the contribution data pattern (S23). Although the method of calculating the contribution abnormality score is not limited, for example, it is defined that the greater the difference between the contribution data pattern of the normal case data stored in the normal contribution data 25 and the pattern of the contribution data 55, the greater the contribution abnormality score becomes.

[0041] Returning to the description of the overall flow (FIG. 8), the integration determination unit 45 then integrates the shape anomaly score and the contribution anomaly score to perform an anomaly determination (S06). For example, the integration determination unit 45 performs the following determination based on the combination of the shape anomaly score and the contribution anomaly score. When the shape anomaly score is determined to be normal and the contribution anomaly score is determined to be normal, the processing result is determined to be normal. When the shape anomaly score is determined to be normal and the contribution anomaly score is determined to be abnormal, the processing result is determined to be normal. This is because it is determined that a new contribution pattern has been found because the processing by the processing device has been performed correctly. When the shape anomaly score is determined to be abnormal and the contribution anomaly score is determined to be normal, the processing result is determined to be normal. This is because it is determined that the accuracy of the prediction model is insufficient because the processing by the processing device has not been performed as expected, even though the contribution pattern is the same as the previous learning data. When the shape anomaly score is determined to be abnormal and the contribution anomaly score is determined to be abnormal, the processing result is determined to be abnormal. This is because it is highly likely that an abnormality in the processing device has caused the shape anomaly and the contribution anomaly.

[0042] The experimental results judged as normal by the integrated judgment unit 45 can be used, for example, as new learning data for updating the machining shape prediction model. On the other hand, for the experimental results judged as abnormal by the integrated judgment unit 45, the knowledge linking unit 46 may present information based on the knowledge data (S07). An example of the data structure of the knowledge data 26 is shown in FIG. 11. The content of the knowledge is arbitrary, but it is assumed to be knowledge related to the observation parameters of the observation data 22. The example in FIG. 11 stores the names of candidate causative substances when plasma emission in a specified band is observed.

[0043] For example, in the example of Fig. 11, when a contribution anomaly is determined for observation data in the 270-300 nm band of the emission spectrum, the user is informed that candidate substances such as SiCl and Si may be involved, making it easier for the user to investigate the cause of the anomaly that occurred in the processing device.

[0044] The abnormality detection results of the shape abnormality detection unit 43 and the contribution abnormality detection unit 44, the judgment result of the integrated judgment unit 45, and the knowledge extracted by the knowledge linking unit 46 are displayed as presentation information 56 on the GUI.

[0045] FIG. 12 shows an example of a GUI for executing the process of FIG. 8. For example, the project designation unit 61 designates a project name that designates the creation of a machining shape prediction model for determining the processing conditions of the processing device 2. The machining recipe data 21, the observation data 22, and the experiment result data 23 are linked to the project name. The data designation unit 62 designates a normal case data ID (the data ID corresponds to an experiment number) and a verification data ID (the data ID corresponds to an experiment number). For example, in this project, a provisional machining shape prediction model is first created, and then, based on the provisional machining shape prediction model, the normal / abnormality of the experiment result is judged, and whether or not it is adopted as learning data is judged, and the provisional machining shape prediction model is updated by the learning data judged to be normal. This enables the generation of a highly accurate machining shape prediction model with less learning data. In this example, an example is shown in which the experiment numbers 1 to 50 of the machining recipe data 21, the observation data 22, and the experiment result data 23 are used to generate a provisional machining shape prediction model, and the experiment number 51 is judged as to whether or not to add it as learning data.

[0046] The shape abnormality score display unit 63 displays the processing result of step S04 (see Figure 8), the contribution abnormality score display unit 64 displays the processing result of step S05, the integrated judgment display unit 65 displays the processing result of step S06, and the knowledge display unit 66 displays the knowledge extracted in step S07.

[0047] The shape anomaly score display unit 63 and the contribution anomaly score display unit 64 display the anomaly score values ​​and thresholds for the normal cases and the verification data. The thresholds may be set by the user or may be set automatically and statistically. For example, the thresholds may be defined as a value obtained by adding twice the variance to the average value of the anomaly scores in the normal cases, i.e., the learning data of the processed shape prediction model. The thresholds may be changed in accordance with improvements in the accuracy of the processed shape prediction model.

[0048] Knowledge display section 66 identifiably displays the observation parameters (here, the wavelengths of the emission spectrum) in which abnormal values ​​have been detected among the contributions of the observation data, and also displays the knowledge extracted in step S07.

[0049] As the above, as the present embodiment, an example of applying the disclosed technology to the process of creating a processing shape prediction model has been described, but the present invention is not limited to this. For example, after the learning of the processing shape prediction model is completed, in mass production processing of semiconductor samples in which conditions are set based on this model, it is also possible to monitor whether a processing abnormality occurs due to a processing device by calculating a contribution abnormality score.

[0050] As an embodiment of the described embodiment, a semiconductor device manufacturing system may be considered in which an application for managing the operation of a line including semiconductor processing equipment is executed on a platform. The semiconductor processing equipment is connected to the platform via a network and is controlled by the platform. In this case, the anomaly detection device 1 is executed as an application on the platform, thereby making it possible to implement the present embodiment in the semiconductor device manufacturing system. [Explanation of symbols]

[0051] 1: Anomaly detection device, 2: Processing device, 2a: Reactor, 2b: Control unit, 3: Sample, 4: Observation device, 5: Evaluation device, 6: Network, 7: Terminal, 11: Processor (CPU), 12: Memory, 13: Storage device, 14: Input device, 15: Output device, 16: Communication device, 17: Bus, 21: Processing recipe data, 22: Observation data, 23: Experimental result data, 24: Prediction result data, 25: Normal contribution data, 26: Knowledge data, 31: Processing shape prediction program, 32: Prediction explanation program, 33: Shape anomaly detection program, 34: 4: Contribution abnormality detection program, 35: Integrated judgment program, 36: Knowledge linking program, 41: Processing shape prediction section, 42: Prediction explanation section, 43: Shape abnormality detection section, 44: Contribution abnormality detection section, 45: Integrated judgment section, 46: Knowledge linking section, 51: Processing recipe data, 52: Observation data, 53: Experimental result data, 54: Prediction result data, 55: Contribution data, 56: Presentation information, 61: Project designation section, 62: Data designation section, 63: Shape abnormality score display section, 64: Contribution abnormality score display section, 65: Integrated judgment display section, 66: Knowledge display section.

Claims

1. An abnormality detection device that determines whether or not there is an abnormality in a processing result obtained by processing a sample using a processing device, a machining shape prediction unit that predicts an evaluation value of the processing by the processing device using a processing result prediction model in which control parameter values ​​of the processing device and observed parameter values ​​obtained by observing phenomena occurring within the processing device during processing by the processing device are used as explanatory variables, and an evaluation value of the processing by the processing device is used as a response variable; an anomaly detection device comprising: a first anomaly detection unit that detects an anomaly in a processing result by the processing device based on a difference between an evaluation value of a processing to be judged and a predicted evaluation value of the processing to be judged that is predicted by inputting control parameter values ​​used in the processing to be judged and observed parameter values ​​observed in the processing to be judged into the processing result prediction model.

2. In claim 1, a prediction explanation unit that calculates a contribution of each explanatory variable to the predicted evaluation value for the process to be determined; and a second anomaly detection unit that detects an anomaly in a processing result by the processing device based on a contribution pattern of explanatory variables in the processing to be determined calculated by the prediction explanation unit.

3. In claim 2, an anomaly detection device having an integrated determination unit that integrates the anomaly detection result of the first anomaly detection unit and the anomaly detection result of the second anomaly detection unit to determine whether or not there is an anomaly in the processing result to be determined.

4. In claim 3, the processing result prediction model is trained based on a plurality of experimental results obtained by processing samples using the processing device; a storage device for storing control parameter values ​​used in the experimental process, observed parameter values ​​observed in the experimental process, and an evaluation value of the experimental process, which are used as learning data for the processing result prediction model; The first anomaly detection unit detects an anomaly in the processing result by the processing device by comparing the difference between the evaluation value and the predicted evaluation value of the processing to be judged with the difference between the evaluation value of the processing in the experiment and the predicted evaluation value of the processing in the experiment predicted by inputting control parameter values ​​used in the processing in the experiment and observed parameter values ​​observed in the processing in the experiment into the processing result prediction model.

5. In claim 4, the prediction explanation unit calculates a contribution of each explanatory variable to the predicted evaluation value for the processing in the experiment, and stores a contribution pattern of the explanatory variables in the processing in the experiment calculated by the prediction explanation unit in the storage device; The second anomaly detection unit detects an anomaly in the processing result by the processing device by comparing a contribution pattern of explanatory variables in the processing to be judged with a contribution pattern of explanatory variables in the processing in the experiment.

6. In claim 5, The processing shape prediction unit is an anomaly detection device that trains the processing result prediction model using the learning data stored in the storage device.

7. In claim 6, When the integrated judgment unit judges that there is no abnormality in the processing result to be judged, the machining shape prediction unit trains the processing result prediction model using an evaluation value of the processing to be judged, a control parameter value used in the processing to be judged, and an observation parameter value observed in the processing to be judged; The integrated judgment unit judges that there is an abnormality in the processing result to be judged when both the first anomaly detection unit and the second anomaly detection unit detect an abnormality, and judges that there is no abnormality in the processing result to be judged when at least one of the first anomaly detection unit and the second anomaly detection unit does not detect an abnormality.

8. In claim 7, an anomaly detection device having a knowledge linking unit that, when the integrated judgment unit judges that there is an anomaly in the processing result to be judged, compares the observed parameter value in which an anomaly has been detected by the second anomaly detection unit with knowledge data and extracts related knowledge.

9. In claim 8, an anomaly detection device that displays, on a display device, the anomaly detection results of the first anomaly detection unit and the second anomaly detection unit, the judgment result of the integrated judgment unit, and the knowledge extracted by the knowledge linking unit.

10. In any one of claims 1 to 9, The processing device is a plasma processing device that etches the sample, An anomaly detection device that uses the emission intensity of the plasma in a predetermined band obtained by observing the emission of plasma generated within the processing device during processing by the processing device as an observation parameter value, and uses the processing dimensions of the sample by the etching process as an evaluation value of the processing by the processing device.

11. An anomaly detection method using an anomaly detection device for determining the presence or absence of an anomaly in a processing result obtained by processing a sample using a processing device, comprising: the abnormality detection device includes a machining shape prediction unit and a first abnormality detection unit, the machining shape prediction unit predicts an evaluation value of the processing by the processing device using a processing result prediction model in which control parameter values ​​of the processing device and observed parameter values ​​obtained by observing a phenomenon occurring within the processing device during processing by the processing device are used as explanatory variables, and an evaluation value of the processing by the processing device is used as a response variable; The first anomaly detection unit detects an anomaly in a processing result by the processing device based on a difference between an evaluation value of the processing to be judged and a predicted evaluation value of the processing to be judged predicted by inputting control parameter values ​​used in the processing to be judged and observed parameter values ​​observed in the processing to be judged into the processing result prediction model.

12. In claim 11, The anomaly detection device further includes a prediction explanation unit and a second anomaly detection unit, the prediction explanation unit calculates a contribution of each explanatory variable to the predicted evaluation value for the process to be determined, The second anomaly detection unit detects an anomaly in a processing result by the processing device based on a contribution pattern of explanatory variables in the processing to be determined calculated by the prediction explanation unit.

13. In claim 12, The abnormality detection device further includes an integrated determination unit, The integrated determination unit integrates the anomaly detection result of the first anomaly detection unit and the anomaly detection result of the second anomaly detection unit to determine whether or not there is an anomaly in the processing result to be determined.

14. In claim 13, When the integrated judgment unit judges that there is no abnormality in the processing result to be judged, the machining shape prediction unit trains the processing result prediction model using an evaluation value of the processing to be judged, a control parameter value used in the processing to be judged, and an observation parameter value observed in the processing to be judged; The anomaly detection method, wherein the integrated judgment unit judges that there is an abnormality in the processing result to be judged when both the first anomaly detection unit and the second anomaly detection unit detect an abnormality, and judges that there is no abnormality in the processing result to be judged when at least one of the first anomaly detection unit and the second anomaly detection unit does not detect an abnormality.

15. In any one of claims 11 to 14, The processing device is a plasma processing device that etches the sample, An anomaly detection method in which the emission intensity of the plasma in a predetermined band obtained by observing the emission of plasma generated within the processing device during processing by the processing device is used as an observation parameter value, and the processing dimension of the sample by the etching process is used as an evaluation value of the processing by the processing device.

Images