Equipment diagnostic device, equipment diagnostic method, plasma processing device, and semiconductor device manufacturing system

Abstract

To provide a plasma processing apparatus in which occurrence of a possible unplanned maintenance is predicted, and a necessary maintenance work and a time point in a planned maintenance to which the work is to be added are immediately determined from the view point of the cost of high-priority maintenance.SOLUTION: An apparatus diagnosis device includes a deterioration level estimation unit that estimates the deterioration level of a plasma processing apparatus upon receiving an output of a sensor mounted on the plasma processing apparatus, a maintenance work occurrence possibility estimation unit that calculates the possibility of occurrence of an unplanned maintenance work which is not included in the original maintenance plan on the basis of the deterioration level estimated by the deterioration level estimation unit, an actual maintenance cost calculation unit that calculates an actual maintenance cost, and a plan-added maintenance cost calculation unit that outputs a maintenance plan amendment obtained by making an amendment to the original maintenance plan to which the unplanned maintenance work is added on the basis of the possibility of occurrence of an unplanned maintenance work estimated by the maintenance work occurrence possibility estimation unit and the actual maintenance cost of the plasma processing apparatus calculated by the actual maintenance cost calculation unit.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a method and an apparatus for diagnosing a plasma processing apparatus that processes a semiconductor wafer by plasma.

Background Art

[0002] A plasma processing apparatus is an apparatus that performs plasma processing to vaporize a substance and remove the substance on a wafer by the action of the substance in order to form a fine pattern on the semiconductor wafer. In a plasma etching apparatus, normally, maintenance such as cleaning inside the apparatus and replacing parts (planned maintenance) is performed according to a pre-established maintenance plan based on, for example, the number of wafers processed. However, due to deterioration of parts caused by aging and accumulation of reaction by-products depending on the usage method, unplanned maintenance work may occur. In order to reduce the non-operating time (downtime) due to unplanned maintenance, it is required to sequentially monitor the deterioration state of the parts and take early measures such as cleaning and replacement according to the deterioration state.

[0003] In order to realize such early measures, in a diagnostic apparatus for an apparatus, generally, using sensor values sequentially acquired from a plurality of state sensors attached to the apparatus, the degree of deterioration is estimated from the deviation from the normal state and compared with a set threshold value to issue an alarm. For example, in the specification of PCT Publication No. WO2018-542408 (Patent Document 1), it is described that "the abnormality detection apparatus applies statistical modeling to a summary value obtained by summarizing observed values, estimates a state in which noise has been removed from the summary value, and generates a predicted value that predicts a summary value for the next period based on the estimation. The abnormality detection apparatus detects the presence or absence of an abnormality in the apparatus to be monitored based on the predicted value."

[0004] In addition, as a diagnostic method after detecting an abnormality, for example, Japanese Patent Application Laid-Open No. 2015-148867 (Patent Document 2) describes "creating a classification criterion for classifying the phenomenon pattern based on the work keyword included in the maintenance history information" and "creating a diagnostic model for estimating the work keyword to be presented to the maintenance worker based on the classified phenomenon pattern and the work keyword".

[0005] Furthermore, Japanese Patent Application Laid-Open No. 2019-133412 (Patent Document 3) describes a method of calculating the maintenance cost in a representative maintenance method by presetting the failure probability.

Prior Art Documents

Patent Documents

[0006]

Patent Document 1

Patent Document 2

Patent Document 3

Summary of the Invention

Problems to be Solved by the Invention

[0007] Patent Document 1 describes a method of detecting the presence or absence of an abnormality in a device and issuing an alarm. However, it does not describe what information should be provided on how to specifically modify the maintenance plan after the alarm is issued.

[0008] Therefore, the present invention provides a device diagnostic device that presents a maintenance plan proposal considering maintenance costs such as device operation rate and costs required for maintenance work in addition to the degradation degree estimation result using sensor values. There are the following two problems in presenting a maintenance plan proposal considering maintenance costs.

[0009] The first point is to estimate the actual maintenance cost for each part and the work related to the part from the maintenance history consisting of free descriptions. From the perspective of recorded man-hours and the like, the maintenance history is often in free descriptions. It is necessary to identify the combination of parts to be maintained and work from such unstructured records and estimate the actual maintenance cost.

[0010] For example, Patent Document 2 describes a method of extracting keywords from words representing work in the maintenance history and assigning importance. However, there are problems such as the inability to identify the combination of parts and work only by keyword extraction using words, and the inability to handle cases where there are the same parts with multiple names.

[0011] The second point is to calculate the maintenance cost (hereinafter referred to as the maintenance cost at the time of planned incorporation) when incorporating unplanned maintenance into planned maintenance at multiple time points using the actual maintenance cost, and present an optimal maintenance plan amendment from the perspective of maintenance cost.

[0012] The plasma processing apparatus is a vacuum apparatus, and it takes time to start up and shut down the apparatus. Therefore, in the method of performing maintenance every time immediately after an alarm is issued, the operation rate may conversely decrease. Therefore, the goal is to adopt a maintenance method of predicting the occurrence of unplanned maintenance in advance and incorporating additional work into the previously established planned maintenance.

[0013] For example, Patent Document 2 describes a method of classifying phenomenon patterns based on sensor values and presenting work keywords to maintenance workers. However, it is difficult for maintenance workers or maintenance plan formulators to determine at which point in time the work should be performed.

[0014] In addition, Patent Document 3 describes a method of calculating the maintenance cost in a representative maintenance method by setting a failure probability in advance. However, since the etching apparatus has a long start-up time, if the apparatus is stopped only for maintenance corresponding to this alarm immediately after an alarm is issued from the apparatus monitor, the problem that the apparatus operation will conversely decrease occurs.

[0015] The present invention solves the above-described problems of the prior art, and in a plasma processing apparatus, predicts in advance the occurrence of unplanned maintenance that may occur, and enables a user such as a maintenance planner or a maintenance worker to immediately determine from the perspective of prioritized maintenance costs what necessary maintenance work should be carried out and at what point in the planned maintenance schedule it should be incorporated. The present invention provides a device diagnosis method and a device diagnosis apparatus for a plasma processing apparatus.

Means for Solving the Problems

[0016] In order to solve the above-described problems, in the present invention, in a device diagnosis apparatus for diagnosing the state of a plasma processing apparatus, a planned incorporation-time maintenance cost calculation unit that corrects the maintenance plan of the plasma processing apparatus based on a second maintenance cost obtained using the probability of occurrence of a maintenance operation of the plasma processing apparatus and a first maintenance cost related to the maintenance operation is provided. The probability is a probability obtained based on the degree of deterioration of components of the plasma processing apparatus estimated using monitor values of the state of the plasma processing apparatus. The first maintenance cost is a maintenance cost obtained based on the actual value related to the maintenance work at the time of maintenance inspection planned in the maintenance plan. and maintenance work not planned in the maintenance plan The second maintenance cost is the maintenance cost when a maintenance operation is added at the time of maintenance inspection planned in the maintenance plan. expected value Each of the maintenance inspections is characterized in that a maintenance operation is added such that the second maintenance cost is minimized.

[0017] In order to solve the above-described problems, in the present invention, in a plasma processing apparatus including a device diagnosis apparatus for diagnosing the state of the apparatus, the device diagnosis apparatus includes a planned incorporation-time maintenance cost calculation unit that corrects the maintenance plan of the apparatus based on a second maintenance cost obtained using the probability of occurrence of a maintenance operation of the apparatus itself and a first maintenance cost related to the maintenance operation. The probability is a probability obtained based on the degree of deterioration of components of the apparatus itself estimated using monitor values of the state of the apparatus itself. The first maintenance cost is the maintenance work at the time of maintenance inspection planned in the maintenance plan. and maintenance work not planned in the maintenance planis a maintenance cost obtained based on the actual value related to [the maintenance plan], and the second maintenance cost is a maintenance cost when new maintenance work is added during the maintenance inspection planned in the maintenance plan. expected value Each of the maintenance inspections is characterized in that maintenance work is added such that the second maintenance cost is minimized.

[0018] Furthermore, in order to solve the above-described problems, in the present invention, in a semiconductor device manufacturing system including a device diagnostic apparatus that is connected to a plasma processing apparatus via a network and executes device diagnostic processing, the device diagnostic apparatus uses a probability that maintenance work of the plasma processing apparatus occurs and a first maintenance cost related to the maintenance work. A planned incorporation-time maintenance cost calculation unit that corrects the maintenance plan of the plasma processing apparatus based on a second maintenance cost obtained, the probability being a probability obtained based on the degree of deterioration of components of the plasma processing apparatus estimated using a monitor value of the state of the plasma processing apparatus, and the first maintenance cost being the maintenance work at the time of the maintenance inspection planned in the maintenance plan and maintenance work not planned in the maintenance plan is a maintenance cost obtained based on the actual value related to [the maintenance plan], and the second maintenance cost is a maintenance cost when new maintenance work is added during the maintenance inspection planned in the maintenance plan. expected value Each of the maintenance inspections is characterized in that maintenance work is added such that the second maintenance cost is minimized.

[0019] Furthermore, in order to solve the above-described problems, in the present invention, in a device diagnostic method for diagnosing the state of a plasma processing apparatus using a device diagnostic apparatus that diagnoses the state of the plasma processing apparatus, the device diagnostic apparatus uses a probability that maintenance work of the plasma processing apparatus occurs and a first maintenance cost related to the maintenance work in order to correct the maintenance plan of the plasma processing apparatus. Executing a step of calculating a second maintenance cost obtained, the probability being a probability obtained based on the degree of deterioration of components of the plasma processing apparatus estimated using a monitor value of the state of the plasma processing apparatus, and the first maintenance cost being the maintenance work at the time of the maintenance inspection planned in the maintenance plan and maintenance work not planned in the maintenance planIt is the maintenance cost obtained based on the actual performance values related to [the relevant item], and the second maintenance cost is the maintenance cost when new maintenance work is added during the maintenance inspection planned in the maintenance plan. expected value Each of the maintenance inspections is characterized in that maintenance work is added so that the second maintenance cost is minimized.

Effect of the Invention

[0020] According to the present invention, in a plasma processing apparatus, the occurrence of unplanned maintenance that may occur is predicted in advance, and for example, users such as maintenance planners and maintenance workers can immediately determine the necessary maintenance work and at which point in the planned maintenance the work should be incorporated, from the perspective of the prioritized maintenance cost. Problems, configurations, and effects other than those described above will be clarified by the description of the following embodiments.

Brief Description of the Drawings

[0021]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Mode for Carrying Out the Invention

[0022] The present invention estimates the maintenance cost when additional work (unplanned maintenance work not incorporated in the original maintenance plan) is incorporated into the maintenance plan based on sensor data, maintenance history, and the original maintenance plan, and outputs a maintenance plan amendment such that the estimated maintenance cost is minimized. The present invention provides a device diagnosis method and a device diagnostic device for a plasma processing apparatus.

[0023] The present invention provides a device diagnosis method and a device diagnostic device that output, as a maintenance plan amendment, additional work such that the estimated maintenance cost is minimized at a plurality of planned maintenance times from the sensor data, maintenance history, and maintenance plan of a device group.

[0024] In the present invention, from the maintenance history consisting of free descriptions of the plasma processing apparatus, the maintenance work, which is a combination of each component and work performed with each maintenance ID, is specified. Based on the maintenance cost information such as the equipment operation rate, the actual maintenance cost for each maintenance work is calculated. The degree of deterioration of each component is estimated using the sensor values sequentially acquired by the plasma processing apparatus and management values such as the number of sample processes. From the probability distribution of the degree of deterioration of the plasma processing apparatus group at the time of occurrence of each maintenance work, the probability of occurrence of maintenance work until a certain degree of deterioration is reached is estimated. At the time of equipment diagnosis, the transition of the degree of deterioration is predicted from the sequentially estimated degree of deterioration, and based on the actual maintenance cost and the probability of occurrence of the maintenance work, the maintenance cost when the additional maintenance work is incorporated into the planned maintenance at multiple time points is calculated and presented.

[0025] The equipment diagnosis apparatus according to the present invention has the following three configurations. (1) Probability of occurrence of maintenance work estimation unit: Estimates the probability of occurrence of each maintenance work at a certain point in time from the distribution of the degree of deterioration of the components estimated using the sensor data of the model construction unit. (2) Actual maintenance cost calculation unit: Identifies the maintenance work (combination of components and work) of each maintenance ID from the maintenance history consisting of free descriptions and the maintenance work dictionary describing component / work keywords consisting of various expressions, and calculates the actual maintenance cost of each maintenance work in combination with the equipment operation rate data. (3) Maintenance cost calculation unit at the time of incorporation into the plan: Calculates the expected maintenance cost and its confidence interval when the necessary additional work is incorporated into the planned maintenance at multiple time points from the outputs of (1) and (2) above and the initial maintenance plan, and outputs an amendment to the maintenance plan such that the specified maintenance cost type (operation time, work cost, etc.) is minimized.

[0026] Hereinafter, embodiments of the present invention will be described with reference to the drawings. In all the drawings for explaining the embodiments, the same parts are generally denoted by the same reference numerals, and the repeated description thereof is omitted.

Example

[0027] The configuration of the device diagnostic apparatus 100 according to this embodiment is shown in FIG. 1. The device diagnostic apparatus 100 according to this embodiment is connected via a communication line 150 to a device group 1 composed of each plasma processing apparatus 11. The device diagnostic apparatus 100 includes an execution unit 2, an analysis unit 3, and a CPU (Central Processing Unit) 4, which are connected by an internal bus 5. The device diagnostic apparatus 100 may be connected to an external control device or storage device via the communication line 150. The following describes each configuration shown in FIG. 1.

[0028] (1) Plasma processing apparatus 11 In the configuration diagram shown in FIG. 1, in each plasma processing apparatus 11 constituting the device group 1, plasma 12 is generated according to the set processing conditions to process the sample 13. A state sensor group 14 (for example, a temperature sensor or a pressure sensor) is attached to the plasma processing apparatus 11, and the measured values of sensor values (for example, temperature and pressure) during the processing of the sample 13 with the plasma 12 or during idling when the generation of the plasma 12 is stopped can be acquired as time-series data. Examples of the plasma processing apparatus 11 constituting the device group 1 include a plasma etching apparatus.

[0029] (2) Device diagnostic apparatus 100 As shown in the configuration diagram of FIG. 1, the device diagnostic apparatus 100 includes an execution unit 2 that receives a sensor signal from the state sensor group 14 attached to each plasma processing apparatus 11 of the device group 1 and executes processing, an analysis unit 3 that analyzes the plasma processing apparatus 11, and a CPU 4 that controls the operations of the execution unit 2 and the analysis unit 3. They are connected by an internal bus 5. Further, the device diagnostic apparatus 100 is connected to each plasma processing apparatus 11 through the communication line 150, and the execution unit 2 acquires data from the state sensor group 14 from each plasma processing apparatus 11 via the communication line 150.

[0030] The execution unit 2 includes a storage unit 20, a degradation degree estimation unit 21, and a maintenance cost calculation unit 22 at the time of planned incorporation. Further, the storage unit 20 includes a sensor value storage unit 200, a management value storage unit 210, a degradation degree storage unit 220, and a maintenance cost storage unit 230 at the time of planned incorporation.

[0031] The analysis unit 3 includes a storage unit 30, a maintenance work occurrence probability estimation unit 31, an actual maintenance cost calculation unit 32, an input unit 33, and a display unit 34. Further, the storage unit 30 includes a maintenance history storage unit 300, a maintenance work dictionary storage unit 310, a maintenance plan storage unit 320, and an actual maintenance cost storage unit 330.

[0032] The sensor value storage unit 200 in the storage unit 20 of the execution unit 2 stores the sensor values (measurement values) acquired from the state sensor group 14 of the plasma processing apparatus 11 via the communication line 150.

[0033] FIG. 2 is a diagram showing an example of in-process data 201 stored in the sensor value storage unit 200 in tabular form. The measured values of the sensor values 202 for each sensor constituting the state sensor group 14 attached to the plasma processing apparatus 11 are stored as time-series data 203. When the plasma 12 is generated in the plasma processing apparatus 11 and the sample 13 is being processed, the process step ID: 204 is attached and stored for each set process step. Further, information specifying the process, such as the wafer ID: 206 and the process condition ID: 205, is stored in association with the sensor value 202.

[0034] The management value storage unit 210 stores management values such as the processing date and time when the sample 13 was processed by the plasma processing apparatus 11, the processing conditions of the sample 13 by the plasma processing apparatus 11 (input power for generating the plasma 12, processing time by the plasma 12, pressure during processing, temperature of the sample 13 during processing, etc.), and the number of processed samples of the sample 13.

[0035] The degradation degree estimation unit 21 uses a degradation degree estimation model pre-learned and constructed for each component to be monitored of each plasma processing apparatus 11, takes the sensor values 202 sequentially acquired from the sensor value storage unit 200 as input, estimates the degradation degree of each component corresponding to each degradation degree estimation model ID, and outputs it.

[0036] FIG. 3 shows, in tabular form, an example of the data 221 stored in the degradation degree storage unit 220. The data 221 stored in the degradation degree storage unit 220 includes the degradation degree 222 of the output for each degradation degree estimation model ID, information for specifying the process (time series data 223, wafer ID: 226, process condition ID: 225, process step ID: 224), and the like.

[0037] When calculating the maintenance cost at the time of planned incorporation in FIG. 1, the maintenance cost calculation unit 22 predicts the transition of the degradation degree 222 of each component corresponding to each future degradation degree estimation model ID from the transition of the degradation degree 222 up to the current time in the data 221 stored in the degradation degree storage unit 220, estimates the occurrence probability of an unplanned maintenance operation at the future planned maintenance time based on the output of the maintenance operation occurrence probability estimation unit 31, and calculates the expected maintenance cost (expected value of the maintenance cost) when incorporating an unplanned maintenance operation as an additional operation into the future planned maintenance using the information in the actual maintenance cost storage unit 330. The calculation result is stored in the planned incorporation time maintenance cost storage unit 230.

[0038] The maintenance history storage unit 300 in the storage unit 30 of the analysis unit 3 stores, as shown in the example of the data 301 stored in the maintenance history storage unit 300 in FIG. 4, the work content 307 for each work ID 302 (one planned maintenance or unplanned maintenance) in free description. In order to be used when calculating the actual maintenance cost, etc., the device ID 303, date and time 304, non-operation time 305, work classification 306, etc. are also stored together.

[0039] The maintenance work dictionary storage unit 310 stores information for extracting keywords of components and work when the actual maintenance cost calculation unit 32 specifies the combination of components and work from the work content of the maintenance history.

[0040] The maintenance plan storage unit 320 stores the time (date and time, number of wafers processed, etc.) and work content of the planned maintenance previously established by the maintenance plan maker.

[0041] The probability estimation unit 31 for unscheduled maintenance operations acquires the degree of degradation at the time of each maintenance operation from the degrees of degradation of the plasma processing apparatuses 11 in the apparatus group 1 stored in the degradation degree storage unit 220, estimates the probability density of the occurrence of maintenance operations according to the degree of degradation using the acquired degree of degradation, and further performs an integration process with respect to the degree of degradation to calculate the probability that an unscheduled maintenance operation will occur before reaching a certain degree of degradation (cumulative probability of occurrence of maintenance operations).

[0042] The actual maintenance cost calculation unit 32 identifies the combination of parts and operations using the information in the maintenance operation dictionary storage unit 310 from the work content stored in the maintenance history storage unit 300. Further, for each combination, by associating maintenance cost information such as the apparatus operation rate, the actual maintenance cost of each combination is calculated and stored in the actual maintenance cost storage unit 330.

[0043] The input unit 33 is an input device that receives information input by a user operation, such as a mouse or a keyboard.

[0044] The display unit 34 is, for example, a display or a printer, and is a device that graphically outputs information to the user based on the information stored in the storage unit 20 of the execution unit 2 and the storage unit 30 of the analysis unit 3, and the final maintenance plan amendment output from the maintenance cost calculation unit 22 at the time of incorporating the plan into the execution unit 2.

[0045] (3) Method for creating a maintenance plan amendment A method for creating an amendment to a maintenance plan that incorporates unscheduled maintenance operations not included in the maintenance plan into the maintenance operations included in the maintenance plan created in advance for each plasma processing apparatus 11 constituting the apparatus group 1 using the apparatus diagnostic apparatus 100 described above will be described with reference to FIG. 5.

[0046] In order to create an amendment to the maintenance plan that incorporates unscheduled maintenance operations into the maintenance plan of each plasma processing apparatus 11 constituting the apparatus group 1, in the apparatus diagnostic apparatus 100, first, the actual maintenance cost calculation unit 32 calculates the actual maintenance cost (S510). Next, the probability estimation unit 31 for unscheduled maintenance operations performs an estimation process of the probability of occurrence of unscheduled maintenance operations (S520).

[0047] Next, in the planned installation-time maintenance cost calculation unit 22, using the actual maintenance cost calculated in S510 and the data of the maintenance work occurrence probability obtained in S520, the calculation process of the planned installation-time maintenance cost is performed (S530), and a maintenance plan amendment is output to the display unit 34 (S540). The details of each step will be described below.

[0048] (3-1) Calculation process of actual maintenance cost: S510 With reference to FIG. 6, an example of a process of identifying the combination of parts of each maintenance work from the maintenance history performed by the actual maintenance cost calculation unit 32 of the device diagnostic device 100 and calculating the actual maintenance cost such as the device operation rate for each maintenance work will be described.

[0049] As a preliminary preparation, based on the knowledge of the device in advance, a maintenance work dictionary that defines the name groups of parts and operations respectively is created and stored in the maintenance work dictionary storage unit 310 (S511). As an example of the maintenance work dictionary 311 for parts shown in FIG. 7, the part names 314 and name groups 315 corresponding to each part ID 312 and part location name 313 are defined.

[0050] When extracting keywords from the maintenance history stored in the maintenance history storage unit 300, the keywords described in the name group 315 stored in the maintenance work dictionary storage unit 310 are extracted as the parts described in the part name 314. By adopting such an extraction method, it becomes possible to cope with the case where there are various names for one part. Also, by describing the part location name 313 together, it becomes possible to extract problems such as a high occurrence frequency of maintenance work around a certain part. The maintenance work dictionary for operations is created in the same way.

[0051] Next, in order to handle fluctuations in common words, the name group 315 is regularized (S512). Examples of fluctuations in common words include the presence or absence of spaces, singular or plural forms, and word endings. Even with such fluctuations, in order to extract keywords, for example, in the case of an "o ring" made of an elastic member with a circular cross-section for sealing a connection part of a vacuum device, it is described using a regular expression such as "o[ -]rings?". Regularization for such typical fluctuations in words can be easily automated.

[0052] Next, the maintenance history for the specified period is acquired from the maintenance history storage unit 300 (S513), and the work content of the maintenance history is split into sentences (S514). From each sentence, keywords that match the regular expression described in the name group are extracted and associated with the part name or site name to which the name group belongs (S515).

[0053] Subsequently, in order to identify the combination of parts and work, tags are assigned to the phrases in the sentence (S516). A part tag is assigned to the phrase extracted as a part, a work tag is assigned to the phrase extracted as work, and a tag indicating the part of speech is assigned to other words. By using the assigned tags to extract the combination of parts and work that matches the specified tag order in a sentence, the combination is identified (S517).

[0054] Regarding the specified tag order, it is only necessary that the combination of parts and work is correctly identified, and it is not limited to a specific tag order. For example, for work content such as ".. replace o ring and pump A..", part tags are assigned to o ring and pump A ( <cmp>) Replace with the work tag ( <work>)、and has the coordinating conjunction tag ( <and>) Tag it and say, " <work> ( <and> * <work> )* <cmp> ( <and> * <cmp>)By specifying the tag order using regular expressions like ")*", the combination of "replace o ring" and "replace pump A" can be correctly extracted.

[0055] Next, the actual maintenance cost for each combination of parts and operations is calculated (S518). For example, when using the device operation rate or non - operation time 305 for each work ID 302 stored in the maintenance history storage unit 300 as maintenance cost information, a regression is performed with the appearance / non - appearance of each combination in each work ID 302 as explanatory variables using dummy variables and the maintenance cost information as the objective variable to calculate the actual maintenance cost for each combination.

[0056] Finally, the calculated actual maintenance cost for each combination of parts and operations is output to the actual maintenance cost storage unit (S519).

[0057] Since the actual maintenance cost calculated as above may change when the period becomes available, it is updated regularly or at any arbitrary point in time.

[0058] (3 - 2) Estimation process of the occurrence probability of maintenance work: S520 When the maintenance work occurrence probability estimation unit 31 estimates the maintenance work occurrence probability according to the degree of deterioration, the degree of deterioration estimation unit 21 accumulates the degree of deterioration regarding the target parts of each plasma processing apparatus 11 in the apparatus group 1.

[0059] The degree of deterioration estimation unit 21 sequentially acquires the sensor values during the processing or idling of the sample 13 by the plasma processing apparatus 11 from the sensor value storage unit 200 using the registered degree of deterioration estimation model for each part, estimates the degree of deterioration, and outputs it to the degree of deterioration storage unit 220. The degree of deterioration estimation model may use a method suitable for estimating the deterioration of each part and is not limited to a specific method.

[0060] For example, when it is expected that the sensor values follow a normal distribution, during the learning of the degradation degree estimation model, the parameters of the normal distribution are learned using the sensor values for a certain period immediately after component replacement. When estimating the degradation degree, the degradation degree may be estimated using the Kullback-Leibler distance or the log-likelihood with respect to the distribution during learning as an index.

[0061] When normality is not expected, the degradation degree may be estimated using a method that can also handle non-normal distributions such as the k-nearest neighbor method. Also, in order to reduce the observation noise, the value obtained by calculating a statistic such as the average value for each processing step for the sensor values may be used as an input.

[0062] An example of the processing of the maintenance work occurrence probability estimation unit 31 when estimating the maintenance work occurrence probability is shown in FIG. 8. The processing flowchart shown in FIG. 8 corresponds to the step of estimating the degradation degree described above, and the learning of the degradation degree estimation model described above is performed in advance in this step.

[0063] First, the degradation degree regarding the component targeted by the plasma processing apparatus 11 is acquired from the degradation degree storage unit 220 (S521). Also, the date and time of the occurrence of the maintenance work content regarding the component targeted is acquired from the maintenance history storage unit 300 (S522). Further, the degradation degree at the time when the maintenance work occurred is extracted using the acquired data (S523). Next, the distribution of the degradation degree at the time of occurrence of the extracted maintenance work (the probability density of the occurrence of the maintenance work according to the degradation degree) is estimated (S524).

[0064] An example of the probability density distribution 901 estimated from the distribution 902 of the degradation degree at the time of occurrence of the maintenance work is shown in the graph 900 of FIG. 9. The horizontal axis of the graph in FIG. 9 indicates the degradation degree, and the vertical axis indicates the probability density. The distribution estimation method is not limited to a specific method, and for example, the Markov chain Monte Carlo method (MCMC) or the kernel density estimation method may be used.

[0065] Finally, integral processing is performed on the maintenance work occurrence probability density estimated in S524 with respect to the degradation degree, and the cumulative maintenance work occurrence probability (the probability that the maintenance work occurs until a certain degradation degree is reached) is calculated (S525).

[0066] With the above-described method for estimating the probability of occurrence of maintenance work, even when the method for constructing the degradation degree estimation model differs for each component, it is possible to calculate the probability of occurrence of maintenance work by a common method.

[0067] (3-3) Calculation process of maintenance cost at the time of plan incorporation: S530 An example of the process of the maintenance cost calculation unit 22 at the time of plan incorporation that outputs a revised maintenance plan considering the maintenance cost using the degradation degree sequentially estimated by the degradation degree estimation unit 21 and the cumulative probability of occurrence of maintenance work calculated by the maintenance work occurrence probability estimation unit 31 will be described with reference to FIG. 10.

[0068] First, the degradation degree from after the maintenance work on the component of interest of the target plasma processing apparatus 11 to the calculation time point is acquired from the degradation degree storage unit 220 (S531). Further, from the transition of the degradation degree up to the calculation time point, the future transition of the degradation degree is predicted (S532). At this time, the confidence interval of the prediction is also calculated. This prediction method is not particularly limited, and for example, an autoregressive model, which is a time series prediction method, may be used.

[0069] Next, as the maintenance plan, the future planned maintenance date and work content of the target plasma processing apparatus 11 are acquired from the maintenance plan storage unit 320 (S533). Further, from the predicted value of the degradation degree and its confidence interval at the future planned maintenance time point and the cumulative probability of occurrence of maintenance work calculated by the maintenance work occurrence probability estimation unit 31, as shown in the graph 110 of FIG. 11, the estimated value 1101 of the probability of occurrence of maintenance work and its confidence interval 1102 at the future planned maintenance time point (at the time points of dates t1, t2, and t3 in the graph 1100 of FIG. 11) are calculated (S534).

[0070] Next, the actual maintenance cost related to the maintenance work of interest is acquired from the actual maintenance cost storage unit 330 (S535). Further, from the estimated value of the probability of occurrence of maintenance work and its confidence interval calculated in S534 and the actual maintenance cost acquired in S535, the expected maintenance cost and its confidence interval at each planned maintenance time point are calculated (S536).

[0071] For example, when non-operation time is selected as the maintenance cost and additional work is performed at time t1 in FIG. 11, the expected value of the maintenance cost can be calculated as (probability that maintenance work occurs by time t1) × (actual maintenance cost when unplanned maintenance is performed) + (probability that maintenance work does not occur by time t1) × (actual maintenance cost when planned maintenance is performed).

[0072] The calculated result is stored in the maintenance cost memory unit 230 at the time of planned incorporation. As shown in the example of the data 231 stored in the maintenance cost memory unit 230 at the time of planned incorporation in FIG. 12, it is stored as the expected value and confidence interval 233 of the maintenance cost when each maintenance work 232 is incorporated at each planned maintenance time point. The example shown in FIG. 12 shows an example when non-operation time is selected as the maintenance cost type 234.

[0073] The above processing is executed for each component degradation degree estimation model, and a final maintenance plan amendment incorporating unplanned maintenance work not included in the initial maintenance plan is output and displayed on the display unit 34 (S537).

[0074] An example of the display screen 341 of the maintenance plan amendment is shown in FIG. 13. When the device ID 342 is selected and the type 343 of the prioritized maintenance cost is selected, the recommended additional work 345 at each planned maintenance date and time 344 is displayed in a list so that the selected maintenance cost is minimized. Thereby, the user can immediately judge from the perspective of the prioritized maintenance cost which unplanned necessary maintenance work not included in the initial maintenance plan and at which planned maintenance time point the work should be incorporated.

[0075] As described above, in this embodiment, a device diagnosis apparatus for diagnosing the state of a plasma processing apparatus includes a degradation degree estimation unit that receives the output of a sensor for monitoring the state of the plasma processing apparatus attached to the plasma processing apparatus and estimates the degradation degree of the plasma processing apparatus; a maintenance work occurrence probability estimation unit that calculates the probability that an unplanned maintenance work not included in the initial maintenance plan of the plasma processing apparatus will occur before the plasma processing apparatus reaches a certain degradation degree based on the degradation degree of the plasma processing apparatus estimated by the degradation degree estimation unit; an actual maintenance cost calculation unit that calculates the actual maintenance cost of the plasma processing apparatus; and a maintenance plan amendment calculation unit that outputs a maintenance plan amendment that modifies the initial maintenance plan of the plasma processing apparatus by incorporating unplanned maintenance work based on the probability that unplanned maintenance work of the plasma processing apparatus occurs estimated by the maintenance work occurrence probability estimation unit and the actual maintenance cost of the plasma processing apparatus calculated by the actual maintenance cost calculation unit.

[0076] Also, in this embodiment, in a device diagnosis method for diagnosing the state of a plasma processing apparatus using a device diagnosis apparatus, the actual maintenance cost calculation unit of the device diagnosis apparatus identifies the parts and combinations of each maintenance work in the maintenance history from the maintenance history of the plasma processing apparatus, calculates the actual maintenance cost such as the device operation rate of the plasma processing apparatus for each maintenance work, estimates the probability of occurrence of unplanned maintenance work of the plasma processing apparatus in the maintenance work occurrence probability estimation unit of the device diagnosis apparatus from the degradation degree of the plasma processing apparatus obtained by receiving the output of a sensor for monitoring the state of the plasma processing apparatus attached to the plasma processing apparatus, creates a maintenance plan amendment that modifies the initial maintenance plan of the plasma processing apparatus by incorporating unplanned maintenance work based on the probability of occurrence of unplanned maintenance work of the plasma processing apparatus estimated by the maintenance work occurrence probability estimation unit and the actual maintenance cost of the plasma processing apparatus calculated by the actual maintenance cost calculation unit in the maintenance plan amendment calculation unit at the time of incorporating the plan, and outputs the maintenance plan amendment created by the maintenance plan amendment calculation unit from the output unit of the device diagnosis apparatus.

[0077] According to this embodiment, it is possible to predict in advance the occurrence of unplanned maintenance that may occur in the plasma processing apparatus 11, and for example, users such as maintenance planners and maintenance workers can immediately determine from the perspective of prioritized maintenance costs what necessary maintenance work should be and at which point in the planned maintenance schedule it should be incorporated.

[0078] In addition, in the above example, an example was shown in which the recommended additional work 345 at each planned maintenance date and time 344 is displayed in a list so that the selected maintenance cost is minimized. However, this embodiment is not limited to this, and a plurality of recommended additional works 345 at each planned maintenance date and time 344 may be displayed in a list such that the selected maintenance cost is the second or third smallest, and it may be possible to select from among these multiple marked recommended additional works 345.

[0079] As described above, the invention made by the present inventor has been specifically described based on the embodiments. However, it goes without saying that the present invention is not limited to the above embodiments, and various modifications can be made without departing from the gist thereof. For example, the above embodiments have been described in detail for the purpose of explaining the present invention in an easy-to-understand manner, and are not necessarily limited to those having all the configurations described.

Explanation of Reference Numerals

[0080] 1... Apparatus group, 2... Execution unit of the apparatus diagnostic device, 3... Analysis unit of the apparatus diagnostic device, 11... Plasma processing apparatus, 20... Storage unit of the execution unit, 21... Degradation degree estimation unit, 22... Calculation unit for maintenance cost at the time of incorporation into the plan, 30... Storage unit of the analysis unit, 31... Estimation unit for the probability of occurrence of maintenance work, 32... Calculation unit for the actual maintenance cost, 33... Input unit, 34... Display unit, 100... Apparatus diagnostic device< / cmp> < / and> < / cmp> < / work> < / and> < / work> < / and> < / work> < / cmp>

Claims

1. In a device diagnostic apparatus for diagnosing the state of a plasma processing apparatus, a planned incorporation-time maintenance cost calculation unit that corrects the maintenance plan of the plasma processing apparatus based on a second maintenance cost obtained using the probability of occurrence of a maintenance operation for the plasma processing apparatus and a first maintenance cost related to the maintenance operation; the probability is a probability obtained based on the degree of deterioration of components of the plasma processing apparatus estimated using a monitor value of the state of the plasma processing apparatus; the first maintenance cost is a maintenance cost obtained based on actual values related to maintenance operations at the time of maintenance inspections planned in the maintenance plan and maintenance operations not planned in the maintenance plan; the second maintenance cost is an expected value of the maintenance cost when a maintenance operation is added at the time of a maintenance inspection planned in the maintenance plan; each of the maintenance inspections is characterized in that a maintenance operation is added such that the second maintenance cost is minimized. A device diagnostic apparatus.

2. In the device diagnostic apparatus according to claim 1, the second maintenance cost includes non-operation time or work cost. A device diagnostic apparatus.

3. In the device diagnostic apparatus according to claim 1, a maintenance operation dictionary storage unit that stores information for extracting a keyword of the component or a keyword of the maintenance operation when identifying a combination of a component of the plasma processing apparatus and a maintenance operation based on the work content of the maintenance operation history of the plasma processing apparatus. A device diagnostic apparatus characterized by this.

4. In the device diagnostic apparatus according to claim 3, a first maintenance cost calculation unit that calculates the first maintenance cost for each of the identified combinations of the component and the maintenance operation by associating maintenance cost information with each of the combinations; the maintenance cost information includes the operating rate of the plasma processing apparatus. A device diagnostic apparatus.

5. In a plasma processing apparatus including a device diagnostic apparatus for diagnosing the state of the device, the device diagnostic apparatus includes a planned incorporation-time maintenance cost calculation unit that corrects the maintenance plan of the self-device based on a second maintenance cost obtained using the probability of occurrence of a maintenance operation of the self-device and a first maintenance cost related to the maintenance operation; the probability is a probability obtained based on the degree of deterioration of components of the self-device estimated using a monitor value of the state of the self-device; The first maintenance cost is a maintenance cost obtained based on actual values related to maintenance work during maintenance inspections planned in the maintenance plan and maintenance work not planned in the maintenance plan. The second maintenance cost is an expected value of the maintenance cost when new maintenance work is added during the maintenance inspection planned in the maintenance plan. Each of the maintenance inspections is characterized in that maintenance work is added such that the second maintenance cost is minimized, in a plasma processing apparatus.

6. In the plasma processing apparatus according to claim 5, The second maintenance cost includes non-operation time or work cost, in a plasma processing apparatus.

7. In a semiconductor device manufacturing system including a device diagnostic apparatus that is connected to a plasma processing apparatus via a network and in which device diagnostic processing is executed, The device diagnostic apparatus includes a planned incorporation time maintenance cost calculation unit that corrects the maintenance plan of the plasma processing apparatus based on a second maintenance cost obtained using a probability of occurrence of maintenance work of the plasma processing apparatus and a first maintenance cost related to the maintenance work. The probability is a probability obtained based on a degree of deterioration of components of the plasma processing apparatus estimated using a monitored value of the state of the plasma processing apparatus. The first maintenance cost is a maintenance cost obtained based on actual values related to maintenance work during maintenance inspections planned in the maintenance plan and maintenance work not planned in the maintenance plan. The second maintenance cost is an expected value of the maintenance cost when new maintenance work is added during the maintenance inspection planned in the maintenance plan. Each of the maintenance inspections is characterized in that maintenance work is added such that the second maintenance cost is minimized, in a semiconductor device manufacturing system.

8. In the semiconductor device manufacturing system according to claim 7, The second maintenance cost includes non-operation time or work cost, in a semiconductor device manufacturing system.

9. In a device diagnostic method for diagnosing the state of a plasma processing apparatus using a device diagnostic apparatus that diagnoses the state of the plasma processing apparatus, The device diagnostic apparatus executes a step of calculating a second maintenance cost obtained using a probability of occurrence of maintenance work of the plasma processing apparatus and a first maintenance cost related to the maintenance work, in order to correct the maintenance plan of the plasma processing apparatus. The probability is a probability obtained based on the degree of deterioration of the components of the plasma processing apparatus estimated using the monitor value of the state of the plasma processing apparatus. The first maintenance cost is a maintenance cost obtained based on the actual values related to the maintenance work at the time of the maintenance inspection planned in the maintenance plan and the maintenance work not planned in the maintenance plan. The second maintenance cost is the expected value of the maintenance cost when new maintenance work is added at the time of the maintenance inspection planned in the maintenance plan. Each of the maintenance inspections is a device diagnosis method characterized in that maintenance work is added so that the second maintenance cost becomes minimum.

10. In the device diagnosis method according to claim 9, The second maintenance cost includes non-operation time or work cost, and is a device diagnosis method characterized by this.

Images