JP2004127989A - Semiconductor manufacturing apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To grasp the true cause promptly when a plurality of alarms are generated mainly by the troubles. <P>SOLUTION: A combination of a plurality of alarms to occur simultaneously or in a specified time period due to a trouble which is the principal cause of the alarms is in advance associated with the trouble, and then registered in a database 12 as a registered trouble occurrence pattern. Upon generation of a plurality of alarms due to a principal cause as a trouble taking place in an apparatus 1, data about the alarms are patterned by a pattern forming means 11 into a detected trouble occurrence pattern. The detected trouble occurrence pattern is compared in a comparing means 13 with registered trouble occurrence patterns from the database 12. In the presence of coincidence, a trouble is isolated in a concentration means 14 out of the registered trouble occurrence patterns as the principal cause of the alarms. Based on the result of concentration, alarms corresponding only to the principal cause are reported to a management controller 4. <P>COPYRIGHT: (C)2004,JPO

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a semiconductor manufacturing apparatus, and more particularly to an apparatus for monitoring an operation state such as a progress state of a processing operation or a failure occurrence state in a semiconductor manufacturing apparatus.
[Prior art]
The semiconductor manufacturing apparatus includes a control controller that monitors the apparatus. When the control controller detects a failure or abnormality of the apparatus, it notifies a management controller such as a host computer of the occurrence of the failure as an alarm. The management controller that has received the notification issues a command to the control controller based on the notification, and stops the processing of the semiconductor manufacturing apparatus.
[0002]
At this time, from the viewpoint of improving the operation rate and the productivity of the semiconductor manufacturing apparatus, when a serious failure occurs in the production of a substrate such as a wafer, it is a major issue to reduce the recovery time. In other words, the occurrence of the failure must be detected immediately, the cause thereof must be analyzed, and countermeasures must be taken to reduce the time of device failure.
[0003]
Conventionally, in order for equipment engineers to quickly ascertain the status of failure occurrence in semiconductor manufacturing equipment, the management controller is used. The method of automatically notifying to is taken.
[0004]
[Problems to be solved by the invention]
However, in recent years, semiconductor manufacturing apparatuses have become more sophisticated and more complicated with the increase in diameter and miniaturization of semiconductor wafers. In semiconductor manufacturing equipment with higher performance and complexity, various sensors simultaneously detect multiple failures due to the occurrence of a failure such as a certain failure or abnormality, and the controller that monitors it detects failure information one by one. It will be recognized as. The plurality of pieces of recognized fault information are all transmitted to a management controller including a host computer or the like, and the occurrence of the fault is automatically notified to the device engineer from the management controller. For this reason, the number of notifications to the device engineer increases, and it becomes difficult to detect important fault information from the received fault information to detect the true cause of the fault.
[0005]
This problem is not limited to the information on the failure occurrence status. It is also common to information on the operation status such as the progress status of the processing operation. For example, regarding the progress of the processing operation, if this is complicated, it may not be possible to quickly grasp the true progress.
[0006]
According to the present invention, a semiconductor manufacturing apparatus capable of resolving the above-described problems of the related art and capable of quickly grasping an operation state even when information related to a plurality of operation states is generated due to a certain operation as a basic factor. Is to provide.
[0007]
[Means for Solving the Problems]
According to a first aspect of the present invention, in a semiconductor manufacturing apparatus, when an operation such as a failure occurs in the apparatus, the operation is a basic factor to generate information relating to a plurality of operation states simultaneously or within a specific time. A registration unit for registering a pattern of information relating to a plurality of operation situations that occur simultaneously or within a specific time as a result of an operation as a basic factor in association with the operation as the basic factor, and registering the pattern as a registered operation status pattern; A pattern forming unit that forms a detection operation status pattern by patterning information relating to a plurality of operation statuses that have occurred simultaneously or within a specific time, and a detection operation status pattern output from the pattern formation unit from the registration unit. Comparing means for comparing with the read registered operation status pattern; and a plurality of operation statuses actually generated from the registered operation status pattern obtained by the comparison means. Means for narrowing down the information according to the above to the operation state of the operation which is a basic factor associated with the registered operation state pattern, so that the information relating to the operation state narrowed down by the narrowing down means is notified. A semiconductor manufacturing apparatus characterized by the following.
[0008]
In the registration means, in advance, a pattern of information relating to a plurality of operation situations that occur simultaneously or within a specific time due to a certain operation as a basic factor is registered as a registered operation situation pattern in association with the operation as the basic factor. deep. In fact, when an operation such as a failure occurs in the device and the operation becomes a basic factor and information about a plurality of operation situations is actually generated simultaneously or within a specific time, the information about the plurality of operation situations becomes a pattern. The pattern is formed by the forming unit, and is output from the pattern forming unit as a detection operation situation pattern. This detection operation status pattern is compared with the registration operation status pattern read from the registration unit by the comparison unit. When the comparison unit obtains a comparison match, the operation that is a basic factor of the detection operation state pattern is narrowed down from the registered operation state patterns obtained by the comparison unit by the comparison unit. Thus, the semiconductor manufacturing apparatus can narrow down information on the operation status corresponding to the basic factor from among the information on the plurality of operation statuses. Therefore, when an operation such as a failure occurs, even when information relating to a plurality of operation situations is issued due to the operation as a basic factor, the operation is performed from the information relating to the plurality of operation situations. It is possible to quickly find the true operating factor that is causing.
[0009]
According to a second aspect, in the first aspect, the information processing apparatus further includes a management controller that manages the semiconductor manufacturing apparatus, wherein the management controller is notified of information on the narrowed operation status from the semiconductor manufacturing apparatus. A semiconductor manufacturing apparatus system configured to transmit management data to the semiconductor manufacturing apparatus based on the notification at the time of management such as remote maintenance. The management controller performs management such as remote maintenance on the semiconductor manufacturing apparatus based on the information related to the operation status notified to the management controller.At this time, management data transmitted from the management controller to the semiconductor manufacturing apparatus includes: Since it is created based on the information related to the narrowed-down operation status, the management data amount can be reduced as compared with the case where the narrowing-down operation is not performed.
[0010]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, embodiments of the present invention will be described. Here, a case will be described where the operation status is not a progress status of the processing operation but a fault occurrence status.
[0011]
FIG. 3 is a configuration diagram of the semiconductor manufacturing apparatus system according to the embodiment. The semiconductor manufacturing apparatus system is a system for managing a plurality of semiconductor manufacturing apparatuses by a management controller, and includes a plurality of semiconductor manufacturing apparatuses 1, a control controller 2 provided in each semiconductor manufacturing apparatus 1, each control controller 2, and a management controller 4. Consists of
[0012]
[Semiconductor manufacturing equipment]
The semiconductor manufacturing apparatus 1 is composed of, for example, a batch type vertical processing apparatus. The processing apparatus includes mechanical units such as a substrate transfer robot and a boat elevator in addition to a processing furnace and a boat. A large number of substrates are loaded on a boat by a substrate transport robot, and the boat is carried into a processing furnace by a boat elevator. After the loading, a process gas is introduced under predetermined temperature and pressure conditions by a process control such as a temperature control, a gas flow rate control, and a pressure control based on a recipe, and a process such as film formation is performed. The plurality of semiconductor manufacturing apparatuses 1 are connected by a dedicated inter-device network.
[0013]
[Control controller]
The controller 2 is installed in the semiconductor manufacturing apparatus 1 to monitor process controls such as temperature control and gas flow control, and to monitor mechanical units such as a substrate transfer robot and a boat elevator, and to give an operation instruction to each mechanical unit. It is connected to a management controller 4 outside the semiconductor manufacturing apparatus 1 by a network 3 composed of a LAN, a WAN, the Internet and the like.
[0014]
In the semiconductor manufacturing apparatus 1, various sensors (not shown) for monitoring the process control and the operation state of the mechanism unit and detecting an abnormality are installed. The controller information is taken in by the controller 2, and when a value exceeding the range of the normal value set in the controller 2 is detected from the sensor information, the controller 2 determines that the operation is abnormal and generates an alarm. Here, the alarm includes alarm information.
[0015]
Failures to be judged as operation abnormalities are basically all failures detected by the control device controller 2. Obstacles include, for example, when the temperature or pressure of the processing furnace rises above a predetermined value, when the substrate falls during transport of the substrate, or when the gate valve to be closed is open. When the process is executed, when the robot during the substrate transfer stops abnormally, when the heater power is shut off due to water leakage from the radiator section, and the like. When these failures occur, the controller controller 2 generates an alarm, and notifies the management controller 4 via the network 3 of the alarm at the same time as the occurrence of the alarm. Here, the alarm information is information including at least the name of the device in which the alarm is generated, the alarm ID for specifying the alarm, and the alarm name, and the recipe executed when the alarm is generated in the information. In some cases, a name, a step name when an alarm occurs, and the like may be added.
[0016]
[Management Controller]
The management controller 4 is installed outside the semiconductor manufacturing apparatus for monitoring and managing a group of the semiconductor manufacturing apparatuses 1, and is connected to the plurality of control controllers 2 via the network 3. The management controller 4 and the network 3 are connected by, for example, Ethernet (registered trademark) or ISDN. The management controller 4 collects information such as various types of monitor data collected by the controller 2 and information such as alarms notified from the controller 2 in the recording medium 5 and manages the information. Further, the management controller 4 is connected to a network device other than a dedicated device-to-device network via a LAN or the Internet. The network device has a web server function, and has a function of browsing semiconductor manufacturing apparatus information and executing a command to the semiconductor manufacturing apparatus 1 using a web and a browser. Further, the management controller 4 has a function of automatically notifying the occurrence of a failure to a preset e-mail address or pager number when an alarm is notified from the control controller 2.
[0017]
[Fault analysis notification device]
The failure analysis and notification device 6 is provided in the control controller 2 described above. The failure analysis and notification device 6 has a function of analyzing a plurality of failures and notifying them as a single failure. As described above, even when a certain failure occurs in the semiconductor manufacturing apparatus 1, the controller 2 may generate a plurality of alarms due to the failure as a basic factor. In order to eliminate the alarm congestion, a failure analysis and notification device 6 is provided in the controller 2. As shown in FIG. 1, the failure analysis and notification device 6 includes a determination unit 10 for determining a failure occurrence status, a pattern forming unit 11 for forming a detected failure occurrence pattern, a database 12 as a registration unit, a comparison unit 13, and a narrowing unit. 14.
[0018]
The judging means 10 judges whether or not the alarms are a plurality of alarms which have occurred simultaneously or within a specific time when the controller 2 monitors the fault occurrence status of the semiconductor manufacturing equipment group and generates an alarm. When it is determined that a plurality of alarms have occurred simultaneously or within a specific time, the alarm information of the plurality of alarms is sent to the pattern forming means 11. Here, the reason for limiting the occurrence status of a plurality of alarms to simultaneous occurrence and occurrence within a specific time is that when a plurality of alarms are caused by a failure such as a certain failure or abnormality, they may occur simultaneously. This is because, if the time at which many or multiple alarms occur exceeds a specific time, it is considered that not a single failure but rather a plurality of failures.
[0019]
The pattern forming unit 11 forms (analyzes) a plurality of pieces of alarm information generated simultaneously or within a specific time to form a detection failure occurrence pattern. Here, the patterning is, for example, an operation of extracting a device name, an alarm ID, and an alarm name from each piece of detected alarm information, and further extracting a target device, a target recipe name, and a target step name. The detection failure occurrence pattern is a combination of each element of a plurality of pieces of alarm information extracted by the above operation. Patterning a plurality of pieces of alarm information to form a detection failure occurrence pattern is performed by comparing a detection failure occurrence pattern with a registered failure occurrence pattern registered in advance and belonging to a category (single failure failure) to which the pattern belongs. ).
[0020]
The database 12 is a recording medium in which a plurality of alarm information patterns that occur simultaneously or within a specific time due to one failure as a basic factor are previously associated with the failure as a basic factor and registered as a registered failure occurrence pattern. is there. That is, a plurality of fault occurrence patterns categorized for each fault cause are registered in the database 12. Here, the registration failure occurrence pattern is basically the same as the detection failure occurrence pattern, and differs in that it is registered in advance and is categorized for each basic factor. Although it is preferable that the number of faults as a basic factor is one, the number may be two or more as long as the number of actually generated alarms can be narrowed down.
[0021]
The comparing means 13 compares the detected fault occurrence pattern output from the pattern forming means 11 with the registered fault occurrence pattern read from the database 12, and outputs a match output when a comparison match is obtained. Is done.
[0022]
When a comparison match output is obtained from the comparing means 13, the narrowing means 14 narrows an alarm which is a basic factor from a plurality of alarms based on the fault associated with the registered fault occurrence pattern at that time, and the narrowed alarm To the management controller 4 together with the alarm information.
[0023]
The operation in the configuration of the semiconductor manufacturing apparatus system as described above will be described. In the configuration as described above, a plurality of alarm information patterns that occur simultaneously or within a specific time due to one failure as a basic factor are registered in the database 12 in advance in association with the failure as a basic factor. Are registered as multiple.
[0024]
When the semiconductor manufacturing apparatus operates, the controller controller 2 monitors the occurrence of an alarm as shown in FIG. 2 (step 101). When a failure occurs and the sensor detects an abnormality, an alarm is generated. When an alarm is detected, the determination means 10 of the failure analysis and notification device 6 determines whether the failure is a plurality of predetermined alarms that have occurred simultaneously or within a specific time (step 103). If it is determined that the alarm is not a predetermined plurality of alarms, the detected alarm is notified (normal notification) to the management controller 4 as it is (step 113).
[0025]
When it is determined that a plurality of alarms have occurred at the same time or within a specific time, the plurality of pieces of alarm information are input to the pattern forming unit 11, and a detection failure occurrence pattern is formed by the pattern forming unit 11 ( Step 105). This detected failure occurrence pattern is compared with the registered failure occurrence pattern read from the database 12 by the comparing means 13 to determine whether or not a comparison match is obtained (step 107). If a comparison match is not obtained by the comparing means 13, the detected alarm is reported to the management controller 4 as it is as a single failure or a pattern not included in the registered patterns (step 113). .
[0026]
When a comparison match is obtained by the comparing means 13, a failure which is a basic factor of the detected fault occurrence pattern is narrowed down from the registered fault occurrence patterns obtained by the narrowing means 14 (step 109). Thus, the failure analysis and notification device 6 specifies an alarm that is a basic cause from among the plurality of alarms, and notifies the management controller 4 of an alarm as a single true failure cause (step 111).
[0027]
As described above, according to the embodiment, a failure occurrence pattern categorized for each failure factor of the semiconductor manufacturing apparatus 1 is managed in the database 12, and a plurality of alarms are simultaneously detected from the monitoring target apparatus by the controller 2. In this case, this is configured to be patterned, collated with a registered failure occurrence pattern, and when they match, the causes of a plurality of alarms are recognized, and notification is made as a single alarm. Therefore, even when a plurality of alarms are issued, the number of fault occurrence notifications from the controller 2 to the equipment engineer is reduced or reduced, so that important information is extracted from the information to determine the true fault factor. You can find it quickly and take quick troubleshooting measures.
[0028]
Further, even if a plurality of alarms are simultaneously generated from various sensors and sub-controllers due to one failure, it is possible to prevent the control controller that controls them from detecting them one by one. Further, by predicting the cause of the failure and adding items to be checked to the notification of the occurrence of the failure, it is possible to more quickly grasp the state of the failure and analyze the cause of the failure.
[0029]
In the semiconductor manufacturing apparatus system, the management controller 4 transmits management data to the semiconductor manufacturing apparatus 1 at the time of management such as remote maintenance of the semiconductor manufacturing apparatus 1 based on a notification from the semiconductor manufacturing apparatus 1 of the narrowed-down alarm. Then, management such as remote maintenance is performed. At that time, since the management data transmitted from the management controller 4 to the semiconductor manufacturing apparatus 1 is created based on the narrowed-down alarm information, the management data is created based on a plurality of alarm information that has not been narrowed down. Thus, the amount of management data can be significantly reduced. Therefore, resources of the recording medium 5 of the management controller 4 can be reduced.
[0030]
In the embodiment, the case where the operation status is the failure occurrence status has been described. However, the present invention can be applied to the case where the progress status of the processing operation is grasped. In the case of the progress of the processing operation, a sensor for detecting the progress of the transport process or the like is provided instead of the sensor for detecting the abnormality. In addition, sensors for both abnormality detection and progress status detection may be installed to grasp both the fault occurrence status and the progress status.
[0031]
Further, in the embodiment, when the registration failure occurrence pattern is registered using the web browser, the registration failure pattern is directly registered in the database 12, but once registered in the recording medium 5 in the management controller 4, May be transferred to the database 12. Further, although the judging means 10 is configured separately from the pattern forming means 11, the judging means 10 may be included in the pattern forming means 11.
[0032]
【Example】
Next, a description will be given of an embodiment of a procedure for registering a fault occurrence pattern in the database described above.
[0033]
[Problem pattern registration procedure]
There are the following three methods for registering registration failure occurrence patterns managed by the database 12.
A. Manually register from all the alarm information registered in the controller 2.
B. The controller 2 manually registers the alarm information based on the actually detected alarm information.
C. The controller 2 automatically registers from the alarm information actually detected.
[0034]
Hereinafter, each of them will be described.
A. Manual registration from all alarm information registered in the controller 2
This method is a case where a new registration is made in a state where no alarm has occurred in the past and present. On the web browser of the management controller 4, a list of alarms registered in advance in each control controller 2 is displayed, and the user selects and registers the alarm from the list. The selective registration is performed in advance based on data in which a plurality of alarm information patterns that occur simultaneously or within a specific time due to a specific failure as a basic factor are associated with an operation as a basic factor. FIG. 4 shows a failure occurrence pattern new registration screen 20. This screen shows a plurality of input screens and selection screens as a user interface. If an item is selected and input, the pattern information and the selected data are sent to the database 12 and recorded as a registration failure occurrence pattern.
[0035]
The registered failure occurrence pattern registered in the database 12 is managed by a unique occurrence pattern name, and the interface is composed of a device name, a plurality of alarms, and their occurrence intervals, and optionally further includes a target device, a target recipe name, and a target step name. Can be added. The registration failure occurrence pattern is associated with the occurrence pattern name. The registration failure occurrence pattern is registered by a series of operations (1) to (5) described below.
[0036]
(1) Enter the occurrence pattern name in the occurrence pattern name input field 21. The input generation pattern name becomes a unique pattern name. When a plurality of failures (such as an increase in the temperature of the processing furnace, an increase in the pressure of the processing furnace, a drop in the substrate, a malfunction of the gate valve, an abnormal stop of the robot, a leak in the radiator, etc.) occur, the unique pattern name is changed from the name. It is preferable to name them so that the basic factors of the obstacle can be immediately understood. For example, if the single true cause of the plurality of faults is water leakage from the radiator unit, the occurrence pattern name is "radiator unit water leakage".
[0037]
The device name column 22 displays the name of the semiconductor manufacturing device. For example, it is a model number for specifying a model.
[0038]
(2) In the registered alarm list display column 23, a plurality of alarms are selected from all the alarms (ALARM001 to ALARM999 (alarm name)) generated in the apparatus, and a combination of the selected alarms is set as an occurrence pattern. By operating the scroll bar 24 next to the display column 23, all alarms can be displayed. In the illustrated example, the alarms (ALARM001, ALARM002, and ALARM004) having three alarm names are selected, and the circles preceding the selected alarm IDs (numbers 001 to 999) are converted to the circles. , These are in the selected state.
[0039]
Here, the meaning of selecting three alarms as described above will be described. For example, it is assumed from empirical rules that the following three alarms occur in a chain due to "water leakage from the radiator section" as a basic factor.
(A) Heater power supply box AC200V blackout (ALARM001)
(B) Water leakage from the radiator (ALARM002)
(C) Heating failure of processing furnace (ALARM004)
In this case, a combination of three alarms (a) to (c) is selected as an occurrence pattern to be associated with the occurrence pattern name “radiator section water leakage”.
[0040]
(3) In the occurrence interval input column 25, the time from when an alarm is received until pattern matching is set as an occurrence interval. When a plurality of alarms are actually generated within the set generation interval, patterns relating to the plurality of alarms are to be compared.
[0041]
(4) The target device to be monitored is selected with the radio button 26. In this case, all devices connected to the network between devices “all”, or devices with the same model type and the same film type “same device”, devices displayed on the new fault occurrence pattern registration screen L (device name column) (Only the device displayed in) is selected from “this device only”. Here, "all" is selected.
[0042]
(5) In the target recipe name input field 27, a recipe name for pattern matching is set. If the target recipe name input field 27 is not set, pattern matching is performed for all recipes. In the target step name input 28 field, a step name for pattern matching can be set. If the target step name input 28 column is not set, pattern matching is performed in all steps. When a target recipe name and a target step name are set, it is checked at the time of registration whether the recipe name and the step name exist in the target device.
[0043]
(6) When all settings are completed on the new fault occurrence pattern screen 20, the OK button 29 is pressed to complete registration of the registered fault occurrence pattern in the database 12. When the registration is completed, the failure identified by the occurrence pattern name becomes the basic factor, and the pattern of multiple alarm information that occurs simultaneously or within a specific time period is associated with the failure that is the basic factor. The pattern is registered in the database 12.
[0044]
By repeating the above operations (1) to (6), a plurality of registered failure occurrence patterns that can be subjected to pattern matching are registered in the database 12 in advance.
[0045]
B. Manual registration from the alarm information actually detected by the controller
This method is for a case where an operator registers from an alarm that has occurred in the past.
A list of alarms generated in the semiconductor manufacturing apparatus 1 is displayed on a web browser, and the user selects and registers them. FIG. 5 shows a failure occurrence pattern history registration screen 30.
[0046]
In the history alarm list display column 33, the device alarm information is displayed in a list in order of date and time, and by selecting a plurality of alarms from among them, a registered failure occurrence pattern corresponding to the occurrence pattern name is registered.
[0047]
The other interfaces of the failure occurrence pattern history registration screen 30, which are the occurrence pattern name input field 31, the device name field 32, the occurrence interval input field 35, and the radio button 36 of the target device, are almost the same as those in the above-described method A. The selection of the target recipe and the target step can be selected by radio buttons 37 and 38, respectively. When all settings are completed on the failure occurrence history registration screen 30, the OK button 39 is pressed to complete registration of the registered failure occurrence pattern. A plurality of registrations are performed by repeating this series of registration work.
[0048]
C. Automatic registration based on alarm information actually detected by the controller
This is a method in which a user sets conditions relating to pattern registration in advance, and automatically recognizes and registers as a registration failure occurrence pattern when the conditions are met. In the automatic registration, after all factors of the failure are removed from the device, the operator activates the device, and then, when an alarm meeting the above condition occurs, recognizes the alarm as a registration failure occurrence pattern (duplicate). Alarm ID, recipe, step, etc.). However, the operator needs to input the occurrence pattern name. The pattern is recognized by the ID on the database 12 similarly to the alarm, and is recognized as, for example, the pattern 001 when no occurrence pattern name is input. Conditions for pattern recognition include, for example, the following.
[0049]
{Circle around (1)} [Three consecutive alarms] in which [three consecutive alarms] occur in [target device 1] [recipe A] [step α] are recognized.
{Circle around (2)} A case where [five consecutive alarms] or [four consecutive alarms] occurs is recognized [four times].
In the case of (1) described above, in order to recognize three consecutive unlimited alarm occurrences as a pattern in a specific step of a specific recipe in a specific device, the state is reproduced three times. Things. In the case of (2), the device, the recipe, and the step are not specified, and five alarm occurrences are used as a pattern, and when the pattern is reproduced four times, the pattern is registered.
[0050]
[Failure pattern recognition method]
Next, how a detected failure occurrence pattern formed based on an actually generated alarm is recognized will be described.
[0051]
FIG. 6 shows an example of detection failure occurrence pattern recognition. The matching method of the detected failure occurrence pattern may be a case where the alarm of the detected failure occurrence pattern occurs continuously in the order of the alarm of the registered failure occurrence pattern registered in the database 12, or may be registered. The alarm may be generated in any order. In the illustrated example, the registered failure occurrence patterns registered in the database 12 are only two, that is, the occurrence pattern A and the occurrence pattern B for convenience. The detection failure occurrence pattern that actually occurred is A. The registered content of the occurrence pattern A is that the alarm IDs 001, 002, and 004 occur at the same time within 10 seconds, and the registered content of the occurrence pattern B is that the 001, 002, and 003 occur at the same time within 10 seconds. is there. Therefore, when the actually detected detected failure occurrence pattern is compared with the registered failure occurrence pattern A or B registered in the database 12, the detected failure occurrence pattern A matches the registered failure occurrence pattern A. With the unique “occurrence pattern name” associated with the coincident registered failure occurrence pattern A, the true cause of the failure can be quickly identified, and the failure notification message can be unified. In addition, information necessary to support analysis of a cause predicted from a failure occurrence pattern, an item to be checked, and the like, such as the contents of individual alarms and recipe execution data at that time, are added to the unified failure notification message. be able to.
[0052]
【The invention's effect】
According to the present invention, even when information relating to a plurality of operation situations is generated due to a certain operation as a basic factor, the true operation situation can be quickly grasped.
[Brief description of the drawings]
FIG. 1 is a block diagram illustrating a control system (failure analysis notification device) of a semiconductor manufacturing apparatus system according to an embodiment.
FIG. 2 is a flowchart of the failure analysis notification device according to the embodiment.
FIG. 3 is a configuration diagram of a semiconductor manufacturing apparatus system according to an embodiment.
FIG. 4 is an explanatory diagram showing a failure occurrence pattern new registration screen according to the embodiment;
FIG. 5 is an explanatory diagram showing a failure occurrence pattern history registration screen according to the embodiment.
FIG. 6 is an explanatory diagram showing an example of generated pattern collation according to the embodiment;
[Explanation of symbols]
1 Semiconductor manufacturing equipment
2 Control controller
4 Management controller
6 Failure analysis notification device
10 Judgment means
11 Pattern forming means
12 Database (Registration means)
13 Comparison means
14 Narrowing means

Claims (1)

When an operation such as a failure occurs in the device, in a semiconductor manufacturing device that generates information related to a plurality of operation states simultaneously or within a specific time as the operation becomes a basic factor,
Registration means for registering, as a registered operation status pattern, a pattern of information related to a plurality of operation situations in which the operation becomes a basic factor and occurs simultaneously or within a specific time in advance in association with the operation as the basic factor,
Pattern forming means for patterning information relating to a plurality of operation situations actually occurring simultaneously or within a specific time to form a detection operation situation pattern,
Comparing means for comparing the detected operation situation pattern output from the pattern forming means with the registered operation situation pattern read from the registration means,
Means for narrowing down information relating to a plurality of operation situations that have actually occurred from the registered operation situation pattern for which a comparison match has been obtained by the comparison means to information relating to an operation situation of an operation that is a basic factor associated with the registered operation situation pattern Wherein the information on the operation status narrowed down by the narrowing means is notified.

Images