JPH07117952B2 - Failure diagnostic equipment for manufacturing equipment - Google Patents

Description

The present invention relates to a failure diagnosis device for manufacturing equipment, and more particularly to a failure diagnosis device suitable for installation in a manufacturing line having a plurality of manufacturing equipment.

[Conventional technology]

As a fault diagnosis method for manufacturing equipment, "Measurement and Control" magazine, Vo
The method described in the paper entitled "Equipment Diagnosis Technology and Safety" by Kageo Akizuki in l.24, No.4 (April 1985) is widely used. In this method, the amount of equipment considered to be involved in an abnormality or failure is measured, the measured quantity is processed, and an amount serving as an index of abnormality, failure, or deterioration is obtained to determine the present situation.

[Problems to be solved by the invention]

In the above-mentioned conventional technology, for each location of individual manufacturing equipment,
The measuring device and the processing device for the measured amount must be individually prepared, and the cost required for that is not necessarily smaller than the cost required for the equipment itself. In addition, it does not determine or instruct post-measures.

In particular, in line production, where production is performed continuously using multiple equipment, it is important to reduce downtime by shortening the time required for subsequent measures, as non-operation equipment stops the entire line. .

In a production line having a plurality of manufacturing equipments, there are cases where the equipments need to be repaired or adjusted due to the failure of the equipments or the change of the operating state with time. When using dozens of types of equipment repeatedly, such as a semiconductor device manufacturing line, to complete product processing hundreds or tens of times, a small downtime due to failure or adjustment of individual equipment is This causes a significant decrease in productivity of the entire production line. Moreover, since it is almost impossible to regenerate the product processed by the faulty equipment, many defective products are generated.

For this reason, in the production line, preventive maintenance is performed to maintain the equipment before the failure, but in the case of semiconductor equipment manufacturing equipment that requires delicate adjustment of the element device that has an extremely clean atmosphere and complicated Requires a few hours to a dozen hours for the maintenance itself, and it is only necessary to inspect and confirm every certain unit of processing, and the entire line is temporarily stopped, dismantled, and readjusted. If not, it is the actual situation.

As described above, in the case of performing the post-repair maintenance in which the repair is started from the time when the failure occurs, the following information is required to perform the repair promptly.

(1) Information related to past failures and countermeasure examples for identifying failure factors and failure points because the entire equipment is composed of complicated element devices.

(2) In order to plan repair methods and repair procedures, detailed equipment configurations, causal relationships regarding failure states and causes, and frequency of occurrence.

Skilled engineers memorize the above information as expert knowledge from past experience, so information such as failure status and product processing accuracy can be used as clues to determine the cause of failure based on the stored causal relationship. The cause, location, countermeasures, etc. can be determined. However, the unskilled person has a lack of experience, so that there is a problem that quick response cannot be made.

An object of the present invention is to provide a failure diagnosis apparatus for manufacturing equipment, which enables any person, regardless of expert or non-expert, to promptly deal with a failure.

[Means for solving problems]

In the case of a semiconductor device manufacturing facility where subtle changes in elemental devices cause product defects, product defects often occur.
An ambiguous failure condition called "unwell" frequently occurs. As described above, in a state where it is difficult to specify the failure mode, there is a problem that it is difficult to find the failure location if the equipment itself is monitored.

A skilled engineer finds such a failure state by the following method.

(1) Recall the product processing state such as the change and tendency of the product processing size and accuracy with time, the size of the dimensional accuracy variation, and the device name where such defects frequently occur, divided into several characteristic patterns. To do.

(2) Next, the failure location is searched based on the knowledge of the failure state and the factor thereof (causal relationship of the failure) that can best explain all of the recalled patterns. When searching for a failure location, the inspection is performed in order from the element devices that are predicted to have the greatest impact on the occurrence of the failure.

In order for even an unskilled person to be able to diagnose a failure by such a method, these product states that can be observed by product inspection are divided into some characteristic patterns, and a failure that best matches all of the patterns is detected. It is necessary to be able to search the causal relationship of.

Therefore, in the present invention, in order to achieve the above-mentioned object, an inspection means for inspecting a plurality of product states of each product manufactured in a manufacturing facility in a time-series manner, and a time-series change of the plurality of product states, respectively. Means for plotting and graphing each based on the cumulative sum method, and superimposing V-shaped masks on the respective graphs so that their vertices are located in the time axis advance direction of each graph, and when the graph intersects the V-shaped mask, The correspondence between the means for determining that an abnormality has occurred in the manufacturing equipment at the time of intersection, the time-dependent change state of the element devices constituting the manufacturing equipment, and the set indicating an abnormal change among the plurality of product states is shown. A failure diagnosis apparatus for manufacturing equipment is configured by means for identifying the cause of the abnormality that occurred at the crossing time from the past pattern classification and instructing maintenance measures for the cause.

[Action]

In this device, by inputting an abnormal state related to the processing size and accuracy of the product observed at a certain point, all the characteristic patterns that match the state are retrieved from the storage device, and the equipment failure corresponding to this pattern is also taken out. The state is searched, and all the basic events related to the failure of the element device necessary for the occurrence of the failure state are shown as failure locations in descending order of the degree of influence. This not only allows unskilled personnel who have insufficient empirical knowledge about failure diagnosis to identify the failure point quickly and accurately, but also to formulate a repair plan by referring to the past failure point and its failure countermeasure history. Can be implemented.

〔Example〕

 An embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 is a configuration diagram of a failure diagnosis device according to an embodiment of the present invention. The failure diagnosis device is a central processing unit (CPU) 1,
It comprises a maintenance database 2 and an input / output device 3. A process database 4 is also connected to the central processing unit 1 so that each manufacturing facility A and the central processing unit 1 can send and receive information data via an online control unit 5.

Such a failure diagnosing device is provided in a manufacturing line having a plurality of manufacturing facilities A1, ..., An as illustrated in FIG. In this manufacturing line, the material B is processed by the manufacturing equipment A, and whether or not this processing is normally performed is inspected by the inspection device C, and the process data such as the processing size and the accuracy is obtained by the online control device 5 (see FIG. 1). ) To the central processing unit 1.

The maintenance database 2 shown in FIG. 1 is a history of past failure points in the manufacturing facilities A1 to An, failure conditions, causes, repair measures for the failures, processing dimensions of products processed in the failed state, and measured values of accuracy. Is a storage device for accumulating a failure location, a failure condition, a cause, a processing accuracy in the failure condition, and a causal relationship of repair contents to be taken.

The central processing unit 1 uses the keywords related to the failure state, the possible cause of the failure, and the accuracy of the processed product so that the worker who discovers the failure state can retrieve the series of causal relationships stored in the storage device 2. When input from the input / output device 3, one of the causal relations of the failure that best matches the keyword is selected from the data in the storage device 2 and the input / output device 3 is selected.
Is displayed on the screen to instruct a countermeasure against the failure, and a history of the location of the repair and the content of the repair actually performed by the operator based on the instruction is stored in the storage device 2 together with a series of causal relationships.

FIG. 3 is an example of a manufacturing facility targeted for failure diagnosis, and is a conceptual diagram of a plasma etching apparatus used for manufacturing a semiconductor integrated circuit.

The plasma etching apparatus etches a wafer-shaped Si single crystal surface along a resist pattern drawn by a photolithography technique. Etching
It is performed by the radicals excited by the high frequency discharge plasma chemically reacting with Si.

11 is a wafer-shaped Si single crystal that is a processing material, 12 is a storage container for a gas that is a source of radicals that react with Si, 13 is an electrode that applies a high frequency to this gas to generate radicals, and 14 is a high frequency power source An apparatus, 15 is a vacuum pump for exhausting a gaseous reactant of Si and radicals to the outside of the etching chamber, and 16 is a trap for trapping the reactant when exhausting the reactant to the outside of the chamber.

Since the amount of etching is proportional to the reaction time of Si and radicals, the reaction is performed for a certain time to obtain the required etching dimensional accuracy. However, the reaction time for etching a certain amount fluctuates due to changes over time in the equipment state, such as reactants adhering to the inner wall of the chamber, the degree of vacuum decreasing, and the density of radicals decreasing due to electrode deterioration. To do. In the worst case, the etching may not proceed and the product may be defective.

In this case, since the progress of etching and the degree of contamination of the inner wall of the chamber cannot be directly measured, the clue to find the cause of failure must depend on the processing accuracy and electrical characteristics of the product.

Such changes in the processing accuracy and electrical characteristics of the product can be associated with the above-described changes in the equipment state over time. Conventionally, maintenance and adjustment of the device have been performed by such a method.

Fig. 4 shows that the time until a certain amount of etching is performed is adjusted by adjusting the vacuum level and cleaning the inner wall of the chamber.
It shows how it changes. In FIG. 4, the oil exchange is to replace the lubricating oil of the vacuum pump 15, the vacuum degree adjustment is to adjust the vacuum degree by adjusting the pressure gauge, and the complete cleaning is to clean the inner wall of the chamber. Say that. As shown in FIG. 4, it can be seen that the cumulative sum of etching times is reduced by changing the oil, adjusting the degree of vacuum, and completely sweeping. However, when the changes over time in the equipment state occur in a complex manner, it is unclear what changes in the product state should correspond to changes over time in the equipment state.

It is difficult to associate any one change in the product state such as variation in processing accuracy or surface resistance value with a change over time in a complex equipment state. However, it is possible to associate a set of changes in a plurality of product states with a composite change in device state over time, for example, by the following method. First, the time series changes of all product states that can be measured by the inspection device are plotted for each equipment used according to the cumulative sum method. The cumulative sum method plots the cumulative sum of the difference between the measured value representing the product state and a constant value (reference value) against time, and by this cumulative sum method, for example, as shown in FIG. A graph is obtained.

When a V-shaped mask having a certain angle and length is superposed on the curve connecting the plotted points, it can be interpreted that an abnormal state change of equipment occurs at the time when they intersect. The V-shaped mask is superimposed on each plotted point with a leading distance d. Here, the shape of the V-shaped mask is determined by the number of samples of the measurement value and the required prediction accuracy, but the standard of the angle θ and the leading distance d is determined as follows.

However, δ is the standard deviation of the average value of the sample, and α ₀ is the risk rate of the test.

Whenever a serious abnormality of these facilities occurs at the manufacturing site,
Shut down the facility and perform ex-post maintenance. On the contrary, the effect of conservation is that the plot curve of the measured value according to the cumulative sum method is V
It can be confirmed by entering the character mask. At this time, the history of the contents of the maintenance and the element device targeted for the maintenance is recorded.

Next, this record and the time of the plot curve of the measured value by the cumulative sum method are matched and overlapped. As a result, among the measures for maintenance performed after the time when the intersection of the plot curve and the V-shaped mask appears, the measure that is effective in that the plot curve of the measured value indicating the product state falls within the V-shaped mask. Can be extracted.

Since each countermeasure is associated with the abnormal state of the element device on a one-to-one basis, it is possible to associate a plurality of abnormal changes in the product state with a set of device states that cause the changes.

In this way, it is possible to grasp the state change, that is, the failure state of the element device that affects some changes in the product state. Further, if the set of basic events that cause such a failure state of the element device is known, the location of the failure can be identified.

A fault tree analysis method is a method for analyzing the relationship between such a failure state and the basic event that causes it.

FIG. 6 is an example of a fault tree of a vacuum pump which is a component device of the above-mentioned etching apparatus. As a cause of the abnormal change in the product state in which the etching amount is insufficient and the gate oxide film thickness is increased, it is conceivable that the vacuum degree is reduced and the radical concentration is reduced. At this time, it is considered that the decrease in the degree of vacuum is caused by a basic phenomenon such as a defective rotor of the vacuum pump system and a defective leak valve, and the failure probabilities are different. The central processing unit outputs to the input / output device a display instructing inspection based on the fault tree stored in the maintenance database and in descending order of failure probability.

In addition, the probability is calculated based on the history of countermeasures for each failure countermeasure, and the value is updated.

As described above, in the present embodiment, the set of basic events characteristic of the product state is patterned by the cumulative sum method, and the causal relationship between the basic event of the equipment state and the failure is one of the failure factor prediction methods. By the fault tree analysis method that is
Each is statistically processed and accumulated. Therefore, even an unskilled person can output an instruction to easily and promptly deal with a failure.

〔The invention's effect〕

According to the present invention, when a failure occurs in the manufacturing equipment in the manufacturing line, even by a non-specialist of each manufacturing equipment, by inputting the recognized equipment state and the factor of the possible failure,
Since it is possible to know the location of the failure and the countermeasure, it is possible to save the labor of the subsequent countermeasure and reduce the downtime due to the failure.

[Brief description of drawings]

FIG. 1 is a configuration diagram of a failure diagnosis apparatus according to an embodiment of the present invention, FIG. 2 is a conceptual diagram of a production line having a plurality of production equipments, and FIG. 3 is a schematic of a plasma etching apparatus which is an example of the production equipments. 4 is a correlation diagram of maintenance work and etching time, FIG. 5 is an explanatory diagram of cumulative sum management using a V-shaped mask, and FIG. 6 is a fault tree diagram of a vacuum pump. A1 to An: Manufacturing equipment, 1 Central processing unit, 2 Maintenance database, 3 Input / output device.

Claims (1)

[Claims] 1. An inspection means for inspecting a plurality of product states of each product manufactured at a manufacturing facility in a time series, and a time series change of the plurality of product states are plotted based on a cumulative sum method. A means for making a graph and a V-shaped mask superimposed on each graph so that the apex thereof is located in the direction of time axis advancement of each graph, and when the graph intersects the V-shaped mask, the manufacturing equipment is abnormal at the intersection time. From the past pattern classification showing the correspondence between the means for judging the occurrence of a change in the manufacturing apparatus, the time-dependent change state of the element devices constituting the manufacturing equipment, and the set showing an abnormal change among the plurality of product states at the crossing time. A failure diagnosis device for manufacturing equipment, comprising: means for specifying a cause of an abnormality that has occurred and instructing a maintenance measure against the cause.