Data processing: measuring – calibrating – or testing – Measurement system in a specific environment – Quality evaluation
Reexamination Certificate
2001-01-16
2003-07-01
Barlow, John (Department: 2863)
Data processing: measuring, calibrating, or testing
Measurement system in a specific environment
Quality evaluation
Reexamination Certificate
active
06587801
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention generally relates to abnormality-cause identifying apparatuses and methods and, more particularly, to an abnormality-cause identifying apparatus and an abnormality-cause identifying method for identifying a manufacturing apparatus or the like that causes an abnormality.
2. Description of the Related Art
FIG. 8
is a flowchart of steps constituting a conventional abnormality-cause identifying method. In
FIG. 8
, step ST
1
is carried out to set conditions (search criteria) for retrieving quality data and history data. Step ST
2
is performed to retrieve from a database the quality data and history data conforming to the retrieving conditions. Step ST
3
is executed to compute a mean value of quality data relating to each of manufacturing apparatuses constituting a production line. Step ST
4
is effected to compare the mean value of quality data relating to each of the manufacturing apparatuses with the currently stored applicable normal-state data (data in effect when no abnormality is present). If the mean value is judged to significantly deviate from the normal-state data, it is estimated that the corresponding manufacturing apparatus causes the abnormality.
Given below is a description of the operation involved in abnormality-cause identification according to the related art.
For example, a semiconductor device is fabricated using a plurality of manufacturing apparatuses in a plurality of processes as shown in FIG.
9
.
The products are inspected in intermediate and final stages of the manufacturing. If an abnormality is found in any of the products, manufacturing apparatuses that cause the abnormality are identified.
More specifically, step ST
1
is first carried out to set conditions for retrieving quality data (e.g., data subject to computation in an intermediate inspection process and denoting a fraction defective of the manufacturing apparatuses) and history data relating to each manufacturing process (e.g. data indicative of processing results of the manufacturing apparatuses) such that the retrieved data are used to identify an apparatus suspected to have caused the abnormality, the quality data and history data being collected in diverse manufacturing processes and stored in a database. The retrieving conditions illustratively include products to be manufactured, processes to be inspected, periods to be covered, applicable quality items, history of applicable processes, and processing units involved.
In step ST
2
, the quality data and history data conforming to the established conditions are retrieved from the database. For example, if a part or all of the processes in
FIG. 9
are suspected, then a part or all of the processes may be designated as the retrieving conditions. Then, the quality data and history data relating to the manufacturing apparatuses involved in the designated part or all of the processes are retrieved.
Step ST
3
is then executed to compute a mean value of the quality data relating to each of the manufacturing apparatuses involved in the designated part or all of the processes. Illustratively, if the manufacturing apparatuses A and B are involved, a mean value of the quality data relating to the manufacturing apparatus A and a mean value of the quality data relating to the manufacturing apparatus B are computed.
Finally, step ST
4
is carried out to compare the mean value of the quality data relating to each manufacturing apparatus with the currently stored normal-state data. Illustratively, where the manufacturing apparatuses A and B are involved in part or all of the manufacturing processes, a comparison is made between the mean value of the quality data relating to the manufacturing apparatus A and the normal-state data relevant to the apparatus A, and another comparison is made between the mean value of the quality data relating to the manufacturing apparatus B and the normal-state data relevant to the apparatus B.
If the comparison yields a difference of, for example, at least 20 percent between the normal-state data and the quality data mean value of a given manufacturing apparatus, that apparatus is identified as having caused the abnormality. The outcome of the identification is printed or otherwise displayed.
After the suspected manufacturing apparatus has been identified as described, the user determines whether or not the abnormality has indeed occurred on the basis of various history data relating to the apparatus in question.
Since the conventional abnormality-cause identifying method is constituted as outlined above, a small number of quality data items on manufacturing apparatuses tend to lower the accuracy of quality data mean values serving as a pointer to an apparatus suspected of the abnormality (hereinafter, such an apparatus will also be referred to as a defective apparatus). Where a defective apparatus cannot be identified accurately, there is a likelihood that manufacturing apparatuses that are not the cause of the abnormality may be processed for determination of the cause of abnormality in advance of the defective apparatus, thus making it difficult to efficiently identify the apparatus that caused the abnormality.
FIG. 10
is a graphic representation of an exemplary fraction defective distribution of manufacturing apparatuses. In the example of
FIG. 10
, the manufacturing apparatus having a fraction defective of about 3.5 percent is most likely to caused the abnormality. In that case, the apparatus with the fraction defective of 3.5 percent should be identified as causing the abnormality. However, as the mean values of quality data decline in their accuracy, a manufacturing apparatus with a fraction defective of about 2.3 percent may be mistakenly identified as having caused the abnormality.
SUMMARY OF THE INVENTION
It is therefore an object of the present invention to overcome the above and other deficiencies of the prior art and, more particularly, to provide an abnormality-cause identifying apparatus and an abnormality-cause identifying method for efficiently identifying an apparatus causing an abnormality.
The aforementioned objects can be achieved by an abnormality-cause identifying apparatus comprising: setting means for setting conditions for retrieving quality data; retrieving means for retrieving quality data conforming to the retrieving conditions set by the setting means; and computing means for computing mean values and standard deviations of the conforming quality data retrieved by the retrieving means, and for computing a likelihood of abnormality of each of candidates that could have caused an abnormality, on the basis of the mean values and standard deviations.
The abnormality-cause identifying apparatus may further comprise determining means for subjecting the candidates to determination as to abnormality, in a descending order of magnitude of likelihood of abnormality.
If an upper limit and a lower limit of quality data are set by the setting means, then the retrieving means may retrieve, from the conforming quality data, quality data in excess of the upper limit and quality data below the lower limit before outputting the retrieved quality data.
The abnormality-cause identifying apparatus may further comprise test value computing means for computing a test value and a fraction defective of each of the candidates, on the basis of the mean values and standard deviations of the quality data, and for computing the likelihood of abnormality of the candidates, on the basis of the test value and the fraction defective.
The aforementioned objects can be achieved by an abnormality-cause identifying method comprising the steps of: setting conditions for retrieving quality data; retrieving quality data conforming to the retrieving conditions; and computing mean values and standard deviations of the conforming quality data, and computing a likelihood of abnormality of each of candidates that could have caused an abnormality, on the basis of the mean values and standard deviations.
The abnormality-cause identifying method may further
Barlow John
Burns Doane , Swecker, Mathis LLP
Mitsubishi Denki & Kabushiki Kaisha
Pretlow Demetrius
LandOfFree
Abnormality-cause identifying apparatus and method does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Abnormality-cause identifying apparatus and method, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Abnormality-cause identifying apparatus and method will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3053441