Data processing: measuring – calibrating – or testing – Measurement system – Statistical measurement
Reexamination Certificate
2000-04-06
2002-12-17
Sherry, Michael (Department: 2829)
Data processing: measuring, calibrating, or testing
Measurement system
Statistical measurement
C702S183000, C438S014000
Reexamination Certificate
active
06496788
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to a data processing method and apparatus for calculating a killer ratio resulting from a specified defect, and an information storage medium for storing programs as software which cause a computer to execute various processing operations.
2. Description of the Related Art
Currently, IC (Integrated Circuit) chips are generally mass-produced by dividing a single semiconductor wafer into a plurality of dies and then manufacturing integrated circuits of the same structure on the respective dies. Various approaches for improving the yields of such mass-produced IC chips are also studied, and for example, the calculation of a killer ratio is performed based on past die research data and yield data.
The killer ratio refers to an expected value of the probability that, when a defect of one type is specified, the specified defect present on a die results in a defective IC chip manufactured from the die. For example, when one die and a defect of one type are specified, the specified defect present on the specified die does not necessarily cause the specified die to be defective. Also, even when the specified defect is not present on the specified die, the specified die may become defective due to another type of defect.
The die research data includes, for example, defect presence/absence data for indicating the presence or absence of defects for each type and result pass/fail data for indicating whether a manufacturing result is favorable or unfavorable, for each die identification data which identifies a plurality of dies. The die research data is registered for each of a plurality of manufacturing steps which are performed on a single wafer.
The aforementioned defect presence/absence data is collected from a plurality of dies for each of a plurality of manufacturing steps when a semiconductor wafer is divided into a plurality of dies to mass-produce a plurality of IC chips through various manufacturing steps, while the result pass/fail data is collected by performing a final test of each of a plurality of finished IC chips.
It should be noted that while a plurality of sets of die research data in a single manufacturing step of a single semiconductor wafer are herein referred as an example as mentioned above for simplifying description, a number of sets of die research data for a plurality of manufacturing steps of a single semiconductor wafer are generally collected and accumulated corresponding to a plurality of semiconductor wafers of a lot, and a number of sets of die research data for each lot is collected and accumulated corresponding to a plurality of lots.
As shown in FIG.
1
(
a
), when a defect of one type is specified, die research data of a plurality of dies from a single semiconductor wafer is classified into four: “specified defect present and manufacturing result favorable”; “specified defect present and manufacturing result unfavorable”; “specified defect absent and manufacturing result favorable”; and “specified defect absent and manufacturing result unfavorable”, as shown in FIG.
1
(
b
).
In this case, assuming that the total number of dies with the specified defect present thereon is T′, and the number of dies with the specified defect present thereon and a favorable manufacturing result is T
G
′, then a killer ratio KR is given by:
KR
=(the number of dies with the specified defect and an unfavorable manufacturing result)/(the total number of dies with the specified defect)=1−(the number of dies with the specified defect and a favorable manufacturing result)/(the total number of dies with the specified defect)=1−
T
G
′/T′
(1)
The calculation of the killer ratio with one of a plurality of types of defects being specified as mentioned above can predict the ratio at which a finished IC chip manufactured from a die becomes defective when the specified defect is present on the die. However, the aforementioned calculation method of the killer ratio does not consider a defective IC chip due to unspecified defects, resulting in unsatisfactory calculation accuracy of the killer ratio.
Thus, a calculation method has been proposed for improving the accuracy of the killer ratio by considering the effects of unspecified defects. The method assumes that all types of defects are randomly distributed over the entire region of a semiconductor wafer and a die has the specified defect and another kind of defect mixed thereon.
A survival rate (a ratio at which its manufacturing result is favorable) is detected for dies with no specified defect and is used as “a baseline yield” which is a survival rate of unspecified defects. Assuming that the total number of dies is T and the number of dies with a favorable manufacturing result is T
G
, then the baseline yield Yb is given by:
Yb
=(
T
G
−T
G
′)/(
T−T′
) (2)
When the aforementioned equation (1) is corrected by the above equation (2), the effects of types of defects other can be eliminated from the killer ratio of dies with the specified defect. The killer ratio KR
M
in this case is given by:
KR
M
=1−(
T
G
−T′
)/
Yb=
1−
[T
G
′(
T−T′
)]/[
T
′(
T
G
−T
G
′)] (3)
The aforementioned calculation method of the killer ratio resulting from a specified defect has satisfactory accuracy since it can eliminate the effects of unspecified other kinds of defects. However, even with the aforementioned calculation method of the killer ratio, the number of the specified defects present on a single die is not taken into account, thereby reducing the accuracy of the killer ratio calculation due to differences in the number.
SUMMARY OF THE INVENTION
The present invention has been made in view of the aforementioned problems, and it is an object thereof to provide a data processing method and apparatus capable of calculating a killer ratio with more satisfactory accuracy than that in the prior art, and an information storage medium for storing programs as software which cause a computer to execute processing operations of the data processing method of the present invention.
According to a first aspect of the present invention, the data processing apparatus comprises data storing means for previously registering die research data which includes at least the number of defects for each type and manufacturing result pass/fail for each of a plurality of dies divided from a single wafer, data input means for receiving type selecting data for selecting a specified defect which is a defect of one type, data retrieving means for retrieving the die research data corresponding to the specified defect from the data storing means with the type selecting data input to the data input means, data classifying means for classifying the die research data retrieved by the data retrieving means based on the number of the specified defects present on a die, and first calculating means for calculating a killer ratio for each of a plurality of groups of die research data classified by the data classifying means.
Therefore, in the data processing method with the data processing apparatus of the present invention, the die research data including at least the number of defects for each type and manufacturing result pass/fail for each of a plurality of dies divided from a single wafer is previously registered in the data storing means. In such a state, when the type selecting data for selecting a specified defect which is a defect of one type is input to the data input means, the data retrieving means retrieves the die research data corresponding to the specified defect from the data storing means with the input type selecting data. The data classifying means classifies the retrieved die research data based on the number of the specified defects present on a die, and the first calculating means calculates the killer ratio for each of a plurality of groups of the classified die research
NEC Corporation
Pert Evan
Sherry Michael
LandOfFree
Data processing method and apparatus to determine killer... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Data processing method and apparatus to determine killer..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Data processing method and apparatus to determine killer... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2995586