Computer-aided design and analysis of circuits and semiconductor – Nanotechnology related integrated circuit design
Reexamination Certificate
1999-06-15
2002-05-21
Smith, Matthew (Department: 2825)
Computer-aided design and analysis of circuits and semiconductor
Nanotechnology related integrated circuit design
C700S121000
Reexamination Certificate
active
06393602
ABSTRACT:
FIELD OF THE INVENTION
The present invention generally relates to semiconductor wafer manufacture in the presence of particle contamination, and more particularly to the field of yield forecasting in a real-time semiconductor wafer manufacturing environment.
BACKGROUND OF THE INVENTION
Fabrication of semiconductor integrated circuits (ICs) is an extremely complex process that involves several hundred or more operations. They are fabricated by selectively implanting impurities into and applying conductive and insulative layers onto a semiconductor substrate. Semiconductor ICs (chips) are not manufactured individually but rather as an assembly of a hundred or more chips on a “wafer,” which is then diced up to produce the individual chips.
Increasing production yield is an ongoing problem in the manufacture of semiconductor chips. Because of various defects that can occur in the fabrication of a wafer, a significant number of wafer die have to be discarded for one reason or another, thereby decreasing the percentage yield per wafer and driving up the cost of the individual chips. Defects are typically caused by foreign particles, minute scratches, and other imperfections introduced during photoresist, photomask, and diffusing operations. Yield impacts the number of wafer starts at the inception of production needed to meet specific customer order quantities for finished chips at the end of the production line. With the high demand for semiconductor chips and more orders than can possibly be filled by a production facility, predicting yield to accurately gauge wafer starts and utilizing defect information to remove yield-detracting operations are important aspects of improving the efficiency and hence the output of the fabrication facility.
Wafer-scanning tools are utilized to identify defects that occur in the chip manufacturing process for the aforementioned purposes. Typically, such tools are located at a variety of positions along the production line and include automated-vision inspection stations for identifying visual irregularities in the wafer die as they move through the line. The irregularities, i.e., defects, are recorded according to their coordinates, estimate of size, or other parameters and are stored as records in a database. The records represent raw information that is then analyzed or otherwise processed offline to determine the impact, if any, of the identified defects on product yield. Some defects, for example, may not adversely affect yield as much as others, and correspondingly must be classified differently for analysis purposes.
Commercially available wafer scanning tools include those made by KLA Instruments Corporation of Santa Clara, Calif.; Tencor Instruments Corporation of Mountain View, Calif.; Inspex, Inc. of Billerica, Mass.; and numerous other manufacturers. Despite significant advances made in wafer-scanning technology, the various tools that are available suffer striking deficiencies. In particular, such tools lack the capability to perform certain advanced classification and analysis of defect information necessary to accurately determine the true impact of wafer defects on yield. While conventional tools offer simple data presentation capabilities, such as the display of wafer maps, histograms and charts, they do not adequately classify or process the defect data.
More specifically, a disadvantage suffered by scanning tools is that they do not adequately perform yield prediction operations beneficial in a manufacturing defect analysis, thereby limiting the utility. It is often desirable to further refine the defect data before manual inspection and classification of individual defects on the review station. Since each wafer can include so many defects, it would not be practical to manually review and classify each of them. It would be desirable to utilize a method to randomly choose a statistically meaningful sample, i.e., subset, of such defects for consideration.
Historically, the review station operator randomly picks sets of defects that seem interesting and then reviews and classifies them. However, it is difficult for humans to systematically choose defects for this purpose that will be representative of all of the defects on the wafer. Some review stations are equipped with the ability to randomly move to different defects which the operator can then review and classify. A problem though with conventional randomizing methods performed on review stations is that they are not necessarily accurate in representing a true sampling of the wafer. For example, picking defects at random tends to result in the inordinate picking of defects that are part of a big cluster, because there are more of them, while defects of other types and in other locations on the wafer are overlooked. Therefore, it would be desirable to adopt an automated and consistent method for randomly identifying for review defects of interest. This method could focus on defect subpopulations defined in terms of defect size ranges or, alternatively, in terms of locations on the wafer, so that the sample of defects chosen best reflects the conditions actually occurring on the wafer.
FIGS. 1 and 2
illustrate a semiconductor wafer
2
, which includes five particles
4
, and the semiconductor wafer
2
′ contains eleven particles
4
′.
FIG. 3
illustrates a schematic illustration of a semiconductor device in a semiconductor wafer. Circuit conductor lines
6
and
8
are designed in the semiconductor wafer to conduct electrical signals independently of one another. Due to imperfections in the semiconductor wafer manufacturing process, particle
10
has been introduced between conductors
6
and
8
. Particle
10
does not interfere with either of conductors
6
and
8
and will generally not affect the functionality (or yield) of the semiconductor device or wafer. Accordingly, even though particle
10
is a result in a defect in the semiconductor wafer manufacturing process, the particle does not cause failure in the semiconductor device by disturbing signals flowing in conductors
6
and
8
.
FIG. 4
is also a schematic illustration of a portion of a semiconductor device similar to the illustration of FIG.
3
. However, in
FIG. 4
, particle
10
′ is much larger than particle
10
of FIG.
3
. In this example, particle
10
′ is in contact with both conductors
6
and
8
at regions
12
and
14
, respectively. If particle
10
′ is able to conduct electricity, the independent operation of conductors
6
and
8
will be jeopardized, creating cross-talk between conductors
6
and
8
. If different devices are connected to conductors
6
and
8
, a single particle
10
′ may destroy the two devices embedded in the semiconductor wafer. Accordingly, particle
10
′ is what is commonly known as a “killer defect” since particle
10
′ may kill or prevent the normal operation of the semiconductor device which utilizes conductors
6
and
8
. While the presence or absence of killer defects may be determined, it is important to utilize the defect characteristics in a semiconductor wafer.
FIGS. 5 and 6
are schematic illustrations of a portion of a semiconductor device for providing some additional background information regarding semiconductor defects. In
FIG. 5
, semiconductor device conductor lines
16
and
18
are separated by the distance
20
. During the manufacturing process, particle
22
is introduced in the semiconductor wafer due to manufacturing defects or imperfections. Particle
22
has a diameter
24
and center point
26
as illustrated. In the situation illustrated in
FIG. 5
, particle
22
is in contact only with conductor
16
and is unable to be extended to contact both conductors
16
and
18
. Therefore, particle
22
is considered to be a non-killer defect. Note that in this situation, the position of center
26
of particle
22
, identified by dashed line
30
, is spaced apart from the center position
28
of conductors
16
and
18
by distance
32
. As particle
22
moves closer toward conductor
18
, the center
26
Atchison Nick
Ross Ron
Brady W. James
Garbowski Leigh Marie
Smith Matthew
Swayze, Jr. W. Daniel
Telecky , Jr. Frederick J.
LandOfFree
Method of a comprehensive sequential analysis of the yield... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method of a comprehensive sequential analysis of the yield..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method of a comprehensive sequential analysis of the yield... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2842960