Optics: measuring and testing – Inspection of flaws or impurities – Surface condition
Reexamination Certificate
1997-12-18
2003-09-02
Rosenberger, Richard A. (Department: 2877)
Optics: measuring and testing
Inspection of flaws or impurities
Surface condition
Reexamination Certificate
active
06614520
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention relates to systems for determining whether a reticle is defective. More particularly, the present invention relates to systems and methods which identify reticle defects that arise at some time after a reticle is “qualified” as being suitable for use.
A normal reticle or photomask is an optical element containing transparent and opaque regions which together define the pattern of coplanar features in an electronic device such as an integrated circuit. A phase shift reticle, which is also well known in the art, may further include graded regions (with varying thickness) that cause a phase shift of the transmitted light. In order to learn more about phase shift reticles, reference may be made to book authored by Van Zant, Peter, entitled “Microchip Fabrication” McGraw-Hill, 1997, which is incorporated herein by reference for all purposes. Reticles are used during photolithography to define masks which protect specified regions of a semiconductor wafer from etching, ion implantation, or other fabrication process. For many modern integrated circuit designs, a reticle's features are between about 4 and about 20 times larger than the corresponding feature size of the mask on the wafer.
Reticles are typically made from a transparent medium such as a borosilicate glass or quartz plate on which is deposited an opaque pattern of chromium or other suitable material. The reticle pattern may be created by a laser or an e-beam direct write technique, for example, both of which are widely used in the art. The reticle is framed and covered by a pellicle which is a thin layer of an optically neutral material such as a polymer attached to the frame. Typically, an adhesive is used to affix the pellicle to the frame. Once in place, the pellicle (positioned about 6 mm from the reticle) protects the reticle from dirt or dust particles in the environment. Such particles may deposit on the pellicle but do not affect the reticle's image because the pellicle is located beyond the focal plane of the reticle.
During the normal course of the reticle's life, however, defects can be introduced into the reticle. For example, particles may be present but hidden (on the chromium region for example) when the reticle is initially formed. Over the course of time, some of these particles may migrate onto the transparent regions where they degrade the image quality. In another example, defects may be introduced into the reticle by “flaking” of the frame or the adhesive material that affixes the pellicle to the frame.
In yet another example, an electrostatic discharge (ESD) generated by a stepper apparatus employed during conventional photolithography may damage the opaque regions of the reticle.
If a reticle becomes defective due to one of the above mechanisms, for example, it may have a very negative impact on the yield of an IC fabrication facility. For example, a particle spanning two opaque lines on a reticle may result in shorting between adjacent metal or polysilicon lines. Other reticle defects may cause more subtle defects that can not easily be detected and may not be manifested until the ICs are in the customers' hands. Undetected, such defects can cost a facility many millions of dollars and potential embarrassment. Thus, many IC manufactures periodically image or otherwise test their reticles to ensure that they are not defective.
FIG. 1A
is an idealized representation of an actual “darkfield” image
10
of a reticle obtained by scanning a light beam onto the reticle and monitoring light scattered therefrom. In the actual image, various regions of the image have varying shades of gray. In
FIG. 1A
, the various shades of gray could not be accurately depicted, so the contrast between features is exaggerated in some cases and reduced in other cases.
Image
10
of the reticle has a dark area
14
, a bright area
16
, and a very dark area
12
. Areas
14
and
16
have certain relatively bright repetitive features created by light scattering off of valid repetitive features on the reticle surface. For example, dark area
14
includes vertical lines
22
created by some repetitive feature on the reticle. Such image patterns created by valid structures may fool a detector into believing that they constitute defects. Therefore these features are sometimes referred to herein as “false defects.” In addition to vertical lines
22
, dark area
14
also has a random bright spot
18
indicative of a reticle defect (hereinafter referred to as a “real defect”), which may be caused by an electrostatic discharge (ESD) for example.
Bright area
16
receives its brightness from bright bands
20
which are light scattered off of valid die features (more examples “false defects”). Very dark area
12
contains very little scattered light and no bright spots that would represent real or false defects on the reticle.
As should be apparent from a study of
FIG. 1A
, various real and false defects may appear in an image. Obviously, a test system must be able to separate the real from the false. Traditionally, this has been accomplished by employing a “die-to-die” comparison which may be carried out in KLA 301 or 351 Reticle Inspection Tool, commercially available from KLA-Tencor of San Jose, Calif.
In systems employing the die-to-die approach, the images of two supposedly identical patterns on a reticle are compared. Note that many reticles contain the patterns of multiple identical die, collectively referred to as a field. Images of two or more of these individual die patterns in a field are compared by optically overlaying the patterns. Such comparisons will screen the false defects because they will be found on the images of both die. Real defects presumably occur randomly and therefore appear only on a single die. Thus, a comparison of two die pattern images will normally find a real defect on only a single die pattern. Thus, the imaging system will flag bright spots appearing on only a single image as real defects.
FIG. 1B
shows some significant components of a scattering or “darkfield” detecting assembly
50
that may be employed to scan a reticle surface and generate an image of the reticle or die pattern. An incident beam
56
generated by an illuminating source
52
, e.g., a laser, is directed at a portion of a reticle surface
54
. Incident beam
56
travels along an incident axis
70
and perpendicular to an axis
72
. First and second detectors
64
and
68
, positioned at an oblique angle, e.g., 45°, with respect to the incident axis
70
, detect a first and second scattered energy signals
58
and
60
, respectively, from reticle portion
54
after the scattered energy signals pass through filters
62
and
66
.
During a typical inspection process of reticle portion
54
, illuminating source
52
directs incident beam
56
to strike reticle portion
54
and a resulting scattered light signal is detected by first and second detectors
64
and
68
. A defect residing at reticle portion
54
may, therefore, be flagged, if the intensity of the detected light signal is equal to or exceeds a predetermined threshold signal intensity. If, however, the intensity of the scattered energy signal detected is less than a predetermined threshold signal intensity, then reticle portion
54
is considered to be free of defects. Typically, the source and detectors are moved in a rasterized fashion to generate an image of the entire reticle.
During a die-to-die comparison in the same reticle, it may be difficult to discriminate between false defects and true defects. This is because there may be subtle differences between the dies that are not necessarily true printable defects. For example, small differences in feature width may fall within acceptable tolerances but still show up as defects on die-to-die comparisons. Further, some reticles contain a pattern for a single die only. Obviously, in such cases die-to-die reticle inspection can not be implemented.
What is needed is an improved inspection system that rapidly and inexpensively determines whether a defect
Bareket Noah
Desplat Christian G.
Glasser Lance A.
Beyer Weaver & Thomas LLP.
Kla-Tencor Corporation
Olynick, Esq. Mary R.
Rosenberger Richard A.
LandOfFree
Method for inspecting a reticle 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 for inspecting a reticle, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for inspecting a reticle will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3063776