Computer-aided design and analysis of circuits and semiconductor – Nanotechnology related integrated circuit design
Reexamination Certificate
1998-05-22
2001-10-30
Smith, Matthew (Department: 2825)
Computer-aided design and analysis of circuits and semiconductor
Nanotechnology related integrated circuit design
C716S030000, C430S005000, C430S022000, C430S030000
Reexamination Certificate
active
06311319
ABSTRACT:
FIELD OF THE INVENTION
The invention relates to the general field of image distortion due to Optical Proximity Effects with particular reference to ways to correct for this effect.
BACKGROUND OF THE INVENTION
The proximity effect is a form of optical distortion associated with photoresist images. For a given development time, whether or not a given area of a photoresist layer will be left or removed after the development process depends on the total amount of energy deposited in that area during its exposure to radiation. Image features whose size and/or separation approach the resolution limit of said radiation will thus be subject to distortion that depends on how the diffraction maxima and minima, that lie on both sides of a ‘sharp’ edge, interact with one another.
The proximity effect can be compensated for, at least in part, by modifying any given feature in the opposite direction to the expected distortion. Thus, a line that would otherwise come out too narrow can be drawn as wider than its true width, etc. The data that represents the information from which a mask suitable for use in photolithography can be generated, is stored in a data file so corrections to allow for the proximity effect will also be stored there. The overall nature and scope of these corrections, and how they get into the file, will vary with the application and the user.
The optical proximity correction (OPC) is commonly calculated by summing two Gaussian functions whose value depend on a critical dimension (CD) defined by the design rules as well as on the wave-length of the exposing radiation. In general, the distortion of lines that are part of a dense assemblage will be more positive than the distortion of isolated lines in optical mode. While OPCs obtained in this manner provide satisfactory results, the computation time can be very long, typically about 16 hours for a single mask file using state of the art computers.
An examination of the changes made to mask images as a result of applying OPC, shows that the OPC takes two principal forms—scatter bars and serifs, the latter category, to which the present invention is limited, includes hammerheads. A serif is a small square that is added to the corner, or vertex, of a stripe. Vertices may be positive or negative, corresponding to whether they are convex or concave. A positive serif extends the boundaries of a positive vertex while a negative serif reduces the boundaries of a negative vertex. A hammerhead may be viewed as the fusion of two serifs, located on adjacent vertices.
Referring now to
FIG. 1
, an example of a pattern of stripes is shown such as might form part of a layout mask for an integrated circuit. Typically the width of such stripes would be between about 0.18 and 0.35 microns and their minimum separation would be between about 0.18 and 0.35 microns. In this particular example, all stripes (except the stripe labelled
2
) have contact holes near their ends so every vertex (except
3
and
4
) will need a serif to compensate for its proximity to another object, even though that object happens to be at a lower level within the IC. Prior to the present invention it had been the practice in our laboratory to attach serifs to all vertices, whether or not they actually needed OPC.
This is shown in FIG.
1
. This approach was taken because making the necessary changes to the mask data file is very easy and is not time consuming. It should, however, be noted that each serif that is added must be formed separately (by exposure to an electron beam of variable shape) when the actual reticle is being drawn. If many serifs are involved, this can add substantially to the time (and hence the cost) required to prepare a full reticle. It should also be noted that the same amount of electron beam time is required to produce a single hammerhead as a single serif, so that wherever adequate OPC can be obtained by replacing two serifs with a single hammerhead, it is cost effective to do so.
As already noted, the cost of computing the full OPC is very high. Furthermore, a method of OPC that adds only serifs and hammerheads where they are needed would be as effective as the full OPC treatment, provided that the distortions that have not been corrected do not introduce shorts, opens, hot spots, etc. in the line patterns that end up being formed in the integrated circuit. Such a method would therefore be attractive if it substantially reduced the cost of OPC. An important additional benefit of such a method would be the reduced cost of reticle formation discussed above.
A number of approaches have been taken in the prior art to dealing with the proximity effect without the need to perform the full OPC calculation. For example, Liebmann (U.S. Pat. No. 5,553,273 September 1996) aims to correct Optical Proximity Effects by biassing critical portions of the design. In particular, this invention attempts to minimize the creation of new vertices so that it actually teaches away from the practice of using serifs.
Liebmann et al. (U.S. Pat. No. 5,657,235 August 1997) use the OPC data to drive the mask writer itself rather than changing the data design file. By assigning relative mask writer doses, as needed, they are able to bring about continuous line width variations (to compensate for OP effects) without increasing the size of the data design file. Serifs and scatter bars are not involved.
Chung et al. (U.S. Pat. No. 5,432,714 July 1995) show how accumulated information on exposure can be used during electron beam lithography to compensate for proximity effects. Sporon-Fiedler et al. (U.S. Pat. No. 5,208,124 May 1993) teach increasing and decreasing line features to compensate for proximity effects. Serifs and hammerheads are not used here either. It is thus apparent that none of the above references offer the simplicity of application provided by the present invention.
SUMMARY OF THE INVENTION
It has been an object of the present invention to provide a process whereby optical proximity corrections may be applied to a mask pattern, such as used in the manufacture of integrated circuits, at minimum cost.
Another object has been that said process lead to a reduction in the cost of preparing a reticle from said mask pattern.
These objects have been achieved by adding serifs and/or hammerheads to the vertices of the mask pattern on the basis of a small number of simple rule checks. The first check is for the presence of an overlapping pattern at the next level. If this is not detected nothing is added at the vertex in question. If some overlap is detected, a predefined search area (at the same mask level) is quickly scanned and, if another stripe is detected to be located within a preset distance, serifs are added at the appropriate vertices. If no stripe was found a second search area, further away, is scanned and if a neighbouring stripe is detected this time, larger serifs are added. If the second search comes up empty, a hammerhead is added at the appropriate line end. The method may be refined by using additional searches, if so desired.
REFERENCES:
patent: 5208124 (1993-05-01), Sporon-Fiedler et al.
patent: 5432714 (1995-07-01), Chung et al.
patent: 5553273 (1996-09-01), Liebmann
patent: 5657235 (1997-08-01), Liebmann et al.
patent: 5707765 (1998-01-01), Chen
patent: 5804340 (1998-09-01), Garza et al.
patent: 5900338 (1999-05-01), Garza et al.
patent: 5965306 (1999-10-01), Mansfield et al.
patent: 6044007 (2000-03-01), Capodieci
Hsieh Ren-Guey
Tu Chih-Chiang
Ackerman Stephen B.
Saile George O.
Smith Matthew
Speight Jibreel
Taiwan Semiconductor Manufacturing Company
LandOfFree
Solving line-end shortening and corner rounding problems by... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Solving line-end shortening and corner rounding problems by..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Solving line-end shortening and corner rounding problems by... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2564217