Radiation imagery chemistry: process – composition – or product th – Imaging affecting physical property of radiation sensitive... – Electron beam imaging
Reexamination Certificate
2000-03-02
2002-08-20
Ashton, Rosemary (Department: 1752)
Radiation imagery chemistry: process, composition, or product th
Imaging affecting physical property of radiation sensitive...
Electron beam imaging
C250S492220
Reexamination Certificate
active
06436607
ABSTRACT:
FIELD OF THE DISCLOSURE
This disclosure relates to lithography and more specifically to beam lithography, for instance electron beam or laser pattern generator lithography, and more particularly to proximity effect correction for such lithography.
DESCRIPTION OF THE RELATED ART
Lithography is a well-known field, especially as used in the semiconductor industry for fabricating both integrated circuits and the masks used in optical stepper lithography to fabricate integrated circuits. Beam lithography involves directing a modulated beam of radiation such as electron beams or laser light onto a sensitive resist which coats a substrate. The substrate is, e.g., a mask blank or semiconductor wafer. Only portions of the resist are exposed by the beam. The resist is then developed and the exposed portions (or the nonexposed portions) are removed depending on the type of resist. Subsequently, the patterned resist is used for etching of the underlying material or growth, for instance, of oxides.
Typically in beam lithography, such as electron beam lithography or laser pattern generator lithography, the beam is modulated and scanned across the surface of the workpiece (wafer or mask blank). The scanning defines two-dimensional features which have particular shapes. Typically, the pattern being imaged is partitioned into a number of such features by pattern generation software and/or hardware which is a portion of the lithography tool.
A process called proximity effect correction is well known in such beam lithography. This especially applies to the field of electron beam lithography but is not so limited. When a resist layer is exposed by an electron beam, the feature edges do not develop exactly as exposed due to electron scattering in the resist and underlying substrate. Particularly, an edge of a feature in close proximity to an edge of another feature is moved after development compared to its nominal (intended) location due to this proximity effect more than is an isolated feature edge. Clearly, this results in undesired distortions (errors) in the resulting developed pattern. It would be highly advantageous to reduce such differences to a constant regardless of the proximity of a feature edge to an adjacent feature.
This problem is illustrated in
FIGS. 1
a
and
1
b
.
FIG. 1
a
shows graphically exposure of a feature. The horizontal axis is distance x in microns (&mgr;m). The vertical axis is the exposure level of the feature (in arbitrary units). The exposure pattern in this case includes the nominal region which is the main portion of the feature and a “prebias” edge region as shown. A corresponding energy density profile in
FIG. 1
b
shows the problematic tail which is the dark area to the right of the curve (“Prebias+Nominal”) causing the influence on an adjacent feature due to the proximity effect. The problem is that the height of this tail varies. Desirably, the height of the tail should be constant throughout the imaged area. Prior art proximity effect correction techniques, such as the “GHOST” technique, do accomplish this. See U.S. Pat. No. 4,463,265 issued Jul. 31, 1984 to Owen et al., incorporated by reference in its entirety, for description of the “GHOST” technique which is one type of proximity effect correction. The GHOST technique arranges the exposure such that there is a uniform background. However, the GHOST technique has the drawback that the modulation of the exposure density is reduced compared to the uncorrected case. This results in a reduced process window and therefore reduced yield due to statistical fluctuations in dose, resist properties, and other process parameters.
Another closely related problem is that of resist profile angle. This is shown in
FIG. 1
c
which is a side view of a substrate S on which is a patterned resist region R. As seen, the left hand side of the resist region R is at angle &thgr; to the principal surface of the substrate S. Nominally and ideally, angle &thgr; is 90 degrees as indicated by the dotted line. That is, the resist edges are preferably vertical. Such vertical resist edges are relatively easily achieved in conventional optical mask (stepper) lithography. It is much more difficult in beam lithography of the type described herein to obtain such a 90 degree angle or even close to it due to scattering effects.
Moreover, the resist profile angles &thgr; may vary on any one workpiece as shown. Without explicit control, the resist profile angle &thgr; from beam lithography varies with feature size and packing density (proximity) even on any one workpiece. It would be desirable to control this resist profile angle of each feature edge to be within a user-specifiable tolerance so that, at a minimum, all feature edges have the same resist profile angle. The reason that variation in resist profile angle is undesirable is that the subsequent processing steps, which typically include etching or material growth, are influenced by the resist profile angle. That is, especially subsequent resist dry etching is likely to follow approximately the sidewall angle of the resist profile angle. Hence, variations in resist profile angles lead to process variations in subsequent steps, which are very undesirable. Again, the variation in the resist profile angle is a result of variations in feature size and packing density and hence has a similar general cause as does the above described edge location problem.
Prior art dose modulation methods (see M. Parikh,
Journal of Vacuum Science and Technology
, vol. 15, No. 3, pp. 931-933, May/June 1978) do not provide the benefits of the GHOST technique but do show less loss of modulation of the exposure density. With prior art dose modulation it is possible, albeit with large computational effort, to position feature edges at their intended locations. However, it is not possible to control the resist profile angles.
SUMMARY
In accordance with this disclosure, the above-described problems are addressed using dose conservation during pattern modification. Prior art, in an attempt to correct for proximity effects, modifies feature exposure doses or moves feature edges, thereby varying the energy scattered into neighboring features.
This undesirably results in complicated iterative calculations.
It also results in the above-described problem of variations in resist profile angle.
In accordance with this disclosure, the data which represents the features, and used by the lithography tool to control the actual exposure process, is modified such that the feature edge in question is corrected while neighboring features suffer little or no change. This decoupling is such that the proximity effect correction need not be iterated as in the prior art (see Parikh, cited above). This substantially reduces computational complexity and hence increases lithography throughput. Thus, the features are modified such that edge locations and resist profile angles are corrected while neighboring features suffer little or no change. This approach recognizes that the exposure energy density deposited in the resist controls not only the location of feature edges, but also the resist profile angle, and seeks in one embodiment to control the profile subject to holding feature edge locations constant.
More broadly, the present method includes estimating (predicting) where a developed feature edge will occur relative to a given feature edge; then modifying the feature in the vicinity of the given feature edge, using dose conservation, to compensate for the estimated misplacement of the developed edge. This modifying includes one of (1) defining a border of the feature and translating the boarder; or (2) defining a border of the feature and modulating the point-wise dose within the border; or (3) or translating the feature edge.
Note further that the feature border region vs. the central region of the feature has no special significance in terms of dimensions; in the extreme case, each border region is half the width of the feature and so the two border regions are the totality of the feature.
Thus the present method also i
Carroll Allen M.
Lozes Richard L.
Muray Andrew
Ashton Rosemary
Kuo Jung-hua
LandOfFree
Border modification for proximity effect correction in... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Border modification for proximity effect correction in..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Border modification for proximity effect correction in... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2890128