Reticle and direct lithography writing strategy

Details Reticle and direct lithography writing strategy Reticle and direct lithography writing strategy

2001-11-16

2003-08-12

Berman, Jack (Department: 2881)

Radiant energy

Irradiation of objects or material

Irradiation of semiconductor devices

Reexamination Certificate

active

06605816

ABSTRACT:

FIELD OF INVENTION
The present invention relates to preparation of patterned reticles to be used as masks in the production of semiconductor and other devices. Methods and devices are described utilizing resist and transfer layers over a masking layer on a reticle.
RELATED ART
Semiconductor devices include multiple layers of structures. The structures are formed in numerous steps, including steps of applying resist, then exposing, developing and selectively removing the resist to form a pattern of exposed areas. The exposed areas may be etched to remove material or sputtered to add material. A critical part of forming the pattern in the resist is exposing it. Resist is exposed to an energy beam that changes its chemical properties. One cost-effective way of exposing the resist is with a stepper. A stepper uses a reticle, which typically includes a carefully prepared, transmissive quartz substrate overlaid by a non-transmissive or masking layer that is patterned with areas to be exposed and areas to be left unexposed. Patterning is an essential step in the preparation of reticles. Reticles are used to manufacture semiconductor and other devices, such as flat-panel displays and television or monitor screens.
Semiconductor devices have become progressively smaller. The feature dimensions in semiconductor devices have shrunken by approximately 40 percent every three years for more than 30 years. Further shrinkage is anticipated. Current minimum line widths of approximately 0.13 microns will shrink to 0.025 microns, if the historical rate of development continues for another 15 years.
The pattern on a reticle used to produce semiconductor devices is typically four times larger than that on the wafer being exposed. Historically, this reduction factor has meant that minimum feature dimensions in the reticles are less critical than the minimum feature dimensions on the surface of the semiconductor. However, the difference in criticality is much less than might be expected and will in the near future disappear.
Critical dimension uniformity, as a percentage of line width, is more exacting in the pattern on a reticle than in the features on the surface of a wafer. On the wafer, critical dimension uniformity of plus or minus 10 percent of the line width has historically been acceptable. In the error budget for the wafer line width, the mask has been allowed to contribute half of the critical dimension variation, or a variation of five percent of a line width. Other factors use the remaining error budget. It has been observed that nonlinearities in transfer of a pattern from a reticle to a wafer magnify any size errors in the mask. This is empirically quantified as a mask error enhancement factor (MEEF or MEF). In current technology, the mask error enhancement factor is typically two. Therefore, the critical dimension uniformity on the reticle is reduced to approximately two and one-half percent of a line width, to remain within the error budget.
It is anticipated that requirements for critical dimension uniformity will tighten in time, particularly for masks. On the surface of the wafer, a critical dimension uniformity of plus or minus five percent of the line width will be required in the future. At the same time, the mask error enhancement factor is likely to increase due to more aggressive lithographic process trade-offs, such as tuning the lithographic process to optimize the manufacture of contact holes, transistors or other critical features in order to use feature sizes closer to the theoretical resolution limit. For masks, a critical dimension uniformity of plus or minus one percent of a line width or feature size is anticipated. At this rate, the tolerance for critical dimension errors on the mask will be smaller in absolute nanometers than it is on the surface the wafer, despite the fact that the stepper takes advantage of a mask that is four times as large as the area on the wafer that is being exposed.
One of the energy beam sources currently used to expose resist is deep ultraviolet (DUV), in the wavelength range of 100 to 300 nanometers. This energy source is used with two types of resist to produce masks: conventional positive, so called Novolac-DNQ, resist and chemically amplified resist. Essentially all DUV exposure in steppers uses chemically amplified resist. The requirements in pattern generators for patterning of reticles are so different than in steppers that chemically amplified resists are unsuitable for patterning reticles. Work to modify conventional Novolac-DNQ resist to produce a resist suitable for DUV exposure of mask patterns reportedly has failed.
Uniformity and feature size requirements have become so demanding that wet etching no longer is suitable. Wet etching is generally not useable when the size of features approach the thickness of the films the features are etched from. A wet etch etches sideways as much as it etches vertically. Deterioration of the three-dimensional shape of small features results. When chrome is wet etched with resist as an etch mask, the etchant removes chrome under the resist, referred to as undercutting. Clear areas produced by wet etching chrome with a resist mask typically come out 0.2 microns too large. A wet etched resist image with alternating lines and spaces equally 0.4 microns wide, produces a chrome mask pattern where the spaces (clear) are 0.6 microns wide and the lines (dark) are 0.2 microns. This is a large deviation. It is difficult to compensate for this deviation by changing the data or the dose. For smaller features, narrow lines will simply disappear. Therefore any pattern with features smaller than 0.5-0.6 microns wide needs to be produced by dry or plasma etching. The plasma process used to etch chrome produces vertical “line-of-sight” etching characteristics. The chrome is removed only where it is within the line of sight from the plasma source; essentially no undercutting results.
Issues Using Positive Non-Amplified Resists
Positive non-amplified resists provide excellent performance in the violet visible and near UV wavelength ranges. This resist is transparent and has high contrast, giving essentially vertical resist walls and good process latitude. It has good shelf life and mask blanks can be precoated with resist at the time of manufacturing, shipped to users, and kept in storage until needed. Although there is a small decay of the latent image, plates can in principle be exposed today and developed after weeks.
In the DUV wavelength range, both the Novolac resin and the photoactive compound used in Novolac absorb strongly. The edge wall angle after development is partly controlled by the absorption of light and partly by the resist contrast. With high absorption, the features will have strongly sloping edge walls, whatever the chemical contrast. No non-amplified resist formulation is known which combines good contrast with high transparency.
The effect of non-vertical trench walls is significant for narrow lines. One reason for non-vertical trench walls is that a resist layer is eroded by the plasma during the etching. The uniformity of resist erosion is difficult to control since, among other things, it depends on the pattern to be etched. Erosion makes the clear areas larger and varying plasma activity from run to run and across the surface of the workpiece gives a varying CD between masks and within each mask. The variation of the resist thickness at the end of the plasma etching step may be 50 nm peak-to-valley or more. For a wall angle of 80 degrees, instead of 90 degrees, a 50 nm variation in resist thickness produces a variation in trench width, at the bottom of the trench, of nearly 20 nm, which may translate into an undesirable three-sigma deviation of 20 nm. This erosion problem is exacerbated by the high optical absorption of non-chemically amplified resists used with DUV radiation. High optical absorption leads to greater development of the resist at the top of the trench than the bottom, further increasing the variation in line width.
Resist sidewall deviation from 90° vertical inevitab

Affiliated with

Also associated with

Rating

Reticle and direct lithography writing strategy does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Reticle and direct lithography writing strategy, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Reticle and direct lithography writing strategy will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-3091365