UV Microscopy Resolves Smaller Features
Features as small as 80 nm can be resolved with new UV microscopy.
Werner Hunn,Leica, Deerfield, Ill. -- Semiconductor International, 7/1/1998
| At a Glance | |||
| |||
This applies first to the lithography field, where optical steppers are utilized at today's high-volume semiconductor factories for the 0.25 µm (250 nm) design rules. This is accomplished with 248 nm DUV exposure technology. The next two technology generations -- 0.18 µm (180 nm), which is upon us, and 0.15 µm (150 nm) design dimensions -- can be handled by 193 nm DUV optical exposure tools. However, for the 0.13 µm (130 nm) design rules and beyond, the lithography options are not so clear, as new competing technologies such as X-ray, extreme UV and e-beam, to name a few, are to be developed.
To measure and defect inspect the produced devices on wafers with optical instruments is equally challenging, even though the microscope systems have the big advantage of much higher imaging aperture over optical lithography tools. While stepper lenses can only be manufactured with numerical apertures (NAs) of around 0.60 plus, the optical microscope utilizes dry objectives of up to 0.95 NA of a possible 1.0, and this is at highest lens correction. As the resolving power is inversely proportional to the NA, a corresponding gain in detail resolution results:
"Raleigh" Resolution =
0.61 X l
/NA
(Eq. 1)
A highly corrected apochromatic microscope objective of 0.95 NA can clearly resolve point pairs of 350 nm spacing in white light. When coupled with a confocal imaging technique, the image contrast can be pushed higher.
The definition of microscope resolution is, however, somehow arbitrary in that the "Raleigh criterion" defines two points as resolved when according to the point spread function they are separated by an intensity saddle of about 75%. This is a mathematically derived number, and based on resolution considerations of line pairs (modulation transfer function), the resolution in terms of feature size is about half the resolving power of an objective at a specified NA.
| Fig. 1. Resolution pattern imaged in brightfield (550 nm). |
180 nm; and DUV (operating at 250 nm) provides a resolution of 80 nm.
To resolve smaller structures, it is obvious (from Equation 1 and accompanying figures) that reducing the wavelength is most effective. Microscopy has been late in adapting to UV/DUV development compared to optical lithography, because of the previously mentioned numerical aperture/resolution advantage of microscope objectives. Also, the availability of suitable microscope light sources and high spatial resolution UV/DUV detectors has greatly improved, favorably affecting the cost/price ratio of a UV/DUV microscope.
|
|
| Fig. 2. The same pattern as Figure 1, imaged in UV (364 nm). |
With changes and improvements in all the above points, UV microscopy is finally realized. Because light absorption and interaction in UV imaging is more material dependent than in visible imaging, we can expect a gain in contrast, at least in certain applications; but, depending on the pattern type and material, the image may look unfamiliar as compared to visible light, and for that reason alone it is important to be able to switch the microscope between UV and visible imaging.
Immersion media between an objective and the wafer is not practical in the production line. Therefore, dry objectives of highest NA, image correction and UV transmission have to be designed. Furthermore, the rest of the optical system (such as tube lenses) has to be optimized.
Figures 1 and 2 are pictures of resolution patterns that respectively show structures imaged in brightfield (550 nm) and UV (365 nm). The gain in information can clearly be seen.