Assessing Damage for UV-Laser-Resistant Fused Silica

Dr. Johannes Moll

Studying laser-induced damage in silica may help scientists improve its resistance and provide another level of flexibility for lens designers.

The heart of any optical lithography tool is its illumination and projection system. The illumination system preconditions the laser beam and provides the optical path from the laser to the photomask, and the projection lens images the pattern from the photomask onto the wafers. The majority of required optical elements are made of high-purity fused silica, and the remaining ones are made of CaF₂ to correct chromatic aberrations. Because lithography tools are generally expected to have a lifetime of approximately 10 years or, in terms of laser exposure of the optical system, 100 billion to 400 billion laser pulses, it is important to understand how deep-UV laser radiation influences these lens materials.

Lens materials are expected to withstand about 100 billion to 400 billion laser pulses to be cost-effective. By exposing fused silica samples at 2000 Hz, automated laser-exposure test benches help quantify laser-induced absorption, wavefront distortion and birefringence.

Laser-induced absorption is well-understood, and quantitative models exist to forecast the long-term performance of fused silica materials when exposed to KrF and ArF excimer-laser radiation. Research has improved the understanding of excimer-laser-induced density and refractive index changes in fused silica.

Photo-induced breaking and forming of molecular bonds is at the root of all laser damage in fused silica. As molecular bonds change, so do both the structural arrangement and the chemical composition of the glass. Understanding this collective reaction becomes critical as lithography tools begin to employ shorter wavelengths, because the energy at these wavelengths increases the effects of laser radiation on the lens material. Primary examples include:

High-purity fused silica lenses represent the majority of optical elements in both the illumination and projection systems in lithography steppers.

• Increased optical absorption caused by color-center formation.

• A change in the optical index of refraction resulting from variations in the chemical composition.

• The change in density of the glass caused by structural rearrangement and alteration of its chemical composition.

Any density change in the material alters the index of refraction and introduces stress-induced birefringence and surface deformation of the exposed glass. Laser-induced optical absorption leads not only to reduced optical transmission, but also to lens heating. This promotes additional index changes and deformation of the lens elements.

Research at Corning Inc. has led to the development of predictive models for the performance of fused silica lens material after long-term exposure to deep-UV radiation. Although these models address exposure trends for a specific grade of fused silica glass, they are representative of the response of fused silica materials in general.

Color-center formation

Exposure of fused silica to ArF and KrF laser radiation gives rise to the formation of color centers, specifically E' centers (undercoordinated silicon atom, ≡Si•) and nonbridging oxygen hole centers (NBOHC, ≡Si-O•).¹ Both types are created by ArF and KrF laser light, but, because the absorption band of nonbridging oxygen hole centers (peaking at 260 nm) does not extend to 193 nm, only E' centers (peaking at 215 nm) are relevant for laser-induced absorption at the ArF wavelength (193 nm). The laser-induced absorption at the KrF wavelength (248 nm), on the other hand, mostly results from formation of nonbridging oxygen hole centers.

The models account for the following reactions and processes:² Absorbed laser radiation leads to the formation of excitons in the glass. Although most of these excitons simply decay — returning the material to its ground state — others become trapped in localized states called precursors. Each precursor decays to form an E' center, as well as a nonbridging oxygen hole color center, both of which react with the molecular hydrogen in the glass to form SiH and SiOH, respectively.

During exposure of the glass, equilibrium develops between the two types of color centers on the one side and SiH and SiOH on the other. The color centers again react with hydrogen to form SiH and SiOH, which are convertible to color centers and hydrogen in a photo-induced reaction (photolysis). The relative number of color centers and SiH and SiOH, and thus the amount of absorption at 193 nm and 248 nm, is fluence-dependent because of the photolysis.

Eventually, after sufficiently long exposure, all precursors that were initially present in the glass react to form color centers, and the induced absorption levels off to a steady state. Because one of the reactions in the equilibrium is photolysis, the relative number of color centers — and thus the optical absorption in the glass — depends on the fluence to which the glass is exposed, even after the steady-state stage has been reached.

At the very low fluence levels used in projection lenses, typically 0.1 mJ/cm²/pulse or less, the number of laser pulses required to reach steady-state induced absorption in the fused silica studied can reach from tens to hundreds of billions. Some lens elements exposed to the lowest fluence levels in a projection lens, however, may never reach steady-state induced absorption within the expected lifetime of the equipment.

The Corning model parameters (rate constants) lead to agreement between model calculations and experimental data for the glass for both ArF and KrF wavelengths, and over a wide range of laser fluences and laser pulse frequencies.³ Although the parameter set is specific to the glass type studied, the models are general. Still, it is likely that the parameters could differ in other fused silica material because of contrasts in the manufacturing process.

A number of reactions in the glass contribute to marathon laser-induced absorption, although only two are photo-induced. Except for those, the same rate constants are used to calculate induced absorption at both wavelengths. These results support the validity of the model because both types of color centers contributing to induced absorption at these wavelengths originate from the same precursor.

Density changes

Two types of density changes occur in fused silica when it is exposed to deep-UV laser radiation. The first effect, generally known as densification, or compaction, is an increase in material density.⁴ The second, recently discovered during a laser damage study conducted by Sematech International at Cymer Inc. in San Diego, is expansion, or rarefaction, which signifies a decrease in density.⁵

Compared with compaction, expansion is significant only at very low fluences, which is why it was overlooked in earlier laser damage studies that applied higher fluences. The effect is thought to be the result of a radiation-induced process that forms β-hydroxyl (SiOH) in the glass.⁶ This process requires hydrogen; so, aside from laser fluence, hydrogen content is one of the key factors determining the amount of expansion in fused silica.

Compaction and expansion occur simultaneously in an exposed piece of glass, but the dominance of one or the other depends on glass parameters and exposure conditions. Nevertheless, both affect lens performance. For example, a change in density alters the index of refraction, induces surface deformation of the lens and incurs stress-induced birefringence. The magnitude of density effects is a function of the geometry of the glass element and the exposure pattern on the glass because unexposed glass surrounding the exposure region reduces the ability of the exposed glass to densify or expand.

The material property used to study density changes and to compare experiments is called the unconstrained density change. It is the change that one would observe in the absence of any constraining material surrounding the exposure region. The parameter can be calculated from the measured density change by using a conversion factor determined through finite-element analysis and by considering the elastic properties of the glass.

As described above, β-hydroxyl formation decreases the density and changes the chemical composition of the material, altering its index of refraction. However, these effects counter one another. Lower density decreases the index of refraction, whereas the β-hydroxyl formation increases it via a photorefractive effect.

In recent studies, this photorefractive effect became apparent, with contradictory results found in samples exposed under certain conditions. Specifically, birefringence measurements indicate a net decrease in density. Interferometric measurements of wavefront distortion in the same samples, however, show a net increase of the exposed material’s optical path length, which, in the absence of any other effects, indicate a density increase. This apparent contradiction can be understood only if there is a positive index change not associated with a density change, such as with a photorefractive effect. Only compaction and expansion contribute to the measured laser-induced birefringence. But three effects — compaction, expansion and photorefraction — contribute to the measured laser-induced wavefront distortion.

When examining net wavefront distortion in typical fused silica as measured interferometrically, expansion is dominant at very low fluence, and the wavefront inside the damage region is advanced. At high fluence, compaction and photorefractive effect are dominant, and the wavefront inside the damage region is retarded.

The distinction between expansion and the photorefractive effect is important because photorefraction is not caused by a density change and thus, unlike compaction and expansion, is not subject to the constraint of surrounding material. Therefore, without knowledge of the magnitude of the photorefractive effect, calculating any unconstrained density change from wavefront distortion measurements is not possible.

One technique for to separating density changes from the photorefractive effect is to measure the laser-induced birefringence in and around the exposure region. This birefringence is stress-induced; that is, the result of compaction and/or expansion, but not of the photorefractive effect. So the magnitude of induced birefringence, considering the same constraint from surrounding material as previously mentioned, can help determine the amount of unconstrained density change. Moreover, the pattern of the direction of the slow or fast axis of birefringence can help determine whether the net density change of the exposed material is positive or negative.

Understanding the type and magnitude of such laser damage effects in fused silica can facilitate development of material grades tailored for optimal performance in microlithography applications. A lens designer may be able to reduce the impact of these effects on lens performance. It is even conceivable that some of these effects could be made to counteract each other.

References

1. T.E. Tsai, D.L. Griscom, E.J. Friebele (1988). PHYS. REV. LETT., 61:444.

2. R.J. Araujo, N.F. Borrelli, C.M. Smith (1998). Proc. SPIE, 3424:2.

3. J. Moll (2001), Proc. SPIE, 4346:1272.

4. N.F. Borrelli et al (1997). JOURNAL OF THE OPTICAL SOCIETY OF AMERICA, B 14:1606.

5. C.K. van Peski, R. Morton, Z. Bor (2000). JOURNAL OF NON-CRYST. SOLIDS, 265:285.

6. C.M. Smith et al (2001). APPL. PHYS. LETT., 78:2452.

Meet the author

Johannes Moll is a senior development scientist at the Sullivan Park Research and Development facility of Corning Inc. in Corning, N.Y.