Issue



Photoblanks for advanced lithography


10/01/2003







Photomask specifications for advanced KrF and ArF lithography tools require improvements in both glass substrate quality and coating development. These improvements are being driven by several factors, including the demanding CD requirements of advanced photomasks and the increased use of phase-shift masks (PSMs).

For the 100nm and below nodes, the ITRS roadmap requires CD uniformity <10nm for the most critical layers. CD control and uniformity will be impacted by optical properties of the mask substrate (transmission, index, and birefringence uniformity) in addition to the mask-manufacturing and lithographic properties of the attenuating layer(s). Advances in coating technology are needed in order to realize attenuating layers for photomasks that will address issues with the current chromium-based absorber for binary masks and be useful for various types of PSMs.

Substrates

The effects of stress birefringence and refractive index homogeneity on the image quality and productivity of optical lithographic step-scan systems has been reported [1, 2]. Wang measured high levels of residual linear birefringence in samples of fused silica mask substrates (6 in. ¥ 6 in.; thickness: 0.25 in.) [1]. In another evaluation of mask substrates, Kasprowicz and Priestley determined that stress birefringence and refractive index homogeneity could affect image quality and productivity (through dose nonuniformity) [2]. Due diligence in the selection of mask substrates is required for advanced reticles.

Click here to enlarge image

null

Several photomask substrate suppliers already have low birefringence products, including Corning, which has developed an advanced photomask substrate material specific to the 248nm and 193nm microlithography technology nodes [3]. The new substrate material is more homogeneous across the area of the piece in both transmission and refractive index with typical properties including birefringence <1nm/cm and an index of refraction homogeneity of <4ppm.

Coatings

Materials and processes are also needed for the preparation of new absorber and phase-shift coatings for future technology nodes (≤0.10µm). Requirements for these coatings include specific optical properties at exposure and/or laser-patterning wavelengths, in addition to certain physical and chemical attributes necessary for mask manufacturing (see table).

The traditional absorber coating for binary masks is a chromium-based thin film with modified surface layers to control reflectivity at the stepper wavelength. A bilayer structure consisting of CrxNy and CrOxNy is used. The top layer is an antireflection (A/R) layer, while the CrxNy layer is the attenuator film with high optical density. There are questions as to whether this chromium-based absorber coating that has been tuned and re-optimized for a number of years will be a viable coating going forward, however.

Issues with the current chromium-based film that will impact advanced optical lithography include challenges in developing optimized etch processes, repairability, and antireflectivity. Recently, International Sematech (ISMT) held a Chrome Absorber Workshop, where the issues and extendibility of the chromium-based absorber were discussed and debated. The general consensus was that, although chrome can continue to be used in the near term, enhancements are needed and a new material solution is likely needed to meet the challenges posed by shrinking feature sizes.

One enhancement to the conventional binary blank for ArF lithography that is being pursued for the 90nm node is a reduction in the Cr absorber film thickness. Conventional Cr blanks are 100nm in thickness. Using thin (59–73nm) Cr photoblanks, researchers have reported excellent results, including reduced reticle process bias due to etching (<50% of conventional ones) and a 50% decrease in etching bias variations.

Another enhancement being pursued for binary mask applications is the use of specialized A/R coatings to minimize the reflectivity of the chromium film at exposure wavelengths of 193nm and 157nm. Using an A/R coating structure based on a three-layer Fabry-Perot structure, reflectance of <1% has been achieved for a chromium film. It is also necessary to optimize the reflectance of the chromium blanks at the wavelengths used for laser pattern generation.

The use of a three-film stack has been proposed to lower the reflectivity of photoblanks for laser patterning and to reduce resist footing. With this approach, a layer of chrome is used for absorption, a layer of chrome-rich chrome oxide is used to reduce the reflection at the pattern generator wavelength and stepper exposure wavelength, and, finally, a very thin layer of stoichiometric chrome oxide is used to prevent chemical interaction with the resist.

Ultimately, an alternative to the chrome absorber may be required. New absorber coatings are indeed being developed for eventual use with the smaller technology nodes. Some industry experts have proposed the use of EUV absorbers such as TaN in order to capitalize on the knowledge gained in the development of EUV blanks. Another material that has been proposed is silicon, given that its etching properties have been widely studied in the IC industry.

The semiconductor industry has become dependent on advanced photomask technologies, such as attenuated phase-shift masks (attPSMs, also known as "soft shifters") to create smaller semiconductor devices. Soft PSMs are expected to be used (along with other lithography extensions) down to the 45nm node. The attPSM utilizes a single embedded layer/coating that allows 5–30% transmission while imparting a 180° phase shift of the light.

AttPSMs are advantageous to alternating aperture phase-shift masks (AltPSMs), which require etching of the fused silica mask substrate to create regions of phase shift. AltPSMs are also more difficult to design and manufacture.

To create an attPSM, a photoblank that contains an attenuating, phase-shift layer between the fused silica substrate and the top chrome layer is required. This embedded, attenuated photoblank is then a bilayer substrate, which tends to have a higher defect density than conventional binary masks
eticles. The most common material used for KrF lithography is molybdenum silicide (MoSiOxNy). The MoSi has a tendency to flake off the edges of the substrate and redeposit during chrome deposition or resist coat. This affects the manufacturing yields of the photoblank as well as the in-process yield for phase-shift masks.

In addition to increased defects, there are also problems with current phase-shift coatings, such as exposure durability and chemical durability to the cleaning process. Phase, transmission, and uniformity of the embedded phase-shift material must be maintained in the production of these blanks and during the use of the final reticle (see table for additional specifications). Some of the most promising candidate materials include TaSiOx, Si-based coatings such as TiN/SixNy and TaN/SixNy, and CrAlOxNy.

Conclusion

The ability to provide a high-quality photomask substrate with low birefringence and improved transmission uniformity has been met by Corning and others. There are still more challenges to overcome in the area of material developments for specialized absorber and phase-shift coatings needed to achieve the high pattern fidelity and precise CD control in reticle fabrication essential for 90nm-node lithography. To support subwavelength lithography with RETs such as phase shifting, materials will be needed with transmission requirements that vary between 5–30% while meeting mask fabrication requirements and lithographic performance.

Robin Walton, Corning Inc., Corning, New York

References

  1. B.B. Wang, "Residual Birefringence in Photomask Substrates," J. Microlith., Microfab., Microsyst., Vol. 1, No. 1, pp. 43–48, 2002.
  2. B.S. Kasprowicz, R. Priestley, "Evaluation of New Mask Materials for Improved Lithography Performance," SPIE, Vol. 4346, pp. 827–830, 2001.
  3. R. Priestley, D. Sempolinski, C. Yu, "Photolithography Method, Photolithography Mask Blanks, and Method of Making," US Patent 6475682 B2 and US Patent 6410192 B1.

For more information, contact Robin Walton at Corning Inc., HP-CB-06-3, Corning, NY 14831; [email protected].