In Search of Calcium Fluoride

Konrad Knapp

Anyone who has ever owned a PC is familiar with the problem of technological obsolescence. Faster models hit the market almost before people have had the opportunity to turn on their new computers. What many people don’t know is that rapid progress in the fabrication of semiconductor circuit components can largely be attributed to advances in optical microlithography, one of the key technologies for creating transistors and memory modules on silicon wafers.

CaF₂ crystal is a key material for microlithography in the production of high-end microchips.

Improved microchip performance requires optics that can generate ever-finer features. Advances in miniaturization are thus closely related to the spatial resolutions of the optical systems employed for microchip fabrication. These finer details can be obtained in two ways: by using either larger lenses or shorter wavelengths. The problem with larger optics is that the number of defects typically increases when the overall size is increased. And with shorter wavelengths, the traditional lens material — fused silica (glass) — behaves poorly because of radiation effects.

Today, state-of-the-art microlithography technology has progressed to the point that ever-shorter wavelengths are being used to “write” features on microchips.

193 to 157 nm

In fact, the latest generation of microlithography tools employs excimer lasers that operate at 193 nm, a wavelength that allows writing features as narrow as 100 nm. Several manufacturers, including Schott Lithotec, Corning and Nikon, are dedicated to making this technology available for mass-production applications in the microchip industry. Implementation of 157-nm microlithography will lower the limit on microchip features to as small as 70 nm. Fabrication of the first semiconductor circuit components that use this technology will most likely begin late this year.

Schott Lithotec produces CaF₂ crystals with a disc diameter of up to 350 mm and a thickness of more than 100 mm.

Microlithography at 157 nm represents a milestone in closing the technology gap between conventional fabrication of semiconductor circuit components, which employ photo?lithographic methods, and extreme-ultraviolet microlithography, which uses wavelengths from 11 to 13 nm. However, neither glass nor quartz is transparent enough for 157 nm. The preferred material here is extremely pure, low-defect, single-crystal calcium fluoride (CaF₂), also known as fluorite.

Although barium fluoride has been considered as an alternative to calcium fluoride, a number of technological hurdles have not yet been resolved. No broadly accepted material can withstand irradiation by short-wavelength UV as well as CaF₂. When used as lenses or prisms, this material can concentrate and deflect UV radiation down to about 130 nm. Its refractive behavior should be as uniform as possible to preserve the quality of chip structures. Another important material requirement is low stress birefringence at application wavelengths.

Octahedrons and tetrahedrons are typical structures of crystals made of calcium fluoride.

Challenges

Demand for CaF₂ is expected to rise sharply over the next three to four years. The large quantity and high quality of the material required call for a good understanding of the crystal growth process. Present-day crystal growth technologies are usually based on the phase transition from the liquid to the solid state, known as melt growth. Only a few materials of industrial importance, such as the semiconductor silicon carbide, are grown from the gas phase.

The challenge to using CaF₂ is that it cannot be used as a lens material in its initial form. To grow CaF₂ in single crystals, the raw material must be purified so that the contaminants are decreased to the parts-per-billion level.

This extremely pure powder is the starting point, and what follows is a multistep process. The synthetic CaF₂ powder is molten and compressed into a single crystal inside a crystal growth furnace. Finally, the single crystal is subjected to an annealing process that results in even higher quality.

The fundamental principles of the melt growth technologies for the manufacture of single crystals were established in the first half of the 20th century. The most important technique for the manufacture of semiconductor and oxide crystals used today is what is known as the Czochralski method.

In this technique, a crucible of molten material is prepared, and a seed crystal is touched to the surface of the liquid. Surface tension causes the molten liquid to cling to the seed crystal as it is slowly raised. As the liquid is pulled up, it cools and solidifies, forming a single crystal with the seed crystal.

For CaF₂ production, however, the polycrystalline raw material is molten in a crucible and then directionally solidified from a single-crystalline seed at the bottom of the crucible. The advantage of this technique over the Czochralski method is that the temperature conditions can be better controlled during the growth process. As a result, crystals can be grown with lower defect density. The drawback is that the crystals grow in contact with the crucible wall and cannot be observed, which can reduce the yield.

During the crystal growth process, careful attention must be paid to the temperature distribution. As in glassmaking, the molten mass must be cooled in a controlled manner to prevent areas of thermal tension that could lead to defects. Yet another problem that often arises is “compaction,” a change of transmitted light resulting from localized density variations that cause spatial variations in refractive index.

Researchers and developers are using what is known as defect engineering to help grow larger crystals and, at the same time, improve their quality in the microscopic or even the atomic range. This is achieved by precisely controlling the growing conditions. The solution strategy takes two routes. First, the crystal growth process is carefully analyzed to discover the relationship between the important growth conditions and to draw up a defect model. Second, a process model is needed that places the growth parameters (i.e., those elements of the process that can be directly manipulated) in relation to the conditions — mainly the temperature field. This is achieved by the combined use of experimental analysis and computer simulation.

The Crystal Growth Laboratory in Erlangen, Germany, has worked closely with Schott Lithotec toward developing furnaces for growing CaF₂ crystals. In addition, the laboratory has built a prototype furnace that will provide a valuable learning experience regarding crystal growth.

The ongoing challenge for any calcium-fluoride producer is to achieve the requirements of lens blanks with large dimensions, while reducing the wavelength and increasing the numerical aperture.

Meet the author

Konrad Knapp is vice president of business development at Schott Lithotec AG in Mainz, Germany, where he is responsible for next-generation lithography.