The Bridgman-Stockbarger method of growing CaF2 has been around for decades. Now that the semiconductor industry has an escalating need for the material a reported "must have" for lens systems in 157nm lithography tools debate has focused on the adequacy of supply and the technique used to obtain it.
"Current technology cannot make CaF2 crystals of sufficient purity and yield to make them economical for mass production of semiconductor or other high-tech products," states Kiril Pandelisev, CEO and chief science officer at Single Crystal Technologies (SCT).
Figure 1. SCT's purification apparatus.
SCT proposes a new crystal-growing method that uses a flat plate or slab (Fig. 1) rather than the traditional round ingot (Fig. 2). Being able to control impurities is key to the performance of 157nm lens systems. Impurities in the crystal affect softness/hardness, which in turn affects creep, as well as built-in stress. (Lenses change shape "creep" over time due to the force of gravity.) Lead fluoride is used to control oxygen in 193nm optics, but because lead has an absorption band at 157nm, it cannot be used at that wavelength. Furthermore, rare earth metals form color centers in CaF2 that start to fluoresce.
Figure 2. Bridgman-Stockbarger method. (Source: Single Crystal Technologies)
Pandelisev explains that in the Bridgman-Stockbarger method the interface separating the solid section of CaF2 from the liquid CaF2 is not a flat surface but forms a shape similar to a meniscus. As the crystal grows, the properties of the liquid change as well. The center of the melt becomes dirtier because it solidifies later than the edges, and over time the melt properties change for the worse.
"The yield is very low," says Pandelisev. "Out of about 100 crystals, you might get only three good ones. Built-in stress due to uneven heat distribution becomes 'frozen' in the crystal. The result is birefringence."
Because the technique developed by SCT has the material in the form of a slab, heat is dissipated faster, and as a result the crystal has lower built-in stress and less birefringence. "The distance between the center of the plate and the outside surface of the plate is only about 2.5 in.," Pandelisev notes. "In the traditional cylindrical boule, the heat barrier [the distance over which heat is dissipated] is about 8 in."
In June, the company sent samples to Los Alamos and Sandia National Labs for purity analysis, and to MIT's Lincoln Lab for other quality testing. Sure to cause discussion is Pandelisev's claim that high-purity LiF can be produced by the company's technology.
"Our technology can open the doors to 121nm lithography that can enable the 30nm technology node," he says. "Ternary and/or quarternary LiF-based compounds might extend the transmission to even shorter than a 121nm wavelength and smaller than 30nm device technology. The current efforts on EUV technology based on very high COO equipment might be obsolete from their inception." D.V.