Litho Faces Daunting Materials Hurdles
Alexander E. Braun, Senior Editor -- Semiconductor International, 1/18/2008 5:54:00 AM
A breakthrough may be coming in the field of high-index lens materials needed for post-32 nm immersion lithography, but the lithography field remains filled with materials hurdles, said Bryan Rice, Sematech’s (Austin, Texas) immersion lithography program manager.
 |
| Bryan Rice, Immersion Lithography Program Manager, Sematech |
Outlining the materials challenges for lithography beyond the 32 nm node, Rice observed that most are using high-index, water-based immersion systems with a numerical aperture (NA) on the order of 1.3-1.35. “We’re converging on using some form of double exposure, double patterning, together with 1.3+ water-based immersion tools,” he said. “Papers are being published about processes using that technology for 32 nm half-pitch. After that, however, everything gets dark.”
The industry prefers to extend a technology instead of going to something new — 193 nm litho is no exception. “The NA approachable with water is limited to 1.35. A new lens material would get past that,” Rice said. “I think that 1.55 is a nice round number, because with a double-timing solution, it’s not much more aggressive than what we are capable of doing today to reach the 22 nm half-pitch. The theoretical maximum, based on angles, lens design and materials — 1.7 — is the best that can be attained with known lens materials. If you do double exposure with 1.7 NA, you could get in the neighborhood of 16-18 nm.”
Rice referred to three generations in lithography. Generation 1 is water-based immersion. Generation 2, chosen because of a commonly occurring organic fluid that has a refraction index of 193 at 1.65, could be combined with a lens material that has a higher index than fused silica. The would enable a 1.55 NA, which Rice said would be capable of a 22 nm half-pitch. Generation 3 is “high-index immersion’s Holy Grail. It requires a 1.8 refraction index fluid. With a good lens, it’s the key to 1.7 NA,” he said.
A seminal study by the National Institute of Standards and Technology (NIST, Gaithersburg, Md.) determined which materials might have sufficiently high refraction indices. The field was narrowed to lutetium aluminum garnet (LuAG), ceramic spinel and pyroxene. Ceramic spinel is not easily commercialized, and pyroxene must be formed under extreme pressure. LuAG is the best candidate and can be grown like the standard crystals used for semiconductor applications.
“We grew a 300 mm LuAG sample,” Rice said, “which wasn’t as highly transmissive as we had hoped. However, we understand the problems and expect a breakthrough in the next couple of months, resulting in a singularly good new lens material.”
The second category of materials needed is immersion fluids. “We have Generation 2 fluids,” Rice said. “They’re readily available and can be produced in bulk. The difficulty is keeping them clean. When you shine light through them they darken; however, this can be corrected through filtration. Generation 3 isn’t that simple, and every single material seems to fall short. We concluded that single-compound materials weren’t going to work and turned to composite materials.”
The Sematech researchers took a high-index inorganic core and dispersed it into a fluid to achieve the required properties. “We found a material that can be made into spherical nanoparticles with a 2.9 refraction index at 193 nm,” Rice said. ”Water is 1.44, and we thought it’d be possible to put enough of these nanoparticles to raise it to 1.8. If it worked, it would result in an aqueous solution with nanoparticles that could be used in an immersion tool — a water-based fluid that doesn’t photochemically darken because it isn’t organic.”
However, a 37% volume fraction would be needed, with a viscosity that would create engineering problems, plus creative solutions to get it on — and off — the wafer. For a Generation 2 fluid, the volume fraction could be reduced by a factor of two and that is doable. “We have nanoparticles around 3 nm in size, we can put them in loading fractions of 10% and we are halfway to produce a Gen 2 nanocomposite fluid,” Rice said.
The last part of this triad is the resist. “This will require platform changes,” Rice said. “First, the lens and fluid materials must get to the point where suppliers and semiconductor manufacturers view it as a good approach. Then suppliers will have to take it seriously and work on high-index resists. This won’t be simple. Just adding materials that raise the resist’s refraction index can make it too dark and ruin its imaging properties.”
In collaboration with resist providers, Sematech worked on a nanocomposite resist. “We took the same nanoparticles we used for fluids and added them to the resist’s casting base,” Rice said, adding that it resulted in very smooth films with an absorbance of ~2.5/µm. Rice is also working on the problem of developing a double exposure resist material that could be exposed twice without having to send it to be etched between lithography passes, reducing processing steps. For this, materials with a non-linear response are needed; these exist, but the hurdle is in applying them to 193.
In extreme ultraviolet (EUV) lithography, Rice sees countless materials questions, such as those linked to source electrodes and high-reflectivity mirrors. “If we had a mirror that was 2% more reflective [considering that the light bounces off 10 or 11 mirrors in the optics system], it would greatly simplify some of the difficulties,” he said.
The way forward for lithography will be difficult, regardless of the choice of technology, he concluded.
