No abstract

In this paper, the potential for sub-10-nm junction formation of partial-melt laser annealing (PMLA), which is a combination of solid-phase regrowth and heat-assisted laser annealing (HALA), is demonstrated. HALA and PMLA are effective for reducing laser-energy density for dopant activation and for improving heating uniformity of device structure. The absence of melting at the dopant profile tail for PMLA results in a negligibly small diffusion at this region. A high activation rate is achievable by melting the upper part of the amorphous-silicon layer. The obtained sheet resistance of 10-nm-deep junctions was about 700 Ω/sq. for both n + /p and p + /n junctions. These results imply that PMLA is applicable for much shallower junction formation.

Laser annealing with nano-seconds pulse width is expected to be a useful tool for ultra-shallow junction formation. To compensate for the deep penetration depth of green laser light into Si, a metal absorber was placed on a specimen. The absorber was effective in reducing the laser energy density required to activate dopant. However, absorber formation that resulted in over-melt that increased junction depth easily occurred compared with specimens without the absorber. This problem was attributable to the absence of the increase in reflectivity by surface Si melting.

The application of plasma doping in three-dimensional (3-D) metal oxide semiconductor (MOS) transistors is proposed. One of the key steps in for fabricating 3-D MOS transistors with high-aspect-ratio channels is 3-D doping. Thus, plasma doping is analyzed and its profile is evaluated. Since conventional evaluation methods are not very effective for impurity concentration profiling along the sidewall of a vertical structure, a new method utilizing impurity-enhanced oxidation is proposed. By this method, an arsenic-doped layer of approximately 1 Â 10 20 cm À3 in concentration has been formed uniformly for a beam of 1 mm height and 0:2 À 1 mm width.

The depth profiling of sub-keV As þ implantation for sub-10 nm junction formation was investigated. The tail slope of profiles measured by secondary ion mass spectrometry (SIMS) using 300 eV Cs þ was much gentler than that of simulated profiles. Model calculation considering the mixing effect due to Cs þ bombardment during SIMS measurement explained the SIMS tail well. This implies that accurate As depth profiling by simply lowering primary Cs þ energy is difficult because of ion gun limitations. At the surface region, a 1.4-nm-thick surface layer was found. This layer is considerably thick for the sub-10 nm junction and is thicker than 0.4 nm for 5 keV implantation.