|AVS 55th International Symposium & Exhibition|
|Plasma Science and Technology||Tuesday Sessions|
|Session:||Fundamentals of Plasma-Surface Interactions I|
|Presenter:||J.J. Végh, University of California, Berkeley|
|Authors:||J.J. Végh, University of California, Berkeley
D.B. Graves, University of California, Berkeley
|Correspondent:||Click to Email|
A firm understanding of fundamental etch limitations is becoming increasingly important as the scale down of feature sizes continues in the manufacture of semiconductor and other thin-film devices. Molecular dynamics (MD) simulations have been conducted to model the formation of small (~2-3 nm) features in silicon, both through the use of confined beams of ions and radicals and through exposure of a substrate to ions and radicals through an explicit masking layer. We compare simulations using an amorphous carbon mask on top of the silicon substrate to other simulations assuming a perfectly confined beam of bombarding species (i.e. to mimic a mask) on the same geometry (~2 nm wide trenches). The presence of the masking layer strongly affects the overall etch process and the minimum achievable feature size. For example, material from the mask is seen to sputter into the feature, where it mixes with the substrate material, and subsequently affects the etch yield and chemistry. Likewise, material from the substrate is seen to sputter and redeposit on the sidewalls of the mask, affecting sticking and transport of subsequent incident species. Ion scattering off the walls of the mask also affects the etch process by altering the angle of incidence and energy of the ions that hit the substrate. For certain bombarding chemistries, the masking layer shows severe degradation and loss of structural fidelity. We illustrate how this loss of fidelity in the masking layer transfers to the underlying substrate material on the small scales examined. The role of sidewall passivating radicals vs. the role of ions (i.e. CFx at 300 K vs. 200 eV CFX+ ions) is also examined: data on sticking coefficients and scattering probabilities are presented and the relative contributions from ions and radicals to the final sidewall composition are elucidated. Ion bombardment alone is sufficient to form a ~1 nm thick damaged/passivation layer along the sidewall, but deposition from radicals also plays an important role in determining the ultimate sidewall structure as the features deepen. We also discuss the challenges of extending MD to include larger and more realistic systems and other important effects, including substrate charging, long-timescale diffusion, and alternative masking materials such as SiC, etc.