Structural characterization

Transmission electron microscopy (TEM) is an experimental technique routinely used to get insight on the particular structure of extended defects at the atomic level. However, due to the resolution limit of TEM, the structure of small defects (under 40 atoms) cannot be resolved experimentally. In such circumstances atomistic simulation techniques become a good alternative to characterize defects. Ab initio and tight-binding techniques can be used to extract useful information such as bond lengths and angles in small defect clusters. Classical molecular dynamics can be used for bigger defects, even at sizes that can be resolved by TEM. In such cases, the comparison of simulation results and TEM images can serve to validate structural defect models.

We developed an atomistic model for the formation of extended {113} defects in silicon. Our model served to validate a novel structural characterization technique known as unprocessed high-angle annular darkfield scanning TEM. We have also used classical molecular dynamics techniques to study the imperfect regrowth of FinFET devices. TEM images show the generation of line defects in the FinFET body and the formation of polycrystalline material, which degrade its performance. Our simulations allowed to relate the formation of such line defects during regrowth of the FinFET body with the particular orientation of the growing amorphous-crystal interface, and even to give technological clues about how to improve regrowth.

(left) TEM image of a FinFET after regrowth (from R. Duffy et al., Appl. Phys. Lett. 190, 241912 (2007)), and (right) classical molecular dynamics simulation results (L. A. Marqués et al., J. Appl. Phys. 111, 034302 (2012) - Molecular dynamics simulation of the regrowth of nanometric multigate Si devices).

TEM image (a) is compared with the structural model obtained in the simulation (b) to characterize a planar {113} defect (c) (K. J. Dudeck, L. A. Marqués et al., Phys. Rev. Lett. 110, 166102 (2013) - Sub-ångstrom experimental validation of Molecular Dynamics for predictive modeling of extended defect structures in Si).