(446b) Atomic-Scale Analysis Of The Role Of Surface Cordination Defects In The Growth Of Amorphous Silicon Thin Films

Valipa, M. S., University of Massachusetts
Singh, T., University of Massachusetts - Amherst
Maroudas, D., University of Massachusetts, Amherst

Hydrogenated amorphous silicon (a-Si:H) thin films grown by plasma-assisted deposition from silane-containing discharges are technologically important semiconductor materials, used widely in the fabrication of solar cells and flat panel displays. Development of systematic strategies for depositing silicon thin films with desirable properties requires a fundamental understanding of the interactions between reactive radicals originating in the plasma and the growth surface. The dominant precursor for deposition of smooth device-quality a-Si:H films is the SiH3 radical. Growth of the a-Si:H film has been assumed to occur through a combination of surface H abstraction by impinging SiH3 radicals to create dangling bonds (DBs), and the passivation of these surface DBs by subsequently impinging SiH3 radicals. However, on a-Si:H surfaces, the DB surface coverage is low (~ 0.001), which is incompatible with the observed high film growth rates. Hence, the precise role of surface coordination defects in mediating the a-Si:H growth process remains unclear.

This presentation focuses on detailed atomic-scale analysis of the role of surface coordination defects in determining the growth mechanism of a-Si:H thin films. Using molecular-dynamics (MD) simulations of impingement of SiH3 radicals on growth surfaces of a-Si:H films, we studied the reactions with the growth surfaces of the SiH3 radical over the temperature (T) range 475 ≤ T ≤ 800 K and monitored the chemical composition of the a-Si:H surfaces. Our MD simulations employed a many-body interatomic potential that has been tested extensively by comparing its predictions with experimental measurements and accurate first-principles calculations based on density functional theory (DFT). In addition, we have carried out analysis of surface silicon hydride SiHx(s) (x = 2,3) dissociation pathways based on DFT calculations using crystalline Si surfaces as representative models of atomic bonding at amorphous film growth surfaces. In our DFT calculations, we used the generalized gradient approximation, plane-wave basis sets, ultra-soft pseudopotentials, slab supercells, and the nudged elastic band method for determining saddle points and optimal surface reaction pathways.

Our MD simulations reveal that the SiH3 radical is very mobile on the a-Si:H surface and diffuses with a low activation barrier. Furthermore, the diffusing SiH3 radical incorporates into the a-Si:H film only when it transfers an H atom and forms a second Si-Si backbond, resulting in the immobilization of the radical on the a-Si:H surface, which contributes to the growth of the a-Si:H film. Specifically, the H atom is transferred from the SiH3 radical only when the Si of the radical becomes five-fold coordinated with a floating bond (FB) and, more importantly, the H atom is transferred to a four-fold coordinated Si atom, which also becomes five-fold coordinated after H transfer, resulting in another FB. Such FB-mediated Si incorporation reactions have energy barriers of 0.35-0.5 eV, and occur frequently, leading to growth of the a-Si:H film even when the DB density is low. In contrast, DB-mediated Si incorporation reactions, where the radical either adsorbs onto a DB or dissociates by transferring an H atom to a surface DB site, occur in less than 10 % of the cases consistently with the low surface DB density. Based on our results, we conclude that the dominant mechanism for a-Si:H film growth is mediated by FBs and excludes the involvement of DBs. These results are consistent mechanistically with our DFT calculations on the H-terminated Si(001)-(2x1) surface, according to which FB-mediated Si incorporation from SiH3 radicals occurs with barriers of 0.4-0.55 eV. Our results also are consistent with experimental measurements of a-Si:H film surface composition, film growth rates, as well as the low surface DB coverage.