Evolution of structural topology of forming nanocrystalline silicon film by atomic-scale-mechanism-driven model based on realistic network

Yixiong Zhang,Shaoji Jiang,Hong Wang

doi:10.1063/1.5048674

Abstract

To establish a description of realistic structural evolution of a growth film, we propose a local definite continuous-random-network (CRN) structure combined with a kinetic Monte Carlo (KMC) method based on an atomic-scale mechanism from first-principles density-functional-theory computations and molecular-dynamics computations. The proposed CRN-KMC method elucidates the evolution of elaborate topological structure and the transformation from amorphous phase to nanocrystalline phase of Si films, which is essentially attributed to the atomic interactive behavior of film growth. The method further predicts the realistic structural networks of a growing film at various temperatures based on various atomic-scale mechanisms competing with each other, mechanisms that not only essentially drive the radical from physisorption to chemisorption with the film surface, but also decidedly influence the film-surface chemical composition. In particular, we find the evolution of topological structure’s critical dependence on the compositions of the film surface and H-induced crystallization mechanism, which provide the important information for the strategy for determining optimized deposition conditions for local crystal formation. The results of the evolution of the structural network indicate that the structure of film is similar the CRN model’s representation at relative lower temperature, and is in full agreement with the inhomogeneous crystalline model at relative higher temperature without an abrupt phase change from polycrystalline to amorphous. Our CRN-KMC realistic structure model has significance for exploring the relation of various atomic-scale mechanisms to the phase transformation of growing films.

Highlights

The ideal continuous random network (CRN) for an amorphous Silicon (a-Si) structural network can well reproduce experimental reduced density functions (RDFs) obtained by diffraction; the model is inconsistent with the variance data of medium-range order (MRO) from fluctuation transmission-electron-microscopy experiments.[9,10,11]
A lattice model combined with the kinetic Monte Carlo (KMC) method has been presented for simulating the CVD of thin films, but this model is seriously limited by the discrete site for atomic accommodated position; the model cannot represent the elaborate structural network of amorphous Si.[14]
In order to establish the description of realistic structural evolution of a growth film, we propose a local definite continuous network structure combination with kinetic Monte Carlo (CRNKMC) method based on the atomic-scale-mechanism-driven film-growth model.[10,18,19,20,21,22,23]

Summary

Introduction

The crystallization mechanism of Si films and evolution topological structure of either the paracrystalline silicon (pc-Si) and nanocrystalline silicon (nc-Si), or microcrystalline Si transformation from hydrogenated amorphous Si in plasma-enhanced chemical vapor deposition (PECVD) with dilution in H, have attracted great interest in the semiconductor materials field for wide utilization in, e.g., thin-film transistors, display devices, memory devices, and thin-film photovoltaic devices, as these films have novel optical and electronic properties.[1,2,3] these wide applications of Si microcrystals are based on their material advantages, including greater doping efficiency, higher electrical conductivity, higher carrier mobility,[1] better quality, and better light stability[2,3] compared to hydrogenated amorphous Silicon (a-Si:H). The ideal continuous random network (CRN) for an a-Si structural network can well reproduce experimental reduced density functions (RDFs) obtained by diffraction; the model is inconsistent with the variance data of medium-range order (MRO) from fluctuation transmission-electron-microscopy experiments.[9,10,11] In addition, a reverse Monte Carlo (RMC) scheme was developed for the computer generation of amorphous structures.[12,13] the well-known RMC is an experimental data-constraint reversion model that cannot describe the deposition growth mechanism of the films Such a model must be based on the experimental data of structural or electronic properties, such as the pair-correlation function, fluctuation transmissionelectron-microscopy variance, and vibrational density of states, as well as the electronic density of states. Physical insight into the transition mechanism of amorphousto-microcrystalline Si and the atomic topological configuration feature in a-Si film different growth mixed phase is still lacking.[9,10]

Objectives

Results

Conclusion