Structure of evaporated pure amorphous silicon: Neutron-diffraction and reverse Monte Carlo investigations

Atomistic Mechanism Underlying the Si(111)-(7×7) Surface Reconstruction Revealed by Artificial Neural-Network Potential

The 7×7 reconstruction of the Si(111) surface represents arguably the most fascinating surface reconstruction so far observed in nature. Yet, the atomistic mechanism underpinning its formation remains unclear after it was discovered sixty years ago. Experimentally, it is observed post priori so that analysis of its formation mechanism can only be carried out in analogy with archaeology. Theoretically, density-functional theory (DFT) correctly predicts the Si(111)-(7×7) ground state but is impractical to simulate its formation process; while empirical potentials failed to produce it as the ground state. Developing an artificial neural-network potential of DFT quality, we carried out accurate large-scale simulations to unravel the formation of the Si(111)-(7×7) surface. We reveal a possible step-mediated atom-pop rate-limiting process that triggers massive nonconserved atomic rearrangements, most remarkably, a critical process of collective vacancy diffusion that mediates a sequence of selective dimer, corner-hole, stacking-fault, and dimer-line pattern formation, to fulfill the 7×7 reconstruction. Our findings may not only solve the long-standing mystery of this famous surface reconstruction but they also illustrate the power of machine learning in studying complex structures.

Shape-fitting analyses of two-dimensional X-ray diffraction spots for strain-distribution evaluation in a β-FeSi2 nanofilm

New fitting analyses for peak shapes in a 2D reciprocal-space map are demonstrated to evaluate the strain, strain distribution and domain size of a crystalline ultra-thin (15 Å) film of β-FeSi2(100) grown epitaxially on an Si(001) substrate, using grazing-incidence X-ray diffraction. A 2D Laue-fit analysis taking into account instrument broadening and the double-domain effect provides residual maps as a function of the inequivalent strains ɛ b and ɛ c along the b and c axes of β-FeSi2, respectively (and domain size D), reflecting the probability of existence of homogeneous domains with fixed ɛ b , ɛ c and D, in addition to the most probable minimum residual. A 2D Laue fit with an inhomogeneous domain distribution provides a population map with ɛ b and ɛ c , reflecting strain components contributing to the film. The population map also leads to a reference residual as a guide for the strains contributing to the residual map. The advantages of the 2D Laue fits are discussed by comparison with the Scherrer, Williamson–Hall and Gaussian fitting methods for equivalent systems. The analyzed results indicate that the β-FeSi2 nanofilm was considerably small strained, which was also confirmed by transmission electron microscopy, implying a weak interface interaction between the film and the substrate.

Extension of the invariant environment refinement technique + reverse Monte Carlo method of structural modelling for interpreting experimental structure factors: the cases of amorphous silicon, phosphorus, and liquid argon

The invariant environment refinement technique, as applied to reverse Monte Carlo modelling [invariant environment refinement technique + reverse Monte Carlo (INVERT + RMC); M. J. Cliffe, M. T. Dove, D. A. Drabold, and A. L. Goodwin, Phys. Rev. Lett. 104, 125501 (2010)], is extended so that it is now applicable for interpreting the structure factor (instead of the pair distribution function). The new algorithm, called the local invariance calculation, is presented by the examples of amorphous silicon, phosphorus, and liquid argon. As a measure of the effectiveness of the new algorithm, the ratio of exactly fourfold coordinated Si atoms was larger than obtained previously by the INVERT-RMC scheme.

Bulk materials made of silicon cage clusters doped with Ti, Zr, or Hf

We investigate the feasibility of assembling the exceptionally stable isovalent X@Si(16) (X = Ti, Zr and Hf) nanoparticles to form new bulk materials. We use first-principles density functional theory. Our results predict the formation of stable, wide band-gap materials crystallizing in HCP structures in which the cages bind weakly, similar to fullerite. This study suggests new pathways through which endohedral cage clusters may constitute a viable means toward the production of synthetic materials with pre-defined physical and chemical properties.

Possibility of reverse Monte Carlo modelling for hydrogenated amorphous Si deposited on reactive ion etched Si substrate

We examined the x-ray diffraction (XRD) patterns of hydrogenated amorphous Si (a-Si:H) and of crystalline Si (c-Si) substrate for high-Q measurements. A structural analysis of thin films on substrates is important for the development of real devices. A transmission geometry with high-energy x-rays was used for this investigation, together with very thin substrates, in an effort to reduce substrate signals. A small area of the substrate was etched using the reactive ion etching (RIE) plasma process to maintain free-standing structures, and a-Si was deposited using catalytic chemical vapour deposition techniques. The x-ray beam was focused on the processed area and a-Si diffraction using a thin Si layer was measured. Unlike a-Si:H films on substrates without etching, we succeeded in detecting amorphous signals from samples deposited on the processed substrate. Application of reverse Monte Carlo (RMC) modelling using these data and subtracting Si substrate peaks was investigated. Direct subtraction and MCGR program (Pusztai and McGreevy 1997 Physica B 234-236 357-8) normalization for the ratio estimation between c-Si and a-Si:H structure factors was employed. MCGR normalization was found to improve subtraction of the c-Si peaks and the first peak at r = 2.3 in the pair distribution function g(r) could be calculated.

Characterizing the nature of virtual amorphous silicon

Virtual samples of approximations to real amorphous silicon, a-Si, have been prepared using several different processing routes. These include a fast quench from the melt followed by a long slow annealing period using molecular dynamics, a Reverse Monte Carlo approach, and an ab initio minimization. The characterization of these virtual a-Si samples includes a consideration of structural data (the radial distribution function, angular order, etc.), electronic properties (through the density of states), and thermodynamic information (chiefly the nature of the phase transformation from a-Si to liquid). The properties of a-Si are compared to network models, via the continuous random network model, and to experiment. We investigated the stability of virtual a-Si and consider its implications for use in future simulation studies. We have demonstrated the necessity for the accuracy provided by ab initio-based models to describe the interatomic potentials. Throughout this study, we have monitored the role of order in determining physical properties, as characterized by traditional routes (such as angular correlations) and more novel ones (the signature cell method).