Reverse Monte Carlo modeling of amorphous silicon

Abstract

An implementation of the Reverse Monte Carlo algorithm is presented for the study of amorphous tetrahedral semiconductors. By taking into account a number of constraints that describe the tetrahedral bonding geometry along with the radial distribution function, we construct a model of amorphous silicon using the reverse monte carlo technique. Starting from a completely random configuration, we generate a model of amorphous silicon containing 500 atoms closely reproducing the experimental static structure factor and bond angle distribution and in improved agreement with electronic properties. Comparison is made to existing Reverse Monte Carlo models, and the importance of suitable constraints beside experimental data is stressed.

Quantification of the bond-angle dispersion by Raman spectroscopy and the strain energy of amorphous silicon

“…From these values calculating ⌬ is straightforward ͑⌬ = ͓͑⌬ G ͒ 2 + ͑ t − ¯͒2 ͔ 1/2 , where t = 109.47°is the tetrahedral bond angle͒ and the ⌬ values are almost identical to ⌬ G ͑⌬ G and ⌬ are also similar for the RDFs measured by Laaziri et al 5,6 commented below͒. Our analysis agrees with the claim 19 that large deviations of bond-angle distributions from Gaussian are due to the finite size of the microscopic models. Consequently, we should consider that the standard deviation of the Gaussian distribution used to fit the experimental RDF is nearly the rms of the actual distribution.…”

Inverse simulated annealing: Improvements and application to amorphous InSb

“…While this allows for an efficient and routine modeling of rather complex disordered structures, the resulting models are not necessarily physical sensible. Furthermore, since no information on the potential energy surface is exploited, a variety of different structural models that are very different from each other, but in similarly good agreement with experiment, can be generated [6][7][8][9][10][11][12][13] . Finite temperature Molecular Dynamics (MD) or Monte Carlo (MC) simulations offer an alternative route to generate glassy models by quenching from the melt using the simulated annealing (SA) algorithm 14 .…”

Realistic inversion of diffraction data for an amorphous solid: The case of amorphous silicon

“…The FEAR model has 96% four-fold coordination, with equal (2% fractions) of 3-fold and 5-fold Si. This is equal to the melt-quench model using environment-dependent interaction potential (96%) [34] and better than models obtained from other techniques [2,20,35]. The average coordination number of our model is 4 which deviates from that of the experimental annealed sample (3.88) [14].…”

“…The situation is different for non-crystalline materials. Evidence from Reverse Monte Carlo (RMC) studies [2][3][4][5] show that the information inherent to paircorrelations alone is not adequate to produce a model with chemically realistic coordination and ordering. This is not really surprising, as the structure factor S(Q) or pair-correlation function g(r) (PCF) is a smooth onedimensional function, and its information entropy [6] is vastly higher (and information commensurately lower) than for a crystal, the latter PCF being a sequence of sharply localized functions.…”