Orsolya Gereben

Possible quantitative measures of order/disorder in models of liquid and amorphous structures

The variance of the pair correlation function, the triplet correlation function, g(3)(r,s,t), and the three-body contribution to the configurational entropy, S3, were studied as possible tools for distinguishing between three different models of amorphous silicon (a-Si) with (nearly) identical pair correlation functions. All of these quantities were shown to be able to make the distinction unequivocally. For practical reasons, short-range S3 is suggested as the best quantitative measure of disorder, in terms of higher-order correlations. For the first time, various projections of the three-body correlation function of a complicated amorphous material are also shown. On the basis of the current study it is proposed that the most realistic structural model of a-Si is the reverse Monte Carlo fit, starting from the well known Wooten model.

Some remarks on the measured structure factor

Two interesting features of the experimentally obtained structure factor are discussed, using the S(Q) of a (model) amorphous silicon sample. In the first part a method is suggested for deciding if the measured data can correspond to a real three dimensional particle arrangement. In the second part it is shown that the Reverse Monte Carlo method is able to provide the correct pair correlation function even if the structure factor is available over only very limited momentum transfer range. The Q range can be as short as 0.4-6 A−1.

The Determination of the Microscopic Density in Liquids and Other Disordered Materials Using Reverse Monte Carlo Simulation

Abstract A procdure using the Reverse Monte Carlo technique was shown to find the correct microscopic density of scattering centers (atoms, ions, etc.) in a model liquid within about 2%, on the sole basis of diffraction data. The method was also tested on solid amorphous systems of low, as well as of high packing fractions. An amorphous tetrahedral network served as a model for the former, while for the latter a model of a metallic glass was used.

Semiempirical Fragment Self-Consistent Field Monte Carlo simulations for liquid chlorosilanes

Abstract We applied the recently developed Neglect of Diatomic Differential Overlap Fragment Self-Consistent Field Monte Carlo method to the simulation of the liquid state of chlorinated monosilanes. This semiempirical technique divides the periodic simulation cell into subsystem, where the random move of an atom takes place, and environment exerting only secondary effects on the former. Expanding the electronic wave function on the basis of atomic hybrid orbitals, that form strictly localised molecular orbitals corresponding to chemical bonds in classical molecules, the wave function of the environment is determined from coupled 2 × 2 secular equations. For the subsystem the conventional Self-Consistent Field equations, with a perturbation term in the Fockian and a much lower dimensionality than for the whole system, have to be solved. Thus the computational efforts drastically decrease as the dependence on the number of atoms in the environment reduces from quartic or cubic, as in conventional ab initio or semiempirical methods, to quadratic. Our simulation for chlorosilanes predicts that in the liquid state the preferred orientation of two neighbouring molecules is the one with the maximum number of SiH…ClSi hydrogen bonds. We found that gas phase SiCl bond lengths increase by 6 to 16 pm in the liquid state as a result of that association. HSiCl and ClSiCl bond angles change much less, by 2–4 degrees, as compared to the gas phase geometry. Since, to our knowledge, there are no experimental data published for these systems, our results may serve as preliminary information on liquid chlorosilanes.

Reverse Monte Carlo approach to the structure of amorphous semiconductors

It is shown that the use of reverse Monte Carlo simulation provides valuable contributions along the entire route from the experimental structure factor to three-dimensional structural models. In particular, the estimation of the microscopic number density and the evaluation of the pair correlation function is described for some elemental amorphous semiconductors. The importance of generating different atomic models of these materials using geometrical constraints is also emphasized.

On the determination of the structure of metallic glasses by reverse Monte Carlo simulation: comparison with hard sphere Monte Carlo results

Abstract Reverse Monte Carlo (RMC) simulation has recently been used for modelling the atomic level structure of several amorphous metallic alloys. Using the particle configuraitons that were provided by the RMC calculations, characteristics of the short-range structure were evaluated. Standard (Metropolis) Monte Carlo simulations on hard sphere systems mimicking three of the metallic glassy materials modelled by RMC have also been carried out. Particle configurations obtained by these two methods were compared. It was shown that dense random packing of hard spheres is most similar to “real” metallic glassy structures when the chemical nature of the components differs the least.

The like-like-like partial triplet correlation functions of amorphous Ni62Nb38

Abstract Projections of the partial three body correlation functions for Ni-Ni-Ni and Nb-Nb-Nb triplets in amorphous Ni62Nb38 are presented, in their full and also, in their reduced forms. The structural model of the material was the result of an earlier reverse Monte Carlo calculation [L, Pusztai et al., J. Non-Cryst. Solids 156-158 (1993) 973], It is shown that the three body correlations in metallic glasses cannot be represented solely on the basis of pair correlations, even though the structure of the material is determined mostly by steric considerations.