Clustering and Extended Range Order in Binary Network Glasses
- PDF / 466,980 Bytes
- 6 Pages / 414.72 x 648 pts Page_size
- 105 Downloads / 195 Views
simulations from random configurations, the amount of charge localized at the position of each atom is updated every time step, based on the number of neighbors within a cut-off distance and an empirical charge transfer function of the following form: k = Cij e-(rij/aij)4, where aij is a constant chosen such that the function vanishes within the cut-off radius, and C4 represents the amount of charge transferred between two bonding atoms. The parameter Cj was adjusted so that the same partial charges on silicon and oxygen are generated as were used in previous work (approximately +2.3 for Si and -1.15 for O).(9) The new form of potential was tested for silica glass and crystalline structures for experimental data mentioned above.(10 -t 1) RESULTS AND DISCUSSION In the following we present data obtained for simulated binary alkali-silicate structures, including a total of 648 atoms, and having the general composition (M20)x(SiO2)1-x, where x ranges between 0 and 0.4. The behaviors of two alkali cations, which differ from each other significantly in terms of size and electrostatic field strength, were studied, i.e., sodium and rubidium. For each composition, both systems were started from the same random configuration. The reason for this was to ascertain the reality of any differences in the final equilibrated structures. The systems were thermalized at 8000 K for 10 ps and cooled to 300 K over 100 ps (50,000 time steps at 2 fs). Structural analysis was performed on the final room temperature configurations. Direct observation of the structures Direct observation of the structures provides a first qualitative impression and is very useful for guiding the choice or design of the statistical analysis to be performed. The use of stereoscopic visualization improves the clarity of this initial examination. Representative snapshots of a sodium silicate and a rubidium silicate are shown in fig. 1. Careful inspection reveals that in (Na2O)x(SiO2)t-x, even at small x (between 0.05 and 0.1), Na ions tend to group in pairs. The positive charge of such pairs is neutralized by that of two negative non-bridging oxygens, which is consistent to MRN model. Rb ions, on the other hand, have a much lower tendency to cluster. They appear to occupy random positions throughout the network structure, and their charge is neutralized by bridging and non-bridging oxygens, arranged within a wider spread of distances from the cation. In the range of x larger than 0.10 clustering in sodium silicates intensifies. The cations form groups of more than two and seemingly align in channels surrounded by non-bridging oxygens. As will be confirmed below by ring statistics, the network apart from these clusters is relatively undisturbed. But as x increases, the effective volume of bulk silica network decreases, and for x>0.30 extended network segments with a structure similar to that of pure silica glass are no longer obvious. For the same range of x in rubidium silicates, proximity between alkali cations is inevitable due to their large concentrations. C
Data Loading...
