Melting of Aromatic Compounds: Molecular Dynamics Simulations
- PDF / 314,018 Bytes
- 6 Pages / 414.72 x 648 pts Page_size
- 76 Downloads / 252 Views
327 Mat. Res. Soc. Symp. Proc. Vol. 408 © 1996 Materials Research Society
the free energy of both of the coexisting phases as a function of the relevant field variable. Assuming a robust free energy scheme and simulation (MD) protocol, the accuracy is then limited primarily by the quality of the forcefield. Broughton and Li [17] showed that this method could give impressive results. Using the forcefield of Stillinger and Weber [43], they calculated a melting point for silicon that agreed with the experimental value to within 0.5%. Furthermore, the MD results of Levesque et al. [44] and free energy calculations of Frenkel [45] on 4He showed the importance of calculating the free energy to determine the thermodynamic stability of the different phases involved in the transitions studied by MD. However, this method has not frequently been used because it is relatively expensive in terms of computer time. Phillpot et al. [20] and Lutsko et al. [21] considered slabs of finite crystals and used the idea that even though the crystalline solid will melt at any temperature above the melting temperature, the speed by which the solid-melt front propagates through the system is a function of the temperature of the system. By extrapolating these data back to the temperature of zero propagation speed, they obtained an estimate of the melting transition accurate to about 10% for Si and Cu. A common method for determining the melting and glass transition points is by calculating either some thermodynamic property such as the density, internal energy, or the mean-square displacement (MSD) of a certain subset of atoms in the system as a function of the simulation time [9, 10, 16]. In our work, we used these indicators to estimate the melting transition of both benzene and the brominated phenyl crystal. In such MD simulations, we locate the melting and glass transition points by subjecting the system to heating (melting) and cooling (freezing) cycles. Due to computer hardware and CPU time constraints, the system size for "real" materials has typically been several hundred to several thousand atoms. Typically, the system has been equilibrated for about I to 100 ps at each temperature step. Thus the effective rate of temperature ramp or quench is much faster than conventional experiments resulting in substantial superheating in the heating / melting cycle and supercooling in the cooling / freezing cycle. In the heating cycle, superheating results in overprediction of the melting point. In the cooling cycle, typically the rapid quench results in glass formation instead of crystallization [8]. Furthermore, MID simulation assumes the crystal to be perfect, which is not the case in the crystals used for experiments on melting. The use of periodic boundary conditions attempts to simulate an infinite bulk system without any surfaces; it is known that crystal surfaces play an important role in the melting process [4 and references therein]. Phillpot et al. [20] showed that for Si, at temperatures above Tme1t, grain boundaries or free surface
Data Loading...