Hi
I am woking on simulation of chloride ion diffusion in nanopores of a ceramic crystal. Crystals are placed at top and bottom of supercell. I created a 60 A pore between substrates. Dimensions of my model are 13.38A x 36.95 A x 100.94 A. The pore is filled by 992 water molecules, 10 chloride ions and 5 calcium ions (using adsorption module).
I used COMPASS as forcefield and total simulation time is 2000 ps and summation method for electrostatics and Van der Waals are Ewald and Atom-based, respectively (using Forcite module).
Using trajectory file we calculated mean square displacements (MSD). So, using Einstein formula we can calculate diffusion coefficient. To do this, we exported MSD data to excel to draw diagrams of MSD versus time. The slope of this curve is diffusion coefficient. If we obtain a straight line, the simulation is converged.
But in my simulations, diagram is not a straight line and linear fitting dosen't provide precise approximation for half of the ions. We tested many option to solve this problem such as changing the initial location of atoms (in the middle of sample or uniform distribution in all points of pore height), increasing simulation time, testing another forcefields (like universal, pcff), ....
I really appreciate if everybody suggest a solution for this problem.
Thanks in advance
Behnam
I am woking on simulation of chloride ion diffusion in nanopores of a ceramic crystal. Crystals are placed at top and bottom of supercell. I created a 60 A pore between substrates. Dimensions of my model are 13.38A x 36.95 A x 100.94 A. The pore is filled by 992 water molecules, 10 chloride ions and 5 calcium ions (using adsorption module).
I used COMPASS as forcefield and total simulation time is 2000 ps and summation method for electrostatics and Van der Waals are Ewald and Atom-based, respectively (using Forcite module).
Using trajectory file we calculated mean square displacements (MSD). So, using Einstein formula we can calculate diffusion coefficient. To do this, we exported MSD data to excel to draw diagrams of MSD versus time. The slope of this curve is diffusion coefficient. If we obtain a straight line, the simulation is converged.
But in my simulations, diagram is not a straight line and linear fitting dosen't provide precise approximation for half of the ions. We tested many option to solve this problem such as changing the initial location of atoms (in the middle of sample or uniform distribution in all points of pore height), increasing simulation time, testing another forcefields (like universal, pcff), ....
I really appreciate if everybody suggest a solution for this problem.
Thanks in advance
Behnam