|From:||Ray Shan <rshan@...1795...>|
|Date:||Sun, 6 Aug 2017 02:32:06 +0000|
Your setup may be alright, but without seeing what actual commands you used it is difficult to say. Your density increasing from 0.3 to 4.2 just indicated that the force field type and/or parameters was not adequate in describing the isolated molecules in gaseous state. Therefore, it is more a problem of your model instead of a problem of LAMMPS.
Green Kubo method is based on the assumption that heat is flowing through a homogenous system. Since your system is not homogeneous, you can’t use Green Kubo.
I built a system which has a solid spherical copper nanoparticle and gas molecules in its surrounding.
I chose the simulation box with the target density of gas molecules at 1 bar and 300K (room temp.) and defined nanoparticle and gas molecules as different subsystems. Then I ran minimization to reduce the high potential collisions. I used fix NVE and fix langevin for both of my subsystems individually and then switched to NVT for the whole system and then switched to fix NPT.
But when I switched to NPT, the density changed and the gas molecules have changed phase and became liquid or solid as the density for the whole system was 0.3 g/cm3 before npt, but after a long run of NPT (1 ns) density shoots up to 4.2 g//cm3. Is this procedure of doing two phase simulation to determine a thermal property correct? Also, when I did the same simulation without doing npt, the code gave me a spurious result of the thermal conductivity using Green kubo. Please help me in understanding what I'm doing wrong and give me suggestions on how to correct this. Thanks!
PhD Student (Mechanical Engineering)