I tried simulation of liquid oxygen at 70 K but after NVE simulation temperature is rising to 80-90 K. which is the boiling point of oxygen. I wanted to simulate liquid oxygen. Is it possible to do by MD simulation.
I agree completely with the comment of Anatoly Belonoshko. I recommend that you perform an MD simulation under constant isotropic pressure (of 1 bar) and constant temperature (70 K), preferably with a weak-coupling method with a rather large time constant. The volume will fluctuate; monitor the density. It should stabilize at some value near the experimental one. If it deviates considerably from that, your model is wrong. Also monitor the homogeneity: if the volume increases and the liquid tends to break up in clusters, your liquid is likely to boil. It is not easy to observe the boiling point, as a liquid may be superheated and still be homogeneous. If you then switch to constant volume, you will see an unphysical negative pressure.
In order to observe a coexistent liquid/gas system, prepare the system at constant V,T with all particles initially in a slab, from where the molecules will evaporate into the empty space adjacent to the slab Then there are two flat interfaces between liquid and gas and the surface (free) energy will not depend on the thickness of the slab. [This is different if you put the liquid initially in a sphere; the total surface free energy will then change with the size and mess-up the phase equilibrium; the equilibrium pressure within the sphere will be larger than the bulk pressure.] It takes a large system and long simulations to get good statistics. You finally get (when equilibrium is reached: may take some time) the boiling pressure at the given temperature.
The best way to study the phase equilibrium is to use the Gibbs ensemble method: exchange molecules between the two phases, depending on an acceptance criterium. See Frenkel and Smith "Understanding Molecular Simulation", chapter 8. You can also compute the change in chemical potential by thermodynamic integration along an isothermal path from within the slab to outside the slab. Considering the gas phase as ideal, this also gives the boiling pressure at the given temperature.
Beware of the low temperature: there may be quantum corrections to the results of classical computations.
When you start an MD simulation, your system, as a rule, is not at equilibrium. Therefore, it takes some time to equilibrate it and, during the equilibration period, the temperature can rise. One needs to scale temperature to the desirable one during the equilibration period or control T in some way. In the case of oxygen there is a possibility that the internal (vibrational) degree of freedom will contribute to the rising of T and it might take quite some time to equilibrate translational and vibrational degrees of freedom. Also, please consider that you are dealing not with oxygen but rather with a model of oxygen. And the model of oxygen can have very much diffferent boiling point as compard to the experimental one. It depends on the quality of the model. The determination of the boiling point is a problem in itself.
I agree completely with the comment of Anatoly Belonoshko. I recommend that you perform an MD simulation under constant isotropic pressure (of 1 bar) and constant temperature (70 K), preferably with a weak-coupling method with a rather large time constant. The volume will fluctuate; monitor the density. It should stabilize at some value near the experimental one. If it deviates considerably from that, your model is wrong. Also monitor the homogeneity: if the volume increases and the liquid tends to break up in clusters, your liquid is likely to boil. It is not easy to observe the boiling point, as a liquid may be superheated and still be homogeneous. If you then switch to constant volume, you will see an unphysical negative pressure.
In order to observe a coexistent liquid/gas system, prepare the system at constant V,T with all particles initially in a slab, from where the molecules will evaporate into the empty space adjacent to the slab Then there are two flat interfaces between liquid and gas and the surface (free) energy will not depend on the thickness of the slab. [This is different if you put the liquid initially in a sphere; the total surface free energy will then change with the size and mess-up the phase equilibrium; the equilibrium pressure within the sphere will be larger than the bulk pressure.] It takes a large system and long simulations to get good statistics. You finally get (when equilibrium is reached: may take some time) the boiling pressure at the given temperature.
The best way to study the phase equilibrium is to use the Gibbs ensemble method: exchange molecules between the two phases, depending on an acceptance criterium. See Frenkel and Smith "Understanding Molecular Simulation", chapter 8. You can also compute the change in chemical potential by thermodynamic integration along an isothermal path from within the slab to outside the slab. Considering the gas phase as ideal, this also gives the boiling pressure at the given temperature.
Beware of the low temperature: there may be quantum corrections to the results of classical computations.
- Badges
- Science topic