I am conducting a molecular dynamics simulation of water evaporation on a copper surface at room temperature. The surface area of the copper plate is 10,000 square angstroms. Based on the existing thermal evaporation limit of water (1.45 kg/m2-hr), I calculated the probability of water evaporation in 1 ns under 10,000 square angstroms and found that the evaporation of water Almost impossible to happen. But when I ran molecular dynamics simulations, water evaporation events at room temperature went from nearly impossible to occasionally occurring. The model of water molecules used is the SPC model. I want to know if the parameters selected are absolutely correct, whether molecular dynamics will enhance the effect of evaporation, thereby turning an almost impossible event into a common event.
Any one can help me to understand this?
None of the existing water models is “absolutely correct”. The very nature of a model is that it is an approximation. But you are also leaving out a lot of relevant information:
- what are the parameters for the copper - water interaction? this can be rather difficult to model correctly and could be crucial to your model.
- how thick is the water layer on top of the copper? when you get to the atomic scale, finite size effects can lead to unexpected behavior.
- why SPC and not SPC/E?
- were you using a thermostat in your simulation? how would that relate to your thermodynamics based estimate of water evaporation.
Is that “evaporation limit” for water under atmospheric pressure? I suspect your MD simulation models water under a vacuum instead, so it would not be surprising to see much higher evaporation rates.
Thank you for answering. I think my previous statement was somewhat wrong. What I want to express is that for the SPC model, I chose a general parameter to ensure that the simulated process can reflect the properties of water to a certain extent. I think these questions you raised are very good：
What I want to do is actually the evaporation of water on the solid surface. What I emphasize more is the interaction between water molecules, so I choose the LJ potential for the interaction between copper and water.
The water layer on the copper surface is about 40 angstroms thick, so I think there is no interaction between the water molecules above and the copper surface.
Since I am a newbie in water evaporation, I am not particularly clear about the physical processes that each model is suitable for. I will try to use the SPC/E model or the TIP4P model to simulate it.
I have no specific requirements for the thermostat used in the simulation. The default thermostat used by lamps is the Nose-Hoover thermostat.
Yes, I think you are right, the simulation command I set is
fix fnpt Water npt temp 295.35 295.35 100 x 0 0 1000 y 0 0 1000, If I want to reflect atmospheric pressure conditions, do we need to modify the external pressure to 1?
If yes, another problem arises: In such a small space-time scale, I cannot observe that water evaporates as it should. How can I judge that the system has reached thermal equilibrium?
Then why include the copper in the model at all?
Then you need to study the published literature about it. People have been simulating water with MD for decades, there should be something along those lines.
There is NO default for time integration and thermostatting in LAMMPS. But the main purpose of a thermostat is to exchange kinetic energy with a larger reservoir to model the behavior of a bulk system. But that doesn’t exists in your case. If at all, you would thermalize the solid and equilibrate your system to have a consistent temperature throughout and then do plain timestepping with fix nve for the water.
My intention was to keep the water at a constant height, I used the
fix freeze command for the bottom Cu. I’m not sure if there is any other way to fix the height of the water。
I did investigate some molecular dynamics simulations of water evaporation, but I did not find any simulation of water evaporation at normal temperature and pressure, which is not consistent with the situation I want to simulate.
That might be my misunderstanding for the lammps documentation, I’ll take a look at it again and try your suggestion
Actually I take back what I said about vacuum vs atmospheric pressure. I’ve just remembered that I used to do these simulations (unpublished) which had a slab of water in vacuum, fully periodic. I saw exactly one “evaporation” in several nanoseconds of simulation in an area of a few nanometers squared.
You don’t need to take my word for it (and you shouldn’t – it’s not scientific). An old standard paper on diffusion in confinement (https://pubs.acs.org/doi/abs/10.1021/jp0375057) describes the methodology of simulating liquid-vapor interfaces – check their graph and you will see essentially no water density in the “vapor” (really vacuum) phase.
So there is in fact something fundamentally wrong with your approach. It is not possible for me to determine what, given the very little information you have supplied, nor is it a useful exercise to try and repair your simulation. Instead I would recommend you try to understand and replicate the Liu, Harder and Berne paper I mentioned above. Their result is replicable (I’ve done it myself) and trying it will teach you some valuable first lessons in molecular dynamics as well as some useful concepts in nanointerfaces.
Thank you very much for providing it, I will take a look and try it !