Hello,
I am trying to reproduce a system following parameters taken from this article : Molecular dynamics simulations of evaporation-induced nanoparticle assembly, S. Cheng and G. S. Grest, J Chem Phys, 138, 064701 (2013)
I do not get an error message but I think I did not set something correctly : the particle are moving in the vapor phase which is not what I expect.
To build the system, I used create_atom command to generate randomly atoms in two regions “vapor” and “liquid”. I took care about boundary conditions. I think I forgot to set the interface. I’m kind of newbie using LAMMPS, and I can’t find what is missing/wrong.
Any though ?
Thank you
Input script :
units lj
boundary p p f
atom_style sphere
dimension 3
[…]
region film block 0 {box_x_length} 0 {box_y_length} 0 {film_z_depth} units box
region vapor block 0 {box_x_length} 0 {box_y_length} {film_z_depth} ${box_z_length} units box
region whole union 2 film vapor units box
create_box 2 whole
create_atoms 1 random {nb_of_np} 99999 film
create_atoms 2 random {nb_of_lj_in_liquid} 99999 film
create_atoms 2 random ${nb_of_lj_in_vapor} 99999 vapor
[…]
fix 1 all wall/reflect zlo EDGE units box # to prevent lost atoms through low z dimension
fix 2 all wall/reflect zhi EDGE units box # to prevent lost atoms through high z dimension
[…]
Hello,
I am trying to reproduce a system following parameters taken from this article : Molecular dynamics simulations of evaporation-induced nanoparticle assembly, S. Cheng and G. S. Grest, J Chem Phys, 138, 064701 (2013)
I do not get an error message but I think I did not set something correctly : the particle are moving in the vapor phase which is not what I expect.
To build the system, I used create_atom command to generate randomly atoms in two regions "vapor" and "liquid". I took care about boundary conditions. I think I forgot to set the interface. I'm kind of newbie using LAMMPS, and I can't find what is missing/wrong.
Any though ?
there is not enough information here to make any specific assessment.
e.g. how you set up a vapor/liquid interface, also depends on the
individual steps of how you equilibrate your system and much more.
axel.
Thank you for your reply.
I followed the paper instruction to equilibrate my system :
Thank you
there is still not much helpful information here. mostly, there are
some questions about your choices.
- why atom style sphere? there is nothing in this input, that cause
torque or depends on the orientation of an "atom". all forces are
pair-wise and between the centers or particles. there are no
tangential forces or dipole/charge or dipole/dipole interactions or
similar
- why use pair style hybrid? the colloid pair style supports "small"
and "large" particles in the same pair style. please re-read the
documentation
- your equilibration procedure, as far as i can tell, is that of a
homogeneous mixture. wouldn't it be better to equilibrate your two
sections separately, by doing time-integration only for one of the two
parts, and to higher (for vapor) and lower (for liquid) temperature,
so that it is guaranteed that you have now phase transitions during
equlibration. after equilibration, you can can rescale the total
temperature to your desired final value with the velocity command and
then continue from there will time-integrating all atoms.
axel.
Dear Axel,
Thank you for your reply.
I'll try to answer your questions, please find them in the body of your message bellow.
Best regards
Charlotte
there is still not much helpful information here. mostly, there are
some questions about your choices.
- why atom style sphere? there is nothing in this input, that cause
torque or depends on the orientation of an "atom". all forces are
pair-wise and between the centers or particles. there are no
tangential forces or dipole/charge or dipole/dipole interactions or
similar
This is the first time I use LAMMPS and also the first time I have to build such system. I've worked on biomolecule (protein, DNA, small molecules like ligands...) before with AMBER software but never with particles or colloids. I thought that this atom_style was the best to fit with the parameters I have to give as input to run my simulation (x y z coordinates, density and diameter).
- why use pair style hybrid? the colloid pair style supports "small"
and "large" particles in the same pair style. please re-read the
documentation
Yes, I am a bit confuse about that because in the paper, the authors give Hamaker constants and cut-offs for NP-NP and NP-solvent interactions but not for solvent/solvent interactions. Thus, I thought it was better to set a lj potential for the least.
- your equilibration procedure, as far as i can tell, is that of a
homogeneous mixture. wouldn't it be better to equilibrate your two
sections separately, by doing time-integration only for one of the two
parts, and to higher (for vapor) and lower (for liquid) temperature,
so that it is guaranteed that you have now phase transitions during
equlibration. after equilibration, you can can rescale the total
temperature to your desired final value with the velocity command and
then continue from there will time-integrating all atoms.
Okay, I'll try to equilibrate both parts separately, thank you for this precision.
Dear Axel,
Thank you for your reply.
I’ll try to answer your questions, please find them in the body of your
message bellow.
Best regards
Charlotte
there is still not much helpful information here. mostly, there are
some questions about your choices.
- why atom style sphere? there is nothing in this input, that cause
torque or depends on the orientation of an “atom”. all forces are
pair-wise and between the centers or particles. there are no
tangential forces or dipole/charge or dipole/dipole interactions or
similar
This is the first time I use LAMMPS and also the first time I have to
build such system. I’ve worked on biomolecule (protein, DNA, small
if this is the first time you have to set up such a system and the first time you are using LAMMPS, then you are trying to do too many things at once and thus setting yourself up for failure. i would suggest you approach the overall problem in stages:
- MD of a bulk liquid system (like liquid argon or the solvent of your system of interest) to learn the overall MD process and how to analyze the data. this could also be done in AMBER, and thus you can compare your settings and results between both codes and published data.
- MD of a liquid vapor system of the solvent from above. as noted, starting from random positions will not automatically equilibrate into a two phase system, and you also have to make certain, that you have the proper coexistence temperature/thermostat settings, or else either the liquid or the vapor phase will grow continuously.
- MD of a build colloid in solvent system. that way you learn to properly set up colloid parameters and check their results
- by now, you should have build up the experience, to tie everything together and do your actual research project.
if you are skipping stages of this, you will have a much harder time to tell whether unexpected results are genuine or a typo or a conceptual problem in your input.
molecules like ligands…) before with AMBER software but never with
particles or colloids. I thought that this atom_style was the best to
fit with the parameters I have to give as input to run my simulation (x
y z coordinates, density and diameter).
it would really only matter, if you have to have polydisperse masses (i.e. all your colloidal particles are of different size and thus different mass and there are too many to treat them with different atom types). as mentioned before, nothing else that is part of what atom style sphere does is used and thus there is no point in using fix nve/sphere for example.
- why use pair style hybrid? the colloid pair style supports “small”
and “large” particles in the same pair style. please re-read the
documentation
Yes, I am a bit confuse about that because in the paper, the authors
give Hamaker constants and cut-offs for NP-NP and NP-solvent
interactions but not for solvent/solvent interactions. Thus, I thought
it was better to set a lj potential for the least.
if you want to reproduce a published calculation, you cannot just make up settings. either the missing parameters are implied somehow, or you should get the info from the corresponding author of the paper.
axel.
Dear Axel,
Thank you for your wise advices. I’ll try to fit to the steps you described.
Regards
Charlotte