Application of UFF potential in LAMMPS

Dear All,
I am new in MD simulation and LAMMPS. I want to use Universal Force Field (UFF) in LAMMPS simulation. Is it possible to use ? If yes, someone please share the detail. How to use it ? what will be the command for it.

In more detail, I want to simulate the diffusion between two metals like Cr and Sc. If someone can suggest any other potential for Cr and Sc it will also be helpful.

Regards
A Biswas

UFF can be set up as a propper combination of bond_style, angle_style, pair_style and so on (along with manual setting of respective parameters). Check the paper of UFF force field to find out what you should use.

Thank you for the reply. Could it be possible to share an example file using UFF.

I have never used it before, but maybe someone else has.

That means you will need a tutor and get some training. Doing meaningful MD simulations for research is more than finding some parameters and figuring out the syntax of the input files needed. You need to discuss this with your adviser or supervisor. Learning and doing this on your own is a very bad idea, as you would be doomed to do many mistakes that are easy to do but not so easy to detect. Your questions indicate that you have not done the necessary “homework” for your specific problem.

It is most certainly possible, but whether it is a good idea is a different question.

If you would be properly trained in MD simulations, you would not need to ask these questions, but would be able to set up your simulation from the information provided in relevant publications and the LAMMPS manual on your own.
More importantly, there are always two questions to consider: 1) what is a suitable potential function and force field parameters 2) what is a suitable simulation method. If you have the answer to one of those, you still are not able to do a meaningful simulation.

At what temperatures? Is the metal going to be liquid or solid? Are you certain that molecular dynamics is a suitable method for this? Have you searched the published literature for any precedents or people having done similar studies on similar compounds? This kind of literature review is a prerequisite for any proper scientific study. You need to make certain that you know what is possible and what suitable approaches are. As a beginner, you will be hard pressed to do some of the advanced work that has been published, but it will be impossible to do a kind of simulation that has not done before. I personally would have my doubts that MD is a suitable method without seeing some credible publications that demonstrate it.

You will find suggestions by searching the published literature. I seriously doubt that UFF would be a good choice. The more generic a force field, the less accurate its results will be, and UFF is one of the most generic of classical force fields. The typical choice of a model for studying metals is the embedded atom method (EAM) which is good for any kind of compound where the interactions are only of the metallic kind. For some metals more sophisticated choices are required, as their interactions have covalent components and thus are more directional. Again, a survey of the published literature for studies of systems similar to yours is the best source of information since this is far more general than using LAMMPS or learning the syntax of commands.

Dear akohlmey,
Thank you for your reply and I am grateful to you for your kind suggestions.
Actually in literature the EAM potential is not available for Sc. In NIST potential repository only the universal OpenKIM is shown. So UFF is thought as an alternative. However, in LAMMPS manual I do not found the ‘pair_style’ command for UFF.

My interest of diffusion is may be up to 1000K.

Regards

That repository is not the only source for information about potentials. There are many that are not listed there.

That doesn’t say anything about its applicability to your system. There must be some molecular dynamics studies in the literature for alloys including Sc. Have you researched what those people were using? Is there any credible study that has successfully applied UFF to such a system? If I were you, I would not waste time on trying to set up a simulation without knowing whether there is a chance for success.

This means, you have not done your “homework” (correctly). You won’t necessarily find a pair style for each force field. Often it consists of a combination of multiple potential functions. Thus, you need to look up the original publication describing the force field you want to employ, study it carefully about its properties and applicability, and then also check which potential functions it uses, and then check the LAMMPS manual, if it supports those potential functions (chances are it does, worst case you have to use pair style table).

At that point you need to make some simple tests to determine whether your choice of force field can represent some simple properties of the material of interest. If the simulations fail to give acceptable results, you need to find a different model.

But I also suggest to check the published literature to see if molecular dynamics itself is a suitable and effective method for the kind of study you want to do. I doubt it, since diffusion is an activated process in solids and to cross those barriers can take a looooong time with classical MD for the system sizes and time scales available to simulations.

In general, this all sounds like you need a discussion with your adviser or a collaborator with experience in simulations. The project you have taken on seems overly ambitious for somebody that is struggling already with rather basic issues and does not seem aware of fundamental problems.

Here are two more suggestions to help you get a better handle on doing the simulations instead of chasing a potential for an elusive element.

  1. set up a simulation study for a similar alloy where a parameter file (say, EAM) does exist for both elements, for example Cr-Co, or Cr-Fe, or Cr-Mn.
  2. set up a simulation study for the same elements as above with the UFF parameters and compare the results.

This will allow you to make an assessment, whether your simulation can yield meaningful results at all and how transferable the UFF parameters are to the kind of simulation you plan to do.

Dear Akohlmey,
Thank you once again. I deeply appreciate your effort for the solving our problem.

I also think similarly. As we have done recently some simulation of Ni-Ti system using MEAM potential. So first I will compare this result with the result of Ni-Ti system using UFF. Then proceed for the new system of Sc.

If you have any tutorial on the technique of using UFF in the input file could you please share it. As I am not able to extract key information related to UFF from manual.

Regards

I have explained this now several times. Your level of ignorance is starting to annoy me. If you are asking for advice, you should follow it. You must study the UFF paper. It will show you the pair style you need. There will be no UFF pair style.

Please also see my comments to this other recent post.