[lammps-users] Issue with Glycerol

Dear Lammps users, I want to build a slab made of Glycerol molecules to be used as substrate for Langmuir-Blodgett simulations.

I tried first with some 6000 molecules and I got a problem, which persisted while I was decreasing the number of molecules.

To make a long story short, I tried with 1 glycerol molecule: again I got an Error: Angle atoms 2 3 6 missing on proc 0 at step 672 immediately after starting the run.

I tried then with a similar molecule (1,3-propanediol) which run smoothly.

Herebelow please find the input and data files.

Any comment woul be very welcome


The behavior you describe strongly hints that there is something wrong with the force field parameters.
Where did you get them from?

The best way to diagnose this is to output data frequently and with more detail, i.e. write out the trajectory and thermo output, say, every 10 steps
and output the potential energy contributions individually, e.g. evdwl, ecoul, ebond, eangle, edihed

If you visualize the trajectory you should see when the molecule takes on unrealistic configurations and can correlate that with the energies.

I would also compare to a similar simulation of the diol.

Also, you can obviously delete all improper data from the data file, since those do not contribute.


Dear Axel, all data on glycerol were obtained from http://zarbi.chem.yale.edu/ligpargen/ , the LigParGen site by the Jorgesen group you certainly know.


I will try to apply your recommendations. Let’s see what happens

thanks a lot


------ Messaggio originale ------

Such automated tools have to be used with care. Generic parameter sets that are simply derived from the molecular geometry often have lower accuracy and may fail for specific molecules. Glycerol is a common enough molecule to make it worth your while to search for specifically tuned parameters.
Given that you want to use it as a substrate should make it particularly important to look for a set of parameters that can represent all relevant properties at the intended thermodynamic state(s). Lower accuracy parameters will taint the reliability of your results.

I succesfully simulated water/glycerol mixtures with
an old force-field indicated here:

         Giovanni La Penna

National research council of Italy
Institute of chemistry of organometallic compounds
via Madonna del Piano 10, 50019 Sesto Fiorentino (FI), Italy

Dear Axel, I applied your recommendations which proved to be very useful: the trajectory showed the molecule blowing to pieces after few fs.

I also made a little investigation through the available literature, again as per your suggestion. I found that the set of atom partial charges generated automatically by LigParGen is surprisingly wrong, i.e. totally different from the “official” data of OPLS-AA.
After having made the necessary corrections, the MD simulation run smoothly.

Thanks a lot for your help


------ Messaggio originale ------

Hi everyone,

To use opls-aa it is necessary to include this comand. Using this I ran molecules like glycerol without problems.

special_bonds lj/coul 0.0 0.0 0.5