Reproducing diblock copolymer phases by DPD


I am trying to reproduce the lamellar phase of diblock copolymer melt from the paper by Groot and Madden, JCP, 1998 (doi:10.1063/1.476300). I am consistently failing (i.e. I observe no phase separation at all). I believe I have run out of options to try, and thus I am asking you for help.

What I tried so far:

  1. In LAMMPS (see attached input file) I varied
  • AB interactions,
  • spring constants,
  • neighbor cutoff distance,
  • box sizes,
  • number of steps,
  • level of parallelisation.
    All without success.
  1. I ran exactly same setting in Materials Studio 8.0, and after 10000 steps I received a fairly clear lamellar phase for eAA = 25 and eAB = 40.

  2. I also tried a simpler binary mixture in LAMMPS, which worked fine.

Can you please help? Perhaps there is some trivial error in my input file that I am overlooking and that some of you can spot quickly.

Thank you in advance for help.

Kind regards, (1.17 KB)

The best way to find trivial errors is to make detailed comparisons with an independent calculation e.g. Materials Studio. For example, directly compare the deterministic forces generated by LAMMPS and Materials Studio on timestep 0. If they don’t match, there is no point waiting around to see if the lamellar phase emerges.

yes, Aidan’s idea is good to compare the 2 codes. Not just forces,

but energy and pressure (which is forces). For your initial few

timesteps, things should match closely if you’re doing equivalent