Lattice drifting

Dear LAMMPS developers,

I have a problem with lattice drifting after minimization or relaxation. After minimization, i have used nve/limit and then nve-langevin. also, I tried decreasing timestep or using other thermostating methods. But, the structure drifts. I couldnot solve this problem. following is my input file. I would be very greatful if you could help me.

Best regards,

#Initialization
units metal
boundary f f p
dimension 3
atom_style atomic

#Atom definition
variable 3_24 equal 3/24
variable 4_24 equal 4/24
variable 7_24 equal 7/24
variable 8_24 equal 8/24
variable 11_24 equal 11/24
variable 12_24 equal 12/24
variable 15_24 equal 15/24
variable 16_24 equal 16/24
variable 19_24 equal 19/24
variable 20_24 equal 20/24
variable 23_24 equal 23/24
variable 1_3 equal 1/3
variable 2_3 equal 2/3

lattice custom 1 a1 3.08 0 0 a2 0 3.08 0 a3 0 0 15.118 basis 0 0 0 basis {1_3} {2_3} {4_24} basis {2_3} {1_3} {8_24} basis 0 0 {12_24} basis {2_3} {1_3} {16_24} basis {1_3} {2_3} {20_24} basis 0 0 {3_24} basis {1_3} {2_3} {7_24} basis {2_3} {1_3} {11_24} basis 0 0 {15_24} basis {2_3} {1_3} {19_24} basis {1_3} {2_3} ${23_24}

region ambient block -2 12 -2 12 -0.5 3.5
region box block 0 10 0 10 0 3
create_box 2 ambient
create_atoms 2 region box basis 1 1 basis 2 1 basis 3 1 basis 4 1 basis 5 1 basis 6 1 basis 7 2 basis 8 2 basis 9 2 basis 10 2 basis 11 2 basis 12 2
group box1 region box

#Interatomic potentials
pair_style tersoff
pair_coeff * * SiC_1994.tersoff Si C
neighbor 1 bin
neigh_modify every 1 delay 0
mass 1 28.0855
mass 2 12.0107

#min_style quickmin
#minimize 1e-5 1e-5 10000 100000
#dump 0 all atom 1000 dump.lamg*.lammpstrj

velocity all zero angular
fix 1 all momentum 1 linear 1 1 1 angular
fix 2 all nve/limit 0

thermo 1000
thermo_style custom step temp etotal
dump 1 all atom 10000 dump.lamg*.lammpstrj
dump_modify 1 scale no
timestep 0.0005
run 10000

unfix 2
velocity box1 create 10 900000 rot yes dist uniform
fix 3 all nve
fix 4 box1 langevin 10 10 0.1 87985456
run 10000

I have a problem with lattice drifting after minimization or relaxation.
After minimization, i have used nve/limit and then nve-langevin. also, I
tried decreasing timestep or using other thermostating methods. But, the
structure drifts. I couldnot solve this problem.

Hint: Look into the lammps archives for similar past issues and solutions.
Your current problem is far from being new to the list and has been faced
by many others in the past. Help yourself by doing a more thorough review
of the archives and get to keep all credit related to solving the apparent
puzzle.
Carlos

Dear Carlos,

Thanks for your advice. I’ve gone through mailing list and considered the points. I’ve tried smaller timestep, minimization, updating neighbour list more often, applying fix nve/limit or langevin, but I couldnot solve the problem. could you please have a wee look at the script?

regards,

Assuming that you are describing your problem properly and indeed your
system is drifting (rotating?) after minimization, lets see:

You have a moving body through space and you want it to stop. You are the
god of such world with unlimited powers. What can you do to achieve such a
goal?...
Maybe setting all the atomic velocities to zero after the minimization step
and before proceeding with any further system setup?... Galileo would be
happy for you to put his equivalence principle into practice....

Carlos

Many thanks for your help. I’ve done this through “velocity all zero angular” and “fix 1 all momentum 1 linear 1 1 1 angular” commands in the script. Yet, It was no use. the problem is that the system encounters a sudden decrease in total energy (from -11622 to -23184). when I dump a snapshot after minimization, the lattice has drifted.

Best regards,

Such a huge change in energy cannot be only the result of a shift. But is this happened during the minimization cycle you should expect the energy to decrease anyways. The minimizer (assuming you are playing well with its input params) is going to try finding you a local energy minimum. If you don’t like the morphology of such config for whatever reasons then it is up to you to figure out why do you think such config is incorrect.
Carlos

Many thanks for your help. I've done this through "velocity all zero
angular" and "fix 1 all momentum 1 linear 1 1 1 angular" commands in the
script. Yet, It was no use. the problem is that the system encounters a
sudden decrease in total energy (from -11622 to -23184). when I dump a
snapshot after minimization, the lattice has drifted.

you should be more precise/descriptive with what you write.

- a "lattice" *cannot* rotate due to its infinite nature.
- a drift in the context of MD is understood as a motion of the center
of mass of the total system. thus a translation, not a rotation.
- why should the energy *not* decrease during minimization? and there
is no way that there is a "drift" during minimization, since there is
no dynamic.
- your input doesn't create a "lattice", but some kind of nanocluster,
since the creation box is smaller than the total system cell.
- without external interactions, systems like yours are energy
invariant against translation, so a drift cannot cause the energy drop

there are multiple things that don't make much sense.
- using fix momentum in *every* step is a *very* bad idea
- using fix nve/limit with a limit of 0 is a pretty pointless operation
- you try to remove rotational and translational momentum on a system
without temperature
- at a temperature range of 0 to 10K there is pretty much nothing
happening. particularly not "drifting".
- what is the point of using periodic boundaries at the smaller
separation of the periodic cells (z) and fixed boundaries elsewhere

most of your issues seem to indicate significant deficiencies in basic
MD knowledge. we cannot help you with that. this is not a classroom
for learning MD. you have to find yourself a proper personal
tutor/adviser that will help you to obtain that knowledge. there is
little point in doing MD, i.e. running LAMMPS, without. you will
mostly upset people by asking questions that don't need to be asked
and are mostly off-topic for this forum.

axel.