temperature of relaxation

[_KIKI]

Dear lammps users
I built a monocrystalline silicon workpiece that needs to be balanced. However,the temperature of "work_newton"in the process of relaxation is particularly high and the atoms of the “work_newton” flew out ，I do not know what caused it .How can I reduce the temperature to normal stability ?Here is my Output data :

here is my program:

units metal
boundary s s p
atom_style atomic
region box block -400 0 -220 0 0 14
lattice diamond 5.431 orient x 1 -1 0 orient y 1 1 1 orient z -1 -1 2 #111 1-10
region work block -400 0 -220 0 0 14 units box
region work_newton block -360 0 -180 0 0 14 units box
region work_t1 block -380 -360 -180 0 0 14 units box
region work_t2 block -380 0 -200 -180 0 14 units box
region work_t union 2 work_t1 work_t2 units box
region work_b1 block -400 -380 -200 0 0 14 units box
region work_b2 block -400 0 -220 -200 0 14 units box
region work_b union 2 work_b1 work_b2 units box
create_box 3 box
create_atoms 3 region work
group work_newton region work_newton
group work_t region work_t
group work_b region work_b
group work region work
set group work_newton type 1
set group work_t type 2
set group work_b type 3
mass 1 28.0
mass 2 28.0
mass 3 28.0
pair_style tersoff
pair_coeff * * SiC.tersoff Si Si Si
neighbor 2.0 bin
neigh_modify delay 5
compute 2 work pe/atom
compute w_pe work reduce sum c_2
compute w_temp work_t temp
compute w_newton work_newton temp
fix b work_b setforce 0.0 0.0 0.0
velocity work_b set 0 0 0 units box
velocity work_newton create 300.0 5812775 temp w_newton
fix 1 work nve
fix res_w_temp work_t temp/rescale 10 293.0 293.0 10.0 1.0
fix_modify res_w_temp temp w_temp
timestep 0.001
thermo 100
thermo_style custom step temp c_w_temp c_w_pe
thermo_modify temp w_newton
run 5000
Here is formed graphics :

Best wishes
Liu Qiqi

[akohlmey]

you are making the same mistake as (too) many other LAMMPS users. the high temperature is the symptom, not the cause. to resolve the situation, you have to find what is causing the high temperature. since most of your input is rather common and not unusual, the most likely cause is the geometry. changing the thermostatting (and what you seem to be doing doesn’t look good and seems overly complex and convoluted in the first place) will at best suppress the symptom, but not resolve the cause. a simulation like yours has to work and conserve energy even without a thermostat.

an “explosion” like you see can easily be caused by “close contacts”, i.e. atoms that are too close (say less than 1\AA) to each other and thus are causing extremely high forces (and pressure!) and then in turn easily “explode” your system because of that, if there are enough of them. usually looking at the pressure of your system is a good indicator of close contacts. and the delete_atoms overlap option makes it easy to confirm, if close contacts exist.

axel.

Dear lammps users
I built a monocrystalline silicon workpiece that needs to be balanced. However,the temperature of "work_newton"in the process of relaxation is particularly high and the atoms of the “work_newton” flew out ，I do not know what caused it .How can I reduce the temperature to normal stability ?Here is my Output data :

[…]