Material response to laser heating and heat reflected

LAMMPS LAMMPS Mailing List Mirror
1

Dear all,

I am trying to study the following problem where I have a layer of carbon anchored to a substrate and a laser light heating it on top. I am hoping to study how the carbon layer respond to the laser light as well as the heat reflected from the substrate.

I am using the fix heat on a cylindrical region to simulate the laser heating. In addition, a fix heat is also used on a block region of 3 layers at the bottom together with the fix wall/lj126 is used to simulate the substrate and the heat reflected from it. Finally, I am using periodic boundary condition on x and y direction with fix boundary condition on the z direction. Potential used is Airebo.

In order to prevent the atoms from being lost, a fix wall/lj126 is used on the top of the surface during the initialization and equilibration process. When it comes to the heating process, the fix wall/lj126 is unfix. However, I am at lost to interpret the results obtained as the carbon seems to be oscillating such that it expands and contracts repeatedly. This is going on while the heat is being given to it continuously.

Any advice would be greatly appreciated. Thank you.

2

Dear all,

I am trying to study the following problem where I have a layer of carbon
anchored to a substrate and a laser light heating it on top. I am hoping to
study how the carbon layer respond to the laser light as well as the heat
reflected from the substrate.

I am using the fix heat on a cylindrical region to simulate the laser
heating. In addition, a fix heat is also used on a block region of 3 layers
at the bottom together with the fix wall/lj126 is used to simulate the
substrate and the heat reflected from it. Finally, I am using periodic
boundary condition on x and y direction with fix boundary condition on the z
direction. Potential used is Airebo.

In order to prevent the atoms from being lost, a fix wall/lj126 is used on
the top of the surface during the initialization and equilibration process.
When it comes to the heating process, the fix wall/lj126 is unfix. However,
I am at lost to interpret the results obtained as the carbon seems to be
oscillating such that it expands and contracts repeatedly. This is going on
while the heat is being given to it continuously.

there is not enough information in those rather vague descriptions
to make any accurate recommendations.

have you equilibrated your system before apply the heat fixes?
what you describe could be simply a result of a badly chosen
and non-equilibrated initial configuration.

axel.

3

Hi,

Sorry. Before the heat part, the system is initialized to amorphous carbon via the liquid quench method from the initial simple cubic structure. I am a bit reluctant to use the change_box command for fear of causing the problem of bad dynamics, hence the initialization of a larger simulation box.

My input script is as follows (I am focusing on the heat from below, the laser heating is not included yet):

units metal
atom_style atomic
boundary p p f

lattice sc 1.821
region box block 0 30 0 30 0 10
region box2 block 0 54.63 0 54.63 -0.9 54.63 units box
region heat_reg block 0 30 0 30 0 3
create_box 1 box2
create_atoms 1 region box
mass 1 12.0

group laser region heat_reg
velocity all create 5000 87287 dist gaussian units box

pair_style airebo 2.5 1 1
pair_coeff * * CH.airebo C
neighbor 0.3 bin
neigh_modify every 10 delay 0 check yes

thermo 1000
timestep 0.0001

dump initial all xyz 1 diamond_ini_pos.xyz

fix 1 all nve
fix 2 all temp/rescale 10 5000 5000 5.0 1.0
run 0
undump initial

fix fix_wall2 all wall/lj126 zlo -0.9 1.0 1.0 2.5 units box
fix fix_wall all wall/lj126 zhi 19.11 1.0 1.0 2.5 units box

run 5000

unfix 1
unfix 2

fix 3 all nvt temp 5000 5000 0.01 drag 0.8
run 25000

unfix 3

dump final all xyz 1000 diamond_final_pos.xyz
fix 5 all nve
fix 6 all temp/rescale 100 5000 300 5.0 1.0
run 47000

unfix 5
unfix 6

fix 7 all nvt temp 300 300 0.01 drag 0.8
run 30000

undump final
unfix 7

unfix fix_wall

dump dump1 all xyz 100 laserHeat.xyz
dump dump2 laser xyz 100 test.xyz
fix fix2 all nve
fix fix1 laser heat 1 100 region box
run 10000

unfix fix1
unfix fix2

#end

4

Hi,

Sorry. Before the heat part, the system is initialized to amorphous carbon
via the liquid quench method from the initial simple cubic structure. I am a
bit reluctant to use the change_box command for fear of causing the problem
of bad dynamics, hence the initialization of a larger simulation box.

well, this input causes bad dynamics right from the first step.

axel.

5

Hi,

I am sorry but I am unclear what your first step refers to? The boundary conditions or the initial simple cubic structure?

I have done the initialization of the amorphous structure. The amorphous structure that I obtained has a sp3 sp2 ratio that is comparable to the ratio of other simulation works.

An additional question, I am wondering is there any possibility of studying laser ablation with periodic boundary condition? Thus far, I have yet to see any previous simulations done on laser ablation with periodic boundary condition.

Thank you very much.

6

Hi,

I am sorry but I am unclear what your first step refers to? The boundary
conditions or the initial simple cubic structure?

you dynamics at the first MD is bad with an insanely high
kinetic energy after just a few MD steps. just set thermo
to 1 and watch.

I have done the initialization of the amorphous structure. The amorphous
structure that I obtained has a sp3 sp2 ratio that is comparable to the
ratio of other simulation works.

that may well be, however, it doesn't change that your input causes
bad dynamics.

An additional question, I am wondering is there any possibility of studying
laser ablation with periodic boundary condition? Thus far, I have yet to see
any previous simulations done on laser ablation with periodic boundary
condition.

it is pointless to discuss anything about this system
unless you manage to set up a decent initial equilibration.

also, please search the mailing list archives for some discussion
on why using temp/rescale is a very bad idea.

axel.

7

Dear Axel,

The bad dynamics is due to the presence of the walls. The wall is there to prevent the volume from changing as I would want the density of the system to be constant. Apart from that, the bottom wall is also there to prevent the atoms at z = 0 to go to negative z positions as we expect the amorphous carbon to be deposited on top of a substrate. I think the kinetic energy will be more stable after 100+ steps due to the temp/scale fix.

I have tried using nvt fix instead of the temp rescale but the temperature tends to fluctuates. However, with a temp/scale fix for 0.5 ps, the temperature will not fluctuates that much and goes to equilibrium quickly when it comes to the nvt fix. This is true when I am using a periodic boundary condition without the presences of the wall. I agree that the temp/scale fix is not a good way but it is thus far the only way which gives me a quick equilibrium of the system.

Instead of having the wall, I tried using the nvt fix to try to maintain the volume. However, I think due to the simple cubic structure which is highly unstable for carbon and the very high temperature of 5000K, the nvt fix is not fast enough to equilibrate the system, causing an expansion of the system which will change the density of the system. The npt fix would not be suitable as the boundary condition along z is not periodic. These are the two methods I can think of to try and replace the walls at the initial stage. However, both do not seem to work very well.

Any suggestions would be deeply appreciated. My apologies for the inconvenience as I am very much inexperience in doing simulation work.

Thanks in advance.

8

Dear Axel,

The bad dynamics is due to the presence of the walls. The wall is there to
prevent the volume from changing as I would want the density of the system
to be constant. Apart from that, the bottom wall is also there to prevent
the atoms at z = 0 to go to negative z positions as we expect the amorphous
carbon to be deposited on top of a substrate. I think the kinetic energy
will be more stable after 100+ steps due to the temp/scale fix.

this makes no sense. if the bad dynamics is due to the walls,
then you placed them wrong (or the carbon atoms).

i repeat. using temp/rescale is a very bad idea because it doesn't
dissipate any energy. on the contrary, it enhances fluctuations in
many scenarios, which is the opposite of what you want.

I have tried using nvt fix instead of the temp rescale but the temperature
tends to fluctuates. However, with a temp/scale fix for 0.5 ps, the

the problem is not there it is because you have a bad initial setup.
also, because of the way it operates, using fix langevin instead of
temp/rescale is a far superior choice. nevertheless, you *first* have
to start from a reasonable starting configuration, i.e. adjust the walls
so that they are located at meaningful positions.

temperature will not fluctuates that much and goes to equilibrium quickly
when it comes to the nvt fix. This is true when I am using a periodic

you are mistaken there. when you use temp/rescale at a frequency
that is a divisor of the thermo output, you don't see the real temperature
of your system.

boundary condition without the presences of the wall. I agree that the
temp/scale fix is not a good way but it is thus far the only way which gives
me a quick equilibrium of the system.

you don't get an equilibrium. that is your problem. in any case
the input you posted *doesn't even get there*, so you need to
correct that first.

Instead of having the wall, I tried using the nvt fix to try to maintain the
volume. However, I think due to the simple cubic structure which is highly
unstable for carbon and the very high temperature of 5000K, the nvt fix is
not fast enough to equilibrate the system, causing an expansion of the
system which will change the density of the system. The npt fix would not be

that makes no sense as well. with fix nvt, the volume of your system
is not changed, and thus the density *cannot* change.

suitable as the boundary condition along z is not periodic. These are the
two methods I can think of to try and replace the walls at the initial
stage. However, both do not seem to work very well.

Any suggestions would be deeply appreciated. My apologies for the
inconvenience as I am very much inexperience in doing simulation work.

you are worrying about the second problem before you have solved the first.

axel.