compute heat flux from a heated gold nanoparticle

Dear all,
I want to compute heat flux from a heated gold nanoparticle (GNP) to the surrounding fluid, I am using the command as it mentioned in the mannual,

compute      myKE GNP ke/atom
compute      myPE GNP pe/atom
compute      myStress GNP stress/atom virial
compute      flux all heat/flux myKE myPE myStress
variable     Jx equal c_flux[1]/vol
variable     Jy equal c_flux[2]/vol
variable     Jz equal c_flux[3]/vol
should the volume which each flux array is divided be the volume of the gold nanoparticle or the volume of the whole system?

Ragards,
Fatemeh

Dear all,
I want to compute heat flux from a heated gold nanoparticle (GNP) to the surrounding fluid, I am using the command as it mentioned in the manual, but I am not sure if I should compute the heat flux from the GNP group or all of the atoms including GNP and fluid particles?!

compute      myKE **GNP** ke/atom
compute      myPE GNP pe/atom
compute      myStress GNP stress/atom virial
compute      flux all heat/flux myKE myPE myStress
variable     Jx equal c_flux[1]/vol
variable     Jy equal c_flux[2]/vol
variable     Jz equal c_flux[3]/vol
if I should calculate the flux from GNP should the volume which each flux array is divided be the volume of the gold nanoparticle or the volume of the whole system? 

Ragards,

Fatemeh

I think it should be just the volume of the nanoparticle (Np). I’m not clear
that you can use the compute heat/flux command this way
since there is no “vector” of directional heat flux for the Np.

Reese can comment.

A sanity check on what you are doing, or an alternative method
would be to monitor the temperature or kinetic energy of the Np.

If you are heating it, then that heat you are adding minus the
heat that moves to the fluid should equal the temperature.

Steve

I think you probably want to work this out on a per atom basis using atom style variables to implement the heat flux computation, then do some sort of spherical averaging over those “atomic heat flux” vectors.

Compute heat/flux is meant to give you the best flux vector the the entire box - if your nanoparticle has spherical symmetry, this should average to zero.

Dear Mr.Jackson

Dear Mr.Jackson
Thank you very much for your answer, I really appreciate it.
You mean I compute PE, Stress, KE for each atom and after adding them
toghether, use the ave spatial in spherical bins to do the averaging? Could
you you please add more information

i think at this point, you should spend some time reading text books,
go back to the basics and work out the theoretical foundation of what
you want to compute.
there obviously is not a ready-made option for it available in LAMMPS
and i doubt that just some random guessing like you describe will give
a satisfactory answer.

axel.

Dear Dr. Plimpton
I use velocity rescaling to keep the GNP temperature constant at each time step. and also How can I compute the volume of GNP.

Thanks in advace for your help.

Dear Dr. Plimpton

I use velocity rescaling to keep the GNP temperature constant at each time
step

that is not a very good model. velocity rescaling in general is a bad
approach for a number of reasons that have been repeated so many times
that i don't want to do it again. you should be easily find
discussions on the subject matter in the mailing list archives.

but also, even if you'd use a more gentle way of thermalization, how
do you explain that kind of exchange of kinetic energy with a real
world model? a thermostat is generally meant to couple the entire bulk
of your system to a larger heat reservoir and thus allow a small box
with periodic boundaries to show the same statistical sampling
behavior as a much larger system.

. and also How can I compute the volume of GNP.

also, this is a frequently discussed topic. so please search the
mailing list archive for further details. in short, the volume of
nano-scale particle is not a well defined entity (what is the volume
of a single atom?) and can at best only be approximated.

Hi Fateme
First if all is the system in thermodynamic equilibrium? If so you need to use something like a green kubo method - I advise against this given your systems configuration. You should drive the system to steady state by heating the particle while cooling the fluid. Then you can estimate the heat flux normal to the surface by

Average heating rate / surface area of the particle.

Reese Jones
[email protected]…3…
925 294-4744

Dear Dr. Jones
Thanks for your answers,the system I am simulating consists a Gold nanoparticle immersed in a fluid, the GNP is heating during the simulation and the fluid in distances more than 10 A is rescaled to a less temperature creating a heat sink. and yes it has reached the steady state, and the problem is NEMD .
And is average heating rate is a lammps command as i can not find it in the mannual?

I think fix rescale outputs a scalar that you can time average

Reese Jones
[email protected]…3…
925 294-4744

Dear Fateme

You can try about radial heat flux for each atoms then sum them to total radial heat flux. if j_x,j_y,and j_z be equal to heat flux for one atom then you can determine heat flux j.
image.png

compute Stress all stress/atom virial
compute KE all ke/atom
compute PE all pe/atom
variable Etot atom c_PE+c_KE

variable xm equal 48.50694 # x component of center of mass GNP
variable ym equal 50.339 # y component of center of mass GNP

variable jcx atom v_Etot*vx # the first term in J formula (x component).
variable jvx atom ((c_Stress[1]*vx)+(c_Stress[4]*vy)+(c_Stress[5]*vz))/1.6021765e6 # nktv2p = 1.6021765e6 # the second term in J formula (x component).
variable jx atom v_jcx-v_jvx # (x component of J for each atom)

variable jcy atom v_Etot*vy # the first term in J formula (y component).
variable jvy atom ((c_Stress[4]*vx)+(c_Stress[2]*vy)+(c_Stress[6]*vz))/1.6021765e6 # nktv2p = 1.6021765e6 #the second term in J formula (y component).
variable jy atom v_jcy-v_jvy # (y component of J for each atom)

variable jr atom ((x-v_xm)*v_jx+(y-v_ym)*v_jy)/sqrt((x-v_xm)^2+(y-v_ym)^2) # jr (for each atom) can be determine via J vector dot radial direction.

compute 2 all reduce/region 2 sum v_jr # summation jr for each atom in this region.

Cheers
Yasti

Dear Yasti,
I really appreciate your answer, I have another question should I follow the same procedure if my system is 3 dimensional, and how should I get the output from this command? with thermo custom command?
Tahnk you and looking forward to hearing from you.

Kind Regards
Fateme

image.png

Dear Yasti,
I really appreciate your answer, I have another question how should I calculate if my system is 3 dimensional, and how should I get the output from this command? with thermo custom command?
and also what is this number indicating (nktv2p = 1.6021765e6)?
Thank you and looking forward to hearing from you.

image.png

Dear Yasti,
I really appreciate your answer, I have another question how should I calculate if my system is 3 dimensional, and how should I get the output from this command? with thermo custom command?
and also what is this number indicating (nktv2p = 1.6021765e6)?
Thank you and looking forward to hearing from you.
Regards
Fateme

image.png

The output from any fix or compute or variable that produces a global

value (as opposed to per-atom) is the same. It is discussed
in Section 6.15 of the manual.

Steve

image.png