Dear all,
In order to investigate confined flows in a slab geometry, I usually
use "frozen" walls -with all wall atoms moving as a block, and in this
case it is easy to use them as a piston to control the pressure:
fix 2 hi aveforce NULL -${fy} NULL
fix 3 hi setforce 0.0 NULL 0.0
fix 4 hi nve
Or to apply a shear stress:
fix 2 hi aveforce ${fx} NULL NULL
fix 3 hi setforce NULL 0.0 0.0
fix 4 hi nve
On the other hand, it is possible to thermostat immobile walls by
enabling wall atoms to vibrate around their equilibrium position with
fix spring/self.
Now I would like to combine both things, namely to enable the
"initial" positions of the atoms in the fix spring/self command to
move as some ghosts atoms...
I was thinking of defining fictitious atoms, with a harmonic bond to
real wall atoms, and use fix aveforce and setforce on these fictitious
atoms to force them to move as a block, but maybe someone can suggest
me a more clever way?
Another possibility would be to use a layer of frozen atoms to apply
pressure and/or shear stress, and to add a layer of free atoms with a
large interaction energy to ensure it remains in the solid state and
stuck to the frozen wall.
What do you think of these two methods, and do you have any suggestion?
Thank you,
Best regards,
Laurent
[...]
Now I would like to combine both things, namely to enable the
"initial" positions of the atoms in the fix spring/self command to
move as some ghosts atoms...
I was thinking of defining fictitious atoms, with a harmonic bond to
real wall atoms, and use fix aveforce and setforce on these fictitious
atoms to force them to move as a block, but maybe someone can suggest
me a more clever way?
Another possibility would be to use a layer of frozen atoms to apply
pressure and/or shear stress, and to add a layer of free atoms with a
large interaction energy to ensure it remains in the solid state and
stuck to the frozen wall.
i would go for the second option, but if you use a large interaction
energy, they will act like frozen atoms. instead i would use a "normal"
non-bonded interaction and then attach them to the frozen wall with
harmonic bonds. the frozen and the attached atoms should have
different atom types so that you can turn off their non-bonded interactions
and dimension the harmonic force constant so that it corresponds to
the non-bonded potential for the attached wall potential in the minimum.
cheers,
axel.
Thank you for your suggestion, I will definitely give it a try. I made
a quick test with a strong non-bonded interaction: epsilon = 3.0 seems
enough to preserve the integrity of the walls, and wall atoms are
still vibrating significantly. But energy transfer between such a
stiff wall and the liquid may not be very good, so I will probably go
for your option since I will be able to tune the bonded interaction
energy over a wider range...
Best,
Laurent