dielectric command problem

Hi all,
I have a problem in using the “dielectric” command. I want to simulate the adsorption process of protein on the TiO2 surface in the implicit solvent. Periodic boundary condition were applied to the x- and y- direction, and the reflecting boundary condition was applied in the z-direction (I used the slab version). The dielectric constant of water at 37℃ is about 74.31, so I think it is necessary to put the “dielectric 74.31” command in the input script to reduce the interatomic coulombic interactions (i.e., the system was in the implicit solvent).
In my simulation, I fixed the bottom of the substrate but freed the top layer of TiO2 (relaxing the top layer of TiO2 in the NVT ensemble during the entire simulation). The results showed that the TiO2 atoms in the free section moved dramatically along the direction perpendicular to the surface (i.e. +z direction) until the atom were lost. If I set the entire substrate free, the thing was more weird. The TiO2 atoms moved laterally in opposite directions (the lower part of TiO2 atoms moved along the –z direction, but the higher part moved along the +z direction). If I delete the “dielectric” command in my script file (the default=1, for the explicit solvent), the TiO2 atoms moved normally even when I freed the whole substrate.
The thing I am wondering is whether it is reasonable to make the free TiO2 atoms in an environment of dielectric=74.31. Does the dielectric command apply to the global system? Can we just apply it to the specific section, such as the interactions between protein and the substrate or the interactions between the protein atoms?
Many thanks

2011/10/17 郑庭 <[email protected]...>:

Hi all,
I have a problem in using the “dielectric” command. I want to simulate the
adsorption process of protein on the TiO2 surface in the implicit solvent.

i think this is your problem. i don't think that using scaled
electrostatics is a good approximation of the impact of
water for a system like this.

Periodic boundary condition were applied to the x- and y- direction, and the
reflecting boundary condition was applied in the z-direction (I used the
slab version). The dielectric constant of water at 37℃ is about 74.31, so I
think it is necessary to put the "dielectric 74.31" command in the input
script to reduce the interatomic coulombic interactions (i.e., the system
was in the implicit solvent).
In my simulation, I fixed the bottom of the substrate but freed the top
layer of TiO2 (relaxing the top layer of TiO2 in the NVT ensemble during the
entire simulation). The results showed that the TiO2 atoms in the free
section moved dramatically along the direction perpendicular to the surface
(i.e. +z direction) until the atom were lost. If I set the entire substrate
free, the thing was more weird. The TiO2 atoms moved laterally in opposite
directions (the lower part of TiO2 atoms moved along the –z direction, but
the higher part moved along the +z direction). If I delete the “dielectric”
command in my script file (the default=1, for the explicit solvent), the
TiO2 atoms moved normally even when I freed the whole substrate.
The thing I am wondering is whether it is reasonable to make the free TiO2
atoms in an environment of dielectric=74.31. Does the dielectric command
apply to the global system? Can we just apply it to the specific section,

yes.

such as the interactions between protein and the substrate or the
interactions between the protein atoms?

no.

axel.