[gmx-users] Surface tension coupling

[Mehmet Sayar]

Hi,

I am trying to obtain the surface tension for the hexane-water interface.
I used Gromacs 4.0.7.

A simulation where I keep the X-Y dimensions fixed and let z fluctuate yields
the following output:

Energy                      Average       RMSD     Fluct.      Drift  Tot-Drift
-------------------------------------------------------------------------------
Box-X                             8          0          0          0          0
Box-Y                             8          0          0          0          0
Box-Z                       9.90151 0.00902362 0.00902139 1.00596e-07
0.000694869
Volume                      633.697   0.577512   0.577369 6.43816e-06  0.0444716
Density (SI)                790.105   0.720108    0.71993 -8.03015e-06
-0.0554683
Pres-XX (bar)               -93.007    111.668    111.668 -5.85703e-05
 -0.404574
Pres-YY (bar)              -93.8345    111.464    111.461 -0.000388743
  -2.68524
Pres-ZZ (bar)               1.04823    102.795    102.795 7.14103e-05   0.493267
#Surf*SurfTen               935.415    1174.75    1174.74 0.00292727    20.2201

In order to test this value, I have done a constant surface simulation.
I expect that this run should also give me a stable system.

In this case I set the X-Y dimensions free, fixed the Z dimension by setting
the compressibility to zero. I set the surface tension to 935, which I
obtained from the first simulation.
This simulation yields completely bizarre results:


Energy                      Average       RMSD     Fluct.      Drift  Tot-Drift
-------------------------------------------------------------------------------
Box-X                       7.22688   0.478216  0.0850938 -0.00115819   -1.63015
Box-Y                       7.22688   0.478216  0.0850938 -0.00115819   -1.63015
Box-Z                       9.90248          0          0          0          0
Volume                      519.449    68.6696    12.4766  -0.166195   -233.919
Density (SI)                980.851    128.888    23.3119   0.311983    439.117
Pres-XX (bar)               8178.74    7814.17    2469.25    18.2466    25682.1
Pres-YY (bar)               8189.38    7827.51    2458.45      18.29    25743.2
Pres-ZZ (bar)               8284.24    7819.63     2462.7    18.2661    25709.6
#Surf*SurfTen               992.022    1516.44    1516.41 -0.0216172   -30.4263


The section from the .mdp file I used is as follows:
; Pressure coupling is  on
Pcoupl              = berendsen
tau_p               = 1.0 1.0
Pcoupltype          = surface-tension
compressibility     = 4.5e-5 0
ref_p               = 935.605 1.0


Some how, the system shrinks in X-Y dimensions rather than expand, and
this yields strange pressure values.
I does maintain the target surface-tension more or less, but the
density is completely unrealistic.

I also tested letting the z-dimension also free, which gave more
reasonable results.
The options are set to:
; Pressure coupling is  on
Pcoupl              = berendsen
tau_p               = 1.0 1.0
Pcoupltype          = surface-tension
compressibility     = 4.5e-5 4.5e-5
ref_p               = 935.605 1.0

Energy                      Average       RMSD     Fluct.      Drift  Tot-Drift
-------------------------------------------------------------------------------
Box-X                       7.88739  0.0533823  0.0365402 -0.000116266
 -0.134811
Box-Y                       7.88739  0.0533823  0.0365402 -0.000116266
 -0.134811
Box-Z                       10.1838    0.13604  0.0933104 0.000295756   0.342929
Volume                      633.462    0.92414   0.917408 -0.000332644
 -0.385701
Density (SI)                790.398    1.15266    1.14426 0.000414983   0.481174
Pres-XX (bar)              -90.6569     105.24    105.186   0.010099    11.7098
Pres-YY (bar)              -91.2499    104.286    104.267 0.00588557    6.82433
Pres-ZZ (bar)               1.79822    95.6962    95.6725 0.00636455    7.37971
#Surf*SurfTen               944.344    1095.81     1095.8   0.010786    12.5064

However in this case also the system does not seem to be stable, but
slowly the z dimension increases.

Any ideas? Am I doing a mistake in the .mdp file, or is there a
problem with the surface tension coupling?

Thanks in advance,

Mehmet