[gmx-users] Water Liquid-vapor interface: Negative surface tension by virial formula?

[WU Yanbin]

Dear GMXers,

I'm computing the liquid-vapor surface tension for SPC/E water using
GROMACS 4.5.3.
The initial water box dimension is 3nm*3nm*3nm, containing 1807 water
molecules, which is extended in z direction to a length of 12nm, to create
two liquid-vapor interfaces. The ensemble is NVT.
The system was equilibrated for 1ns and ran for 60ns for data collection.

Below is the pressure and virial output using g_energy:

Energy                      Average   Err.Est.       RMSD  Tot-Drift
-------------------------------------------------------------------------------
Vir-XX                       4809.7        1.3    721.805   -5.05917
(kJ/mol)
Vir-YY                      4810.68          1    723.077   -3.29302
(kJ/mol)
Vir-ZZ                      4536.49        0.1    715.838  -0.434853
(kJ/mol)
Pres-XX                    -97.9413       0.39    223.717     1.4721  (bar)
Pres-YY                    -98.2664       0.33    224.085    1.21738  (bar)
Pres-ZZ                   -0.204398      0.012    221.921  0.0134426  (bar)

If I compute the surface tension using gamma = (pzz-0.5*(pxx+pyy))*Lz/2,
the resulting surface tension is 58.7mN/m.
But if the surface tension is computed using
gamma=(PIzz-0.5*(PIxx+PIyy))/Axy, the resulting surface tension is
-25.3mN/m, which is negative.

Is there anything wrong with my calculation?
I will be happy to provide any further information regarding the simulation
setup if needed.

Thank you in advance.

Best,
Yanbin
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20111107/58dc9fcf/attachment.html>