Water molecules stick to wall during evaporation simulation

Dear all,

I am running a molecular dynamic simulation to model evaporation of water from a hot solid substrate to cold solid substrate. I followed the same simulation strategy as described in the paper https://doi.org/10.1016/j.ijheatmasstransfer.2011.09.035. My simulation domain is a 50 x 80 x 320 box. Two solid silicon blocks with a thickness of 18 A are placed at the top and bottom of the simulation domain. Initially, two water blocks with a thickness of 32 A are placed on the bottom silicon block and beneath the top silicon block, respectively. During the simulation, the top and bottom layer of the solid block are fixed in space. The whole system is first minimized, then equilibrated at 350K for 50 ps under NVT ensemble. Next, the vapor water molecules are relaxed under NVE ensemble for 100 ps before evaporation. During the NVE process, however, all the water molecules still stick to the silicon wall even after relaxation in NVE for 200 ps. The mid space which was supposed to be occupied by the vapor molecules are still completely empty. I modified the potential parameters with multiple values and but no water molecules diffuse into the vapor domain. Does anyone know any possible cause for this?

The potential force I used are:


   1  15.999400 # o
   2   1.007970 # h
   3  28.086000 # si

Pair	Coeffs # harmonic

   1   0.15207	3.15065
   2   0.00000	0.00000
   3   49.99530	2.10000
Bond Coeffs # harmonic

   1   540.6336     0.9600 # o-h
   2   180.9273     2.3515 # si-si

Angle Coeffs # harmonic

   1    50.0000   104.5000 # h-o-h
   2   119.5029   109.4700 # si-si-si

Dihedral Coeffs # harmonic

  1    15.4900   1   3 # si-si-si-si