"Big hole" in the solution after running MD

Dear All,

I'm running a MD on a solid-solution-solid sandwich-like system to see the adsorption of ions in solution to the surface of the solid phase. After running the simultion, some water molecules penetrated into the solid, ions in the solution were adsorbed on the surface. But there was a big less concondensed area in the middle of the solution, which looked like a "big hole". In the concentration profile for oxygen in water molecules, water molecules showed big peaks at the near surface area, and there was a big dent  in the middle plane area of the solution, and this phenomenon became more evident with increaing concentration of the solution. However, this "big hole" was not alway located at the same position, if I compare different runs for the system with same solution concentration. This brought a big problem when I tried to calculate the interaction energy between ions and the solid phase, it looked like the "big hole" prevent the free diffusion of ions, the interaction energy showed big difference between different runs. I ran NVT for the dynamic. I also tried to run NPT first to equilibrate the system, then use NVT to get the dynamic date, but there was still a big hole in the solution.

Does any one have a solution for this? Thank you

Regards,

Jingjing