On the surface tension of plasmas