Was this published? Even if it did not work, I would like to read any paper on it. It is always good to know what does not work. I am working on a similar problem (3-5 nm square nanochannels) which will have a varying number charges or different chemical groups attached to the inside. I can get by with the slab geometry, unless the corners turn out to be important. As Steve suggested, there should be a way to correct 3D to 1D like the 3D to 2D, although it would be more complicated and the correction would depend on at least 2 directions of the dipole moment of the simulation box instead of just 1.