Hi DR,
I use something similar for my water where I use the C18 disks for filtration but instead of a C18 column, I use a penta-flouro-phenyl. Anecdotally, I see the exact same quality of impurity removal on the PFP column when compared to the C18 column. The synthesis I run on the PFP column has an almost identical chromatogram when I run it on the C18 column (barring a few select peak-- the diastereomers) with the same method. This case might be unique because the phases provide me with nearly identical chromatograms, and suggests that perhaps the phases are more-similar than I would anticipate, but I don't see any ghost peaks between the two. My guess would be that any of the ghost peaks you are used to seeing on your C18 (from the water) would be identically removed when switching to the CN. If you didn't filter them, then these impurities will likely still form ghost peaks on the CN column, albeit with potentially different retention, etc. However, the CN column would probably retain a completely different set of ghost-peaks when compared to the C18 phase and you may be seeing new ghost peaks resulting at those later gradients. The best way to check is to run a sample! I would certainly expect most of the ghost peaks from your original water on C18 issues to still vanish, but potentially there are other impurities in your solution that aren't retained or are irreversibly absorbed onto your C18 that may not be on the CN column. Either way, it definitely does not hurt to continue filtering in this manner (unless the filter/disk contributes some impurity when you're filtering).