I think this is (as with most things) a complex issue. It's going to depend on your type of sample, type of extraction, sequencing platform, etc.
You might find it worth looking at the kathroseq protocol for low biomass samples, as well as a paper on well-to-well contamination in robotic extraction, which might give you a better idea about both controls and possible sources of contamination. Index hopping is a known issue with Illumina sequencing, as well, so you might find this useful.
You might also want to look at the big thread on decontamination on this forum.
Personally, I think I'd replace your nuclease free water with a single organism control, FWIW.
Best,
Justine