the density is close to water, but that all depends on how much nanoparticles you mix in. We have shown no pumping effects nor any scouring of the impeller etc. We(me and a friend) did a research project for the airforce looking at an optimum amount. I am currently running some in my liquid cooling system. We showed a 16% increase in heat transfer over water. My CPU temp dropped almost 16% after I changed from the stock liquid to the nanofluid.

We did all of that. Remember, we did this project for the US Air Force. Everything was carefully monitored, weighed, calibrated, etc etc. We are presenting a paper at the Heat Transfer Conference in Jacksonville, FL. You can't replace heat pipe fluid with nanofluids because heat pipes use phase-change, which would NOT work with nanofluids.

First off, it doesn't look like mercury..it looks like milk to be honest. When you add nanoparticles to a fluid you can do it by volume or weight..as in Volume% or Weight%. We did ours by Volume..it is sort of like measuring by feet or inches. The Volume% has to be optimized or it will become to heavy and then you will run into pumping problems, scouring...etc.
#1. If you were to spill this it would be no different from spilling water on an electrical part. You just have to clean it up.
#2 It is not smaller than a water molecule (which is WAY small)..on the order of 10^-19..we are at 10^-9..no leaking issues.
#3 If the system is running you will not have algomeration, but if you turn it off for a week you will get some settling. If you are turning your system off for weeks at a time you don't need to worry about nanofluids.