The holy-grail of smoke color for modeling is black. It sure would be interesting to get a handle on how ultrasonic "atomization" transducers handle the difference between dyes vs. pigments ... or suspended vs. dissolved particles.
There is no shortage of self-proclaimed expertise on these ultrasonic transducers such as whether you should use distilled water vs. "tap" water and so on. I have zero expertise in this regard. But I do see vendors selling these atomizers for aromatherapy so "something" more than plain-old H20 can be emitted.
The good news is you won't break the bank messing with these gadgets. If you "clog up" the transducer with an incompatible liquid, it can be replaced for maybe $1. As with our heater-resistors where you can burn/char the wick if operating dry; you should not operate ultrasonic transducers "dry". While the mist is cool, the vibrating transducers get hot to the touch if dry...and possibly irreversibly fail. The metaphor is almost uncanny! That is, the concept of replacing the smoke wick and possibly even the smoke heater-resistors is akin to replacing the ultrasonic transducers which themselves have a finite life albeit typically in the thousands of hours.
OTOH if you're not after black, in previous threads I've shown how a colored LED can provide a reasonable illusion of colored smoke by illuminating the smoke stream. Here's photo from an earlier OGR thread where a blue LED models the "blue" smoke from a diesel engine. AFAIK there is no such thing as a black LED. 
