Glass / silicates - 0.005mg
Sapphire - 0.008mg
Steel - 0.016mg
Lead - 0.025mg
Gold - 0.05mg
This means that relatively weak forces, e.g. electrostatics, could be enough to move him.
Of course this light mass gives its own problems. Ignoring the possibility that he's something really exotic like neutronium, he still has tiny mass and a relatively big perimeter, about 600mm, with a surface area around the edges of 0.0006mm3 (yes, I know there's all sorts of reasons why this is nonsense, but bear with me, this is all hand-waving anyway).
So... with that big a perimeter he must experience a significant number of molecular collisions every second. My question - and this is where I need help - are we talking so many that they even out, or few enough that Brownian motion could occur? And if so, how do we go about converting random impacts into controlled movement? Would it be possible to change the number of impacts in a particular part of the perimeter by e.g. vibrating an edge slightly?