Space junk is becoming a more and more of a problem as nation after nation haphazardly chucks stuff up into orbit, and a variety of schemes have been proposed to deal with it, including lasers, explosions and janitor satellites. The latest idea comes from the Naval Research Lab, and it involves giant clouds of dust.
NASA estimates that "there are more than 21,000 orbital debris objects larger than 10 centimeters (roughly four inches) in diameter in low Earth orbit (LEO), and approximately 500,000 object particles between one and 10 centimeters, with the number of debris particles smaller than one centimeter in excess of 100 million." It's those little tiny particles that are the real problem, since they're quite dangerous despite their size, but difficult to spot and keep track of. To deal with junk on this scale, the Naval Research Laboratory is proposing to launch rockets into space that'll release 10 to 20 tons of micron-scale particles of tungsten dust, forming giant space-clouds that can "sweep" huge orbital areas for debris.
Functionally, this dust cloud idea is similar to the atmospheric explosion idea, just 100% less explodey. In both cases, the goal is to increase the density of space itself (using air molecules or dust particles) so that any space junk that passes through the area encounters more drag, slows down and de-orbits itself. The advantage of techniques like these (as opposed to lasers or janitor satellites) is that you don't necessarily have to be able to find and track every single little tiny particle of space junk beforehand: you can just toss a bunch of dust up there knowing that it'll do its job on whatever space junk it encounters, no matter what size.
As for the dust itself, the particles are too small to cause damage to anything, and they'll naturally fall out of orbit and burn up in the atmosphere, just like the estimated 100 tons of cosmic dust that micrometeorites leave behind in our upper atmosphere every day. The NRL is proposing this idea because it's safe, it seems reliable, it'll be relatively cheap and the technology exists to do it right now if we decide that it's a good plan.
One thing's for sure, though: we definitely need to do something, 'cause if we don't, there's a substantial risk of space junk collisions spawning more space junk collisions to the point where space is simply too dangerous for us to visit anymore.
Via NRL