Like a collection of flotsam propelled along the surface by waves traveling across the ocean, it turns out that a thermal wave — a moving pulse of heat — traveling along a microscopic wire can drive electrons along, creating an electrical current.
The key ingredient in the recipe is carbon nanotubes — submicroscopic hollow tubes made of a chicken-wire-like lattice of carbon atoms. These tubes, just a few billionths of a meter (nanometers) in diameter, are part of a family of novel carbon molecules, including buckyballs and graphene sheets, that have been the subject of intensive worldwide research over the last two decades.
This so-called "thermopower" may or may not be an interesting breakthrough with some clean energy applications: it's always very difficult to predict how quickly or to what extent a new insight can bring forth new innovations. It is a reminder, though, of how rapidly the frontier of knowledge is still advancing. What is possible today is not what will be possible next year, much less in 20 years. This is particularly true when it comes to nanotechnology (which I think it's clear will have some major implications for materials design, smart zero-waste systems and post-ownership in particular).
At the same time, it's also worth seeing the complexities inherent in a story like this.
For one thing, this is not (despite tweets to the contrary) the "solution" to our sustainability problems. It's pretty clear that barring truly revolutionary discoveries (ones that change our understandings of physics, pretty much), no new technologies are going to make the Swap possible. Too many aspects of the current suburban American version of prosperity are simply unsustainable (for reasons that don't appear to be changeable) to make it likely that even big improvements in some of the component technologies can save the whole system. The technofix is not a menu option.
The other thing is that nanotechnologies like thermopower are almost certain to create their own attendant problems. Second-order effects come with the territory. We will need new regulatory systems to anticipate and control those impacts, certainly. But on many, many questions -- like How serious a threat is nanopollution? -- we don't even know enough yet to know how to sensibly regulate deployment, and since that means we also don't know how to undo any damage we might cause, we probably ought to be defaulting to a precautionary stance. The alternative is trying to respond to unanticipated nanoindustrial catastrophes. We ought to get the frameworks we need in place before we find ourselves dealing with buckyjunk or brain-abrading nanoparticles (or gray goo for that matter).