Comment by tbrownaw

A previous time this came up, I had the impression that what it really does is extract energy from a temperature difference between the graphene and the load, which is in principle perfectly normal even is it's a new way to arrange the details.

Edit: missed the link to the abstract.

> However, there is power dissipated by the load resistor, and its time average is exactly equal to the power supplied by the thermal bath.

So yeah, extracting power from a thermal bath sounds exactly like capturing energy as it flows down a temperature gradient.