There is a lot of Oxygen and a tiny tiny tiny amount of Hydrogen in the air. It's in not equilibrium, so they "burn" slowly and "produce" some energy following the first and second thermodynamics law. Dew to the second law, part of that energy has to be dissipated as heat, and only a part can be potentially harvested in a useful form like electricity.

The idea is to harvest that energy, to do that they use the graphene membrane as a part of a device that separate the Hydrogen. The separation requires that some additional energy is dissipated as heat, because of the entropy requirements of the second law.

If they can get more useful energy "burning" the separated Hydrogen than the energy necessary to separate the Hydrogen, then they get a useful device. I'd have to do the calculations using the concentration of the Hydrogen in the atmosphere, but I'm too lazy.

I guess that this is theoretically possible is you use the maximal efficiency in each and every step, but in the real word the efficiency is much lower and you get no useful energy.

I think there is a little hope that this can be useful with industrial fumes that have more Hydrogen to separate the contaminants, but the application with air is only a linkbait or a grantbait.