I read that part a few times, but it makes no sence.

I prefer another quote

> What we did was reroute the current in the circuit and transform it into something useful.

Let's suppose that they get some "useful" energy to turn on the lamp like in the animation or power a small device.

Let's suppose that you use it to power a laser and send a beam that heat some object far away. As a side effect, you are extracting energy from the device with the graphene in the lab so it will get cooler.

So the net effect is that the device in the lab gets cooler and the object far away get hotter. So you have a flux of heat. But if the other object is hotter, you have a lux of heat from in the wrong direction, that is impossible according to the Second Law.

There are a million ways to rewrite this https://en.wikipedia.org/wiki/Second_law_of_thermodynamics in more abstract or more concrete ways. With some oversimplifications, another is that you need at least two heat baths and the efficiency of the device to transform heat from the "hot" bath to of "useful" energy can be calculate using the temperatures of the baths. When the difference of temperature is zero, the efficiency is zero and the device can produce no "useful" energy. Or in other words, with only one heath bath, you can produce no "useful" energy.

>So the net effect is that the device in the lab gets cooler and the object far away get hotter. So you have a flux of heat. But if the other object is hotter, you have a lux of heat from in the wrong direction, that is impossible according to the Second Law

I don't think your hypothesis would necessarily break the Second Law. Wouldn't the hotter distant object result in making Brownian motion around it? The same Brownian motion which would eventually end up as energy input at the graphene membrane - provided a thermodynamically closed system?

As so I'm not convinced this study breaks the Second Law, nor do I see how it produces useful work. To me it only seems so, because the system isn't being modeled as a closed system...

EDIT: Actually, I'm pretty sure the following statement is completely false:

>the graphene and circuit are at the same temperature and heat does not flow between the two.

I'd suspect heat does flow between the two, it's just outside their modeling of the system.

So, spooky thermocouples at a distance? This paper is one that really needs further attempts at replication.