I guess it uses tunneling to create a thermal gradient. Instead of moving energy itself (violating the 2nd law of thermodynamics), move more energetic particles themselves to one side and isolate less energetic particles on the other side. Somehow, the particles are encouraged to self-order without the application of external motivation.
Then you can generate a small current through the thermoelectric effect. That’s neat!
But I’m skeptical that random tunneling of large particles will be common enough that it creates a long lasting thermal gradient (or even one that’s statistically significant enough to be measurable). Also, every time one of those particles collides with something in the environment, they trade energy to reach equilibrium.
And the De Broglie wavelength of anything much larger than a free electron becomes short enough that the tunneling effect is negligible. Whatever is tunneling must be large enough to carry no net charge. So tunneling seems infinitesimally low probability.
In other words, I’m skeptical.