The paper is pay-walled so could not read the mandatory discussion on 2nd law of thermodynamics. The underlying idea is not new. Brillouin published Can the Rectifier Become a Thermodynamical Demon? in 1950 [0] [1]. I hope these are cited in the paper.
[0] https://journals.aps.org/pr/abstract/10.1103/PhysRev.78.627....
[1] https://pubs.aip.org/aip/apl/article-abstract/14/2/54/41611/...
The paper you mention is indeed cited.
Huh. Is this for real? Most of the discussion assumes an ideal diode. Real diodes have diode drop. All this is powered by very small fluctuations, right. They'd have to overcome the diode drop.
(There is a real component called an "ideal diode", which is a power MOSFET with control circuitry to turn it on when the voltage is higher in the forward direction. MOSFETs have ON resistances in millohms, which is better than diodes can do. Often used with solar powered battery chargers, large DC-DC converters, etc. Not sure that applies here.)