A quantum-dot Szilard engine puts fluctuation theorems to the experimental test
David Barker, Ville Maisi, Patrick Potts and collaborators implemented a Szilard engine in a semiconductor quantum dot with real-time single-electron charge detection and feedback.
Running a forward and two distinct backward protocols, they experimentally tested detailed and integral fluctuation theorems together with thermodynamic uncertainty relations on one platform. A key finding is that the entropy production inferable from measurement can be a more relevant quantifier of the information powering the engine than the widely used mutual information.
The experiment demonstrates measurement-and-feedback work extraction from an equilibrium environment — the canonical Maxwell's-demon scenario — while sharpening the conceptual accounting of how information converts into work.