AQP Seminar: Stochastic Thermodynamics with single-electron devices: how much work can we extract from thermal fluctuations at the single thermodynamic trajectory level?

Stochastic thermodynamics considers single realizations of work and heat relative to a given thermodynamic transformation, unlike standard thermodynamics that deals with averaged quantities over an ensemble of realizations, more appropriate for systems with a large number of degrees of freedom. While the first law of thermodynamics (energy conservation) remains untouched in this context, the second law (entropy increase over time) does not apply because of the stochastic nature of heat and work at the level of a single realization. The probability distributions of work and heat must then be considered: while the second law of thermodynamics is restored only for ensemble averaged quantities, fluctuations of these quantities, e.g. work, obey the so-called nonequilibrium fluctuations relations. In particular, it is possible, for one realization of a thermodynamic process, to extract from a system connected to a single heat bath work greater than the free energy difference, i.e. the bound provided by the second law.

On the experimental side, the ongoing miniaturization of physical devices enable fluctuations of their thermodynamic quantities to become comparable with or larger than their mean values. In this talk we will present an experiment based on a single-electron box (SEB), made of a submicronic metallic island with a gate-tunable electrostatic potential and tunnel coupled to a electrical lead acting as a thermal bath. Single electron tunneling in and out of the box is detected with a sensitive electrometer, which thus monitors stochastic heat transfer events. As such, the SEB is an ideal experimental system for stochastic thermodynamics. From this building block we will then show that the SEB driven with a far from equilibrium gate protocol can provide more work than the free energy difference either to an arbitrary large extent or with an arbitrarily high probability, thus questioning the extent of apparent violations of the second law at the stochastic level.