Physicists at the University of Konstanz and the Max Planck Institute for Dynamics and Self-Organization have demonstrated that Maxwell’s demon can extract up to three times more energy than standard Markovian limits allow by exploiting hidden memories in a complex viscoelastic fluid.
Measuring a Laser-Trapped Bead in a Viscoelastic Fluid
Most experimental setups assume a memoryless environment where any information leaking between the system and its surroundings vanishes instantly, extending from the classic thought experiment first proposed in 1867 that imagines a microscopic gatekeeper sorting gas molecules to generate usable work without an apparent energy input, though modern experimentalists have built real-world versions of this information engine using lasers, colloids, and electronic circuits.
A study published in Nature Physics by Lokesh Muruga, Felix Ginot, Sarah A. M. Loos, and Clemens Bechinger challenges that premise. The team confined a microscopic glass bead inside a complex liquid using a tightly focused laser beam. Unlike water, this specialized medium contains a spaghetti-like network of microscopic structures called micelles that continuously rearrange and interact with the trapped particle.
Harvesting Energy Through a Double-Measurement Strategy
As the glass bead undergoes Brownian motion and moves through the fluid, it deforms the surrounding micellar network. The network relaxes only gradually over time, storing traces of the bead’s past motion and influencing its future trajectory. Because observers cannot track the internal dynamics of the microscopic background medium directly, part of the relevant control information remains hidden.
To recover that missing knowledge, the researchers implemented a double-measurement protocol. They recorded particle trajectories at 100 frames per second with a spatial resolution of roughly five nanometres. The demon eavesdrops on the environment’s memory by comparing the particle’s position against a fixed threshold at a first moment, and then performing a second binary measurement after a controllable delay. This yields four possible two-bit states: |00>, |01>, |10>, and |11>.
We measured the particle twice with a defined temporal delay in between, In this way, we can gain access to otherwise hidden environmental dynamics, ultimately increasing the amount of exploitable work.
Clemens Bechinger and Sarah Loos, via Nanowerk
Surpassing Markovian Work Limits in the Laboratory
The team found an optimal sweet spot at an interval of around one second after determining that if the measurements occurred too close together, they revealed little about the fluid’s hidden dynamics, whereas if spaced too far apart, the network had already forgotten the bead’s earlier movements. The researchers distilled this discrepancy into an integrated order parameter that vanishes for Markovian dynamics and peaks sharply at this temporal window, which is set jointly by the micellar relaxation time and the particle’s relaxation in the trap.
By applying this technique in designated parameters, the team successfully gathered as much as threefold the work output compared to the theoretical threshold for a memory-free baseline configuration. The |10> state delivered the largest relative boost in work beyond the Markovian limit, while the |11> state yielded the largest absolute work due to ordinary potential energy contributions.
Implications for Microscopic Information Engines
Takahiro Sagawa and Masahito Ueda formulated the theoretical basis in 2010 through their work on the generalized second law of thermodynamics tailored for feedback-controlled systems. Their equation establishes that extractable net work obeys bounds determined by free-energy differences and mutual information gained from measurements. In a cyclic machine where free-energy differences vanish, all extracted work must be paid for in information.
These microscopic engines operate without a temperature difference or macroscopic energy gradient, using information about fluctuations as a thermodynamic resource instead. The work was supported by an ERC Advanced Grant awarded to Clemens Bechinger for research on Brownian particles in non-equilibrium baths.