🔬 condensed matter

Piston-Like Information Engine II: Boundary-Controlled Optimum in Active Matter

This paper demonstrates that a boundary-controlled information engine utilizing self-propelled bristle bots exhibits a qualitatively distinct work-output relationship and a density-dependent regime switch compared to thermal systems, a nonequilibrium feature attributed to the accumulation of active particles near boundaries.

Original authors: Laura Hoek, Neta Ben Ari, Rémi Goerlich, Saar Rahav, Yael Roichman

Published 2026-09-03

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Laura Hoek, Neta Ben Ari, Rémi Goerlich, Saar Rahav, Yael Roichman

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine a world where the smallest movements of matter are not just random jitters, but a source of potential energy waiting to be harvested. For over a century, scientists have been fascinated by the idea of turning information into work. This concept, rooted in the famous thought experiments of the 19th century, suggests that if you know exactly where a particle is, you can use that knowledge to push it in a useful direction, effectively extracting energy from the chaos of heat. In the last decade, researchers have built tiny machines that do exactly this, using light or magnetic fields to trap and move microscopic particles. These "information engines" work beautifully when the particles are passive, simply bouncing around due to thermal energy, much like dust motes in a sunbeam. But nature offers a more chaotic and energetic class of particles: active matter. These are self-propelled entities, like bacteria or synthetic robots, that generate their own motion. They do not just jiggle; they push, swarm, and crash into walls with their own internal power. The question that drives this new research is simple yet profound: does the same logic of information-to-energy conversion hold true when the fuel is not heat, but the relentless, self-driven activity of these living-like machines?

A team of researchers in Israel has built a large-scale version of this engine to find out. Instead of using invisible light traps on microscopic beads, they constructed a macroscopic machine using a rectangular arena and a collection of small, vibrating robots they call "bristle bots." Each bot is about four centimeters long, powered by a tiny battery and motor that vibrates its flexible bristles, causing it to scuttle across the floor. The researchers placed a movable wall, acting as a piston, at one end of the arena. The engine operates on a simple, repetitive cycle. First, the system waits and watches a specific zone near the piston. If that zone happens to be empty, the piston is allowed to slide inward, compressing the swarm of bots. If the zone is occupied, the piston stays put. By repeating this process, the bots are gradually squeezed into a smaller space without the machine doing any direct pushing work; the energy comes entirely from the bots' own movement and the clever timing of the wall's motion. The researchers measured how much work was stored in the compressed gas of bots and compared it to the probability of finding an empty space to compress.

In the world of passive, thermal particles, physics predicts a very specific relationship between the work extracted and the probability of an empty space. There is a universal rule that dictates the most efficient way to run such an engine, suggesting that the best performance happens when the chance of finding an empty spot is roughly one in three. However, when the researchers ran their experiment with the active, self-propelled bristle bots, the results were strikingly different. The engine did not follow the standard rule. Instead, the most efficient way to run the machine changed depending on how crowded the arena was. At low densities, the engine worked best when the detection zone was large, sweeping up a wide area to find a rare empty spot. But as the researchers added more bots, making the arena more crowded, the optimal strategy flipped. Suddenly, the best performance came from using a very small detection zone, looking only for a tiny gap right next to the wall.

This switch in strategy is caused by a unique behavior of active particles: they tend to pile up against the walls. Because these bots are constantly pushing themselves forward, they accumulate in dense layers along the boundaries of their container, leaving the center of the arena relatively empty. When the bots are few, this accumulation layer is thin, and the engine can afford to look at large areas. But as the density increases, the wall becomes packed with bots, making it very hard to find a large empty space near the piston. The engine adapts by focusing its search on the tiny, fleeting gaps that appear within this dense wall layer. The researchers confirmed that this behavior is not just a quirk of their specific setup; they simulated the process with dry friction and found that the same transition between strategies occurs even when the machine is not running perfectly, but with realistic delays and resistance.

The study reveals that while the mechanical pressure of these active bots behaves somewhat like a standard gas, the way they distribute themselves in space completely rewrites the rules of the information engine. The accumulation of bots near the boundaries creates a complex trade-off: a large search area stores more energy if successful, but it is rarely empty; a small search area is almost always empty, but it stores less energy. The machine must constantly balance these two factors. The researchers found that the point where the machine switches from one strategy to the other is a direct result of how the bots crowd the walls. This discovery suggests that to build efficient engines using active matter, one cannot simply rely on the average properties of the material. Instead, the design must account for the specific way these self-driven particles cluster and interact with their container. The work demonstrates that in a world of active matter, the geometry of the boundaries and the statistics of the crowd are just as important as the energy the particles carry, offering a new blueprint for how to harness the chaotic power of self-propelled systems.

Technical Summary: Piston-Like Information Engine II: Boundary-Controlled Optimum in Active Matter

Problem Statement
Information engines convert fluctuations into extractable work through measurement and feedback. While the thermodynamics of such engines operating on thermal (equilibrium) working media are well-characterized—specifically, the mean work per measurement follows a universal form dependent on the probability of an empty detection region—less is known about engines utilizing active matter. Active particles, being inherently out-of-equilibrium due to self-propulsion, exhibit distinct phenomenology, including flocking and phase separation. The central question addressed is how the non-equilibrium nature of active working media alters the operation and optimization of information engines compared to their thermal counterparts. Specifically, the authors investigate whether the universal relation for mean work, $\bar{W} = -k_B T p_1 \ln p_1$ (where $p_1$ is the probability the detection region is empty), holds for active systems and how particle activity influences the optimal operating regime.

Methodology
The authors experimentally realized a macroscopic information engine using a gas of self-propelled "bristle bots" (bbots) confined in a two-dimensional rectangular arena with a movable wall acting as a piston.

Experimental Setup: The engine operates on a cycle where a detection region of width $\Delta x$ near the movable wall is periodically monitored. If the region is empty (probability $p_1$ ), the wall moves inward by $\Delta x$ , compressing the gas without performing direct work during the compression step. If occupied, the wall remains stationary.
Force and Pressure Calibration: To quantify the work stored, the authors measured the pressure exerted by the bbots. They replaced the rigid wall with a flexible nylon barrier to measure cumulative collision forces via strain. A force-strain calibration ( $F = \mu \epsilon^\sigma$ ) was established, allowing the conversion of boundary deflection into force and subsequently into two-dimensional pressure ( $P$ ).
Density Profiling: High-resolution tracking of bbot trajectories was used to construct spatial density profiles. The authors analyzed the probability of finding an empty detection region ( $p_1$ ) and the mean work per measurement ( $\bar{W}$ ) as functions of the detection region size ( $\Delta x$ ) and the area fraction ( $\Phi$ ).
Modeling and Simulation: The experimental data were compared against theoretical models, including an ideal gas model, a "two-density" model (high density near walls, low in bulk), and an "active profile" model (exponential decay from walls). Additionally, a phenomenological model incorporating dry friction (stick-slip dynamics) was simulated to analyze finite-time cycle performance and power output.

Key Contributions and Results

Deviation from Universal Thermal Behavior: The study demonstrates that the mean work per measurement for the active engine, $\bar{W}$ , deviates qualitatively from the universal thermal form $-p_1 \ln p_1$ . While the mechanical equation of state remains linear ( $P \propto \Phi$ ), analogous to an ideal gas but with an active energy scale $\tilde{\alpha}$ replacing $k_B T$ , the feedback statistics are fundamentally altered by activity.
Boundary Accumulation Mechanism: The deviation is attributed to the persistent accumulation of active particles near the boundaries (the "boundary layer"). This accumulation creates a non-uniform density profile where particles spend more time near the walls than in the bulk.
Density-Dependent Regime Switching: A primary finding is a transition in the optimal operating regime as the particle density (area fraction $\Phi$ $Φ$ ) increases.
- Low Density: The maximum work is achieved with large detection regions ( $\Delta x$ ) extending into the bulk, corresponding to a low probability of emptiness ( $p_1 < 1/e$ ).
- High Density: As $\Phi$ increases, the probability of finding an empty region within the dense boundary layer drops sharply. Consequently, the optimal regime switches to smaller detection regions confined within the accumulation layer, operating at a higher probability of emptiness ( $p_1 \ge 1/e$ ).
Robustness in Finite-Time Cycles: The authors investigated a non-ideal cycle involving finite-time expansion against dry friction. The simulations and experimental data confirm that the boundary-controlled switch in optimal operation persists even under realistic constraints (finite cycle time and friction). The trade-off between stored work and the waiting time for a successful measurement remains central, with the optimal detection size shifting similarly to the quasi-static case.

Significance and Claims
The paper claims that the operation of active information engines is governed not just by the equation of state of the working medium, but critically by the spatial statistics of fluctuations selected by the feedback mechanism. The accumulation of active particles near boundaries acts as a "fluctuation selector," where the engine extracts work by selecting rare boundary configurations (empty regions) rather than harvesting bulk active forces directly.

The authors conclude that the boundary-controlled switch between operating regimes is a robust nonequilibrium feature. This suggests a design principle for active information engines: optimizing performance requires tuning boundary geometry and confinement to control the probability $p_1$ and the associated feedback statistics, rather than relying solely on thermodynamic parameters. The findings indicate that the rich behavior of active matter, specifically boundary accumulation, fundamentally reshapes the feedback statistics governing engine operation, distinguishing them from thermal engines.

More like this