Realisation of a demon

A new paper published in this week's Physical Review Letters reports on a photonic realisation of the 'Maxwell's demon' thought experiment by Oxford physicists. Dr Oscar Dahlsten and Mihai Dorian Vidrighin, from the Clarendon Laboratory in Oxford's Department of Physics, spoke to Science Blog about their work.

What is Maxwell's demon?

Maxwell's demon is a famous thought (imaginary) experiment by James Clerk Maxwell that has been around since 1867. In the original version, the demon is an imaginary creature controlling a small door in a wall that separates two boxes, each box containing the same gas at the same pressure and temperature. The gas is made up of molecules moving around and bouncing off the walls of the boxes. In the simplest version of the experiment, when the demon sees a molecule approaching from the right box, it opens the door. When a molecule approaches from the left, the demon closes the door. After a while, there will be more molecules on the left of the wall and therefore a pressure difference between the boxes. The demon can use this pressure difference, for example by letting it push the wall towards the right and using a rope and pulley to lift a weight.

The type of energy used to lift a weight is 'work'. Work entails directionality, which makes it different from the energy stored in an equilibrated gas, called 'heat'. Heat belongs to the completely disordered movement of gas molecules. Thermodynamics is the theory that deals with the transformation of energy, and its second law states that work cannot be extracted from an equilibrium system (such as the gas that the demon works with). A simpler way to express this is: directionality does not appear out of disorder. The demon, however, seems to achieve precisely this.

This apparent paradox has fascinated physicists, and the imaginary demon has driven theoretical understanding, like a real experiment would. In particular, it has helped to reveal the close connection between thermodynamics and information theory. The explanation for why work can be extracted from a disordered gas is that the demon itself is changed during the process. Its memory fills up with the information of the events it has observed. The second law of thermodynamics is preserved if we recognise that resetting the demon’s memory must cost work, at least as much as he can extract by his actions.

Maxwell's demon has found a place in modern physics, and it has been called upon in many instances, even to argue that some basic characteristics of modern theories such as quantum mechanics and general relativity can be derived from the laws of thermodynamics.

How does your new paper relate to the original thought experiment?

In this paper, which is co-authored by Marco Barbieri, M S Kim, Vlatko Vedral and Ian Walmsley, we have realised a Maxwell's demon using photonics. Our setup is analogous to the original thought experiment, using light instead of gas. The demon is looking at two light fields with random amounts of energy and uses single-photon detection to obtain extra knowledge about the light’s energy and to eventually extract work. We find it particularly satisfying that the final effect is easy to see – we can use the setup to charge a battery.

We also derive theoretical results that take into account the realities of the experiment (such as having detectors at a lower temperature than the light field), allowing us to show a clear relation between the information acquired by the demon and the work produced in our setup.

What are the potential implications of these findings?

There has been much theory recently on the link between information and thermodynamics but a dearth of experiments to anchor the theory in experiment, and we think our work can influence which theoretical models are used and lead to new experiments. An application of the general research direction will be in making energy harvesting technology more efficient.

Our work shows that photonics is a natural system for investigating the relation between work and information, which is fundamentally interesting. For instance, it’s worth observing how the particle nature of light affects the way in which our setup operates. Also, we are already thinking of ways in which features such as entanglement can be introduced in future experiments based on this one, as our interests gravitate around quantum information.