22/12/2025
Alhun Aydın
Sabancı University Faculty of Engineering and Natural Sciences
Imagine a demon whose hobby is to precisely observe the motion of molecules in a container. The container is divided into two parts, with a tiny door in the middle. Whenever a fast molecule approaches from the left, the demon opens the door and lets it pass to the right. Whenever a slow molecule approaches from the right, it opens the door and lets it pass to the left. Over time, the right side becomes hotter (more fast molecules) and the left side colder (more slow ones). A temperature difference emerges purely through observation and clever sorting. This famous thought experiment is called Maxwell’s demon. And once you have a temperature difference, you can run a thermal machine. However, creating a temperature difference requires work. Think of an air conditioner: it cools the room and dumps heat outside, but the cost is electricity. Yet in Maxwell’s setup, it seems as if the demon creates a temperature difference and a directed heat flow without expending any work.
Leo Szilard took the idea further by turning it into an explicit heat engine. Picture a box containing a single molecule. Insert a partition in the middle so the molecule is either on the left or right, exerting pressure on the partition. If you know which side the molecule is on, you can attach a piston and a tiny weight and let the molecule’s expansion lift the weight as it reoccupies the entire box. The molecule’s location is a binary fact (left or right) equivalent to one bit of information. The knowledge is converted into work without any cost, seeming at odds with thermodynamics.
While these thought experiments seem philosophical gymnastics, they actually reveal profound insights on how work, heat and energy behave in physical systems. Later, Rolf Landauer figured out what both stories omitted: memory. Once you learn something, you must store it (whether in the demon’s mind or in an information-processing device). And if you want a cyclic engine, you eventually have to erase that memory to begin a new cycle. Landauer’s key insight was that erasing information is an irreversible operation that must produce heat. Erasing a bit means taking something that could be in state 0 or 1 and forcing it into a reference state (say, 0) regardless of its prior value. You compress two logical possibilities into one, and that loss of information carries a fundamental energy cost, called the Landauer limit. You can’t delete information for free.
Every time your phone, laptop, or a supercomputer processes data, it is ultimately overwriting and resetting bits. Modern chips operate far above the Landauer limit. They waste many orders of magnitude more energy per operation due to resistive losses and imperfect architectures. Still, as electronics are miniaturized and computation becomes more densely packed, Landauer’s principle sets a hard lower bound: no computer can use less energy than the thermodynamic cost of erasing the bits it discards.
Up to this point, the story was classical. But things get even richer in the quantum physics, where information can be encoded not only as 0 or 1 but also in superpositions of 0 and 1. It can also be stored in the correlations between the system and its environment, e.g. as entanglement. In quantum thermodynamics, resources such as quantum correlations and coherence can be traded against heat and work in ways with no classical analogue, for example, enabling heat to be driven from colder to hotter regions, which is impossible using purely thermal resources alone. Moreover, quantum confinement and spectral engineering can produce thermodynamic behaviors that seem counterintuitive from a classical perspective, such as spontaneous transitions into lower-entropy configurations under conditions where classical intuition would expect the opposite. These features open the door to quantum heat engines with enhanced capabilities.
These insights point toward conceptually deeper and practically promising routes for improving energy efficiency. At the smallest scales, energy loss is often not constrained by how clever our circuitry is, but by how much unwanted heat is generated when we manipulate information and move energy. Quantum systems provide new degrees of control: we can engineer discrete energy levels, modulate transitions between them, and harness entanglement and coherence effects that have no classical counterpart. This enables computation, sensing, and energy conversion to be designed with a greater precision. The long-term vision includes ultra-low-power logic, more efficient nanoscale refrigerators, and quantum heat engines whose performance is governed by the structure and dynamics of their quantum states.
As always, theoretical physics continues to seed and guide future technologies: when you move bits, you move heat; when you control information, you control energy.
