Thermal machines in the quantum world

Thermodynamics, developed in its basic structure in the 19th century, is based on a coarse-grained viewpoint, abstracting from details, to give a description of a plethora of physical systems. This way she advanced to the bases for many key technologies that made our modern society possible. The heat engine converts heat (disordered energy) into mechanical energy (work). Examples are the steam engine, steam turbine and all internal combustion engines. A power heat engine transports thermal energy from a lower temperature level to a higher one using mechanical energy. Examples are a heat pump or a refrigerator. However, quite new questions - going beyond the traditional picture of thermodynamics - arise when looking for a description of small systems for which stochastic fluctuations are relevant, or even of machines for which quantum effects play a central role. In recent years, the potential and limitations of the resulting quantum thermodynamics and the resulting thermodynamic transformations for entanglement, quantum fluctuations, quantum information exchange, and coherences have begun to be explored. Despite some important progress, many conceptual issues are still unclear. Perhaps even more urgently, there are no experimental realizations that convincingly demonstrate the potential of the intended advantage of quantum effects. This planned Research Unit faces up to these challenges. It brings together leading researchers in quantum thermodynamics in experiments - on trapped ions, ultracold atoms or NV centers - and theory to explore new approaches. On one hand, quite significant developments of experimental platforms are needed to actually detect and exploit true quantum effects in work and power extraction. On the other hand, many conceptual questions are still wide open: in what sense can realistic quantum machines be more powerful than classical machines? How do small (quantum) systems thermalize? Can the traditional separation of system and bath always be carried out? What is the role of quantum correlations and entanglement? Is quantum error correction understandable within this framework? The proposed research not only promises profound insights into the fundamentals of thermodynamic processes, in a world ultimately determined by quantum processes, but also technological implications, such as new cooling techniques, quantum heat engines and quantum refrigerators. Ultimately, this Research Unit promises a breakthrough on the question of whether quantum thermodynamics has the potential to improve quantum machines and provides a framework for discussion and exchange.

Engines, refrigerators and power plants obey the laws of heat and work. Do these laws still hold when a machine becomes so small that quantum physics takes over? In this project we built a laboratory system in which that question can be settled by measurement rather than by theory alone. The answer: the fundamental laws survive, but the microscopic reasons behind them are surprising. Our experiments use a few thousand rubidium atoms, cooled to below a millionth of a degree above absolute zero and held in a microchip by magnetic fields. At such temperatures the atoms lose their separate identities and behave as a single collective quantum wave a BEC. It is small enough to be fully quantum, yet large enough to have a temperature, an energy content and internal flows. It therefore serves as the working substance of a thermal machine made from a quantum field that we can prepare, manipulate and photograph with great precision. Our first task was to find out what carries energy through this gas. The textbook picture of sound waves rippling through the cloud turned out to be only roughly correct. The true carriers behave like particles that refuse to share the same state of motion, and this reluctance shields the gas from disorder at far higher temperatures than expected. This told us in which regime the cloud behaves as a clean quantum field, and guided all later work. We then turned to information. Using a method that reconstructs the complete pattern of correlations in the cloud from series of images, we showed that two neighbouring regions share information only across the surface between them, not throughout their volume - a rule long predicted for quantum fields but never measured. We then carried out the first quantitative test of Landauer's principle in a genuine many-body quantum system. The principle states that erasing information always costs a minimum amount of heat - the ultimate limit on the energy efficiency of any computer. Our measurements confirm it precisely and show where the cost comes from: information shared between the system and its surroundings, carried across the boundary by pairs of excitations. Finally, we measured for the first time how such a quantum gas conducts particles and energy, and found that it flows without friction even at finite temperature. To extract these tiny currents from faint images we trained neural networks that are forced to obey the physical conservation laws. These results also matter for quantum technology: waste heat and loss of information are the main obstacles facing quantum computers and quantum sensors. Knowing the limits nature sets, and being able to measure them in a real system, helps in designing devices that come close to them.