Author: Herwig Ott

**A quantum engine in the BEC–BCS crossover (B5)**

Jennifer Koch, Keerthy Menon, Eloisa Cuestas, Sian Barbosa, Eric Lutz, Thomás Fogarty, Thomas Busch, Artur Widera

Heat engines are at the heart of thermodynamics. They can be used to perform work or generate cold. What happens if heat engines are operated at ultralow temperatures, where quantum physics is relevant?

The work demonstrates a new quantum mechanical cyclic process that takes place near absolute zero temperature. The medium is an ultracold quantum gas of fermionic lithium atoms in two different spin states. The special feature of the system is that the interaction between the lithium atoms can be changed using a so-called Feshbach resonance. This means that the two spin states can either form bosonic molecules, for which the Bose-Einstein statistics apply, or a strongly interacting Fermi gas, for which the Fermi-Dirac statistics are decisive. The cyclic process now consists of the familiar elements of compression and expansion, but in between, it is not the temperature of the medium that is changed, but its quantum statistics. This changes the energy ratios in the medium (see figure). For example, during fermionization (transition from Bose-Einstein statistics to Fermi-Dirac statistics), energy is added to the system, as the Fermi energy is significantly higher than the ground state energy of the molecular Bose-Einstein condensate.

This is the first time that it has been shown that new types of cyclic processes are possible in the quantum world in which quantum statistics can be used as a thermodynamic resource.

Figure: Cycle of the heat engine. The atomic cloud is compressed and expanded via changing the confinement of the harmonic trapping potential. After compression, the quantum statistics is changed from bosonic to fermionic by changing the interaction parameter with help of a Feshbach resonance. After the subsequent expansion, the interaction parameter is set back to its original value and the statistics change from fermionic to bosonic.

**Observation of a Topological Edge State Stabilized by Dissipation (B1, C1, C4)**

Helene Wetter, Michael Fleischhauer, Stefan Linden, and Julian Schmitt

Phys. Rev. Lett. 131, 083801 (2023)

Topology is a non-local property of a system. It is known from band structures as they occur in solids, arrays of waveguides, but also in mechanical and electronic periodic systems. Whether the band structure of a system is topological or not depends on the couplings and the geometry of the unit cell. When a system is topological, it exhibits remarkable properties and so-called topological phases are formed, which are characterized by invariants. The topological phases are not influenced by local changes in the system, which makes them robust. If two topological phases now touch and form an interface, another concept becomes important: topological edge states. They are protected as well. Particles that are in such a state move along the edge and are not entering any of the two phases.

While in most systems the topology is generated by alternating coupling parameters between neighboring lattice sites, the current work uses alternating losses to achieve the same effect. With the help of coupled waveguides (using so-called surface plasmon polaritons) and tailored losses (chromium strips under the waveguides), the experiment shows topological protection in a dissipative system. This can be seen in figure (a), where the polariton propagation after excitation at the edge or in the bulk of such a waveguide array is shown. In (I) and (II), where no topological order is present, the quasiparticles additionally move either in the x-direction across the lattice towards the bulk (I) or oscillate near the edge (II). In the topologically protected variant (III), however, the signal propagates exactly along the edge and sticks to it. Figure (b) shows the corresponding theoretical simulations. The work has shown that losses in optical waveguides are capable of generating topology in a system that would have no topology at all without losses.

Figure: (a) Propagation of polaritons in a waveguide array (experiment). For a suitable choice of periodically alternating losses (III), the light propagates along the edge and thus remains localized in the x direction. (b) Theoretical simulation of the polariton propagation in waveguide arrays with corresponding loss patterns.

**Mean-field approach to Rydberg facilitation in a gas of atoms at high and low temperatures (B2)**

Daniel Brady and Michael Fleischhauer

Phys. Rev. A 108, 052812 (2023)

Rydberg atoms have special properties that make them interesting as model systems for various fields of physics. In particular, the long-range interaction between them holds great potential. The so-called Rydberg facilitation is one of the resulting effects. It describes the consecutive excitation of a whole series of atoms into a Rydberg state when they are at the right distance from each other and are driven by a laser field. This excitation is counteracted by spontaneous decay. If we now assume a random arrangement of atoms in an atomic cloud (see figure), the question arises as to whether a large excitation cascade can propagate through the entire cloud. This is called the active phase. It is characterized by the fact that there are always enough neighboring atoms at the right distance and that the excitation probability is high enough for the next neighbor to be excited within the lifetime of the Rydberg atom. If this is not the case, only very small cascades occur and it is referred to as the absorbing phase. The picture is now made more complicated by the fact that the atoms are allowed to move and can decay into "inert" states not participating in the excitation dynamics. The spatial arrangement of the atomic positions is therefore constantly changing. Interestingly, such models have a very close analogy to infection models used in medicine. Forest fires can also be modeled using such approaches.

In order to describe such a system theoretically, a new mean-field approach has been developed. In contrast to existing methods, it takes into account the microscopic processes and conditions during excitation and is therefore much more meaningful. A comparison with a Monte Carlo simulation shows impressive agreement. With these results, both the concrete Rydberg systems can now be better understood and the properties of infection models can be investigated in much greater detail and at a much deeper level of understanding.

Left: three atoms within a cloud have the right distance rf to each other in order to be excited to a Rydberg state. Right: Time evolution of the ground state atom density (blue) and the Rydberg atom density (red) of the whole cloud, resembling the three phases of epidemic dynamics. After an initial growth the Rydberg atom density enters the active phase and is constant over time. When the cloud is getting too dilute by decay of Rydberg atoms into "inert" states ("immunization"), the active phase can no longer prevail.