Many of us have sat in a boat that is being rocked by the waves from another passing vessel. This situation is an example of a mediated interaction: even though the two boats do not touch, they interact through the exchange of waves in the water between them. Mediated interactions have a central role in modern physics. For instance, the fundamental interactions between elementary particles are mediated by particles such as photons1. And the attraction between electrons that leads to superconductivity in metals arises from the exchange of vibrations in the surrounding crystal lattice2. In a paper in Nature, DeSalvo et al.3 report the first observation of mediated interactions in an ultracold mixture of lithium and caesium atoms (Fig. 1a).
Consider a mixture of two kinds of atom, A and B, and suppose, for clarity, that there are many more B atoms than there are A atoms. Let us first assume that the A and B atoms attract each other. Because of this attraction, there would be a high density of B atoms around a given A atom (Fig. 1b). Another A atom would be attracted to these B atoms, and the net effect would be an attractive interaction between the A atoms that is mediated by the B atoms.
Had we instead assumed that the two kinds of atom repel each other, there would be a low density of B atoms around a given A atom. Another A atom would be attracted to this absence of B atoms, and the end result would again be an attractive mediated interaction between the A atoms. This simple argument explains why mediated interactions are inherently attractive at short distances. At longer distances, the story is more complicated, because intricate quantum effects can make a mediated interaction repulsive.
DeSalvo and colleagues used a mixture of bosonic caesium-133 atoms and fermionic lithium-6 atoms — bosons have integer spin angular momentum, whereas fermions have half-integer spin. The atoms were confined near the minimum of a harmonic trap, which is a type of potential-energy surface (often simply called a potential) that has a parabolic form. The authors cooled the mixture to extremely low temperatures, so that the caesium atoms formed an exotic state of matter known as a Bose–Einstein condensate (BEC), whereas the lithium atoms instead formed a quantum gas called a Fermi gas.
The direct interaction between the caesium atoms was weakly repulsive, whereas the direct interaction between the caesium and lithium atoms could be tuned to be either attractive or repulsive. There was no direct interaction between the lithium atoms because of a fundamental principle of quantum mechanics that states that two fermions cannot occupy the same space. This principle also meant that the lithium gas had a much larger volume than the BEC of caesium atoms, which was located at the centre of the gas. DeSalvo et al. demonstrated that the presence of the lithium gas had two effects on the caesium BEC.
First, when the BEC was set to oscillate back and forth in the harmonic trap, the oscillation frequency changed when the lithium gas was present. The reason for this change is that the energy of the caesium atoms was shifted by the direct interaction between the caesium and lithium atoms. This energy shift was spatially dependent because the density of the lithium gas decreased as a function of distance from the centre of the trap. As a result, the BEC was subjected to an additional trapping potential from the lithium gas that altered the oscillation frequency.
The second, and arguably more intriguing, effect was that the lithium gas made the BEC decrease in size. By carefully analysing their experimental data, DeSalvo and colleagues concluded that this effect was caused by an attractive interaction between the caesium atoms that was mediated by the lithium gas. The results were in fair agreement with theoretical predictions based on a type of interaction called a Ruderman–Kittel–Kasuya–Yosida (RKKY) interaction, which has been shown4–6 to exist as an interaction mediated by a Fermi gas precisely like the lithium gas.
RKKY interactions give rise to a great variety of magnetic phenomena in rare-earth elements — the collective name for 17 chemically similar metallic elements — in which electrons play the part of the Fermi gas. In addition, electronic devices, such as hard drives, that exploit a phenomenon known as the giant-magnetoresistance effect contain magnetic layers that are thought to be coupled by RKKY-like interactions7.
Although the observed mediated interaction between the caesium atoms was quite weak, DeSalvo et al. used a clever trick to show that it could still have spectacular effects. The authors tuned the direct interaction between the caesium atoms to be only very weakly repulsive so that the attractive mediated interaction was comparatively stronger. The combination of the direct and mediated interactions then gave rise to a net attraction between the caesium atoms. Because, unlike for fermions, nothing prevents bosons from occupying the same space, the BEC collapsed. The authors observed this collapse through the formation of small, soliton-like blobs of caesium atoms — solitons are spatially localized states that are characteristic of BECs.
Given the unrivalled versatility of atomic gases, DeSalvo and colleagues’ results open up the possibility of exploring mediated interactions in detail and probing interactions that have never been seen before. So far, only a weak mediated interaction has been observed, and it would be useful to study stronger interactions. Such interactions should greatly affect the energy spectrum of excitations of the BEC8 and give rise to a range of exotic phases of matter in mixtures of bosons and fermions9,10.
Future work should also explore the reciprocal case of an interaction between fermions that is mediated by a BEC. This mediated interaction is, in general, much stronger than the fermion-mediated interaction because of the large compressibility of a BEC compared with a Fermi gas, and it could also lead to previously unobserved phases of matter11–13. It will be exciting to see what discoveries follow the breakthrough results of DeSalvo and colleagues.
Nature 568, 37-38 (2019)