When light passes through matter, it slows down. Light can even be brought to a standstill when it travels through carefully designed matter. One way in which this occurs is when the velocity of individual particles of light (photons) in a material is zero. Another, more intriguing, way is when photons, which normally pass through each other unimpeded, are made to repel each other. If the repulsion is strong enough, the photons are unable to move, and the light is frozen in place. Writing in Nature, Ma et al.1 report that such a phase of matter, known as a Mott insulator, can be produced by exploiting energy loss in a system in which photons move through an array of superconducting circuits.
A wide variety of experiments aim to engineer large quantum systems for use in computation and high-precision sensing, and to design materials that have unprecedented properties. These experiments often treat dissipation — the loss of energy or, more generally, information about a system — as anathema to producing large quantum systems. The reason is that tiny perturbations, including dissipation, often destroy quantum effects in systems of more than a few particles. Great care is therefore taken to minimize dissipation.
However, when dissipation is carefully engineered, it can also be a resource, and its utility for realizing exotic quantum matter is beginning to be harnessed2–8. One common way to use dissipation to produce unusual quantum states is to lower the temperature of a material. This can be accomplished by immersing the material in a coolant to extract energy. Just as cars on a motorway can go from a high-energy, smoothly flowing state to a jammed state by dissipating energy into heat and sound, quantum systems can go from a relatively hot initial state to a cold, jammed state by dissipating energy into photons.
Ma and colleagues used a more sophisticated method to dissipate energy from superconducting circuits, which are similar to ordinary circuits, but have some elements replaced by superconducting wires. The authors used a chain of eight superconducting circuits called transmons. Each transmon can be thought of as a site that photons can occupy. If two photons occupy the same site, they interact with each other at a substantial energy cost.
The authors implemented two different schemes to produce a Mott insulator, a simplified picture of which is a state that has exactly one photon per site. Here, I will describe only the simpler scheme (although the second one was observed to yield a higher-quality final state). To understand this scheme, consider a single site, and how this site might be driven to be occupied by a single photon. It is easy to add a photon to an empty site, without dissipation, by applying a microwave-frequency electric field. However, such a process is equally likely to remove a photon from an already occupied site.
What is needed is an irreversible process that adds a photon to an empty site. Such a process can be engineered by combining non-dissipative processes with a ‘reservoir’ that rapidly loses photons. In the authors’ scheme, an applied microwave field causes photons to be added in pairs to a site (or subtracted in pairs, in the unavoidable reverse process). When the site is occupied by a pair of photons, it has the same energy as the reservoir and one photon can move to the reservoir and be lost, resulting in the site being left with a single photon (Fig. 1a). A single photon on the site is not energetically matched with the reservoir, and therefore remains on the site (Fig. 1b).
This idea extends to more than one site. In Ma and colleagues’ scheme, the chain of sites in which the Mott insulator is to be prepared is attached at one end to a reservoir (Fig. 1c). If any site in the chain is doubly occupied, one of the photons from this site will wander through the chain until it reaches the end site. This photon will then move to the reservoir and be lost, ultimately resulting in a Mott-insulator state (Fig. 1d).
Although this simple picture makes it look as if the sites could be prepared independently to have exactly one photon, the actual Mott-insulator state is more complicated, and has quantum fluctuations that excite the system to include empty and doubly occupied sites. The authors’ scheme prepares the system to include just the right excitations to be in the true Mott-insulator state. This state is an example of a strongly correlated phase of matter, which has been studied for decades in condensed-matter and ultracold-matter physics owing to its importance in quantum materials.
The strongly correlated state prepared in the current experiment is relatively simple, but the superconducting-circuit platform is flexible, and could be used to realize systems that have different geometries and different couplings between sites. Ma and colleagues’ scheme is predicted to be able to prepare any gapped phase of matter — defined as having a non-zero energy cost to add or remove a particle — that occurs in any of these systems. One type of gapped phase that will surely be the target of future work is ‘topological’ matter. Fundamental open questions about the behaviour of topological matter can be explored in superconducting-circuit systems using dissipative preparation schemes9.
At least two further advances are needed if researchers are to use the authors’ scheme to prepare complex quantum states. First, the technique needs to be extended to larger quantum systems. Second, the quality of state preparation needs to be increased. Currently, there is a relatively small (roughly 10%) density of defects — sites that have either zero or two photons. Nevertheless, this density is at least ten times larger than that of comparable experiments that use, for example, ultracold atoms10. Reducing the defect density is challenging, but it seems to be an engineering issue rather than a fundamental one.
The ability to engineer quantum states promises to revolutionize areas ranging from materials science to information processing. Because quantum states are often as fragile as they are useful, robust state preparation will be essential to realizing their potential. Ma and colleagues’ technique engineers dissipation to drive the system to the desired state and is therefore intrinsically robust to perturbations and noise. The robustness and generality of this scheme will ensure that, as it is refined, it will find a home in the quantum mechanic’s toolbox.
Nature 566, 45-46 (2019)