Historically, what has been the most pressing issue in the field of photonic structures?

Mikael Rechtsman (left) and Mordechai Segev (right) in their lab. Credit: TAL SEGEV

Magnetic effects in optical media are of central importance in the physics of photonic structures, for example in transformation optics, such as cloaking and perfect lensing, or non-reciprocal devices, such as optical isolators. One way around this is by using metamaterials, but they exhibit significant shortcomings due to their large absorption losses. This is a crucial limitation of their applicability due to large absorption losses as a result of their resonant nature and the involvement of metallic elements.

How did you come up with the idea of straining photonic structures?

We realized that certain ideas whose origins lie in graphene physics could be applied to photonics. We had already been working with 'photonic graphene', a structure that has honeycomb symmetry and therefore the same photonic dispersion relation as graphene's electronic dispersion. We had previously been able to observe various effects of graphene physics and also some phenomena that are inaccessible in carbon-based graphene, such as graphene solitons, tachyonic dispersion and nonlinear dynamics. The idea that a 'pseudomagnetic field' could emerge from an inhomogeneous strain in graphene started with Kane and Mele in the late 1990s. Subsequently, Guinea, Geim and co-workers showed that if strain was applied in a certain way, the pseudomagnetic field could be made to be constant and therefore the electronic spectrum would show Landau levels. We realized that if a photonic structure could be designed in this spirit, the photonic bands would show Landau levels too. The reason this caught our attention is the Purcell effect. Landau levels are frequencies at which many degenerate states coexist, and the Purcell enhancement is proportional to the density of states at a given frequency. A countless number of photonic applications rely on this type of enhancement, including single-photon sources, nanolasers, optomechanical coolers and optical antennas. We thought that bringing this 'graphene technology' to photonics would provide a whole new approach to the Purcell enhancement problem.

What did you do and what challenges did you face?

We designed a honeycomb photonic lattice — an array of dielectric optical waveguides organized in a honeycomb fashion. We strained the honeycomb in a special way such that the underlying vector potential caused a d.c. pseudomagnetic field. We demonstrated that such a structure exhibits 'photonic Landau levels' in its spatial photonic spectrum. The evidence for the existence of these Landau levels was the experimental observation of localized edge modes residing in the gaps between them. We also engineered a defect waveguide to 'drag' the excited states through the spatial spectrum, thus experimentally proving that the Landau levels were positioned at the centre of the spectrum. It was technologically challenging to engineer a photonic lattice with targeted strains and with defect waveguides that had the desired refractive indices. It was also conceptually difficult to figure out how to probe the existence of the Landau levels. In two-dimensional photonic lattices such as our strained honeycomb lattice, it is difficult to engineer the input beam so as to excite only the eigenmodes one wishes to probe. After much trial and error, we realized that in order to establish the presence of Landau levels, we had to probe the 'armchair'-shaped edge of the lattice. As we explain in our Article, this showed the spreading of light when there were no Landau levels, but confinement when there were Landau levels present.

What is the future outlook?

Sticking to periodic photonic structures, such as Bragg gratings, photonic crystals, coupled resonator arrays, metamaterials and plasmonic arrays, is too constraining. The original goal of photonic crystals, as proposed by Eli Yablonovitch in 1987, is to manipulate the ambient density of photonic states. Our work provides a new mechanism to achieve a high density of states that is fundamentally aperiodic. In the future, perhaps relaxing the constraint of periodicity will allow us to push this much further. For photonics in particular, there are a number of natural research directions. Can we realize strong pseudomagnetic effects in photonic crystal slabs? Can we get a higher number of states at a given frequency using a mechanism other than pseudomagnetism? Is it possible to achieve high emission from quantum dots when they are placed in a Landau-level photonic environment? Can we enhance nonlinear effects in this way, as they are directly proportional to the density of states? Will such a high density of states improve the absorption of thin-film solar cells? Can we think of photonic structures more generally, breaking away from perfect periodicity, and what new applications will that bring? Finally, there is still the long-standing challenge of engineering a non-reciprocal structure that can be used as an optical diode. In our structure, despite the presence of a pseudomagnetic field, time-reversal symmetry is preserved and propagation is thus fully reciprocal. This issue is conceptual; breaking the spatial symmetry, as we did in our strained photonic honeycomb, is insufficient for non-reciprocity. Rather, one would need to add temporal modulation or to make the system nonlinear. An important future challenge is creating non-reciprocal structures, perhaps by pseudomagnetic effects that are nonlinear or explicitly depend on time.

Interview by Rachel Won

Mordechai Segev and co-workers have an Article on strain-induced pseudomagnetic fields in photonic structures on page 153 of this issue.