Strong backscattering in valley Hall photonic crystal waveguides

Just published in Nature Photonics this week: Observation of strong backscattering in valley-Hall photonic topological interface modes

Researchers from the Technical University of Denmark report experiments with state-of-the-art photonic crystal waveguides, comparing the performance of valley Hall waveguides against conventional line defect waveguides. In the slow light regime, of most interest for enhancing light-matter interactions or nonlinear optical effects, the conventional line defect waveguides have substantially lower propagation losses (0.5 dB / cm) compared to the topological valley Hall waveguides (15-200 dB / cm). 

Interestingly, while earlier theoretical estimates (e.g. in Phys. Rev. Research 2, 043109 (2020) and discussed in Section 5 of our review article) focused on potential out-of-plane scattering losses, which might in principle be mitigated by embedding the 2D photonic crystal in a suitable cladding material, the present experiments establish in-plane disorder-induced backscattering leading to Anderson localization as the dominant loss mechanism. There is no obvious way to inhibit this kind of backscattering except for improving fabrication precision.

The stronger losses be intuitively understood intuitively by comparing scanning electron microscope images of the conventional and valley Hall waveguides:

Scattering arises because the boundaries between the air holes and silicon are not perfectly smooth, but have nanometer-scale roughness. The conventional waveguide design minimizes potential roughness-induced by scattering by localizing the guided mode to the hole-free region. In contrast, the valley Hall waveguide has additional triangular air holes leading to a larger area of rough sidewalls.

Experts working on topological photonics will probably not be that surprised by this result; several pessimistic theoretical estimates of valley Hall waveguide performance have been published in the past few years. Unfortunately, high profile journals generally prefer positivity and bold claims over more subdued but realistic assessments. It is great to see important negative results such as this published in a high profile journal. Hopefully this will stop numerous authors from repeating uncritically misleading claims that topological photonic crystals offer robustness against all forms of disorder.