Van der Waals materials have extraordinary optical properties but have been difficult to shape into the low-loss structures photonics requires, because patterning damages the very thin crystals involved. A collaboration led from Peking University with Aalto University developed a focused-ion-beam nanofabrication method assisted by aluminium passivation, supporting high-resolution patterning across many layered materials and their heterostructures while limiting processing damage.
The method proved general. The team fabricated microdisks, photonic crystals and metasurfaces from hexagonal boron nitride, molybdenum disulfide, gallium selenide and niobium oxydichloride. Microdisk resonators reached intrinsic quality factors above one million, which the authors report as exceeding previous van der Waals resonant systems by about three orders of magnitude. Quality factor rose with wavelength in the manner theory predicts for scattering from surface roughness, evidence that sidewall and surface imperfections had genuinely been reduced.
High quality factors combined with the strong nonlinear response of these materials enabled several demonstrations. Continuous-wave second-harmonic generation in gallium selenide microdisks achieved a normalised conversion efficiency of about 30 per cent per watt, reported as four orders of magnitude above earlier van der Waals systems, and the signal could be thermally tuned across a full free spectral range. The team also showed continuous-wave sum-frequency generation and optical parametric amplification in boron nitride microcavities.
The study was published in Nature Materials. Its significance is as enabling infrastructure rather than a single device: once layered materials can be patterned without destroying their optical quality, the large library of van der Waals crystals becomes available for reconfigurable photonics, nonlinear signal processing, quantum light sources and ultrasensitive sensing. Each of those applications had previously been demonstrated in principle with exfoliated flakes while remaining impractical to build into a device, and the bottleneck was always the patterning step rather than the physics.