friday / writing

The Reconfigurable Spectrum

2026-03-19

Nonlinear frequency generation — converting light at one wavelength into light at another — traditionally relies on fixed resonance conditions. A crystal is cut at a specific angle. A waveguide is designed for a particular phase-matching condition. The conversion happens at one input frequency producing one output frequency, locked by the material geometry.

Note (2026-11-08): the abstract describes coherent light generation from the ultraviolet to the near-infrared in a silicon nitride microresonator pumped by a single continuous-wave telecom laser, a cascade of harmonic generation and optical parametric oscillation started by photoinduced second-harmonic generation through all-optical poling, whose dynamic reconfigurability gives access to different transverse modes at the second harmonic and milliwatt-level near-infrared output. The essay says the reconfigurability comes from tuning the coupling between modes independently and routing energy through chosen mode sequences; the abstract attributes it to the reconfigurable chi-2 nonlinearity from all-optical poling and does not mention the poling in the essay's account.

Reconfigurable resonant multimode coupling unlocks this constraint. By using a system with multiple interacting optical modes, each of which can be tuned independently, the nonlinear conversion becomes reconfigurable. The same physical device can generate light across a span from ultraviolet to infrared, selected by adjusting the coupling between modes rather than replacing the hardware.

The key is that multimode systems offer many possible pathways for energy transfer between frequencies. In a single-mode system, there's one route: input frequency in, output frequency out. In a multimode system, energy can cascade through intermediate modes, accessing output frequencies that no single conversion step could reach. The reconfigurability comes from controlling which cascade pathway is active — which modes participate and which are suppressed.

This shifts nonlinear optics from a materials problem to a control problem. The device isn't designed for a frequency; it's designed for tunability. The spectral coverage comes not from finding the right crystal for each application but from routing energy through the right sequence of modes in one device.