SOI Nanophotonics

AI Photonic Chip Inverse Designer

Click & Drag to perturb silicon permittivity voxels Mode: Active FDTD Wave Sim
Voxel Brush:
Gen 100

Real-Time Diagnostics

AI Pareto Optimum
Insertion Loss (1310nm) 0.42 dB Transmission: 90.8%
Insertion Loss (1550nm) 0.58 dB Transmission: 87.5%
Crosstalk Suppression -26.4 dB Isolation Port 1 vs 2
Wave Routing Fidelity 97.8% Phase & Mode Overlap
Topology Benchmark 43.9x Area Reduction
Human Classical MZI 150.0 µm length (450 µm²)
AI Inverse Topology 3.2 µm × 3.2 µm (10.24 µm²)
Grid Voxel Resolution 64 × 64 (50 nm pixel pitch)

Mechanism: Conventional photonic design enforces smooth rectilinear curves to avoid scattering. AI inverse gradient algorithms intentionally exploit sub-wavelength multiple scattering to engineer constructive interference within micrometers.

Sub-Wavelength Multiple Scattering in Inverse Nanophotonics

Read the explanation

Traditional integrated optics requires broad hundred-micrometer bends. In inverse photonic design, an optimized sixty-four by sixty-four dielectric matrix deliberately scatters light across sub-wavelength voxels. High-permittivity silicon voxels reduce phase velocity by the square root of epsilon. Lower-index silica lets wavefronts surge forward rapidly, tailoring interference phases directly into output channels. Manual voxel painting breaks this calibrated interference pattern, scattering optical energy into loss channels. Clicking Reset Optimum instantly restores the AI Pareto matrix.

Super generates helpful tools and automates fact-checking across the internet proactively. If you enjoyed this tool, build your own with Super and share it with a friend.