Stopped 200 nm SiN Fourier atom-hole campaign

This page freezes the 200 nm thickness experiment at the checkpoint selected for replacement by the finite SiO2-clad stack. It remains an exploratory, continuous-density result; it did not run through binarization and is not presented as a finished device.

Executed configuration and outcome

Quantity	Value
Material stack	200 nm Si3N4 slab in infinite SiO2
Atom	z-oriented, 780.24 nm
Immutable opening	1.0 µm-diameter through-slab air hole
Trainable region	9.6 µm-diameter extruded 2-D mask
Grid / symmetry	25 nm / x-y-z (8× reduced solve)
Active controls	40,000 real Fourier coefficients
Accepted updates	69
Initial / final / best LDOS	0.517264 / 0.595742 / 0.595742
LDOS gain	15.17%
Final grayness	0.6140
Final field / energy fitted Q	34.03 / 65.17, rejected

The largest apparent conservative Q in three pulsed audits was 174.39, but it failed the field/energy residual and window-stability gates. The optimizer therefore never entered shifted-pole mode. This is the intended fail-closed behavior, not evidence of a Q=174 cavity.

Geometry at the stopping point

The initial spectral seed, final continuous mask, and literal 200 nm material cross-section at update 69.

Optimization and pole-audit history

LDOS improved monotonically over the short run. Untrusted Q fits are shown as diagnostics and never used to recenter the objective.

Spatial spectrum and electric field

The complete xz panel includes the PML entrances. A stronger validation doubled the physical z interior from 1.8 to 3.6 µm at update 67; LDOS changed only -0.000160% (0.59500849 to 0.59500754). The visible upward and downward field is therefore outgoing radiation rather than a measurable finite-z artifact at this checkpoint.

The resumable artifact remains at benchmarks/artifacts/invdes_fourier_atom_hole_ldos_sin200_isotropic_v1.npz.