A thin dielectric gap sandwiched between two metal half-spaces (a metal-insulator-metal, or MIM, waveguide) supports a bound "gap plasmon" mode with no cutoff width — unlike an ordinary dielectric waveguide, it keeps guiding light however narrow the gap gets. Its fundamental (symmetric) mode satisfies the transcendental dispersion relation:
tanh(κ_d · d/2) = −(ε_d κ_m) / (ε_m κ_d)
κ_d = √(β² − ε_d k₀²), κ_m = √(β² − ε_m k₀²)
k₀ = 2π/λ₀, β = β' + iβ'' (complex propagation constant)
This simulator solves that equation live by complex Newton–Raphson every time a slider moves. The metal's permittivity ε_m(ω) comes from the free-electron Drude model,
ε_m(ω) = 1 − ω_p² / (ω² + iωγ)
with plasma frequency ω_p and collision rate γ fit to gold or silver (a standard teaching simplification — real gold deviates below ~550 nm due to interband d-band transitions, which the free-electron model ignores).
- neff = Re(β)/k₀ — how much the mode's wavelength shrinks inside the gap versus free space; it climbs sharply as d narrows.
- Propagation length Lp = 1/(2·Im(β)) — the distance over which guided power falls by 1/e. Tighter confinement always costs propagation length: the field bars along the waveguide decay visibly faster for a narrow gap and barely decay at all for a wide one.
- The bars extending sideways from the metal plates show the transverse mode profile — roughly uniform across the gap and exponentially evanescent (∝ e−κ_m x) inside each metal, the same evanescent tail that lets the mode beat the diffraction limit.
This confinement/loss trade-off is the central design problem in nanophotonic plasmonic circuitry — routing light through structures far smaller than its own wavelength always trades propagation distance for mode size.