What Produces a Photonic Band Gap
A photonic crystal is a dielectric medium with a spatially periodic modulation of its refractive index — typically alternating layers, rods, holes, or spheres on a length scale comparable to optical wavelengths. When electromagnetic waves propagate through this structure, Bragg scattering at each dielectric interface creates interference. At specific frequency ranges, destructive interference prevents wave propagation in some or all directions: these forbidden frequency ranges are the photonic band gaps.
The analogy with electronic structure is precise. Just as electrons in a crystal lattice experience a periodic potential that opens gaps in the electronic dispersion relation at Brillouin zone boundaries, photons in a periodic dielectric experience a periodic "optical potential" (the spatially varying ε) that opens gaps in the photonic dispersion relation.
Maxwell's Equations in a Periodic Medium
The photonic band structure emerges directly from Maxwell's equations with a spatially periodic dielectric function ε(r) = ε(r + R) for any lattice vector R. By Bloch's theorem — the electromagnetic analogue — solutions take the form of Bloch modes indexed by a crystal momentum k within the first Brillouin zone. The master equation governing the magnetic field H is an eigenvalue problem entirely analogous to the Schrödinger equation.
Master Eigenvalue Equation
The Hermitian nature of Θ̂ guarantees real eigenfrequencies (no gain or absorption) and orthogonal eigenmodes. The gap arises because two degenerate plane waves at the zone boundary (k = ±π/a) are split by the Fourier component of ε at the corresponding reciprocal lattice vector: one mode concentrates its energy density in high-ε regions (lower frequency, "dielectric band"), the other in low-ε regions (higher frequency, "air band").
Key Parameters
| Parameter | Symbol | Role | Effect on gap | Typical range |
|---|---|---|---|---|
| Dielectric contrast | Δε = ε₂ − ε₁ | Fourier component magnitude | Gap width ∝ Δε/ε̄ | Δε = 1–100 |
| Fill fraction | f = V_high / V_cell | Sets ε̄ and Fourier components | Optimum f ≈ 0.2–0.4 | 0 < f < 1 |
| Lattice constant | a | Sets midgap frequency | ω_gap = πc / (n_eff · a) | 100 nm – 1 µm |
| Lattice symmetry | space group | Determines BZ geometry | Higher symmetry → wider gap | sq, tri, fcc, dia |
| Gap-midgap ratio | Δω/ω₀ | Standard gap figure of merit | Maximized at optimal f | 0% – 30% |
Photonic Band Diagram — Live Render
The canvas below draws a schematic photonic band structure for a 2D triangular lattice of high-index rods in air, plotted along the high-symmetry k-path Γ → M → K → Γ. The band gap region is highlighted in orange. Bands are rendered with an animated draw-on effect.
Horizontal axis: k-path through the first Brillouin zone. Vertical axis: normalized frequency ωa/2πc. The orange shaded band gap is the range of frequencies with no propagating modes for this polarization.
Photonic Crystal Dimensionality
The 1D photonic crystal — the Bragg stack — is the simplest and oldest realization. Alternating layers of high and low refractive index materials with optical thickness nd = λ/4 produce the strongest reflectance at the design wavelength. The multilayer thin-film coating on every laser mirror, camera lens, and telescope mirror is a 1D photonic crystal engineered for a specific band gap.
The 1D gap is polarization-independent at normal incidence but splits into TE and TM bands at oblique angles. An omnidirectional reflector requires a sufficiently large gap so that the angular-shifted TE and TM bands still overlap for all angles.
A 2D photonic crystal is periodic in two dimensions and uniform (or slab-confined) in the third. The most common realizations are triangular or square lattices of air holes in a high-index dielectric slab, or high-index rods in air. The band structure must be computed numerically (plane-wave expansion, FDTD, or transfer matrix methods). The gap properties depend strongly on polarization: TE modes (E in-plane) and TM modes (H in-plane) generally have gaps at different frequency ranges.
The triangular lattice of rods opens a particularly large TM gap because the circular cross-section naturally breaks the connectivity of the high-index region, forcing TM modes to concentrate in the (blocking) low-index gaps. The honeycomb and Kagome lattices have been explored for their multiple Dirac-point-adjacent band gaps and topological properties.
A complete 3D photonic band gap — a frequency range in which no propagating modes exist for any direction or polarization — requires a high dielectric contrast and a lattice geometry that equalizes the BZ extent in all directions. The woodpile (layer-by-layer) structure, inverse opal (FCC), and diamond-lattice configurations are the principal 3D geometries with experimentally verified complete gaps.
The diamond and woodpile structures achieve large gaps because their FCC-like Brillouin zone (a truncated octahedron) is nearly spherical — the ratio of shortest to longest BZ radius is maximized, minimizing the frequency mismatch between gap edges in different directions. Silicon (n ≈ 3.45 at 1550 nm) and GaAs (n ≈ 3.6) comfortably exceed the contrast threshold for near-IR complete gaps.
The photonic band gap enables a unique form of confinement: introducing a structural defect (a missing rod, an extra layer, a modified hole) creates a localized mode at a frequency inside the gap. Because the bulk crystal supports no propagating modes at that frequency, the defect mode decays exponentially into the surrounding crystal — analogous to a donor/acceptor level in an electronic semiconductor.
Line defects (a row of removed holes/rods) form photonic crystal waveguides: the guided mode lies inside the gap of the bulk crystal and is therefore lossless in the plane (though not necessarily in 3D for slab geometries). Point defects form high-Q resonators with mode volumes approaching (λ/2n)³. The combination of ultrahigh Q and ultrasmall V enables strong-coupling cavity QED, single-photon sources, and on-chip nonlinear optics at milliwatt powers.
Animated Field Propagation in a 2D Crystal
The canvas below simulates wave propagation through a 2D periodic dielectric. A continuous wave is launched from the left. Inside the crystal, a frequency near the band gap edge shows strong Bragg scattering and field localization. The color map represents field amplitude.
Propagating field (blue/cyan) versus evanescent decay inside gap (field collapses to zero beyond a few unit cells). The standing-wave pattern at the crystal edge is the hallmark of Bragg reflection.
Core Physical Concepts
Band Structure Calculation Approaches
Approach: Expand H and ε⁻¹ in plane waves; diagonalize matrix
Accuracy: Exponential convergence in N_pw (plane waves)
Typical N: 256–2048 plane waves for 2D; 10⁴+ for 3D
Gold-standard for band structures. Handles arbitrary geometry via supercell. Cannot directly compute transmission or time-domain response.
Approach: Discretize Maxwell's equations on Yee grid; time-step
Strength: Transmission spectra, defect modes, nonlinear
Cost: O(N³) in 3D per time step; dispersion from Fourier
Best for computing transmission/reflection spectra, near-field profiles, and resonator Q-factors. Band structure requires Fourier analysis of time-domain response.
Approach: Propagate 2×2 or 4×4 matrices through layers
Speed: Extremely fast — milliseconds per spectrum
Limit: Strictly valid only for planar geometries
Ideal for 1D Bragg stack design and optimization. Produces exact analytic band edges and reflectance. Generalizes to 2D via rigorous coupled-wave analysis (RCWA).
Approach: Variational formulation on unstructured mesh
Strength: Complex geometries, curved boundaries, anisotropy
Cost: High memory for 3D; requires careful mesh refinement
Best for fabrication-realistic geometries with smooth walls, tilted sidewalls, and mixed materials. Required when geometric imperfections must be modeled.
Language: Python (meep/mpb via conda)
Output: ω(k) arrays + gap-midgap ratio
See code tab below for a working example computing the band structure and identifying the gap.
Language: Python (meep)
Output: T(ω), R(ω), field animations
FDTD is the workhorse for transmission/reflection spectra. A broadband Gaussian pulse excites all modes; Fourier transform yields the full spectrum in one run.
Code Sketch — Band Structure (MPB)
# Triangular lattice PhC band structure via MPB (python-meep)
import meep as mp
from meep import mpb
import numpy as np
import matplotlib.pyplot as plt
# ── Geometry: triangular lattice of Si rods in air ────────
a = 1.0 # normalized lattice constant
r = 0.2 # rod radius / a (fill fraction ≈ 0.145)
n_si = 3.4 # Si refractive index (near-IR)
eps_si = n_si**2 # ε = 11.56
geometry_lattice = mp.Lattice(
size=mp.Vector3(1, 1),
basis1=mp.Vector3(1, 0),
basis2=mp.Vector3(0.5, np.sqrt(3)/2) # triangular basis
)
geometry = [mp.Cylinder(r, material=mp.Medium(epsilon=eps_si))]
# ── k-path: Γ → M → K → Γ ────────────────────────────────
Gamma = mp.Vector3(0, 0)
M = mp.Vector3(0, 0.5)
K = mp.Vector3(1/3, 1/3)
k_points = [Gamma, M, K, Gamma]
k_interp = 20 # interpolation points between each pair
# ── Run plane-wave expansion ──────────────────────────────
ms = mpb.ModeSolver(
geometry_lattice=geometry_lattice,
geometry=geometry,
k_points=k_points,
num_bands=8,
resolution=32 # grid points per unit cell axis
)
ms.run_tm() # TM polarization
tm_freqs = ms.all_freqs # shape: (n_k, num_bands)
tm_gaps = ms.gap_list # [(gap%, f_low, f_high), ...]
# ── Identify and report gaps ─────────────────────────────
for gap in tm_gaps:
gap_pct, f_low, f_high = gap
print(f"TM gap: {f_low:.4f} – {f_high:.4f} ({gap_pct:.1f}% gap-midgap)")
# ── Plot band structure ───────────────────────────────────
fig, ax = plt.subplots(figsize=(8, 5))
n_k = len(tm_freqs)
for band in range(ms.num_bands):
ax.plot(tm_freqs[:, band], color='steelblue', lw=1.5)
# Shade gap region
for gap in tm_gaps:
ax.axhspan(gap[1], gap[2], alpha=0.15, color='orange', label='Band Gap')
ax.set_xlabel('k (Γ → M → K → Γ)')
ax.set_ylabel('Frequency (ωa/2πc)')
ax.set_title('TM Band Structure — Triangular Lattice PhC')
plt.tight_layout(); plt.savefig('phc_bands.png', dpi=200)
Photonic Crystal Technologies
Realization Methods by Wavelength
Feature size: a ≈ 400–500 nm, holes ≈ 100–200 nm
Materials: Si, GaAs, InP on SiO₂ or suspended membrane
Process: EBL → ICP-RIE etch → undercut for membrane
Standard platform for telecom-band PhC lasers and waveguides. Achieves lattice precision ~5 nm — critical for Q-factors above 10⁵.
Feature size: a ≈ 500 nm (193 nm DUV stepper)
Throughput: Wafer-scale, CMOS-compatible
Platform: Si photonics on SOI wafers
Enables volume production of PhC devices integrated with Si photonics. Geometry limited to 2D slabs; 3D requires layer-by-layer approaches.
Material: SiO₂ or PS spheres → inverse opal (TiO₂, Si)
Lattice: FCC via sedimentation or convective assembly
Gap width: ~5–8% for direct opal; ~20% for inverse Si opal
Low-cost route to 3D photonic crystals. Disorder limits Q-factors; best used for structural color, sensors, and proof-of-concept 3D gap demonstrations.
Resolution: ~100–200 nm voxel (Nanoscribe GT2)
Materials: Photoresist; infiltrated with high-n materials
Advantage: Arbitrary 3D geometry, direct-write
Only method for truly arbitrary 3D PhC geometries including woodpile, gyroid, and Schwartz-P structures. Index limited to ~1.6 unless infiltrated.
Approach: Interfere 4 laser beams to expose photoresist in 3D
Symmetry: BCC, FCC, diamond lattices via beam geometry
Advantage: Large-area, single-exposure 3D patterning
Produces periodic 3D structures over cm² areas in a single exposure. Requires high-quality spatial coherence and vibration isolation. Demonstrated complete gaps at near-IR in Si-infiltrated templates.
Materials: Alumina rods (ε ≈ 10) or metal rod arrays
Advantage: Easy fabrication; perfect analog to optical PhC
Measurement: Network analyzer transmission; direct field mapping
The scale invariance of Maxwell's equations makes cm-scale microwave structures perfect analogs of nm-scale optical crystals. Topological, nonlinear, and defect physics were all first demonstrated at microwave scales.