Birefringence

Two indices, one crystal

In an isotropic medium (water, glass), the refractive index is a single number — the same in every direction. In an anisotropic crystal, the speed of light depends on the polarization direction and the direction of propagation. The simplest case (uniaxial) has just two principal indices:

• Ordinary index n_o. Applies to light polarized perpendicular to the crystal's optical axis. Obeys Snell's law like a normal isotropic refractor.
• Extraordinary index n_e. Applies to light polarized along the optical axis. Its effective index depends on the angle of propagation — and the e-ray's Poynting vector (energy flow) doesn't follow the wavevector. That's the "walk-off."

When unpolarized light enters such a crystal at an angle to the optical axis, it splits into two beams that travel separate paths. Each beam emerges linearly polarized, with the two polarizations mutually perpendicular. Bartholin saw this in 1669 with Iceland spar (calcite from Iceland); it baffled Newton because his corpuscular theory had no way to explain two-by-two refraction.

Worked example — calcite at 589 nm

Calcite (CaCO₃) is the textbook birefringent crystal because its Δn is enormous compared to most natural materials.

n_o = 1.658   (ordinary, E ⊥ axis)
n_e = 1.486   (extraordinary, E ∥ axis at 90° to propagation)
Δn  = n_e − n_o = −0.172   (negative uniaxial)

For a 1 cm thick calcite rhomb at normal incidence:

• The o-ray passes straight through.
• The e-ray walks off by angle ρ ≈ tan⁻¹(|Δn|·sin·cos / ...) — roughly 6° in calcite, depending on the crystal cut. Inside a 1 cm rhomb, that's ~1 mm lateral displacement.
• Place a Polaroid above the rhomb and rotate it: one image vanishes, then the other.

Waveplates — turning polarization with a slab

Cut a birefringent crystal so the optical axis lies in the face. Light entering at normal incidence sees both indices, but the two orthogonal polarizations now propagate along the same path at different speeds. They emerge with a relative phase:

Δφ = (2π / λ) · Δn · d

Pick d so Δφ = π/2 — that's a quarter-wave plate. Linear polarization at 45° to the axis becomes circular. Pick Δφ = π — half-wave plate. Linear polarization at angle α to the axis becomes linear at angle −α (the axis flips the orientation).

Material	Δn at 589 nm	λ/4 thickness	Notes
Quartz	+0.0092	16 µm	Standard waveplate material — durable, transparent into UV
Calcite	−0.172	0.86 µm	Too thin to be practical alone; used in prisms instead
Mica	+0.0040 (avg)	37 µm	Cheap natural — historic standard before quartz dominated
MgF₂	+0.012	12 µm	Excellent transmission into UV
Liquid crystal cell	0.10 to 0.30 (tuneable)	0.5–2 µm	Voltage-tunable Δn — every LCD pixel

Common birefringent materials

Material	n_o	n_e	Δn = n_e − n_o	Why it's used
Rutile (TiO₂)	2.616	2.903	+0.287	Highest-Δn natural crystal; used in white pigment too
Calcite (CaCO₃)	1.658	1.486	−0.172	Polarizing prisms (Nicol, Glan-Thompson, Wollaston)
Sapphire (Al₂O₃)	1.768	1.760	−0.008	Hard, scratch-resistant; mild birefringence
Quartz (SiO₂)	1.544	1.553	+0.009	Stock waveplate material
Ice	1.309	1.310	+0.001	Atmospheric haloes; auroral polarization
Stressed acrylic / glass	~1.49	~1.49 ± stress·C	~10⁻⁵ to 10⁻³	Photoelastic stress visualization

Costed claims you can verify

• Calcite Δn = 0.172 at 590 nm. Verified by every introductory optics textbook; cited indices n_o = 1.658, n_e = 1.486.
• Calcite walk-off ≈ 6° at the principal section. A 1 cm rhomb produces ~1 mm separation between the two emerging images.
• Quartz λ/4 plate at 589 nm: 16 µm thick. d = λ/(4·Δn) = 589e-9 / (4 × 0.0092) ≈ 16 µm.
• LCD switching ~10 ms. The liquid-crystal Δn change in response to voltage settles in milliseconds; that's the display refresh ceiling for many panels.
• Photoelastic stress-optic coefficient (typical polymer): C ≈ 5 × 10⁻¹⁰ Pa⁻¹. A 100 MPa stress produces Δn ≈ 5 × 10⁻⁵ — fringes visible in a 5 mm sample.

JavaScript — birefringence calculations

// Waveplate phase retardation
function retardation(deltaN, d, wavelength) {
  return 2 * Math.PI * deltaN * d / wavelength;
}

const lambda = 589e-9;
const quartzDn = 0.0092;

// λ/4 plate: Δφ = π/2 → d = λ / (4·Δn)
const dQuarterQuartz = lambda / (4 * quartzDn);
console.log(`Quartz λ/4 at 589 nm: ${(dQuarterQuartz * 1e6).toFixed(2)} µm`);
// 16.0 µm

// λ/2 plate
const dHalfQuartz = lambda / (2 * quartzDn);
console.log(`Quartz λ/2 at 589 nm: ${(dHalfQuartz * 1e6).toFixed(2)} µm`);
// 32.0 µm

// Calcite walk-off (simplified; principal section)
// tan(ρ) ≈ (n_o² − n_e²) · sin(2θ) / (2 · (n_o²·sin²θ + n_e²·cos²θ))
function walkOff(theta, no, ne) {
  const s = Math.sin(theta), c = Math.cos(theta);
  const num = (no*no - ne*ne) * Math.sin(2*theta);
  const den = 2 * (no*no * s*s + ne*ne * c*c);
  return Math.atan(num / den) * 180 / Math.PI;
}

console.log(`Calcite walk-off at θ=45°: ${walkOff(Math.PI/4, 1.658, 1.486).toFixed(2)}°`);
// ~5.7° — close to the ~6° rule of thumb

// Photoelastic fringes in stressed polymer
// Number of full-wave fringes = Δn · d / λ
function photoelasticFringes(stressMPa, d_mm, stressOpticCoeff_perPa, wavelength_nm) {
  const dn = stressOpticCoeff_perPa * stressMPa * 1e6;
  return dn * d_mm * 1e-3 / (wavelength_nm * 1e-9);
}

// 50 MPa stress, 5 mm sample, C = 5e-10 /Pa, λ = 550 nm
console.log(`Fringes: ${photoelasticFringes(50, 5, 5e-10, 550).toFixed(1)}`);
// ~22.7 fringes

// Number of distinct walked-off image copies in a calcite stack of N rhombs
function nestedImageSeparation(N, perRhomb_mm) {
  return N * perRhomb_mm; // for aligned rhombs
}

console.log(`5 stacked rhombs (1 mm each): ${nestedImageSeparation(5, 1)} mm total split`);

Where birefringence shows up

• LCD displays. Voltage rotates the liquid-crystal director → controls Δn → controls phase retardation → controls how much light each pixel transmits between two polarizers.
• Waveplates. λ/4 plates make linear-to-circular polarization (used in 3D cinema, optical isolators); λ/2 plates rotate linear polarization (used in laser power control with a polarizer).
• Polarizing prisms. Glan-Thompson, Wollaston, Nomarski — all use calcite birefringence to split or filter polarizations with high extinction (10⁻⁶).
• Polarization microscopy. Geologists identify minerals by birefringence colors in thin sections under crossed polarizers; biologists detect collagen, amyloid plaques, starch granules.
• Photoelasticity. Engineers visualize stress in transparent models (sometimes called "stress patterns in plastic"); now mostly superseded by FEA but still used in teaching.
• Polarization-maintaining fiber. Intentional birefringence keeps two polarization modes from mixing — used in fiber-optic gyroscopes and quantum-key distribution.

Common mistakes

• Assuming all crystals are birefringent. Cubic-symmetry crystals (NaCl, diamond, silicon) are isotropic — no birefringence. Only crystals with at least one preferred axis can split light.
• Confusing walk-off with refraction. The e-ray's wavevector and energy-flow vector aren't parallel inside a birefringent crystal. Walk-off is a real spatial displacement even at normal incidence; refraction is angle-change at the boundary.
• Ignoring wavelength dispersion. Δn varies with λ. A perfect λ/4 plate at 589 nm is roughly λ/4 at 633 nm but visibly off near the violet. Achromatic (zero-order or compound) waveplates correct this.
• Forgetting temperature dependence. Δn drifts ~10⁻⁵/K. For precision interferometry, waveplates need thermal control.
• Using a "false zero" thickness for waveplates. The phase wraps every 2π. A "5/4 wave" plate behaves like λ/4 but is more dispersive; "zero-order" plates have the thinnest design and best achromaticity.
• Skipping the optical axis specification. A birefringent crystal cut at the wrong angle splits light very differently. Always specify the optical-axis orientation in your setup.