Visual walkthrough — Birefringence — ordinary and extraordinary rays

This page is the visual companion to Birefringence — ordinary and extraordinary rays. It leans on Dielectric tensor and anisotropic media for why the crystal has two indices, and pays off into Wave plates (quarter and half wave).

---

Step 0 — The three words we are allowed to use

Before any equation, let us pin down every symbol in plain language and glue each to a picture.

Everything below is a story about how these two fixed numbers, $n_o$ and $n_e$, blend as the angle $\theta$ swings from $0$ to $90^{\circ }$.

---

Step 1 — Draw the two "stiffnesses" as two lengths

WHAT. We place the optic axis vertically ($z$) and the sideways directions horizontally ($x,y$). Along $z$ the crystal responds with index $n_e$; sideways it responds with index $n_o$.

WHY. Dielectric tensor and anisotropic media tells us the crystal's electrons sit on springs: stiff one way, soft another. Stiffness $\to$ permittivity $\to$ index. So before we do any geometry, the physics has already handed us exactly two numbers, one per direction. We just need a picture that stores both.

PICTURE. Two arrows from the centre: a short/long one up the axis (length $n_e$) and one sideways (length $n_o$). Two lengths, nothing more yet.

---

Step 2 — Grow those two lengths into an ellipse

WHAT. Sweep the two lengths around the centre. Where a direction is horizontal, the radius is $n_o$; where vertical, the radius is $n_e$; in between, it smoothly interpolates. The result is an ellipse (in 3D, an ellipsoid) — the index ellipsoid.

WHY. We do not want just two directions; we want a rule that gives an index for every possible wiggle direction. An ellipse is the simplest curve that hits $n_o$ sideways and $n_e$ vertically and blends between them. This single curve now stores the answer for all directions at once.

PICTURE. The ellipse below: horizontal semi-axis $=n_o$, vertical semi-axis $=n_e$. Each point on the curve is a possible wiggle direction; its distance from the centre is the index for that wiggle.

---

Step 3 — Point the light, and slice the ellipsoid

WHAT. Choose a travel direction $\hat{k}$ making angle $\theta$ with the optic axis. Cut the ellipsoid with the flat plane that passes through the centre and is perpendicular to $\hat{k}$. The cut is an ellipse.

WHY. A light wave's wiggle ($\vec{E}$, and its partner $\vec{D}$) must lie perpendicular to the travel direction — that is what "transverse wave" means (see Polarization of light). So the only wiggle directions the wave is even allowed to use are the ones lying in that perpendicular plane. Slicing the ellipsoid by that plane collects exactly those allowed directions.

PICTURE. The travel arrow $\hat{k}$ tilts by $\theta$ from vertical. The grey disk is the perpendicular slice; its rim is the small ellipse of allowed wiggle directions.

---

Step 4 — The two semi-axes = the two rays

WHAT. Any ellipse has a longest radius and a shortest radius — its two semi-axes, at right angles. On our sliced ellipse these two lengths are the two allowed indices: one is the o-ray, the other is the e-ray.

WHY. Nature only lets a wave travel with a wiggle along a semi-axis of this slice — those are the "stable" polarizations that keep their shape as they move. So a single beam is forced into two independent waves, each with its own index. That is the very splitting birefringence is named for.

PICTURE. On the slice: one semi-axis stays horizontal (its length is exactly $n_o$ — this is the o-ray); the perpendicular semi-axis is tilted and has some in-between length $n_e(\theta )$ — this is the e-ray.

---

Step 5 — Set up the right triangle for the e-ray

WHAT. The e-ray's wiggle direction $\vec{D}$ lies in the vertical ($xz$) plane and makes angle $\theta$ with the sideways direction — equivalently, its perpendicular makes $\theta$ with the axis. Resolve this unit direction into a piece along the $n_o$-direction and a piece along the $n_e$-direction.

WHY. We need a number for the e-ray's index. The ellipse "knows" how to answer if we feed it a direction split into its two natural axes. So we project the e-ray direction onto the two axes — that projection is just trigonometry on a right triangle.

PICTURE. A unit arrow at angle $\theta$; drop perpendiculars to get two legs: $\cos \theta$ along the horizontal ($n_o$) axis and $\sin \theta$ along the vertical ($n_e$) axis.

---

Step 6 — Combine the pieces as inverse squares

WHAT. Feed those two legs into the ellipse. A point on an ellipse with semi-axes $n_o$ and $n_e$, whose direction has components $\cos \theta$ and $\sin \theta$, sits at distance $n_e(\theta )$ obeying the ellipse rule: $\frac{1}{n_e(\theta {)}^2}=\frac{{\cos }^2\theta }{n_o^2}+\frac{{\sin }^2\theta }{n_e^2}.$

WHY. The ellipse equation from Step 2 said "$(\text{sideways}{)}^2/n_o^2+(\text{axis}{)}^2/n_e^2=1$". A point at distance $r=n_e(\theta )$ in direction $(\cos \theta ,\sin \theta )$ has sideways part $r\cos \theta$ and axis part $r\sin \theta$. Plug those in, and $r^2$ divides out to leave $1/r^2$ on the left — exactly the boxed formula. The inverse squares are not magic; they are the ellipse's own $1/n^2$ weights.

PICTURE. The final ellipse with the e-ray radius drawn and its length labelled $n_e(\theta )$, sitting between $n_e$ and $n_o$.

---

Step 7 — The edge cases (never leave a corner unshown)

WHAT. Test the two extreme angles and confirm the picture behaves.

WHY. A formula you cannot check at its endpoints is a formula you do not trust. These two limits must reproduce the definitions of $n_o$, $n_e$ and the optic axis.

PICTURE. Left panel: $\theta =0$, the slice is horizontal, both semi-axes equal $n_o$ — one circle, no splitting. Right panel: $\theta =90^{\circ }$, the slice is vertical, e-ray semi-axis reaches full height $n_e$ — maximum splitting.

---

The one-picture summary

WHAT. One figure carrying the entire chain: the ellipse (semi-axes $n_o$, $n_e$), a tilted $\hat{k}$, the $\theta$ triangle, and the e-ray radius $n_e(\theta )$ landing between the two. Reading it left to right is the derivation.

Recall Feynman: the whole walk in plain words

Picture a squashed balloon — round side-to-side (radius $n_o$), flatter top-to-bottom (radius $n_e$). That balloon is the crystal's memory of "how slow is light for each wiggle direction." Now shoot a beam through it at some slant angle $\theta$. The beam can only wiggle in the flat plate that sits square across its path, so slice the balloon with that plate. The cut is a little oval with a long side and a short side — those two sides are the only two ways light is allowed to wiggle. One side always stays flat and keeps the fixed length $n_o$: that's the calm ordinary ray. The other side tips as you tip the beam, and its length is a blend of $n_o$ and $n_e$ set purely by the angle — that's the restless extraordinary ray. Split the tilted direction into a "sideways" part ($\cos \theta$) and an "up-the-axis" part ($\sin \theta$), let the balloon weight each part by its own $1/n^2$, add them, and out drops $\frac{1}{n_e(\theta {)}^2}=\frac{{\cos }^2\theta }{n_o^2}+\frac{{\sin }^2\theta }{n_e^2}$. Point the beam straight up the axis and the blend gives back $n_o$ — the two rays fuse. Point it dead sideways and you get the full $n_e$ — maximum split. Everything birefringence does lives in that one squashed balloon.

Recall Quick self-test

Why is the o-ray index constant for all $\theta$? ::: Its wiggle semi-axis always lies flat in the $xy$-plane, never tilting toward the optic axis, so it always equals $n_o$. In the formula, what does ${\cos }^2\theta$ weight? ::: The sideways ($n_o$) contribution to the e-ray index. What is $n_e(\theta )$ at $\theta =0$? ::: Exactly $n_o$ — the optic-axis case with no splitting. Between which two values is $n_e(\theta )$ always trapped? ::: Between $n_o$ and $n_e$.