Band Gaps Are Just Bragg Reflection

This should sound too cheap to be true: the "forbidden energy gap" — the single fact that decides whether a material is a metal, an insulator, or the semiconductor your processor is etched into — is not a quantum mystery at all. It is the same interference that paints a soap bubble. The same interference an X-ray crystallographer uses to read the structure of a protein. The same interference that makes an oil slick on wet tarmac flash green and magenta. One idea — waves bouncing off a regular array and adding up in step — explains the rainbow in the puddle and why a transistor can switch. The band gap is not exotic physics layered on top of ordinary physics. It is ordinary physics, in a place you did not expect to find it.

The usual telling buries this. Open a solid-state textbook and the gap arrives wrapped in Bloch's theorem, periodic potentials, and a wall of operator algebra, as if it had no classical shadow. It does, and the shadow is the whole story. An electron in a crystal is a wave; a crystal is a regularly spaced array of scatterers. When the electron's wavelength hits the one value that makes every scattered wavelet add up in phase, the lattice reflects it completely — the way a properly tuned crystal reflects an X-ray. That reflection is the gap. Everything else is bookkeeping.

A free electron is a plane wave

Before the lattice does anything, take the electron on its own. Strip away the crystal and a single electron in empty space is the simplest object in quantum mechanics: a plane wave, $\psi \sim e^{ikx}$, travelling with a wavevector (momentum, in disguise) $k$. Its energy is pure kinetic, and it depends on $k$ the way kinetic energy always depends on momentum — quadratically.

$$E(k) = \frac{\hbar^2 k^2}{2m}$$

Copy LaTeXwith $$ delimitersCopy mathexpression only

That is the parabola you saw with Potential V₀ at zero. It is smooth and gapless: pick any energy and there is a value of $k$ that has it. Nothing is forbidden, because there is nothing for the wave to bounce off. The crystal introduces the prohibition — and to see how, you stop treating the lattice as a backdrop and start treating it as a diffraction grating.

The wavevector $k$ relates to the de Broglie wavelength the way it always does, $\lambda = 2\pi / k$. Hold onto that, because the entire argument turns on one wavelength hitting one special value.

The lattice is a diffraction grating

A crystal is a row of atoms spaced a fixed distance $a$ apart — the lattice constant. To the travelling electron wave, each atom is a weak scatterer: a little bump in the potential that reflects a sliver of the wave back the way it came. Most of the time these slivers are a mess — each wavelet carries a different phase, they interfere destructively, and they cancel to almost nothing. The electron sails straight through as if the lattice were barely there. This is why metals conduct so well: to a good approximation the electrons are nearly free, and the parabola of equation (1) is nearly right.

But there is one wavelength where the slivers stop cancelling. When the wave reflected off atom $n+1$ comes back exactly one wavelength behind the wave reflected off atom $n$, every reflected wavelet is in phase with every other. The round trip between adjacent atoms is $2a$, so the condition is that $2a$ equals a whole number of wavelengths.

$$2a = n\lambda \quad\Longleftrightarrow\quad k = \frac{n\pi}{a}$$

Copy LaTeXwith $$ delimitersCopy mathexpression only

The left side is the Bragg condition, written for a beam at normal incidence. The general form an X-ray crystallographer uses is $2d\sin\theta = n\lambda$, where $d$ is the spacing between planes of atoms and $\theta$ is the grazing angle of the beam. Our electron travels straight down the line of atoms, so $\theta = 90°$, $\sin\theta = 1$, $d = a$, and the famous equation collapses to $2a = n\lambda$. Substitute $\lambda = 2\pi/k$ and you get the right-hand side: the special momenta are $k = n\pi/a$. Those are precisely the zone boundaries where the gaps opened in the lab.

When the electron hits $k = \pi/a$, the lattice no longer lets it pass. The Bragg condition is met, every scattered wavelet adds in phase, and the wave is thrown straight back. An electron at exactly $k = \pi/a$ cannot propagate forward; it is reflected as hard as the X-ray is. A travelling wave that is constantly and totally reflected cannot survive as a travelling wave. Something else takes its place.

At the boundary, two standing waves, two energies

When a rightward wave $e^{i\pi x/a}$ is perfectly reflected into the leftward wave $e^{-i\pi x/a}$, the two no longer move. They lock into a standing wave. And there is not one way to combine them — there are two, the symmetric sum and the antisymmetric difference, and this is the hinge of the entire argument.

The two standing waves are the cosine and the sine combinations of the counter-propagating pair. Up to normalisation:

$$\psi_{+} \propto \cos\!\left(\frac{\pi x}{a}\right), \qquad \psi_{-} \propto \sin\!\left(\frac{\pi x}{a}\right)$$

Copy LaTeXwith $$ delimitersCopy mathexpression only

Both have the same wavelength, the same $|k|$, the same kinetic energy. If kinetic energy were all that mattered, they would be degenerate — equal in energy, and there would be no gap. But there is a second term: the potential energy of the electron's charge sitting in the lattice's electrostatic landscape. The ions are positively charged; the dips in the potential sit at the ion sites. And here is where the two standing waves stop being equal.

Put the atoms at $x = 0, a, 2a, \dots$. Then $|\psi_{+}|^2 = \cos^2(\pi x/a)$ peaks exactly on the ions — this standing wave piles the electron's charge density right at the positive cores, where the potential is most attractive and the energy is lowest. Meanwhile $|\psi_{-}|^2 = \sin^2(\pi x/a)$ has its nodes on the ions and its peaks in the gaps between them — it parks the electron's charge as far from the attractive cores as it can get, where the energy is highest. Same wave number, same kinetic energy, but two different electrostatic bills.

The gap is twice the relevant Fourier component

How big is the gap? Take the energy difference between the two standing waves. The kinetic parts are identical and cancel, so what is left is the difference in how the two charge distributions sample the periodic potential $V(x)$. Because $V(x)$ is periodic with period $a$, decompose it into a Fourier series — plane waves with wavevectors $G = 2\pi n / a$, the reciprocal lattice vectors. The one that matters at the boundary $k = \pi/a$ is the harmonic whose wavevector connects the incident wave to the reflected one, $G = 2\pi/a$. Call its amplitude $V_G$. The cosine state lowers its energy by $|V_G|$; the sine state raises its by $|V_G|$. The gap is the distance between them.

$$E_\text{gap} = E_{-} - E_{+} = 2\,|V_G|$$

Copy LaTeXwith $$ delimitersCopy mathexpression only

Equation (4) is the whole result, and it is clean. The forbidden gap is twice one Fourier coefficient of the crystal's potential — the single harmonic that matches the Bragg condition. A weak lattice (small $|V_G|$) gives a narrow gap; a strong one gives a wide gap. That is precisely the behaviour you dragged out of the lab: pushing Potential V₀ up widened every gap, because $|V_G|$ scales with the potential's strength. Set $V_G = 0$ and the gap closes, leaving the bare free-electron parabola of equation (1). A flat potential reflects nothing, so there is no Bragg reflection and no gap. The gap exists if and only if the lattice has a Fourier component to reflect from.

The band gap is twice one Fourier coefficient of the crystal — the single harmonic of the lattice that satisfies the Bragg condition, and nothing more.

Now you can read off the entire periodic table's worth of electrical behaviour from one number per material — not the gap itself, but where the highest occupied electrons land relative to it. Fill the bands with the material's electrons up to a level called the Fermi energy, and ask one question: does the Fermi level fall inside a band or inside a gap?

The transistor, the solar cell, the LED, the camera sensor — every one is an engineered answer to "where does the Fermi level sit relative to the gap, and how do we move it?" Doping shifts the Fermi level toward one band edge. A photon with energy above $E_\text{gap}$ lifts an electron clean across — that is how a solar cell turns light into charge, and why it ignores photons too red to clear the gap. The gap you opened by dragging a slider is the gap a chip foundry spends billions controlling.

Silicon gap

1.1 eV

small enough to dope, large enough to switch — the semiconductor sweet spot

Diamond gap

5.5 eV

a wide-gap insulator; the same equation (4), a bigger Fourier component

Thermal energy at 300 K

0.025 eV

kT — the budget an electron has to climb the gap unaided

The same interference, everywhere you have already seen it

The band gap is not a special case. It is one instance of a pattern you have been looking at your whole life without naming it. Anywhere a wave meets a periodic structure with a spacing comparable to its wavelength, the Bragg condition decides which wavelengths reflect and which pass — and the reflected ones are forbidden to propagate, exactly as a band-gap electron is.

The soap film is the cleanest example. A film of soapy water is two surfaces a few hundred nanometres apart. Light reflecting off the back surface travels an extra $2d$ — the same round trip as our electron between atoms — and that path difference sets which colours come back in phase and blaze, while their neighbours interfere away. (A film adds one twist the lattice does not: the reflection off the front surface flips phase by half a wave, nudging the bright condition to $2d = (n + \tfrac{1}{2})\lambda$. The geometry is identical; only the bookkeeping shifts by half a wavelength.) The shifting greens and magentas on a bubble are a photonic Bragg condition, equation (2) with photons, and the colours it cancels are a band gap for light. An oil slick on a wet road does the same with a thinner film. You have been reading equation (2) off puddles since you were a child.

A photonic crystal makes this deliberate: a material whose refractive index varies periodically in space, engineered so that a band of light frequencies satisfies the Bragg condition and cannot propagate through it. Physicists call that band a photonic band gap and mean it literally — it is the optical twin of the electronic gap, built from the same standing-wave argument, the same $\psi_\pm$ piling field energy in high-index versus low-index regions. The iridescent blue of a morpho butterfly's wing is a natural photonic crystal: structural colour, no pigment, just a lattice tuned to Bragg-reflect blue. Opals do it with stacked silica spheres.

And the original instrument runs the same physics in reverse. Where the band-gap argument assumes the lattice and asks what it does to a known wave, X-ray crystallography measures where the diffracted waves land and works backward to the lattice. W. L. Bragg used $2d\sin\theta = n\lambda$ in 1913 to read the spacing of atoms in rock salt off the angles of the reflected X-ray spots — among the first crystal structures ever solved this way. The same condition that forbids an electron's energy gives the crystallographer the atomic structure. One equation, pointed two directions: tell me the wave and I will tell you which energies are forbidden; tell me which angles are bright and I will tell you the lattice.

That is the unification worth carrying away. The rainbow in the oil slick, the diffraction pattern of a protein crystal, the blue of a butterfly, and the reason your processor can hold a 1 or a 0 are not four phenomena. They are one — a wave, a periodic array, and the Bragg condition deciding what gets through.

Reading further

• Kittel, Introduction to Solid State Physics, 8th ed., chapter 7 — the standard undergraduate derivation of the energy gap from the nearly-free-electron model; equations (3) and (4) come straight from its standing-wave argument.
• Ashcroft & Mermin, Solid State Physics, chapters 8–9 — the graduate treatment, with Bloch's theorem and the full reciprocal-lattice machinery behind the Fourier-component gap; read this when you want the algebra under the geometry.
• W. L. Bragg, "The Diffraction of Short Electromagnetic Waves by a Crystal" (Proc. Camb. Phil. Soc., 1913) — the original paper that turned $2d\sin\theta = n\lambda$ into an instrument and founded X-ray crystallography; the same condition, pointed at structure instead of energy.
• Joannopoulos et al., Photonic Crystals: Molding the Flow of Light, 2nd ed. — the photonic-band-gap case, where the identical standing-wave argument forbids bands of light instead of electrons; the cleanest cross-field bridge from the gap to the soap film.