47
Part One · Quantum mechanics

WKB approximation

In 1913 Bohr quantized the atom with a rule that seemed arbitrary: the action around an orbit must be a whole number of Planck’s constants. Thirteen years later, the Schrödinger equation replaced it. But Bohr’s rule was not wrong. It is the first term of an expansion of the Schrödinger equation in powers of ℏ\hbar, and with one small correction it predicts energies to a fraction of a percent. This chapter derives it, and uses it to explain neutrons that bounce on a mirror in quantized steps of a few micrometres.

On the plate: an exact wavefunction, and the envelope the semiclassical method predicts for it.

Waves with slowly changing wavelength

Where the potential is constant, ψ=eipx/ℏ\psi = e^{ipx/\hbar} with p=2m(E−V)p = \sqrt{2m(E - V)}. If VV changes slowly, let p(x)p(x) change too. Substituting ψ=eiS(x)/ℏ\psi = e^{iS(x)/\hbar} and expanding SS in powers of ℏ\hbar gives, to the first two orders,

ψ(x)≈Cp(x)exp⁡ ⁣(±iℏ∫xp(x′) dx′)\psi(x) \approx \frac{C}{\sqrt{p(x)}}\exp\!\left(\pm\frac{i}{\hbar}\int^x p(x')\,dx'\right)

This is the Wentzel–Kramers–Brillouin (WKB) approximation of 1926. The phase advances by the local wavenumber; the amplitude goes as 1/p1/\sqrt p, so ∣ψ∣2∝1/p|\psi|^2 \propto 1/p, which is the classical rule that a particle is most likely found where it moves slowest. Step through the states of a quartic well, V=x4V = x^4, with 8. The amber envelope is WKB’s; the white wave is the exact solution.

Where it breaks

At a turning point, where E=VE = V, p=0p = 0 and the formula blows up. The wavelength is no longer short compared with the distance over which it changes. The cure: near the turning point, the potential is nearly a straight line, and the Schrödinger equation for a linear potential is solved exactly by the Airy function. Matching the Airy function to WKB on both sides gives the connection formulas: the oscillating wave inside joins the decaying one outside with a phase of π/4\pi/4.

The plate zooms in on one turning point. Amber is WKB, which diverges; white is the exact Airy function, which passes smoothly into the wall.

Bohr’s rule, corrected

Fit the wave between two turning points, with a phase of π/4\pi/4 lost at each, and you get

∮p dx=(n+12)h\oint p\,dx = \left(n + \tfrac12\right)h

The area enclosed by the classical orbit in the plane of position and momentum is a whole number of hh, plus a half. That half is the zero-point energy of Chapter 35, which Bohr’s original rule missed. Choose a potential: , or .

For the harmonic oscillator, WKB is exact. For the quartic well it is 18% off for the ground state, 1.3% for the next, and 0.06% by n=7n = 7: better and better as the motion becomes more classical. Established

Neutrons on a mirror

A particle above a hard floor in gravity has the potential mgzmgz above z=0z = 0 and a wall below. The exact levels involve the zeros of the Airy function; WKB, with a wall on one side, gives (n−14)(n - \tfrac14) instead of (n+12)(n + \tfrac12) and matches them to within 1% even for the lowest.

For neutrons, the natural length is 5.87 µm, and the lowest states have heights of about 13.7, 24.0, 32.4 and 39.8 µm. In 2002 Valery Nesvizhevsky and colleagues at the Institut Laue–Langevin sent ultracold neutrons over a mirror under an absorbing ceiling. Lowering the ceiling below about 15 µm stopped transmission almost completely: the neutrons could not fit. Gravity, the weakest force, quantized in the lab. Established

Tunnelling, approximately

Under a barrier, pp becomes imaginary and WKB’s wave becomes a decaying exponential, e−∫κ dxe^{-\int\kappa\,dx}. So the transmission through any smooth barrier is roughly

T≈exp⁡ ⁣(−2ℏ∫x1x22m(V−E) dx)T \approx \exp\!\left(-\frac{2}{\hbar}\int_{x_1}^{x_2}\sqrt{2m(V - E)}\,dx\right)

This is Gamow’s factor for alpha decay from Chapter 34. Applied to a Coulomb barrier, it gives the Geiger–Nuttall law. The same formula sets the rates of fusion in the Sun and of hydrogen transfer in some enzymes, which Part Four will examine.

When to trust it

The condition is that the wavelength changes little over one wavelength: ∣dλ/dx∣≪1|d\lambda/dx| \ll 1. That fails near turning points, at sharp steps, and for low quantum numbers. Michael Berry and Kenneth Mount showed in 1972 that the full series in ℏ\hbar is asymptotic, like the perturbation series of Chapter 44, and that its failures encode real physics, such as the exponentially small reflection from a smooth step that no finite order captures.

What you now have

The WKB wavefunction and its 1/p1/\sqrt p amplitude, turning points and Airy functions, the corrected Bohr–Sommerfeld rule, the quantum bouncer and the neutron experiment, the Gamow factor, and the limits of validity.

· · ·

Three questions to carry with you.

What does the ½ in (n + ½)h represent?tap
The phase of π/4 lost at each turning point; it produces the zero-point energy.
Why is WKB worst for the ground state?tap
The wavelength is long compared with the well, so it changes rapidly relative to itself.
What did the 2002 neutron experiment show?tap
Quantized states of neutrons in Earth’s gravity, with the lowest at about 14 µm height.

And one to work out. The neutron’s natural length in gravity is ℓ0=5.87\ell_0 = 5.87 µm, and the first Airy zero is 2.338. How high is the lowest state, 2.338 ℓ02.338\,\ell_0, in µm?

z₁ =µm

Sources: G. Wentzel, Z. Phys. 38, 518 (1926); H. A. Kramers, Z. Phys. 39, 828 (1926); L. Brillouin, C. R. Acad. Sci. 183, 24 (1926). M. V. Berry and K. E. Mount, Rep. Prog. Phys. 35, 315 (1972). V. V. Nesvizhevsky et al., Nature 415, 297 (2002). Checked by code: quartic well exact 0.668, 2.394, 4.697 against WKB 0.546, 2.364, 4.671; Airy zeros 2.338, 4.088, 5.521 against WKB 2.320, 4.082, 5.517; ℓ₀ = 5.87 µm, heights 13.7, 24.0, 32.4, 39.8 µm.

Next · Chapter 48
Scattering theory

Everything we know about the inside of a proton came from throwing things at it and watching how they bounce.

Turn the page