32
Part One · Quantum mechanics

Free particle and wavepackets

Pin an electron down to within a nanometre and let go. Thirty femtoseconds later its wavefunction is twice as wide. Do the same to a speck of dust and you would wait about a year. Same equation, same physics; only the mass is different. This chapter is about why quantum objects refuse to stay put, and why you never notice.

On the plate, a free wave packet, launched with no forces on it at all, drifts and spreads.

A plane wave is nowhere

The simplest solution of the free Schrödinger equation is a plane wave, ei(kx−ωt)e^{i(kx - \omega t)} with ω=ℏk2/2m\omega = \hbar k^2/2m. It has a perfectly definite momentum ℏk\hbar k and is spread evenly over all of space, so it cannot describe a particle that is here. To build something localized you add plane waves with a spread of kk, exactly as Chapter 6 built a pulse from sines.

Add 5 waves with nearby wavenumbers. With one wave, nothing is localized. With a few, lumps appear and repeat. With many, a single packet stands alone: the waves agree in one place and cancel everywhere else.

Two speeds

Watch the packet move. The ripples inside it, marked by the small white dot, travel at the phase velocity ω/k=ℏk/2m\omega/k = \hbar k/2m. The envelope, marked in amber, travels at the group velocity dω/dk=ℏk/md\omega/dk = \hbar k/m, which is twice as fast. The ripples are born at the back of the packet and die at the front.

Only the group velocity matches the classical p/mp/m. The phase velocity has no physical meaning on its own; you can change it by adding a constant to the potential energy, which shifts every frequency without changing anything you can measure. Established

How fast a packet spreads

For a Gaussian packet that starts with width σ0\sigma_0 and no internal chirp, the Schrödinger equation can be solved exactly:

σ(t)=σ01+(tτ)2,τ=2mσ02ℏ\sigma(t) = \sigma_0\sqrt{1 + \left(\frac{t}{\tau}\right)^{2}}, \qquad \tau = \frac{2m\sigma_0^2}{\hbar}

Start narrower and it spreads faster. Set the starting width to 1.0. The plate runs the solver and plots the measured width in amber against the formula. They lie on top of each other.

This is the uncertainty principle in motion. A narrow packet must contain a wide spread of momenta, σp≥ℏ/2σ0\sigma_p \ge \hbar/2\sigma_0, and the fast parts outrun the slow ones. At late times the width grows as σpt/m\sigma_p t/m, exactly like a cloud of classical particles flying apart at different speeds.

The momentum never changes

Here is something that surprises people. While the packet spreads in position, its momentum distribution, the plate’s lower curve, does not change at all. With no force, each plane wave keeps its own momentum; only their relative phases change. The packet gets wider, but it is not getting more uncertain about its speed. It is becoming correlated: fast parts at the front, slow parts at the back.

Electron or dust grain

The time to double in width is 3 τ\sqrt3\,\tau. Localize each of these to one nanometre and see how long it takes: , , , .

The electron doubles in 30 femtoseconds. A buckyball, the 60-carbon molecule that has been sent through double slits, takes 40 nanoseconds. A virus particle, about 30 milliseconds. A microgram of dust, about a year. Spreading is universal, but its rate falls with mass, and for anything you can see it is hopelessly slow.

NoteThis is not the whole reason everyday objects look classical. Even slow spreading would eventually make a dust grain delocalized; what stops it is that air molecules and light keep scattering off it, which destroys the coherence between distant parts of its wavefunction long before it can spread. That process, decoherence, is the subject of Part Two.

What you now have

Wave packets as sums of plane waves, phase and group velocity, the exact spreading law σ(t)\sigma(t) and its time scale τ=2mσ02/ℏ\tau = 2m\sigma_0^2/\hbar, the constant momentum distribution, and a feel for why heavy things do not spread noticeably.

· · ·

Three questions to carry with you.

Which velocity of a packet matches the classical p/m?tap
The group velocity, dω/dk = ħk/m; the phase velocity is half that and not measurable on its own.
Why does a narrower packet spread faster?tap
It contains a wider range of momenta (σ_p ≥ ħ/2σ₀), so its parts separate faster.
Does the momentum uncertainty grow as a free packet spreads?tap
No. The momentum distribution is unchanged; position and momentum become correlated instead.

And one to work out. An electron is localized to σ0=0.1\sigma_0 = 0.1 nm, about an atom’s size. What is τ=2mσ02/ℏ\tau = 2m\sigma_0^2/\hbar, in femtoseconds? Use m=9.109×10−31m = 9.109\times10^{-31} kg and ℏ=1.0546×10−34\hbar = 1.0546\times10^{-34} J s.

τ =fs

Sources: D. J. Griffiths and D. F. Schroeter, Introduction to Quantum Mechanics, §2.4. M. Arndt et al., Nature 401, 680 (1999) (C₆₀ interference). Checked by code: τ for 1 nm localization: electron 1.73×10⁻¹⁴ s, C₆₀ 2.27×10⁻⁸ s, 1 fg virus 0.019 s, 1 µg dust 1.9×10⁷ s; doubling time √3τ.

Next · Chapter 33
Infinite and finite square wells

Quantum dots glow different colors just by changing size. The color is set by the box.

Turn the page