66
Part Two · Quantum fields and particles

Quantizing the electromagnetic field

Place two uncharged, perfectly flat metal plates a micrometre apart in a vacuum. Nothing is between them, nothing is on them, and yet they are pushed together. In 1948 Hendrik Casimir predicted the force and its size exactly; in 1997 Steve Lamoreaux measured it to within 5%. The usual story is that the vacuum between the plates holds fewer modes of the electromagnetic field than the vacuum outside. This chapter quantizes that field, and then asks whether the usual story is right.

On the plate: two plates, and the field modes that fit between them.

Too many components

The electromagnetic field is described by the four-potential AμA^\mu with Lagrangian −14FμνFμν-\tfrac14F_{\mu\nu}F^{\mu\nu} (Chapter 58). Four components, but light has only two polarizations. The extra two are not physical: changing Aμ→Aμ+∂μχA_\mu \to A_\mu + \partial_\mu\chi leaves FF and all physics unchanged (Chapter 42). To quantize, fix the freedom with a gauge. In the Coulomb gauge, ∇⋅A=0\nabla\cdot\mathbf A = 0, and A0A^0 is fixed by the charges; what remains are two transverse components for each wave vector.

Each wave vector and polarization is a harmonic oscillator, quantized exactly as in Chapter 60. One quantum is one photon: energy ℏω=ℏc∣k∣\hbar\omega = \hbar c|\mathbf k|, momentum ℏk\hbar\mathbf k, two possible polarizations. Established

Two polarizations

A massless particle moving along k\mathbf k can have its spin only along or against its motion, helicity ±1\pm1: a spin-1 particle loses the zero state when it has no mass. Choose or polarization to see the two basis choices. A massive spin-1 particle, like the W and Z, has three states; Chapter 77 shows where the third comes from.

Counting modes between plates

Between perfectly conducting plates a distance dd apart, the field must vanish at the metal, so only waves with kz=nπ/dk_z = n\pi/d fit, like the box of Chapter 33. Outside, every kzk_z is allowed. Both regions have infinite zero-point energy (Chapter 60), but the difference between them, carefully computed, is finite. The result is an attractive pressure

FA=−π2ℏc240 d4\frac{F}{A} = -\frac{\pi^2\hbar c}{240\,d^4}

Set the gap to 1.0 µm. At 1 µm the pressure is 1.3 mPa, a hundred-millionth of atmospheric pressure; at 100 nm it is 13 Pa, and at that scale it makes the tiny moving parts of microelectromechanical devices stick.

DeeperThe mode sum involves ∑n3\sum n^3, which diverges. Regularized with a smooth cutoff or with the Riemann zeta function, ζ(−3)=1/120\zeta(-3) = 1/120, the cutoff-dependent part cancels between inside and outside, and π2/240\pi^2/240 remains. The same trick famously assigns ζ(−1)=−1/12\zeta(-1) = -1/12 to 1+2+3+⋯1 + 2 + 3 + \cdots.

Measured

Parallel plates are nearly impossible to align, so Lamoreaux in 1997 used a lens-shaped surface near a flat plate on a torsion pendulum, at separations of 0.6 to 6 µm, and found the predicted force within 5%. Umar Mohideen’s group, with an atomic-force microscope, soon reached 1%. Established

Is it really the vacuum?

Robert Jaffe pointed out in 2005 that the Casimir force can be derived without mentioning zero-point energy at all: it is the relativistic, retarded version of the van der Waals attraction between the fluctuating charges in the two metals. His calculation shows the force depends on the fine-structure constant and vanishes if the plates’ coupling to light is turned off, which a pure vacuum-energy effect would not. So the Casimir effect is consistent with vacuum energy but does not prove it exists. Whether vacuum energy is “real”, independent of the matter it interacts with, is argued over, and it bears directly on the cosmological constant problem of Chapter 60. Debated

What you now have

Gauge freedom and gauge fixing, the photon as a quantum of the electromagnetic field with two polarizations, the Casimir force and its 1/d41/d^4 law, the experiments, and Jaffe’s caution about what it proves.

· · ·

Three questions to carry with you.

Why does the photon have only two polarizations?tap
Two of A^μ’s four components are gauge freedom; a massless spin-1 particle has only helicity ±1.
How does the Casimir pressure scale with gap?tap
As 1/d⁴: halving the gap multiplies it by 16.
Does the Casimir effect prove the vacuum has energy?tap
Not uniquely; it can be derived as a retarded van der Waals force between the plates’ charges.

And one to work out. The Casimir pressure at 1 µm is 1.30 mPa. What is it at 0.5 µm, in mPa?

F/A =mPa

Sources: H. B. G. Casimir, Proc. K. Ned. Akad. Wet. 51, 793 (1948). S. K. Lamoreaux, Phys. Rev. Lett. 78, 5 (1997). U. Mohideen and A. Roy, Phys. Rev. Lett. 81, 4549 (1998). R. L. Jaffe, Phys. Rev. D 72, 021301 (2005). Checked by code: π²ħc/(240d⁴) = 13.0 Pa at 0.1 µm, 1.30 mPa at 1 µm; ζ(−1) = −1/12, ζ(−3) = 1/120.

Next · Chapter 67
Quantum electrodynamics

One rule ("the phase can vary from place to place") forces light to exist.

Turn the page