← quantum · contents

§ Phase v · Superposition & Time · Chapter 03

Quantum beats

When two quantum states share a single atom, the atom doesn't just sit there. The combined wavefunction throbs, with a tempo set by the gap between the two energies. Drop the right photon onto a cesium atom and it pulses 9,192,631,770 times every second. That pulse is what we now call a second.

The previous chapter ended with a quiet revelation. Each energy eigenstate evolves in time by twisting in the complex plane at its own private rate, e^\{-i E_n t / ℏ\}. A single eigenstate, on its own, has nothing to show for that twist. Multiply a state by an overall phase and every measurable probability comes out identical. The clock is hidden. You cannot read it. You cannot use it. It might as well not be there.

But the moment you put two eigenstates into the same atom, the situation changes completely. The two phases no longer cancel against themselves. They interfere with each other. The difference of the two frequencies (the beat between the two clocks) suddenly becomes visible in the probability of finding the system in this state or that one. The probability oscillates. You can listen to it. You can count the swings. You can build a watch out of it.

This was not obvious in 1925, and it took the better part of two decades for physicists to learn how to drive these oscillations on demand. The story moves from Isidor Rabi’s molecular beam table in New York in the late 1930s to Norman Ramsey’s separated-pulse trick in the 1950s to today’s optical lattice clocks at NIST, which keep time to one part in ten to the nineteenth. That last number means a clock that, started at the Big Bang, would today be wrong by less than a second. All of it (every digit of that absurd precision) comes from one fact: quantum systems beat at the frequency of their energy gap.

To see why energy differences are clock frequencies, start with the simplest possible superposition. Take a two-level system: one atom, two energy levels we will call |a⟩ and |b⟩, with energies E_a and E_b. Prepare it in an equal mixture:

|ψ(0)⟩ = (1/√2) ( |a⟩ + |b⟩ )

Each piece evolves with its own twist:

|ψ(t)⟩ = (1/√2) ( e^{-i E_a t / ℏ} |a⟩ + e^{-i E_b t / ℏ} |b⟩ )

Now ask the question you can actually answer: if I make a measurement, what is the probability of finding the atom in some particular state? Suppose the thing the detector responds to is not |a⟩ or |b⟩ but a third combination, say (1/√2)(|a⟩ + |b⟩) itself. Project, square the amplitude, and stand back. The cross-terms between the two energies do not cancel. The overall phase, the boring one, drops out. What survives is the difference, and the probability of detection oscillates as cos²(ΔE · t / 2ℏ + φ), with ΔE = E_b − E_a. The beat frequency is the gap divided by ℏ. That is the entire idea. Everything else is engineering.

This sounds abstract, so let us pin it down with a concrete experiment that you can almost build in your kitchen. Hit a sodium lamp with a flashbulb. Sodium has two yellow lines very close together (the famous D_1 and D_2, separated by about 0.6 nanometers in wavelength but corresponding to two excited states only 17 cm⁻¹ apart in energy). If the flashbulb pulses faster than the gap can resolve, you populate both excited states coherently. They re-emit yellow light, but the brightness of that yellow light is not constant. It pulses, at the beat frequency between D_1 and D_2. The two excited states interfere with each other on the way down, and the interference is written into the time-history of the emitted photons. This is a literal quantum beat. People measured it for the first time in the late 1960s with subnanosecond electronics, and it remains a workhorse technique in atomic and molecular spectroscopy. Two clocks, slightly out of step, beating against each other in a single atom.

The Wikipedia framing is correct as far as it goes, but it makes superposition sound like a static fact about wavefunctions. The truth is more alive. Superposition is a verb. It happens through time. The relative phase between the components winds at the rate of the energy gap, and that wind is what we measure when we say “the atom is oscillating between state a and state b.” Strictly speaking the atom is doing no such thing. The state vector is rotating in a two-dimensional complex subspace, and the projections onto your favourite basis are sliding up and down as a consequence. But the everyday language (Rabi oscillation, quantum beat, Bohr frequency) all points at the same arithmetic.

This is also where the language of frequency and energy quietly fuses. The Planck-Einstein rule says E = hν for a photon. Bohr’s atom says the photon’s energy must equal the gap between two atomic states. Combine them and the photon’s frequency, the gap divided by Planck’s constant, is also the beat frequency of any superposition of those two states. The photon and the atom are speaking the same dialect. Drop a photon at frequency ν = (E_b − E_a) / h onto an atom prepared in state |a⟩ and the atom will start oscillating back and forth between |a⟩ and |b⟩. That is the trick at the heart of every laser, every NMR scan, every MRI machine, and every superconducting qubit.

A Rabi oscillation is just a quantum beat, observed while you keep applying the driving field. Picture a single two-level atom in its ground state, bathed in a resonant radio wave. As the wave’s electric field oscillates, it couples the two states, and the population of the upper state climbs from zero, sweeps through one, and returns to zero, then climbs again. The frequency of this population swing is set by how strongly the driving field couples to the atom, the so-called Rabi frequency. If you switch the drive off after a quarter cycle, you have prepared a 50-50 superposition: a π/2 pulse. Switch off after a half cycle and you have moved the atom from purely ground to purely excited: a π pulse, the spin flip Rabi was hunting. A full cycle (2π) returns you to where you started, having driven a complete trip around the Bloch sphere.

The remarkable thing about a Rabi oscillation is what it tells you about coherence. The cosine of the probability does not soften into a flat 50-50 mixture, the way classical statistics would predict. It stays sharp. The atom remembers, at every moment, what phase it is at, because the drive and the atom share a common clock. Now imagine turning the drive off after a quarter cycle, walking away, and coming back later. The atom is in a 50-50 superposition. Its internal phase continues to advance at the gap frequency (E_b − E_a)/ℏ, but without the drive there is no external reference to compare it to. If you now turn the drive back on with a second pulse, the relative phase between drive and atom decides what happens. This is the Ramsey trick.

|a(t)⟩ = e^{-i E_a t / ℏ} |a⟩,    |b(t)⟩ = e^{-i E_b t / ℏ} |b⟩

|ψ(t)⟩ = c_a e^{-i E_a t / ℏ} |a⟩ + c_b e^{-i E_b t / ℏ} |b⟩

P_+(t) = |⟨+|ψ(t)⟩|²
       = (1/2) | c_a e^{-i E_a t / ℏ} + c_b e^{-i E_b t / ℏ} |²

P_+(t) = (1/2) | c_a + c_b e^{-i ΔE t / ℏ} |²
       = (1/2) ( c_a² + c_b² + 2 c_a c_b cos(ΔE t / ℏ) )

P_+(t) = (1/2) ( 1 + cos(ΔE t / ℏ) ) = cos²(ΔE t / 2ℏ)

Norman Ramsey’s idea, which earned him the 1989 Nobel Prize, was to replace one long pulse with two short ones, separated by a long wait. Here is the choreography. Send the atom through a π/2 pulse: now it is in a 50-50 superposition. Let it fly freely for a time T with no drive at all. During that wait the relative phase between |a⟩ and |b⟩ winds at the rate ΔE/ℏ, accumulating a total phase ΔE · T / ℏ. Then hit it with a second π/2 pulse, identical to the first. Whether the second pulse completes the rotation (so the atom is fully excited) or undoes it (so the atom is back in the ground state) depends entirely on the accumulated phase. Scan the wait T and the detected probability sweeps through a beautiful sinusoid, called the Ramsey fringe.

The reason the Ramsey method is so powerful is that the fringe spacing depends only on ΔE and the wait T, not on the details of the pulse shape. If you can make T long (a second, a minute, an hour) the fringes get arbitrarily narrow. You are no longer measuring the atomic frequency through a short blurry window; you are measuring it across a long interval, where the phase has had time to advance through millions of cycles. The accuracy improves linearly with T, limited only by how long the atom can hold its phase coherence before noise washes it out. That coherence time, in turn, depends on how well you can isolate the atom from random kicks: stray magnetic fields, photon recoils, thermal motion, collisions with the walls.

This is exactly the strategy of every modern atomic clock. In a cesium fountain, a cloud of cesium atoms is laser-cooled to near absolute zero, then tossed upward through a microwave cavity that delivers the first π/2 pulse. The atoms rise on a ballistic trajectory, fall back, and pass through the same cavity again for the second π/2 pulse. The wait T is the full flight time of about a second. The fringe spacing is therefore about one hertz around the cesium frequency of 9.192631770 gigahertz, which gives the clock an instantaneous accuracy of about one part in ten billion. Average for a few days and you reach one part in ten to the sixteenth. The current optical lattice clocks, which use a higher-frequency transition in strontium or ytterbium rather than the microwave hyperfine transition in cesium, beat at hundreds of terahertz instead of nine gigahertz, so the fringes are tens of thousands of times finer for the same T. They have demonstrated fractional accuracies of 10⁻¹⁹. A clock that good, started at the Big Bang, would today disagree with reality by less than half a second.

There is one more piece of choreography worth admiring. The optical clocks would be useless if you could not connect them to the rest of the world; their ticks come at 429 terahertz, far too fast for any electronic counter. The trick that closed the gap is the optical frequency comb, invented by Theodor Hänsch and John Hall in the late 1990s. A femtosecond laser produces a train of pulses whose Fourier spectrum is a forest of equally spaced spikes, each one phase-coherent with the others. The comb is then a ruler in the frequency domain: count how many teeth fit between an optical transition and a radio reference, and you have translated the optical frequency into a radio one a microwave counter can handle. Hänsch and Hall shared the 2005 Nobel Prize for the comb, and it is the reason we can now compare two optical clocks across continents and watch them stay in step at the eighteenth decimal place.

There is one more loose thread to tie off. Throughout this chapter we have spoken as though the energy gap ΔE is a fixed number that we are simply measuring. In real atoms it is not. The gap depends on environmental conditions: the local magnetic field (Zeeman effect), the local electric field (Stark effect), gravity (relativistic time dilation), the velocity of the atom (Doppler). Modern clocks are so good that they detect gravitational time dilation across a height difference of a single centimetre on the lab bench. Lift the apparatus by a finger’s width and the upper atoms tick faster than the lower ones by about one part in ten to the seventeenth, exactly as Einstein’s general relativity predicts. The beat between two energy levels is not just an abstract tempo. It is a clock, and it knows where you stand on the planet.

This is the deep lesson of the chapter. The Schrödinger equation makes every energy difference a frequency. Every frequency is a beat. Every beat can be counted. Counting beats is what humans have done since they first looked at the sun and called the gap between sunrises a “day.” Quantum mechanics has done nothing more, and nothing less, than push that ancient act of counting down to the level of single atoms. The atomic age, in the most literal possible sense, is the age in which we stopped using the Earth to keep time and started using the trembling phase between two electrons in a single ion. And the trembling, of course, is just the wavefunction doing what it has always done: rotating in the complex plane at the rate of its energy, in lockstep with every other state that shares the atom.