Renormalization That ClicksEpisode 2 / Phase, or probability?

The neutron's 880 seconds is not the time God spends rolling dice ── make the mass complex and decay falls out

Phase, or probability? When exactly did the neutron decay?
Decay is the phase rotating into the complex plane, and what separates oscillation from decay is whether the destination is discrete or continuous.
Coarse-graining hides inside a single particle too.

Tools you'll need: Episode 1's coarse-graining, exponentials, complex numbers (having an i is enough) The heart of this episode: m → m − iΓ/2

A free neutron decays into a proton, an electron and an antineutrino after about 880 seconds. Textbooks say "there is a fixed probability of decaying per unit time," and from that derive the exponential law \(N(t)=N_0e^{-t/\tau}\) ── as if every neutron were rolling dice each second. But there is no "roll dice" term anywhere in quantum mechanics. There is only the rotation of phase. In fact, making the neutron's mass ever so slightly complex ── \(m\to m-i\Gamma/2\) ── makes exponential decay come out automatically. Decay is the phase you were tracking escaping into the complex plane. So why is it "never back" decay rather than "there and back" oscillation? The answer is whether the destination is discrete or continuous, and Episode 1's coarse-graining shows up there once more.

01Is 880 seconds a dice roll?

The exponential law of radioactive decay fits a probabilistic picture so comfortably that it is usually explained this way: "each particle is memoryless and decays with probability \(1/\tau\) per second, so the survivors fall off exponentially." A surviving neutron is indistinguishable from a newborn one (memorylessness). This matches experiment well.

But the basic equation of quantum mechanics (Schrödinger's) is completely deterministic. A state evolves by a rotation of phase, \(e^{-iEt/\hbar}\), and probability only enters at measurement. So why does a neutron decay "on its own" when nobody is looking?

First, a check ── "both a neutron and p+e+ν̄" is an allowed state Counterintuitively, a mid-decay neutron is a superposition: $$|\psi(t)\rangle=\alpha(t)\,|n\rangle+\beta(t)\,|p\,e^-\bar\nu_e\rangle$$ Nothing forbids this. The charge is 0 on both sides (\(0\) and \(+1-1+0\)), baryon number is 1 on both, lepton number is 0 on both. Every conserved quantity matches, so they may be superposed coherently. A state in which "it hasn't been settled yet whether this is a neutron or a proton+electron+antineutrino" exists in principle.

02Make the mass complex and decay falls out

A stable particle evolves by a phase turning at constant rate: \(e^{-imc^2t/\hbar}\). Now add a tiny imaginary part to the mass.

The heart of this episode ── decay is the phase turning into the complex plane
$$m\;\longrightarrow\;m-i\frac{\Gamma}{2c^2}\qquad\Longrightarrow\qquad e^{-imc^2t/\hbar}\;\longrightarrow\;\underbrace{e^{-imc^2t/\hbar}}_{\text{ordinary phase}}\cdot\underbrace{e^{-\Gamma t/2\hbar}}_{\text{amplitude shrinks}}$$

Probability is the square of the amplitude, so the survival probability is \(P(t)=e^{-\Gamma t/\hbar}=e^{-t/\tau}\). The exponential law came out without rolling a single die. \(\Gamma\) is called the decay width, and \(\Gamma=\hbar/\tau\). This is the standard treatment, known as the Wigner–Weisskopf approximation.

Working out Γ for the neutron

With \(\hbar=6.58\times10^{-16}\ \mathrm{eV\cdot s}\) and \(\tau\approx 879\ \mathrm{s}\):

$$\Gamma=\frac{\hbar}{\tau}=\frac{6.58\times10^{-16}}{879}\approx 7.5\times10^{-19}\ \mathrm{eV}$$

The neutron's mass-energy is 939 MeV \(=9.4\times10^{8}\) eV, so the imaginary part is \(10^{-27}\) times the real part. Making the mass just that slightly complex gives you an 880-second lifetime. Decay is a very small imaginary part of the mass.

03Two pieces of evidence that it isn't dice

You might think the two ways of speaking are equivalent. They are not ── they differ observably. A pure probabilistic process would give an exactly exponential survival curve forever; quantum mechanics departs from the exponential in two places.

Departure ① Short times ── t², not exponential (the quantum Zeno effect)

In the very first moments the survival probability falls as $$P(t)\approx 1-\left(\frac{\Delta E\,t}{\hbar}\right)^2$$ ── quadratically (an exponential would give the linear \(1-t/\tau\)). Quadratic means the slope is zero at \(t\to0\) ── so if you measure repeatedly and keep resetting the clock, decay stops. This is the quantum Zeno effect, confirmed in ion-trap experiments in 1990 (Itano et al.). If it were dice, watching would change nothing.

Departure ② Very long times ── slower than exponential (a power-law tail)

Many lifetimes later, the survival probability is known theoretically to leave the exponential and cross over to a power-law tail (Khalfin's theorem). The reason is that energy is bounded below (there is nothing underneath). A pure probabilistic process has no such reason. Catching this experimentally is extremely hard and reports are very few.

That short-time \(t^2\) is precisely the stage where the phase is only beginning to slip. The amplitude is still undecided about where to go and is leaking with its phase intact. The exponential law appears only afterwards ── once the phases have scattered. The exponential law is not a fundamental law; it emerges from coarse-graining. Same shape as Episode 1.

04What separates oscillation from decay ── discrete or continuous

Here is the main point. Two kinds of behaviour exist in the world, both of them "phase leaking."

Oscillation (there and back)Decay (never back)
Examplesneutrino oscillation, kaon mixingneutron decay, atomic emission
Destination statesdiscrete (a countable number)continuous (an infinitely dense spectrum)
Energynearly degenerate (tiny splitting)a big release (Q = 782 keV for the neutron)
Behaviourperiodic, reversibleexponential, irreversible
The phasekeeps turning in stepscatters and never lines up again

The neutron's destination is a three-body continuum. Proton, electron and antineutrino can share the energy in any proportion, so the destination states form a continuous infinity. And they carry 782 keV as they fly away. The phases scatter and never realign. If the destination had been a single discrete level at matching energy, the neutron would have oscillated back and forth with p+e+ν̄.

Order of magnitude ── the "can't tell any more" time is not 880 seconds

How long until you can no longer tell a neutron from p+e+ν̄? That is set by the energy spread \(Q\) of the destination:

$$t_{\text{distinguish}}\sim\frac{\hbar}{Q}=\frac{6.58\times10^{-16}\ \mathrm{eV\cdot s}}{7.82\times10^{5}\ \mathrm{eV}}\approx 8\times10^{-22}\ \mathrm{s}$$

The decay lifetime, meanwhile, is 879 seconds. The ratio is 10²⁴. The time for the superposition to break is 24 orders of magnitude shorter than the time to decay. The destination escapes long before a single period can complete, so you get exponential decay rather than oscillation. Those 24 orders of magnitude are what separates oscillation from decay.

05Try it ── add destinations and oscillation becomes decay

The figure below solves, on the spot, the minimal model of one initial state (the neutron) coupled to \(M\) destinations (final states). The coupling strength is auto-adjusted so that the decay rate stays the same. The only thing you move is the number of destinations \(M\).

Start at \(M=1\) (one destination) ── the survival probability swings cleanly between 0 and 1. That is oscillation. Push \(M\) up and the swings get ragged, then melt into a smooth exponential. That is decay. The underlying equation is the same unitary phase rotation throughout; not one die was rolled. The only thing that changed is how many destinations there are.

Figure: survival probability P(t) of the initial state. Increase the number of destinations M and you go from oscillation (reversible) to exponential decay (irreversible). The dashed line is e^(−Γt); the thin curve is the short-time quadratic 1−(ΔE t/ħ)². The "return" bump ── the recurrence ── runs off to the right as M grows
survival probability P(t) (exact time evolution) exponential e^(−Γt) short-time quadratic (Zeno regime)

Watch the "return" bump. When \(M\) is small, the survival probability dips and then climbs back (the phases realign; information returns). As \(M\) grows, that recurrence time runs off to the right. A real neutron has a continuous infinity of destinations, so the recurrence time is effectively at infinity. That's why it never comes back. Irreversibility just means the recurrence time is longer than your observation time ── exactly the reason the coffee and milk in Episode 1 didn't separate.

06The bottle and the beam disagree ── the neutron lifetime puzzle

This "decay = where does the amplitude leak" view formulates an experimental puzzle very cleanly. The neutron lifetime is measured two ways, and the values don't agree.

MethodWhat is countedWhat is measuredValue
Bottle (trap ultracold neutrons, count survivors)remaining neutronstotal width Γtot = the rate at which neutrons disappear877.8 ± 0.3 s
Beam (count protons emerging from neutrons in flight)protons producedpartial width Γp = the rate at which protons appear888 ± 2 s

The gap is about 10 seconds, roughly 1%. And the sign is suggestive ── more neutrons disappear than protons appear. Read plainly: "about 1% of neutrons turn into something that makes no proton." There is a second leak.

So is it "different environments change the decay rate"? ── the orders don't reach

A bottle (confined, magnetic field, slow) and a beam (free flight) are very different environments. Could the phase turn differently? The yardstick is the energy spread of the destination, \(Q=782\) keV.

EffectEnergy shiftRatio to Q
Ultracold-neutron confinement (~100 neV)10⁻⁷ eV10⁻¹³
1 T magnetic trap (μnB)6×10⁻⁸ eV10⁻¹³
Time dilation in flight (hundreds of m/s)10⁻¹²
Gravitational redshift at the Earth's surface10⁻¹⁶

The largest is 10⁻¹³. The gap to explain is 10⁻² ── eleven orders of magnitude short. The decay rate is set by MeV-scale phase space, so nudging it at the neV scale does nothing. Consistently, magnetic-trap experiments varied field, trap size and holding time and the lifetime did not move. This route is closed both theoretically and experimentally.

So what's left? Two families of explanation.

① An unknown leak (new physics). For instance a dark decay \(n\to\chi+\gamma\) with a branching ratio near 1%. Or ── exactly this episode's theme ── the neutron mixing very weakly with a mirror-sector partner \(n'\) and oscillating into it; the bottle counts that as "disappeared" while the beam, watching only protons, is largely unaffected. Sign and size both fit. However, dark decay is strongly constrained by neutron stars (such a leak softens the equation of state and caps the maximum mass near \(0.7\,M_\odot\), contradicting the observed \(2\,M_\odot\) neutron stars), and most of the mirror-neutron parameter space has been excluded by searches.

② A systematic error in the beam method. This is the current majority view. Bottle measurements agree with one another across genuinely different principles (material bottles, wall-free magnetic traps), while the beam method depends on absolute neutron-flux calibration and proton collection efficiency ── both places where a 1% systematic could hide. A higher-precision beam experiment is expected to settle it.

And yet mixing really is there ── at the quark level The idea that "decay rates are set by phase and mixing" is already realised inside the Standard Model. Neutron decay is really the quark transition \(d\to u+W^-\), and the "d quark" the weak interaction sees is a mixture of the real d, s and b (CKM mixing): $$d'=V_{ud}\,d+V_{us}\,s+V_{ub}\,b,\qquad |V_{ud}|\approx0.974$$ The neutron lifetime goes as \(|V_{ud}|^{-2}\). So a real mixing angle is already baked into the number 880 seconds. The intuition "surely this is decided by a phase" is literally correct here.
◇ ◇ ◇
The honest line ── what is established and what is hypothesis

Established: that decay follows from unitary time evolution of amplitudes; that the Wigner–Weisskopf approximation with complex energy \(E-i\Gamma/2\) gives the exponential law; \(\Gamma=\hbar/\tau\); the short-time quadratic behaviour and the quantum Zeno effect (experimentally confirmed); the long-time power-law tail (Khalfin's theorem; experimental verification is limited); that a discrete final state gives Rabi oscillation while a continuum gives irreversible decay (Friedrichs model / Fermi's golden rule); CKM mixing with \(|V_{ud}|\approx0.974\); the roughly 4σ discrepancy between bottle and beam neutron lifetimes; and that environmental effects fall orders of magnitude short.

Hypothesis / unsettled: the cause of the lifetime puzzle is not settled. New physics such as dark decay or mirror-neutron oscillation is attractive but strongly constrained by neutron-star masses and direct searches; the current mainstream view is an unidentified systematic in the beam method. The experimental values quoted here are representative values as of writing and may be updated by new measurements or reanalyses. Also, the "complex mass" is an approximation rather than an exact solution, and we are not claiming decay is exactly exponential ── on the contrary, its not being exact was the evidence that the amplitude picture, not the probability picture, is the right one.

Exercises (solvable with this episode's ideas)
  1. If a particle's lifetime \(\tau\) is 10 times longer, what happens to its decay width \(\Gamma\)?
    See the answer
    \(\Gamma=\hbar/\tau\), so it is 1/10 as large. Long-lived means narrow. Conversely, short-lived unstable particles have "blurred" masses (broad resonances).
  2. Give two reasons why neutron decay is "decay" rather than "oscillation."
    See the answer
    (1) The destination is a three-body continuum with infinitely dense states (a discrete destination could be returned from). (2) The products carry away 782 keV and fly off, so the phases never realign. The recurrence time is effectively infinite.
  3. What does each of the bottle and beam methods measure? What would a discrepancy mean?
    See the answer
    The bottle measures the total width (rate of disappearance); the beam measures the partial width for making protons. τbottle < τbeam means "disappearance is faster than proton production" ── i.e. there is another leak that makes no proton (or one of the methods has a systematic).
  4. Why does "the environment changes how the phase slips, so the lifetime changes" fail?
    See the answer
    The decay rate is set by the destination's energy spread \(Q\sim\)MeV. Confinement (neV) and magnetic fields (10⁻⁸ eV) are only \(10^{-13}\) of that ── eleven orders short of the 1% \(=10^{-2}\) to be explained. Experimentally, varying field and holding time left the lifetime unchanged.

Episode 2 summaryDecay is the phase turning into the complex plane

Make the mass just slightly complex, \(m-i\Gamma/2c^2\), and exponential decay comes out ── no dice required. For the neutron the imaginary part is \(10^{-27}\) of the real part, and that is the 880 seconds. Two pieces of evidence that the amplitude picture, not the probability picture, is right: the short-time \(t^2\) behaviour (the quantum Zeno effect, experimentally confirmed) and the long-time power-law tail. The exponential law is not fundamental; it is the result of coarse-graining.

And what separates oscillation from decay is whether the destination is discrete or continuous. For the neutron the "can no longer tell n from p+e+ν̄" time is \(\hbar/Q\approx8\times10^{-22}\) s, 24 orders of magnitude shorter than the 879-second lifetime. It cannot complete even one period, so it decays. Irreversibility is nothing but "the recurrence time exceeds the observation time" ── the same thing as the coffee and the milk. The 1% bottle-versus-beam discrepancy is neatly formulated in this framework as total-vs-partial width, i.e. is there a second leak? (Cause unsettled.)

This document is Episode 2 of the "Renormalization That Clicks" series, a reading piece for physics-loving high-schoolers and undergraduates. That the exponential law follows from unitary time evolution of an unstable state (Wigner–Weisskopf approximation, complex energy \(E-i\Gamma/2\), \(\Gamma=\hbar/\tau\)); the short-time quadratic regime and the quantum Zeno effect (experimentally confirmed by Itano et al. 1990 among others); the long-time power-law tail (Khalfin's theorem); that a discrete final state gives Rabi oscillation while a continuum gives irreversible decay (Friedrichs model, Fermi's golden rule); that the CKM element \(|V_{ud}|\approx0.974\) governs the neutron lifetime; and the roughly 4σ bottle-versus-beam discrepancy ── all of this is established physics and standard understanding. That the cause of the lifetime puzzle (new physics such as dark decay or mirror-neutron oscillation, versus a beam-method systematic) is unsettled, and that the quoted experimental values are representative as of writing, is spelled out in the body's "honest line." The figure numerically integrates, in the browser, a minimal model of one level coupled to M levels; it does not reproduce the actual phase space of neutron decay (the coupling is normalised so that the decay rate stays fixed). ── To print, use your browser's "Print" and "Save as PDF" (in the print version the slider and answers are frozen and hidden). Adjacent episodes: Episode 1, Throw it away, same answer / Episode 3, Bigger means more classical / Contents / sister series Quantum That Clicks.

Print / make a PDF: ⌘+P (Ctrl+P on Windows). On screen, moving the slider changes the number of destinations M and turns oscillation into exponential decay. The button switches the y-axis to logarithmic. "See the answer" opens each solution.