Tunneling That ClicksEpisode 7 (main-series finale) / The vacuum tunnels too

Make the thing that tunnels as large as possible ── the universe itself may not have finished falling

The vacuum tunnels too The vacuum we live in need not be the lowest state.
If it is not, it tunnels to a lower one ── with probability, again, \(e^{-S_E/\hbar}\).
And with the measured Higgs and top masses, the answer comes out barely metastable.

Tools you'll need: \(e^{-2S_E/\hbar}\) from Episode 1, the non-perturbative from Episode 6, the Higgs from "Fields That Click" The heart of this episode: the bounce, Γ/V = A e^(−S_E/ℏ)

The finale of the main series. Episode 1 wrote that "the wall is a valley in imaginary time." Now we apply that line to the largest thing there is ── the vacuum. A vacuum is a state in which a field sits at the bottom of a valley; but what if there are two valleys? The one we are in need not be the deepest. If it is not, the vacuum tunnels to the deeper one. In 1977 Coleman computed the probability ── \(\Gamma/V=Ae^{-S_E/\hbar}\), exactly the same form as everything before. And remarkably, this is not an armchair exercise. Put in the Higgs mass measured in 2012 and the top quark mass, and the Standard Model vacuum comes out neither "stable" nor "unstable" but metastable ── it does decay, but takes vastly longer than the age of the universe, a delicately balanced position indeed. We live on an exponent.

01What if the vacuum has two valleys?

"Vacuum" does not mean empty space; it means a field sitting still at the bottom of an energy valley (see the sister series "Fields That Click"). The Higgs field is one, currently settled at a value of 246 GeV.

The trouble is that there need not be only one valley. If there is another, deeper valley further out, then ours is a false vacuum.

How a false vacuum falls

Classically it cannot: there is a hill in between. But as Episode 1 showed, quantum mechanics can tunnel.
In field theory, though, the tunneling has a distinctive shape. Space does not shift all at once; instead a "bubble" of true vacuum nucleates at some point. The bubble has two competing budgets ── the energy its interior gains (proportional to volume) and the cost of building its wall (proportional to area).

$$\Delta E(R)=-\frac{4\pi}{3}R^3\varepsilon+4\pi R^2\sigma$$

Small bubbles are beaten by surface tension and collapse; large ones are won by volume and expand at nearly the speed of light. The dividing line is \(R_c=3\sigma/\varepsilon\).
"The probability of nucleating a critical bubble" is the tunneling probability, and with Coleman's bounce solution

$$\frac{\Gamma}{V}=A\,e^{-S_E/\hbar},\qquad S_E=\frac{27\pi^2\sigma^4}{2\varepsilon^3}$$

Once more, \(e^{-S_E/\hbar}\). A particle in Episode 1, a proton in Episode 2, a phase in Episode 4, and now the universe itself. One equation, working at wildly different scales.

02Put in the measured Higgs and top masses

Now the real subject. In the Standard Model, the shape of the Higgs valley changes with energy scale (running couplings, from the sister series "Renormalization That Clicks"). We can compute how \(\lambda\), the steepness of the valley, evolves toward high energy.

Who pushes \(\lambda\) down

The one-loop flow is dominated by these two terms:

$$16\pi^2\frac{d\lambda}{d\ln\mu}=\underbrace{24\lambda^2}_{\text{itself: pushes up}}\underbrace{-\,6y_t^4}_{\textbf{the top quark: pushes down}}+\cdots$$

The top quark threatens the stability of the vacuum. It is the heaviest fermion (Yukawa coupling \(y_t\approx0.94\)) and enters at the fourth power. So \(\lambda\) is pushed down hard at high energy and crosses zero somewhere. \(\lambda<0\) means the valley opens downward ── i.e. there is a lower vacuum out there.

03Try it ── where does our vacuum sit?

The figure runs the Standard Model Higgs self-coupling \(\lambda\) at one loop. The horizontal axis is energy scale (GeV, log, from \(10^2\) up to the Planck scale \(10^{19}\)).

Move the top-quark mass slider. A change of just 1 GeV moves the place where \(\lambda\) crosses zero by orders of magnitude. And with the measured values (\(m_t\approx172.5\) GeV, \(m_H\approx125.25\) GeV) it lands ── just inside the boundary between "metastable" and "stable." This uncanny positioning is called near-criticality and is one of the Standard Model's great puzzles.

Figure: the running of the Higgs self-coupling λ (one loop, Standard Model). Horizontal = energy scale (log). Where λ crosses zero is the "instability scale." The red dashed curve marks the boundary below which the vacuum would decay within the age of the universe (metastable/unstable). Changing the top mass by 1 GeV moves the answer by orders of magnitude
λ(μ) the line λ = 0 the metastable/unstable boundary

04What "metastable" means

CategoryMeaningAre we here?
stable\(\lambda\) stays positive up to the Planck scale. No lower vacuum existsjust barely not
metastablea lower vacuum exists, but falling into it takes vastly longer than the age of the universehere
unstableit would decay within the age of the universe. We could not existclearly not

Lifetime estimates vary with the calculation, but they run far beyond \(10^{100}\) years, often above \(10^{300}\). The universe is \(1.4\times10^{10}\) years old, so the comparison is meaningless.

Why there is nothing to worry about ── cosmic rays have already tried The worry that "an accelerator might trigger vacuum decay" was cleanly disposed of by Hut and Rees in 1983.
Cosmic rays have been colliding all over the universe for 13.8 billion years at energies far above anything humanity can build. The highest-energy cosmic rays observed are of order \(10^{20}\) eV ── orders of magnitude above the LHC's centre-of-mass energy. And the universe is still here. Nature has already run an enormous number of these experiments and nothing has happened ── this is the strongest safety argument there is, and it is re-confirmed with each new accelerator.

05Hawking radiation can be read as tunneling too

One more large thing. In 2000 Parikh and Wilczek derived Hawking radiation as a tunneling process.

Tunneling through the horizon

For a particle just inside a black hole, the horizon is an "impassable wall." Compute the action in the manner of Episode 1 and the escape probability is

$$\Gamma\propto e^{-2\,\mathrm{Im}S}=e^{-8\pi GME/\hbar c^3}=e^{-E/k_BT_{\rm H}}$$

── exactly the Boltzmann factor at temperature \(T_{\rm H}=\hbar c^3/8\pi GMk_B\). The Hawking temperature from the finale of "Temperature That Clicks" emerges in the language of tunneling.
And the derivation has a sequel. Emitting a particle makes the black hole lighter, so imposing exact energy conservation modifies the exponent to \(e^{-8\pi GE(M-E/2)/\hbar c^3}\). Then the spectrum departs from exactly thermal and correlations appear between emitted quanta ── which feeds the argument that information might come out after all (see the information paradox in the sister series "Black Holes That Click").

06The series' conclusion ── the scales of the world are written on the exponent

EpisodeWhat tunnelledWhat rode on the exponent
1a particle (a wall)\(2S_E/\hbar\)
2a proton (the Coulomb barrier)\(\sqrt{E_G/E}\)
3the divide between heat and tunneling\(2\pi E_b/\hbar\omega_b\)
4the phase difference (one macroscopic variable)\(\Delta U/k_BT\) and \(2S_E/\hbar\)
5(not tunneling)\(1/N(0)V\)
6the outside of perturbation theory itself\(S/g\)
7the vacuum\(S_E/\hbar\)
The series' conclusion

The "sizes" of the world are usually written on an exponent.

The Sun's lifetime, the superconducting transition temperature, the proton's mass, the vacuum's lifetime ── all of them settle in the form \(e^{-(\text{something})}\), where a change of a few tens on the exponent moves the answer by a few tens of orders. That is precisely how phenomena spanning wildly different scales can coexist in one universe. If everything were linear, the Sun would burn out instantly, the vacuum would have collapsed first, and we would not be here.

And the quantities riding on that exponent share a common origin ── classical motion in imaginary time. The wall is a valley in imaginary time (Episode 1), heat is a circle in imaginary time (Episode 3), vacuum decay is a bubble in imaginary time (this episode). What looks "impossible" in real time is usually "ordinary" in imaginary time.

On the collection's map, this series walked the non-perturbative side of the \(\hbar\) axis ── and at the end touched \(G\) (vacuum decay, Hawking) and \(k_B\) (Episode 3's \(T_0\)) as well. Somewhere near the diagonal of the Physics Cube.

◇ ◇ ◇
The honest line ── where the assumptions are

Established: the theory of false-vacuum decay (Coleman 1977, Callan–Coleman 1977) and \(\Gamma/V=Ae^{-S_E/\hbar}\), the thin-wall action \(S_E=27\pi^2\sigma^4/2\varepsilon^3\) and critical radius \(R_c=3\sigma/\varepsilon\); that the Higgs self-coupling \(\lambda\) can turn negative at high energy in the Standard Model, driven mainly by the \(-6y_t^4\) term from the top Yukawa coupling; that with the measured values the Standard Model vacuum is assessed as metastable with a lifetime exceeding the age of the universe by many orders (Degrassi et al. 2012, Buttazzo et al. 2013 and others); the cosmic-ray safety argument (Hut–Rees 1983); and the tunneling picture of Hawking radiation (Parikh–Wilczek 2000) reproducing the Hawking temperature. All standard results.

Assumptions and caveats: (1) The metastability conclusion rests on the strong assumption that the Standard Model holds unchanged all the way to the Planck scale. A single new particle changes the running and could change the conclusion. (2) The result is extremely sensitive to \(m_t\), and the systematic uncertainty in which top-mass definition is meant (pole mass, MS-bar, their relation to the Monte-Carlo mass) is the dominant uncertainty. (3) The figure runs at one loop and is simplified. Real analyses use two or three loops and place the zero-crossing of λ at around \(10^{10\text{–}11}\) GeV. Read the figure's values as order-of-magnitude guides. (4) The metastable/unstable boundary is likewise an approximate formula. (5) Bubble nucleation in curved spacetime (an expanding universe, near a black hole) is not simply the flat-space calculation, and remains debated. (6) Parikh–Wilczek is one derivation of Hawking radiation, and whether information can genuinely be extracted is unresolved. (7) Vacuum decay is not something that can be triggered artificially (the cosmic-ray argument).

Exercises (solvable with this episode's ideas)
  1. Why does vacuum decay happen as a "bubble" rather than all of space shifting at once?
    See the answer
    Moving all of space at once would require infinite energy (infinite action), sending \(e^{-S_E/\hbar}\) to zero. Only a local bubble costs a finite action. A bubble gains by volume and loses by surface, so only those exceeding the critical radius \(R_c=3\sigma/\varepsilon\) expand.
  2. Why does the top quark threaten the vacuum's stability?
    See the answer
    Because it enters the running of \(\lambda\) as the negative term \(-6y_t^4\). Being the heaviest fermion it has a large \(y_t\approx0.94\), and it enters at the fourth power. So it pushes \(\lambda\) down at high energy and drives it through zero.
  3. Why can we say there is no danger of an accelerator triggering vacuum decay?
    See the answer
    Because cosmic rays have already tried (Hut–Rees 1983). The highest-energy cosmic rays observed are of order \(10^{20}\) eV, orders of magnitude above human accelerators, and they have been colliding all over the universe for 13.8 billion years. The universe is still here ── the enormous experiment nature has already run gives the strongest bound.
  4. Throughout this series the form \(e^{-(\text{exponent})}\) recurred. What does that mean for the world?
    See the answer
    That phenomena at wildly different scales can coexist. Since a change of a few tens on the exponent moves the answer by a few tens of orders, one universe can host the Sun lasting ten billion years, superconductivity at a few kelvin, and a vacuum lasting \(10^{300}\) years. If everything were linear it would not work out this way.

Episode 7 summary / main series completeWe live on an exponent

We made the thing that tunnels as large as possible ── the vacuum. With two valleys, the one we occupy may be false, and a bubble nucleates and tunnels to the true one. The probability is \(\Gamma/V=Ae^{-S_E/\hbar}\) (Coleman 1977) ── exactly the form from Episode 1.

And this is not hypothetical. The Higgs self-coupling \(\lambda\) is pushed down at high energy by the top quark's \(-6y_t^4\) and turns negative; with the measured values the Standard Model vacuum comes out metastable ── it decays, but on a timescale of order \(10^{300}\) years. And it sits just inside the boundary with "stable," an uncannily delicate place (near-criticality). Though the result is acutely sensitive to \(m_t\) and rests on assuming no new physics up to the Planck scale.

Even Hawking radiation reads as "tunneling through the horizon," producing \(e^{-E/k_BT_{\rm H}}\). The conclusion: the "sizes" of the world are usually written on an exponent. The Sun's lifetime, the proton's mass, the vacuum's lifetime. And the quantities on that exponent share one origin ── classical motion in imaginary time. What looks impossible in real time is usually ordinary in imaginary time.

This document is Episode 7 (the main-series finale) of the "Tunneling That Clicks" series, a reading piece for physics-loving high-schoolers and undergraduates. The theory of false-vacuum decay (Coleman 1977, Callan–Coleman 1977), \(\Gamma/V=Ae^{-S_E/\hbar}\), the thin-wall action and critical radius, the running of the Higgs self-coupling and its destabilisation by the top Yukawa coupling, the assessment of the Standard Model vacuum as metastable with the measured values (Degrassi et al. 2012, Buttazzo et al. 2013 and others), the cosmic-ray safety argument (Hut–Rees 1983), and the tunneling picture of Hawking radiation (Parikh–Wilczek 2000) are all standard results. That the metastability conclusion assumes the Standard Model up to the Planck scale, that the dominant uncertainty is the systematic one in the top-mass definition, that the figure is a simplified one-loop running while real analyses use two or three loops (placing the zero-crossing of λ at around \(10^{10\text{–}11}\) GeV), that bubble nucleation in curved spacetime remains debated, that Parikh–Wilczek is one of several derivations with the information question unresolved, and that vacuum decay cannot be triggered artificially ── all spelled out in the body's "honest line." The figure numerically integrates the Standard Model's one-loop renormalization-group equations live in the browser. ── To print, use your browser's "Print" and "Save as PDF" (in the print version the sliders and answers are frozen and hidden). Previous episode: Episode 6, What perturbation theory can never see / on to the bonuses: Does cold fusion happen? / Contents / sister series Temperature That Clicks, Black Holes That Click, Fields That Click, The Physics Cube.

Print / make a PDF: ⌘+P (Ctrl+P on Windows). On screen, moving the top-quark mass by 1 GeV shifts the place where λ crosses zero by orders of magnitude. Check that the measured values sit just inside the boundary between "stable" and "metastable." "See the answer" opens each solution.