The Cosmic Domain-Wall Problem

What a Domain Wall Is: The Physical Basis

A domain wall is a sheet of concentrated field energy separating two regions of space that sit in different but energetically equivalent vacuum states. The textbook example is a real scalar field φ with a double-well ('Mexican hat' in one dimension) potential V(φ) = λ(φ² − v²)²/4, symmetric under the discrete reflection φ → −φ. This Z₂ symmetry has two degenerate minima at φ = +v and φ = −v.

When the universe cools below the critical temperature, causally disconnected patches independently 'choose' either +v or −v — a process called the Kibble mechanism. Where a +v region abuts a −v region, the field must pass through φ = 0 (the top of the potential barrier), trapping energy in a thin transition layer. That layer is the wall.

• Topological requirement: walls form if and only if the vacuum manifold is disconnected, i.e. the zeroth homotopy group π₀(M) is nontrivial.
• Wall thickness δ ~ 1/(√λ · v) ~ m_φ⁻¹, set by the inverse mass of the field.
• Surface tension (energy per unit area) σ ~ √λ · v³, the single number that governs everything.

The Mechanism: Why Walls Overclose the Universe

The danger lies in how wall energy dilutes as space expands. A domain-wall network reaches a scaling regime: through reconnection, self-intersection, and straightening, the network keeps roughly one wall of Hubble-scale curvature per horizon volume at all times. Numerical lattice simulations confirm this 'one wall per Hubble volume' attractor.

The energy in that one wall is (tension) × (area) ~ σ × (1/H)², spread over a Hubble volume ~ (1/H)³. So the wall energy density is:

• ρ_wall ~ σ · H ~ σ / t (using H ~ 1/t in a decelerating universe).

Compare this to the backgrounds: radiation dilutes as ρ_rad ∝ a⁻⁴ ∝ t⁻² and matter as ρ_matter ∝ a⁻³ ∝ t⁻³/². Walls fall off only as t⁻¹ — the slowest of all. Whatever fraction of the energy budget the walls hold, that fraction grows without bound. Sooner or later ρ_wall overtakes everything, the universe becomes wall-dominated, expansion turns anisotropic and power-law, and the smooth Friedmann cosmology we observe is destroyed. That is the overclosure catastrophe.

Key Numbers: The Zeldovich Bound and a Worked Example

How light must walls be to avoid disaster? Set the requirement that walls have not yet dominated today and do not distort the CMB. Because ρ_wall ~ σ·H, and today H₀ ~ 10⁻³³ eV, the wall density stays below the critical density ρ_crit ~ 10⁻¹¹ eV⁴ only if σ is tiny.

• Requiring ρ_wall/ρ_crit ≲ 1 today gives the classic Zel'dovich bound: σ^(1/3) ≲ ~1 MeV, i.e. the symmetry-breaking scale v of a wall-forming transition must be below about 1 MeV.
• Sharper CMB isotropy limits push this further: a persistent wall network must contribute ≲ 10⁻⁵ of the critical density, since the induced quadrupole would otherwise exceed the observed C₂ ~ 10² μK².

Worked example — the electroweak scale: If a discrete symmetry broke at v ~ 100 GeV (with λ ~ 1), then σ ~ v³ ~ 10⁶ GeV³. That is roughly 10¹⁵ times above the Zel'dovich σ^(1/3) ~ 1 MeV limit — such walls would have overclosed the universe by a factor of order 10⁴⁵ in energy density. Stable walls at any 'interesting' particle-physics scale are therefore flatly excluded.

How It Shows Up: Signatures and Constraints

Because truly stable heavy walls are forbidden, the observational program focuses on near-miss networks — walls that decay just before dominating — and on the constraints ruling out the rest.

• CMB anisotropy: A wall crossing the last-scattering surface imprints a large-angle temperature step and boosts the low-ℓ multipoles. Planck's measured quadrupole (C₂ ≲ ~800 μK²) and octopole cap any wall abundance at ≲ 10⁻⁵ of ρ_crit — the tightest late-time bound.
• Gravitational waves: A collapsing/annihilating wall network radiates a stochastic gravitational-wave background peaked at the Hubble frequency at decay. For walls decaying near the QCD epoch this lands in the nanohertz band now probed by pulsar timing arrays (NANOGrav, EPTA); the 2023 PTA signal has been fit by domain-wall models among others.
• Big Bang nucleosynthesis: Walls decaying after ~1 second would inject entropy and disturb the primordial ²H/⁴He abundances, giving an independent lower bound on how early they must vanish.
• Cosmic birefringence: Axion-like walls can rotate CMB polarization, a hint tentatively seen in Planck data.

Related Regimes: Walls, Strings, Monopoles, and Escape Routes

Domain walls are one of a trio of cosmological topological defects, distinguished by the topology of the broken vacuum. Strings (π₁ ≠ 0) scale as ρ ∝ t⁻² and are cosmologically tolerable; monopoles (π₂ ≠ 0) behave like matter (ρ ∝ a⁻³) and give the 'monopole problem'. Walls are the most dangerous because they dilute slowest.

The standard escape routes are:

• Inflation: If the symmetry breaks before inflation, the walls are stretched to sizes far beyond the horizon, leaving at most a fraction of one wall in our universe. (Inflation similarly cures the monopole problem.)
• Biased vacua: Add a small explicit breaking (a 'bias' energy ε lifting the degeneracy). The true-vacuum regions grow and eat the false ones, collapsing the network before it dominates — this is the leading fix for the QCD-axion wall problem (N_DW > 1).
• Symmetry restoration or non-formation: Keep the discrete symmetry unbroken today, or make the transition never trap walls.

Significance and Open Questions

The domain-wall problem is a cornerstone constraint on beyond-the-Standard-Model physics: any new theory that spontaneously breaks a discrete symmetry in the early universe — spontaneous CP violation, Peccei–Quinn axion models with domain-wall number N_DW > 1, discrete flavor symmetries, left-right and grand-unified models — must build in a way to remove the walls. It is, alongside the monopole problem, one of the historic motivations for cosmic inflation.

The most famous live case is the QCD axion domain-wall problem: when the Peccei–Quinn symmetry breaks after inflation with N_DW > 1, a stable wall network forms at the QCD scale (~150 MeV) and would overclose the universe. Resolving it requires either N_DW = 1 or a carefully tuned bias.

Open questions include: the precise scaling exponent and gravitational-wave spectrum of realistic networks (still debated in lattice studies); whether the PTA nanohertz signal is partly walls rather than merging supermassive black holes; and whether collapsing walls seed primordial black holes. Whether any relic wall network could act as dark energy or address the Hubble tension remains speculative but actively studied.

The three cosmological topological defects, set by the topology of the broken vacuum, and their cosmological danger

Defect	Dimension / topology	Energy density scaling	Cosmological status
Domain wall	2D surface, π₀(M) ≠ 0 (disconnected vacua)	ρ ∝ 1/t (∝ a⁻¹); dilutes slowest	Catastrophic — overcloses universe unless removed
Cosmic string	1D line, π₁(M) ≠ 0 (non-simply-connected)	ρ ∝ 1/t² (∝ a⁻²), scaling network	Tolerable — subdominant, Gμ ≲ 10⁻¹¹ from PTAs
Monopole	0D point, π₂(M) ≠ 0 (S² vacua)	ρ ∝ a⁻³ (like matter), no dilution by motion	Problematic — 'monopole problem', diluted by inflation
Texture	unstable, π₃(M) ≠ 0	collapses and radiates away	Harmless — decays