Gravitational Waves from Confinement in $SU(N)$ Yang-Mills Theory

We provide a detailed analysis of the gravitational wave spectrum of $SU(N)$ pure Yang-Mills theory. The confinement phase transition is described with an effective Polyakov loop model, using the latest lattice data as an input. In particular, recent lattice studies clarified the large-$N$ scaling of the surface tension, which we incorporate through a modification of the kinetic term. We demonstrate that the thin-wall approximation agrees with the Polyakov loop model at small $N$ while it breaks down at large $N$. Furthermore, we include reliable estimates of the bubble wall velocity using a recently developed framework based on a large enthalpy jump at the phase transition. Altogether, this allows us to derive the gravitational wave signals for all $SU(N)$ confinement phase transitions and clarifies the behaviour at large $N$. The strongest signal arises for $N=20$, but overall the predicted signals remain rather weak. Our work paves the way for future studies of other gauge groups and systems with fermions.

💡 Research Summary

This paper presents a comprehensive study of the stochastic gravitational‑wave (GW) background generated by the confinement (deconfinement) first‑order phase transition in pure SU(N) Yang‑Mills theory. The authors adopt an effective Polyakov‑loop model (PLM) as the thermodynamic description of the transition, but crucially modify its kinetic term so that the model reproduces the most recent lattice results for the interface (surface) tension, which scale as σ ∝ N² at large N. By fitting the PLM parameters to lattice data for N = 3–20, they obtain a unified effective potential that accurately captures the latent heat, critical temperature, and surface tension across the whole range of N.

The paper then compares this PLM‑based approach with the traditional thin‑wall approximation, which relies only on σ and the latent heat L. While the thin‑wall method works reasonably well for small N (where the nucleation temperature Tₙ lies close to the critical temperature T_c), it dramatically overestimates supercooling at large N because it assumes the Euclidean action scales as (T − T_c)⁻². In the PLM, however, the barrier between the false and true vacua disappears already near T_c, limiting the amount of supercooling and rendering the thin‑wall approximation invalid for N ≳ 8.

A key novelty of the work is the treatment of the bubble wall velocity v_w, a non‑equilibrium quantity that is difficult to extract from lattice simulations. The authors employ a recently developed “large enthalpy jump” framework, which predicts that when the number of relativistic degrees of freedom changes sharply at the transition, the wall accelerates to relativistic speeds. Using the lattice‑determined latent heat, they estimate v_w≈c for all N, and compute the inverse duration parameter β/H from the temperature dependence of the nucleation rate obtained from the PLM.

With the four standard GW parameters (α, β/H, Tₙ, v_w) in hand, the authors calculate the GW spectrum using state‑of‑the‑art numerical fits to hydrodynamic simulations of bubble collisions, sound‑wave contributions, and magnetohydrodynamic turbulence. Their results show that the peak amplitude h²Ω_peak reaches a maximum for N ≈ 20, where h²Ω_peak ≈ 10⁻¹⁴ at a frequency of a few × 10⁻³ Hz. For larger N the signal falls off as h²Ω_peak ∝ N^{−14/3}, becoming rapidly undetectable. Even at the optimal N = 20 the predicted signal lies well below the projected sensitivities of planned space‑based detectors such as LISA, DECIGO, and BBO.

The authors provide a detailed error analysis, propagating uncertainties from the lattice inputs through the PLM fits and the GW parameter extraction. They also discuss the physical origin of the weak signal: the strong coupling of the dark sector leads to a rapidly changing effective potential near T_c, which shortens the phase‑transition duration (large β/H) and suppresses GW production.

In the concluding section, the paper emphasizes that the PLM, calibrated to lattice data, offers a reliable tool for studying confinement transitions in a wide class of gauge theories. The methodology can be straightforwardly extended to other gauge groups (SO(N), Sp(N), exceptional groups) and to theories with fermions in various representations, where the order‑parameter dynamics may differ and potentially yield stronger GW signals. Nonetheless, for pure SU(N) Yang‑Mills dark sectors, the study demonstrates that the stochastic GW background is intrinsically faint and unlikely to be observed with any foreseeable detector.