Recoupled Dark Radiation reconciling CMB and DESI BAO measurements

Technical Summary: Recoupled Dark Radiation Reconciling CMB and DESI BAO Measurements

Problem Statement
Recent measurements from the Dark Energy Spectroscopic Instrument (DESI) regarding baryon acoustic oscillations (BAO), when combined with Cosmic Microwave Background (CMB) data, reveal a significant tension within the standard $\Lambda$CDM model. This discrepancy manifests primarily in the fractional matter density parameter ($\Omega_m$) and the combination of the Hubble constant and sound horizon at baryon drag ($H_0 r_d$). A critical consequence of this tension is the cosmological preference for a summed neutrino mass that is vanishingly small or even "negative," falling below the minimum values established by neutrino oscillation experiments in normal ordering. Standard models attempting to resolve this often struggle to simultaneously address the Hubble tension and the neutrino mass anomaly without conflicting with other observational constraints.

Methodology
The authors investigate a specific extension to the standard cosmological model involving a "recoupled interacting radiation" component. The theoretical framework posits sterile neutrinos ($\nu_s$) interacting via a light pseudoscalar mediator ($\phi$). The coupling is described by the Lagrangian term $\mathcal{L} \sim g \phi \bar{\nu}_4 \gamma_5 \nu_4$, where $g$ is the coupling constant and $m_\phi$ is assumed to be negligible, keeping the mediator ultra-relativistic.

The thermal history of this system proceeds through three distinct phases:

1. Partial Thermalisation: New interactions suppress the in-medium mixing angle, delaying the production of sterile neutrinos and keeping their contribution to the effective number of relativistic species ($\Delta N_{\text{eff}}$) below unity.
2. Recoupling: As the universe cools, the collisional scattering rate ($\Gamma_s \propto g^4 T$) overtakes the Hubble expansion rate ($H \propto T^2$), causing sterile neutrinos and pseudoscalars to recouple before recombination. They subsequently behave as a single, tightly coupled perfect fluid with negligible anisotropic stress.
3. Annihilation: Once sterile neutrinos become non-relativistic, they annihilate into pseudoscalars ($\nu_s \bar{\nu}_s \to \phi \phi$). This process effectively removes the massive sterile neutrinos from the cosmic neutrino background, preventing the late-time suppression of structure formation typically associated with massive free-streaming species.

The authors modified the Boltzmann solver CLASS to include this sterile neutrino plus pseudoscalar fluid. They interfaced these theoretical predictions with MontePython for Bayesian parameter estimation and used OLE for on-the-fly emulation. The analysis utilized a baseline dataset comprising primary CMB anisotropies from Planck (PR4 and PR3) and the South Pole Telescope (SPT-3G), along with CMB lensing data. These were analyzed both in isolation and in combination with DESI DR2 BAO measurements. The baseline model included six standard parameters plus two new parameters: the effective number of interacting sterile neutrinos ($N_{\nu_s, \phi}$) and the sterile neutrino mass ($m_{\nu_s}$).

Key Results

• CMB Fit: The pseudoscalar model provides a better fit to CMB data alone than $\Lambda$CDM, reducing the chi-squared value by $\Delta\chi^2 = -2.86$. It prefers a higher Hubble constant ($H_0 = 69.2 \pm 1.1$ km/s/Mpc), reducing the tension with SH0ES measurements to $2.6\sigma$.
• Combined CMB + DESI DR2: When DESI DR2 BAO data are included, the model shows a $2.7\sigma$ preference for a nonzero interacting sterile neutrino component, with a best-fit value of $N_{\nu_s, \phi} = 0.253 \pm 0.094$. The mass is constrained to $m_{\nu_s} = 5.8 \pm 2.2$ eV within the prior range.
• Resolution of Tensions:
  • Neutrino Mass: The model eliminates the preference for "negative" neutrino mass. The summed active neutrino mass remains consistent with oscillation experiments ($\sum m_\nu < 0.10$ eV at 95% CL).
  • $\Omega_m$ and $H_0 r_d$: The model favors a lower fractional matter density ($\Omega_m = 0.2991 \pm 0.0037$), significantly improving the consistency between CMB and DESI BAO constraints in the $\Omega_m$ vs. $H_0 r_d$ plane.
  • Hubble Tension: The inclusion of DESI data further reduces the Hubble tension to the $2.4\sigma$ level ($H_0 = 70.01 \pm 0.74$ km/s/Mpc).
• Goodness of Fit: The combined analysis yields a total improvement of $\Delta\chi^2 = -8.98$ relative to the standard $\Lambda$CDM model.
• Robustness: The results remain stable when Type Ia supernovae data (Pantheon+) are added. However, the inclusion of ACT DR6 primary anisotropy data reduces the preference for the model, with $\Delta\chi^2 \approx 0$, consistent with ACT's preference for $N_{\text{eff}} < 3.044$.

Significance and Claims
The paper claims that a recoupled interacting radiation component offers a viable mechanism to reconcile the emerging discrepancy between CMB and DESI BAO data. Specifically, the sterile neutrino plus pseudoscalar scenario addresses three simultaneous cosmological challenges:

1. The tension between CMB and DESI BAO measurements regarding $\Omega_m$ and $H_0 r_d$.
2. The apparent cosmological preference for a neutrino mass lower than oscillation bounds allow.
3. The Hubble tension with local distance ladder measurements.

The authors emphasize that while their specific realization focuses on sterile neutrinos, the underlying phenomenology—where a relativistic component recouples and annihilates near recombination—could be a general feature of other new light particle models. They conclude that the model's success stems from the fluid-like behavior of the recoupled component (modifying CMB acoustic peaks) followed by the annihilation of massive species (avoiding late-time structure suppression). The authors note that further testing using DESI full-shape measurements and future large-scale structure data is required to fully validate the viability of this model.