Improved constraints on ultralight axions using latest observations of the early and late Universe

TL;DR

ULAs may contribute to dark matter or act as dark energy, leaving subtle imprints across the CMB and late-time cosmic expansion. The authors implement ULAs in the axionCAMB framework with a cosine potential $V(\phi)=m_a^2 f_a^2[1-\cos(\phi/f_a)]$ (n=1) and perform a joint Bayesian analysis of Planck $2018$ CMB spectra and DESI-DR2 BAO to probe masses in $10^{-31}<m_a<10^{-27}$ eV. They validate the pipeline with mock CMB/BAO datasets and priors, reporting a 0.7% upper bound on $f_{ax}$ at $m_a=10^{-28}$ eV and finding the symmetry-breaking scale $f_a$ near, or above, the GUT scale for this mass range, with only modest shifts to the $w_0$–$w_a$ plane. The results imply that future CMB polarization and BAO measurements, plus multi-tracer datasets, can dramatically improve ULAs constraints and potentially reveal signatures such as isocurvature or photon-axion couplings. This work thus tightens the cosmological parameter space for ULAs and informs the relevance of ULA physics for dark energy and early-universe dynamics.

Abstract

Ultralight axions (ULAs) are hypothetical particles which can behave like dark matter (DM) or dark energy (DE) depending on masses generated at the symmetry-breaking scale. It remains a mystery whether the ULAs can make up a fraction of DM or DE. Although theoretical predictions indicate that the ULAs may leave distinct imprints on cosmological signals, these signatures may exist in a broad spatial and temporal scales, and may be degenerate with the known effects of the standard model. The ULA signatures are extremely subtle and the observational evidence of the ULAs remain elusive. In this work, we infer the ULA properties using both the early and late universe observations from the cosmic microwave background (CMB) and baryon acoustic oscillations (BAO). We validate modeling of the ULA effects using the CMB and BAO mock data and perform different tests to cross-check the results. By analyzing the Planck 2018 CMB measurements and the BAO measurements from the Data Release 2 of Dark Energy Spectroscopic Instrument (DESI), we constrain the energy density fraction ratio of the ULAs to total dark matter $Ω_a/Ω_d$ and obtain a new upper bound of $Ω_a/Ω_d$. Future CMB and BAO measurements will achieve unprecedented precision and will be crucial for understanding the nature of the ULAs.

Figures (6)

• Figure 1: Parameter inference from CMB and BAO mock datasets. In this test, we generate the mock data assuming an ULA fraction $\Omega_a/\Omega_d=0.01$ at a representative mass $m_a=10^{-28}$ eV. The inferred contours are consistent with the fiducial values, verifying that the Markov chain Monte Carlo (MCMC) analysis can fully recover the input cosmological parameters for the ULA masses considered in this work.
• Figure 2: (Left) 95% C.L. upper bounds on the ultralight axion (ULA) energy density fraction $f_{ax} = \Omega_a/\Omega_d$ across a range of masses. Results are shown for both ULA-only (orange) and ULA+CPL (blue) scenarios. The colored bands refer to the 95% upper bounds and the data points with error bars denote the best-fit values with 1$\sigma$ uncertainties. The upper bounds from Planck+BOSS rogersetal2023 are also shown for comparison. (Right) The inferred symmetry-breaking scale from the ULA energy density fractions. We use the $\Omega_a\hbox{-}\phi_i$ relation derived in renee2015 to estimate the initial values of the ULA field $\phi_i$ which is related to the symmetry-break scale $f_a$ by $\phi_i=\theta_i f_a$ with the initial misalignment angle $\theta_i$ being order of unity. The shaded region is the grand unified theory (GUT) energy scale between $10^{15.5}<f_a<10^{16.5}$ GeV.
• Figure 3: Reconstructed PDFs of cosmological parameters from Planck 2018 CMB and DESI-DR2 BAO data. Two models are considered in this work: the ULA-only model (left) and the ULA+CPL model (right). The red dashed lines in the right panel denote the $\Lambda$CDM predictions.
• Figure 4: Constraints on the $w_0$, $w_a$ parameters from the ULA+CPL model. The 95 % confidence levels do not show as a strong tension as the model with CPL-only. The vertical and horizontal dashed lines represent the $\Lambda$ CDM model. The tilted dot-dashed line corresponds to the constraint $1+w_0+w_a=0$ and the tilted dotted line is the constraint at $z=1$ when $w_0+(1-a)w_a=-1$, i.e., $1+w_0+w_a/2=0$.
• Figure 5: Constraints on the $H_0r_d$ and $\Omega_m$ parameters from different models. The constraint from the ULA-only model is consistent with the $\Lambda$CDM model and the main difference is driven by the inclusion of the CPL model.