Constraints on Large-Scale Dark Acoustic Oscillations from Cosmology

TL;DR

This work investigates whether a fraction of dark matter could remain coupled to a bath of dark radiation in the early universe, producing dark acoustic oscillations (DAO) that imprint a characteristic sound horizon on the matter power spectrum. Adopting an atomic dark matter (dark atoms) model, the authors parameterize the interacting DM with $f_{ m int}$, the dark-to-visible temperature ratio $\xi$, and the DAO strength proxy $\Sigma_{ m DAO}$, and solve the perturbation equations for DM, DR, and standard components using a modified CAMB. By fitting Planck+CMB lensing, BAO, and BOSS DR9 data, they derive stringent constraints: for strong DAO ($\Sigma_{ m DAO}\gtrsim 10^{-3}$), the DR fraction must be tiny ($\xi\lesssim 0.2$) and the interacting DM fraction must be $f_{ m int}\lesssim 5\%$, with a transition near $\Sigma_{ m DAO} \sim 10^{-4.5}$ where DR behaves more like extra relativistic species. They translate these bounds into a DAO sound horizon limit and show that most DM must behave as standard CDM on large scales, though a narrow region around $\xi\approx0.5$, $\Sigma_{ m DAO}\approx10^{-2.5}$, and $f_{ m int}\approx2\%$ remains mildly compatible with data. They also assess the impact on galaxy formation via double-disk DM, finding that cosmological data largely exclude robust double-disk regions, though tiny interacting fractions at specific DM parameters could still be viable. Overall, the study demonstrates how precision cosmology places powerful, general constraints on hidden DM–DR interactions and highlights the complementarity between large-scale cosmology and galactic-scale physics for probing dark sectors.

Abstract

If all or a fraction of the dark matter (DM) were coupled to a bath of dark radiation (DR) in the early Universe we expect the combined DM-DR system to give rise to acoustic oscillations of the dark matter until it decouples from the DR. Much like the standard baryon acoustic oscillations, these dark acoustic oscillations (DAO) imprint a characteristic scale, the sound horizon of dark matter, on the matter power spectrum. We compute in detail how the microphysics of the DM-DR interaction affects the clustering of matter in the Universe and show that the DAO physics also gives rise to unique signatures in the temperature and polarization spectra of the cosmic microwave background (CMB). We use cosmological data from the CMB, baryon acoustic oscillations (BAO), and large-scale structure to constrain the possible fraction of interacting DM as well as the strength of its interaction with DR. Like nearly all knowledge we have gleaned about dark matter since inferring its existence this constraint rests on the betrayal by gravity of the location of otherwise invisible dark matter. Although our results can be straightforwardly applied to a broad class of models that couple dark matter particles to various light relativistic species, in order to make quantitative predictions, we model the interacting component as dark atoms coupled to a bath of dark photons. We find that linear cosmological data and CMB lensing put strong constraints on existence of DAO features in the CMB and the large-scale structure of the Universe. Interestingly, we find that at most ~5% of all DM can be very strongly interacting with DR. We show that our results are surprisingly constraining for the recently proposed Double-disk DM model, a novel example of how large-scale precision cosmological data can be used to constrain galactic physics and sub-galactic structure.

Figures (16)

• Figure 1: Comoving DAO scale as a function of the parameter $\Sigma_{\rm DAO}$ for strongly-coupled atomic DM models ($\alpha_D>0.025$). In the upper panel, we fix $\xi=0.5$ and vary the fraction of interacting DM. In the lower panel, we fix $f_{\rm int}=5\%$ and let $\xi$ vary. Here, take $H_0=69.57$ km/s/Mpc, $\Omega_{\rm m}=0.3048$, $\Omega_{\rm DM}h^2=0.1198$, and $N_{\rm eff}=3.046$.
• Figure 2: Angle averaged galaxy correlation function $\tilde{\xi}_0(r)$ for different PIDM models. In the upper panel, we take $f_{\rm int}=5\%$, $\xi=0.5$ and vary $\Sigma_{\rm DAO}$ and $\alpha_D$. In the lower panel, we fix $\Sigma_{\rm DAO}=10^{-3}$, $\alpha_D=0.01$ and $\xi=0.5$, but let the fraction of interacting DM vary. We set the galaxy bias to $b=2.2$ and the dilation scale to $\alpha=1.016$. We compare theoretical predictions with BOSS-DR9 measurements from Ref. Anderson:2012ve, and we also show a standard $\Lambda$CDM model with an equivalent number of effective neutrinos. In this work, we focus uniquely on linear scales, which lie to the right of the dashed vertical line on the plot.
• Figure 3: Linear galaxy power spectra for different PIDM models. In the upper panel, we fix $f_{\rm int}=5\%$, $\xi=0.5$ and vary $\Sigma_{\rm DAO}$. The lower panel uses $\Sigma_{\rm DAO}=10^{-3}$ and $\xi=0.5$ but let the fraction of interacting DM vary. To compare with galaxy power spectrum from the CMASS catalogue, we have convolved our linear matter power spectra with the BOSS window function and multiplied the results by a scale-independent galaxy bias $b=2.01$ (see section \ref{['sec:pk_data']} for more details). For comparison, we also show a standard $\Lambda$CDM model with an equivalent number of effective neutrinos. In this work, we focus uniquely on linear scales, which lie to the left of the dashed vertical line on the plot.
• Figure 4: Upper panel: Time-evolution of the gravitational potential $\psi$ for different values of $\Sigma_{\rm DAO}$. Here, $\psi$ is given in units of $2\zeta/3$, where $\zeta$ is the curvature perturbation on constant density hypersurfaces, which is conserved on super-horizon scale for pure adiabatic fluctuations. Before horizon entry, we have $\psi\rightarrow (2\zeta/3)(1+\frac{4}{15}R_{\rm f-s})^{-1}$, where $R_{\rm f-s}$ is the free-streaming fraction of the total radiation content of the Universe. Lower panel: Monopole source term for the CMB temperature anisotropies. For both panels, we have taken $k = 0.15$ Mpc$^{-1}$, $\xi=0.5$, and $f_{\rm int}=100\%$. For comparison, we also show a standard $\Lambda$CDM model with an equivalent number of effective neutrinos.
• Figure 5: Difference between PIDM models and a $\Lambda$CDM model with an equivalent number of neutrinos for the pure ISW contribution to $C_l^{\rm TT}$, $\Delta C_l^{\rm ISW} = C_l^{\rm ISW,PIDM}-C_l^{\rm ISW,CDM}$. In the upper panel, we take $\xi=0.5$, $f_{\rm int}=100\%$, and vary $\Sigma_{\rm DAO}$. In the lower panel, we take $\xi=0.5$, $\Sigma_{\rm DAO}=10^{-2}$, but let $f_{\rm int}$ vary.