Constraining the axiverse with reionization

TL;DR

This work investigates how a spectrum of decaying axions from Type IIB string theory—produced via freeze-in and decaying to photons—can ionize the early universe and alter the CMB optical depth. By computing the full reionization history for ensembles of axions and comparing to a model-independent Planck posterior on high-$z$ ionization, the authors constrain the reheating temperature and thus portions of the string axiverse. They analyze CY-based models with $h^{1,1}=20,50,100$, finding that a non-negligible fraction of models prefer low $T_{\rm reh}$ (around $10^{10}\,\mathrm{GeV}$) at 95% CL, and up to about a quarter of models can be excluded at high $T_{\rm reh}$ depending on $h^{1,1}$. The results, together with the publicly released code, offer a path to integrating multi-axion decays with other observables to further constrain the axiverse and the reheating history of the early universe.

Abstract

Axions that couple to electromagnetism are produced in the early Universe by, among other channels, freeze-in via the Primakoff process. For sufficiently large axion masses, the same coupling causes the axions to decay into two photons, which subsequently ionize the intergalactic medium. If this decay occurs in the redshift range $20 \lesssim z \lesssim 1100$, then the contribution to the cosmic microwave background optical depth $τ_{\rm reio}$ can lead to a conflict with observations, excluding models with sufficiently strongly coupled, heavy axions and high reheating temperatures, $T_{\rm reh}$. Using large ensembles of explicit type IIB string theory models with up to $h^{1,1} = 100$ axions, we compute the full cosmic reionization history caused by the decays of multiple axions. We compare this to the posterior on the high-$z$ component of $τ_{\rm reio}$ derived from model-independent constraints on the ionization state of the Universe, obtained in a full \textit{Planck} analysis presented in a companion paper. For $h^{1,1} = 20, 50, 100$, we find that approximately 15\%, 15\%, and 10\% of the models in the ensemble prefer $T_{\rm reh} \lesssim 10^{10}\,\text{GeV}$ at 95\% CL. We provide a publicly available code at:~\href{https://github.com/ZiwenYin/Reionization-with-multi-axions-decay}{github.com/ZiwenYin/Reionization-with-multi-axions-decay}, which computes the reionization history for arbitrary ensembles of decaying axions. Our analysis opens the door for future large-scale work studying the preference for low-temperature reheating in models with multiple axions.

Figures (12)

• Figure 1: Left: Axion production by the Primakoff process. Right: Axion to two photon decay.
• Figure 2: Single axion case. Ionization fraction in the $\Lambda$CDM scenario (black line), compared to a model with one additional axion component, assuming instantaneous energy injection with the "beyond on-the-spot" assumption and using the full energy deposition efficiency function provided by DarkHistory. The axion, constituting a fraction $\approx 10^{-7}$ of the DM, decays at redshift $z_{\rm decay}<1100$, leading to efficient ionization of the intergalactic medium and a CMB optical depth $\tau_{\rm reio} \approx 1.92$.
• Figure 3: The energy deposition rates from representative models with different numbers of axions in the mass window relevant for decay and ionization. Each curve corresponds to a randomly selected model from the $h^{1,1} = 50$ ensemble of Ref. Gendler:2023kjt. The black line is, for reference, the energy injection rate and redshift range of Population III core collapse supernovae (Pop-III CCSNe) Hartwig:2022lon.
• Figure 4: Multiple axions case. Ionization history for a specific string axion model in an ensemble with $h^{1,1}=50$, containing two axions within the relevant mass window. The axion parameters are: $m_{a}^{(1)}=8.27$ GeV, $g_{a\gamma\gamma}^{(1)}=1.87\times 10^{-20}\,\mathrm{GeV}^{-1}$, and $m_{a}^{(2)}=0.72\,\mathrm{GeV}$, $g_{a\gamma\gamma}^{(2)}=2.1\times 10^{-20}\,\mathrm{GeV}^{-1}$. Different line styles correspond to the reheat temperatures in the legend.
• Figure 5: Normalized posterior distribution, $P/P_{\mathrm{max}}$, for the high–redshift contribution to the optical depth, for the critical redshift $z_c=30$ defining the separation in Eq. \ref{['eq:tau_lowhighz']}. Results are based on the Planck low-$\ell$ EE dataset alone and the analysis in Paper I. This posterior distribution gives the upper limit $\tau_{\mathrm{highz}} < 0.108$ at 95% CL, which we use to set limits on axion models.