Signatures of inhomogeneous dark matter annihilation on 21-cm

The annihilation events of dark matter (DM) can produce secondary particles that can potentially be detected through astrophysical probes Slatyer (2022); Hooper (2019). Cascades of these secondary particles can lead to extra ionization and heating of the intergalactic medium (IGM). The ionization effect enhances the scatters between the cosmic microwave background (CMB) photon on free electrons and photons, leading to observable effects in the CMB anisotropy measurements Aghanim et al. (2020); Padmanabhan and Finkbeiner (2005); Slatyer (2013, 2016a, 2016b); Liu et al. (2016). The latest observation of Planck Aghanim et al. (2020) yielded DM annihilation constraints that are competitive to those from high-energy cosmic ray searches Slatyer (2022); Hooper (2019); Ahnen et al. (2016); Bergstrom et al. (2013); Giesen et al. (2015). The heating effect from DM annihilation, on the other hand, can be efficiently probed by 21-cm signal from neutral hydrogen.

The 21-cm signal arises from the transition between the neutral hydrogen’s singlet and triplet states, which offers an invaluable glimpse into the cosmic dark ages and the Epoch of Reionization (EoR) Pritchard and Loeb (2012). During these epochs the formation of DM halo structures is expected to significantly enhance DM annihilation rate. The 21-cm signal strength is sensitive to the thermal and ionization conditions in the IGM, thereby providing a unique avenue to detect the possible heating and ionization induced by DM annihilation. The primary 21-cm observation window below redshift $20$ lies deep in the nonlinear structure growth epoch, during which spatial inhomogeneity is expected to be present in the DM halo-boosted heating and ionizing sources. This can in turn lead to an enhancement in the 21-cm power spectrum at scales where the spatial inhomogeneity manifests itself.

In this paper, we examine the effects of inhomogeneity in DM distribution on 21-cm power spectrum. Our results show that the formation of DM halo structures induces inhomogeneity in DM annihilation rate that closely traces density fluctuations. Assuming that DM annihilation products have a short absorption length, which is generally feasible for relatively light-dark matter and the resultant radiation is at low energy, we find that the DM-induced heating and ionization exhibit distinctive inhomogeneous structures, which further enhances spatial fluctuation and power spectrum for 21-cm signal. Such features can be particularly helpful for discrimination of possible DM signatures from complex astrophysical background. For an annihilation rate of $\left<\sigma v\right>/m_{\chi}\sim 10^{-27}{\rm cm^{3}s^{-1}GeV^{-1}}$, which is roughly the same level as that constrained by cosmic ray and CMB Aghanim et al. (2020); Slatyer (2022); Hooper (2019); Ahnen et al. (2016), the corresponding 21-cm power spectrum can be easily detected by the Square Kilometer Array (SKA) telescope Sitwell et al. (2014).

This paper is organized as follows: Sec. II briefly reviews the basics of cosmic 21-cm signal, in Sec. III we describe our model for inhomogeneous DM annihilation, the 21-cm power spectrum results are presented in Sec. IV and we conclude in Sec. V.

The hydrogen 21-cm signal arises from the hyperfine energy split between singlet and triplet states of neutral hydrogen. In cosmological context, the strength of this signal is measured by 21-cm brightness temperature $T_{21}$ Mesinger et al. (2011); Pritchard and Loeb (2012),

where $z$ is redshift, $x_{\rm HI}$ is the neutral fraction of the IGM, $\delta_{\rm b}$ is baryon density contrast, $H(z)$ is the Hubble parameter, ${\rm d}v_{\rm r}/{\rm d}r$ is velocity gradient along the line of sight, $T_{\gamma}$ indicates the temperature of the radiation background, which is commonly assumed to be CMB temperature. $\Omega_{\rm m}$ and $\Omega_{\rm b}$ are fractional densities for matter and baryon respectively (therefore fractional density of cold dark matter is $\Omega_{\rm c}=\Omega_{\rm m}-\Omega_{\rm b}$), $h$ is the Hubble constant in the unit of $100\,{\rm km/s/Mpc}$. The spin temperature $T_{\rm s}$ quantifies the number density ratio of hydrogen atoms in singlet and triplet states and is coupled to both radiation temperature $T_{\gamma}$ and gas kinetic temperature $T_{\rm k}$ through collisional coupling and Wouthuysen-Field effect Furlanetto et al. (2006); Mesinger et al. (2011)

where the color temperature $T_{\alpha}$ is closely coupled to $T_{\rm k}$, $x_{\alpha}$ and $x_{\rm c}$ are coefficients for collisional and Wouthuysen-Field coupling (see Ref. Pritchard and Loeb (2012)).

As can be seen from Eq. (1), spatial fluctuations of ionization, density, gas temperature and velocity gradient means that $T_{21}$ is inherently inhomogeneous, therefore in addition to the global average $\bar{T}_{21}$, 21-cm signal is also characterized by its power spectrum, which is generally defined as follows,

where $\phi$ denotes the physical quantity under consideration, which can take values of $T_{21}$, density field $\delta$ and boost factor $B$ (see next section) in the context of this work. The brackets $\left<\right>$ denotes the ensemble average, $\tilde{\phi}({\vec{k}},z)$ refers to the Fourier transform of $\phi({\vec{x}},z)$, $\delta_{\rm D}$ is the three-dimensional Dirac function, and $k=|\vec{k}|$. We compute all power spectrum in our analysis using the powerbox package Murray (2018).

Equations (1, 2) show that 21-cm signal is encoded with information about the thermal and ionization states of IGM ($T_{\rm k},\ x_{\rm HI}$). For the epochs of interests ($5\lesssim z\lesssim 40$), IGM is affected by energy injection from annihilating DM as well as the radiation from the first galaxies, therefore discrimination of possible DM signal using $T_{21}$ requires thorough knowledge about the astrophysical background. Our calculations are built primarily on the 21cmFAST code Mesinger et al. (2011); Park et al. (2019), which is a fast semi-analytic package for simulating both the density field and astrophysical radiation.

Throughout this work, we adopt a $\Lambda$CDM cosmology with Planck 2018 parameters Aghanim et al. (2020). For the background astrophysics, we adopt the default 21cmFAST setting, which has been detailly presented in Ref. Park et al. (2019). Note that these setting has been shown to be consistent with measurements of UV luminosity function Bouwens et al. (2015a, b); Oesch et al. (2018), optical depth Adam et al. (2016) and reionization timing McGreer et al. (2015). For numerical processes, we make use of the 21cmFAST framework, and interested readers can refer to Park et al. (2019); Mesinger et al. (2011) for review and program details.

Assuming that DM particles $\chi$ annihilate through $s$-wave with a thermally averaged cross-section $\left<\sigma v\right>$, the energy injected per unit volume and time (referred to as injection rate hereafter for convenience) can be written as

where $m_{\chi}$ denotes DM mass, $\rho_{\rm c}$ is DM density, $n_{\chi}=\rho_{\rm c}/m_{\chi}$ is the number density of DM particle, and $\left<\sigma v\right>n^{2}_{\chi}$ is the number of DM annihilation events per unit volume and time, $g$ is a symmetry factor which we take $1/2$ following Aghanim et al. (2020). For homogeneous distribution, using $\rho_{\rm c}=\bar{\rho}_{\rm c}=\Omega_{{\rm c}}\rho_{\rm cr}(1+z)^{3}$, where $\rho_{\rm cr}$ is current critical density, Eq. (5) can be expressed as,

where the subscript HMG denotes homogeneous distribution.

As can be seen from Eq. (5), the global injection rate of DM is proportional to $\bar{\rho^{2}_{\rm c}}$, which is simply $\bar{\rho_{\rm c}}^{2}=\Omega^{2}_{\rm{c}}\rho^{2}_{{\rm cr}}(1+z)^{6}$ for homogeneous distribution. As matter overdensity grows at lower redshifts ($z\leq 50$), the homogeneity assumption in Eq. (6) is no longer valid, and $\bar{\rho^{2}_{\rm c}}$ can be enhanced above $\bar{\rho_{\rm c}}^{2}$ by orders of magnitude. As the result, the overall DM injection rate can also be significantly enhanced. Therefore at low redshifts we model DM injection by combining the contribution from both collapsed halos and that from un-collapsed regions.

We consider the following 3 different simulation settings,

• •

  Inhomogeneous Boost (IHM): Our main simulation assumes inhomogeneous injection boost factor detailed in Sec. III and the background astrophysics described in Sec. II. We set DM annihilation rate to $\left<\sigma v\right>/m_{\chi}=10^{-27}{\rm cm^{3}s^{-1}GeV^{-1}}$, which roughly corresponds to the current CMB constraints from Planck Aghanim et al. (2020). This gives a reionization optical depth of $\tau_{\rm rei}=0.073$ which is also in agreement with Planck Aghanim et al. (2020); Muñoz et al. (2022).

• •

  Homogeneous Boost (HMG): Similar to IHM case but instead of the inhomogeneous boost factor, we use its global average value $\bar{B}$. Note that all global quantities in this simulation are identical to that in IHM scenario.

• •

  Fiducial: Simulation for the fiducial astrophysical background detailed in Sec. II in absence of DM injection ($\left<\sigma v\right>/m_{\chi}=0$).

All our simulations are performed with $300^{3}$ resolution and a box length of 500 comoving Mpc. Note that if DM annihilation products have a long mean free path before absorption, the heating and ionization rates will remain widely dispersed despite the inhomogeneity in injection rate. The HMG simulation can be seen as a representation of this scenario, and to a large extent aligns with the situation in previous analyses in Taylor and Silk (2003); Liu et al. (2016); Lopez-Honorez et al. (2016); Short et al. (2020); Huetsi et al. (2009); Poulin et al. (2015); Diamanti et al. (2014); Natarajan and Schwarz (2010); Valdes et al. (2013). In contrast, as represented by the IHM simulation, if the annihilation products have a short absorption length, which is typically the case if they are injected below electroweak energy scales or if they consist mainly of electrons Slatyer (2013); Slatyer et al. (2009); Slatyer (2016b), the injected energy will be deposited locally, and the corresponding ionization and heating rates would be inhomogeneous.

Fig. 1 shows the inhomogeneous lightcone evolution for the boost factor, density contrast and various observables ($x_{\rm{e}},\ T_{\rm{k}}$ and $T_{21}$) along with comparisons between IHM and HMG simulations. From the upper 2 panels, it can be seen that the boost factor $B$ exhibits distinctive inhomogeneity patterns that closely traces that in density contrast $\delta$. This is further demonstrated quantitatively in the middle and right panels of Fig. 2, where we show that after normalization, $B$ and $\delta$ shares remarkably similar power spectrum. The amplitude of $B$ also traces density fluctuation level, as can be inferred from the top 2 panels of Fig. 1. At high redshifts ($z>50$) when the inhomogeneity in matter distribution is negligible, $B$ takes unity and increases with the growth of density fluctuation. Since the collapse fraction $f_{\rm coll}$ indirectly reflects the density fluctuation level, this can also be seen in the left panel of Fig. 2, which shows that $B$ grows with $f_{\rm coll}$. By redshift $z=4$ when about $70\%$ of matter collapsed into halos, DM annihilation rate is boosted by roughly a factor of $10^{5}$ compared to the uniform background.

In Fig. 3 we present comparisons of global signal (top) and power spectrum (middle and lower panels) for 21-cm temperature from our simulations. We highlight the power spectrum at $k=0.08{\rm Mpc^{-1}}$ and $k=0.18{\rm Mpc^{-1}}$, corresponding to scales which are large enough for efficient foreground removal and yet small enough for experiments to achieve high signal to noise ratio Mesinger et al. (2014); Lidz et al. (2008); Dillon et al. (2014); Pober et al. (2013). We also include the forecasted $1\sigma$ power spectrum sensitivity for SKA (Square Kilometer Array) telescope computed in Refs. Sitwell et al. (2014); Mesinger et al. (2014), which assumed 2000 hours of observation time and an observational strategy carefully chosen to minimize thermal noise. The noise power spectrum at each $uv$ cell was calculated in Mesinger et al. (2014) as,

where the factor $F$ describes experiment beam, frequency coverage, observational strategy (see Mesinger et al. (2014) for details). As we have chosen $k$ scales that are likely free from foreground contamination, the systematic temperature $T_{\rm sys}$ can be expressed as the sum of sky temperature $T_{\rm sky}$ and receiver noise temperature $T_{\rm rec}$, given by $T_{\rm sky}=237(0.15{\rm GHz}/\nu)^{2.5}\ {\rm K}$ and $T_{\rm rec}=50{\rm K}+0.1T_{\rm sky}$ respectively Rogers and Bowman (2008); Dewdney et al. (2013), here the frequency $\nu$ is related to redshift via $z=1.43{\rm GHz}/\nu-1$. For the experimental specifications,  Mesinger et al. (2014) adopted the SKA Low Phase 1 design Dewdney et al. (2013), which has 866 station locations each with $17\times 17$ array and a frequency resolution of 1 kHz.

Compared to the IHM simulation, $x_{\rm e}$, $T_{\rm k}$ and $T_{\rm 21}$ all display remarkably different inhomogeneity in th IHM simulation (see Fig. 1). For redshift window $z\in[16,\ 30]$ in particular, inhomogeneity levels are significantly enhanced in IHM simulation due to fluctuations in DM heating/ionization rate. 21-cm power spectrum in Fig. 3 provides quantitative comparisons of these fluctuations. In presence of heating from DM, the 21-cm absorption signal ($T_{21}<0$) in the Fiducial setting is weakened or shifted into emission ($T_{21}>0$). For the HMG simulation, the change in $\Delta^{2}_{21}$ relative to the Fiducial setting is largely driven by the difference in 21-cm amplitude, and we found that our global and power spectrum 21-cm signals are similar to that in  Lopez-Honorez et al. (2016). However for the IHM scenario, $\Delta^{2}_{21}$ is also affected by spatial variation of $T_{21}$ induced by inhomogeneous DM heating/ionization. As shown in Fig. 3, the inhomogeneity in boost factor can enhance $\Delta^{2}_{21}$ by more than a factor $10^{4}$ for $16\lesssim z\lesssim 30$, and such enhancement is potentially detectable at the SKA telescope Sitwell et al. (2014).

Particles injected from dark matter (DM) annihilation events can heat up the intergalactic medium and change the 21-cm signal from neutral hydrogen during the cosmic dawn. At low redshifts, the growth of structures can significantly boost DM annihilation rate relative to the uniform background. This paper examines the inhomogeneity in DM annihilation boost factor and its impact on 21-cm brightness temperature power spectrum $\Delta^{2}_{21}$. Building on the 21cmFAST simulation framework and the Press-Schechter conditional halo mass function, we obtain the lightcone evolution and the power spectrum for the inhomogeneous boost factor. We showcase the effect for an annihilation rate of $\left<\sigma v\right>/m_{\chi}=10^{-27}{\rm cm^{3}s^{-1}GeV^{-1}}$, and our result shows that compared to the case with homogeneous boost factor, which approximate long propagation length for DM annihilation products, the inhomogeneous boost factor can induce distinctively different fluctuation features in 21-cm signal in the redshift window of $16\lesssim z\lesssim 30$, and the corresponding 21-cm power spectrum $\Delta^{2}_{21}$ can be enhanced by more than a factor of $10^{4}$. Such features can potentially be detected by the SKA (Square Kilometer Array) telescope.

Our analysis for 21-cm signal with inhomogeneous boost factor relied on a simple on-the-spot prescription for energy deposition process, which is generally a decent approximation below electroweak energy scale or if DM annihilation primarily produce electrons. If particles injected by DM annihilation have long propagation length before absorption, the effect of inhomogeneous boost factor on 21-cm signal can be weakened or washed out. A more comprehensive analysis of the energy deposition process would necessitate a detailed study of particle cascade and propagation in an inhomogeneous background, and we reserve such investigations for future work.

Acknowledgements
We thank Andrei Mesinger, Yuxiang Qin and Steven Murray for their helpful communications. This work is supported by the National Natural Science Foundation of China (grant No. 12275278), the National Research Foundation with grant No. 150580, and the research program “New Insights into Astrophysics and Cosmology with Theoretical Models Confronting Observational Data” of the National Institute for Theoretical and Computational Sciences of South Africa.