Constraints on Dark Matter-Neutrino Interaction from 21-cm Cosmology and Forecasts on SKA1-Low

The standard model of Cosmology, the $\Lambda$CDM model, is so far the most widely accepted theory of the Universe, thanks to its compatibility with the highly precise data of Cosmic Microwave Background (CMB) anisotropy, Large Scale Structure (LSS) along with many other observations, barring couple of moderate to high tensions. In its vanilla version, the Universe is composed of a cosmological constant $\Lambda$ that takes into account $\sim 70\%$ of cosmic budget and $\sim 26\%$ is attributed to cold dark matter (CDM). The CDM, assumed to be non-relativistic (hence dubbed as ”cold") and mostly non-interacting (other than the gravitational interactions), conforms with a number of astrophysical and cosmological observations at all scales starting from galaxies (Rubin & Ford, 1970), cluster of galaxies (Randall et al., 2008), to large scale structures (Tegmark et al., 2004) and CMB (Aghanim et al., 2020). In order to find out the exact constituent of DM, one needs to propose particle candidates. However, in the framework of particle physics, it is unnatural to incorporate DM without (non-gravitational) interactions. Thus one of the major objectives of dark matter research is to search for its possible interactions, either with self or with any other species, either in terrestrial experiments or in astrophysical/cosmological observations.

Several well-motivated Beyond Standard Model (BSM) scenarios give rise to possible dark matter species with interactions with the Standard Model (SM) particles. Of them, a widely accepted candidate is weakly interacting massive particle (WIMP), in which the WIMP DM particles are in thermal equilibrium with the SM particles in the early Universe, leading to energy injection into the SM thermal bath due to DM annihilations, and the effect is observable in the CMB anisotropy spectrum (Chluba, 2010), (Finkbeiner et al., 2012), (Galli et al., 2013), (Slatyer et al., 2009). Scattering between DM and baryons leave imprint in the cosmological observables, hence have been constrained both in the context of CMB (Boddy et al., 2018) and 21 cm Cosmology (Barkana, 2018). Other interesting possibilities include DM-DR (dark radiation) interactions (Cyr-Racine et al., 2016), (Escudero et al., 2018a) and DM-neutrino interactions, which have been studied in the light of CMB Power Spectrum (Mangano et al., 2006), (Wilkinson et al., 2014), (Escudero et al., 2015), (Escudero et al., 2018b), (Di Valentino et al., 2018), (Olivares-Del Campo et al., 2018), (Stadler et al., 2019), (Ghosh et al., 2020), (Mosbech et al., 2021), (Paul et al., 2021). DM annihilation into highly energetic neutrinos have also well studied in the literature (Blennow et al., 2008) constraining DM annihilation cross-section at current epoch (Argüelles et al., 2021), (Frankiewicz, 2015).

Among the above-mentioned possibilities, we consider the interaction between DM and neutrinos and explore its possible signatures in reionization history. For the early stages of the Universe, when the perturbation theory is valid, the evolution of the fluctuations of DM and neutrino fluids can be studied using the first order Boltzmann equations, hence estimating its impact in CMB anisotropy spectrum. This have been exhaustively studied in the literature as mentioned before.

On cosmological scales, sizable DM-neutrino interactions may significantly impact the structure formation of the Universe, resulting from the modified evolution of DM over-density. This non-standard interaction provides with a pressure-like term in the DM fluid, thus preventing DM from collapsing significantly at small length scales. Similar to the acoustic waves in the baryon-photon fluid, a dark acoustic oscillation along with the collisional and silk damping-like phenomenon is also observed in this case. Although the imprint of this oscillation may be observed in the linear matter power spectrum, non-linear evolution at late time washes out some of these features, leaving only a suppression of power at small scales. The suppression of the power at small scales may hence significantly reduce the halo formation and number of stars and galaxies in those halos, resulting in a smaller number of ionizing photons in the intergalactic medium (IGM). This lower number of ionizing photons reduce the amount of ionized hydrogen in the IGM and consequently, affects the reionization history.

This is the primary target of the present article. We have made a thorough investigation of the possible effects of DM-neutrino interaction on reionization history. In the process, we have constrained the interaction strength using 21 cm Cosmology. We found that the constraints on DM-neutrino interaction from 21 cm missions is much more stringent compared to the constraints from existing cosmological observations. We also find a non-trivial effect of DM-neutrino interaction in HI maps. We have done forecast analysis in the upcoming 21 cm mission Square Kilometer Array 1 - Low (SKA1-Low).

The paper is organized as follows: in section 2 we have discussed about the cosmological scalar perturbation equations in presence of DM- neutrino scattering, in section 3 we describe the semi-numerical reionization simulation and constraints of these parameter from different observations, in 4 constraints on the DM-neutrino parameter from reionization physics is discussed and detection techniques of this interaction are presented in 5 for upcoming telescope SKA$1$-Low.

Throughout this work, we have adopted cosmological parameters $\Omega_{\text{m}}=0.3183,\Omega_{\Lambda}=0.6817,h=0.6704,\Omega_{\text{b}}h^{2}=0.022032,\sigma_{8}=0.8347,n_{\text{s}}=0.9619$ consistent with Planck+WP best-fit values (Aghanim et al., 2020).

In order to consider DM-neutrino interaction one needs to go beyond the six parameter description of vanilla $\Lambda$CDM model and analyze the new set of perturbation equations. Let us describe the present situation briefly. Here we will work on Newtonian gauge for our analysis. Presence of scattering between two cosmic species (DM and neutrino in the present case) results in a drag term in the perturbed Euler equation for the corresponding species. Assuming a flat Universe, the modified perturbation equations for dark matter and neutrino fluid, in presence of DM-neutrino interaction is given by (Stadler et al., 2019),

In principle, $\sigma_{DM-\nu}$ can be velocity dependent (Wilkinson et al., 2014). However, in the present analysis we will consider $\sigma_{DM-\nu}$ to be constant.This dimensionless quantity $u$ will act as a new parameter on top of the 6-parameter in the present scenario.

Over the years, the constraint on $u$ have improved considerably with the availability of new datasets. For example, using Planck 2013 and WP measurement the constraints on u was found to be $u<3.27\times 10^{-2}$ (Wilkinson et al., 2014). Recent studies using Planck 2018 data-set (high-$\ell$ TT+TE+EE, low-$\ell$ TT, low-$\ell$ EE) the constraint comes out to be a tighter $u<3.34\times 10^{-4}$ (Mosbech et al., 2021), (Paul et al., 2021), the scattering cross-section constraint being $\sigma_{DM-\nu}<7.44\times 10^{-29}cm^{2}$. As mentioned earlier, in this article, our primary intention is to investigate if 21 cm Cosmology can put further constraints on the parameter $u$ and the prospects of SKA1-Low in this regard, which will in turn help us have better idea on more stringent constraints on DM-neutrino interaction from future observations.

Let us first find out the effects of DM-neutrino interaction on the matter power spectrum in Fig. 1 using a modified version (Stadler et al., 2019) of the CLASS code (Blas et al., 2011). We have illustrated this for specific values of the interaction parameter $u$ irrespective of observational constraints. For this, we have considered flat $\Lambda$CDM model with DM-neutrino interaction on top of it, where the other cosmological parameters are taken as the Planck 2018 best-fit values (Aghanim et al., 2020). Fig. 1 reveals that the effects of DM-neutrino interaction (with $u<10^{-4}$) on the matter power spectrum is being reflected by a series of damped oscillations with a suppression of power in the non-linear regime ($k\geq 0.2hMpc^{-1}$).

Having convinced ourselves on the non-trivial effects of DM-neutrino interaction, we are now in a position to explore more realistic situation, namely, the reionization simulation, corresponding HI maps and error estimation for SKA1-Low, that will help us drawing important conclusions on the prospects of 21 cm cosmology in this regard. We will do that stepwise in the subsequent sections.

Let us now discuss the major results obtained from our analysis.

• •

  In order to find out the effect of DM-neutrino interaction on the reionization epoch, we have performed the semi-numerical simulation for different strength of interaction parameter ‘$u$’ (as given in Table 1) and we have considered the same initial random seed for all the simulations. We have considered benchmark values for which neutral hydrogen fraction $x_{H\Romannum{1}}\sim 0.5$ at redshift $z=8.0$. Introduction of the DM-neutrino interaction suppresses the matter power spectrum w.r.t. the $\Lambda$CDM model in the non-linear regime, thereby hampering dark matter halo formation. As a result, it will reduce star formation and lesser number of ionizing photons will be produced, hence delaying the reionization process. So, in order to achieve $50\%$ ionization $z=8$ in the interacting dark scenario, $N_{\rm ion}$ has to be increased. However, for Population $\Romannum{2}$ stars, $N_{\rm ion}$ beyond $500$ is not physically allowed. So, we have restricted ourselves to $N_{\rm ion}<500$ condition. Keeping two reionization parameters $M_{\rm min}$ and $R_{\rm mfp}$ fixed at $1.08\times 10^{9}M_{\rm Sun}$ and $20Mpc$, we have varied the parameters $u$ and $N_{\rm ion}$ in order to achieve $x_{H\Romannum{1}}=0.5$. In order to make the analysis robust, we have taken three different values of $N_{\rm ion}$ and searched for possible range of the interaction parameter $u$ for which $50\%$ ionization condition can be achieved. In order to satisfy two conditions $N_{\rm ion}<500$ and $x_{H\Romannum{1}}=0.5$ simultaneously we have found that the the maximum value of the interaction parameter can be $u\leq 6.6\times 10^{-7}$ that can be achieved from the reionization physics using these two criteria.¹¹1A somewhat similar analysis has been put forward in arXiv after this paper, that investigated for the bounds on ’$u$’ for massive neutrinos (Mosbech et al., 2022) that reassures our analysis.

• •

  It is noteworthy that the current constraints on the DM-neutrino interaction in light of different existing cosmological observations are not that stringent. For example, using Planck 2018 data (high-$\ell$ TT+TE+EE, low-$\ell$ TT, low-$\ell$ EE) the allowed range of the interaction parameter is of the order $u<3.34\times 10^{-4}$ (Mosbech et al., 2021), (Paul et al., 2021). Our analysis reveals that putting 21 cm observation on top of existing observations results in a much more tighter constraint on the DM-neutrino interaction ($u\leq 6.6\times 10^{-7}$). So, the upcoming 21 cm missions like SKA would have very good prospects to constrain DM-neutrino interaction further compared to existing observations and hence to improve our understanding about this cosmic entities.

  DM-$\nu$ interaction Model (u)	$N_{\rm ion}$	$x_{H\Romannum{1}}$
  0.0 ($\Lambda$CDM)	24	0.5
  8.8 $\times 10^{-8}$	300	0.51
  6.6 $\times 10^{-7}$	500	0.51

• •

  We have done the N-body simulation for three different cases and generated a non-linear matter power spectrum (Fig. 2) at redshift $z=8.0$. From the figure, it is obvious that no oscillatory signature in the power spectrum is observed, as non-linearities washes out these features. Nevertheless, the maximum suppression of the power is observed for $u=6.6\times 10^{-7}$.

  

• •

  For different values of the interaction parameter $u$ and $N_{\rm ion}$, different ionization features appear. As justified earlier, we have restricted our analysis to $x_{H\Romannum{1}}=0.5$ and $N_{\rm ion}<500$ condition at $z=8.0$. However the assumption of $x_{H\Romannum{1}}=0.5$ at $z=8.0$ has considerable uncertainties, which may change the bounds on $u$ to some extent. With this we have performed the reionization simulation and have generated HI maps for the above three representative value of $u$. These have been depicted in Fig. 3. The maps reveal that increase in the value of $u$ within its allowed range decreases ionization and delays reionization physics. In order to achieve $50\%$ ionization condition $N_{\rm ion}$ has to be increased. We observe slight increase in bubble size as $N_{\rm ion}$ increases, confirming the non-trivial effect of interaction on HI maps. Further, as u increases, we found slightly lower number of dark matter halos forming in the universe, as expected from the suppression of dark matter power spectrum. However, this change resulting from the figures may not be significant enough so as to quantify and find out its edge over standard $\Lambda$CDM scenario, mostly because of the feeble value of the interaction strength.

• •

  We argue that our results are solely depend on the two constraints which are $x_{HI}=0.5$ at $z=8.0$ which is justified from Planck 2018 data and $N_{ion}<500$ for Pop-II stars. The two main observable quantities which constraints reionization epoch are the optical depth to reionization ($\tau_{Th}$ = 0.054 $\pm$ 0.007) and Gunn Peterson bound from quasar absorption spectra which tells abount the end of reionization. We have considered a conservative limit on $x_{HI}$ so that it can satisfy the two observational bounds. However, if we consider the $50\%$ ionization criteria at some smaller redshift (Bouwens et al., 2015), (Wang et al., 2020), (Mason et al., 2019), (Villanueva-Domingo, 2021), say at $z=7$, then it will affect the constraints on $u$ and the conclusion may differ slightly from our case. While these aspects worth investigating, we do not expect a significant deviation from the conservative approach taken up in this article.

  

  

  

Having convinced ourselves of the role of 21 cm observations in putting tighter constraints on DM-neutrino interaction, let us now investigate the prospects of the upcoming radio observations in this regard. To this end, we have done a forecast analysis on the upcoming SKA1-Low telescope. The specifications for SKA1-Low are shown in Table 2. Considering SKA1-Low antenna layout (Dewdney & Braun, 2016) we have simulated the distribution of antenna pair separations ’d’ corresponding to 8 hours of observations with an integration time of 60 seconds towards a fixed sky direction located at DEC= $-30^{\circ}$. In addition to the signal i.e 21 cm brightness temperature fluctuation, observed 21 cm differential brightness temperature will have a contribution from random Gaussian system noise corresponding to each and every grid point in the simulation. The noise power spectra is given by (Chatterjee & Bharadwaj, 2018), (Shaw et al., 2019), (Mondal et al., 2020).

where $P_{0}$ is the system noise power spectrum for a single visibility measurement with 60 seconds integration time. $\tau_{k_{g}}$ quantifies the number of independent visibility measurements. Here $t_{\rm obs}$ is the observation performed over number of nights, considering a run in each night consists of 8 hours. So, effectively, the noise power spectrum will depend on the number of observation nights (in the denominator) scaled by 8 hours (in the numerator).

We have compared the signal i.e. differential brightness temperature fluctuation with the noise power spectra for 1024, 10000, 50000 hours of observation and the results have been shown in Fig. 5. While computing the noise power spectrum we have considered the effect of cosmic variance. We can write the cosmic variance as (Shaw et al., 2019),

where $\bar{P}(k_{i})$ is bin averaged power spectrum and $N_{k_{i}}$ is number of independent k modes in each bin. In our analysis we have assumed 21 cm signal is completely Gaussian so we have neglected the trispectrum $(\bar{T}(k_{i},k_{j}))$ contribution of the cosmic variance. If one incorporates the tri-spectrum contribution in the calculation then it will effect noise power spectrum slightly and the correlations which we have estimated for the astrophysical parameters will change slightly. However, if the signal is intrinsically non-gaussian at this redshift range, it should affect the calculations considerably. We hope to explore that direction in future.

We have assumed the foregrounds which plays an important role for detecting low frequency radio signal, is removed completely after modelling it so that the entire k space is allowed for detecting the signal. We have observed that in each of these cases the signal is higher w.r.t. the noise for all ranges of k, thereby giving a sizable SNR as required for detection.

We have done fisher forecast analysis for our astrophysical parameters and model parameter. The fisher matrix corresponding to the model parameter $q_{\alpha}$ is (Repp et al., 2015)

where $C_{ij}$ is covariance matrix and $\bar{P}(k_{i})$ is bin averaged power spectrum. In Fig. 4 the forecast on the parameter $u$, $N_{\rm ion}$, $M_{\rm min}$ and $R_{\rm mfp}$ has shown. From the analysis we have found a positive correlation on $u$ and $N_{\rm ion}$ and a negative correlation between $u$ and $M_{\rm min}$. Further from our analysis we found there is no significant dependencies on $R_{\rm mfp}$ and $u$.

In Fig. 5 a comparison between Signal and Noise Power Spectra for SKA1-Low configuration has been demonstrated and explained in details. For $u>0$, the power at small length scales in both the linear and non-linear matter power spectra are suppressed, as a result less number of low mass halos are formed, resulting in lower number of ionizing sources. Hence the ionization at small length scales are suppressed, giving rise to more HI fraction, resulting an increase in the HI power spectrum. An essential conclusion that comes out of the error analysis as in Fig. 5 is that the prospects of detection of any interaction between DM and neutrino, if any, of strength as constrained by reionization epoch in the upcoming SKA1-Low telescope is quite high. So, this guarantees further investigation in this direction, some of which we hope to take up as we go along.

In this work we have examined the effect of DM-neutrino interaction in the reionization epoch. We have run the semi-numerical N-body simulation to compute the reionization observables. We have put a constraint in the dark matter neutrino interaction parameter $u\equiv 6.6\times 10^{-7}$ using $50\%$ ionization criteria and $N_{\rm ion}\leq 500$ condition at redshift $z=8.0$. Reionization puts stronger constraints to the interaction parameter $u$ than any other observations studied in the literature. We also found a non-trivial effect of the said interaction on HI maps via a slight yet distinguishable change in the sizes of the bubbles. Further, we have done the forecast analysis for the upcoming SKA1-Low telescope and found that the DM-neutrino interaction signal using this telescope can be observed for a very small time of observation and we observe possible correlations of the reionization parameters with interaction parameter. Thus, future 21 cm mission like SKA1-Low can help us comment in a more convincing way whether or not these two cosmic species interact, and if so, what can be the possible interaction strength. This will not only shed some light on these entities on top of their properties as constrained by particle physical theory and experiments but also help us have better understanding of these cosmic species on top of the existing cosmological observations like CMB.

The analysis done in the present article mostly focuses on reionization era with typical redshift of our consideration $z=8$. However, if at all DM and neutrinos interact, that may reflect on post-reionization era as well. A thorough investigation of the effects of such interaction on post-reionization era needs to be done that will act as a complimentary to the constraints as obtained from 21 cm Physics during reionization. We look forward to explore these aspects in near future.

Authors gratefully acknowledge the use of publicly available code CLASS (Blas et al., 2011), Particle Mesh N-body code (Bharadwaj & Srikant, 2004), Friends-of-Friends code (Mondal et al., 2015), ReionYuga code (Choudhury et al., 2009), (Majumdar et al., 2014), (Mondal et al., 2016) and thank the computational facilities of Indian Statistical Institute, Kolkata. AD thanks ISI Kolkata for financial support through Senior Research Fellowship and Abinash Kumar Shaw for useful discussions. SP thanks the Department of Science and Technology, Govt. of India for partial support through Grant No. NMICPS/006/MD/2020-21.

The simulated data underlying this work will be shared upon reasonable request to the corresponding author.