Radio constraints on dark matter annihilation in the galactic halo and its substructures

Radio constraints on dark matter annihilation in the galactic halo and its substructures

E. Borriello111eborriello@na.infn.it, A. Cuoco222cuoco@phys.au.dk and G. Miele333miele@na.infn.it Università “Federico II”, Dipartimento di Scienze Fisiche, Napoli, Italy & INFN Sezione di Napoli
Department of Physics and Astronomy, University of Aarhus, Ny Munkegade, Bygn. 1520 8000 Aarhus Denmark
Instituto de Física Corpuscular (CSIC-Universitat de València), Ed. Institutos de Investigación, Apartado de Correos 22085, E-46071 València, Spain.
July 13, 2019
Abstract

Annihilation of Dark Matter usually produces together with gamma rays comparable amounts of electrons and positrons. The gyrating in the galactic magnetic field then produce secondary synchrotron radiation which thus provides an indirect mean to constrain the DM signal itself. To this purpose, we calculate the radio emission from the galactic halo as well as from its expected substructures and we then compare it with the measured diffuse radio background. We employ a multi-frequency approach using data in the relevant frequency range 100 MHz–100 GHz, as well as the WMAP Haze data at 23 GHz. The derived constraints are of the order  cms for a DM mass  GeV sensibly depending however on the astrophysical uncertainties, in particular on the assumption on the galactic magnetic field model. The signal from single bright clumps is instead largely attenuated by diffusion effects and offers only poor detection perspectives.

pacs:
95.35.+d, 95.85.Bh, 98.70.Vc
preprint: DSF/21/2008, IFIC/08-48

I Introduction

Cosmology and Astrophysics provide nowadays a compelling evidence of the existence of Dark Matter (DM) Komatsu:2008hk (); Bertone:2004pz (). Nevertheless, its nature still remains elusive, and Dark Matter constituents have escaped a direct detection in laboratory so far. Promising candidates are DM particles produced in thermal equilibrium in the early universe, the so-called Weakly Interacting Massive Particles (WIMPs). Theoretically, models of WIMPs naturally arise, for example, in SUSY as the Lightest Super-symmetric Particle or as the Lightest Kaluza-Klein Particle in the framework of extra-dimensions. These candidates are self-conjugate and can thus annihilate in couples to produce as final states: neutrinos, photons, electrons, light nuclei (as wells as their antiparticles), etc., which can in principle be detected.

Among the indirect DM detection channels, gamma-ray emission represents one of the most promising opportunity due to the very low attenuation in the interstellar medium, and to its high detection efficiency. See for example Ref.s Bertone:2004pz (); Jungman:1995df (); Bergstrom:2000pn () for a review of this extensively studied issue. The expected neutrino detection rates are generally low although forthcoming km detectors offer some promising prospect Bergstrom:1998xh (); Barger:2001ur (). Finally, positrons and protons strongly interact with gas, radiation and magnetic field in the galaxy and thus the expected signal sensibly depends on the assumed propagation model Baltz:1998xv (); Hooper:2004bq (); Donato:2003xg (). However, during the process of thermalization in the galactic medium the high energy and release secondary low energy radiation, in particular in the radio and X-ray band, that, hence, can represent a chance to look for DM annihilation. Furthermore, while the astrophysical uncertainties affecting this signal are similar to the case of direct , detection, the sensitivities are quite different, and, in particular, the radio band allows for the discrimination of tiny signals even in a background many order of magnitudes more intense.

Indirect detection of DM annihilation through secondary photons has received recently an increasing attention, exploring the expected signature in X-rays Bergstrom:2006ny (); Regis:2008ij (); Jeltema:2008ax (), at radio wavelengths Blasi:2002ct (); Aloisio:2004hy (); Tasitsiomi:2003vw (); Zhang:2008rs () , or both Colafrancesco:2005ji (); Colafrancesco:2006he (); Baltz:2004bb (). In the following we will focus our analysis on the radio signal expected from the Milky Way (MW) halo and its substructures. It is worth noticing that the halo signal has been recently discussed in Ref.s Hooper:2007kb (); Hooper:2008zg (); Grajek:2008jb () in connection to the WMAP Haze, which has been interpreted as a signal from DM annihilation. In this concern we will take in the following a more conservative approach, by assuming that the current radio observations are entirely astrophysical in origin, and thus deriving constraints on the possible DM signal. The main point will be the use of further radio observations besides the WMAP ones, in the wide frequency range 100 MHz-100 GHz, and a comparison of the achievable bounds. Furthermore, the model dependence of these constraints on the assumed astrophysical inputs will be analyzed. We will also discuss the detection perspectives of the signal coming from the brightest DM substructures in the forthcoming radio surveys.

The paper is organized as follows: in section II we will discuss the astrophysical inputs required to derive the DM signal such as the structure of the magnetic field, the DM spatial distribution and the radio data employed to derive the constraints. In section III we describe in detail the processes producing the DM radio signal either when it is originated from the halo or from the substructures. In section IV we present and discuss our constraints, while in section V we analyze the detection sensitivity to the signal coming from the single DM clump. In section VI we give our conclusions and remarks.

Ii Astrophysical Inputs

ii.1 Dark matter distribution

Our knowledge of the DM spatial distribution on galactic and subgalactic scales has greatly improved thanks to recent high resolution zoomed N-body simulations Diemand:2005vz (); Diemand:2006ik (); Kuhlen:2008aw (); Springel:2008cc (). These simulations indicate that for the radial profile of the galactic halo the usual Navarro-Frank-White (NFW) distribution Navarro:1996gj ()

 ρ(r)=ρhrrh(1+rr%h)2, (1)

still works as a good approximation over all the resolved scales. We will thus use this profile in the following. Note, anyway, that this choice is quite conservative with respect to other proposed profiles like the Moore profile Moore:1999nt (), which exhibits an internal cusp that would give in principle a divergent DM annihilation signal from the center of the halo. Observationally the situation is more uncertain: Baryons generally dominate the gravitational potential in the inner kpc’s and fitting the data thus requires to model both the baryon and DM component at the same time. The NFW profile is in fair agreement with the observed Milky Way rotation curve Klypin:2001xu (), although, depending on the employed model, it is possible to find an agreement for many different DM profiles (see also Bertone:2004pz () and references therein). We emphasize, however, that the various profiles differ mainly in the halo center (for kpc) where the uncertainties, both in numerical simulations and from astrophysical observations are maximal. Thus, our analysis which explicitly excludes the galactic center, does not crucially depend on the choice of the profile.

A problem related to the profile of Eq.1 is that the mass enclosed within the radius is logarithmically divergent. A regularization procedure is thus required to define the halo mass. Following the usual conventions we define the mass of the halo as the mass contained within the virial radius , defined as the radius within which the mean density of the halo is times the mean critical cosmological density which, for a standard cosmological model ( Komatsu:2008hk ()) is equal to GeV  cm. The parameters describing the halo are then determined imposing the DM density to be equal to GeV cm near the Solar System, at a galactocentric distance of kpc.

Simulations, however, predict a DM distribution sum of a smooth halo component, and of an additional clumpy one with total masses roughly of the same order of magnitude. Hereafter we will assume for the mass of the Milky Way , where and denote the total mass contained in the host galactic halo and in the substructures (subhaloes) distribution, respectively. The relative normalization is fixed by imposing that subhaloes in the range , have a total mass amounting to 10% of Diemand:2005vz (). Current numerical simulations can resolve clumps with a minimum mass scale of . However, for WIMP particles, clumps down to a mass of are expected Hofmann:2001bi (); Green:2005fa (). We will thus consider a clump mass range between and .

Finally, to fully characterize the subhalo population we will assume a mass distribution and that they are spatially distributed following the NFW profile of the main halo. The mass spectrum number density of subhaloes, in galactocentric coordinates , is thus given by

 dncldmcl(mcl,→r)=A(mclMcl)−2(rrh% )−1(1+rrh)−2 , (2)

where is a dimensional normalization constant. The above expression assumes some approximations: for example, a more realistic clump distribution should take into account tidal disruption of clumps near the galactic center. Numerical simulations suggest also that the radial distribution could be somewhat anti-biased with respect to the host halo profile. However, with our conservative assumptions the DM annihilation signal is dominated by the host halo emission within up to from the galactic center so that the details of the sub-dominant signal from the clumps have just a slight influence on the final results. Recent results also show that mass distribution seems to converge to rather than Kuhlen:2008aw (). This would also produce only a minor change in the following results.

Following the previous assumptions the total mass in DM clumps of mass between and results to be

 M(m1,m2) = ∫d→r∫m2m1mcldncldmcl(mcl,→r)dmcl (3) = 4π[ln(1+ch)−ch1+ch% ](Ar3hMcl) × ln(m2m1)Mcl ,

where denotes the host halo concentration; while their number is

 N(m1,m2) = ∫d→r∫m2m1dncldmcl(mcl,→r)dmcl (4) = 4π[ln(1+ch)−ch1+ch% ](Ar3hMcl) × (Mclm1−Mclm2) .

Imposing the normalization condition , we finally get for the mass due to the entire clumps distribution:

 Mcl=M(10−6M⊙,1010M⊙)∼53.3%MMW , (5)

while for the number of these clumps we obtain

 N(10−6M⊙,1010M⊙)∼2.90×1017. (6)

Finally by using the previous constraints one can fix the values of free parameters , and , hence obtaining kpc, that corresponds to a halo concentration of ,  GeV cm and kpc M.

A further piece of information is required to derive the annihilation signal from the clumps, namely how the DM is distributed inside the clumps themselves. We will assume that each clump follows a NFW profile as the main halo with and replacing the corresponding quantities of Eq.1. However, for a full characterization of a clump, further information on its concentration is required. Unluckily, numerical simulations are not completely helpful in this case, since we require information about the structure of clumps with masses down to , far below the current numerical resolution. Analytical models are thus required. In the current cosmological scenario Komatsu:2008hk () structures formed hierarchically, via gravitational collapse, with smaller ones forming first. Thus, naively, since the smallest clumps formed when the universe was denser, a reasonable expectation is , where is the clump formation redshift. Following the model of ref. Bullock:1999he () we will thus assume with and . With this concentration the integrated DM annihilation signal from all the substructures dominates over the smooth halo component only at about from the galactic center (see section III), so that the constraints on the DM signal do not crucially depend on the unresolved clumps signal, coming basically only from the smooth halo component. However, given the large uncertainties in the models, larger contributions from the unresolved population of clumps are in principle possible considering a different parametrization of the concentration (see for example the various models considered in Pieri:2007ir (); Kuhlen:2008aw ()). We will not investigate further this possibility here. An enhancement of the clumps signal is also possible considering different choices of the clump profile other than the NFW: Differently from the case of the halo, in fact, the clump signal depends sensibly from the chosen profile and a Moore profile or an Isothermal profile can in principle enhance the signal of several orders of magnitude. Also in this case we choose to quote conservative constraints and we will not consider these possibilities further.

ii.2 Galactic Magnetic Field

The MW magnetic field is still quite uncertain especially near the galactic center. The overall structure is generally believed to follow the spiral pattern of the galaxy itself with a normalization of about G near the solar system. Eventually, a toroidal or a dipole component is considered in some model.

We will consider in the following the Tinyakov and Tkachev model (TT) Tinyakov:2001ir () which is a fair representative of the available descriptions of MW magnetic field. Within this model , the field shows the typical spiral pattern, an exponential decrease along the axis and a behavior in the galactic plane. The field intensity in the inner kiloparsecs is constant at about G. We will use the slightly modified parametrization of this model as described in Kachelriess:2005qm (). Higher normalizations are in principle possible considering more complex structures as for example a dipole or a toroidal component Prouza:2003yf (). Indeed some recent analyses Sun:2007mx (); Noutsos:2008jw () including new available data seems to favor the presence of these further structures. We will thus consider as possible also an “high normalization model” that we simply parameterize as a constant 10 G field. This choice is also motivated by a comparison with the results of Hooper:2007kb (); Hooper:2008zg () where the same magnetic field is used.

Further, beside the regular component, the galactic magnetic field presents a turbulent random component. The r.m.s. intensity of this component is generally expected of the same order of magnitude of the regular one, but both its spatial distribution and its spectrum are poorly known, thus here we neglect its effects. Naively, this random component is expected to affect the synchrotron maps that we will show in the following producing a blurring of the otherwise regular pattern. Also, the random component contributes to increase the overall normalization of the field. Thus without this component the synchrotron signal is slightly underestimated so that we can regard this choice as conservative.

In the following we will derive constraints on the DM emission comparing the expected diffuse emission from the smooth halo and the unresolved population of clumps with all sky observation in the radio band. In the frequency range 100 MHz-100 GHz where the DM synchrotron signal is expected, various astrophysical processes contribute to the observed diffuse emission. Competing synchrotron emission is given by Cosmic Ray electrons accelerated in supernovae shocks dominating the radio sky up to 10 GHz. At higher frequencies the Cosmic Microwave Background (CMB) and its anisotropies represent the main signal. However, thanks to the very sensitive multi-frequency survey by the WMAP satellite, this signal (which represents thus a background for DM searches) can be modeled in a detailed way and can thus be removed from the observed radio galactic emission Tegmark:2003ve (). Other processes contributing in the 10-100 GHz range are given by thermal bremsstrahlung (free-free emission) of electrons on the galactic ionized gas, and emission by small grains of vibrating or spinning dust.

In the following our approach will be to compare the DM signal with the observed radio emission where only the CMB is modeled and removed. For this purpose we use the code described in deOliveiraCosta:2008pb () where most of the radio survey observations in the range 10 MHz-100 GHz are collected and a scheme to derive interpolated, CMB cleaned sky maps at any frequency in this range is described.

A more aggressive approach would be of course to try to model and subtract also the remaining emissions (synchrotron, free-free, dust) in order to compare the expected DM signal with the residual radio map. This is indeed the approach followed in Finkbeiner:2003im (); Dobler:2007wv () where residual maps at the 5 WMAP frequencies are derived using spatial templates for the various expected astrophysical components. The residual maps then exhibit the feature called the WMAP Haze, which has been indeed interpreted as radio emission related to DM annihilation Finkbeiner:2004us (); Hooper:2007kb (). However, the fit procedure used for the Haze extraction is crucial, and using more degrees of freedom to model the foregrounds as performed by the WMAP team Gold:2008kp () fails in finding the feature. We will anyway show in the following for comparison the constraints derived using the Haze residual map at the WMAP frequency of 23 GHz HazeSkymap (). A map of the Haze and of the 1 GHz emission is shown in Fig.1. We will see however that within our conservative approach comparable or better constraints can be obtained thanks to the use of multi-frequency information. For a given DM mass, in fact, 23 GHz is generally not the best frequency to use and better constraints are instead obtained using observations at lower frequencies even without further foreground modeling.

Definitely, a detailed foreground modeling at all radio frequencies would clearly give much stronger constraints on the DM signal and/or eventually confirm the DM nature of the WMAP haze. To this purpose consistent progress will be achieved in the next years with the new high quality data coming from the PLANCK mission and from low frequency arrays like LOFAR and SKA.

Iii DM Synchrotron Signal

iii.1 Particle Physics

In a standard scenario where WIMPs experience a non exotic thermal history, a typical mass range for these particles is , while a simple estimate for their (thermally averaged) annihilation cross section yields Jungman:1995df (), giving for as resulting from the latest WMAP measurements Komatsu:2008hk (). However, this naive relation can fail badly if, for example, coannihilations play a role in the WIMP thermalization process Griest:1990kh (), and a much wider range of cross sections should be considered viable. In this work we consider values of from about 10 GeV to about 1 TeV, and in the range (-

The annihilation spectrum for a given super-symmetric WIMP candidate can be calculated for example with the DarkSUSY package Gondolo:2004sc (). However, the final spectrum has only a weak dependence on the exact annihilation process with the channels giving basically degenerate spectra. For leptonic channels like the decaying mode the spectrum differs significantly, although this channel has generally a quite low branching ratio. For simplicity we will assume hereafter full decay into channel, hence () will be emitted by decaying muons (anti-muons) produced in pions decays. In this framework, the resulting , spectrum can be written as a convolution, namely

 dNedEe(Ee)=∫mχc2EedEμdN(μ)edEe(Ee,Eμ) ×∫Eμ/ξEμdEπWπ(Eπ)dN(π)μdEμ(Eπ) (7)

with where

 dN(μ)edEe(Ee,Eμ)=2Eμ[56−32(EeEμ)2+23(EeEμ)3], (8) dN(π)μdEμ(Eπ)=1Eπm2πm2π−m2μ, (9) Wπ(Eπ)=1mχc21516(mχc2Eπ)32(1−Eπmχc2)2. (10)

In particular, Eq. (8) is the electron (positron) spectrum produced in the muon (anti-muon) decay (). Eq. (9) stands for the () spectrum from () decay process, and, finally, Eq. (10) provides a reasonable analytical approximation of the spectrum of pions from hadronization Hill1983 (). It is worth noticing that to be more accurate Eq. (10) should be substituted by a numerical calculation, which however results not necessary for the aim of the present paper as discussed in the following.

In this approximation the final electron (positron) spectrum can be cast in a simple polynomial form of the ratio :

 dNedEe(Ee)=1mχc2∑j∈Jaj(Eemχc2)j, (11)

where and the coefficients are listed in Table 1.

The main advantage of using the above analytical approximation instead of a more accurate numerical input is that, as will be clear in the next section, most of the observables for the radio emission will be expressed in an analytical form as well. This, in turn, is of help for a better understanding of the physical results. Nevertheless, the difference with the complete numerical calculation turns to be small, arising only for quite low electron energies, and thus for very low radio frequencies. At low energies, in fact, the analytical form has an asymptotic behavior while the numerical spectrum has a turn down. From a comparison with the numerical output from DarkSUSY for a 100 GeV WIMP with 100% branching ratio into the analytical form is a fair approximation until GeV, which for a magnetic field G translates into a minimum valid frequency MHz, thus below the frequency window we are going to explore (see Eq.20 below).

iii.2 Electrons equilibrium distribution

Dark matter annihilation injects electrons in the galaxy at the constant rate

 Q(Ee,r)=12(ρ(r)mχ)2⟨σAv⟩dNedEe . (12)

On the other hand, the injected electrons loose energy interacting with the interstellar medium and diffuse away from the production site. In the limit in which convection and reacceleration phenomena can be neglected, the evolution of the fluid is described by the following diffusion-loss equation Moskalenko:1997gh (); Strong:1998pw (); Strong:1998fr ()

 ∂∂tdnedEe = →▽⋅[K(Ee,→r)→▽dnedEe]+∂∂Ee[b(Ee,→r)dnedEe] (13) +Q(Ee,→r),

where stands for the number density of , per unit energy, is the diffusion constant, and represents the energy loss rate. The diffusion length of electrons is generally of the order of a kpc (see section V) thus for the diffuse signal generated all over the galaxy, and thus over many kpc’s, spatial diffusion can be neglected. This is not the case for the signal coming from a single clump for which the emitting region is much smaller than a kpc. We will further analyze this point in section V. By neglecting diffusion, the steady state solution can be expressed as

 dnedEe(Ee,→r)=τEe∫mχc2EedE′eQ(E′e,→r), (14)

where is the cooling time, resulting from the sum of several energy loss processes that affect electrons. In the following we will consider synchrotron emission and Inverse Compton Scattering (ICS) off the background photons (CMB and starlight) only, which are the faster processes and thus the ones really driving the electrons equilibrium. Other processes, like synchrotron self absorption, ICS off the synchrotron photons, annihilation, Coulomb scattering over the galactic gas and bremsstrahlung are generally slower. They can become relevant for extremely intense magnetic field, possibly present in the inner parsecs of the galaxy Aloisio:2004hy (), and thus will be neglected in this analysis.

For synchrotron emission the energy loss is given by (for ex. see Longair ()) with the magnetic energy density so that the time scale of the energy loss is:

 τsyn=τ0syn(BμG)−2(EeGeV)−1 (15)

with .

Similarly, for Inverse Compton emission the energy loss is given by . The relevant radiation background for ICS is given by an extragalactic uniform contribution consisting of the CMB with eV/cm, the optical/infrared extragalactic background and the analogous spatially varying galactic contribution, the Interstellar Radiation Field (ISRF). For the latter we use as template the Galprop distribution model Porter:2005qx () which reduces to the extragalactic one at high galactocentric distances. In this model, the ISRF intensity near the solar position is about 5 eV/cm, and reaches values as large as 50 eV/cm in the inner kpc’s. With this model the ICS is always the the dominant energy loss process, also near the galactic center (see Fig.2). We thus have

 (16)

with .

Finally, considering both the energy losses we have

 τ(Ee,→r) = (EeGeV)−1μ(→r)τ0syn, (17) μ(→r) = ⎡⎣(B(→r)μG)2+τ0synτ0ICSUrad(→r)eV/cm3⎤⎦−1, (18)

with the function enclosing the whole spatial dependence.

By substituting the above expressions into Eq.(14) we get the following equilibrium distribution for electrons

 dnedEe = ⟨σAv⟩τ0% syn2mχc2μ(→r)(ρ(r)mχ)2 (19) ×∑k∈Kbk(mχc2GeV)−k(EeGeV)k−1,

being , , if , while .

iii.3 Synchrotron spectrum

The synchrotron spectrum of an electron gyrating in a magnetic field has its prominent peak at the resonance frequency

 ν=ν0(BμG)(EeGeV)2 , (20)

with . This implies that, in practice, a –approximation around the peaks works extremely well. Using this frequency peak approximation, the synchrotron emissivity can be defined as

 jν(ν,→r)=dnedEe(Ee(ν),→r)dEe(ν)dνbsyn(Ee(ν),→r). (21)

This quantity is then integrated along the line of sight for the various cases to get the final synchrotron flux across the sky.

iii.3.1 Single clump signal

According to the description of previous section II.1, let us consider a clump of mass , whose center of mass is placed at and with a sufficiently small size. In this case it is possible to neglect the spatial variation of inside the clump itself, and thus the flux can be calculated as:

 Iν(ν,→Rcl)=14πd2cl∫d→rjν(ν,→Rcl+→r), (22)

with the distance between the observer and the clump. This can be rewritten as

 Iν(ν,→Rcl)=I0νμ(→R%cl)∑kAk⎛⎝B(→Rcl)μG⎞⎠1−k2(νHz)k2 (23) Ak(mχ)=bk(mχc2GeV)−k(ν0Hz)−k2−1, (24)

with

 I0νGeV cm−2s−1 Hz−1= =2.57×10−12(mχc2100GeV)−3⟨σAv⟩10−26cm3s−1 ×(rclkpc)3(dclkpc)−2(ρclGeVc−2cm−3)2. (25)

Fig. 3 shows some examples of signal, produced by three clumps of our simulation. An important feature to notice is that the synchrotron signal sensibly depends on the magnetic field both in the normalization and in the covered frequency range. In particular, the signal frequency cutoff, remnant of the energy spectrum cutoff near , depends on following Eq.20.

Fig. 4 shows instead the positions and radio intensities for a realization of the clumps distribution with masses . It can be seen that all the clumps with a non negligible signal lie near the galactic plane where most of the galactic magnetic field is concentrated. Few clumps are visible at high latitude just because of projection effects, being located very near and slightly up or below the solar position with respect to the galactic plane.

iii.3.2 Diffuse signals

The diffuse halo signal is similarly given by the integral along the line of sight of Eq.(21)

 d2Iνdldb=cosb4π∫∞0jνds, (26)

where are coordinates on the sphere and the line of sight coordinate. To calculate the total contribution from the substructures, instead, we have to sum over all haloes

 d2Iunrνdldb=cosb∫dmcl∫dss2dncldmcl(mcl,→r)Iresν(ν,→r), (27)

with given by Eq.(23) and .

Interestingly, the sum of the two diffuse contributions can be rewritten as

 d2IDMνdldb=cosb4π∫jDMνds, (28)

where

 jDMν = 14(mχc2GeV)−3⟨σAv⟩cm3s−1⎧⎨⎩[ρh/GeVc−2cm−3(r/rh)(1+r/rh)2]2+ρCL/% GeVc−2cm−3(r/rh)(1+r/rh)2⎫⎬⎭ (29) ×μ(→r)∑kAk(mχ)(B(→r)μG)1−k/2(νHz)k/2GeV\,cm−3s−1Hz−1sr−1.

Thus, from the point of view of DM annihilation the unresolved clumps signal behaves like a further smooth NFW component with the same scale radius of the halo profile, but with a different effective density , and with an emissivity simply proportional to the density profile instead of its square.

We see that the halo component dominates in the central region of the galaxy, where

 rrh(1+rrh)2<ρ2hρCL⇒r<4.39kpc (30)

which corresponds to a disk of radius 27.3 degrees (see fig. 5).

Iv DM Annihilation constraints

The pattern and intensity of the DM radio map resulting from the sum of the contributions from the smooth halo and unresolved clumps is shown in Fig. 6 for GeV and cms. Similar maps are obtained at different frequencies and different and to obtain DM exclusion plots. For our analysis we use a small mask covering a region around the galactic center where energy loss processes other than synchrotron and ICS start possibly to be relevant. We include the galactic plane although this region has basically no influence for the constraints on the DM signal.

In Fig.7 we show the radio constraints on the DM annihilation signal in the plane for various frequencies and various choices of the foreground. Several comments are in order. First, we can see that, as expected, the use of the haze at 23 GHz gives about one order of magnitude better constraints with respect to the synchrotron foregrounds at the same frequency. However, using also the information at other frequencies almost the same constraints can be achieved. This information in particular is complementary giving better constraints at lower DM masses. This is easily understood since a smaller DM mass increases the annihilation signal () at smaller energies, and thus smaller synchrotron frequencies. In particular, the constraints improve of about one order of magnitude at GeV from 23 GHz to 1 GHz while only a modest improvement is achieved considering further lower frequencies as 0.1 GHz. This saturation of the constraints is due to the frequency dependence of the DM signal, that below 1 GHz becomes flatter than the astrophysical backgrounds so that the fraction of contribution from DM is maximal at about 1 GHz. Further, the constraints show a threshold behavior given basically by Eq.20 which settles a maximum emitted radio frequency for a given DM mass . This threshold behavior is, for example, clearly seen at 23 GHz in the right panel of Fig.7 where only for masses above GeV the cross section is constrained.

Notice that although the astrophysical background which we compare with at 1 GHz is an interpolation, the derived constraints are still valid given the smooth behavior and the broad frequency extent of the DM signal, which does not exhibit narrow peaks at particular frequencies. However, effective measurements have been performed for example at 408 MHz and 1.4 GHz (see deOliveiraCosta:2008pb ()). Quoting our constraints at these exact frequencies would change the results only slightly.

The DM signal has thus a broad frequency extent and also below 1 GHz is still relevant. This is a potential problem for the DM interpretation of the WMAP Haze given that, in the Haze extraction procedure, the observed radio emission at 408 MHz is used as template of the synchrotron background. In fact, naively, a DM signal at 23 GHz should be relevant at 408 MHz as well, unless either the DM mass or the magnetic field is so high to shift the DM contribution to higher frequencies and making it negligible at 408 MHz.

The second relevant point to notice is that the constraint depends quite sensibly on the magnetic field assumptions. The constraints we obtain with the TT model are generally almost two orders of magnitude weaker with respect to the results reported in Hooper:2008zg (). They are instead more in agreement with Grajek:2008jb () where Galprop has been employed to calculate the DM synchrotron signal. For a closer comparison with Hooper:2008zg () we choose, as they do, a constant magnetic field of 10 G although still keeping the Galprop ISRF model. Even in this case our derived constraints are a factor of 5 weaker (despite the inclusion of the contribution from substructures). The remaining factor of 5 can be finally recovered using a constant ISRF with  eV/cm as assumed in Hooper:2008zg (). In this case, in fact, the smaller values of reduces the ICS losses enhancing in turn the synchrotron signal. It should be said however that, while the magnetic field normalization is still quite uncertain, the ISRF is instead more constrained and a large variation with respect to the Galprop model seems unlikely.

The constraints shown in Fig.7 extend down to GeV, which is somewhat the mass limit for a conservative analysis. It is clear that for low masses the constraints come more and more from lower frequencies: For example for a WIMP of 30 GeV the data at 100 MHz are 2 orders of magnitude more constraining than the data at 10 GHz. However, extremely low frequencies are not experimentally accessible. For a WIMP of 1 GeV, from Eq.20 with a magnetic field of (G) only frequencies MHz would be useful to place constraints on the DM signal. Although observations at this frequency exist deOliveiraCosta:2008pb (), in general the survey sky coverage is quite incomplete and the data quality is non-optimal. Observations in this very low frequency range should substantially improve with the next generation radio arrays LOFAR and SKA. WIMP masses below 1 GeV still would produce observable synchrotron radiation at the galactic center where the magnetic field is likely much higher than G scale (possibly (mG) ). This kind of analysis would be however quite model dependent and would face further background uncertainties.

V Single clumps detectability

To have a reliable estimate of the sensitivity to a single clump detection diffusion effects cannot be neglected. Although the integrated synchrotron clump signal is given by Eq.(23), the clumps appear extended rather than pointlike with a dimension typically of several degrees. As a reasonable approximation we can assume that the signal is spread over an area of radius equal to the diffusion length of the electrons , where is the diffusion coefficient and is the energy loss time given by Eq.(17). We use for the Galprop model Moskalenko:1997gh ()

 K=K0(EeEe0)δ, (31)

with a reference energy GeV, a Kolmogorov spectrum and cm/s.

Taking as reference the parameters of a very bright clump like the #3 in table 2, we get ()

 lD= ⎷K(Ee)(EeGeV)−1μ(→x)τ0syn≈1kpc, (32)

for GeV and for a radiation density  eV/cm. The energy losses are basically dominated by ICS thus the result is almost independent of the magnetic field value. Moreover the dependence on the electron energy and the radiation density itself is very weak. Of course the clumps will have a certain profile peaked in the center and will not be perfectly smoothed all over . However the dilution of the signal in the much larger volume with respect to the region of emission makes it quite hard to detect the clump. We can consider for example the signal from a very bright clump at a distance of 5 kpc with a flux of GeV cmsHz at 20 GHz, corresponding approximately to the characteristics of clump #3 in Fig.3. With a dilution over 1 kpc, the clump emission is seen under a steradian sr with , giving a diffuse clump flux of GeV cmsHzsr. The WMAP sensitivity of about K translates into a flux sensitivity444At radio frequencies the Rayleigh-Jeans law is employed to translate fluxes into brightness temperatures of GeV cmsHzsr, meaning that the expected, optimistic signal is about 3 order of magnitude below the reach of the current sensitivity. The situation is only slightly better at 150 MHz where the expected LOFAR sensitivity is 50 mK Jelic:2008jg () i.e. GeV cmsHzsr.

The chance of clump radio detection seems thus quite poor even with the next generation experiments. On the other side, the fact that the signal is anyway extended and not pointlike makes the clump signal not really complementary to the diffuse component sharing the same systematics with a much fainter signal. It is likely thus that the a role for DM investigations in the radio will be played basically by the diffuse signal.

Vi Summary and conclusions

Using conservative assumptions for the DM distribution in our galaxy we derive the expected secondary radiation due to synchrotron emission from high energy electrons produced in DM annihilation. The signal from single bright clumps offers only poor sensitivities because of diffusion effects which spread the electrons over large areas diluting the radio signal. The diffuse signal from the halo and the unresolved clumps is instead relevant and can be compared to the radio astrophysical background to derive constraints on the DM mass and annihilation cross section.

Constraints in the radio band, in particular, are complementary to similar (less stringent but less model dependent) constraints in the X-ray/gamma band Mack:2008wu (); Kachelriess:2007aj () and from neutrinos Yuksel:2007ac (). Radio data, in particular, are more sensitive in the GeV-TeV region while neutrinos provide more stringent bounds for very high DM masses ( TeV). Gammas, instead, are more constraining for GeV. The combination of the various observations provides thus interesting constraints over a wide range of masses pushing the allowed window significantly near the thermal relic possibility.

More into details, we obtain conservative constraints at the level of  cms for a DM mass  GeV from the WMAP Haze at 23 GHz. However, depending on the astrophysical uncertainties, in particular on the assumption on the galactic magnetic field model, constraints as strong as  cms can be achieved. Complementary to other works which employ the WMAP Haze at 23 GHz, we also use the information in a wide frequency band in the range 100 MHz-100 GHz. Adding this information the constraints become of the order of  cms for a DM mass  GeV. The multi-frequency approach thus gives comparable constraints with respect to the WMAP Haze only, or generally better for GeV where the best sensitivity is achieved at GHz frequencies.

The derived constraints are quite conservative because no attempt to model the astrophysical background is made differently from the case of the WMAP Haze. Indeed, the Haze residual map itself should be interpreted with some caution, given that the significance of the feature is at the moment still debated and complementary analyses from different groups (as the WMAP one) miss in finding a clear evidence of the feature. In this respect, the multifrequency approach will be definitely necessary to clarify the nature of controversial DM signals as in the case of the WMAP Haze. Progresses are expected with the forthcoming data at high frequencies from Planck and at low frequencies from LOFAR and, in a more distant future, from SKA. These surveys will help in disentangling the various astrophysical contributions thus assessing the real significance of the Haze feature. Further, the low frequency data in particular, will help to improve our knowledge of the galactic magnetic field. Progresses in these fields will provide a major improvement for the interpretation of the DM-radio connection.

Acknowledgments

We wish to thank P.D. Serpico for valuable comments and T. Di Girolamo for a careful reading of the draft. Use of the publicly available HEALPix software Gorski:2004by () is acknowledged. G.M. acknowledges supports by the Spanish MICINN (grants SAB2006-0171 and FPA2005-01269) and by INFN–I.S.Fa51 and PRIN 2006 “Fisica Astroparticellare: Neutrini ed Universo Primordiale” of Italian MIUR.

References

• (1) E. Komatsu et al. [WMAP Collaboration], arXiv:0803.0547 [astro-ph].
• (2) G. Bertone, D. Hooper and J. Silk, Phys. Rept. 405 (2005) 279 [arXiv:hep-ph/0404175].
• (3) G. Jungman, M. Kamionkowski and K. Griest, Phys. Rept. 267 (1996) 195 [arXiv:hep-ph/9506380].
• (4) L. Bergstrom, Rept. Prog. Phys. 63 (2000) 793 [arXiv:hep-ph/0002126].
• (5) L. Bergstrom, J. Edsjo and P. Gondolo, Phys. Rev. D 58 (1998) 103519 [arXiv:hep-ph/9806293].
• (6) V. D. Barger, F. Halzen, D. Hooper and C. Kao, Phys. Rev. D 65 (2002) 075022 [arXiv:hep-ph/0105182].
• (7) E. A. Baltz and J. Edsjo, Phys. Rev. D 59 (1999) 023511 [arXiv:astro-ph/9808243].
• (8) D. Hooper and J. Silk, Phys. Rev. D 71 (2005) 083503 [arXiv:hep-ph/0409104].
• (9) F. Donato, N. Fornengo, D. Maurin and P. Salati, Phys. Rev. D 69 (2004) 063501 [arXiv:astro-ph/0306207].
• (10) L. Bergstrom, M. Fairbairn and L. Pieri, Phys. Rev. D 74, 123515 (2006) [arXiv:astro-ph/0607327].
• (11) M. Regis and P. Ullio, Phys. Rev. D 78 (2008) 043505 [arXiv:0802.0234 [hep-ph]].
• (12) T. E. Jeltema and S. Profumo, arXiv:0805.1054 [astro-ph].
• (13) P. Blasi, A. V. Olinto and C. Tyler, Astropart. Phys. 18 (2003) 649 [arXiv:astro-ph/0202049].
• (14) R. Aloisio, P. Blasi and A. V. Olinto, JCAP 0405 (2004) 007 [arXiv:astro-ph/0402588].
• (15) A. Tasitsiomi, J. M. Siegal-Gaskins and A. V. Olinto, Astropart. Phys. 21 (2004) 637 [arXiv:astro-ph/0307375].
• (16) L. Zhang and G. Sigl, arXiv:0807.3429 [astro-ph].
• (17) S. Colafrancesco, S. Profumo and P. Ullio, Astron. Astrophys. 455 (2006) 21 [arXiv:astro-ph/0507575].
• (18) S. Colafrancesco, S. Profumo and P. Ullio, Phys. Rev. D 75 (2007) 023513 [arXiv:astro-ph/0607073].
• (19) E. A. Baltz and L. Wai, Phys. Rev. D 70 (2004) 023512 [arXiv:astro-ph/0403528].
• (20) D. Hooper, D. P. Finkbeiner and G. Dobler, Phys. Rev. D 76 (2007) 083012 [arXiv:0705.3655 [astro-ph]].
• (21) D. Hooper, Phys. Rev. D 77 (2008) 123523 [arXiv:0801.4378 [hep-ph]].
• (22) P. Grajek, G. Kane, D. J. Phalen, A. Pierce and S. Watson, [arXiv:0807.1508 [hep-ph]].
• (23) J. Diemand, B. Moore and J. Stadel, Nature 433 (2005) 389 [arXiv:astro-ph/0501589].
• (24) J. Diemand, M. Kuhlen and P. Madau, Astrophys. J. 657 (2007) 262 [arXiv:astro-ph/0611370].
• (25) M. Kuhlen, J. Diemand and P. Madau, arXiv:0805.4416 [astro-ph].
• (26) V. Springel et al., arXiv:0809.0898 [astro-ph].
• (27) J. F. Navarro, C. S. Frenk and S. D. M. White, Astrophys. J. 490 (1997) 493 [arXiv:astro-ph/9611107].
• (28) B. Moore, S. Ghigna, F. Governato, G. Lake, T. R. Quinn, J. Stadel and P. Tozzi, Astrophys. J. 524 (1999) L19.
• (29) A. Klypin, H. Zhao and R. S. Somerville, Astrophys. J. 573 (2002) 597 [arXiv:astro-ph/0110390].
• (30) S. Hofmann, D. J. Schwarz and H. Stoecker, Phys. Rev. D 64 (2001) 083507 [arXiv:astro-ph/0104173].
• (31) A. M. Green, S. Hofmann and D. J. Schwarz, JCAP 0508 (2005) 003 [arXiv:astro-ph/0503387].
• (32) J. S. Bullock et al., Mon. Not. Roy. Astron. Soc. 321 (2001) 559 [arXiv:astro-ph/9908159].
• (33) L. Pieri, G. Bertone and E. Branchini, Mon. Not. Roy. Astron. Soc. 384 (2008) 1627 [arXiv:0706.2101 [astro-ph]].
• (34) P. G. Tinyakov and I. I. Tkachev, Astropart. Phys. 18 (2002) 165 [arXiv:astro-ph/0111305].
• (35) M. Prouza and R. Smida, Astron. Astrophys. 410 (2003) 1 [arXiv:astro-ph/0307165].
• (36) M. Kachelriess, P. D. Serpico and M. Teshima, Astropart. Phys. 26 (2006) 378 [arXiv:astro-ph/0510444].
• (37) X. H. Sun, W. Reich, A. Waelkens and T. Ensslin, Astron. Astrophys. 477 (2008) 573-592 arXiv:0711.1572 [astro-ph].
• (38) A. Noutsos, S. Johnston, M. Kramer and A. Karastergiou, Mon. Not. Roy. Astron. Soc. 386 (2008), 1881-1896 arXiv:0803.0677 [astro-ph].
• (39) M. Tegmark, A. de Oliveira-Costa and A. Hamilton, Phys. Rev. D 68 (2003) 123523 [arXiv:astro-ph/0302496].
• (40) A. de Oliveira-Costa, M. Tegmark, B. M. Gaensler, J. Jonas, T. L. Landecker and P. Reich, arXiv:0802.1525 [astro-ph].
• (41) D. P. Finkbeiner, Astrophys. J. 614 (2004) 186 [arXiv:astro-ph/0311547].
• (42) G. Dobler and D. P. Finkbeiner, arXiv:0712.1038 [astro-ph].
• (43) D. P. Finkbeiner, arXiv:astro-ph/0409027.
• (44) B. Gold et al. [WMAP Collaboration], arXiv:0803.0715 [astro-ph].
• (45) The map of the Haze at 23 GHz can be downloaded from http://www.skymaps.info/
• (46) K. Griest and D. Seckel, Phys. Rev. D 43 (1991) 3191.
• (47) P. Gondolo, J. Edsjo, P. Ullio, L. Bergstrom, M. Schelke and E. A. Baltz, JCAP 0407 (2004) 008 [arXiv:astro-ph/0406204].
• (48) C.T. Hill, Nucl. Phys. B 224 (1983) 469.
• (49) I. V. Moskalenko and A. W. Strong, Astrophys. J. 493 (1998) 694 [arXiv:astro-ph/9710124].
• (50) A. W. Strong and I. V. Moskalenko, Astrophys. J. 509 (1998) 212 [arXiv:astro-ph/9807150].
• (51) A. W. Strong, I. V. Moskalenko and O. Reimer, Astrophys. J. 537 (2000) 763 [Erratum-ibid. 541 (2000) 1109] [arXiv:astro-ph/9811296].
• (52) M.S. Longair, High Energy Astrophysics (2nd ed.), Cambridge University Press, Cambridge, 1982.
• (53) T. A. Porter and A. W. Strong, arXiv:astro-ph/0507119.
• (54) V. Jelic et al., arXiv:0804.1130 [astro-ph].
• (55) G. D. Mack, T. D. Jacques, J. F. Beacom, N. F. Bell and H. Yuksel, arXiv:0803.0157 [astro-ph].
• (56) M. Kachelriess and P. D. Serpico, Phys. Rev. D 76 (2007) 063516 [arXiv:0707.0209 [hep-ph]].
• (57) H. Yuksel, S. Horiuchi, J. F. Beacom and S. Ando, Phys. Rev. D 76 (2007) 123506 [arXiv:0707.0196 [astro-ph]].
• (58) K. M. Gorski, E. Hivon, A. J. Banday, B. D. Wandelt, F. K. Hansen, M. Reinecke and M. Bartelman, Astrophys. J. 622 (2005) 759 [arXiv:astro-ph/0409513].
You are adding the first comment!
How to quickly get a good reply:
• Give credit where it’s due by listing out the positive aspects of a paper before getting into which changes should be made.
• Be specific in your critique, and provide supporting evidence with appropriate references to substantiate general statements.
• Your comment should inspire ideas to flow and help the author improves the paper.

The better we are at sharing our knowledge with each other, the faster we move forward.
The feedback must be of minimum 40 characters and the title a minimum of 5 characters