Arxiv:0707.3260V3 [Astro-Ph] 18 Sep 2007 O Akmte.Silms Ftewr Sn Hs The- These Using Work the of Need Most the Still Without Observations Matter

arXiv:0707.3260v3 [astro-ph] 18 Sep 2007 o akmte.Silms ftewr sn hs the- these using work the of need most the Still without observations matter. the dark explain for [14] to Bekenstein by tried theory TeVeS that recent force more Newtonian the modified and a law using [13] Milgrom by proach fluc- density scales. sound of smaller spectrum unable of on power probably tuations speed matter it the makes a ap- reproduce which having This to times of late 12]. at 11, difficulty increasing 10, within the [9, energy faces have models dark proach gas authors and Chaplygin matter Some called dark 8]. so unify [7, to tried e.g. even see inhomogeneity local con- or ΛCDM called constant so a model. the cold cordance just constitues of this amount is sizeble matter energy a dark with dark universe. Combined models energy. the vacuum simplest of expansion the by ob- exponential the characterized In explain time mostly to late allows is that served it pressure negative clear, a less having even is energy, galaxies two of ob- collision a lensing of weak [6]. clusters aftermath and the x-ray mat- of combined dark servation the by for paradigm support there independent ter addition recent In a of interactions. been consisting weak has gravitation- by be interact perhaps to and only ally appears that and matter particles matter Dark non-relativistic mainly dark and energy. components, flat unknown dark is two universe of the consists been have that [5] [4], indicating Ia luminosity- of strongly type of the of evolution abundancies supernovae of of the observation relation the of redshift by of and parameter [3] Hubble [2], universe the the structure in scale elements radiation light large background microwave of cosmic [1], the of analysis loteehsbe h hnmnlgclMN ap- MOND phenomenological the been has there Also scale large a using energy dark to alternative an For dark component, unknown second the of nature The precision high the and observations years recent In BN acltosadWA-aafrafltuies ihdar compati with be universe a flat to particles a potential for model chemical WMAP-data standard and and all calculations mass (BBN) least for at ranges when allowed particles the of h . ie h akmte eprtr.I hscs h resul the case this In ch temperature. matter matter rest dark f m dark initial least that the the the times constrain find even to 6.3 We present us allow (LSS). are results particles data WMAP model structure standard scale all large to comparison g oodnr amdr atrmdl h atcems a ob to has mass particle the models matter dark warm ordinary to h akmte hmclptniloe eprtr ai c ratio temperature over potential chemical matter dark the eff s .) uhroew siaetefe temn eghfor length streaming free the estimate we Futhermore 0.7). = eepoeamdlfrafrincdr atrpril famil particle matter dark fermionic a for model a explore We ≤ .INTRODUCTION I. ( T 3e,weeteuprbudsae ierywith linearly scales bound upper the where eV, 53 Dec ) > ntttfu hoeicePyi srpyi,JW Goet J.W. Astrophysik, / Physik für Theoretische Institut 10 3 a-o-aeSrß ,D648Fakuta an Germany Main, am Frankfurt D-60438 1, Max-von-Laue-Straße ob optbewt bevtoso h Ly the of observations with compatible be to ) omlg ffrincdr matter dark fermionic of Cosmology ilanBekladJürgen Schaffner-Bielich and Boeckel Tillmann Dtd coe 4 2018) 24, October (Dated: xli l h etoe needn observations. independent mentioned can that the con- paradigm all no matter present explain dark at the is weak to there or alternative all structure vincing in scale large All whether CMB, unclear observations. the lensing highly for is so do it matter can scales dark they galactic for need below ap- the and different remove on to mentioned able the be may Although a proaches using galaxies. [16] relativity disk general grav- al. for linearized of et and theory Cooperstock metric new symmetric by axially a done or been matter has dark itation exotic for without curves need rotation the galactic ap- least recent at interesting explain quite to A proach the data. year to BOOMERanG first from WMAP fit extracted and includ- peak bad when acoustic rather anisotropies third the CMBR a ing the gives of density spectrum matter power model low MONDian simple prove a a matter. that to with find dark have [15] to Ref. still alternative example For the cosmological they viable explaining so an in scales being succeed these It even at anomaly. they stellar observations Pioneer whether curves, the unclear rotation and is galaxies galactic dwarf on in done velocities been has ories int h up-oepolm u e.[3 on that found [23] Ref. but solu- problem, a cuspy-core as the example matter to For dark tion self-interacting case. proposed CDM properties [22] standard matter Ref. the dark the from that different show are matter, actually baryonic these the from if effects or in- more that be example simulations/models to for realistic clude must can more problems in matter these resolved whether dark unclear be of is It was density it constant. core Instead rather the 21]). 20, that [19, found Ref. confirmed not e.g. was (see matter that observations dark galaxies, by the spiral of is and profiles dwarf model in density CDM halos inner by the cuspy found of of as problem prediction Another simulations, n-body 18]. approaches [17, numerical formation Ref. in structure as well linear as larger both orbiting halos in matter galaxies dark exces- mass an low of predicts amount currently sive it although observations the h CMmdli ucsfli xliigms of most explaining in successful is model ΛCDM The g eff s itv obndBNcluainand calculation BBN combined rictive nece unity. exceed an ( T igms ag sa ot18eV 1.8 most at is range mass ting ei qiiru.W aclt the calculate We equilibrium. in re mclptnilt ihs au of value highest a to potential emical Dec emosadatfrin oallow to antifermions and fermions l ihbgbn nucleosynthesis bang big with ble nry(Ω energy k α rdr atrdculn when decoupling matter dark or agrthan larger e .Fo S efidta similar that find we LSS From ). hc eope rmterest the from decouples which y oeta ihrdhf,btstill but redshift, high at forest eUniversitäthe Λ 0 .2 Ω 0.72, = ∼ 0 V(meaning eV 500 M 0 0.27, = ≤ 2 the interesting parameter range in which self-interacting at the GUT scale and at the time of BBN. The tempera- dark matter could flatten halo cores can probably be ex- ture of equilibrium particles therefore is shifted upwards cluded by other observations. Another solution could be compared to the temperature of decoupled particles (like warm dark matter (WDM) as studied by Ref. [24], where our dark matter particles) leading to the temperature the core densities are reduced by the higher velocity dis- relation: persion of WDM. 1/3 It is usually assumed that the dark matter constituent, gs (T ) whether it is bosonic (a candidate would be the axion) or eff Dec T = TDM s (3) fermionic (for example the neutralino, the gravitino or a geff (T ) ! massive sterile neutrino), initially has a negligible chemical potential (or to be more exact that the condition The equilibrium relations µ = −µ¯ and T = T¯ will still T ≫ µ − m is fullfilled). The reason for this assumption be fullfilled as long as both fermions and antifermions is the expectation that the matter-antimatter asymmetry are ultra-relativistic (where barred quantities refer to an- in the universe is of the same magnitude for all particle tiparticles), although they are decoupled and no inter- species. So spin statistics are mostly ignored for dark actions take place anymore. The transition from ultra- matter and the usual approach for cold dark matter is relativistic to non-relativistic behaviour of both fluids then a free Boltzmann gas. In contrast to that we will needs to be treated numerically, meaning that we can use a free Fermi gas in our approach to find constraints on start our numerical calculation at relatively late times the allowed matter-antimatter-asymmetry through the (i.e. after BBN) and simply add the energy density, related chemical potentials and the resulting mass win- pressure etc. of the underlying photon, neutrino and dow of a thermal fermionic dark matter relic from cosmo- baryon fluids where their behaviour and magnitude is logical observations. We assume an additional conserved known from standard cosmology, WMAP-data and BBN- quantum number beyond those of the standard model to calculations. Neutrinos are treated as massless for sim- keep the particle stable, but this is common in many dark plicity. matter models. We will also explore possible implications for structure formation for such a particle species.

III. ANALYTIC LIMITS FOR ENERGY II. INITIAL ASSUMPTIONS AND DENSITY AND EQUATION OF STATE MOTIVATION The energy density and equation of state (EoS) of a We assume a thermal distribution of a fermionic parti- 1 non-interacting Fermi gas with non-zero chemical poten- cle family with 4 degrees of freedom (Spin 2 ) was created tial can be calculated analytically in the ultra-relativistic asymmetrically (i.e. a surplus of particles over antiparti- as well as in the non-relativistic limit, an outline of the cles or the other way around) at an early stage somewhere 15 derivations will be given below, a more detailed descrip- around the GUT-scale (approx. 10 GeV). The assym- tion can be found in [25]. metric process then became ineﬀective and the surplus The energy density ǫ+ = ǫ+¯ǫ and pressure p+ = p+¯p and of fermions was conserved leading to a non-zero chem- net number density n− = n − n¯ of a free non-interacting ical potential of these particles. We then assume ther- Fermi gas in chemical equilibrium are given in natural mal decoupling at a later stage while it was still ultra- units (c = ~ = kB = 1) by the following integrals over relativistic. Thus the particles will be unaﬀected by the the momentum ~k of the particle: following numerous phase transitions and the subsequent reduction of the number of equilibrium particle species ∞ and their degrees of freedom gi. Assuming overall en- + g 3~ ~ 1 tropy conservation from the dark matter decoupling scale ǫ = d k E(k) − (2π)3  exp( E µ )+1 up to today demands: Z0 T ∞ gs a3T 3 = gs a3T 3 (1) 1 eff Dec eff later + d3~k E(~k) (4) exp( E−µ¯ )+1 where Z0 T  T 3 7 T 3 gs = g i + g i (2) eff i T 8 i T i=bosons i=fermions ∞ X X g k2 1 p+ = d3~k which allows to write the entropy using just one overall (2π)3  ~ E−µ 3E(k) exp( T )+1 temperature (usually the photon temperature). In eq. (2) Z0 ∞ it is assumed that all the standard model particles (or 2 3~ k 1 whatever other particles may be present at dark matter + d k − (5) 3E(~k) exp( E µ¯ )+1 decoupling) have negligible chemical potentials at least Z0 T  3

∞ cases to distinguish. Both limits refer to particles as well − g 1 n = d3~k as to antiparticles, since both fluids are still present in (2π)3  exp( E−µ )+1 Z0 T the non-relativistic regime. ∞ 1 − d3~k (6) exp( E−µ¯ )+1 1. Non-relativistic degenerate limit Z0 T  Where the energy-momentum-dependence reads: In the degenerate case with (µ−m) ≫ T , only particles E(~k) = ~k2 + m2. Here g denotes the degeneracy with a momentum higher or equal the Fermi momentum factor (i.e. in our case g = 2 for spin), not counting par- kF can be added to the system since all momentum states p ticles/antiparticles since this has already been accounted with lower energy have already been filled with particles. for by the two separate integrals. Eq. (4) simplifies then to:

kF g g m k3 A. Ultra-relativistic limit ǫ ≈ k2dk m = F (9) 2π2 6π2 Z0 In the ultra-relativistic limit E ≈ k the integrals are 1/3 6π2ǫ solvable giving us the equation of state p = ǫ/3: k = = µ2 − m2 F gm ∞ p g k3 ǫ+ ≈ dk Using eq. (9) one sees that the pressure scales as 2π2  k−µ exp( T )+1 Z0 kF ∞  g k2 g k5 3 2 F 5/3 k + p ≈ 2 k dk = 2 ∝ ǫ (10) + dk − =3p (7) 2π 3m 30π m exp( k µ¯ )+1 Z0 Z0 T  which is a polytropic equation of state of the form by using the subtitutions k − µ k − µ¯ 4 1/3 γ 5 36 π x1 = , x2 = p ≈ Bǫ with γ = and B = (11) T T 3 125 g2 m8 for the ﬁrst and the second integral, respectively. By renaming x1 and x2 to x and rearranging one ends up with the following three integrals (assuming µ = −µ¯ to 2. Non-relativistic non-degenerate limit be true as said before): ∞ In the non-degenerate case (µ − m) ≪ T , the Fermi- gT 4 2x3 Dirac-distribution function can be well approximated by ǫ+ ≈ dx 2π2  exp x +1 the Maxwell-Boltzmann-limit so that: Z0 ∞ 2 ∞ 0 g k 2 2 µ 6x µ 3 ǫ ≈ 2 dkk m exp − + dx + dx x + 2π 2mT T 2 exp x +1 T  Z0 Z0 −µ/TZ µ − m  × exp = mn (12)  T So ﬁnally we arrive at the ultrarelativistic limit of the energy density of a free Fermi gas with non-zero chemical 2 k potentials : Using the substitution x = 2mT one arrives at 2 g 7π g g 3/2 ǫ+ ≈ T 4 + µ2 T 2 + µ4 (8) mT µ − m 120 4 8π2 ǫ ≈ gm exp = mn (13) 2π T This formula can also be derived using the so-called Sommerfeld-expansion for eq. (4), which is a power series whereas a similar calculation for the pressure gives in (T/µ)2, as for m = 0 all terms above second order will vanish and one obtains eq. (8). mT 3/2 µ − m p ≈ gT exp = Tn (14) 2π T B. Non-relativistic limit Again, the equation of state is a polytrope of the form

1/3 In the non-relativistic limit E ≈ m and consequently 5 8π3 p ≈ Cǫγ with γ = and C = (15) ǫ ≈ mn (andǫ ¯ ≈ m n¯). But there are still two limiting 3 g2 m8 D2 4

Interestingly, this leads to almost the same equation of Normalizing to a0 = 1 as before we arrive at the consis- µ−m state because the exponential factor D ≡ exp T tency condition apole >aRel in the form turns out to become approximatly constant in the non- − − 2 2 1/3 relativistic limit since µ − m ∝ a and T ∝ a , which −A 3/2 we will outline below in more detail. −2/3 > ǫ0(−3A) (23) sǫ − A 0

C. Generalized Chaplygin gas which can be rewritten to 2 ǫ2/3A> (24) A polytrope resembles a generalized Chaplygin gas 0 3 which has an equation of state of the form This condition is never fullfilled since A is negative, A meaning that the gas will become relativistic before it p = − (16) ǫΓ can run into the pole. where A and Γ are normaly taken to be positive. Then the Chaplygin gas can be used as an unified model for IV. CHEMICAL POTENTIAL AND dark matter and dark energy, see e.g. in [9, 10, 11, 12]. TEMPERATURE The scaling behaviour of the generalized Chaplygin gas can be found by solving A. Ultra-relativistic behaviour dǫ 3(ǫ + p) + = 0 (17) da a After chemical decoupling of the dark matter par- which can be deduced from energy-momentum conserva- ticle/antiparticle distribution from the standard model tion. The general solution is : particle background, the net number, i.e. the surplus of particles over antiparticles of the fermion distribution in 1/(Γ+1) ǫ(Γ+1) − A a comoving volume, will be frozen in (as well as the num- ǫ = A + 0 (18) ber of particles and antiparticles separately). This can a3(Γ+1) ! be expressed by the conservation equation where ǫ is the energy density at a = 1. We will now 0 n˙ − +3Hn− = 0 (25) examine the non-relativistic Fermi gas solution in comparison. For Γ = −5/3 the solution is where H is the Hubble parameter, leading to a simple −3/2 dependence on the scale parameter a −2/3 2 ǫ = A + (ǫ0 − A)a (19) − −3 n ∝ a (26) For negative A the energy density runs into a pole at the −2/3 The ultrarelativistic limit of the net number density can scale parameter apole = −A/(ǫ − A). The energy 0 be calculated as a derivative of the pressure or the energy density of a free Fermi gas should not become infinite in a q density, using eq. (8): finite volume. The way out of this problem is to consider that the gas will become relativistic, i.e. the EoS should − ∂p 1 ∂ǫ g g change to that of an ultrarelativistic gas p = ǫ/3, which n = = = µT 2 + µ3 (27) ∂µ 3 ∂µ 6 6π2 can be estimated in the following way. First we formulate the constraint in terms of the scale parameter: can aRel showing that the net number density goes to zero for a be smaller than aPole? The point where both limits of the vanishing chemical potential as expected. Furthermore non-relativistic and the ultra-relativistic EoS approach since the net number density has to scale like a−3 for each other any value of T and µ we can deduce the ultrarelativistic ǫ p ≈ Rel ≈−Aǫ5/3 (20) scaling of the temperature and the chemical potential: Rel 3 Rel −1 −1 leads to T ∝ a and µ ∝ a (28) 1 3/2 so that ǫRel = − (21) 3A − − ǫ ∝ a 4 and p ∝ a 4 (29) for the energy density of the transition from the non- relativistic to the relativistic EoS. Now we can use the like for standard relativistic particles (with µ = 0). This non-relativistic scaling to eliminate ǫRel is of cause only true as long as the chemical potentials and temperatures of fermions and antifermions are well −3 ǫRel aRel above the mass of the particle. For a Fermi gas in chem- ≈ (22) ǫ a ical equilibrium with a surplus of particles that becomes 0 0 5 non-relativistic, particles and antiparticles annihilate un- elements are sensitive to the expansion rate of the uni- til all antiparticles are gone (exponentially suppressed verse during BBN. The expansion rate on the other hand with respect to particles, sinceµ ¯ →−m) while conserv- is determined by the energy density of relativistic parti- ing the net number per comoving volume a3 · n−. In our cles (radiation domination) during this short epoch (a case not only the net number, but also the numbers of few MeV ≥ Tphoton ≥ 30 keV) which is usually expressed particles and antiparticles are conserved separately, forc- in terms of the effective number of neutrino families Nν , ing both chemical potentials to approach the mass and for a detailed description see Ref. [3]. For standard cos- the temperatures of the non-relativistic limit to be dif- mology Nν = 3 (or to be more precise Nν =3.046, which ferent (see below for more details). is due to the small overlap of neutrino decoupling and + − e e - annihilation) while the values of Nν deduced from comparison of the observed abundances of light elements B. Non-relativistic behaviour to BBN-calculations and from WMAP data differ and have relatively large error bars. For example the au- +1.55 As soon as the temperatures of the fermion fluids ap- thors of Ref. [26] find Nν = 3.08−1.28 at 2σ error us- proach the rest mass energy the scaling of the temper- ing a combined 4He-abundance and CMB results for the atures and chemical potentials will start to be different. baryon abundance. On the other hand Ref. [27] arrives +2.7 In the following equations n,T and µ refer to particles or at Nν = 5.2−2.2 at 2σ error using CMB and large scale antiparticles, not both. From eqs. (13) and (26) we see structure data. Ref. [28] find a similar upper limit for +4.8 that in the non-degenerate limit Nν =3.2−2.2 at 2σ error using WMAP and BBN including 4He- and deuterium-abundances. In a more recent 3/2 µ − m n0 +1.1 n ≈ C T exp = (30) paper de Bernardis et al. [29] find Nν = 3.7− at 2σ 1 T a3 1.2 using CMB data, LSS, Supernova data and other inde- If we now assume that µ−m and T just scale like powers pendent measurements of the Hubble parameter, which of the scale parameter is more restictive than Ref. [27, 28]. So although Nν is in any case compatible with the standard value of Nν = 3 x y µ − m = C2/a and T = T0/a (31) it can be stressed that some of the observations seem to favour a larger value, as for example found in Ref. [27] by where C1, C2, T0, n0, x and y are constants, we arrive at including large scale structure data. We will give upper and lower bounds on the chemical potential and particle n C 0 = C T 3/2a−3y/2 exp 2 ay−x (32) mass to allow comparison to these different results. The a3 1 0 T 0 standalone BBN results from Cyburt et al. [26] are nev- or ertheless the most relevant (and most restrictive) for our approach since all the other measurements of Nν refer x−y n0 3 C2 to later stages in the evolution of the universe at which a ln 3/2 + (y − 2) ln a = (33) 2 T0 our dark matter particle will already be non-relativistic C1 T0 ! and thus does not contribute to the effective number of This can only be fullfilled for arbitrary scales if x = y = 2, relativistic particle species. meaning that Next we will estimate the allowed window for the value of the chemical potential of the dark matter particle. One T ∝ a−2 and (µ − m) ∝ a−2 (34) has to be careful with the definition of Nν at this point Note that this infers that the Boltzmann factor since the important quantity is the temperature of the exp ((µ − m)/T ) stays constant in the non-relativistic neutrino distribution, which will differ from the photon temperature at BBN. This is due to the annihilation of e+ limit. The energy density and pressure will then scale − as ǫ ≈ mn ∝ a−3 and p ≈ Bǫ5/3 ∝ a−5 in the non- and e below 0.5 MeV, which occurs after the decoupling relativistic degenerate limit. We stress that this is ex- of neutrinos at around 1 MeV, and the subsequent release actly the same scaling as that of a Boltzmann gas, where of entropy almost completely into the photon fluid. So p ≈ nT ∝ a−5 as T ∝ a−2. It is noteworthy that the we find the following constraint on the energy density of evolution of energy density and pressure with the scale a fermionic dark matter particle ∆ǫrel ≥ ǫDM : parameter does not change in both the ultrarelativistic and the non-relativistic limit when a finite chemical po- 7π2 7π2 tential is added. ∆N T 4 ≥ g · T 4 ν 120 ν DM 120 DM 1 1 + T 2 µ2 + µ4 (35) C. Big bang nucleosynthesis 4 DM DM 8π2 DM The most important constaint on a chemical potential for dark matter can be deduced from big bang nucleo- where Tν is the neutrino temperature. Using eq. (3) and synthesis (BBN) calculations. The abundances of light defining 6

to 2). For ∆Nν = 0.094 a non-zero chemical potential can be excluded, for ∆Nν < 0.094 even a non-degenerate s 4/3 4 g (TDec) fermion family can be excluded (the T -term in eq. (35) Θ ≡ eff (36) DM gs (T ≈ 1MeV) always remains!). This limit will of course be shifted to a eff ! s higher value of ∆Nν if geff is smaller at decoupling than max and consequently our reference value of 106.75. For ∆Nν = 5 [27, 28] the maximum allowed value for the chemical potential is 106.75 4/3 6.3 times the dark matter temperature. Futhermore even ΘSM = ≈ 21.3 for ∆N = 1.63 [26], the ratio µ /T can be as large 2+ 7 (2 · 2+2 · 3) ν DM DM 8 as 4.4. Figure 1 also shows the allowed chemical potential s for decoupling when all standard model particles are for geff = 50 at DM-decoupling and for hypothetical val- present and in equilibrium. We arrive at the following ues of 200 and 1000 showing that even such an enormous ´ıncrease in gs will allow a maximum chemical potential constraint on µDM /TDM : eff of only a factor of about 2 larger compared to decoupling 2 4 while gs = 106.75. ∆Nν Θ 30 µDM 15 µDM eff ≥ 1+ 2 + 4 (37) s gDM 7π TDM 7π TDM In principle geff (TDec) could be much higher than the standard model value for example due to numerous addi- Solving for the largest value of µDM /TDM that still full- tional super symmetric particles. E.g. in a gas of strings fills this relation one arives at in the very early universe there are exponentially many states present (see for example Ref. [30, 31]). As we will s µ 7π4 ∆N Θ 8 see later very large values of geff (TDec) are essential for DM = −π2 + ν + (38) this approach to work. T v 15 g 7 DM max u s DM u Of course there are other mechanisms that could lead t to a variation in the extracted effective number of neu- This constraint is of course only valid for fermions that trino families from the standard value of N = 3.046. are still relativistic at BBN. It does not apply for particles ν For example an evolving gravitational constant G could that were non-relativistic at decoupling or non-thermally also lead to a difference in the expansion rate at BBN produced. This is due to the fact that the contribution of non-relativistic particles to the total energy density at (see e.g. Ref. [32]), as it could also be the case in cosmological models with more than the standard 4 spacetime BBN must be negligible not to overclose the universe to- dimensions (see e.g. Ref. [33]). day, consequently one can hardly probe µDM /TDM for such particles via BBN. The constraints on the maximum allowed values for

V. NUMERICAL RESULTS

10 10 In the following we examine one speciﬁc parameter set of a ﬂat universe (Ωtot = 1) in our numerical treatment / T / 0 with the present day values ΩΛ = 0.72 (simple vacuum 0 0 0 energy), ΩDM = 0.236, ΩB = 0.044 and Ωγ+ν = 8.5 · − 1 1 5

10 , similar to the parameters obtained from WMAP- 0 data using the ΛCDM-model [1]. ΩX as usual denotes the fraction of the critical energy density of the respec-

max. allowed allowed max. tive energy component today. We use a Hubble param- −1 −1 eter of the value H0 =70 km s Mpc . We start our

0,1 0,1 calculation of the Fermi-integral-sets (ǫ,p,n) and (¯ǫ, p,¯ n¯) from eqs. (4), (5), (6) when the particles and antiparti-

0,01 0,1 1 cles are still ultra-relativistic and use number and energy- N momentum-conservation, eq. (17), for both fermions and antifermions separately to find the scaling of the chem- FIG. 1: Maximum allowed dark matter chemical potential ical potentials and temperatures. When the latter have over temperature as a function of the number of additional reached non-relativistic scaling behaviour with sufficient effective neutrino families for different values of the effective precision we continue using the non-relativistic expres- s number of degrees of freedom geff at decoupling. The differ- s sion of eq. (13). ent lines correspond to the cases geff (TDec) = 1000 (dashed), We add photons, three massless non-degenerate neutrino 200 (solid gray), 106.75 (solid) and 50 (dash dotted) families and baryons as particle background. To find the time dependence of the scale parameter a we use the µDM /TDM are plotted vs. ∆Nν in figure 1 (gDM is set first Friedmann-equation for a flat universe (i.e. with zero 7 curvature): gDM is in this case equal to four since particles and antiparticles contribute equally to the energy density. The 2 max a˙ 8πG lower limit on the mass (in other words ∆Nν = ∆Nν ) H2 = = ǫ (39) a 3 tot can be estimated by: 0 Finally we calculate the free streaming length for TDM ǫγ+ν Ω m ≈ 3 · T 0 DM h2 fermions and anti-fermions to find additonal constraints min γ T ǫ Ω0 γ DM TDM >m γ+ν on the particle mass and show that the scaling of the 1/3 Jeans length and Jeans mass is exactly that of ordinary 106.75 5 g ≈ 1.6 DM warm dark matter (WDM). s geff (TDec)! ∆NMax 2 Ω0 h 2 × DM eV A. Mass range 0.236 0.70 where in both cases we assume that the fermions be- For a given initial value of µDM /TDM we then use the 0 come non-relativistic at T ≈ m/3. Most of the parame- condition ΩDM = 0.236 to pin down the particle mass which is sufficient to close the universe. Figure 2 shows ter range is dominated by the contribution of the second max term in eq. (8), only for the values above µDM /TDM ≈ the allowed range of the mass, assuming ∆Nν = 5 (i.e 4.4 the third term ∝ µ4 becomes dominant. For a larger µDM /TDM ≤ 6.3), were we find 1.8 eV ≤ m ≤ 53 eV. For max effective number of degrees of freedom at decoupling the ∆Nν = 1.63 (or a maximum initial value of µDM /TDM = 4.38) the corresponding mass constraint is 4.4 eV ≤ mass window is shifted to higher masses by a factor of gs (T )/gs (SM), where gs (SM) corresponds to m ≤ 53 eV. For small initial values of µDM /TDM one eff Dec eff eff

all standard model particles. In this case the highest allowed value of the initial chemical potential will be larger 100 (see section IV C) meaning that the size of the allowed mass window will grow. It is also interesting to notice

)[eV] that the lower limit on the mass only has a slight dependence on the degrees of freedom at decoupling and is

(SM) therefore always close to 1eV for all discussed values of eff

s s geff (TDec). / g /

10 In ﬁgure 3 we show the dependence of the particle mass on the degrees of freedom at decoupling for initial val- (Dec)

eff ues of µ/T = 0.1, 1.0, 2.0, which is the most interesting s range of parameters as we shall see later when discussing

x (g x structure formation. DM

m / T = 0.1 / T = 1.0 / T = 2.0

0,1 1

Initial dark matter / T 1000

FIG. 2: particle mass suﬃcient to close the universe for the allowed range of the initial chemical potential over temperature ratio [eV] DM 4

ﬁnds the expected T -plateau due to the ﬁrst term in m 100 the ultra-relativistic energy density, see eq. (8), which is independent of the chemical potential. The critical mass is determined just by the relativistic degrees of freedom at decoupling and the expected density today (or at any other arbitrary point in the non-relativistic regime). A

good estimate of the upper limit on the mass (meaning 10 for µDM /TDM ≈ 0) is 100 1000

effective degrees of freedom at decoupling

T ǫ Ω0 m ≈ 3 · T 0 DM γ+ν DM h2 FIG. 3: dark matter particle mass suﬃcient to close the uni- max γ T ǫ Ω0 γ DM TDM >m γ+ν verse vs. the degrees of freedom at decoupling for several s 0 2 values of the initial chemical potential over temperature ratio 4 g (TDec) Ω h ≈ 51 eff DM eV g 106.75 0.236 0.70 DM 8

B. Scaling of the temperature and chemical tory for number conservation of both species (in figure 5 potential the difference is too small to be visible). The chemical potential of the antifermions,µ ¯, approaches the mass from Figures 4 and 5 show the scaling of the dark matter negative values with a similar rate as µ making it look temperature and chemical potential for initial values of like it is shooting up exponentially in the log-log-plot. µDM /TDM =5.5, 0.1 and a fermion mass of 2.54 eV and In figure 5, the case for small initial µDM /TDM , we see 53.3 eV, respectively. In the ultra-relativistic regime (i.e. that both the particle and the antiparticle chemical po- for small values of a) we recover the simple 1/a - scaling tential are temporarily below the mass and change their for µDM and TDM as expected from eq. (28). In the non- scaling only when the temperature approaches the mass. relativistic regime we find that both chemical potentials are aproaching the mass and the temperatures scale as −2 a like in eq. (13). Furthermore, one can see that the C. Equation of state

T fermion T antifermion

fermion antifermion As outlined in sections IIIA and IIIB one expects a

1000 radiation-like EoS in the ultra-relativistic limit as well as a polytropic equation of state with an exponent of 100 5/3 in the non-relativistic limit for both a degenerate and a non-degenerate Fermi gas. Actually one does not 10 expect to see the non-relativistic degenerate limit since the highest allowed initial value of µDM /TDM of 6.3 is 1 only slightly higher than the value at which the µ4-Term in (8) becomes larger than the µ2T 2-Term at a value of 0,1 4.4. So we actually examine only the medium- to non- degenerate region of a free Fermi gas. In Figure 6 the

0,01

EoS fermions EoS antifermions temperature, chemical potential [eV] potential chemical temperature,

1E-3

1E7

100000

1E-4

1E-6 1E-5 1E-4 1E-3 0,01

UR limit scale parameter a 1000

p = /3

10 ] FIG. 4: Temperatures and chemical potentials for an initial 4

µ/T = 5.5 and a mass of 2.54 eV 0,1

1E-3 pressure[eV

T fermion T antifermion 1E-5

fermion antifermion

10000

1E-7

NR limit

5/3

1E-9

1000

1E-11

1E-7 1E-3 10 100000

energy density [eV ]

100

FIG. 6: EoS for an initial µ/T = 5.5 and a mass of 2.54 eV

EoS of fermions and antifermions for an initial µ/T =5.5 1 and a mass of 2.54 eV is shown. The EoS diﬀers in the non-relativistic regime for a non-zero initial µDM /TDM temperature, chemical potential [eV] potential chemical temperature,

0,1 because the fermions are (at most) semi-degenerate while the antifermions are non-degenerate and very di-

1E-7 1E-6 1E-5 1E-4 lute (in comparison to the fermions). When comparing

scale parameter a the fermion-pressure and the antifermion-pressure in the non-reltivistic regime at the same energy density for each FIG. 5: Temperatures and chemical potentials for an initial particle-distribution the antifermions have a higher pres- µ/T = 0.1 and a mass of 53.3 eV sure because their temperature is higher for the same value of the energy density. The pressure of fermions is temperature of fermions will be higher than that of an- of course always larger than the pressure of antifermions tifermions in the non-relativistic regime, which is manda- at the same scale parameter for positive initial fermion 9

chemical potential, because the energy density of an- 1 tifermions can in fact be orders of magnitude smaller.

0,1

D. Scaling of energy densities, number densities and pressures 0,01

For a ﬂat universe the total density is equal to the crit-

1E-3 ical density at all values of the scale parameter ǫtot(a)= ǫcrit(a). This can be used to deduce the scaling of the in- dividual normalized energy densities (usually denoted by 1E-4

Ω). For example the normalized radiation energy density Fraction of the critical density is given by ǫ 1E-5 γ+ν 1E-7 1E-6 1E-5 1E-4 1E-3 0,01 0,1 1 Ωγ+ν = (40) ǫγ+ν + ǫ +¯ǫ + ǫB + ǫΛ scale parameter a where ǫ is the energy density of fermions andǫ ¯ is the FIG. 8: As ﬁg. 7 but for an initial µ/T = 0.1 one of antifermions. We show the scaling of the mean

1 In Figure 9 we show an example of the scaling of energy density, pressure and number density of fermions and antifermions an initial µDM /TDM = 5.5. In the 0,1 ultra-relativistic limit we recover radiation like scaling of energy density and pressure being proportional to a−4,

0,01 in the non relativistic limit the energy densities scale like matter (ǫ ∝ a−3) and the pressure drops rapidly (p ∝ a−5). The number densities always have a slope of 1E-3 −3 as demanded by number conservation of particles and antiparticles.

1E-4 Fraction of the critical density

1E8

1E-5

1E-7 1E-6 1E-5 1E-4 1E-3 0,01 0,1 1 10.000

scale parameter a ] 3

1 [eV FIG. 7: Fractions of the critical density for an initial µ/T = 5.5, photons and neutrinos (solid gray), dark matter(fermions:

1E-4 and n and ]

solid, antifermions: dash dotted gray), baryons (dash dotted) 4

and dark energy (dashed) [eV

1E-8 , p , energy densities normalized to the critical density in ﬁg- 1E-12 ures 7 and 8 for the cases µDM /TDM = 0.1 and 5.5. The plots show nicely what is often called the ”coinci- 1E-16 dence problem”, namely that dark energy seems to to 1E-6 1E-5 1E-4 1E-3 0,01 0,1 1

1 scale parameter a become dominant just below a redshift z = a − 1 ≈ 0.5. On can see that the contribution from fermions and antifermions is constant during radiation domination until FIG. 9: Evolution of the energy density (solid), pressure they become non-relativistic and can constitute a sub- (dashed) and number density (dash dotted) of fermions (black) and antifermions (gray) for an initial µ/T = 5.5 and stantial fraction of the total energy budget in the very a mass of 2.54 eV early universe. Then their normalized energy densities grow linearly with a until non-relativistic particles dominate. In contrast to that cold dark matter would in these plots scale just like the baryon density and exactly coin- cide with the scaling behaviour of the sum of the fermions E. Age of the universe and antifermion contributions in the matter dominated 0 0 regime, i.e. the CDM contribution is highly suppressed It is generic of the cases studied (ΩΛ = 0.72, ΩDM = at early times. 0.236) to produce an almost fixed age of the universe, 10 independent of the initial chemical potential over tem- which is normaly in between the scale where particles be- perature ratio of the dark matter component. This is come non-relativistic and the scale where matter starts due to the fact that the dark matter has to become non- to dominate. The mean velocity of fermions can be cal- relativistic before the decoupling of photons and baryons culated from the distributions function occurs at around 380 000 years after the big bang. So ∞ gDM k 1 any increase in the energy density at an earlier stage will hvi = d3~k (43) (2π)3n ~ exp( E−µ )+1 only reduce the age of the universe by less than 380 000 Z0 E(k) T years due to an increased Hubble parameter at earlier The corresponding mass enclosed in a sphere of radius times. λFS is the free streaming mass: 0 0 So for ΩΛ = 0.72, ΩDM = 0.236 we find the age of the 4π 3 universe is the well known 13.7 Gyrs (as in ΛCDM), for MFS = λFSρ¯M (44) 0 3 example in a flat ΩM = 1 universe the age would be only 9.3 Gyrs(!). The latter case (although already being ex- From the comparison of numerical n-body simulations of cluded by WMAP, if the spectrum of initial fluctuations warm dark matter and cold dark matter Ref. [40] claim is a simple power law [1]) has the common problem of all that a (thermal) warm dark matter particle of a mass be- flat matter dominated models to be in violation of some low 750 eV would be incompatible with observations of independent lower limits on the age of the universe. This the Ly α forest at a redshift of z ≈ 3. On the other hand, is due to the today well messured value of the hubble pa- Boehm et al. [41] find with higher resolution simulations rameter of around 70 km s−1 Mpc−1. For example from that warm dark matter particle masses down to 600 eV stellar evolution [34] one finds a lower limit of 10 Gyrs on would produce a power spectrum that is virtually indis- the age of ω Cen and Ref. [35] finds M92 to be at least 13 tinguishable from a simple cold dark matter spectrum be- Gyrs old. From the cooling of white dwarfs in globular low z ≈ 2. This effect arises from the fact, that as soon as clusters [36] derives an age of more than 10.3 Gyrs (even large scale fluctuations go non-linear, small scale fluctua- ata2σ lower limit) on the age of M4, while in [37] NGC tions will start to grow non-linear as well and quickly re- 6397 is found to be 11.47±0.47 Gyrs. generate an initially suppressed power spectrum on small A significantly lower value of the hubble parameter (h< scales. The conclusion of [41] is that the primordial power 0.5) would be needed to solve this age problem and reach spectrum could have been exponentially surpressed ini- 9 an age of more than 13 Gyrs for a flat matter dominaded tially up to mass scales of ∼ 10 M⊙ without being ob- universe also. servable in large scale structure surveys today. We will now use this upper limit on the free streaming mass to find a lower limit on the particle mass or the degrees of F. Structure formation freedom at decoupling using the prescription of eq. (42).

Now we will discuss more restrictive limits on the particle mass and degrees of freedom at decoupling which can / T = 0.1 / T = 1.0 / T = 2.0 be deduced from the shape of the matter power spec- 10 trum. In the second part of this section we will show that the general scaling of the relevant quantities for linear structure formation do not change when adding a

ﬁnite chemical potential. 1

1. Free streaming

For collisionless particles (that have decoupled while 0,1 being ultra relativistic) the most important damping free streaming lenght [Mpc] scale is the free streaming length, as found by [38]. The comoving free streaming length is given by t hv(t′)i λ = dt′ (assuming t ≪ t) (41) 100 1000 FS a(t′) Dec ef f ective degrees of f reedom at decoupling Z0 which is basically the distance a particle can travel in FIG. 10: Free streaming length of fermions vs. the degrees of an expanding background since decoupling. Interestingly freedom at decoupling for several values of the initial chemical Bond and Szalay [39] found that the relevant comoving potential over temperature ratio damping scale from free streaming of collisionless particles can be estimated by In Figure 10 we show the resulting free streaming hv(t)i t length and free streaming mass of fermions as a func- λ =3.2 (42) FS a(t) tion of the degrees of freedom at decoupling for several Max

/ T = 0.1 / T = 1.0 / T = 2.0 magnitude. [As a side remark: of course for a Majorana s 1E14 fermion geff (TDec) does only need to be half as large to reach the same mass as for a Dirac fermion with µ/T ≈ 0]

1E13

] So structure formation gives far more restrictive

SUN bounds on the initial µ/T , namely that the ratio should 1E12 not exceed unity by far (keeping in mind that BBN al- s 3 1E11 lowed µ/T to be up to 14 for geff (TDec) = 10 ). Nev- ertheless one can in principle only give a lower bound 1E10 s on geff (TDec) since the free streaming mass is mostly

1E9 ﬁxed by the particle mass (as one can see in Figure s 12), which increases linearly with geff (TDec), meaning 1E8 that higher values of µ/T could be possible for even free streaming mass [M higher values of the degrees of freedom at decoupling 1E7 s (geff (TDec) > 4000). The slight decrease in the free

100 1000 streaming mass for rising µDM /TDM at ﬁxed mDM is

ef f ective degrees of f reedom at decoupling s caused by the increase in geff (TDec) in order to keep 0 s ΩDM constant. A higher geff (TDec) means a lower tem- FIG. 11: Free streaming mass of fermions vs. the degrees of perature relative to the temperature of equilibrium par- freedom at decoupling for several values of the initial chemical ticles, meaning the dark matter particles become non- potential over temperature ratio relativistic at a smaller scale parameter aNR. Remem- bering eqn. (3) this means

/ T = 0.1 / T = 1.0 / T = 2.0 1E15 TDM s −1/3 aNR ∝ ∝ geff (TDec) (45) 1E14 Tγ TDM >m ]

1E13

SUN at a ﬁxed dark matter particle mass, which explains the small reduction in MFS. 1E12 If the dark matter distribution should not exactly coin-

1E11 cide with the baryon distribution up to such small scales

as measured by the Ly α data (possibly measurable by

1E10 future weak lensing surveys) then this could of course

1E9 relax the given restrictions. Another way of giving a lower limit on the mass of 1E8 free streaming mass [M dark matter candidates is the dark matter phase space 1E7 density in galaxies. A more realistic version of the origi- nal Tremaine-Gunn-limit [42] takes into account changes 1000000

100 1000 of the initial particle distibutions functions during col- m [eV] DM lapse to a galaxy halo as discussed by Madsen [43]. He finds a lower limit on the mass of a fermionic dark mat- FIG. 12: Free streaming mass vs. the fermion mass for several values of the initial chemical potential over temperature ratio ter particle of ∼ 380 eV, which is less restrictive than the limits from the shape of the matter power spectrum. values of the initial ratio of µ/T . The free streaming 2. Linear structure formation length and mass is the same for antifermions at vanishing chemical potential and slightly lower for increasing The simplest approach to describe the onset of struc- chemical potential because antifermions will then become ture formation after matter radiation equality is to look non-relativistic earlier as we have seen in Figure 4. Since at the linear equation for the density contrast of mat- the relative number of antifermions is also suppressed in ter δ(t, ~x) in comoving coordinates (~x = ~r/a) which is the latter case the total damping scale should still be defined by ρ(t, ~x)=ρ ¯(t)(1+ δ(t, ~x)).ρ ¯ is the mean well described by the damping scale of fermions. One mass density of the corresponding matter component. In can also see that even for a µ/T = 2.0 one can stay 9 fourier space the evolution of the density contrast with below the threshold of ∼ 10 M⊙ although this requires time can be found by solving more than 2000 degrees of freedom at decoupling. Inter- estingly, even for a vanishing chemical potential a value k2 v2 s 3 ¨ ˙ of g (T ) ≈ 10 (corresponding m ≈ 500 eV sim- δk +2Hδk + 2 − 4πGρ¯ δk = 0 (46) eff Dec DM a ! ilar to the lowest mass value used in Ref. [41]) is needed to be in agreement with Ly α measurements, which al- where δk(t) corresponds to the density contrast for one ready exceeds the standard model value by an order of specific wave number k describing density pertubations 12

10000 at the comoving lengthscale λcom =2π/k or the physical lengthscale λphys = 2πa/k. For a non-relativistic collisionless ﬂuid the velocity dispersion v2 replaces the > 0.236

DM square of the speed of sound v2 of a collisional ﬂuid and s is in our approach given by (see Ref. [44] and section III) /T =1

/T =2

/T =3

1000 [eV]

−4/3

3¯p 9 ∂p¯ t /T =4 2 −2 DM

v ≈ = ∝ ∝ a (47) m /T =5

mn¯ 5 ∂ρ¯ t 9 eq /T =6

> 10 M

FS SUN

On the other hand

2 ∂p¯ 5A 2/3 m < 380 eV

v = = ρ¯ DM s ∂ρ¯ 3

100

100 1000 10000 for a collisional non-relativistic fermion ﬂuid. One can see effective degrees of freedom at decoupling here that the velocity dispersion (or the speed of sound) s drops like for an ordinary warm dark matter scenario FIG. 13: geff (TDec) vs. mDM parameter plane. Shaded re- (Ref. [40, 41, 44, 45]) contrary to the generalized Chap- gions are excluded by present day energy density (diagonal lygin gas which has a late time growing speed of sound. pattern), by the Tremaine-Gunn-limit (horizontal pattern) or Now inserting the velocity dispersion into eq. (46) one by a too high free streaming mass (vertical pattern). ﬁnds the critical physical wavelength, the Jeans length,

1/3 2 s 4 πhv i t 1/2 g (TDec)=5 · 10 . The region with the diagonal pat- λJ = ∝ ∝ a (48) eff Gρ¯ t tern is excluded because of a too high present day energy s eq density (ΩDM > 0.236 for h = 0.7) even for an initial The corresponding Jeans mass reads: µ/T = 0. The border line is just the case of vanishing chemical potential (compare to eqn. (40)) and therefore 3 5/2 2 3/2 −1 4π λJ π hv i t −3/2 also applies to an ordinary warm dark matter relic. For MJ = ρ¯ = 3/2 1/2 ∝ ∝ a 3 2 6 G ρ¯ teq all points below this line there is always one combina- s (49) tion of mDM , geff (TDec) and µ/T that gives the right Hence, the scalings derived here turn out to be the same present day energy density. The dashed lines show these as for standard WDM or CDM. In principle one has right combinations for several initial values of µ/T from s to do a full calculation of the linear power spectrum of 1 to 6. To make this clear, a (mDM , geff (TDec)) point on density fluctuations in this model, but we expect results the µ/T = 1 line gives the right ΩDM =0.236 today for a similar to warm dark matter models, although the final fermion with µ/T =1 and a too high ΩDM for µ/T > 1 shape of the spectrum may be a bit more complicated and vice versa for µ/T < 1. A free streaming mass of due to the interplay of two (more or less) different dark more than ∼ 109 solar masses is excluded for warm dark matter components. matter as found by [41]. The region excluded by this constraint is the area with the vertical pattern. Note that the limiting line has been extrapolated beyond the ΩDM limit and values of µ/T above 6. The region with the VI. CONCLUSION horizontal pattern is excluded by the Tremaine-Gunn- limit on the phase space density of fermions in galaxies We have shown that a relic fermionic particle family as found by Madsen [43], which is independent of the de- with sizeable chemical potential cannot be excluded even grees of freedom at decoupling. Fermion masses of less from the most recent BBN-calculations and large scale than 380 eV are excluded, but as one can see this limit s 4 structure data. Using different constraints on the energy becomes relevant only for geff (TDec) > 10 . The con- density at BBN we find that even for decoupling when straints from BBN are actually so much less restrictive s g (TDec) is equal to the maximum standard model that they lie way outside of this plot. Remember that eff s value of 106.75 the initial ratio of µ/T could have been as for geff (TDec) = 106.75 the upper bound on the initial s 3 large as 4.4 (when using the results of [26]), 4.5 (for the µ/T was 4.4 (for ∆Nν < 1.63) and for geff (TDec)=10 limits of [29]) or 6.3 (according to [27, 28]). In this case it would be even µ/T ≤ 10.7. Now note that the limit on the resulting mass range is at most 1.8 eV ≤ m ≤ 53 eV, the free streaming mass rules out a non-zero µ/T even for s s 3 where the upper bound scales linearly with geff (TDec). geff (TDec) = 10 . This case would require for example s −3 In figure 13 we show the combined results on geff (TDec) only a tiny additional ∆Nν ≈ 5 · 10 to the number of and mDM from ΩDM , the free streaming mass MFS effective neutrinos families at BBN. The BBN constraints s and the Tremaine-Gunn-limit up to very high values of become even less restrictive with growing geff (TDec) and 13 thus do not appear in this summary plot. As one can see velocity dispersion, the Jeans length and the Jeans mass from eqn. (38) the maximum allowed value of µ/T grows all have the same scaling as for regular cold or warm dark s 1/3 s as (geff (TDec)) for large values of geff . So to summa- matter after adding a finite chemical potential. rize these results, we only find absolute lower limits on An interesting feature about the model studied is prob- the fermionic dark matter mass and the effective degrees ably that two different dark matter components can of freedom at dark matter decoupling. The initial ratio emerge from the same source, that do not necessarily of chemical potential over temperature can be bounded clump with the same magnitude in the same region of s from above for a given value of geff (TDec) but no abso- space. This is due to a different temperature and veloc- lute upper limit has been found. ity dispersion in the non-relativistic regime. A detailed We have seen that at least an order of magnitude more numerical simulation of structure formation in such an effective degrees of freedom than the standard model approach would be necessary to make a clear statement value are needed to allow a fermionic dark matter parti- about differences in the matter power spectrum today cle to be the dominating dark matter component (even compared to a warm dark matter particle with the same with vanishing chemical potential) and this is also the mass but zero initial chemical potential. case for an ordinary warm dark matter relic. In a gas of Finally, it is important to keep in mind that it is of strings in the very early universe there are exponentially course assumed that the dark matter particle was in many states present (see for example Ref. [30, 31]), so statistical equilibrium at some point in the evolution of this would be a possible scenario to account for the very the universe, otherwise it will not follow a Fermi-Dirac s high values of geff (TDec) that are needed. distribution. For example in the already mentioned It is noteworthy that the limits on mDM from the free string gas cosmology approach such an early period streaming mass and from the phase space limit cross, of equilibrium would naturally occur. No assumption implying that the two main problems of CDM that stan- was made regarding the production mechanism and the dard warm dark matter can probably only resolve with source of the matter-antimatter asymmetry. The scope two different WDM particle masses (i.e. the excessive of this work was primarily to determine if a fermionic amount of small structures and the cuspy core issue in dark matter particle could in principle have a sizable standard CDM, see e.g. Ref. [44]) may be resolvable in chemical potential using cosmological observations and the fermionic dark matter approach without this discrep- not to motivate such a relic from an underlying theory. ancy. s Note that decoupling at geff (TDec) > 100 needs a super-weak interaction between dark matter and standard model particles, therefore constraints from cooling of supernovae and red giants that apply for light weakly interacting particles or axions do not apply here. Acknowledgments We have calculated the scaling of temperature, chemical potential, equation of state, energy density, number We would like to thank Jens Niemeyer, Stefan Hof- density and pressure of fermions and antifermions and mann and Robert Brandenberger for constructive dis- have shown that even their temperatures and equations cussions and critical remarks. We are grateful to Paolo of state (as well as all other mentioned quantities) will Salucci for bringing our attention to the cuspy-core issue. be different in the non-relativistic regime for an initially J. Schaffner-Bielich would like to thank Richard Cyburt non-zero chemical potential. We demonstrated that the for conversations and insights about his BBN results.

[1] D. N. Spergel, R. Bean, O. Dor´e, M. R. Nolta, C. L. Ben- [7] H. Alnes and M. Amarzguioui, Phys. Rev. D75, 023506 nett, J. Dunkley, G. Hinshaw, N. Jarosik, E. Komatsu, (2007), astro-ph/0610331. L. Page, et al., Astrophys. J. Supp. 170, 377 (2007), [8] H. Alnes, M. Amarzguioui, and O. Gron, Phys. Rev. astro-ph/0603449. D73, 083519 (2006), astro-ph/0512006. [2] M. Tegmark, D. J. Eisenstein, M. A. Strauss, D. H. Wein- [9] L. Amendola, F. Finelli, C. Burigana, and D. Carturan, berg, M. R. Blanton, J. A. Frieman, M. Fukugita, J. E. JCAP 0307, 005 (2003), astro-ph/0304325. Gunn, A. J. S. Hamilton, G. R. Knapp, et al., Phys. Rev. [10] O. Bertolami, in ESA Special Publication, edited by D 74, 123507 (2006), astro-ph/0608632. F. Favata, J. Sanz-Forcada, A. Gim´enez, and B. Battrick [3] G. Steigman, Int. J. Mod. Phys. E15, 1 (2006), astro- (2005), vol. 588 of ESA Special Publication, pp. 343–+. ph/0511534. [11] V. Gorini, U. Moschella, A. Kamenshchik, and [4] P. Astier et al. (The SNLS), Astron. Astrophys. 447, 31 V. Pasquier, in General Relativity and Gravitational (2006), astro-ph/0510447. Physics, edited by G. Espositio, G. Lambiase, G. Marmo, [5] A. G. Riess et al., Astrophys. J. 656 (2007), astro- G. Scarpetta, and G. Vilasi (2005), vol. 751 of American ph/0611572. Institute of Physics Conference Series, pp. 108–125. [6] D. Clowe et al., Astrophys. J. 648, L109 (2006), astro- [12] A. A. Sen and R. J. Scherrer, Phys. Rev. D72, 063511 ph/0608407. (2005), astro-ph/0507717. 14

[13] M. Milgrom, Astrophys. J. 270 (1983). hep-th/0703173, March 2007. [14] J. D. Bekenstein, Phys. Rev. D70, 083509 (2004), astro- [31] R.H. Brandenberger, A. Nayeri, S. P. Patil, and ph/0403694. C. Vafa. String Gas Cosmology and Structure Forma- [15] A. Slosar, A. Melchiorri, and J. I. Silk, Phys. Rev. D 72, tion. hep-th/0608121, August 2006. 101301 (2005), astro-ph/0508048. [32] J. P. Kneller and G. Steigman, Phys. Rev. D 67, 063501 [16] F. I. Cooperstock and S. Tieu (2006), astro-ph/0610370. (2003), astro-ph/0210500. [17] A. Klypin, A. V. Kravtsov, O. Valenzuela, and F. Prada, [33] L. Randall and R. Sundrum, Phys. Rev. Lett. 83, 4690 Astrophys. J. 522, 82 (1999), astro-ph/9901240. (1999), hep-th/9906064. [18] B. Moore, S. Ghigna, F. Governato, G. Lake, T. Quinn, [34] B. Chaboyer and L. M. Krauss, Astrophys. J. 567, L45 J. Stadel, and P. Tozzi, Astrophys. J. L. 524, L19 (1999), (2002), astro-ph/0201443. astro-ph/9907411. [35] N. E. Q. Paust, B. Chaboyer, and A. Sarajedini, Astro- [19] G. Gentile, P. Salucci, U. Klein, D. Vergani, and phys. J. 133, 2787 (2007), astro-ph/0703167. P. Kalberla, MNRAS 351, 903 (2004), astro-ph/0403154. [36] B. M. S. Hansen, H. B. Richer, G. G. Fahlman, P. B. [20] P. Salucci (2007), arXiv:0707.4370. Stetson, J. Brewer, T. Currie, B. K. Gibson, R. Ibata, [21] G. Gentile, P. Salucci, U. Klein, and G. L. Granato, R. M. Rich, and M. M. Shara, Astrophys. Supp. 155, MNRAS 375, 199 (2007), astro-ph/0611355. 551 (2004), astro-ph/0401443. [22] Davé, R., Spergel, D. N., Steinhardt, P. J., & Wandelt, [37] B. M. S. Hansen, J. Anderson, J. Brewer, A. Dotter, B. D. 2001, Astrophys. J. , 547, 574 G. G. Fahlman, J. Hurley, J. Kalirai, I. King, D. Reitzel, [23] Hennawi, J. F., & Ostriker, J. P. 2002, Astrophys. J. , H. B. Richer, et al. (2007), astro-ph/0701738. 572, 41 [38] C. Boehm and R. Schaeffer, A&A 438, 419 (2005), astro- [24] Bode, P., Ostriker, J. P., & Turok, N. 2001, Astrophys. ph/0410591. J. , 556, 93 [39] J. R. Bond and A. S. Szalay, Astrophys. J. 274, 443 [25] I. Sagert, M. Hempel, C. Greiner, and J. Schaffner- (1983). Bielich, Eur. J. Phys. 27, 577 (2006), astro-ph/0506417. [40] V. Narayanan, D. Spergel, R. Davé, and C. Ma, Astro- [26] R. H. Cyburt, B. D. Fields, K. A. Olive, and E. Skillman, phys. J. 543, L103 (2000). Astropart. Phys. 23, 313 (2005), astro-ph/0408033. [41] C. Boehm, H. Mathis, J. Devriendt, and J. Silk, MNRAS [27] G. Mangano, A. Melchiorri, O. Mena, G. Miele, and 360, 282 (2005). A. Slosar, JCAP 3, 6 (2007), astro-ph/0612150. [42] S. Tremaine and J. E. Gunn, Phys. Rev. Lett. 42, 407 [28] V. Barger, J. P. Kneller, P. Langacker, D. Marfatia, (1979). and G. Steigman, Phys. Lett. B 569, 123 (2003), hep- [43] J. Madsen, Phys. Rev. D 64, 027301 (2001), astro- ph/0306061. ph/0006074. [29] F. de Bernardis, A. Melchiorri, L. Verde, and R. Jimenez [44] C. J. Hogan (1999), astro-ph/9912549. (2007), arXiv:0707.4170. [45] A. Knebe, J. E. G. Devriendt, B. K. Gibson, and J. Silk, [30] R.H. Brandenberger. String Theory, Space-Time MNRAS 345, 1285 (2003), astro-ph/0302443. Non-Commutativity and Structure Formation.