4.1 Eetive number of neutrinos
4.1.1 ALP bounds
At the beginning of this hapter we have desribed some limiting ases to
un-derstand the matters related to dilution and photon injetion. The key quantity
was
T
. But onsidering the large ALP parameter spae and all the intermediate ases it inludes, it is unavoidable to numerially solve the problem. The ativespeiestoonsiderare photons,eletronsand neutrinos. Eletrons atsas
messen-gers between photons and neutrinos. Photonsthermalise veryfast with eletrons,
whih,throughweakreations,heattheneutrinostokeeptrakofthetemperature
hangesinthe eletromagnetibath. Themostrelevantoftheseenergy
redistribu-tion proesses is
e + e − → νν ¯
. We negletsattering proesses suh ase ± ν → e ± ν
,sinethey annothangetheneutrinonumberand arelesseetiveintransferring
energy to the neutrino bath 1
. The energy ow from ALPs to neutrinos an then
be modelled by a set of Boltzmann equationsfor omoving energies, dened as
X i = ρ i R 4 ,
(4.10)1
Assume that eletronshave temperaturelarger than neutrinos like in our ase. Then the
speedoftheenergytransferperunitvolumeinsatteringproessesisproportionalto
h δE i T e 4 T ν 4
,with
h δE i ∼ (T e − T ν )
athermal-averagedenergytransferpersattering. Forannihilationsitis∝ T e (T ν 8 − T e 8 )
,whihismuh moresensitivetononequilibrium situations.andtheALPphasespaedistributionfuntion
f (K φ )
asafuntionoftheomovingmomentum,
K φ = Rk φ
,d
d t f (K φ ) = − (c γ + c P )(f − f eq ) ,
(4.11a)d
d t (X γ + X e ) = 3Hδp e R 4 +
Z d 3 K φ
(2π) 3 W φ (c γ + c P )(f − f eq ) + Γ eν
R 4
h c e X ν 2 e − X ν eq 2 e + c µτ
X ν 2 µτ − X ν eq 2 µτ i
,
(4.11b)d
d t X ν e = − Γ eν
R 4 c e X ν 2 e − X ν eq 2 e
,
(4.11)d
d t X ν µτ = − Γ eν
R 4 c µτ
X ν 2 µτ − X ν eq 2 µτ
,
(4.11d)d
d t R = 1 R
s 8π
3m 2 Pl X γ + X e + X ν e + X ν µτ + ρ φ R 4
,
(4.11e)where
ω φ = q
k φ 2 + m φ 2
is the ALP energy and
W φ = ω φ R
. The set ofequa-tions (4.11) desribes the evolution of the omoving energy density stored in
γ
togetherwith
e ±
, thethreespeies ofν
's,theALPsand theosmisale fatorR
.Thesolutionofequation(4.11a)isthe distributionfuntion
f
forthepseudosalar partile. The ollisionterms for the deay and Primako proesses are [1℄c γ = m 2 φ − 4m 2 γ m 2 φ
m φ
ω φ
1 + 2T k φ
log 1 − e −(ω φ +k φ )/2T 1 − e −(ω φ −k φ )/2T
1
τ ,
(4.12)c P ≃ α g φ 2
16 n e log
"
1 + [4ω φ (m e + 3T )] 2 m 2 γ [m 2 e + (m e + 3T ) 2 ]
#
.
(4.13)Sinetheenergystoredineletronsandpositronsistransferreddiretlytothe
pho-tonbath,wekept
X γ
andX e
togetherinequation(4.11b). Thetermδp e = p e − ρ e /3
aounts of the omoving energy density gain as eletrons beome inreasingly
non-relativisti,experiening the transitionfrom
ρ e ∝ R −4
toρ e ∝ R −3
. Here,p e
and
ρ e
are the pressure and energy density ofe ±
. The seond term in the righthand side rekons up the photon-ALPenergy exhange, while in the lastlast line
of equation (4.11b) are the neutrino-eletron interations. The equations (4.11)
and(4.11) provide the evolutionof neutrino omovingenergies. Neutrinoavour
inuenetheeletron-neutrinoenergyexhange rate. Wehaveseparatedthe three
urrent interations with eletrons. The rst is the ase of
X ν e
for eletronneu-trinos, while in the seondone we have
X ν µτ
for muon and tau avours together.The energy exhange rate is given by the fator
Γ eν ≡ G 2 F T γ
, withG F
the Fermionstant, multiplied by
c e ≃ 0.68
for the eletron avour andc µτ ≃ 0.15
for themuon and tau ones, whih follow from the appropriate thermally averaged ross
setion. Weassumethatneutrinosalwayshaveathermaldistribution,determined
onlybyaneetivetemperature,whihshouldbeareasonablerstapproximation
and aurate enough for our purposes. We negletthe energy reshuing between
dierent neutrino speies, whih does not inuene the total neutrino density at
leading order. Finally, through equation (4.11e) we alulate the expansion rate
ofthe universethrough theevolutionoftheosmisalefator
R
. Theinitialon-ditions are speiedat
T ≫ MeV
byhaving allspeies at aommontemperature and the ALPnumberdensity given by equation (2.53).Forvalues
g φ . 10 −7 GeV −1
,thePrimakoproessisdeoupledinthetemper-aturerangeof interestandanbenegleted. Iftheinverse deay isalsonegligible,
weanintegratethe ALPphasespaedistributionexpliitlyanddiretlyompute
the evolutionof the number density
d
d t n φ R 3
= − n φ R 3
τ
(4.14)instead of using equation (4.11a). In this way we reover the exponential deay
law
N φ ∝ e −t/τ
. The integralin equation (4.11b)is thenZ d 3 K φ
(2π) 3 W φ (c γ + c P )(f − f eq ) ≃ m φ n φ R 4 /τ .
(4.15)WehavesannedtheALPparameterspaein[2℄,andwepresentourresultsfor
N eff
intheg φ
τ
andm φ
τ
planes ingure4.3. Werequired astandard osmologysenarioattemperaturesbelowthestandardmatter-radiationequality
T mr ∼ 1
eV.Thuswe have analysedonlythe part ofparameter spaewhere
T d > T mr
,inordertonot perturbthestandardpiturewiththedeay. Fromequation2.63,weobtain
a mass lower bound for the validity of our alulations
g φ
GeV −1 & 10 −3 eV
m φ 3/2
,
(4.16)0.01 0.1
0.5 0.5
1
1.5 2.11
2.3
2.4 2.5
N eff 3
- 5 -3 -1 1 3 5
- 3 - 1 1 3 5 7 9
Log 10 m Φ @MeVD Log 10 Τ @ s D
(a)
0.01
0.1
0.5
0.5 1
1.5
2.11
2.3
2.4 2.5 3
N eff
-14 -12 -10 -8 -6 - 4
- 3 - 1 1 3 5 7 9
Log 10 g Φ @GeV -1 D Log 10 Τ @ s D
(b)
Figure4.3: Contourplotsoftheeetivenumber ofneutrinos
N eff
asafuntionoftheALPmassandlifetime(left)andoftheouplingparameter
g φ
andlifetimeτ
(right). ALP osmologiesleading toN eff < 2.11
an be safelyexluded.whih largely overlaps with the limits obtained from CMB spetral distortions
obtained inthe previous hapter.
The sharp features of the isoontours in gure 4.3are due to the abrupt and
sizeable derease of
g ∗S
during the QCD phase transition. The reli abundane of ALPsdepends ong φ
through the freezing out temperature, and for ALPs with oupling aroundg φ ∼ 10 −7.5 GeV −1
we haveT fo ≃ Λ QCD
from equation (2.58).In a small range of
g φ
the ALP abundane onsiderably hanges, aording if the deoupling is late enough for ALPs to share the entropy transferred fromtheQCDdegreesoffreedomornot. ThelargeristheALPabundane,thelargeristhe
subsequent entropy release at the time of the deay, hene the peuliar shape of
theisoontours. Ingure4.3a weobservethesame features,asthe
g φ
dependeneishidden in
τ
.In gure 4.4, we present some illustrativeexamples depiting the evolution of
the
X
s of eletrons, neutrinos and ALPs as a funtion of the temperature. Note that in all of them, when ALPs beome non-relativisti, the ratioX φ /X γ
risesbeause it beomes proportional to
m φ /T
, until the age of the universe beomesomparable with
τ
.Figure 4.4a shows a typial ase of very early deay of a massive ALP, when
neutrinos are still partially oupled to eletrons. This example depits the ALP
behaviourinthe lowerrightornerofgure4.3a,andorrespondinglyinthelower
leftoneofgure4.3b. Here
N eff = 2.6
,marginallydierentfromthestandardvalueof
3
. Even ifthe ALPenergy dominatesthe universe andthe entropy injeteddu-ring the deay ishuge, the reheating temperatureis large enoughfor neutrinos to
almost fully reovertheir thermalabundane. In general,the nal value of
N eff
isrelated not to the total entropy released, but only to the part of it injeted after
the freeze out of the neutrino-eletron interations. In this partiular region of
the parameter spae,the neutrino dilutionismainlysensitiveto theALP lifetime
and not to
m φ
org φ
individually. The outome of a deay earlierthant ∼ 10 −2
sis undistinguishable from standard osmology, sine neutrinos regain ompletely
their thermal abundane. Around
T ∼ m e
, eletrons and positrons beomenon-relativisti and annihilate, heating the photon bath but not the neutrinos, whih
have deoupled. The ratio
X ν /X γ
therefore dereases. In this period,the photontemperature inreases with respet to the neutrino temperature by the standard
fator
(4/11) 1/3
due toentropy onservation.The piture hanges onsidering later deays, after neutrino deoupling:
pho-tons andeletrons get allthe ALP enthropy,asshown ingure4.4b. Thisiswhat
happensingure4.3a andingure4.3b abovethe thik isoontour. The neutrino
dilution is omputable in this ase, as the ratio of the nal and initialomoving
entropies of the photon-eletron bath is given by equation (4.7). The
tempera-ture of the eletromagneti bath inreases with respet to the neutrino one by a
fator
(S f /S i ) 1/3
, makingN eff = 3(S f /S i ) −4/3
beause also the eletron entropyends up in photons. The neutrino energy density is therefore strongly diluted by
the energy gain of photons plus eletrons. Note that this is mainly a funtion of
m φ √
τ ∝ g φ √ m φ
−1
as we an evine from equation (4.7) whih produes
the harateristi slope of the isoontours at long ALP lifetimes
τ ∝ m φ −2
in
gure 4.3a and
τ ∝ g φ 4
in gure 4.3b. Forg φ . 10 −9 GeV −1
, this is the onlydependene on
g φ
, sineY φ
is onstant. Another example of late ALP deay, butwith a smaller mass, is shown in gure 4.4. In this ase we observe rst the
e ±
m Φ = 81 GeV g Φ =H10 13 GeVL -1 Τ= 25 ms N eff = 2.60
10 -1 1 10 10 2
10 -1 1 10 10 2
T @MeVD
X i X Γ 2
(a)
m Φ =9.4 GeV g Φ =H10 12 GeVL -1 Τ= 0.15 s N eff = 0.45
10 -1 1 10 10 2
10 -1 1 10 10 2
T @MeVD
X i X Γ 2
(b)
m Φ = 32 keV g Φ =H6.3´10 8 GeVL - 1 Τ= 1.6´10 9 s N eff = 0.26
10 -5 10 -4 10 -3 10 -2 10 -1 1 10 0
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
T @MeVD
X i
X Γ 2
()
m Φ = 32 keV g Φ =H4´10 5 GeVL -1 Τ= 631 s
N eff = 2.34
10 0 -3 10 -2 10 -1 1 10
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
T @MeVD
X i
X Γ 2
(d)
Figure 4.4: Examples of the evolution of the omoving energy of all speies
of neutrinos (solid), ALPs (dashed) and eletrons (double-dashed), as funtions
of the temperature of the universe. All energies are normalized to one thermal
annihilation, whih heats photons with respet to neutrinos. A sizeable neutrino
dilutionis observable afterthe ALP deay.
Finally,ingure4.4dweshowanexampleforwhihinversedeaysarerelevant,
whih is on the left side of gure 4.3a and on the right in gure 4.3b. As the
temperature drops, we observe a rst derease of the ALP energy density due to
the eletrons heating the photon bath. The inverse deay hannel opens around
T ∼ 70
keV and helps ALPs to regain equilibrium before disappearing atT ∼ m φ
. During rethermalization, the photon energy dereases, whih an be seenas a slight rise in
X ν /X γ
. Due to this mehanism, in the small-mass andshort-lifetimeregionoftheparameterspaewehaveseenthatentropyonservationgives
N eff = 3(11/13) 4/3 ≃ 2.4
. Ifm φ
islargerthanafewMeV,the deay-in-equilibrium happenswhenneutrinosarestilloupled,soN eff
approahes3. Thedisappearane from the thermal bath is governed only by the mass, and the isoontours ofN eff
exatly followthe isoontoursof
m φ
.The exluded region of the parameter spae are determined omparing these
numbers with the limits (4.9), taking the 99% C.L. value of
N eff > 2.11
. We areompelledtothis onservativehoie beause the 95%C.L. valueof
N eff > 2.39
isjustbelowthe
N eff ≃ 2.4
obtainedfromtheaseofdeay-in-equilibrium. Given all the unertainties of this alulation, we an onsider this large part of parameterspaedisfavouredbut notexluded. Ingure4.3, theexludedregionisabovethe
thik line, whih is oloured in yellow in the summarising plot for this hapter,
gure 4.10.