You are on page 1of 18

Chapter 3

Thermal History

Dodelson for quick look,


Kolb & Turner for details.

In this chapter we investigate some of the phenomena in the early Universe in which there were
departures from thermodynamical equilibrium. These phenomena and the time/temperature/energy
in which they happened are important to understand the physics of the early Universe and how it
affected its observed properties later on. They include

i) the neutrino decoupling that we already studied in the previous chapter,


ii) the freeze out of neutrons,
iii) the formation of nuclei of light elements known as nucleosynthesis,
iv) the recombination of electrons and protons allowing the decoupling of the photons and
v) possibly the production of dark matter particles.

All these phenomena are connected by the Boltzmann equation. In this chapter, we will still be
working in the background metric though so the Boltzmann equation will still be simplified as in
the previous chapter. However, now we will be dealing with non-equilibrium phenomena and we
need to include the Collision terms of each process that we study. In following chapters we will
then consider perturbations in the metric and in the Boltzmann equations.

3.1 Boltzmann Equation


As seen in the previous chapter, the Boltzmann equation is given by
 
df ∂f dxi ∂f dp ∂f dp̂i ∂f ∂f
= + i
+ + i
= . (3.1)
dt ∂t dt ∂x dt ∂p dt ∂ p̂ ∂t C
We may write this equation as
 
∂f
L[f ] = , (3.2)
∂t C

where the operator L is given by


∂ dxi ∂ dp ∂ dp̂i ∂
L= + i
+ + (3.3)
∂t dt ∂x dt ∂p dt ∂ p̂i

103
104 CHAPTER 3. THERMAL HISTORY

The collision term on the right is typically computed within the relativistic context of Quantum
Field Theory. Therefore, for consistency, we must use the relativistic covariant version of the
operator L, which is given by
dxα ∂ dP α ∂
L = + (3.4)
dλ ∂xα dλ ∂P α
Notice that the classical operator has derivatives with respect to 7 variables (t, xi , pi ), but the
relativistic operator has derivatives with respect to 8 variables (xµ , P µ ) = (t, xi , P 0 , P i ), so we
seem to have gained 1 more variable. But the condition gαβ P α P β = −m2 allows us to get rid of
one variable. Typically we choose to eliminate P 0 ∝ E, so the operator is simply:

dxα ∂ dP i ∂
L = +
dλ ∂xα dλ ∂P i
∂ ∂
= P α α − Γiβγ P β P γ (3.4)
∂x ∂P i
where we used the Geodesics equation in the second line. For the FRW metric, we have
∂ ∂ ∂
L = P 0 0 +P i i − 2Γi0j P 0 P j i
∂x
| {z } ∂x | {z ∂P }
∂ ∂
P 0 ∂t 2Hδij P 0 P j
∂P i
 
0 ∂ Pi ∂ ∂
= P + 0 i − 2HP i (3.4)
∂t P ∂x ∂P i

From Eq. 2.274 we have P i = P p̂i where P = p/a and p2 = gij P i P j . Since P i = P i (P, p̂i ), we may
express
∂ ∂P ∂ ∂ p̂i ∂ 1 ∂ 1 ∂
i
= i
+ i i
= i + (3.5)
∂P ∂P ∂P ∂P ∂ p̂ p̂ ∂P P ∂ p̂i
Therefore
∂ ∂ ∂
Pi i
=P + p̂i i (3.6)
∂P ∂P ∂ p̂
The operator then becomes
 
∂ Pi ∂ ∂ ∂
L = P0 + 0 i − 2HP − 2H p̂i i (3.7)
∂t P ∂x ∂P ∂ p̂

Since we are neglecting spatial and directional dependences, we set ∂f /∂xi = ∂f /∂ p̂i = 0, but
we will allow for departures from equilibrium induced by the collision term. Therefore
 
0 ∂ ∂
L = P − 2HP (3.8)
∂t ∂P

Now recall p ∝ a−1 , so we may set p = α/a for constant α. Therefore P = p/a = p2 /α, which
implies dP/dp = 2p/α = 2/a. We may then replace P by p as

∂  p   ∂p ∂   p  a ∂ 1 ∂
P = = = p (3.9)
∂P a ∂P ∂p a 2 ∂p 2 ∂p
3.1. BOLTZMANN EQUATION 105

and L becomes
 
0 ∂ ∂
L = P − Hp . (3.10)
∂t ∂p

Since in the FRW metric we still have P 0 = E, the Boltzmann equation becomes
 
∂f ∂f 1 ∂f
− Hp = (3.11)
∂t ∂p E ∂t C

But we saw in the previous chapter that for a single species with number density n integrating the
left-hand side of this equation in (the collisionless case) over all momenta produced Eq. 2.255, and
therefore we have
3 Z  
−3 d(na ) d3 p 1 ∂f
a =g (3.12)
dt (2π)3 E ∂t C

Therefore, the Boltzmann equation formalizes the statement that the departure in the number
density evolution of a certain particle from the simple dilution effect n ∝ a−3 is due to the collisions
producing and eliminating the particle that fall out of equilibrium. Suppose the reaction

1+2↔3+4 (3.13)

The number densities are all coupled. The Boltzmann equation for particle 1 is obtained by
integrating over the momenta of all 4 particles. Defining

d 3 p1
d3 q1 = g 1 (3.14)
(2π)3 2E1

we have
4 Z
! 4
! 4
!
d(n1 a3 ) Y X X
a−3 = d3 qi (2π)4 δ 3 pi δ Ei |M|2 {f3 f4 − f1 f2 } (3.15)
dt | {z } | {z }
| i=1 {z } | i=1
{z } | i=1
{z } Amplitude Distributions
Momentum cons. Energy. cons.
Integral over all momenta

Without interactions: n ∝ 1/a3 .

Rate of producing n1 : ∝ to occupation number f3 , f4 . Loss: f1 , f2

[1 ± f1 ][1 ± f2 ] neglected: + (Bosons, Bose Enhancement), − (Fermions, Pauli blockin)

Delta functions: Energy + Momentum conservation.

|M|2 : Interaction Amplitude: fundamental physics (e.g. Feynmann diagrams in QFT)

Another way to see how the factors of 2E on the denominator appear is to consider integration
over relativistic phase-space, i.e. over the 4-momentum P α = (E, p), but constrained to lie on the
3-sphere E 2 = p2 + m2 :
106 CHAPTER 3. THERMAL HISTORY

Z Z Z Z Z p
4 α 3 2 2 2 3 δ(E − p2 − m2 )
d P = d p dE δ(E − p − m ) = d p dE p
3−sphere 2 p2 + m2
Z Z
d3 p d3 p
→ p = (3.15)
2 p 2 + m2 2E

where we used
δ(E − ξ1 )
,
δ(f (E)) = where f (ξ1 ) = 0. (3.16)
|f ′ (ξ1 )|
p p
Here f (E) = E 2 − p2 − m2 , so ξ1 = p2 + m2 , f ′ (E) = 2E and f ′ (ξ1 ) = 2ξ1 = 2 p2 + m2 .

3.1.1 Approximations

In equilibrium 1 + 2 ↔ 3 + 4 happen efficiently in both directions.


In this case, chemical equilibrium:

µ1 + µ2 = µ3 + µ4 (3.17)

Out of equilibrium: annihilation not efficient (e.g. neutrino decoupling from last chapter).

Scattering processes → kinetic equilibrium, i.e. FD and BE distributions, but with µ(t).
Still, must solve ordinary differential equation for µ, or equivalently for number density n

As temperature decreases, we will be interested in systems with T < E − µ


(E − µ)/T > 1, so ±1 in BE and FD distributions can be neglected and

1
f (E) = → e−(E−µ)/T < 1 (3.18)
e(E−µ)/T ±1
Also neglect Pauli blocking/Bose enhancement factors [1 ± f1 ][1 ± f2 ]. Therefore

f3 f4 − f1 f2 → e−(E3 +E4 )/T e(µ3 +µ4 )/T − e(E1 +E2 )/T e(µ1 +µ2 )/T (3.19)

And because of energy conservation E1 + E2 = E3 + E4 , so

h i
f3 f4 − f1 f2 → e−(E1 +E2 )/T e(µ3 +µ4 )/T − e(µ1 +µ2 )/T (3.20)

Now express results in terms of number density n, instead of µ. We have

Z Z
d3 p d3 p −Ei /T
ni = g i f (E) = gi eµi /T e (3.21)
(2π)3 (2π)3
3.1. BOLTZMANN EQUATION 107

(0)
Define the species-dependent equilibrium number density (µi = 0) as ni :

  3/2
Z
(0) d3 p −Ei /T  gi m2πi T e−mi /T mi ≫ T
ni = gi 3
e = (3.22)
(2π)  g T 3
mi ≪ T
i π2

(0)
Therefore eµi /T = ni /ni and the piece appearing on the Boltzmann equation

" #
n3 n4 n1 n2
f3 f4 − f1 f2 = e−(E1 +E2 )/T (0) (0)
− (0) (0)
(3.23)
n3 n4 n1 n2

The Boltzmann equation becomes then

3) Z Z Z Z
−3 d(n1 a d 3 p1 d 3 p2 d 3 p3 d 3 p4
a =
dt (2π)3 2E1 (2π)3 2E2 (2π)3 2E3 (2π)3 2E4
(2π) δ (p1 + p2 − p3 − p4 ) δ (E1 + E2 − E3 − E4 ) |M|2
4 3

{f3 f4 − f1 f2 }
Z Z Z Z
d 3 p1 d 3 p2 d 3 p3 d 3 p4
= e−(E1 +E2 )/T
(2π)3 2E1 (2π)3 2E2 (2π)3 2E3 (2π)3 2E4
(2π)4 δ 3 (p1 + p2 − p3 − p4 ) δ (E1 + E2 − E3 − E4 ) |M|2
" #
n3 n4 n1 n2
(0) (0)
− (0) (0) (3.19)
n3 n4 n1 n2

We then define the (thermally) averaged cross-section, as

Z Z Z Z
1 d 3 p1 d 3 p2 d 3 p3 d 3 p4
hσvi ≡ e−(E1 +E2 )/T
n1 n2
(0) (0) (2π)3 2E1 (2π)3 2E2 (2π)3 2E3 (2π)3 2E4
(2π)4 δ 3 (p1 + p2 − p3 − p4 ) δ (E1 + E2 − E3 − E4 ) |M|2

and the Boltzmann equation becomes

" #
d(n1 a3 ) (0) (0) n3 n4 n1 n2
a−3 = n1 n2 hσvi (0) (0)
− (0) (0)
(3.18)
dt n3 n4 n1 n2

We will solve this equation for different cases to track down the evolution of abundances.

Note that left-hand side is of order ∼ n1 /t ∼ n1 H for typical cosmological times

The right-hand side is ∼ n1 n2 hσvi.

If reaction rate n2 hσvi ≫ H, for the equation to be satisfied, the bracket must be small, i.e.
108 CHAPTER 3. THERMAL HISTORY

n3 n4 n1 n2
(0) (0)
= (0) (0)
(3.19)
n3 n4 n1 n2

This equation is known as the chemical equilibrium, or nuclear statistical equilibrium or Saha
equation, depending on the context it is used.

Obviously it is only valid while chemical equilibrium still holds. For a more complete treatment,
we must solve the differential equation while only kinetic equilibrium holds.

3.2 Big Bang Nucleosynthesis


3.2.1 Nuclear Physics

Proton p: 1 H (hydrogen nucleus)

Proton+neutron p + n: 2 H = D (deuterium)

Proton+2 neutrons p + n + n: 3 H = T (tritium)

Atomic number Z = protons

Mass number A = protons + neutrons

A − Z= neutrons

Total mass m of a nucleus with Z protons and A − Z neutrons is different from the sum of the
masses of isolated protons and neutrons. The difference is the binding energy

B ≡ Zmp + (A − Z)mn − m (3.20)

For instance mD = 1875.63 MeV while mp + mn = 1877.84 MeV, so BD = 2.22 MeV. Typical
binding energies are of this order, and therefore we expect nucleosynthesis to occur when the
Universe has cooled down to T ∼ 1 MeV.
Neutrons and protons can convert on one another via weak interactions:

p + ν̄ → n + e+ ; p + e− → n + ν ; n → p + e− + ν̄ (3.21)

Light elements are formed via electromagnetic interactions

p+n → D+γ (3.22)


3
D + D → n + He (3.23)
3 4
He + D → p + He (3.24)

Our main goal then is to find first the neutron abundance, which will roughly monitor the abundance
of the other elements.
3.2. BIG BANG NUCLEOSYNTHESIS 109

3.2.2 Baryon-to-photon ratio


Let us estimate the current baryon-to-photon ratio (in number density). Since ρb = nb mp , and
ρb = Ωb ρcrit , we have nb = Ωb ρcrit /mp . Moreover, nγ = (2.4/π 2 )Tγ3 . With Tγ = 2.75K we then find
the ratio
 
nb Ωb ρcrit π 2 −10 Ωb h
2
η= = = 5.5 × 10 (3.25)
nγ 2.4mp Tγ3 0.02

If the baryon number NB = Nb −N̄b (difference in number of baryons and anti-baryons) is conserved,
its ratio to the entropy, also conserved, will be constant

NB Nb − N̄b nb − n̄b
= = = const. (3.26)
S S s
If there is slightly more baryons than anti-baryons early on, this remains true in the absence
of processes that violate baryon number. At some point the anti-baryons are annihilated away
(n̄b = 0), and only the baryons remain, so the ratio above becomes simply nb /s. But since s ∝
a−3 ∝ Tγ3 ∝ nγ , then

nb − n̄b nb
∼ ∼ 10−10 (3.27)
s nγ

So there had to be an asymmetry of this precise amount in the primordial abundances of baryons
and anti-baryons.

These baryons will eventually settle into atomic nuclei, and to find out how much of each we
have to solve the coupled set of Boltzmann equation for all nuclei.

Here we will make the simplifying assumption that no elements heavier than helium are pro-
duced.

So we only need to track 1 H, 2 H and 3 H and 3 He, 4 He.

Consider Deuterium production:

n+p↔D+γ (3.28)
(0)
For photons µ = 0 so they are always at chemical equilibrium, i.e. nγ = nγ . The equilibrium
condition for this reaction is

(0) (0)
nD nγ nD nγ
= (0) (0)
nn np nn np
(0)
nD nD
= (0) (0)
(3.28)
nn np nn np

Inserting Boltzmann factors n = g(mT /2π)3/2 e−m/T , with


110 CHAPTER 3. THERMAL HISTORY

 
nD gD 2πmD 3/2 [mn +mp −mD ]/T
= e (mD ≈ 2mn = 2mp , but BD = mn + mp − mD )
nn np gn gp mn mp T
 
3 4π 3/2 BD /T
= e (3.28)
4 mp T

and since nb = nn + np , we have nn ∝ nD and nD ∝ nb , so nn ∝ nb (similarly np ∝ nb ). So using


the baryon-to-photon ratio η = nb /nγ and nγ ∝ T 3 , we have
 3/2
nD 1
∼ nb eBD /T
nb mp T
 3/2
nb 1
∼ nγ eBD /T
nγ |{z} mp T
|{z} ∝T 3
η
 3/2
T
∼ η eBD /T (3.27)
mp
| {z }
∼10−14 (T /1MeV)

For T > 0.1 MeV, all baryons are in protons and neutrons.

3.2.3 Neutron Abundance


Let us now look at the neutron-proton ratio. Important reactions here are

p + e− → n + ν e (3.28)

which keep neutrons and protons in equilibrium until T ∼ MeV. After that we must solve the
Boltzmann equation.
We have for protons and neutrons in equilibrium
(0)  
np mp 3/2 (mn −mp )/T
(0)
= e = eQ/T (3.29)
nn mn
| {z }
≈1

with Q = mn − mp = 1.293 MeV, so at T ∼ 1 MeV the neutron fraction starts to go down. If weak
interactions were efficient neutron abundance would go to zero.
However, we must solve the Boltzmann equation to see what really happens. Define the ratio
of neutrons to total nuclei:
nn nn 1
Xn = = = (3.30)
nb nn + np 1 + (np /nn )

In equilibrium

1
Xn,EQ ≡ (0) (0)
(3.31)
1+ (np /nn )
3.2. BIG BANG NUCLEOSYNTHESIS 111

For

1+2 → 3+4 (3.32)


n + lepton → p + lepton (3.33)
(0)
with leptons in equilibrium nl = nl , we have
" #
d(n1 a3 ) (0) (0) n3 n4 n1 n2
a−3 = n1 n2 hσvi (0) (0)
− (0) (0)
dt n3 n4 n1 n2
" #
(0)
d(nn a3 ) (0) np nn
a−3 = nl hσvi (0)
− nn
dt np
(3.32)

Now nn = (nn + np )Xn , so

d(nn a3 ) d[(nn + np )Xn a3 ] dXn dXn


a−3 = a−3 = a−3 (nn + np )a3 = (nn + np ) (3.33)
dt dt | {z } dt dt
nb a3 =const.

(0) (0)
Therefore, using nn /np ∼ e−Q/T we have
dXn h i
(0)
(nn + np ) = n2 hσvi np e−Q/T − nn
dt
(3.33)
(0)
or defining λnp = nl hσvi ( rate of neutron → proton convertion) and (1 − Xn ) = np /(nn + np ) we
have
dXn h i h i
= λnp (1 − Xn )e−Q/T − Xn = λnp e−Q/T − Xn (1 + e−Q/T ) (3.34)
dt
Let x = Q/T . Then dXn /dt = ẋdXn /dx and ẋ = −(Q/T 2 )Ṫ = −xṪ /T . Because T ∝ a−1
r
Ṫ ȧ 8πGρ
= − = −H = − (3.35)
T a 3
and dXn /dt = xHdXn /dx. Nucleosynthesis is in the radiation-dominated era, so

π2g∗ 4
ρ= T (3.36)
30
where g ∗ = g ∗ (T ) changes when species annihilate, but remain constant otherwise. At T = 1 MeV,
we have gγ = 2, gν = 6 and ge− = ge+ = 2, so g ∗ ≈ 10.75. Then
r r
8πG π 2 g ∗ 4 4π 3 Gg ∗ Q4 H(x = 1)
H= T = = (3.37)
3 30 45x4 x2
Then
dXn dXn H(x = 1) dXn  
= xH = = λnp e−x − Xn (1 + e−x ) (3.38)
dt dx x dx
112 CHAPTER 3. THERMAL HISTORY

Figure 3.1: Fractional Abundance of light elements.

or finally,
dXn xλnp  −x 
= e − Xn (1 + e−x ) (3.39)
dx H(x = 1)

where, since g ∗ (x = 1) = g ∗ (T = 1.293MeV) = 10.75, we have


r
4π 3 Gg ∗ (x = 1)Q4
H(x = 1) = ≈ 1.13 sec−1 (3.40)
45
The neutron-proton conversion can be computed (Problem Set) from the collision term definition
and we find
255
λnp = (12 + 6x + x2 ) (3.41)
τn x 5
where the neutron lifetime τn = 886.7 sec. For instance, when T = Q, i.e (x = 1), we have
λnp = 5.5 sec−1 > H(x = 1), so equilibrium still maintains. As T < 1 MeV, we start to have
λnp < H and conversion becomes inneficient.

We must then solve the equation numerically. See plot.

At T ∼ 0.5 MeV, Xn ∼ 0.15 freezes.

At T < 0.1 MeV, two reactions become important:

n → p + e− + ν̄ (Neutron decay) (3.42)


n+p → D+γ (Deuterium production) (3.43)
3.2. BIG BANG NUCLEOSYNTHESIS 113

Decays can be modelled via

d(n1 a3 ) n1 d(n1 a3 ) n1 a 3
a−3 = − → =− (3.44)
dt τn dt τn

So the effect of decays can be added as a term e−t/τn to the results of the figure. When decays
become relevant e± annihilated and g ∗ = 3.36.

In the radiation era H 2 ∝ a−4 , so H ∝ a−2 and


Z Z Z Z  2
da da 2 1
t= dt = = = ada ∼ a ∼ (3.45)
aH aa−2 T

more precisely
 2
0.1MeV
t = 132sec (3.46)
T

Production of deuterium and other light elements start at Tnut ∼ 0.07 MeV (see next section
for estimation). The decays deplete the neutron fraction a little further by a factor of

exp [−(132/886.7)(0.1/0.07)2 )] = 0.74 (3.47)

and the neutron abundance at the onset of nucleosynthesis becomes 0.15 × 0.74

Xn (Tnuc ) = 0.11 (3.48)

and Xp = 1 − Xn = 0.89. This is the starting point for light element formation.

3.2.4 Light Element Abundances


Approximation: Assume that light element prodution occurs instantaneously at Tnuc when ener-
getics compensates the small value of η (baryon-to-photon ratio).
For instance deuterium. We have that nD ∼ nb when
 3/2
nD T
∼ η eBD /T ∼ 1 (3.49)
nb mp

when
3 BD
ln(η) + ln(Tnuc /mp ) ∼ − (3.50)
2 Tnuc
and we see that this equation holds when Tnuc ∼ 0.07.

nHe
Similarly for He: nb ∼ eBHe /T

Since BHe > BD , the exponential eB/T favors helium over deuterium.

As a result, nearly all neutrons are processed into 4 He at T ∼ Tnuc ∼ 0.07 MeV. Since one needs
2 neutrons for each 4 He, then n4 He = nn /2. Results often quoted in terms of mass fraction:
114 CHAPTER 3. THERMAL HISTORY

4n4 He 2nn
X4 = = = 2Xn (Tnuc ) ≈ 0.22 (3.51)
nb nb

It is more common to refer to the Helium mass fraction Yp relative to hydrogen H or proton p.
Since Xn = 0.11 and XH = Xp = 1 − Xn = 0.89, we have nn /np = Xn /Xp = 0.12, and

4n4 He 2nn Xn (Tnuc )


Yp = He/H = = =2 ≈ 0.24 (3.52)
np np Xp (Tnuc )

A more exact approach, solving the differential equations and not making approximations give

Yp = 0.2262 + 0.0135 ln(ηb /1010 ) (3.53)

We see that the Helium mass fraction does not depend too much on the baryon density.

As for Deuterium, a trace amount is left over, because the reaction that eliminates it

D + p →3 He + γ (3.54)

also becomes inefficient at some point.

After Tnuc , deuterium is depleted but eventually freezes out at 10−5 − 10−4 . See figure above.
It turns out that the left-over abundance of deuterium is highly sensitive to the baryon density Ωb .

Measurements
Helium measured by extrapolating Y versus O/H to O/H → 0. The lower the value of O/H, the
less processed is system, so its Helium abundance is closer to primordial. This measurement is one
of the pillars of the standard cosmological model.
As for measuring the abundance of deuterium, it is obtained from spectral lines of quasars.
The quasar spectra is modified to include absorption lines of deuterium in gas clouds at z ≈ 3.
This measurement helps pin down the baryon fraction. For example, O’Meara et al. 2001 measure
D/H ∼ 3 × 10−5 for which Ωb h2 ∼ 0.0205.

3.3 Recombination
While T > 1 eV, photons remain coupled to electrons via Compton scattering and electrons to
protons via Coulomb scattering. Almost no neutral hydrogen, only hydrogen nuclei at this point.

Energetics favor neutral atoms (binding ǫ0 = 13.6 eV), but the prefactor with η ∼ 10−10
prevents this from happening, i.e. the large number of photons ionize a neutral atom as soon as it
forms.

Here the reaction of interest:


3.3. RECOMBINATION 115

Figure 3.2: Measured mass-fraction of Helium relative to Hydrogen (Y = He/H) in systems as a function
of O/H. All elements above He are called metals by astronomers. Systems with more metals are more
processed in their history. Systems with less metals are more pristine. Extrapolating the data to O/H → 0,
we obtain the primordial Helium abundance at Y ∼ 0.24.

Figure 3.3: Primordial abundances of light elements. Consistency of all of them, mostly Deuterium, pins
down the value of Ωb ∼ 0.04 − 0.05. From Burles, Nollett and Turner, 1999.
116 CHAPTER 3. THERMAL HISTORY

1+2 ↔ 3+4 (3.55)



e +p ↔ H +γ (3.56)
(0)
For photons, as usual nγ = nγ , so the Saha equation:
(0) (0)
ne np ne np
= (0)
(3.57)
nH nH

The Universe is overall neutral, so ne = np . Define the free electron fraction:

ne np nH
Xe = = , (1 − Xe ) = (3.58)
ne + nH np + nH ne + nH

The equilibrium condition becomes


 ǫ 
 3/2 z }|0 {
Xe2 1  me T
e− [me + mp − mH ] /T 

=  (3.59)
1 − Xe ne + nH 2π

where we set mp ≈ mH in the prefactor. The denominator is

np + nH = nb = ηnγ ≈ 10−9 T 3 . (3.60)

The pre-factor on the right-hand side is therefore


"  #  3/2  
1 me T 3/2 9 −3 3/2

9 me
3/2
9 me ǫ0 3/2
≈ 10 T (me T ) ∼ 10 ∼ 10
ne + nH 2π T ǫ0 T
5 3/2  ǫ 3/2
   ǫ 3/2
9 5.1 × 10 0 0
∼ 10 ∼ 1015 (3.60)
13.6 T T

which is very large, even for T = ǫ0 . That means the denominator on the left-hand side of the
equilibrium equation must be small, i.e. Xe ∼ 1, all electrons are free.

But as the temperature falls further, the exponential changes, Xe will fall. We must solve the
Boltzmann equation.

In this case, using the equilibrium solution we have for the Boltzmann equation,
" #
d(n a 3) n n 2
e H
a−3 = n(0) (0)
e np hσvi (0)
− (0) e (0)
dt nH ne np
" #
(0) (0)
ne np 2
= hσvi nH (0)
− ne (3.60)
nH
3.3. RECOMBINATION 117

Figure 3.4: Free electron fraction Xe versus redshift z. Recombination happens at z ∼ 1000 or T ∼ 0.03 eV.
Without electrons to bound, photons decouple from the plasma.

(0) (0) (0)


or using ne np /nH = (me T /2π)3/2 exp[−ǫ0 /T ], ne = nb Xe and nH = (1 − Xe )nb
" #
3) (0) (0)
−3 d(ne a ne np
a = hσvi nH (0)
− n2e
dt nH
"  3/2 #
me T −ǫ0 /T
= nb hσvi (1 − Xe ) e − Xe2 nb (3.60)

Finally, since nb a3 = const. we have ne a3 = Xe nb a3 and


"   #
dXe me T 3/2 −ǫ0 /T 2
nb = nb hσvi (1 − Xe ) e − X e nb (3.61)
dt 2π

or
dXe
= (1 − Xe )β − Xe2 nb α(2) (3.62)
dt
where
 3/2
me T
β ≡ hσvi e−ǫ0 /T (Ionization rate, increase free electrons) (3.63)

α(2) ≡ hσvi (Recombination rate, decrease free electrons) (3.64)

We can then solve this numerically.


118 CHAPTER 3. THERMAL HISTORY

Recombination is related to Decoupling of photons from matter. Compton scattering:

γ + e− → γ + e− (Compton Scattering) (3.65)

The condition for decoupling is that the rate of interation ne σT is lower than the expansion rate
H. We can compute the ratio:

  1/2  3/2  −1/2


ne σ T Ωb h2 0.15 1+z 1 + z 0.15
= 113Xe 1+ (3.66)
H 0.02 Ωm h2 1000 3600 Ωm h2

So when Xe ∼ 10−2 , at z ∼ 1000, this ratio will be smaller than 1 and photons decouple from
matter, so decoupling is tightly connected with recombination.
Notice however that even if the universe remained ionized forever, i.e. Xe = 1 above, there
would still be decoupling at z ∼ 40. Later, the Universe was reionized at z ∼ 6 − 12, when the first
stars formed.

3.4 Dark Matter Production


Let us consider dark matter is composed of a Weakly Interacting Massive Particle (WIMP), which
was in equilibrium with the cosmic plasma early on, but that experienced freeze-out, when the
temperature dropped below its mass, turning its interactions ineffective to allow it to remain in
equilibrium.

After freeze-out (falling out of equilibrium) what is its relic abundance?

If we fix Ωm ∼ 0.3 (observations), can we learn about the particle properties, i.e. interaction
cross-section and mass?

Typical case: heavy dark particle X producing light particle l. The latter is assumed to be
(0)
always coupled and in equilibrium, i.e. nl = nl )

1+2 → 3+4
X +X → l+l (3.66)

Therefore

 
 
3)  
−3 d(n1 a (0) (0)  n3 n4 n1 n2 
a = n1 n2 hσvi  (0) (0) − (0) (0)  (3.67)
dt n n
| 3 {z 4 } |n1 {zn2 } 
| {z } 
(0) 2
(nX )
1 (0)
n2 /(n )2 X X

becomes
d(nX a3 ) h
(0)
i
a−3 = hσvi (nX )2 − n2X (3.68)
dt
3.4. DARK MATTER PRODUCTION 119

Figure 3.5: Abundance Y = nX /T 3 of heavy stable particle as a function of x = m/T . Relic abundance Y∞
is obtained at very late times when x ∼ ∞.

Since T ∼ 1/a, defining Y = nX /T 3 , we have

dY  2 
= T 3 hσvi YEQ −Y2 (3.69)
dt
(0)
where YEQ = nX /T 3 . Then change
m
x= , (3.70)
T

and since T ∝ 1/a and Ṫ /T = −ȧ/a = −H, we have

dx m Ṫ
= − 2 Ṫ = − x = Hx (3.71)
dt T T
and the differential equation becomes

dY m3  2 
Hx = 3 hσvi YEQ −Y2 (3.72)
dx x

or, since H(x) = H(x = 1)/x2 , we finally obtain in terms of H(x = 1) = H(T = m)

dY λ  2 2

= − 2
Y − YEQ (3.73)
dx x
m3 hσvi
λ = (3.74)
H(m)

As one change the specific model, one can predict solve the differential equation and find the relic
abundance of heavy stable particles. See figure.
120 CHAPTER 3. THERMAL HISTORY

Since we already know more or less the dynamics of freeze-out, we know that for x ≪ 1, we
have Y ∼ YEQ and for x ≫ 1, we have Y ≫ YEQ , so that the equation becomes

dY λY 2
= − (x ≫ 1)
dx x2
dY dx
= −λ 2 (3.74)
Y2 x
Integrating from the time of freeze-out (x = xf ) until very late times (x = 0), we have

1 x=∞ 1
− |x=xf = λ |x=∞ (3.75)
Y x x=xf
or
1 1 λ
− = (3.76)
Y∞ Yf xf

Typically, after freeze-out, Yf still falls by at least an order of magnitude to Y∞ , so we may set
Yf ≫ Y∞ and find
xf
Y∞ ≈ (3.77)
λ
and the full numerical solution (or order of magnitude estimates) tell us that xf ∼ 10.
From this one can estimate the density ρX and its fraction relative to critical ΩX today, as a
function of mass m and cross-section. The fact that we observe ΩX ∼ 0.25 today then puts bounds
on masses and interactions of dark matter candidates.

You might also like