MP467 - Astrophysics and Cosmology PDF

MP467: Astrophysi
s and Cosmology
Brian Dolan
De ember 12, 2014

Contents
1 Overview 4
1.1 Newtonian Gravity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2 Why is the moon round? . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 How high is a mountain? . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.4 How old is the Earth? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.5 Distan e S ales . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2 Stellar Formation and stru ture 11

2.1 Energy sour e . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2 Dynami al equilibrium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.3 How do stars form? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.4 How long does the ollapse take? . . . . . . . . . . . . . . . . . . . . . . . 14
2.5 Virial Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.6 Stellar Equilibrium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.7 Equation of State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.8 Energy Sour es in Stars . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.9 Energy transport . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.9.1 Radiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.9.2 Conve tion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.10 Mass-Luminosity Relation . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.11 Eddington Limit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
2.12 Maximum Stellar Mass . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.13 Minimum Stellar Mass . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.14 Degenerate Stars . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
2.14.1 White Dwarves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
2.14.2 Neutron stars . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.14.3 Pulsars . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
2.14.4 Bla k Holes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
2.15 A tive Gala ti Nu lei (AGN) and Quasars (QSO) . . . . . . . . . . . . . 43
3 Cosmology and Expansion of the Universe 45

3.1 Cosmologi al Constant . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.2 Redshift-distan e relation . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
3.3 Friedmann Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
3.4 The Friedmann Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
3.5 Mi rowave ba kground . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
2
3.6 The Horizon Problem and Ination . . . . . . . . . . . . . . . . . . . . . 59
3.6.1 Inationary Universe . . . . . . . . . . . . . . . . . . . . . . . . . . 61
3.7 The rst 3 minutes (Weinberg) . . . . . . . . . . . . . . . . . . . . . . . . 61
3
1 Overview
1.1 Newtonian Gravity
The a eleration due to gravity at the surfa e of a spheri al mass M of radius R is
GM
g=− r̂
R2
where r̂ G is Newton's
is a unit radial ve tor and universal onstant of gravitation. The
gravitational for e on a small mass m is thus
GM m
F = mg = − r̂ = −∇U
R2
where U = − GM
R
m
m in the gravitational eld of M . The
is the potential energy of
GM
gravitational potential energy per unit mass is V = −
R and g = −∇V .
2π
For a small mass in a ir ular orbit of radius r , with period T =
ω , around a mu h
2
larger mass M the entrifugal a eleration rω is balan ed by the a eleration due to
gravity
GM GM
ω2 r = ⇒ ω2 =
r2 r3
giving Kepler's third law, that the square of the period is proportional to the ube of the
distan e 2
2 2π r3
T = = 4π 2 .
ω GM
Kepler's third law an easily be dedu ed by dimensional analysis: Newton's universal
onstant of gravitation has dimensions [G] = length3 mass−1 time−2 . For a given mass
the length
3 time−2 gives Kepler's third law.
1.2 Why is the moon round?

Consider a spheri al ro ky body with mass M and radius R. The gravitational potential
V = − GM
energy per unit mass on the surfa e is
28 16
R . Ro ks (e.g. SiO2 with 14 Si and 8 O )
have a mole ular mass ≈ 60mp . SiO2 melts at 1880K ⇔ the hemi al binding energy is
≈ kB T = 2.6 × 10−20 J = 0.16eV = −EB per mole ule. Gravitational energy of a sili a
mole ule on the surfa e of M is
EGrav = mSiO2 V.
4
Ro ks will disintegrate if
|EGrav | > |EB | (1.1)
GM
EGrav = −60mp .
R
For simpli ity assume M has onstant density
1/3
4π 3 3M
⇒M = ρR ⇒ R =
3 4ρπ
1/3
4ρπ
EGrav = −60mp GM 2/3 .
3
Ro ks disintegrate if
1/3
2/3 4πρ
M (60mp G) > |EB |
3
1/2 3/2
3 |EB |
M> := MP (1.2)
4πρ 60mp G
M > MP then we expe t M will

If be roughly spheri al (a planet or large moon), if
M < MP then M will not even be approximately spheri al it will be an amorphous
shape. For example
ρ⊕ = 5500kg/m3
ρmoon = 3400kg/m3
ρSiO2 = 2800kg/m3
let ρ = 2800 × α × kg/m3 1<α<2
1
MP = √ × (2.2. × 1021 )kg
α

3MP 1/3 570km
RP = = √
4πρ α
Mmoon = 7 × 1022 kg >> MP .

The largest asteroid, Ceres, has R = 500km and it is roughly spheri al. Small asteroids
are not even approximately spheri al.
The stru ture of a planet, su h as the Earth, is more ompli ated than this model
allows for: mu h of the interior of the Earth is not solid, due to heat produ ed by
radioa tive isotopes. Also the giant planets, su h as Jupiter and Saturn, are mostly gas.
But this is not an issue for obje ts ≈ 600 km a ross and the above al ulation does give
good intuition into the reasons for the shapes of solar system obje ts.
5
1.3 How high is a mountain?
Consider a mountain of mass Mm = N mSiO2
Gravitational energy stored in the mountain
EGrav ≈ Mm gh.
Chemi al Binding energy of the mountain is N |EB | Mountain is stable if
EGrav < N |EB |

60mp gh < |EB |
GM
60mp h 2 < |EB |
R

h |EB |R |EB | 3 1/3
< =
R 60mp GM 60Gmp M 2/3 4πρ

MP 2/3 RP 2
= =
M R
e.g. for the earth with R⊕ = 6400km and α=2 we get
2
h 1 570
= = 4 × 10−3
R α 6400
and
h = 25km.
Compare this to the height of Mount Everest, 8, 840m. This analysis ignores weathering,
i.e. wind and freeze/thaw a tion, as well as te toni events su h as earthquakes. Moun-
tains on the Earth are worn down over the millennia by weathering and earthquakes but
the Himalaya, whi h were formed by plate te toni s when the Indian plate rashed into
the Asian plate 50 million years ago, are young enough that weathering has not yet had
a signi ant impa t. On moons or planets with little or no atmosphere and no te toni
a tivity, these ee ts are absent. We have also ignored the rotation of the Earth, whi h
makes it bulge out at the equator.
1.4 How old is the Earth?

The age of the Solar System an be estimated by isotope dating, similar in on ept
to Carbon dating of an ient organi material but using dierent isotopes with a longer
and more appropriate half-life. The element Rubidium, with atomi number 37, has an
87
isotope of atomi weight 87, 37 Rb, that de ays via β -de ay to Strontium 87 (38 Sr, whi h
87
is stable with 38 protons and 49 neutrons) with a half-life of τ = 5 × 10

10 years
87 87
37 Rb → 38 Sr + e− + ν̄. (1.3)
6
87
When meteorites from around the world are analysed for their 37 Rb and 38 Sr ontent it
87
87 87
is found that a graph of the number of 38 Sr atoms against the number of 37 Rb atoms for
various meteorites lies on a straight line with slope 0.0656.
87 Sr v. 87 Rb ontent for various meteorites.

38 37
87 87
If the number of 38 Sr and 37 Rb atoms as a fun tion of time are denoted by S(t) and
R(t) respe tively then
R(t) = R(0)e−t/τ
S(t) = S(0) + R(0)(1 − e−t/τ ) = S(0) + R(t)(et/τ − 1).
If all the meteorites are assumed to have formed about the same time, with the same
87
initial ontent of 38 Sr , then a plot of S(t) against R(t) should be indeed a straight line
with slope
S(t) − S(0)
= et/τ − 1 ,
R(t)
where τ = 5 × 1010 / ln(2) and t is measured in years. Setting
et/τ − 1 = 0.0656 ⇒ t = τ ln(1.0656) = 4.6 × 109 yrs.
The urrently a epted age of the solar system is 4.56 × 109 years and the Sun has been
shining, with almost onstant luminosity, for this time.
7
6 5
R=6,400km 1.5x10 km R=7x10 km
9
R=4.5x10 km
Nearest star: 4.3 lyr away
6
R=50,000 lyr 2x10 lyr
Clusters of galaxies
R=5x10 7 lyr
9
R=2x10 lyr
8
1.5 Distan e S ales
Astronomi al distan es are usually quoted in either light years or parse s. A light year is
the distan e a beam of light travels through empty spa e in one year. One year is 365.25
days, or 31, 557, 600secs, so a light year is
1 lyr = (2.998 × 108 ) × (3.156 × 107 )m = 9.462 × 1015 m.
A parse is dened using trigonometry with the Earth-Sun distan e, R = 1.50×1011 m,

as a base-line. Most stars are so far away that they appear essentially stationary in the
sky as the Earth moves around the Sun, hanging its position by 3 × 1011 m in 6 months.
Some stars are mu h loser than most however and these appear to move per eptibly in
the sky over a 6 month period, des ribing a tiny ellipse in 12 months. This is motion is
alled parallax. Simply trigonometry gives the angle φ as
R/d = tan φ ≈ φ ⇒ d ≈ R/φ,

where d is the distan e to the star and φ is measured in radians.
*
** ** ** ** * ** * Distant Stars
φ
* Nearby Star
φ
d
Sun Earth
R
Angles in the sky are measured in degrees divided into 60 units alled minutes of ar ,
or ar minutes, and ea h ar minute is further divided into 60 units alled se onds of ar ,
or ar se onds. The symbol
′ is used to denote minutes of ar and ′′ to denote se onds of
ar (these should not be onfused with minutes and se onds of time ). There are thus
1
3600′′ in 1◦ and 1, 296, 000′′ in 360◦ whi h is 2π radians. Hen e
1′′ = 4.85 × 10−6 rad.

A parse (1 ps ) is dened to be the distan e at whi h a star would exhibit a parallax of
1′′ , that is φ = 4.85 × 10−6 rad, so 1 psc orresponds to
1.50 × 1011
d = R/(4.85 × 10−6 ) = m = 3.09 × 1016 m = 3.26 lyr.
4.85 × 10−6
1
The sky appears to rotate through 360◦ in 24 hours whi h is 1◦ in 4 minutes, or 15 minutes of ar in
1 minute of time.
9
Inter-gala ti distan es are millions of light years, often quoted in millions of parse s, or
megaparse s (M pcs).
10
2 Stellar Formation and stru ture
2.1 Energy sour e
A typi al star (su h as the Sun) is a ball of hot plasma, mostly ionised Hydrogen. The Sun
itself, for example is, 71% hydrogen, 27% Helium and only about 2% heavier elements,
by mass. The surfa e temperature of the Sun is T⊙ = 5800K . It radiates energy at a
rate
2
L⊙ = 4πR⊙ σSB T⊙4 = 4 × 1026 J/s
where σSB is the Stefan-Boltzmann onstant. L⊙ is alled the luminosity. Using
E = mc2 ⇒ 4 × 1026 J/s ⇔ 4.4 × 109 kg/s = −Ṁ .

The mass of the Sun is M⊙ = 2 × 1030 kg so in prin iple it ould sustain this rate of mass
loss for 5× 1020 s or 1.5 × 1013 yrs. However no energy produ tion me hanism is 100%
e ient, there will always be residual mass, or ash, left when the sour e of energy runs
out. Possible energy sour es are:
1. Chemi al Energy: e.g. H + H → H2 + energy (typi al hemi al energy ǫ ≈ 1eV )

H2 has mass ≈ 3.4 × 10−27 kg ≈ 2000M eV /c2 . If ea h rea tion produ es about 1eV
of energy then the e ien y η would be:
ǫ 1eV
η= = = 5 × 10−10 .
m H2 c2 2000M eV
This ould burn M × η = (2 × 1030 ) × (5 × 10−10 )kg = 1021 kg of the Sun's mass.
The rest is left as ash H2 . This way the Sun ould only last
M ×η 1021
= 10 s = 1011 s ≈ 3000 years.
|Ṁ | 10
The solar system was formed about 5 × 109 years ago and so the age of the Sun is
τ⊙ ≈ 5 × 109 years
2. Gravitational Energy: The gravitational energy between two point masses (M1 =
2
M2 = M a distan e R apart is EGrav = − GM R (for a single sphere of mass M ,
3 GM 2
radius R and onstant density the gravitational self energy is EGrav = −
5 R ).
GM⊙2
41
Using solar values
R⊙ = 3 × 10 J . If the Sun were shining by onverting gravi-
tational potential energy into heat, through ontra tion from R = ∞ to its present
size, assuming onstant luminosity, this would take a time
|EGrav | 4 × 1041 J
= ≈ 1015 s = 3 × 107 years.
|Ė| 4 × 1026 J/s
11
This is known as the Kelvin-Helmholtz time.
3. Nu lear Energy: 2p + 2n → 4 He + 25M eV

25M eV 25M eV
mp ≈ mn ⇒ η= = = 6.6 × 10−3
4mp × c2 3752M eV
η × M⊙ (6.6 × 10−3 ) × (2 × 1030 kg)
= ≈ 3 × 1018 s = 1011 years.
|Ṁ | 4.4 × 109 kg/s
Only about the entral 10% is a tually hot enough to burn hydrogen so the expe ted
lifetime is a tually about 1010 years, the Sun is almost exa tly at the mid-point of
its life as a star.
2.2 Dynami al equilibrium

Consider a spheri al mass, su h as a star or a spheri al gas loud (ignoring rotation).
Density, ρ(r), and pressure, P (r), will depend on the distan e from the entre in general
Consider the for es on a small volume δV = δAδr of matter in a shell of radius r,
ontaining mass δM = ρ(r)δV .
r+δr
g
r
P P+δP
The thermal pressure on the interior fa e of δV is P (r) (outwards) and on the exterior
fa e is P (r + δr) = P (r) + δP (r) (inwards). For a thin shell with δr << 1,
dP (r)
δP (r) ≈ δr.
dr
The a eleration due to gravity is g(r) = − GM rr
(r)
where
Z r
M (r) = 4π ρ(r)r 2 dr (2.1)
0
is the mass inside the shell (gravitational for es δV due to the mass outside the shell all
an el). Newton's 2nd Law applied to the mass δM = ρ(r)δV in δV gives
d2 r GM (r)δM
δM 2
=− + P (r)δA − (P (r) + δP (r))δA
dt r2
d2 r GM (r)ρ(r) (P (r) + δP (r)) − P (r)
ρ(r) 2 = − − .
dt r2 δr
12
Let δr → 0
GM (r)ρ(r) dP (r)
ρ(r)r̈ = − − . (2.2)
r2 dr
Together with equation (2.1) in the dierential form
dM
= 4πr 2 ρ(r)
dr
this equation governs the dynami s of star and gas louds. In summary
GM (r)ρ(r) dP (r) dM
ρ(r)r̈ = − − , = 4πr 2 ρ(r) (2.3)
r2 dr dr
2.3 How do stars form?

Stars are formed in interstellar gas and dust louds whi h start ontra ting under the
gravitational for e. A spheri al interstellar gas loud with uniform density ρ, temperature
T and radius R, whi h is initially at rest, will start ollapsing if R̈ < 0 and will expand
if R̈ > 0. These two possibilities are separated by the equilibrium ase R̈ = 0. To nd
where this boundary lies we set r = R and r̈ = 0 in (2.2) and onsider a loud with
uniform density ρ and total mass M . Then
3M
ρ(R) =
4πR3
and (2.3) gives
GM 1 dP (R)
=− .
R2 ρ(R) dR
Assume the loud onsists of N parti les with average parti le mass m and obeys the
ideal gas law, with temperature, T, independent of r. Then
ρ(R) dP kB T dρ 3ρ kB T
P V = N kB T ⇒ P (R) = kB T ⇒ = =− ,
m dR m dR R m
1
sin e ρ(R) ∝ R3
,
1/3 3/2
kB T 3M kB T 9 kB T 1
⇒ GM = 3R =3 ⇒ M= √ .
m 4πρ m 2 π Gm ρ1/2
M is known as the Jean's Mass, MJ , and is a riti al mass for a loud of temperature
T and density ρ. If the mass of the loud is greater than MJ the gravitational for e
dominates the thermal pressure and it will ollapse, if the mass is less than MJ the
thermal pressure an sustain the loud against its own gravitational ontra tion.
For an interstellar loud of mole ular hydrogen H2 , m = 2mp = 3.4 × 10−27 kg, a
typi al T is ≈ 100K and ρ≤ 10−19 kg/m3 ≈ 100 mole ules/cm3 ,1 we get a Jean's Mass
1
For omparison the best man-made va uum so far, reported by va uum s ientists at CERN the
European parti le physi s fa ility near Geneva (Ireland is not a member state of CERN!), is 5×
−17 −3
10 kg m
13
of
MJ ≥ 3 × 1033 kg ≈ 1000M⊙ .
Individual stars annot be this massive, the most massive stars known are only about
100M⊙ , and these are quite rare. As the loud ollapses it must break up into smaller
pie es, ea h of whi h eventually be omes a star. Stars are never formed in isolation, they
are always formed in groups In fa t the initial stages of star formation is mu h more
involved than this and is one of the most poorly understood stages of stellar history. Gas
louds are heated up by osmi rays and an lose heat through radiation. Also magneti
elds and rotation are often signi ant as well as large s ale turbulent gas ows indeed
star formation is often losely asso iated with sho k waves in the interstellar medium
whi h is learly not a ounted for in the above analysis. Nevertheless the on lusion that
stars form in lusters rather than individually is orre t.
2.4 How long does the ollapse take?

For simpli ity ignore P for the moment and onsider a gas ollapsing under free fall (a
proto-star)
GM (r)
r̈ = − .
r2
Let r = R(t), M0 = M (R) = const
GM0
R̈ = − .
R2
dR
Substitute v= dt
dv GM0
=− 2
dt R
dv GM0 dR d GM0
⇒v =− 2 =
dt R dt dt R

1 d 2 d GM0
⇒ (v ) =
2 dt dt R
1 GM0
⇒ v2 = + const
2 R
Suppose the loud ollapses from rest starting from t = 0, so v(0) = 0, with initial radius
R(0) = R0 ⇒ const = − GM R0 . Dene the ollapse time, tc , to be the time it takes to
0
ollapse to zero size (ignoring heating), so R(tc ) = 0. Taking v(t) to be negative for
14
t > 0, sin e the loud is ollapsing,

p 1 1 1/2
v = − 2GM0 −
R R0
√
dt 1 R
⇒ = −√ q
dR 2GM0 1 − R
R0
Z tc Z 0 √
1 RdR
⇒ dt = − √ q .
0 2GM0 R0 1 − R
R0
R
Substituting u= R0 = sin2 θ we get
s √
Z 1
R03 udu
tc = √ =
2GM0 0 1−u
s
π/2 Z
R03 2 sin2 θ cos θdθ
=
2GM0 0 cos θ
s
π R03
= .
2 2GM0
4πρ0 3
Assuming a onstant initial density ρ0 (so M0 = 3 R0 ) we obtain
r
3π
tc =
32Gρ0
the free fall ollapse time (zero pressure).

e.g. interstellar gas loud ρ = 10−19 kg/m3 ⇒ tc = 2 × 1014 s = 6 × 106 years. For
omparison: ρ⊙ = 1.4 × 103 kg/m3⇒ tc = 30min
2.5 Virial Theorem

For a stati , spheri ally symmetri star in equilibrium, i.e. the radius r(t) of a shell is
independent of t, so from (2.2) we get
GM (r)ρ(r) dP
− 2
= (2.4)
r dr
The total gravitational potential energy is
Z Z R Z R
GM (r)ρ(r) GM (r)ρ(r) 2 dP 3
EGrav = − dV = −4π r dr = 4π r dr
Star r 0 r 0 dr
15
Now assuming that P (0) is nite and P (R) = 0
Z R Z R
dP 3
EGrav = 4π r dr = 4π[P (r)r 3 ]R
0 − 12π P (r)r 2 dr =
0 dr 0
Z R
= −12π P (r)r 2 dr
0
Dene the average pressure
R RR
StarP dV 4π 0 P (r)r 2 dr
P̄ = = (2.5)
V V
so
EGrav
P̄ = − .
3V
We an also relate P̄ to the thermal energy for an ideal monatomi gas using the Equipar-
tition Theorem from thermodynami s whi h states that, for a gas in thermal equilibrium
at temperature T , every dynami al degree of freedom asso iated with the gas parti les
arries energy kB T /2. For a monatomi gas for example, in whi h the only dynami al
degrees of freedom are translations in ea h of the three independent dire tion of spa e,
ET hermal = 32 N kB T.
Essentially the kineti energy of ea h parti le is
m 2
ǫKinetic = (v + vy2 + vz2 )
2 x
kB T
and the equipartition theorem states that there is an energy
2 asso iated with motion
in ea h dire tion in spa e
m 2 m m kB T
v = vy2 = vz2 =
2 x 2 2 2
giving
3
ET hermal = N ǫKinetic = N kB T.
2
To use the equation of state for an ideal gas in modelling a star we repla e the pressure
P with its average value in the star, P̄ in (2.5), and the temperature T with the average
temperature T̄ , dened similarly
R RR
Star T dV 4π 0 T (r)r 2 dr
T̄ = = .
V V
We an then write
EGrav
P̄ V = N kB T̄ = 23 ĒT hermal = −
3
16
For a (non-relativisti ) ideal gas in thermodynami equilibrium we obtain the Virial
Theorem
2ĒT hermal + EGrav = 0 (2.6)
If a star ontra ts slightly, in quasi-equilibrium,

2 then T̄ → T̄ + ∆T̄ and EGrav →
EGrav + ∆EGrav (we expe t ∆EGrav < 0 for a ontra tion). Then the virial theorem
says: ∆ET hermal = − 12 ∆EGrav > 0 ⇒ ∆T̄ > 0, as the star ontra ts the temperature
goes up. Half of |EGrav | goes into heating the star up. The other half must be radiated
away ( onservation of energy) as thermal radiation.
If a star expands, temperature de reases and ∆EGrav > 0. Half of ∆EGrav omes from
thermal energy, the other half must ome from somewhere else! e.g. when the Sun runs
out of hydrogen fuel (about 5 billion years from now), initially the ore will ontra t and
heat up as EGrav is onverted into ĒT hermal . Eventually the ore temperature be omes
high enough for He to burn whi h releases more nu lear energy whi h is transferred to
the outer envelope whi h provides the extra energy ne essary for the outer envelope to
expand and ool - T̄ de reases and the star be omes a Red Giant. Without some sour e
of energy other than gravitation a star in quasi-stati equilibrium ould never expand,
ontra tion would be a one-way pro ess.
2.6 Stellar Equilibrium

For a stati equilibrium (normal state) ṙ = 0 we had
dP GM (r)ρ(r)
=− (2.7)
dr r2
The mass in a shell of thi kness δr is
δM = 4πρ(r)r 2 δr
dM
= 4πr 2 ρ(r) (2.8)
dr
Combining these two equations we get

1 d r 2 dP
= −4πGρ(r). (2.9)
r 2 dr ρ(r) dr
We have one dierential equation for two unknown fun tions P (r), ρ(r). We need another
relation between P and ρ to solve the problem and there will be more on this later.
2
Quasi-equilibrium means that the state hanges very slowly relative to the time s ale ne essary to
a hieve thermal equilibrium. Imagine hanging the radius of a star in a series of very small steps,
allowing time for thermal equilibrium to be maintained at ea h step.
17
For the moment we take a short ut by making a simplisti assumption, that the
4π 3
density is uniform: ρ(r) = ρ0 = const so M (r) = 3 ρ0 r . Then it is probably easier to
solve (2.7) dire tly:
dP (r) 4πGρ20 r 2πGρ20 r 2

=− ⇒ P (r) = C −
dr 3 3
where C is a onstant. Assuming that the pressure vanishes at the surfa e of the star
2πGρ20 R2
P (R) = 0 xes C = 3 giving
2πGρ20 2
P (r) = (R − r 2 )
3
and the entral pressure is
2πGρ20 R2 π 1/3 4/3

Pc := P (0) = = GM 2/3 ρ0 .
3 6
For the Sun: M⊙ = 2 × 1030 kg, R⊙ = 7 × 108 m and ρ⊙ = 1.4 × 103 kgm−3 = 1.4gcm−3
giving
Pc = 1.3 × 1014 N m−2 .
Better, more sophisti ated, models give
Pc = 2.5 × 1016 N m−2 and ρc = 1.62 × 105 kgm−3
ρ = const is too simplisti ; better is to use the relation between pressure, density and
temperature: the equation of state, P (ρ, T ).
2.7 Equation of State

i) Ideal gas law: P V = N kB T
ρkB T
P =
m
where m is the average mass of the gas parti les.
mp +me mp
For a star made of ionised hydrogen, H + + e− , ⇒ m = 2 ≈ 2
2ρkB T
P = .
mp
e.g. in our simple model (ρ = ρ0 = const)

1/3
mp Pc 1 4π Gmp 2/3 1/3 Gmp M
Tc = = M ρ0 = .
2kB ρ0 4 3 kB 4R
For the Sun we get
Tc ⊙ = 6 × 106 K
18
(more a urate models give Tc ⊙ = 1.56 × 107 K ).
Noti e that
1/3 1/3 4/3

1 4π 4/3 1 4π mp Pc
Pc = GM 2/3 ρ0 = GM 2/3
2 3 2 3 2kB Tc
4
6 2kB Tc4
⇒ Pc = (2.10)
πG3 mp M2
so, if two stars have the same entral temperature, the more massive one has a lower
entral gas pressure.
The pressure is modied slightly for a more realisti hemistry. For a star made of
a mixture of Hydrogen and
4 He, both ompletely ionised, let X denote the mass
fra tion of Hydrogen and Y the mass fra tion of Helium, so X + Y = 1. Then if
np is the number density of Hydrogen nu lei (protons) and nHe the number density
Helium nu lei (ea h with mass mHe = 4mp )
ρ = (X + Y )ρ = mp np + mHe nHe = mp np + 4mp nHe
with
mp n p Xρ
X= ⇒ np =
ρ mp
mp nHe Yρ
Y =4 ⇒ nHe = .
ρ 4mp
The number density of gas parti les is

Xρ Yρ ρ 3
n = ne + np + nHe = 2np + 3nHe = 2 +3 = 2X + Y ,
mp 4mp mp 4
sin e there is an ele tron for every proton and 2 ele trons for every Helium nu leus.
So the average mass of the gas parti les is
ρ mp
m= = .
n 2X + (3/4)Y
For example the Sun has X = 0.7 and Y = 0.3 on average, but it is onverting
Hydrogen into Helium only in the ore and has already used up half of its fuel so in
the ore one expe ts X = 0.35 and Y = 0.65 at the moment giving
mp
m= ,
1.2
rather than the valuemp /2 that we got for pure Hydrogen. This redu es the pressure
7
by a fa tor 1.2/2=0.6. Using a entral temperature Tc ⊙ = 1.56 × 10 K and ρc =
5 −3 16 −2
1.62 × 10 kgm the ideal gas law gives Pc = 2.5 × 10 N m . As the Sun ages
X in the ore will de rease and Y will in rease, thus redu ing the pressure.
19
ii) Thermal Radiation pressure:
a 4
P = T
3
a = 4c σSB = 7.5 × 10−16 Jm−3 K −4 .
7
Using Tc ⊙ = 1.6 × 10 K gives
PRadiation = (2.5 × 10−16 ) × (6.6 × 1028 ) = 1.7 × 1013 N m−2

whi h is mu h less than the gas pressure Pc = 2.5 × 1016 N m−2 .
However for a given temperature the radiation pressure is independent of the mass
of the star while we have seen that the gas pressure de reases as ∼ 1/M 2 as the
mass is in reased. The pressures are equal when
4
6 2kB Tc4 a 3 6 2kB
= Tc4 ⇒ M = 2
= 4 × 1031 kg = 20M⊙ .
πG3 mp M 2 3 a πG3 mp
For the Sun it is gas pressure that supports it, radiation pressure is negligible. For a
star with M > 2M⊙ radiation pressure be omes more signi ant. For M > 100M⊙
thermal pressure dominates and the enter of the star is supported purely by thermal
radiation pressure!
iii) Degenerate Fermi Gas: Relevant for very high densities of parti les with intrinsi
1
spin of
2 ~. Su h parti les obey the Pauli ex lusion prin iple and are alled Fermions.
If we try to put a lot of Fermions (e.g. ele trons or neutrons) into a small box, the
ex lusion prin iple generates a pressure whi h resists the addition of more parti les.
This is alled the degenera y pressure be ause, as we shall see, it depends on the
fa t that quantum me hani al energy levels have nite degenera y. It is partly
responsible (together with ele trostati repulsion) for the for e on the soles of your
feet that stops you falling through the oor
Consider N parti les onned in a ubi al volume V = a3 . Quantum states are su-
perpositions of standing waves, with wave-length λ. Dene the wave-number, k, to be
k= 2π ~ 2π
λ . Quantum me hani s only allows dis rete wave-ve tors k = (mx , my , mz ) a ,
where mx , my and mz are integers, be ause a whole number of wavelengths must t
in the box of width a.
The number of quantum states in a volume d3 k = dkx dky dkz of wave-ve tor spa e
a
3
is dmx dmy dmz = 2π d3 k at least for large integers mx , my , mz .
Now the number of quantum states in a spheri al shell of thi kness dk and radius k
in k-spa e is
a 3 V
4πk2 dk = 4πk2 dk.
2π (2π)3
h
In quantum me hani s the de Broglie relation is p = λ = ~k, so the number of
quantum states for parti les with momentum in the range p → p + dp is
V p 2 p 4πV
4π d = 3 p2 dp =: g(p)dp.
(2π)3 ~ ~ h
20
g(p) is alled degenera y fa tor. Be ause of ele tron (or neutron) spin, there an be
two states with the same p, spin up and spin down ⇒ for ele trons and neutrons
8πV 2
g(p) = p . (2.11)
h3
Possible values of a parti le's energy, ǫ(p) are also dis rete (when onned to a box).
Let the lowest possible energy be ǫ0 . If N ele trons go into the box, all possible
energy levels will be lled sequentially up to some top energy ǫF = ǫ(pF ). ǫF (the
Fermi-energy) and pF (the Fermi-momentum) depend on N and are given, when N
is large, by approximating the sum over energy levels, or equivalently momenta, by
an integral
Z pF Z pF
8πV 8πV 3
N= g(p)dp = 3 p2 dp = p ⇒
0 h 0 3h3 F
1/3
3 N 1/3 3n
pF = h =h (2.12)
8π V 8π
where n := N/V is the number of parti les per unit volume.
We want to al ulate P (ρ, T ), but the al ulation is simplied when T is "small"

(by whi h is meant kB T ≪ energy level dieren es, ∆ǫ), in whi h ase the Fermi
gas is said to be degenerate. The internal energy is then
Z pF Z pF
8π
E= ǫ(p)g(p)dp = 3 V ǫ(p)p2 dp (2.13)
0 h 0
p2F p
with ǫ(p) = 2m for non-relativisti momenta and ǫ(p) = m2 c4 + p2 c2 for relativis-
ti momenta (the relativisti form is used below as the non-relativisti form an then
follows by taking c → ∞).
The rst Law of Thermodynami s, for an adiabati pro ess, is
3
dE = −P dV + T dS + µdN.
For
T ≈ 0 this redu es to dE = −P dV + µdN so, using (2.13) and noting that
∂g(p)
∂V = g(p)
V from (2.11),
N

∂E E 8πV 2 ∂pF
P =− = − − 3 ǫ(pF )pF
∂V N V h ∂V N

∂E 8πV ∂pF
µ= = 3 ǫ(pF )p2F .
∂N V h ∂N V
3
µ is the hemi al potential, i.e. the energy required to add one extra parti le we shall prove
momentarily that µ = ǫF .
21
Now from (2.13)

∂pF 1 ∂pF 1
=− pF , = pF
∂V N 3V ∂N V 3N
hen e
E 8πV pF E
P =− + 3 ǫ(pF )p2F = − + nǫ(pF )
V h 3V V
8πV 2 pF 8πV
µ = 3 ǫ(pF )pF = 3 ǫ(pF )p3F = ǫ(pF ).
h 3N 3h n
ǫF N − E
⇒ P = , µ = ǫF . (2.14)
V
Energy
εF
. .
. .
. .
ε1
ε0
Dis rete energy levels, lled a ording to the ex lusion prin iple.
22
Now
E
P = ǫF n − =
r V r
Z
2 p2F 8π pF 3 8πmc2 pF p2 2
= mc 1 + 2 2 − 1 + p dp
m c 3 h h3 0 m2 c2
q Z zF p
8πmc2 3 1 2 3 2 2 p
= 3
(mc) 1 + zF zF − 1 + z z dz (z = , dp = mcdz)
h 3 0 mc
Z
8πm4 c5 zF z4 p
= √ dz (z = sinh α, 1 + z 2 = cosh α, dz = cosh αdα)
h3 0 3 1 + z2
q
πm4 c5
q
2 2 2 2
= zF z −1 1 + zF + ln zF + 1 + zF
h3 3 F
(
8π m4 c5 5
−→ 2πm15 h3 zF , zF ≪ 1, pF ≪ mc (non-relativisti )
4 c5
4
3h3 zF , zF ≫ 1, pF ≫ mc (relativisti )
(
3 2/3 h2 5/3
P = 8π 5m n non-relativisti
(2.15)

3 1/3 ch 4/3
8π 4 n relativisti .
It has been assumed that the temperature is negligible, these expressions are valid,
provided kB T ≪ ǫF . This is the equation of state for a old degenerate gas of
ele trons or neutrons but note that, when n is large, ǫF an be so large that T is
negligible even for temperatures of many thousands of degrees.
Summary: Three dierent equations of state, depending on the situation:

ρkB T
 Ideal gas pressure

 m

 aT 4 Radiation pressure
P (ρ, T ) = 3 (

 β 5/3 non-relativisti
∝ n β= where n = ρ/m.


4/3 relativisti
To solve the problem ompletely we need T (r) - depends on energy transport and energy
produ tion.
23
2.8 Energy Sour es in Stars
Three main nu lear rea tions in stars
1. Hydrogen burning. 4H →4 He, relevant for T ≥ 106 K (the pp- hain). This is
the main sour e of power in the Sun.
p ◦B eO + p◦ γO
BB O
BB O
BB O
BB
!
=⋆
/ d⋆ / 3 He
{{ 33
{ { 33
{{ 33
{{ 33
p◦ νe 33 p O ◦
33
33
3
⋆ / 4 He
E

p ◦C νO e p ◦
CC
CC

CC
C!
=⋆
/ d⋆ / 3 He
|| O
| O
|| O
|| O
||
p◦ e+ p γ
Net result:
4
4p → He + 2e+ + 2νe + 2γ (26.2M eV ).
Ea h step has a hara teristi time-s ale:
p + p → d + e+ + νe τ ≈ 1010 yr = 3 × 1017 s
3
p+d→ He τ = 1s
3 3 4
He + He → He + 2p τ = 3 × 105 yrs
(d =2 H is a deuterium nu leus, an isotope of Hydrogen with one neutron and one

proton). The rate determining step, pp → d, is so slow that it annot be measured
in the laboratory and must be al ulated. Estimates range from 6 × 109 yr to
14 × 109 yr .
For the Sun X ≈ 0.7 on average, but in the ore Hydrogen is being turned into
Helium and X is smaller. At the present day the Sun has used up approximately
half of its nu lear fuel, so the mass fra tion of Hydrogen in the ore is loser to
X ≈ 0.35,
ρc 0.35 × (1.6 × 105 kgm−3 )

ρc = 1.6×105 kgm−3 ⇒ np = X = = 3.3×1031 m−3 .
mp 1.7 × 10−27 kg
24
The rate per unit volume for the rea tion p+p→d is then
np 3.3 × 1031 m−3

= = 1.1 × 1014 s−1 m−3 .
τ 3 × 1017 s
The power/unit volume requires an extra fa tor of 1/2 be ause it requires two
p+p→d rea tions to produ e one
4 He nu leus, liberating 26.2 M eV :
1
wpp = (1.1 × 1014 s−1 m−3 ) × (26.2M eV ) =
2
1
= (1.1 × 1014 ) × (26.2 × 1.6 × 10−13 Js−1 m−3 )
2
= 231W m−3
wpp 231
⇒ = = 1.4 × 10−3 W kg−1 . (2.16)
ρc 1.6 × 105
This is the dominant sour e of energy produ tion in the Sun.
The rate at whi h energy is produ ed is very sensitive to the temperature: typi ally
1 np
τ ∼ ρT k , with k in the range 10 ∼ 20, and w∝ τ ∼ ρ2 T k .
2. CNO- y le
For M ≫ M⊙ entral temperatures are higher and another rea tion ki ks in
p .

αO 2 12 C /o o/ o/ /o / γ

p / 15 N 13 N e+
O 6 + νe
v
e+ + νe 15 O 13 C o p
U
O
O
O

γ o o/ o/ o/ o/ 14 N s γ
O
Net result:
4
4p → He + 2e+ + 2νe + 3γ (23.8M eV ).
Slowest step: p +14 N →15 O + γ , τ = 5 × 108 yr = 1.5 × 1016 s. But in the Sun only
≈ 0.6% is
14 N and only 1.6% of the Sun's energy is CN O. For stars with mass a
little more than the Sun the CNO y le dominates.
25
3. Helium burning. The CN O y le does not produ e C, N or O. 12 C an be
produ ed by Helium burning. In the Sun, on e H is used
4
up, He an burn
(10
8 <T <2× 108 K ).
4
He +4 He ⇋8 Be unstable, τ = 2.6 × 10−16 s
4
He +8 Be ⇋12 C ∗ unstable, τ = 1.8 × 10−16 s
12
C ∗ →12 C + γ (7.3M eV )
On e
12 C is present other rea tions are possible, su h as 4 He
+12 C→16 O+γ . The
CN O- y le an ontribute signi antly, even when H is plentiful, if M is high
8
enough. For M > 8M⊙ temperatures an rea h T ∼ 5 × 10 , whi h is hot enough
to burn higher nu lei up to
56 Fe. Nu lei higher than 56 Fe are destroyed in stellar
interiors, not reated.
The main sour e of energy in the Sun is the pp- hain, the CNO y le and Helium
burning are important in more massive stars and Helium burning will be ome important
when the Sun runs out of Hydrogen 5 billion years from now. We saw that the pp- hain
gave a value
wc
= 1.4 × 10−3 W kg−1 . (2.17)
ρc
If the whole Sun were burning nu lear fuel we would expe t the total power produ tion
to be
wc
L = M⊙ = (2 × 1030 ) × (1.4 × 10−3 ) = 2.8 × 1027 Js−1
ρc
whereas the observed luminosity is L⊙ = 3.8 × 1026 Js−1 , implying that slightly more
than 10% of the Sun's mass is involved in nu lear burning. This is in the entral ore
where it is hottest. We ould try to estimate the radius, Rc , of the nu lear burning
entral region by demanding that it ontains ≈ 1/10 of the total mass. Assuming the
ore has uniform density,
1/3
M⊙ 4π 3 3M⊙
= ρRc ⇒ Rc = .
10 3 40πρ
For example ρ = ρ⊙ = 1.4 × 103 kg m−3 gives Rc = 3.2 × 108 m = 0.5R⊙ while ρ = ρc =
1.6 × 10 kgm−3 gives Rc = 6.7 × 107 m = 0.1R⊙ . Clearly the rst is an overestimate and
5
the se ond an underestimate. A more a urate value is Rc ≈ 0.3R⊙ .

An important onsequen e of the pp- hain nu lear rea tion is neutrino produ tion.
Using
4 × 1026 eV s−1
L⊙ = 4 × 1026 Js−1 = = 2.5 × 1045 eV s−1 = 2.5 × 1039 M eV s−1
1.6 × 10−19
and ea h pp- hain rea tion produ es 26.2M eV so the number of rea tions per se ond is
2.5 × 1039 /26.2. Ea h rea tion produ es two neutrinos so the expe ted rate of neutrino
produ tion is

2.5 × 1039
2× s−1 = 1.9 × 1038 s−1
26.2
26
giving an expe ted ux at the Earth of
1.9 × 1038 s−1 1.9 × 1038

2 = = 6.7 × 1014 m−2 s−1 .
4πd⊕−⊙ 4π(150 × 109 m)2
Only 1/2 of this is measured and this is the famous Solar neutrino problem whi h has
important impli ations for parti le physi s some of the ele tron neutrinos in the Sun
are hanging into other parti les, a dierent kind of neutrino, as they pass between the
Sun and the Earth.
2.9 Energy transport

Only the entral ore of stars a tually generates energy by nu lear rea tions and this must
be transported to the surfa e in order to explain the surfa e luminosity. Energy an be
transported by three dierent me hanisms: onve tion, ondu tion or radiation. For a
typi al star like the Sun, radiation dominates energy transport in the ore ( onve tion
dominates further out). For the most part ondu tion is not signi ant for normal stars
and will not be onsidered here (though it an be important in the nal stages of a star's
life: white dwarves).
2.9.1 Radiation
Absolute luminosity = total power radiated
L⊙ = 3.8 × 1026 J/s from the surfa e.
For an obje t with surfa e area A at a temperature T radiating into empty spa e
L = σSB AT 4 (2.18)
(this is alled bla k body radiation). For a spheri ally symmetri al obje t with radius R
L = 4πR2 σSB T 4
ln L = 4 ln T + 2 ln R + ln(4πσSB ).
This relation is evident if the luminosity of a large number of stars is plotted against
their surfa e temperature on a log-log plot. Stars with similar radii fall on a straight
line. Su h a plot is alled a Hertzprung-Russel diagram.
27
L(LO. )
supergiants
4
10
10 2 ma
red
giants
in s
equ
1 enc
e
−2 white
10 dwarves
−4
10
40000 20000 10000 5000 2500

T(K)
Hertzsprung-Russel diagram.
(It is onventional to plot the temperature as in reasing to the left.) Most (normal)
stars fall on a single line, alled the main sequen e, indi ating that they all have mu h
the same radius. There is a group of hot smaller stars falling below the main sequen e,
alled white dwarves. There are also two groups of larger stars above the main sequen e,
indi ating larger than normal radii, one slightly to the right (indi ating a lower surfa e
temperature a group alled red giants) and with lower luminosity than the other.
Members of the higher luminosity group have very large radii are alled supergiants.
The Hertzsprung-Russel diagram indi ates the surfa e luminosity of stars but the en-
ergy sour e is only in the ore, the outer layers are not hot enough to burn nu lear fuel.
There must therefore be a ux of energy through any shell of matter of radius r<R
L(r)
Flux: F(r) = . (2.19)
4πr 2
The ux of energy through a given sphere depends on r in general, as ea h nu lear
burning shell is produ ing energy. Nu lear rea tions in the ore of a star produ e energy
at a rate per unit volume w(r) and w will depend on r through ρ and T , w(ρ, T ) and
this requires a se ond equation of state (e.g. w∝ ρT q ).
In equilibrium onservation of energy demands that the total power radiated outward
from the outer surfa e of a shell at radius r of thi kness δr , L(r + δr), must equal the
28
total power going into the shell at r plus the power produ ed by nu lear rea tions within
the shell:
L(r + dr) − L(r) = 4πr 2 drw(r)

dL
⇒ = 4πr 2 w(r) (2.20)
dr
Ignoring onve tion and thermal ondu tion, the energy ux, F(r), is driven by radiation
pressure,
a 4 4σSB
PRad = T a= .
3 c
dPRad
It is reasonable to assume that F∝ dr
c dPRad
F =− (2.21)
ρκ dr
where κ is alled the "opa ity" and has dimensions of area over mass.
L(r) c dPRad ca d 4
F(r) = 2
=− =− (T )
4πr ρκ dr 3ρκ dr
d 4 3 ρκ
(T ) = − L(r). (2.22)
dr 4πr 2 ca
This equation relates the temperature gradient at r to the luminosity and is a se ond
important equation for understanding stellar equilibrium, equal in importan e to (2.3).
To understand where κ omes from we need to know something about the history of a
photon trying to leave the ore of the star. Consider a parti le (a photon) hitting a slab
of material (the plasma), of area A and thi kness d ontaining N parti les, onsisting of
parti les (ele trons and protons) ea h presenting area σ to the in oming parti le (σ is
alled the ross-se tion). The probability of the proje tile hitting a slab parti le is
Nσ
P=
A
N σd
=
V
P Nσ
⇒ = = nσ,
d V
where V = Ad is the volume of the slab. Dene the mean free path of one of the in oming
parti les as being the average distan e they travel before being s attered. This an be
estimated as the value of d at whi h P=1
1
l= .
nσ
The opa ity κ is inversely proportional to the mean free path and is dened as
1 nσ
κ= = . (2.23)
ρl ρ
29
As was done for the dynami al equation for stellar equilibrium are (2.3), where the
density an be eliminated to give a se ond order dierential equation for the pressure,
(2.20) an be used to eliminate the luminosity from (2.22) to give a se ond order equation
for the temperature, leading to

1 d r 2 dP
= −4πGρ(r) (2.24)
r 2 dr ρ dr

1 d r 2 dT 4 3
= − w(r). (2.25)
r 2 dr κρ dr ac
We need to know equations of state P (ρ, T ) and w(ρ, T ) and κ(ρ, T ) to ompletely spe ify
the system. A full understanding of the solutions requires a detailed knowledge of the
opa ity, whi h depends sensitively on the hemi al omposition of the star. This an
only be a hieved numeri ally and will not be taken further here.
2.9.2 Conve tion

The above equations assume that it is radiation that dominates energy transport, but
this is not always true. Consider a rising blob of gas in a star, from r to r + ∆r . (∆
designates hanges in the the environment and δ designates hanges in the blob itself ).
T+ δ T
T+ ∆ T
P+ δ P r+ ∆ r
P+ ∆ P
ρ+ ∆ ρ ρ+ δ ρ
T, P, ρ
T, P, ρ r
Assume:
1. the pressure equalizes immediately
∆P = δP.
ρ kB
2. ideal gas law for the star: P = m kB T , ∆P = m (T ∆ρ + ρ∆T )
∆P ∆ρ ∆T
⇒ = + .
P ρ T
30
3. the blob expands adiabati ally ( i.e.without heat energy being ex hanged between
the blob and its environment): P V γ = const or P ρ−γ = const (γ = 5/3 for a
monatomi gas).
δP δρ ∆P
=γ = .
P ρ P
The blob will ontinue to rise and onve tion sets in if the density of the blob is less
than the ambient density, δρ < ∆ρ while for δρ > ∆ρ it will stop rising and start sinking.
So there will be no onve tion if
1 ∆P δρ ∆ρ ∆P ∆T
= > = −
γ P ρ ρ P T

∆T 1 ∆P dT γ − 1 T dP
⇒ > 1− ⇒ >
T γ P dr γ P dr

dT
< γ − 1 T dP

⇒ dr γ P dr

where the limit ∆r → 0 has been taken (both dT /dR and dP/dr are negative and γ > 1).
The star is stable against onve tion, and energy is transmitted primarily by radiation,
if
dT γ−1 T dP

dr < dr .

Radiation γ P
So, assuming quasi-stati equilibrium and using equation (2.3) with r̈ = 0,

3κρ L γ − 1 T dP
<
16πac r 2 T 3 γ P dr
3κρ L γ − 1 T GM ρ
⇒ 2 3
<
16πac r T γ P r2
3κ γ − 1 T4
⇒ L< GM.
16πac γ P
Conversely onve tion is more important than radiation for energy transport, if
L(r) γ − 1 T 4 G 16πac γ − 1 PRadiation 16πGc L

> = := (2.26)
M (r) γ P 3κ γ P κ M critical
where P = PRadiation + PGas is the total pressure. In the entre of the Sun we saw
earlier that PGas >> PRadiation , so P = PGas in the solar entre. For example, using
realisti values for the entral ore of the Sun, take a monatomi gas γ = 5/3 with
Pc = 2.5 × 1016 N m−2 and Tc = 1.6 × 107 K , κc = 0.1m2 kg−1 (a value that will be
derived in the next se tion)
L
= 3 × 10−3 Js−1 kg−1 .
M critical
31
Compare this to the power per unit mass arising from Hydrogen burning in the solar
ore (2.16)
wc
= 1.4 × 10−3 Js−1 kg−1 . (2.27)
ρc
Model the Sun as a entral Hydrogen burning ore of radius Rc with onstant wc and ρc ,
surrounded by an envelope in whi h there is no nu lear burning so w = 0 in the envelope.
Then, for the ore
L(Rc ) 4πRc3 wc /3 wc L
= 3
= <
M (Rc ) 4πRc ρc /3 ρc M critical
and we on lude that the entre of the Sun is not onve tive.
The observed value at the surfa e:
L⊙ 3.8 × 1026
= = 1.9 × 10−4 Js−1 kg−1 (2.28)
M⊙ 2 × 1030
is an order of magnitude less than (2.27) but the radiation pressure at the surfa e is
also de reased, by the mu h greater fa tor of (T⊙ /Tc )4 = (5.8 × 103 )4 /(1.6 × 107 )4 =
1.7 × 10−14 , so that (2.28) is greater than (L/M )critical near the surfa e: the outer layers
of the Sun are onve tive. In fa t the Sun be omes onve tive about 5/7 of the way out
from the entre: energy is transferred by radiation, and equation (2.22) is valid, from
the entre out to about 0.7R⊙ and energy is transferred by onve tion the rest of the
way, where equation (2.22) is not valid. This is due in part to the fa t that gas pressure
dominates radiation pressure in the ore but we saw earlier that this is not true for more
massive stars stars a few times the mass of the Sun are onve tive in the ore and
equation (2.22) is not appli able to su h stars. The theory of onve tive energy transport
is not as simple as that of radiation transport and will not be pursued further here.
2.10 Mass-Luminosity Relation

Nu lear rea tions in stars lead to typi al internal temperature Tc ≈ 107 K . If this were
the surfa e temperature the luminosity of the Sun would be
2
Lnaive = 4πR⊙ σSB Tc4 = 2.3 × 1040 J/s, (2.29)
using Tc = 1.6 × 107 K , whereas the observed solar Luminosity is
L⊙ = 3.8 × 1026 J/s (2.30)
be ause surfa e temperature is a tually T⊙ = 5800K ≪ Tc . What auses this dis rep-
an y?
A thermal photon trying to es ape from the ore s atters o plasma parti les
4 with a
mean free path l= 1

ρκ , the photon performs what is alled a random walk. Consider a
4
ele trons are more important than protons in this regard be ause the ross-se tion is inversely pro-
portional to the the square of the target's mass
32
photon taking k steps, ea h of the same length l but in random dire tions
~ = ~l1 + ~l2 + . . . + ~lk

D
k
X
D2 = ~l1 .~l1 + ~l2 .~l2 + . . . + ~lk .~lk + 2 ~li .~lj
i<j
Pk ~ ~
For k large i<j li .lj = 0 be ause the dire tions are random
√
D2 = kl2 ⇒ D= kl
For a photon to es ape from the entre of the Sun to the surfa e: D = R⊙ ,
2
R⊙
k=
l2
With no s attering, the photon would travel from the entre to the surfa e in a straight
R⊙
line at the speed of light in a time τ0 = c . S attering retards the photon: the distan e
R2⊙
it has to travel is a tually kl = l . The photon is retarded and a tually takes a time
R⊙
τ = τ0
l
l
to rea h the surfa e. This retardation de reases the power radiated by a fa tor
R⊙ , sin e
a power is an energy divided by time. Sin e energy is onserved these two powers are
related by a ratio of times,
τ0 l
L⊙ ≈ Lnaive ≈ Lnaive .
τ R⊙
So 4 4
L⊙ T⊙ 8 6000
l ≈ R⊙ ≈ R⊙ ≈ (7 × 10 ) × ≈ 10−4 m
Lnaive Tc 107
Sin e the photon has to travel a distan e
2 /l >> R
R⊙ it takes a time
⊙
2
R⊙
τ≈ ≈ 5 × 105 years
cl
to get to the surfa e! As long as the Sun is in equilibrium the total power radiated is
the same at the ore as at the surfa e, L⊙ = 3.8 × 1026 W , the ore is very mu h hotter
be ause it is well insulated.
This an be used to estimate the opa ity κc in the entre of the Sun. Generally κc
depends rather sensitively on the temperature and the hemi al omposition, but a rough
estimate an be obtained using l ≈ 10−4 m and ρc = 1.6 × 105 kg ⇒
1
κc = ≈ 10−1 m2 kg−1 . (2.31)
lρc
33
We an now derive a relation between the surfa e luminosity and the mass of a typi al
star, su h as the Sun. In our simple model, with uniform density, we found the ore
1

4π 1/3 Gmp 2/3 ρ1/3 . This is related to the surfa e temperature
temperature Tc = 4 3 kB M 0 Ts ,
using the Stefan-Boltzmann law and the photon diusion length l, by
l
L = 4πR2 σSB Ts4 ≈ 4πR2 σSB Tc4
R
!4
1 4π 1/3 Gmp 2/3 1/3
≈ 4πσSB M ρ0 lR
4 3 kB
4 4/3
7/3 1 1 Gmp 4 8/3 4/3
≈ (4π) σSB M ρ0 lR.
4 3 kB
4π 3

3 1/3 −1/3
Now M ≈ 3 ρ0 R ⇒R≈ 4π M 1/3 ρ0 . The magi is that, with l ≈ 1/κρ0 , the
density an els leading to
4
π2 Gmp M3
L≈ σSB
48 kB κ
independent of ρ0 , where κ is an average opa ity in the star. Ignoring approximate
numeri al fa tors we get the Mass-Luminosity relation
4
Gmp M3
L ≈ σSB (2.32)
kB κ
Observationally main sequen e stars satisfy L ∝ M 3.5 while M 2.3 works better for low
mass stars.
2.11 Eddington Limit

This is an upper limit on the luminosity a star an have for a given mass be ause radiation
pressure an blow a star apart!
The radiation ux was given as
c dPRad c dPRad L(r)

F(r) = − =− = .
ρκ dr nσ dr 4πr 2
dPRad
dr has dimensions of for e/volume so dene a for e/volume, fRad ,
dPRad nσ nσ L(r)
fRad = − = F=
dr c c 4πr 2
For e/volume = a eleration × density so
fRad σ L(r)
aRad = = (2.33)
ρ mp c 4πr 2
34
is the a eleration due to Radiation pressure. The a eleration due to gravity at a
distan e r from the entre is
GM (r)
g=−
r2
so stability at the surfa e requires that, at r = R, aRad < |g| ⇒
4πR2 mp c GM 4πmp Gc
L< 2
= M := LEddington (2.34)
σ R σ
At the surfa e take σ to be the ross-se tion for a photon to s atter o an ele tron in
free spa e (this is the Thomson ross-se tion, σT = 6.7 × 10−29 m2 ) so
4πmp Gc
LEddington = M (M in kgs, L in Js−1 )
σT
LEddington M
⇒ = (3.3 × 104 ) ,
L⊙ M⊙
where in the last equation the luminosity and mass have been expressed in solar units,
L⊙ = 3.8 × 1026 Js−1 and M⊙ = 2.0 × 1030 kg.
2.12 Maximum Stellar Mass

Combining Mass-Luminosity relation
4
Gmp
L ≈ σSB M 3 ρ⊙ l
kB
with the Eddington limit

4πGmp c
L< M
σT
gives
4
Gmp 4πGmp c
⇒ σSB M 3 ρ⊙ l < M
kB σT
k4 c 1
M 2 < 4π 3B 3 (2.35)
G mp σSB σT ρ⊙ l
M 2 < 1.8 × 1065 kg
M < 4.3 × 1032 kg ≈ 200M⊙ .
2.13 Minimum Stellar Mass

As a gas loud (a "protostar") ollapses, the temperature in reases. Assume an ideal gas
mp Pc
in quasi-stati equilibrium. Temperature at the ore of the loud Tc = 2kB ρc (no fa tor
1/2 for neutral hydrogen atoms). Eventually Tc gets high enough to ionize Hydrogen,
35
and as Tc in reases further either it gets high enough for nu lear rea tions to swit h on
( 106 K ), or the density gets high enough for degenera y pressure due to ele trons to halt
ontra tion.
Assume non-relativisti ele trons, degenera y pressure
2/3
3 h2 5/3
P = n
8π 5me e
where ne is the number density of ele trons (ne = np = ρ/mp if the star is pure Hydro-
gen). Total ore pressure
2/3 5/3
2ρc kB Tc 3 h2 ρc
PT = +
mp 8π 5me m5/3
p

π 1/3 4/3
For our simple model (ρ = const) we had a ore pressure of PT = 6 GM 2/3 ρc .
Therefore
2/3
2ρc kB Tc 3 h2 5/3 π 1/3
+ n = GM 2/3 ρ4/3
mp 8π 5me e 6 c
Solving for kB T and using ne = np = ρc /mp gives
2/3 2/3
1 π 1/3 1 3 h2 ρc
kB T = Gmp M 2/3 ρ1/3
c −
2 6 2 8π 5me m2/3
p
:= Aρc1/3 − Bρ2/3
c .
dT
Maximum temperature requires
dρc =0
1 −2/3 2 −1/3
Aρ − Bρc =0
3 c 3
3
2B 1 A
⇒ ρ−1/3
c = ⇒ ρc = ,
A 8 B

max A A 2 A2
kB T =A −B =
2B 2B 4B
π
2/3
G2 m2p M 4/3
= 48 2/3 2
3 h 1
2 8π 5me 2/3 mp
8/3
5 π 4/3 G2 mp M 4/3 me
=
21/3 6 h2
Tcmax 4/3
∝M . (2.36)
Fusion requires Tcmax ≥ 1.5 × 106 K so
!3/4
21/3 6 4/3 2
h kB × (1.5 × 106 ) M⊙
M > Mmin = 5 π
8/3
≈ 2 × 1028 kg =
G2 mp me 100
36
In summary it has been shown that stellar masses should all lie in the range
0.01M⊙ < M < 200M⊙
Observationally
0.08M⊙ < M < 150M⊙ .
2.14 Degenerate Stars

2.14.1 White Dwarves
When nu lear fuel runs out, a star ools by radiation and thermal pressure an no longer
sustain it against gravitational ollapse. It ollapses until degenera y pressure be omes
dominant. For a degenerate gas of ele trons the pressure is
(
3 2/3 h2 5/3
P = Knγe = 8π 5me ne non relativisti ele trons

3 1/3 hc 4/3
8π 4 ne relativisti ele trons

3ne 1/3
The Fermi momentum is pF = h 8π . The mass of an atomi nu leus is mN = Amp ,
where A is the number of neutrons plus the number of protons in the nu leus. Assuming
full ionisation the ele tron density is ne = ZnN where the atomi number Z is the positive
m n
harge on the nu leus. Therefore ρ = mN nN = (Amp )( nZe ) = pµ e where µ = Z 1
A ≈ 2 for
most stable atomi nu lei. So
γ
µρ µρ
ne = ⇒ P =K (2.37)
mp mp
i) Non relativisti : our simple model with uniform density gave pressure
1

4π 1/3
Pc = 2 3 GM 2/3 ρ4/3 . If Pc were due to ele tron degenera y pressure
2/3
3 h2 µρ 5/3 1 4π 1/3
= GM 2/3 ρ4/3
8π 5me mp 2 3
2/3
1 3 1/3 1 4π 1/3 5me mp 5/3
ρ = GM 2/3
22/3 4π 2 3 h2 µ

1 4π 5me G mp 5/3 2/3
ρ1/3 = 1/3 M
2 3 h2 µ

1 20πme G 3 mp 5 2
ρ= M (2.38)
2 3h2 µ

(6.2 × 108 ) M 2
= kg m−3 .
µ5 M⊙
With µ = 1/2
2
10 M
ρ = 2 × 10 kg m−3 . (2.39)
M⊙
37
Che k: for onsisten y ρ must give non relativisti ele trons, whi h requires pF <<
me c for realisti values of M.

e 1/3

pF h 3n8π h 3 µρ 1/3
= =
me c me c me c 8π mp
1/3
h 3 µ 1 4π 5me G mp 5/3 2/3
= M (using (2.38))
me c 8π mp 21/3 3 h2 µ

20G π 3 1/3 2/3 mp 4/3
= 1/3 M
2 hc 3 8π µ
2/3 4/3
5G 2π mp
= M 2/3
hc 3 µ

0.9 M 2/3
= 4/3 . (2.40)
µ M⊙
This ratio must be signi antly less than one for the non-relativisti degenera y
pressure to be valid, for example µ = 1/2 ⇒ M << 0.3M⊙
ii) For M ≈ 0.3M⊙ and larger ele trons are relativisti .
1/3 4/3 1/3

3 hc µρ 1 4π
= GM 2/3 ρ4/3
8π 4 mp 2 3
1/3 4/3
1 3 hc µ 2π 1/3
1
= 2/3 GM 2/3
22/3 2π 4 mp 2 3
2/3
hc 3 µ 4/3
M 2/3 =
4G 2π mp
3/2 2
3 hc µ
M= := MChandraskhar = 1.7µ2 M⊙ . (2.41)
16π G mp
This has assumed uniform density and better models imply
MChandrasekhar ≈ 5.8µ2 M⊙ .
√
This is independent of the density. Thus for low density (non-relativisti ) M∝ ρ while
for higher densities (relativisti ) M is independent of ρ. A graph of M as a fun tion of
ρ looks something like this:
38
M(ρ)
M Ch
0.1M .
8 9
10 10 ρ (kg m −3)
There are no stable stars held up by ele tron degenera y pressure for M > MCh . More
sophisti ated models (numeri al integration with ρ(r)) ⇒ MCh = 5.8µ2 M⊙ ≈ 1.4M⊙
for µ = 1/2. Stars with M > MCh ollapse atastrophi ally when their fuel runs out to
a small dense ore and rebound as a Supernova (type II), leaving behind a neutron star
or a bla k-hole.
For M < MCh the star is stable and very small a White Dwarf, held up by ele tron
degenera y pressure. To al ulate the radius onsider e.g. M = M⊙ (ele trons are

3ne 1/3 8π

me c 3
relativisti ). Then pF = me c, ⇒ me c = h 8π ⇒ ne = 3 h . Charge neutrality
mp np m n
ne = np , ρ = µ = pµ e ,
8π mp me c 3
ρ= ≈ 2 × 109 kgm−3
3 µ h
Assuming onstant density
1/3 1/3
3M⊙ 3 × 2 × 1030
R= = m
4πρ 4π × 2 × 109
≈ 0.62 × 107 m ≈ 6200km ≈ 10−2 R⊙ ,
hen e "dwarf".
e.g. Sirius B is a "white dwarf" with R = 0.007R⊙ = 5000km Numeri al al ulations
⇒ white dwarves should be stable up
11
to ρ ≈ 10 kg m
−3
2.14.2 Neutron stars

For M > MCh , the star ontinues to ollapse until ele trons and protons are ompressed
together and they ombine and form neutrons by inverse β -de ay. In free spa e n →
p + e− + ν̄e (β -de ay). Inverseβ -de ay: e− + p + energy → n + νe . We then have a
star onsisting solely of neutrons (M⊙ ≡ 10
57 neutrons) made of "nu lear matter". For
ρ > 1.4 × 1014 kg/m3 ele trons begin to fuse with protons resulting in atomi nu lei
whi h are neutron ri h, i.e. have a larger ratio of neutrons to protons than usual. The
39
ex ess neutrons leak out in a pro ess known as neutron drip until eventually there are
no protons left and no nu lei, we have a state of pure neutron matter. The entire star is
ρnuclear ≈ 3 × 1017 kgm−3 .
a giant ball of neutrons with a density of
Neutrons are fermions, they have spin 1/2, and there is therefore a degenera y pressure
asso iated with then. If the mass is not too high, the "star" an be supported by neutron
degenera y pressure.
1. Non-relativisti :
2/3
3 h2 5/3
P = n ,
8π 5mn n
ρ 4×1014
e.g. ρ = 4 × 1014 kgm−3 ⇒ nn = mn = 1.7×10−27 m
−3 = 2.4 × 1041 m−3 . The Fermi
momentum is then
1/3
3nn pF
pF = h = 2.0 × 10−20 kgms−1 ⇒ = 0.04
8π mn c
and neutrons are non-relativisti at these densities. The al ulation exa tly paral-
lels that of white dwarves, ex ept that µ=1 and me → mn in (2.38), in reasing
the density by (mn /me )3 = 6.2 × 109 , giving
2
18 M
ρ = 3.8 × 10 kgm−3
M⊙
for non-relativisti neutrons. Neutrons be ome relativisti for M ≈ M⊙ (equation

(2.40) with µ = 1), when pF ≈ mn c ⇒ ρ ≈ 6.1 × 1018 kg m−3 .
2. for M > M⊙ neutrons are relativisti and we get (2.41) with µ=1
M = 1.7M⊙ .
This ignores nu lear intera tions between the neutrons and in reality the upper limit for
the mass of a stable neutron star depends on the strength of the nu lear for es. It lies
somewhere between 1.5M⊙ ∼ 2.7M⊙ .
The radius for M = 1.7M⊙ and ρ = 6.2 × 1018 kgm−3 is:
1/3
3M
R= ≈ 5km
4πρ
Generi ally, more sophisti ated al ulations give R ≈ 10km.
40
M(ρ)
unstable
region Final mass
M Ch
} depends on
nuclear
=
1.4M . equation of
state.
10
8
10
9
10
14
10
18 ρ (kg m −3)
2.14.3 Pulsars
A pulsar is a rotating neutron star with a strong magneti eld. All stars are believed
to have magneti elds, the Sun for example has a eld B⊙ ≈ 2 × 10−4 T ( ompared to
that of the Earth, B⊕ ≈ 3.1 × 10−5 T ). A pulsar is a rotating dipole and rotating dipoles
emit ele tromagneti radiation, thus losing energy.
5
B
α
m0
Energetic
particles
2π
e.g. pulsar in the Crab Nebula: ω = 190 radians s−1 , T = ω = 3 × 10−2 s, and it is
slowing down slowly over time, ω̇ = −2.4 × 10−9 s−2 .
How large an ω be? Assuming, again, uniform density for simpli ity:
GM
2 2 GM 4πGρ
ωmax2
R= ⇒ ωmax = 3 =
R r R 3
r
2π R 3 3π
Tmin = = 2π = .
ωmax GM Gρ
q
1012
For a neutron star with M = M⊙ , R = 10km: Tmin = 2π
6.7×10−11 ×2×1030
s = 5 × 10−4 s
5
This is not a tually the sour e of the radiation that is dire tly observed. The ele tromagneti pulses
seen from a pulsar are due to radiation from beams of harged parti les emitted along the axis of the
dipole and sweeping past the Earth like a lighthouse beam.
41
This is the extreme value for the period, most pulsars are rotating with angular velo ity
less than ωmax so
GM 3π
ω2R < ⇒ ρ> ,
R2 GT 2
so, for the Crab pulsar, the average density must be greater than 1.6 × 10−14 kg m−3 ,
otherwise it would break up due to entrifugal for es.
The total time-averaged power radiating by a rotating dipole with magneti dipole
moment m0 is:
2 µ0 4 2 2
P= ω m0 sin α,
3 4πc3
so, for the Crab pulsar with T = 3 × 10−2 s, we know that the average density must be
more than 1.614 ks m−3 , otherwise it would disintegrate due to entrifugal for es.
The rotational energy is E = 21 Iω 2 so we expe t
dE
P=− = −Iω ω̇.
dt
For a sphere of onstant density I = 25 M R2
P 5 µ0 ω 3
ω̇ = − =− (m0 sin α)2
Iω 3 4πc3 M R2
For M = M⊙ , R = 10km, ω = 190s−1 , ω̇ = −2.4 × 10−9 s−2
⇒ (m0 sin α) = 2.7 × 1027 Cm2 s−1 .
27 )
So the magnitude of ~
|B| at the surfa e is B≈ µ0 m0
4π R3 = 10−7 (2.7×10
1012 T = 2.7 × 108 T ≈
1012 B⊙ . This is not unreasonable, if the Sun had the same dipole moment at the entre
the eld at the surfa e of the Sun would be related to that of the neutron star by a fa tor
3
R
R⊙ ≈ 1/(7 × 104 )3 ≈ 3 × 10−15 , whi h is not too far o.
5 µ0 (m0 sin α)2

We now have a relation between ω and ω̇ , ω̇ = −Aω 3 with A= 3 4πc3 M R2
dω
= −Aω 3
dt
dω
= −Adt
ω3
1
− (ω −2 − ω0−2 ) = −A(t − t0 )
2
where ω0 = initial period at time t0 . Choose the time of formation t0 = 0, then the
urrent age is

1 1 1 1 1 1 ω
t= 2
− 2 < 2
=
2A ω ω0 2A ω 2 |ω̇|
giving an upper bound on the age.
1 ω 1 190
For the rab pulsar t< 2 |ω̇| = 2 2.4×10−9 s = 4 × 1010 s = 1250yr . This is remarkably
lose to the orre t value as the exa t age is known form histori al re ords: the supernova
that gave birth to the Crab nebula pulsar was observed in 1054 AD.
42
2.14.4 Bla k Holes
If the nal mass of a dead star M > 3M⊙ it ollapses even further and R de reases.
1 2 GM
Naïvely the es ape velo ity from the surfa e is given by
2v = R . For v=c
2GM
R= := RS (2.42)
c2
alled the S hwarzs hild-radius. This argument is in orre t for relativisti velo ities, but
R < RS
the on lusion is orre t. If light annot es ape and the star forms a bla k hole.
e.g. for M = M⊙ : RS ≈ 3km.
Young bla k holes are often surrounded by matter whi h spirals in and heats up ⇒
bright X-ray sour es or intense radio sour es. e.g. Cygnus X-1 is an intense X-ray
sour e showing u tuations on a time-s ale 10−3 s ⇒ R < 300km and yet M >> M⊙ ;
6
Sagittarus A* is a ompa t mass at the entre of our Galaxy with M ≈ 4 × 10 M⊙ whi h
is believed to be a bla k-hole.
3M 3c 2
1 1
For a given mass the density is ρ= = 8πGR
4πR3 2 ∝ R2 ∝ M 2 e.g. for one solar mass
M = M⊙ a neutron star would have R ≈ 10km, ρ ≈ 1018 kgm−3 , while a bla k hole
19
would have R ≈ 3km, ρ ≈ 3 × 10 kgm
−3 . For M = 108 M (e.g. Andromeda Galaxy:
⊙
7 1019 −3 = 103 kgm−3 = 1 g/cc, about
entral bla k hole with M = 3 × 10 M⊙ ) ρ ≈ 8 kgm
10
the density of water!
2.15 A tive Gala ti Nu lei (AGN) and Quasars (QSO)

These are extremely energeti :
L ≈ (1010 − 1013 )L⊙ ≈ (1037 − 1040 )Js−1
Extra gala ti (extremely distant):
d ≈ 109 lyr
They are believed to be very young galaxies. The AGN re ord is L = 1041 Js−1 but
38
about 10 Js
−1 is more typi al,
L ≈ 1012 L⊙ , Eddington limit: ⇒ M ≥ 108 M⊙ .
The power radiated is equivalent to loss of mass:
1040 Js−1 1040 1

|Ṁ | = 2
= 8 2
kgs−1 ≈ 1023 kgs−1 ≈ M⊙ s−1 ≈ M⊙ yr −1 .
c (3 × 10 ) 2 × 107
Nu lear burning is only ≈ 1% e ient and it does not seem possible to produ e these
w
kinds of energies by nu lear for es. For the Sun the power per unit mass was
ρ =
1.4 × 10−3 W kg−1 . An AGN or QSO with M≈ 109 M⊙ ould only produ e
109 × (2 × 1030 ) × (1.4 × 10−3 )Js−1 = 2.8 × 1036 Js−1
43
The energy sour e annot be nu lear energy. The only other known possible sour e of
GM 2
energy is gravity. The gravitational energy of a mass M in radius R is |EGrav | = R .
For a bla k hole
2GM 1
R= ⇒ |EGrav | = M c2
c2 2
|EGrav |
Using M ≈ 109 M⊙ and |Ė| ≈ 1040 Js−1 gives
|Ė|
≈ 109 yrs, or about 1/10th the age
of the Universe. Assuming that the gravitational for e is the sour e of power for AGN's
explains why there is none observed very lose the our own Galaxy there is none left
today, we an only see them from a time mu h earlier than the present day.
44
3 Cosmology and Expansion of the
Universe
On very large length s ales (> 100M pc) the distribution of galaxies appears to be
isotropi . If we assume that all points in spa e are equivalent, i.e. we are not at a
spe ial point (this is alled the Coperni an Prin iple), then the distribution must be is
isotropi about all other points and this a tually implies that distribution of galaxies
is homogeneous on very large s ales, i.e. it is uniform. Therefore mass density ρ is
independent of position on large enough length s ales.
3.1 Cosmologi al Constant

Consider a spheri al shell of matter, radius R and thi kness δR, of mass δm = 4πR2 δRρ
4πR3 ρ
around a mass M (R) = 3 . In Newtonian Gravity, the dynami s of the shell is the
same as it would be if the entral mass were on entrated at a point in the entre and any
mass outside the shell doesn't ae t it at all if it is spheri ally distributed ( onsequen e
of the inverse square for e). The Newtonian equation of motion for the shell is
GδmM
δmR̈ = − . (3.1)
R2
We shall assume the following:
1. Ṙ ≪ c (non-relativisti velo ities);
2GM 8πG 2
2.
c2 R
= 3c2
R ρ ≪1 (weak gravitational eld, Newtonian Gravity is valid);
3. R≫ gala ti separation, so we approximate ρ by a smooth fun tion ρ(R).

Ṙ2 2GM
It might be thought that 1) and 2) are related by the virial theorem ⇒ c2
≈ c2 R
,
but the virial theorem only applies to stable gravitationally bound systems and is not
appli able to galaxies that are so far apart that they are not bound to ea h other by
gravitational for es.
q
2) and 3) ⇒ gala ti separation ≪ R ≪ 3
8πGρ c. Provided R lies in this range we
shall assume we an use equation (3.1) and immediately get the rst integral
1 GδmM
δmṘ2 − =E
2 R
1 2 GM E E
Ṙ − = = := ǫ.
2 R δm 4πR2 δRρ
45
where E is the total energy of the shell and ǫ the energy of the shell per unit mass,
whi h is a onstant if δm is onstant and has dimensions of velocity 2 . Assuming uniform
density M an be eliminated in favour of ρ,
1 2 4π
Ṙ − GR2 ρ = ǫ. (3.2)
2 3
This equation was derived using Newtonian gravity and non-relativisti physi s. In
2
relativisti physi s Newton's 1/R for e is invalid (it's only valid for small velo ities).
We should use the full power of Einstein's General theory to determine the dynami s of
the Universe and this lies outside the s ope of this ourse. Remarkably General Relativity
gives the same equation but with a dierent interpretation. In General Relativity R is
not labelling a radial position, it is a length s ale determining the physi al size of lengths
in 3-dimensional spa e. IfR = R(t) then when Ṙ > 0 spa e is said to be expanding while
when Ṙ < 0 R as the physi al distan e between
spa e is ontra ting. We an interpret
any two xed galaxies, provided their separation is of the order of 100M P c, or more.
This distin tion between the meaning of R in a Cosmologi al ontext, as dis ussed here,
and in the ontext of Newtonian gravity, as dis ussed earlier in the se tion on stellar
equilibrium, is very important. For example if the mass M inside a sphere of raduis R,
4π 3 1
M= 3 ρR , is independent of R then ρ(R) ∝ R3 . If R labelled positions in spa e then ρ
would not be independent of position, ontradi ting the Coperni an Prin iple. If instead
R des ribes osmologi al length s ales rather than position then it is perfe tly onsistent
with the Cosmologi al Prin iple for ρ R, ρ is then independent of position
to depend on
1
but it depends on the osmologi al length s ale. ρ ∝ then means that the average
R3
density goes down as R in reases (the Universe expands), but ρ is the same everywhere
at any one time.
Equation (3.2) is the same as the energy of a point parti le of unit mass moving on
the half-line R>0 in a quadrati potential, like a harmoni os illator equation but with
a negative o-e ient. Suppose ρ = ρ0 is independent of R. Dene the onstant
8πGρ0
Λ := ,
c2
whi h has dimensions of (length)2 , then
1 2 c2 Λ 2
Ṙ − R =ǫ (3.3)
2 6
(the 1/6 is a standard onvention in osmology). We an get a qualitative understanding
of the behaviour simply by plotting the potential V (R) = − Λ6 R2 ,
46
R
V(R)
If the energy ǫ > 0 then R an have any value in the range 0 < R < ∞, with |Ṙ|
in reasing with R; if ǫ < 0 then there is a repulsive barrier and R annot rea h zero.
Neither of these ases allows for a stati solution with Ṙ = 0. A stati solution is
only possible when ǫ = 0, whi h allows for R = Ṙ = 0, but this is learly unstable
the slightest deviation away from R = 0 and R will start to grow, eventually rea hing
innity. We on lude from this that ρ = const does not allow for a stable stati solution:
on su iently large s ales the distan es between galaxies annot be onstant, so ρ must
hange with time, the Universe is either expanding or ontra ting observationally it
is expanding and R(t) is growing as a fun tion of time.
Consider two galaxies a distan e R apart with M being the total amount of mass in
a sphere of radius R entred on one of the galaxies. Even if R is hanging with time it
is reasonable to assume, if mass annot be reated or destroyed, that M is onstant. In
that ase
4πρ 3 3M 1
M= R ⇒ ρ(R) = .
3 4π R3
Note that, with the interpretation of R as being the physi al distan e between the galax-
ies, ρ ∝1/R3 does not mean that ρ depends on position ρ is independent of position
but is a de reasing fun tion of time if R(t) is an in reasing fun tion of time.
For notational onvenien e let A = 2GM , then equation (3.2) gives
1 2 A
Ṙ − = ǫ. (3.4)
2 2R
47
This equation has the same mathemati al form as that of the energy of a proje tile with
unit mass thrown verti ally upward from the surfa e of the Earth, moving in a potential
A
V (R) = − 2R ,
V(R)
There is an attra tive for e towards R = 0: ifṘ > 0 initially the late time behaviour
depends on the sign of ǫ. ǫ has dimensions 2
of (velocity) so dene a dimensionless
parameter K by ǫ = − 12 Kc2 . Then there are three types of behaviour:
K<0 R in reases indenitely and Ṙ > 0 always;
K>0 R rea hes a maximum value and de reases again;
K=0 R in reases indenitely but Ṙ → 0 as t → ∞.
K =0 is a riti al value, orresponding to the notion of es ape velo ity in Newtonian

dynami s but the physi al interpretation of K in General Relativity is very dierent. K
is a measure of the urvature of 3-dimensional spa e.
• K=0 is at Eu lidean spa e;
• K > 0 is a 3-dimensional analogue of the surfa e of a 2-dimensional sphere. In

a spa e with positive K, the traje tories of two proje tiles will bend toward ea h
other, rather like the paths of two ships following great ir les on the surfa e of the
Earth.
48
Mathemati ally a 3-dimensional sphere of radius r an be des ribed by imposing
the onstraint v 2 + x2 + y 2 + z 2 = r 2 on Cartesian o-ordinates (v, x, y, z) in at 4-
dimensional Eu lidean spa e. This is a natural extension of the geometry of a ir le
in 2-dimensional Eu lidean spa e with Cartesian o-ordinates (y, z), y 2 + z 2 = r 2
(a 1-dimensional sphere), and the usual 2-dimensional
2 2 2
sphere, x + y + z = r ;
2
• K<0 is a 3-dimensional spa e in whi h traje tories diverge. It is a 3-dimensional

analogue of a hyperbola y2 − z2 = r2 in 2-dimensional Eu lidean spa e, and an be
des ribed by imposing the onstraint v 2 +x2 +y 2 −z 2 = r 2 on Cartesian o-ordinates
(v, x, y, z) in at 4-dimensional Eu lidean spa e.
The full story appears to be that, at the present time, the mass density of the Universe
at length s ales of 100M P c and greater appears to have two omponents, one with ρ=
const orresponding to a onstant mass density, and one with ρ ∝ 1/R3 orresponding
to a onstant amount of mass in a sphere of radius R(t). Denoting the latter by ρM at a
ombination of these two possibilities an produ e a stati solution. Let
ρ = ρM at + ρΛ = 3A
8πGR3
+ Λc2 /8πG
where ρM at is the density of ordinary matter. Then
8πG Λc2 R2
Ṙ2 = ρM at R2 − Kc2 +
3 3
A Λc2 R2
⇒ Ṙ2 = − Kc2 +
R 3
A 2Λc2 R
⇒ 2R̈ = − 2 +
R 3
3A
Choosing Λ= 2c2 R3
(a repulsive for e) gives a solution with no a elaration and
A A
Ṙ2 = − Kc2 + ,
R 2R
3A
so K = 2c2 R
> 0 gives a stati solution with R a onstant. ( alled Einstein's stati
universe, in whi h spa e is a 3-dimensional sphere with nite volume). This is stati , but
unfortunately unstable. The osmologi al onstant Λ was introdu ed by Einstein in order
to obtain stati solutions be ause he did not know at the time that R was hanging and
he assumed that it should be onstant. At the present day R(t) appears to be in reasing
49
but only hanges very slowly, on osmologi al time-s ales of the order of billions of years.
Nevertheless there is by now onsiderable observational eviden e that Λ is positive.
We have arrived at the Friedmann-Equation
!2
Ṙ c2 K 8πGρM at Λc2
+ = + (3.5)
R R2 3 3
This is a dynami al equation that determines the behaviour of the osmologi al length
s ale R(t) for a given mass density ρM at and onstants K and Λ. In the General Theory
of Relativity it is valid for relativisti velo ities Ṙ ≈ c and for strong gravitational elds
8πG 2
3c2
R ρM at ≈ 1, but still requires that R >> gala ti separations.
3.2 Redshift-distan e relation

Consider two galaxies A and B with oordinates (rA , θA , φA ) and (rB , θB , φB ). Choose
π
o-ordinates so that θA = θB = 2 and φA = φB = 0. Take rA > rB , xed for ea h
galaxy (the oordinates are xed to the galaxies and are alled o-moving o-ordinates,
analogous to Lagrangian oordinates). The distan e between the galaxies is R(t)(rA −rB )
where R(t) is the osmologi al s ale fa tor.
Consider a beam of light passing between A and B. It travels a distan e R(t)δr in the
R(t)δr
time δt = c . Suppose light leaves A at time t1 and arrives at B at time t2
Z t2 Z rB
dt 1
= dr. (3.6)
t1 R(t) c rA
At a later time light leaves A at t1 + ∆t1 , light would rea h B at t2 + ∆t2

Z t2 +∆t2 Z Z t2
dt 1 rB dt
= dr =
t1 +∆t1 R(t) c rA t1 R(t)
Z t2 +∆t2 Z t1 +∆t1
dt dt
⇒ = .
t2 R(t) t1 R(t)
Take ∆t1 and ∆t2 R(t) = R(t1 ) is
to be the inverse of opti al frequen y and assume
essentially onstant between t1 and t1 +∆t1 and R(t) = R(t2 ) between t2 and t2 +∆t2 (this
always is an extremely good approximation: for opti al frequen ies ∆t ≈ 10
−15 s while
9 16
the s ale fa tor R only hanges appre iably on time s ales of order 10 yrs ≈ 3 × 10 s),
then
1 1
∆t1 = ∆t2
R(t1 ) R(t2 )
R(t1 ) ∆t1 ν2
= = (3.7)
R(t2 ) ∆t2 ν1
If R(t2 ) > R(t1 ) then ν2 < ν1 , i.e. the light is redshifted.
50
Assuming R(t) is a slowly varying analyti fun tion of time it an be Taylor expanded
around t2 as
Ṙ
R(t1 ) = R(t2 ){1 − (t2 − t1 )H + . . .} H= ,
R t2
with (t2 − t1 )H << 1. Let sAB = c(t2 − t1 ) (time-of-ight distan e) then
R(t2 ) sAB H
≈ 1 + (t2 − t1 )H = 1 + .
R(t1 ) c

Ṙ
If t2 = t0 is the present day then H0 :=
R is alled the Hubble onstant. Suppose
t0
light left a distan e galaxy at a time t = t1 and arrives at our teles ope at the present
day t0 = t2 . Then, assuming t0 − t << H0−1 , we have
δν ν1 − ν2 ν1 R(t0 ) sAB H0
z := = = −1= −1= .
ν2 ν2 ν2 R(t) c
z is alled the redshift of the galaxy, it is a measure of the amount by whi h light
from a distant galaxy is shifted toward the red end of the spe trum, i.e. towards longer
wave-lengths. So
H0
z= sAB ∝ sAB (3.8)
c
giving the redshift distan e relation
c
sAB = z. (3.9)
H0
H0 has dimensions of (time)

−1 but is usually quoted as kms−1 M pc−1 . The urrent best
estimate of H0 from observations is
H0 = 67.3 ± 1.2 km s−1 M pc−1

:= h × 100 km s−1 M pc−1 where h = 0.673 ± 0.012.
This means that a galaxy at a distan e of 1 M pc exhibits a redshift orresponding to a

velo ity of about 70 kms−1 . The linear relation (3.9) is only valid for "small" (t2 − t1 ),
δν
i.e.
ν ≪ 1.
For example there is a luster of galaxies in the onstellation of Virgo (≈ 2500 galaxies)
with an average redshift c δν
ν = 1150kms
δν
−1 ⇒
ν = 0.00383 giving sAB = H0 ν =
c δν
16M pc = 50M lyrs. The urrent redshift re ord is δν ν ≈ 10 whi h gives a naïve redshift
4
distan e relation of sAB ≈ 4.5×10 M pc, but this is not the true physi al distan e be ause
δν
ν > 1 is not small.
1 1 1 17 1 10
H0 has dimensions of time: H0 = h (3×10 s) = h 10 yr ≈ 14 billion years whi h is the
9
approximates age of the universe (more a urate modelling gives t0 = 13.7 × 10 years).
If R(t) were linear then R would have been zero 14 billion years ago, but this is only a
1 1
rough approximation. In this approximation R0 = H0 c = h × 1026 m is the approximate
size of the observable universe.
51
3.3 Friedmann Equation
Hubble dis overed the linear redshift-distan e relation in 1929. In re ent years the obser-
vational data have be ome good enough to go beyond a linear approximation for R(t).
The Friedmann equation (3.5) an be written as
8πG Λc2 c2 K
H2 = ρM at + − 2
3 3 R
whereρM at is the mass density of matter, as in stars, gas and dust in galaxies.
When Λ = K = 0,
3
ρΛ=K=0
M at = H 2.
8πG
Using present day values
3
H 2 = h2 × (1.89 × 10−26 kgm−3 ) = 8.54 × 10−27 kg m−3 .
8πG 0
Observationally, ounting the visible luminous galaxies gives ρLuminous ≈ 3×10−29 kgm−3 .
Using observations of orbital dynami s of galaxies in lusters (Kepler's Law) ρM at ≈
10−27 kg m−3 (equivalent to about 1 proton per ubi metre). The most a urate value
we have to date omes from the dynami s of galaxies on osmologi al s ales and is
ρM at = 2.5 × 10−27 kg m−3 . So ρM at < ρΛ=K=0
M at . Returning to the more general ase with
Λ 6= 0
8πGρM at c2 K Λc2
H2 = − 2 +
3 R 3
8πGρM at 2
c K Λc2
⇒ 1= 2
− 2 2+
3H H R 3H 2
Dene the following 3 onstants using present day values H0 and R0 = R(t0 )
8πGρM at (t0 ) A c2 K Λc2

ΩM := 2 = 2 3, ΩK := − and ΩΛ := .
3H0 H 0 R0 H02 R02 3H02
Then only two of these are independent, sin e
ΩM + ΩK + ΩΛ = 1.
These an be related to a Taylor expansion of R(t),

1
R(t) = R0 + (t − t0 )Ṙ0 + (t − t0 )2 R̈0 + . . .
2 !
1 R̈ 0
= R0 1 + H0 (t − t0 ) + (t − t0 )2 + ... ,
2 R0
52
where, in what I hope is an obvious notation, Ṙ0 = Ṙ(t0 ) and R̈0 = R̈(t0 ). Dene the
de eleration parameter, q0 = − 21 HR̈2 R

0
, then
0 0
8πGρM at R2 Λc2 R2 3 A
Ṙ2 = − c2 K + , ρM at =
3 3 8πG R3
A 2
Λc R 2
⇒ = − c2 K +
R 3
AṘ 2Λc2 RṘ
⇒ 2ṘR̈ = − 2 +
R 3
R̈ A Λc 2 4πGρM at Λc2
⇒ =− 3 + =− +
R 2R 3 3 3
R̈ A Λc 2
⇒ 2
=− 2 3
+
H R 2H R 3H 2
1
⇒ 2q0 = ΩM − ΩΛ .
2
So ΩΛ and ΩM are dire tly related to the parameters in a Taylor expansion of R(t).
What we a tually measure is the redshift z, but we an onvert from t to z . From
ν1 R0
(3.7), with t2 = t0 and t1 = t, and the denition z= ν2 −1 we have
R(t) =1+z
R0 dz dR Ṙ ż
⇒ dz = − dR ⇒ =− ⇒ =− .
R2 1+z R R 1+z
Now use this to write the Friedmann equation as
!2 3 2
Ṙ A c2 K Λc2 2 R0 2 R0
= 3 − 2 + = ΩM H 0 + ΩK H 0 + ΩΛ H02
R R R 3 R R

ż 2 = (1 + z)2 (1 + z)3 H02 ΩM + (1 + z)2 H02 ΩK + H02 ΩΛ .
This is a non-linear dierential equation (no approximations) for z(t) in terms of the
onstants H0 , ΩM , ΩK and ΩΛ . Now we invoke (3.6) with B being our Galaxy, so we
set rB = 0, to write the distan e dA to a galaxy A at redshift z as
Z t0 Z t0 Z t0
dt (1 + z)dz
dA = R0 rA = cR0 =c (1 + z)dt = −c
R ż
Z zt t t
dz
⇒ dA = c p . (3.10)
0 (1 + z)3 H02 ΩM + (1 + z)2 H02 ΩK + H02 ΩΛ
This is an exa t non-linear redshift-distan e relation.
For example
cz
dA (z) =
H0
if ΩΛ = 1, ΩM = ΩK = 0, while

2c 1
dA (z) = 1− √
H0 1+z
53
ifΩM = 1, ΩΛ = ΩK = 0
Although the ase ΩΛ = 1, ΩM = ΩK = 0 looks very like (3.9) they are not exa tly
the same sin e the distan e dA = R0 rA is not the same as the time-of-ight distan e
sAB = c(t0 − t) (with tA = t and tB = t0 ), though their dieren e is negligible at small
z << 1.
Observational data plotting gala ti distan e against redshift (the verti al

axis is the magnitude, whi h is essentially the logarithm of the distan e).
Taken from http://supernova.lbl.gov/
Observationally
(
0.0487 ± 0.0011 "ordinary matter" (neutrons, protons)
ΩM = 0.315 ± 0.017
0.266 ± 0.017 "Dark matter" (unknown, not yet dete ted in the lab)
ΩΛ = 0.685 ± 0.017 often alled "Dark Energy".
1 = ΩM + ΩK + ΩΛ ⇒ ΩK = −0.000 ± 0.024

ΩK = 0 ⇒ 1 = ΩM + ΩΛ ⇒ q= 12 21 (1 − ΩΛ ) − ΩΛ = 14 − 34 ΩΛ = −0.264 ± 0.013 < 0
implying that the expansion rate of the Universe is a elerating.
54
A value of 0.27 for ΩM means that the density in matter at the present time is
3H02
ρM at = 0.31 × = 2.9 × 10−27 kgm−3 ,
8πG
whi h is equivalent to about 1.5 protons per ubi metre. We shall see later that this
annot all be due to ordinary matter with whi h we are familiar, protons and neutrons.
Only about 1/7 an be protons and neutrons, the other 6/7 appears to be some new form
of matter about whi h very little is known this is known as Dark Matter be ause it
is not visible. The mass density in protons and neutrons, known as baryoni matter
be ause protons and neutrons are lassied as 'baryons' by parti le physi ists, is then
ρb = 17 ρM at = 4.1 × 10−28 kgm−3 . This is still ten times the amount of visible matter
in the Universe, in the form of luminous stars and galaxies, but this is not in onsistent
it just means that a lot of the ordinary matter is emitting visible light. There are
other observations however that imply that the whole of the 2.9 × 10
−27 kg m−3 in matter
annot be protons and neutrons, this is overed later when the nu lear synthesis of light
elements in the very early Universe and the ratio of proton to neutron density is dis ussed.
3.4 The Friedmann Models

The Friedmann equation is
A Λc2 R2
Ṙ2 = − c2 K + .
R 3
Pursuing the analogy with 1-dimensional parti le me hani s:

1 2 1 Λc2 R2 A c2 K
Ṙ − + = −
2
|{z} 2 3 R | {z2 }
| {z }
kineti energy potential energy total energy
A 2 2
Think of V (R) = − 2R − Λc 6R as the potential energy per unit mass of a parti le moving
in one dimension.
55
lambda <0
lambda =0
Einstein’s static sol.
lambda >0
R
The potential V (R) plotted as a fun tion of R.
The behaviour of solutions depends, among other things, on the sign of K. Res ale
R to set K = ±1 (or zero). Look for solutions with Ṙ > 0 and R(0) a non-negative
onstant (possibly zero). The most general ase, with the three onstants K , A and Λ all
non-zero requires numeri al solution. Analyti solutions an be found in various spe ial
ases, by setting one or other of the onstants to zero.
q
2
i) Empty Models: A = 0 ⇒ Ṙ = c ΛR
3 −K
Z
dR dR
q = cdt ⇒ q = ct
ΛR2 ΛR2
3 −K 3 −K
a) Λ = 0, K = 0 ⇒ R = const
b) Λ = 0, K = −1 ⇒ R = ct, Milne-universe (R(0) = 0)
q
Λ
) Λ > 0, K = 0 ⇒ R(t) = R(0) exp 3 ct
q q q
3 Λ 3
d) Λ > 0, K = 1 ⇒ R(t) = Λ cosh 3 ct de Sitter spa e (R(0) = Λ)
q q
3 Λ
e) Λ > 0, K = −1 ⇒ R(t) = Λ sinh 3 ct
56
q q
3 Λ
f) Λ < 0, K = −1 ⇒ R(t) = Λ sin 3 ct Os illating universe
ii) Matter dominated: Λ=0

r
A dR
Ṙ = − c2 K ⇒ dt = q
R A
− c2 K
R
q
R
a) K = 0, dt = A dR⇒ t = 3√2 A R(t)3/2 − R(0)3/2 ;
with initial ondition R(0) = 0 (universe "started" at t = 0 with zero size)
2
t = √ R(t)3/2
3 A
1/3
9A
R(t) = t2/3 .
4
q
A
b) K = 1, Ṙ = R − c2 ; solution in parametri form:
A
R(ψ) = (1 − cos ψ)
2c2
A
t(ψ) = 3 (ψ − sin ψ) (Cy loid)
2c
Che k:
A
dR dψ 2 sin ψ c sin ψ
Ṙ = = A 2c =
dψ dt 2c3
(1 − cos ψ) 1 − cos ψ
r 2
2
c 1 − 2cA R − 1 r
A 4c2 R 4c4 2
= 2c2 R
= − 2R
A
2cR A A
r
A
= − c2 .
R
3.5 Mi rowave ba kground

The universe is "glowing" at T = 2.725 ± 0.001K (⇒ λ = 2mm)
i) Ordinary matter: In sphere of radius R areN galaxies, ea h mass mG . The sphere

ontains Mass M = N mG = 4π 3
3 R ρM at , ⇒ ρM at = 3N4πmG R13 = 8πG
3A 1
R3
1
ρM at ∝ .
R3
57
ii) Thermal Radiation, energy density ǫ. A sphere of radius R ontains energy E =
4π 3
3 R ǫ, but E ∝ ν (for thermal radiation kB T ≈ hν ). As R in reases, wavelength λ
stret hes, λ = νc ⇒ ν de reases ⇒ ν ∝ R1 , ⇒ E ∝ R1 , ⇒ R4 ǫ = const, ǫ ∝ R14 . Let
ǫ 3B 1
ρRad = c2
= 8πG R4
, with B a onstant, be the mass density equivalent to ǫ. Then
1
ρRad ∝ .
R4
Radiation mass density ontributes to the Friedmann equation,
!2
Ṙ A B Λc2 c2 K
= + + − 2 (3.11)
R R3 R4 3 R
At the present day, assuming K = 0, ρM at = 2.5 × 10−27 kg m−3 For thermal radiation
4σSB T 4 ǫ
ǫ= c so ρRad = c2 = 4.642 × 10−31 kg m−3 , with T = 2.725◦ K . So, at the present
day,
ρRad
= 1.9 × 10−4 .
ρM at
At earlier times
ρRad (t) ρRad (t0 ) R0 R0
= = (1.9 × 10−4 ) .
ρM at (t) ρM at (t0 ) R(t) R(t)
For matter dominated expansion (ignoring Λ),
1/3
9A
R(t) = t2/3 ∝ t2/3 ,
4
where t is in se onds. The energy density in radiation equals that in matter when
2/3
t0 1
= ⇒ t = (1.9 × 10−4 )3/2 t0 = 2.6 × 10−6 t0 ,
t 1.9 × 10−4
a few millionths of its present age. With t0 = 13.7 × 109 yr

we get that ρRad = ρM at
when t = 36, 000 yr . For t < 36, 000 yr the energy density in radiation dominates over
that in matter, so it is safe ignore A and the Friedmann equation be omes
!2
Ṙ B Λc2
= 4+ .
R R 3
B Λc2
But we also have
R4
≫ 3 at the same time (though not today), so in the early universe
(t < 104 yr ) a very good approximation is
B
Ṙ2 =
R2
√ √ √
d
⇒ ṘR = B ⇒ 21 dt (R2 ) = B ⇒ R2 = 2 Bt + a. With initial ondition R(0) = 0
we get
√
R(t) = (2 B)1/2 t1/2 .
58
Summary
!2 r !
10 Ṙ Λc2 Λ
t > t0 = 10 yr = R(t) ∝ exp ct
R 3 3
!2
Ṙ A
36, 000yr < t < t0 = R(t) ∝ t2/3 matter dominated
R R3
!2
Ṙ B
t < 36, 000yr = R(t) ∝ t1/2 radiation dominated
R R4
K
R2 never was and never will be signi ant.
When the universe was younger, it was hotter
1
T ∝ν∝ .
R
R0
At the present day: R0 ≈ 1026 m, T = 3K , at t = 36, 000yr , R = 104 ⇒ T = 3 × 104 K .
All the matter (mostly Hydrogen and Helium) was ionised, when T > 4000K . This
o urred when
R0 t0 1.4 × 1010 yr
≈ 103 ⇒ = 109/2 = 3 × 104 ⇒ t = = 500, 000yr
R t 3 × 104
(a more a urate gure is ts = 370, 000 yrs). Before this time the matter in the Universe
was an ionised plasma, afterwards it is mostly neutral Hydrogen unless it gets re-ionised
due to the heat from stars.
t→0
As R ∝ t1/2 for small t we get Ṙ −−→ ∞. We annot trust the Friedmann equation
ba k to t = 0. We an only go ba k to some time ti (a fra tion of a se ond) where the
temperature and energies are still understood and repla e our ignoran e of t < ti with
initial onditions R(ti ).
3.6 The Horizon Problem and Ination

Mi rowave photons from two points 180◦ apart in the sky ome from the "surfa e of last
s attering" when neutral hydrogen was formed at ts ≈ 370, 000yr when T ≈ 4000K .
So two diametri ally opposite points in the sky appear to be the same temperature (to
within a few parts in 106 ), so presumably they were in thermal onta t at some point in
the past. But this is in onsistent with our model.
The radius Rs of the surfa e of last s attering an be al ulated. Let our own Galaxy
sit at r=0 and we observe photons oming in from the surfa e of last s attering at a
o-ordinate distan e rs . In a short time intervaldt at an intermediate time t, a photon
dt
travels a physi al distan e −R(t)dr = cdt ⇒ dr = −c R(t) (dr < 0 be ause the photon is
travelling inwards, towards us)

Z 0 Z t0
dt
dr = −c .
rs ts R(t)
59
The physi al radius at ts is
Z t0
dt
Rs = R(ts )rs = cR(ts ) .
ts R(t)
For ts < t < t0 we have R(t) = bt2/3 . Therefore

Z t0 h it
1 dt 2/3 c 1/3 0
Rs = R(ts ) c = bt s 3t =
b ts t2/3 b ts
" #
2/3

1/3 1/3
t0 1/3
= 3cts t0 − ts = 3cts −1 .
ts
With t0 ≈ 1.4 × 1010 yr, ts ≈ 3.7 × 105 yr we get
m
Rs ≈ c × 105 × 30 yr = (3 × 108 ) × (3 × 106 ) × (3 × 107 s) ≈ 3 × 1023 m.
s
Let the distan e a photon ould have travelled at time ts , sin e t = 0, be dH . For t < ts
2 1
let R(t) = b′ tn (n = 3 for matter dominated and n= 2 for radiation dominated). Now
Z ts −n+1 ts
dt t c
dH = cb′ tns ′ n
n
= cts = ts
0 bt −n + 1 0 1−n
1
e.g. for n= 2 ⇒ dH = 2cts but then
" 1/3 #
Rs 3 t0
= − 1 ≈ 50.
dH 2 ts
t0
ts
t=0
2d H
2Rs
60
How an two points on the surfa e of last s attering, 180◦ apart in the sky and therefore
a physi al distan e 2Rs apart be in thermal equilibrium with ea h other when 2Rs ≈
50(2dH ) is 50 times the distan e a photon ould have travelled sin e the beginning of the
Universe? This is known as the Horizon Problem.
3.6.1 Inationary Universe

Possible solutions
1. Friedmann equation is wrong - it breaks down for some t < ts (must happen at
some early time anyway, sin e annot allow Ṙ → ∞) (e.g quantum gravity).
2. Friedmann equation is orre t, but hange the R.H.S. In the "inationary universe"
pi ture it is assumed that a very large positive osmologi al onstant "swit hed
on" for a very brief period at a very early time (t ∼ 10−35 s) and was large enough
q
Λ
2 Λc2 R2 ct
to dominate the dynami s. If Λ dominates Ṙ = 3 and R = R(0)e 3
exponential expansion for a period of time between t1 and t2 . This solves the
horizon problem if R(t2 )/R(t1 ) ≈ 1025 .
As a by-produ t of ination we get a natural explanation of why K ≈ 0. Non-zero

K is asso iated with urvature of 3-dimensional spa e, for example positive urvature,
K > 0, results in parallel lines onverging, like great ir les interse ting on the surfa e of
a 2-dimensional sphere. The greater the radius of the sphere the smaller the urvature,
for example it is di ult to dete t the urvature of the Earth's surfa e on length-s ales
of a few metres. If K 6= 0 and is signi ant before ination, when t < 10−35 s, then its
signi an e is de reased by a fa tor of 10
25 if R(t) in reases by 10
25 and the relevan e of
K in the Friedmann be omes negligible for all times after the period of ination. This
ould explain why attempts to measure K today give a null result.
3.7 The rst 3 minutes (Weinberg)

Looking ba k at t < 36, 000yr ≈ 1012 s when T = 104 ◦ K , R ∝ T1 ∝ t1/2 . kB T is an
energy, in fa t kB T ≈ √1t if the energy is given in M eV (1 M eV ≈ 10
10 K ) and t in
se onds.
1
T (1010 K) ≈ T (M eV ) ≈ p
t (secs)
It is remarkable how mu h we an dedu e about the early Universe from this equation.
61
Time Temperature/
Energy q
Λ
ct
R∝e 3 ↑
1.37 × 1010 yr 2.7K t0 , present day
R ∝ t2/3 Matter dominated ↓
4.56 × 109 yr 4K Solar System formed
109 yr 5K Galaxies formed. Era of Quasars and a tive gala ti nu lei
Heavy elements are reated in stars and supernovae
370, 000yr 4000K Hydrogen ionises, surfa e of last s attering
R ∝ t2/3 Matter dominated ↑
104 yr 25, 000K R ∝ t1/2 Radiation dominated ↓
No heavy elements; Plasma, p, 4 He, e− , 7 Li, D, γ , ν , ν̄
4 1 7
75% p and 25% He by mass,
8 neutrons, 8 protons
100s 0.1M eV α-parti les disintegrate as do D and Li, Nn /Np = 1/7
9
(10 K ) Era of nu leo-synthesis
4s 0.5M eV photons in thermal ba kground an produ e e+ , e− pairs
9
(5 × 10 K) p, n, e+ , e− , ν, ν̄, γ
1s 1M eV β -de ay, n → p + e− + ν̄ , starts to deplete neutrons
−2
10 s 10M eV p + e− ↔ n + ν works both ways
11
(10 K ) protons and neutrons in thermal equilibrium, Np /Nn = 1
5 × 10−5 s 150M eV p and n evaporate into quarks and gluons
(1.5 × 1011 K) Plasma: q, q̄, e− , e+ , µ− , µ+ ν, ν̄, γ, g
10−10 s 100GeV Ele tromagnetism and weak nu lear for e
unify into the ele tro-weak for e
10−33 s Era of ination
10−38 s 1016 GeV Grand Uni ation (ele tromagnetism, strong and weak);
Grand Unied Theory, Supersymmetry, Superstrings?
10−43 s 1019 GeV Era of Quantum Gravity;
Friedmann equation annot be orre t
Between 2s and 100 s free neutrons an de ay to protons, n → p + e− + ν̄e . After

100 s neutrons are bound into Helium nu lei and are stable be ause the only energy
levels available to a proton arising from neutron de ay are blo ked, due to the Pauli
ex lusion prin iple, by the other two protons already present in the Helium nu leus. We
an estimate how many neutrons are left at 100s using the lifetime for neutron de ay in
free spa e, n → p + e− + ν̄ , whi h is τ = 900 s. We should allow for the fa t that the
neutron to proton ratio Nn /Np is not quite 1 at 2 s, due to the neutron-proton mass
2
dieren e ∆E := (mn − mp )c = 1.3 M eV . The Maxwell-Boltzmann distribution for
√
−E/kB T . Using T = 1/ t,
parti les of energy E in a gas at temperature T is n(E) ∼ e
with T measured in M eV and t in se onds, gives
Nn √ √
= e−∆E t ⇒ Nn = Np e−∆E t
.
Np
If neutrons drop out of equilibrium at a time te and subsequently de ay to protons until
62
they are bound into Helium nu lei at time tHe , then we expe t the neutron-proton ratio
at tHe to be
Nn (tHe −te ) √
= e− τ e−∆E te .
Np
Using the values ∆E = 1.3 M eV , τ = 900 s, te = 2 s and tHe = 100 s gives
Nn 1
≈ ,
Np 7
as observed.
If the rate of Helium produ tion in the early Universe were larger, more Helium would
be produ ed earlier and the neutrons would have less time to de ay, resulting in a larger
Nn /Np ratio and hen e a larger He/H ratio. If the rate of Helium produ tion were
smaller more of the Helium would be produ ed later and the neutrons would have more
time to de ay, resulting in a smaller Nn /Np ratio and hen e smaller He/H ratio. The
rate for Helium produ tion in reases if the density of neutrons and protons in reases
and the observed ratio of primordial Hydrogen to Helium in the Universe puts a limit
on the maximum allowed density of neutrons and protons onsistent with observations.
At the present day it annot be more than 15% of the Dark Matter. There are similar
onsiderations for other light elements that were produ ed in the Big Bang deuterium,
3 He and 7 Li.
63
Abundan es of primordial elements: the verti al axis is the abundan e of ea h isotope
relative to hydrogen. The red re tangles ree t the observational data their verti al extent
are the measured primordial abundan es and their horizontal extent is obtained by omparing
their verti al extent with the theoreti al predi tions as represented by the various green urves.
4
The top urve shows the mass fra tion of primordial He relative to Hydrogen: sin e 1990
the abundan es as measured by independent observations do not agree, probably indi ating
that un ertainties have been underestimated and the dotted bla k box may be a more a urate
4
ree tion of the observational un ertainties, giving a He abundan e lying between 0.23 and
4
0.25. Noti e that the theoreti al predi tion for He in reases as the baryon density (the density
of protons and neutrons) in reases, as des ribed above. The tightest onstraints ome from
observations of inhomogeneities in the osmi mi rowave ba kground from a mi rowave dete tor
alled the Wilkinson Mi rowave Anisotropy Probe (WMAP - the verti al yellow band) indi ating
(4.1 ± 0.1) × 10−28 kgm−3 , putting the density of protons plus

a value for the baryon density of
neutrons at about 15% of the total mass density (or 4% of the riti al density), implying the
7
existen e of another, unknown, type of matter. Dire t observations if Li are also somewhat
7
lower than the WMAP value, perhaps due to the re-pro essing if Li in stars not being fully
understood. (Taken from Physi s World, Vol. 28, No. 8, August 2007.)
64
At extremely high temperatures and energies the energy density in thermal radiation
is so large that every photon behaves like a bla k-hole. Suppose a thermal photon has
wavelength λ and frequen y ν , so λ = c/ν , and energy E = hν , with mass equivalent
M = E/c2 = hν/c2 . The wavelength is the same as the S hwarzs hild radius when
r
GM hc3 hc5 hc5
λ≈ 2 ⇒ E = hν = = ⇒ E= .
c GM GE G
It is onventional to use ~ rather than h (this is only an order of magnitude estimate)
and dene the Plan k Energy as
r
~c5
EP lanck = = 2 × 109 J = 1.2 × 1019 GeV
G
and the mass equivalent is the Plan k Mass,

r
2 ~c
MP lanck = EP lanck /c =
G
whi h is about 1019 mproton or 10−5 gm. Dividing the Plan k energy by Plan k's onstant
gives one over the Plan k Time,
r
~G
tP lanck = ≈ 10−43 s,
c5
and multiplying this by the speed of light gives the Plan k Length,
r
~G
lP lanck = ≈ 10−35 m.
c3
The Friedmann equation is unlikely to be valid when the energy rea hes the Plan k energy
over the Plan k volume (the Plan k length ubed). At these fantasti energy densities
quantum ee ts probably require some, as yet unknown, quantum theory of gravity.
In fa t the Plan k energy appears in the Friedman equation naturally, at mu h more
modest energies. Using the expli it expression for the Stefan-Boltzmann onstant,
π 2 kB
4
σS−B = ,
60~3 c2
the energy density in thermal photons is
4 π2
ǫRad = σS−B T 4 = (kB T )4 ,
c 15c3 ~3
equivalent to a mass density
π2
ρRad = (kB T )4 ,
15c5 ~3
65
At early times, when the universe was less than about 10,000 years old, ρRad dominates
the Friedmann equation,
8πG 8π 3 G 8π 3 (kB T )4
H2 = ρRad = (kB T )4
= .
3 45 c5 ~3 45~2 (EP lanck )2
Then the osmi s ale fa tor a(t) ∝ t1/2 , so H= 1 1

2t , and
2
4 45 ~EP lanck
(kB T ) = .
32π 3 t
This formula is valid for times as late as 10,000 years, the appearan e of the Plan k
energy here is not a signal of quantum gravity ee ts, it is merely due to the fa t that a
lassi al gravitational eld is being sour ed by quantum matter (thermal photons).
1 2
A tually, as written, this equation is only valid for temperatures for whi h kB T << me c , that is
2
times later than about 4s. For temperatures of order 2me c /kB and greater ele tron-positron pairs
8π 3
an be reated out of thermal energy and they ontribute to ρRad . This modies the prefa tor
45~ 2
but the general on lusion is un hanged.
Quantity Symbol Value
Speed of light (in va uum) c 299 792 458 m s−1 (exa t)

Newton's onstant G 6.673 × 10−11 kg−1 m3 s−2
Plan k's onstant h 6.626 × 10−34 J s
Ele tron harge (magnitude) e 1.602 × 10−19 C
1
Ele tri permittivity (va uum) ǫ0 = mu0 c2
8.854 × 10−12 C 2 N −1 m−2
Magneti permeability (va uum) µ0 4π × 10−7 N s2 C −2
e2
Fine stru ture onstant α = 2ǫ0 hc 7.297 × 10−3
Thompson ross-se tion σT 6.652 × 10−29 m2
Ele tron mass me 9.109 × 10−31 kg
Proton mass mp 1.673 × 10−27 kg
Neutron mass mn 1.675 × 10−27 kg
Atomi mass unit (mass of
12 C atom /12) a.m.u. 1.661 × 10−27 kg
Boltzmann's onstant kB 1.381 × 10−23 J K −1
Stefan-Boltzmann onstant σSB 5.670 × 10−8 J s−1 m−2 K −4
Avagadro's number NA 6.022 × 1023 mol−1
Earth mass M⊕ 5.97 × 1024 kg

Earth radius (equatorial) R⊕ 6.38 × 103 km
Lunar mass M
7.35 × 1022 kg
Lunar radius R
1.74 × 103 km
Earth-Moon distan e (mean) d⊕−
3.84 × 105 km
Earth-Sun distan e (mean) d⊕−⊙ 1.50 × 108 km
Solar mass M⊙ 1.99 × 1030 kg
Solar radius (equatorial) R⊙ 6.961 × 105 km
Solar luminosity L⊙ 3.85 × 1026 J s−1
Temperature of mi rowave ba kground T0 2.725 ± 0.002◦ K

Hubble onstant H0 67.3 ± 1.4 km s−1 M pc−1
(H0 = 100h km s−1 M pc−1 h 0.673 ± 0.014)
3H 2
Criti al density ρc = 8πG0 1.89 × 10−26 h2 kg m−3
Dark energy density (Cosmologi al onstant) ΩΛ 0.685 ± 0.017
Baryon density ΩB = ρB /ρcrit 0.0487 ± 0.0011
Dark matter density ΩM = ρM /ρcrit 0.266 ± 0.017
Total density Ωtot 1.000 ± 0.024
Age of the Universe t0 13.7 ± 0.2 × 109 yr
Ele tron Volt eV 1.602 × 10−19 J

year yr 3.156 × 107 s
light year lyr 9.461 × 1015 m
parse (1pc=3.26 lyr ) pc 3.086 × 1016 m

MP467 - Astrophysics and Cosmology PDF

Transféré par

Informations du document

Titre original

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

MP467 - Astrophysics and Cosmology PDF

Transféré par

Droits d'auteur :

Formats disponibles

MP467: Astrophysi

De ember 12, 2014

2 Stellar Formation and stru ture 11

3 Cosmology and Expansion of the Universe 45

1.2 Why is the moon round?

M > MP then we expe t M will

Mmoon = 7 × 1022 kg >> MP .

Chemi al Binding energy of the mountain is N |EB | Mountain is stable if

EGrav < N |EB |

e.g. for the earth with R⊕ = 6400km and α=2 we get

1.4 How old is the Earth?

is stable with 38 protons and 49 neutrons) with a half-life of τ = 5 × 10

87 Sr v. 87 Rb ontent for various meteorites.

et/τ − 1 = 0.0656 ⇒ t = τ ln(1.0656) = 4.6 × 109 yrs.

Nearest star: 4.3 lyr away

1 lyr = (2.998 × 108 ) × (3.156 × 107 )m = 9.462 × 1015 m.

A parse is dened using trigonometry with the Earth-Sun distan e, R = 1.50×1011 m,

R/d = tan φ ≈ φ ⇒ d ≈ R/φ,

3600′′ in 1◦ and 1, 296, 000′′ in 360◦ whi h is 2π radians. Hen e

1′′ = 4.85 × 10−6 rad.

E = mc2 ⇒ 4 × 1026 J/s ⇔ 4.4 × 109 kg/s = −Ṁ .

1. Chemi al Energy: e.g. H + H → H2 + energy (typi al hemi al energy ǫ ≈ 1eV )

3. Nu lear Energy: 2p + 2n → 4 He + 25M eV

2.2 Dynami al equilibrium

2.3 How do stars form?

2.4 How long does the ollapse take?

the free fall ollapse time (zero pressure).

2.5 Virial Theorem

Dene the average pressure

Essentially the kineti energy of ea h parti le is

2ĒT hermal + EGrav = 0 (2.6)

If a star ontra ts slightly, in quasi-equilibrium,

2.6 Stellar Equilibrium

dP (r) 4πGρ20 r 2πGρ20 r 2

2πGρ20 R2  π 1/3 4/3

Pc = 1.3 × 1014 N m−2 .

Better, more sophisti ated, models give

Pc = 2.5 × 1016 N m−2 and ρc = 1.62 × 105 kgm−3

2.7 Equation of State

e.g. in our simple model (ρ = ρ0 = const)

 1/3  1/3  4/3

ρ = (X + Y )ρ = mp np + mHe nHe = mp np + 4mp nHe

The number density of gas parti les is

PRadiation = (2.5 × 10−16 ) × (6.6 × 1028 ) = 1.7 × 1013 N m−2

where n := N/V is the number of parti les per unit volume.

We want to al ulate P (ρ, T ), but the al ulation is simplied when T is "small"

Summary: Three dierent equations of state, depending on the situation:

(d =2 H is a deuterium nu leus, an isotope of Hydrogen with one neutron and one

ρc 0.35 × (1.6 × 105 kgm−3 )

np 3.3 × 1031 m−3

interiors, not reated.

the se ond an underestimate. A more a urate value is Rc ≈ 0.3R⊙ .

1.9 × 1038 s−1 1.9 × 1038

2.9 Energy transport

L⊙ = 3.8 × 1026 J/s from the surfa e.

40000 20000 10000 5000 2500

L(r + dr) − L(r) = 4πr 2 drw(r)

2.9.2 Conve tion

1. the pressure equalizes immediately

L(r) γ − 1 T 4 G 16πac γ − 1 PRadiation 16πGc L

2.10 Mass-Luminosity Relation

using Tc = 1.6 × 107 K , whereas the observed solar Luminosity is

L⊙ = 3.8 × 1026 J/s (2.30)

A parse is dened using trigonometry with the Earth-Sun distan e, R = 1.50×1011 m,

Dene the average pressure

2πGρ20 R2 π 1/3 4/3

1/3 1/3 4/3

We want to al ulate P (ρ, T ), but the al ulation is simplied when T is "small"

Summary: Three dierent equations of state, depending on the situation:

1/3 4/3 1/3

K<0 R in reases indenitely and Ṙ > 0 always;

K=0 R in reases indenitely but Ṙ → 0 as t → ∞.

• K=0 is at Eu lidean spa e;

with (t2 − t1 )H << 1. Let sAB = c(t2 − t1 ) (time-of-ight distan e) then

3.6 The Horizon Problem and Ination

3.6.1 Inationary Universe

As a by-produ t of ination we get a natural explanation of why K ≈ 0. Non-zero

3.7 The rst 3 minutes (Weinberg)