Lecture 4

Hydrostatics and
Time Scales
Glatzmaier and Krumholz 3 and 4
Prialnik 2
Pols 2
 Assumptions – most of the time
• Spherical symmetry
Broken by e.g., convection, rotation,
magnetic fields, explosion, instabilities
Makes equations a lot easier. Also facilitates
the use of Lagrangian (mass shell) coordinates

• Homogeneous composition at birth

• Isolation frequently assumed

• Hydrostatic equilibrium
When not forming or exploding
Uniqueness

One of the basic tenets of stellar evolution is the

Russell-Vogt Theorem, which states that the mass
and chemical composition of a star, and in particular
how the chemical composition varies within the star,
uniquely determine its radius, luminosity, and internal
structure, as well as its subsequent evolution.

A consequence of the theorem is that it is possible

to uniquely describe all of the parameters for a star
simply from its location in the Hertzsprung-Russell
Diagram. There is no proof for the theorem, and in fact,
it fails in some instances. For example if the star has
rotation or if small changes in initial conditions cause
large variations in outcome (chaos).
 Hydrostatic Equilibrium
Consider the forces acting upon a spherical mass shell
dm = 4π r 2 dr ρ
The shell is attracted to the center of the star by a force
per unit area
−Gm(r )dm −Gm(r ) ρ
Fgrav = = dr
(
4π r r
2 2
) r 2

where m(r) is the mass interior to the radius r

M dm
It is supported by the pressure
gradient. The pressure
P+dP
on its bottom is smaller
P
than on its top. dP is negative

FP = P(r ) i area − P(r + dr ) i area m(r)

or
FP dP
= dr
area dr
 Hydrostatic Equilibrium
If these forces are unbalanced there will be an acceleration
per unit area
F Fgrav −FP dm 4 π r 2 ρdr
= = r = r
Area Area Area 4π r 2

⎛ Gm(r ) ρ dP ⎞
( )
ρ dr r =− ⎜
⎝ r 2
+ ⎟
dr ⎠
dr

−Gm(r ) 1 dP
r = −
r 2
ρ dr
r will be non-zero in the case of stellar explosions
or dynamical collpase, but in general it is very small
in stars compared with the right hand side, so
dP Gm(r ) ρ
=−
dr r2
where m(r ) is the mass interior to the radius r. This is
called the equation of hydrostatic equilibrium. It is the
most basic of the stellar structure equations.
Examples of hydrostatic equilibrium:
1)Consider, for example, an isothermal atmosphere composed
of ideal gas, P = nkT (T= constant; n is the number density).
Let the atmosphere rest on the surface of a spherical mass M
with radius R. R and M are both constant. h is the height
above the surface:

dP GM ρ
=− 2 r =R+h h << r
dr r
dr = dh
NAkT d ρ GM ρ ρNA
=− = − gρ P = nkT n= T const.
µ dh R 2
µ
ρ h
dρ µg µgh
∫ρ ρ = −
NAkT ∫0
dh ⇒ ln ρ − ln ρ0 = −
NAkT
0

⎛ ρ⎞ h NAkT
ln ⎜ ⎟ = − H= = density scale height
⎝ ρ0 ⎠ H µg
−h −
h
kT
ρ = ρ0 e H equivalently n = n0e H
H=
g
Examples:

2) Or water in the ocean (density constant; incompressible)

dP GM ρ
=− 2 r = R − h h << r
dr r
dr = − dh h = depth below surface
dP GM ρ
− =− 2
= − gρ
dh R

P = g ρh e.g., at one km depth

= (980)(1)(105 ) = 9.8 × 107 dyne cm−2
= 98bars
Examples:
3) Stellar photospheres
As we will disuss next week, a beam of light going through
a medium with opacity, κ (cm2 g−1), suffers attenuation
dI
= − κρI ⇒ I = I0 e −κρr ≡ I0e −τ
dr
where τ is the "optical depth". The photosphere is defined
by place where, integrating inwards, τ ≈1 (a value of 2/3 is
also sometimes used).
Consider hydrostatic equilibrium in a stellar atmosphere
where M and R can assumed to be constant
dP GM
= − 2 ρ = − gρ
dr R
Multiplying by κ and dividing by ρ
κ dP dP dP g
= κ = −g or =−
κρ dr dτ dτ κ
This was first noted by K. Schwartzschild in 1906
Solar photosphere
Integrating inwards and assuming
g and κ are constant (the latter is
not such a good approximation)
g
Pphotosphere ≈
κ
GM
For the sun g = = 2.7 ×10 4

R2
So if κ  1, Pphotosphere ≈ 0.027 bars.

κ is due to the H− ion

and varies rapidly with temperature
but is of order unity (between 0.1
and 1 in the region of interest).
Actually P ~ 105 dyne cm−2 = 0.1 bar
As we learned last time, the solar photosphere is
largely neutral. H and He are not ionized much at all
(~10-4 for H). The electrons come mostly from elements
like Na, Ca, K, etc. that are very rare. Consequently the
ion pressure overwhelmingly dominates at the solar
photosphere (radiation pressure is negligible). This is
not the case if one goes deeper in the star.

The electron pressure at the photosphere, though small,

gives the electron density which enters into the Saha
equation and determines the spectrum. This is in fact
how it is determined.
 Table 2-1: The Holweger-Müller Model Atmosphere7
Optical Temper- Electron
Height8 Depth ature Pressure Pressure Density Opacity
(km) (τ5000) (°K) (dynes cm-2) (dynes cm-2) (g cm-3) (κ5000)
550 5.0×10-5 4306 5.20×102 5.14×10-2 1.90×10-9 0.0033
507 1.0×10-4 4368 8.54×102 8.31×10-2 3.07×10-9 0.0048
441 3.2×10-4 4475 1.75×103 1.68×10-1 6.13×10-9 0.0084
404 6.3×10-4 4530 2.61×103 2.48×10-1 9.04×10-9 0.012
366 0.0013 4592 3.86×103 3.64×10-1 1.32×10-8 0.016
304 0.0040 4682 7.35×103 6.76×10-1 2.47×10-8 0.027
254 0.010 4782 1.23×104 1.12 4.03×10-8 0.040
Holweger and Muller
Solar Physics, 1974 202 0.025 4917 2.04×104 1.92 6.52×10-8 0.061
176 0.040 5005 2.63×104 2.54 8.26×10-8 0.075
149 0.063 5113 3.39×104 3.42 1.04×10-7 0.092
121 0.10 5236 4.37×104 4.68 1.31×10-7 0.11
94 0.16 5357 5.61×104 6.43 1.64×10-7 0.14
66 0.25 5527 7.16×104 9.38 2.03×10-7 0.19
29 0.50 5963 9.88×104 22.7 2.60×10-7 0.34
0 1.0 6533 1.25×105 73.3 3.00×10-7 0.80
-34 3.2 7672 1.59×e+005 551 3.24×10-7 3.7
-75 16 8700 2.00×e+005 2.37×103 3.57×10-7 12
 Lagrangian coordinates
The hydrostatic equilibrium equation
dP Gm(r ) ρ
=−
dr r2
can also be expressed with the mass as the
independent variable by the substitution dm = 4π r 2 ρ dr
dP dr Gm(r ) ρ 1
=−
dr dm r2 4π r 2 ρ

dP Gm(r )
=−
dm 4π r 4

This is the "Lagrangian" form of the equation. We will find

that all of our stellar structure equations can have either
an Eulerian form or a Lagrangian form.
Merits of using Lagrangian coordinates

• Material interfaces are preserved if part or

all of the star expands or contracts. Avoids
“advection”.

• Avoids artificial mixing of of composition

and transport of energy

• In a stellar code can place the zones “where

the action is”, e.g., at high density in the
center of the star

• Handles large expansion and contraction (e.g.,

to a red giant without regridding.
• In fact the merit of Lagrangian coordinates
is so great for spherically symmetric problems
that all 1D stellar evolution codes are writen
in Lagrangian coordinates

• On the other hand almost all multi-D stellar codes

are written in “Eulerian” coordinates.
This Lagrangian form of the hydrostatic equilibrium
equation can be integrated to obtain the central pressure:
Psurf M
Gm(r ) dm
∫
Pcent
dP = Psurf − Pcent = − ∫
0 4π r 4

and since Psurf = 0

M
Gm(r ) dm
Pcent = ∫
0 4π r 4

To go further one would need a description of how

m(r) actually varies with r. Soon we will attempt that .
Some interesting limits can be obtained already though.
For example, r is always less than R, the radius of
1 1
the star, so 4 > 4 and
r R
M
G GM 2
4 ∫
Pcent > m(r ) dm =
4π R 0 8π R 4
A better but still very approximate result comes from
assuming constant density (remember how bad this is
for the sun!) m(r)= 4/3 π r 3 ρ  4/3 π r 3 ρ0 (off by 2 decades!)
4/3 4/3
⎛ 3m(r ) ⎞ m(r ) ⎛ 4πρ0 ⎞
4
r =⎜ ⎟ ⇒ 4 = ⎜ m(r )
⎝ 4πρ 0 ⎠
r ⎝ 3m(r ) ⎟⎠
4/3 M 4/3
−G ⎛ 4πρ0 ⎞ dm G ⎛ 4πρ0 ⎞ 3M 2/3
Pcent ≈
4π ⎜⎝ 3 ⎟⎠ ∫0 m1/3 (r ) = 4π ⎜⎝ 3 ⎟⎠ 2
4/3 4/3
⎛ 4πρ0 ⎞ ⎛M⎞
but ⎜ =⎜ 3⎟ so
⎝ 3 ⎟⎠ ⎝R ⎠
3GM 2 GM ρ0 3M
Pcent = or since ρ0 is assumed =
8π R 4
2R 4π R 3
This is 3 times bigger but still too small
The sun's average density is 1.4 g cm−3 but its
central density is about 160 g cm−3 .

Interestingly, most main sequence stars are supported primarily by

ρNAkT
ideal gas pressure (TBD), P= ,so for the constant density
µ
(constant composition, ideal gas) case
ρ0NAkT GM ρ0
=
µ 2R
so the density cancels and one has
GM µ M
Tcent = ∝
2NAkR R
For a typical value of µ =0.59
⎛ M / M ⎞
Tcent = 6.8 ×10 6
⎜ R/R ⎟ K
⎝  ⎠

This of course is still a lower bound. The actual present value

for the sun is twice as large, but the calculation is really for
a homogeneous (zero age) sun. A similar scaling (M/R)
will be obtained for the average temperature using the
Virial Theorem
 Hydrodynamical Time Scale
We took r=0 to get hydrostatic equilibrium. It is also
interesting to consider other limiting cases where r
is finite and the pressure gradient or gravity is negligible.
The former case would correspond to gravitational
collapse, e.g., a cloud collapsing to form a star or galaxy.
The latter might characterize an explosion.

−Gm(r ) 1 dP
r = −
r 2
ρ dr
dP
If → 0, then r= - Gm(r ) / r 2 = g(r )
dr
which is the equation for free fall.
 Free fall time scale
If g(r) is (unrealistically) taken to be a constant = g(R),
the collapse time scale is given by
1 2
gτ ff ≈ R and for the whole star,
2
GMτ ff2 4πρR 3
2
≈R but M = (exact)
2R 3
4π G ρRτ ff2
≈R
6
3
τ ff2 =
2π G ρ
So
1 2680 s
τ ff ≈ =
2π ρ
Gρ
3
Clearly this is an overestimate since g(r) actually increases
during the collapse.
Sometimes in the literature one sees instead
1/2
R R ⎛ 3R 3
⎞
τ ff = = =⎜
⎝ 8π GR ρ ⎟⎠
3
v esc 2GM / R
1/2
⎛ 3 ⎞
=⎜ = 1340 / ρ sec
⎝ 8π G ρ ⎟⎠
And for the density, which changes logrithmically
3 times as fast as the density
1/2
1⎛ 3 ⎞
τ ff = ⎜ = 446 / ρ sec
3 ⎝ 8π G ρ ⎟⎠

In any case all go as ρ −1/2 and are about 1000 s for ρ =1

 Explosion Time Scale

A related time scale is the explosive time scale.

Say g suddenly went to zero. An approximate expansion
time scale for the resulting expansion would be

⎛ dP ⎞ 1 ⎛ dP ⎞
r =4π r 2 ⎜ = ⎜ 2
R ~1/2 r τ exp
⎝ dm ⎠ ρ ⎝ dr ⎟⎠
⎟
2R
2R P r~ 2
~ τ exp
τ2 ρR
exp
1/2
⎛ ρ⎞
τ exp ~R ⎜ ⎟ ≈ R/csound
⎝P⎠

Usually the two terms in the "hydrostatic equilibrium" equation

for r are comparable, even in an explosion andτ ff ≈ τ exp We shall
just use 2680 s / ρ for both.
 Examples

a)The sun ρ = 1.4 g cm−3

τ HD = 2680. / 1.4 =1260 s = 38 minutes
If for some reason hydrostatic equilibrium were lost
in the sun, this would be the time needed to restore it. If the
pressure of the sun were abruptly increased by a substantial
factor (~2), this would be the time for the sun to explode

b) A white dwarf ρ ~10 6 −108 g cm−3

τ HD = 2680. / 10 6 = 0.26 − 2.6 s
~ Time scale for white dwarfs to vibrate (can't be pulsars)
~Time for the iron core of a massive star to
collapse to a neutron star
 Examples

c) A neutron star ρ ~1015 g cm−3

τ HD = 2680. / 1015 = 0.084 ms
~ Time for a neutron star to readjust its structureafter
core bounce

d) A red giant - solar mass, 1013 cm ⇒ ρ ~5 × 10 −7 g cm−3

τ HD = 2680. / 5 × 10 −7 = 4 ×105 s ≈ 5 days

~Time for the shock to cross a supergiant star making

a Type IIp supernova
~Typical Cepheid time scale
Generally the hydrodynamical, aka free fall,
aka explosion time scale is the shortest of all
the relevant time scales and stars, except in
their earliest stages of formation and last explosive
stages, are in tight hydrostatic equilibrium.
 Some definitions
dm
The total gravitational binding energy of a star of
m
mass M, is

M
Gm GM 2
Ω=−∫ dm = − α α ~ 1 (see next page)
0
r R

Similarly, the total internal energy is the integral

over the star's mass of its internal energy per gram, u.
3 NAkT 3 P
For an ideal gas u = = (TBD)
2 µ 2ρ

U = ∫ u dm
0
M
3 P
= ∫ dm for an ideal gas
20 ρ
 Gravitational binding energy for a
sphere of constant
4π r 3 ρ0
If ρ =constant = ρ0 , m(r )= dm = 4π r 2 ρ0 dr
3

M
Gm(r )
R
4π Gr 2 ρ0
Ω=− ∫ dm = − ∫ 4π r 2 ρ0 dr
0
r 0
3
16π 2G ρ02 R 4 16π 2G ρ02R 5
=−
3 ∫
0
r dr = −
15
2
3G ⎛ 4π R ρ0 ⎞
3
3GM 2
=− ⎜ ⎟ =−
5R ⎝ 3 ⎠ 5R

i.e., α =3/5 for the case of constant density. In the

general case it will be larger.
Similarly
The total kinetic energy is

M
r2 (m) 1
T=∫ dm  Mv 2 if the velocity is the same everywhere
0
2 2
The total nuclear power is

Lnuc = ∫ ε (m) dm where ε is the energy generation rate

0

in erg g−1 s−1 from the nuclear reactions

ε is a function of T, ρ, and composition
and the luminosity of the star

L = ∫ F(m) dm where F(m) is the energy flux entering

0

each spherical shell whose inner

boundary is at m
 The Stellar Virial Theorem
The Lagrangian version of the hydrostatic equilibrium equation is
dP Gm(r )
=−
dm 4π r 4
4
Multiply each side by the volume inside radius r, V= π r 3 ,
3
and integrate
⎛4 ⎞ ⎛ Gm(r ) ⎞ 1 Gm(r )dm
V dP(r ) = ⎜ π r 3 ⎟ ⎜ − dm = −
⎝3 ⎠ ⎝ 4π r 4 ⎟⎠ 3 r
P (r ) m(r )
1 Gm'dm'
∫
Pcent
V dP = −
3 ∫0
r
The integral on the right hand side is just the total gravitational
potential energy in the star interior to r, Ω(r ).
The left hand side can be integrated by parts out to radius r < R
P (r ) r V (r ) V (r )
since V = 0
∫
Pcent
V dP = PV ⎤⎦ − ∫
0
P dV = P(r )V(r ) − ∫
0
P dV
at the center
cent
 P (r ) r V (r ) V (r )

∫ V dP = PV ⎤⎦
Pcent
− ∫
0
P dV = P(r )V(r ) − ∫
0
P dV
cent

4π r 2
ρ dr dm
Substituting dV = 4π r 2dr = =
ρ ρ
V (r ) m(r )
P
∫
0
P dV = ∫
0
ρ
dm so

m(r )
P 1
P(r )V(r ) − ∫
0
ρ
dm = Ω(r )
3
This is true at any value of r, but pick r = R where
P(R)= 0, then
M
P 1
- ∫ dm = Ωtot (Prialnik 2.23)
0
ρ 3 for any EOS
P 2
but for an ideal gas = u
ρ 3
 The Virial Theorem for ideal gases
M2
u dm =− 1Ω
So for ideal gas ∫3 3
0

The left hand side is 2/3 of the total internal energy of

the star, hence
2U = - Ω (Prialnik 2.26) (Ω is defined < 0)
The internal energy is, in magnitude, 1/2 the binding energy
of the star. We shall see later that similar though somewhat
different expressions exist for radiation and relativistic degeneracy
pressure. This is the Virial Theorem.
Note that we can also define a mass averaged temperature
in the star
M
1
T = ∫ T(m)dm
M0
 The Virial temperature
NAk ρT
For an ideal gas, P= (same as nkT but in terms of ρ )
µ
M M M
3 P 3 NAkT
U = ∫ dm = ∫ dm but ∫ T dm =TM
20 ρ 20 µ 0

3 NAkMT 1
= =− Ω
2 µ 2
3MNAkT α GM 2
=
µ R
where e.g., α =3/5 for a sphere of constant density
αµG M
T= Virial temperature
3NAk R
note again the inverse dependence of T on R
This is similar to the value obtained from hydrostatic equilibrium
α
but (about 0.2) instead of 0.5, so cooler, about
3
2.6 x 10 6 for the sun.

Note that the temperature of nearly homogeneous

stars like the sun is set by their bulk properties, M and R.
As the mass on the main sequence rises, R empirically
rises only as about M2/3 so the central temperature of
massive stars rises gently with their mass. We shall
see later that the density actually decreases.
 Conservation of Energy
(aka First law of thermodynamics)
Consider the changes of energy that can be experienced
by a small (spherical) mass element m = 4  r2  r.
r << R; m << M. If the zone is sufficiently small, we
can also think of m as dm and it is customary to
do so In stellar evolution codes. The zone’s internal energy,
u, can change as a consequence of:

• energy flowing in or out of its upper or lower

boundary by radiation, conduction, or convection

• compression or expansion

• nuclear reactions generating or absorbing energy

The change in internal energy in a thin shell of mass,
δ m, during a small change of time, δ t, is then dm and δ m used
interchangably here
δ u dm = δ Q + δ W
P is the pressure
in the zone. ε is the
where δ Q = ε δ m δ t + F(m)δ t − F(m + δ m) δ t is energy generation
the internal energy generation plus the net in the zone.
accumulation or loss of energy from fluxes at its
F is the flux of energy at
upper and lower boundaries and m or m+dm
⎛ 1⎞
δ W = − P δ V = − Pδ ⎜ ⎟ δ m
⎝ ρ⎠
is the energy lost to work because the zone expands and pushes
on its boundaries or is compressed and gains energy. Note that
δ V =4π r 2 δ r and δ m = constant = 4π r 2 ρδ r so
⎛ 1⎞
δ V =δ m δ ⎜ ⎟
⎝ ρ⎠
δ Q = ε δ m δ t + F(m)δ t − F(m + δ m) δ t F is the flux of energy at
and since m or m+δ m

⎛ dF ⎞
F(m + δ m)= F(m) + ⎜ ⎟ δm
⎝ dm ⎠
⎛ dF ⎞
δQ = ⎜ ε − ⎟ δm δt
⎝ dm ⎠
⎛ 1⎞ ⎛ dF ⎞
so δ u δ m + Pδ ⎜ ⎟ δ m = ⎜ ε − ⎟ δm δt
⎝ ρ⎠ ⎝ dm ⎠
dividing by δ m and δ t and taking the limit as δ t → 0

du d ⎛ 1⎞ dF The energy conservation

+P ⎜ ⎟ = ε − equation aka “the first law
dt dt ⎝ ρ ⎠ dm
of thermodynamics
 du d ⎛ 1⎞ ∂F
+P ⎜ ⎟ = ε −
dt dt ⎝ ρ ⎠ ∂m
du
is the rate at which the internal energy in erg g−1 is
dt
changing, e.g., in a given zone δ m of a (Lagrangian)
stellar model

d ⎛ 1⎞
P is the PdV work being done on or by the zone
dt ⎜⎝ ρ ⎟⎠
as it contracts (ρ ↑⇒PdV is negative) or expands
(ρ ↓⇒PdV is positive). Units are erg g−1 s−1

ε is the nuclear energy generation rate minus neutrino

losses in erg g−1 s−1
dF
dm is the difference in energy (per second) going out the
dm
top of the zone (by diffusion, conduction, or convection)
minus the energy coming in at the bottom. It is positive
if more energy is leaving than entering.
 Thermal Equilibrium
If a steady state is reached where in a given zone the
internal energy (u), density (ρ ), and pressure (P) are not changing
very much, then the left hand side of the 1st law becomes
approximately zero and
dF
ε=
dm
Energy flows into or out of the zone to accomodate what
is released or absorbed by nuclear reactions (plus neutrinos).

If this condition exists through the entire star

M M

∫ ε dm = L
0
nuc
= ∫ dF = L
0

then the star is said to be in thermal equilibrium. It is also

possible for thermal equilibrium to exist in a subset of the star.
Examples:

• Main sequence stars are in thermal equilibrium

• Massive stars becoming red giants are not. More

energy is being generated than is leaving the
surface. The star’s envelope is expanding

• White dwarfs are in a funny thermal equilibrium

where nuc = o and PdV is zero, but u is decreasing
in order to provide L ⇒ U = L

• Late stages of massive stellar evolution may also

approach a funny equilibrium where neutrino
losses balance nuclear energy generation, at least
in those portions of the star where burning is going
on at a rapid rate
 Integrating the first law of thermodynamics
du d ⎛ 1⎞ dF
+P ⎜ ⎟ = ε −
dt dt ⎝ ρ ⎠ dm
Integrate over mass:
d ⎛ 1⎞
M M M
du
∫0 dt dm +∫0 P dt ⎜⎝ ρ ⎟⎠ dm = ∫0 ε dm − F(M) + F(0)
= Lnuc − L
Since m does not depend on t the leftmost term can be
rewritten
M M
du d
∫0 dt dm = ∫ u dm = U
dt 0
The second term takes some work, one can rewrite 1/ρ
(since dm = dV / ρ )
d ⎛ 1 ⎞ d ⎛ dV ⎞ d ⎛ dV ⎞
= ⎜ =
dt ⎝ ρ ⎠ dt ⎝ dm ⎠ dm ⎜⎝ dt ⎟⎠
⎜ ⎟ ⎟
 Continuing:

d ⎛ 1 ⎞ d ⎛ dV ⎞ d ⎛ dV ⎞
= ⎜ =
dt ⎝ ρ ⎠ dt ⎝ dm ⎠ dm ⎜⎝ dt ⎟⎠
⎜ ⎟ ⎟

d ⎛ 2 dr ⎞
= 4π r
dm ⎜⎝ dt ⎟⎠
Integrate by parts:
d ⎛ 1⎞ ⎛ 2 dr ⎞
M M
(2nd
∫ P ⎜ ⎟ dm = ∫ Pd ⎜ 4π r
term) 0 dt ⎝ ρ ⎠ 0 ⎝ dt ⎟⎠
M M
dr ⎤ 2 dr dP
= 4π r P ⎥ − ∫ 4π r
2
dm
dt ⎦0 0
dt dm
The first term is zero at r = 0 and M (where P = 0) so
d ⎛ 1⎞
M M
2 dr dP
∫0 dt ⎜⎝ ρ ⎟⎠
P dm = − ∫0 4π r
dt dm
dm
Continuing:

d ⎛ 1⎞
M M
2 dr dP
∫0 P dt ⎜⎝ ρ ⎟⎠ dm = − ∫0 4π r dt dmdm
dP −Gm r
But = −
dm 4π r 4
4π r 2
d ⎛ 1⎞
M M M
Gm dr dr
∫0 P dt ⎜⎝ ρ ⎟⎠ dm = ∫0 r 2 dt dm + ∫0 r dt dm


dr d ⎛ r2 ⎞ Gm dr d ⎛ Gm ⎞
and r = ⎜ ⎟ = − ⎜
dt dt ⎝ 2 ⎠ 2
r dt dt ⎝ r ⎟⎠
So
d ⎛ 1⎞
M M M
d Gm 1d  + T
∫0 dt ⎜⎝ ρ ⎟⎠ = −
dt ∫0 r
+
2 dt ∫0
 = Ω
2
P dm dm r dm
So all together
 Ω
U+  + T = L − L
nuc

which expresses succinctly (and obviously) the conservation

of energy for the star. Power generated or lost on the right
balances the change in internal, gravitational, and kinetic
energies on the left.

Suppose the star is static so that T=0, and the Virial
 = -1/2 Ω
Theorem also applies so U  (for an ideal gas)
Then
1 
Ω = Lnuc − L
2

This is interesting.
 1 
Ω = Lnuc − L remember Ω is negative
2

• If Lnuc = L then the star is in a state of balanced

power. Over long period of time it neither expands
or contracts. It is in thermal and dynamic equilibrium

• If Lnuc > L, more energy is being generated by

nuclear reactions than is being radiated.  &
becomes less negative. The star expands to
absorb the excess

• If Lnuc < L, the star is radiating more energy than

nuclear reactions are producing. If possible, the
star makes up the deficiency by contracting to
a more tightly bound state.  becomes more negative
  in the same
Also interesting is the solution for U
situation. Then

 = -1 Ω
U  = L−L
nuc
2

In the absence of nuclear energy input, the star increases

its internal energy (heat) as it radiates. The more a star,
supported by ideal gas pressure, radiates, the hotter it
gets. The heat capacity is defined by C(T) = Δq/ΔT where
Δq is the energy flowing into or out of the matter. Here ΔT
is positive when Δq is negative. We say that stars have a
"negative heat capacity". The origin of the energy is gravity.
 The Kelvin Helmholtz time scale

The time scale for adjusting to an imbalance in

energy generation (historically Lnuc = 0) is the Kelvin
Helmholtz timescale
Ω Ω
τ KH = ≈

Ω 2L

M R
Gm Gm(r )
Ω=− ∫ dm = − ∫ 4π r 2 ρ(r ) dr
0
r 0
r
α GM 2
=− where α depends on ρ(r )
R
I
 Kelvin Helmholtz time scale for the sun

For the sun the constant density expression gives

3GM2
τ KH ()= = 2.95 ×1014 s = 9.3 My
10RL
but the actual sun is far from constant density and the
actual value is closer to 30 My. We will obtain a better
value when we study polytropes later.
 This is (roughly) the time the sun could shine without nuclear
reactions at its present radius and luminosity.
In fact, if the radius is allowed to change gravity could
power the sun or any star much longer. Even keeping the
luminosity the same the sun could, in principle, continue
to contract until it became a white dwarf. Taking a
white dwarf radius of ~5000 km, τ KH could be increased
to 1.3 Gy, but then the sun wouldnot look like it does
(with the same luminosity it would radiate in the ultraviolet).

Taking this to the absurd limit (not allowed by quantum

mechanics for a 1 M star), Ω~Mc 2 , for a black hole
τ KH ~1013 years. Gravity, natures weakest force, can, in
the end provide more power than any other source
including nuclear reactions.
One could in principle explain the modern sun
without recourse to nuclear reactions by allowing
it to become extremely centrally condensed (e.g., a
black hole at the center). But this would be quite
inconsistent with

• Stellar physics and known equations of state

• The solar neutrino experiments

Other places the Kelvin Helmholtz time scale enters:

• Star formation – it is the time required for a

protostar to settle down on the main sequence
and ignite nuclear burning

• The time scale between major nuclear burning

stages where degeneracy does not enter in,
e.g., between helium depletion and carbon
ignition in a massive star

• The time a proto-neutron star requires to release

its binding energy as neutrinos. During this
time it can power a supernova
As the star contracts, so long as it remains approximately
an ideal gas, the Virial Theorem (or hydrostatic equilibrium)
implies:
Ω 3 GM 2 GM ρ
U=− ⇒ NAkTM = α ; or Pc ∝
2 2µ 2R R
M M
T∝ but ρ ∝ 3
R R
So T ∝ M2/3 ρ 1/3
As the the star contracts, the temperature rises as the cube
root of the density and, for a given density, is less in stars of
smaller mass.
We will make these argumentsmore quantitative later
Evolution of the central
density and temperature of
15 M and 25 M stars

Kelvin Helmholtz
nuclear
Kelvin Helmholtz

nuclear

Kelvin Helmholtz

nuclear
 The nuclear time scale

As we shall see later nuclear fusion – up to the

element iron is capable of releasing large energies
per gram of fuel (though well short of mc2).

Roughly the energy release during each phase is

the fraction of the star that burns times

Fuel fraction mc2

Hydrogen 7 x 10-3
Helium 7 x 10-4
Carbon 1 x 10-4
Oxygen 3 x 10-4
Silicon 1 x 10-4
 The nuclear time scale
One can also define a nuclear time scale
Mqnuc
τ nuc =
L
but everywhere except the main sequence, one must
be quite cautious as to what to use for L and M because
more than one fuel may be burning at a give time and
neutrinos can carry away appreciable energy. Also
only a fraction of the star burns, just that part that is hot
enough. For the sun if 10% burns
( 0.1) ( 0.007c ) (M )
2

τ nuc ≈
L
= 10Gy
which is fortuitously quite close to correct.
There are several things to notice here.

First the energy yield is a small fraction of the rest

mass energy. It just comes from shuffling around
the same neutrons and protons inside different
nuclei. The number of neutrons plus protons is
being conserved. No mass is being annihilated.

Less energy is produced once helium has been

made. The binding energies of neutrons and
protons in helium, carbon, oxygen and silicon
are similar, all about 8 MeV/nucleon.
More energy is released by oxygen
burning simply because the oxygen abundance
is about 5 times the carbon abundance.

We shall see later that the nuclear time scale can

be greatly accelerated by neutrino losses
Radiation diffusion
time scale

In a random walk, how

far are you from the
origin in n steps, and
how long did it take
to get there?
Random walk:

=3 cm2
x3 = 3 cm
 The thermal time scale

The average of s itself is zero

 
s is =
 ⎛ (6.9 × 1010 )2 ⎞
R2
τ dif ( )
 ≈ =⎜ 10 ⎟
c ⎝ (0.1)(3 × 10 ⎠
=1.6 ×1012
sec

or about 50,000 years.

A more accurate value is 170,000 years

http://adsabs.harvard.edu/full/1992ApJ...401..759M
 Ordering of Time scales

To summarize we have discussed 4 time scales

that characterize stellar evolution.

• Hydrodynamical time – 446/1/2 to 2680/1/2 sec

– the time to adjust to and maintain hydrostatic
equilibrium. Also the free fall or explosion time scale

• Thermal – R2/lc – the time to establish thermal

equilibrium if diffusion dominates (it may not)

• Kelvin-Helmholtz – GM2/(2RL) – the time to adjust

structure when the luminosity changes.

• Nuclear – qnucM /L – the time required to fuse a

given fuel to the next heavier one
In general, and especially on the main sequence
τ HD << τ therm < τ KH < τ nuc

There are however interesting places where this ordering

breaks down:
τ therm ~ τ KH ~ τ nuc during the late stages of massive
star evolution
τ HD ~ τ nuc for SN Ia; for explosive
nucleosynthesis in SN II
There are also other relevant time scales for e.g.,
rotational mixing, angular momentum transport,
convective mixing, etc. Usually a phenomenon can
be characterized and its importance judged by examining
the relevant time scales.