Astronomy 112: The Physics of Stars

Class 16 Notes: Post-Main Sequence Evolution of Low Mass Stars

Our topic for today is the first of three classes about post-main sequence evolution. Today
the topic is low mass stars.

I. Leaving the Main Sequence

A. Main Sequence Lifetime

Stars remain on the main sequence as long as their hydrogen fuel lasts. While on
the main sequence, their properties do not change much, but they do change some
due to the gradual conversion of H into He. As a result of this conversion, the
hydrogen mass fraction X decreases, while the helium mass fraction Y increases.
As a result, the mean atomic weight changes. Recall that
1 1 1−X −Y
≈ X+ Y +
µI 4 hAi
1 1
≈ (1 + X)
µe 2
1 1 1
= + .
µ µI µe

Insterstellar gas out of which stars form has roughly X = 0.74 and Y = 0.24 (in
contrast to X = 0.707 and Y = 0.274 in the Sun, which has processed some of
its H into He), which gives µ = 0.61. In contrast, once all the H has been turned
into He, X = 0 and Y = 0.98, which gives µ = 1.34.
In stars like the Sun that are radiative in their cores, the changes occur shell by
shell, so different shells have different compositions depending on their rate of
burning. In stars that are convective in their cores, convection homogenizes the
composition of the different shells, so the entire convective zone has a uniform
composition.
As we mentioned briefly a couple classes ago, this difference between convective
and non-convective stars affects how long it takes stars to leave the main sequence.
Stars with convection in their cores do not leave the main sequence until they have
converted all the mass in the convective region to He. As the convective zone fills
more and more of the star, the main sequence lifetime therefore approaches the
naively computed nuclear timescale tnuc = M c2 /L.
In contrast, stars with radiative cores, like the Sun, leave the main sequence
once the material in the very center where nuclear burning occurs is converted to
He. This makes their lifetimes shorter than tnuc , with the minimum of tms /tnuc

1
 occurring near 1 M , since that is where stellar cores are least convective. This
general expectation agrees quite well with numerical results.
[Slide 1 – main sequence and nuclear lifetimes]
A general complication to this story is mass loss, which, for massive stars, can be
significant even while they are on the main sequence. The mass loss mechanism is
only generally understood. All stars have winds of gas leaving their surfaces, and
these winds become more intense for more massive stars. These numerical results
include a very approximate treatment of mass loss, but on the main sequence it
is only significant for stars bigger than several tens of M .
B. Luminosity Evolution
Regardless of convection, the increase in µ results in an increase in luminosity.
One can estimate this effect roughly using an Eddington model. The Eddington
quartic is !2
M L
0.003 µ4 β 4 = 1 − β =
M LEdd
and so the luminosity of a star in the Eddington model is
!3
4πcGM 4 4 M
L = 0.003 µβ .
κs M
Thus the luminosity at fixed mass is proportional to (µβ)4 .
For a low mass star, the first term in the Eddington quartic is negligible, so β ≈ 1
independent of µ, and thus L ∝ µ4 . For very massive stars the first term in
the Eddington quartic dominates, which means that µβ ≈ constant, so L stays
constant. However, this will apply only to very, very massive stars. Thus in
general we expect L to increase with µ, with the largest increases at low masses
and smaller increases at high masses.
If an entire star were converted from H to He, this would suggest that its lumi-
nosity should go up by a factor of (1.34/0.6)4 = 25 at low masses. Of course
the entire star isn’t converted into He except in fully convective stars that are
uniform throughout, and stars are fully convective only below roughly 0.3 M .
These stars have main sequence lifetimes larger than the age of the universe, so
none have ever fully converted into He. In more massive stars that have reached
the end of the main sequence, µ increases to 1.34 in their cores, but not elsewhere,
so the mean value of µ and the luminosity increase by a smaller amount.
This simple understanding is also in good agreement with the results of numerical
calculations. What is a bit less easy to understand analytically, but also happens,
is that stars radii swell, reducing their effective temperatures. The swelling is
greatest for the most massive stars, so, although they do not move very far in L,
they move a considerable distance in Teff .
[Slide 2 – HR diagram tracks]

2
II. The Red Giant Phase

A. The Schönberg-Chandrasekhar Limit

As stars reach the end of their main sequence lives, they accumulate a core of
helium that is inert, in the sense that it is too cold for 3α reactions to take place,
so the core generates no energy. The consequences of this become clear if we
examine the two stellar structure equations that describe energy generation and
transport:
dF
= q
dm
dT 3 κ F
= − .
dm 4ac T (4πr2 )2
3

Strictly we should write down the possibility for convective as well as radiative
transport in the second equation, but we will see in a moment that is not necessary.
If there is no nuclear energy generation in the He core, then q = 0, which means
dF/dm = 0 in the core. Thus the heat flow through the core must be constant,
and, since there is no heat flow emerging from m = 0, this means that the core
must have F = 0. It immediately follows that dT /dm = 0 in the core as well
– that is, the core is isothermal. This is why we do not need to worry about
convection: since dT /dr = 0, the temperature gradient is definitely sub-adiabatic.
Thus if there was any convection going on in the core, it shuts off once the nuclear
reactions stop due to lack of fuel.
The star as a whole is not necessarily pushed out of thermal equilibrium by this
process because nuclear burning can continue in the material above the core that
still has hydrogen in it. This can be enough to power the star. However, as this
material depletes its hydrogen, it too becomes inert, adding to the mass of the
helium core. Thus the core grows to be a larger and larger fraction of the star as
time passes.
We can show that this configuration of a growing isothermal core cannot continue
indefinitely, and, indeed, must end well before the entire star is converted to He.
This point was first realized by Schönberg and Chandrasekhar in 1942, and in
their honor is known as the Schönberg-Chandrasekhar limit. There are several
ways to demonstrate the result, but the most straightforward is using the virial
theorem.
We will apply the virial theorem to the isothermal core. It requires that
Z Mc
P 1
Ps Vc − dm = Ωc ,
0 ρ 3
where Vc is the volume of the core, Mc is its mass, and Ωc is its binding energy.
The term Ps is the pressure at the surface of the core, and it is non-zero. This is
somewhat different than when applying the virial theorem to the star as a whole:

3
normally when we do so, we drop the surface term on the grounds that the surface
pressure of a star is zero. In this case, however, the core is buried deep inside the
star, so we cannot assume that the pressure on its surface is zero.
Evaluating both the integral and the term on the right hand side is easy. For
the term on the right hand side, we will just use our standard approximation
Ωc = −αGMc2 /Rc , where α is a constant of order unity that depends on the
core’s internal structure. For the integral, because the core is isothermal and of
uniform composition, with a temperature Tc and a mean atomic mass µc = 1.34,
appropriate for pure helium. Assuming the core is non-degenerate (more on this
in a bit), we have P/ρ = (R/µc )Tc , which is constant, so

R 1 GMc2
Ps Vc − Tc Mc = − α .
µc 3 Rc

Re-arranging this equation, we can get an expression for the surface pressure:
3 RTc Mc αG Mc2
Ps = − ,
4π µc Rc3 4π Rc4

where we have replaced the core volume with Vc = 4πRc3 /3.

An interesting feature of this expression is that, for fixed Mc and Tc , the pressure
Ps reaches a maximum at a particular value of Rc . We can find the maximum in
the usual way, by differentiating Ps with respect to Rc and solving:
dPs 9 RTc Mc αG Mc2
0= = − +
dRc 4π µc Rc4 π Rc5
4αG Mc µc
Rc = .
9R Tc
Plugging this in, the maximum pressure is
37 R4 Tc4
Ps,max = .
210 πα3 G3 µ4c Mc2

The physical meaning of this maximum is as follows: if one has a core of fixed
mass and temperature, and exerts a certain pressure on its surface, it will pick
a radius such that it is in equilibrium with the applied surface pressure. At low
surface pressure Rc is big. In such a configuration self-gravity, represented by the
term αGMc2 /(4πRc4 ) in the equation for Ps , is unimportant compared to internal
thermal pressure, represented by the term 3RTc Mc /(4πµc Rc3 ). As the external
pressure is increased, the radius shrinks, and the thermal pressure of the core
goes up as Rc−3 .
However, if the pressure is increased enough, the self-gravity of the core is no
longer unimportant. As self-gravity grows in importance, one has to decrease the
radius more and more quickly to keep up with an increase in surface pressure,

4
because more and more of the pressure of the core goes into holding itself up
against self-gravity, rather than opposing the external pressure. Eventually one
reaches a critical radius where the core is exerting as much pressure on its surface
as it can. Any further increase in the external pressure shrinks it further, and
self-gravity gets stronger faster than the internal pressure grows.
We can estimate the pressure exerted on the surface of the helium core by the
rest of the star (the envelope). To calculate this, we note that the envelope must
obey the equation of hydrostatic equilibrium, and that we can integrate this from
the surface of the isothermal core to the surface of the star:
Z 0 Z M
dP Gm Gm
=− 4
=⇒ dP = −Ps = − dm.
dm 4πr Ps Mc 4πr 4

Approximate this as
Gm Gm
4
≈ .
4πr 4πR4
Plugging this into the integral gives
Z M
Gm G ZM G  2 2
 GM 2
Ps ≈ dm = m dm = M − M c ≈ .
Mc 4πR4 4πR4 Mc 8πR4 8πR4

Combining this with our previous result gives a rough condition that the star
must satisfy if it is to remain in hydrostatic equilibrium:
GM 2 37 R4 Tc4
≈
8πR4 210 πα3 G3 µ4c Mc2

To see when this is likely to be violated, consider the gas just above the surface
of the isothermal core. The temperature and pressure must change continuously
across the core edge, so the envelope pressure and temperature there obey Tenv =
Tc and Penv = Ps . Applying the ideal gas law to the envelope we have
Ps µenv
Tenv = Tc = ,
Rρenv
where µenv and ρenv are the mean molecular weight and density just above the
envelope. The maximum temperature occurs when Ps is at its maximum value,
and substituting in Ps = Ps,max gives
37 R4 Tc4
!
µenv
Tc =
Rρenv 210 πα3 G3 µ4c Mc2
210 πα3 G3 µ4c Mc2 ρenv
Tc3 =
37 R3 µenv

As an extremely rough estimate we can also take ρenv ∼ 3M/(4πR3 ), and plugging
this in gives
3 28 α3 G3 µ4c Mc2 M
Tc ≈ 6 3 .
3 R µenv R3

5
 Thus we have now estimated Tc terms of the properties of the star. Plugging this
into our condition for stability gives
!4/3
GM 2 > 37 R4 1 28 α3 G3 µ4c Mc2 M
8πR4 ∼ 210 πα3 G3 µ4c Mc2 36 R3 µenv R3
s !2
Mc < 27 µenv
M ∼ 2048α3 µc
Doing the analysis more carefully rather than using crude approximations, the
coefficient turns out to be 0.37:
!2
Mc µenv
≤ 0.37 .
M µc

Since µenv < µc , this implies that the core can only reach some relatively small
fraction of the star’s total mass before hydrostatic equilibrium becomes impossi-
ble. Using µenv = 0.6 and µc = 1.3, the limit is that Mc ∼ < 0.1M . Once a star
reaches this limit, the core must collapse.
This limit applies to stars that are bigger than about 2 M . For smaller stars,
the gas in the He core becomes partially degenerate before the star reaches the
Schönberg-Chandrasekhar limit. Since in a degenerate gas the pressure does not
depend on the temperature, the pressure can exceed the result we got assuming
isothermal gas. This allows the core to remain in hydrostatic equilibrium up to
higher fractions of the star’s mass.
B. The Sub-Giant and Red Giant Branches
Collapse of the core causes it to cease being isothermal, because it provides a new
source of power: gravity. The collapse therefore allows hydrostatic equilibrium
to be restored, but only at the price that the core shrinks on a Kelvin-Helmholtz
timescale.
The core also heats up due to collapse, and this in turn heats up the gas around
it where there is still hydrogen present. This accelerates the burning rate in
the shell above the helium core. Moreover, it does so in an unstable way. The
increase in temperature is driven by the KH contraction of the core, which is not
sensitive to the rate of nuclear burning because none of the burning goes on in
the collapsing core. Thus the burning rate will accelerate past the requirements
of thermal equilibrium, and Lnuc > L.
Consulting the virial theorem, we can understand what this implies must happen.
Recall that we have shown several times that for stars with negligible radiation
pressure support,
dE 1 dΩ dU
Lnuc − L = = =− .
dt 2 dt dt
Since Lnuc > L, the left hand side is positive, and we conclude that Ω must
increase and U must decrease. The potential energy −Ω ∝ GM 2 /R, and the

6
 thermal energy U ∝ M T . Since the mass is fixed, the only way for Ω to increase
is if R gets larger (since this brings Ω closer to zero), and the only way for U to
decrease is for the mean temperature T to decrease.
Thus the unstable increase in nuclear burning causes the radius of the star to
expand, while its mean temperature drops. In the HR diagram, this manifests as
a drop in Teff . As a result, the star moves to the right in the HR diagram. The
phase is called the sub-giant branch.
[Slide 2 – HR diagram tracks]
In low mass stars the migration is slow, because the core is restrained from outright
collapse by degeneracy pressure. In more massive stars the migration is rapid,
since the core collapses on a KH timescale. For this reason we only see fairly low
mass stars on the sub-giant branch. More massive stars cross it too rapidly for
us to have any chance of finding one.
There is a limit to how red a star can get, which we encountered last time during
our discussion of protostars: the Hayashi limit. As a post-main sequence star
moves to the right in the HR diagram, it eventually bumps up against the ∼ 4000
K limit imposed by H− opacity. Since it can no longer deal with having Lnuc > L
by getting any colder at its surface, it instead has to increase its radius. This allows
the internal temperature and the gravitational binding energy to drop, complying
with energy conservation, and it also increases the luminosity, decreasing the
difference between Lnuc and L. This phase of evolution is known as the red giant
phase, and stars that are at low temperature and high and rising luminosity are
called red giants.
[Slide 2 – HR diagram tracks]
Red giants also display an interesting phenomenon called dredge-up. The high
opacity of the low-temperature envelope of the red giant guarantees that it will be
convectively unstable, and the convective zone reaches all the way down to where
the region where nuclear burning has taken place. It therefore drags up material
that has been burned, changing the visible composition of the stellar surface.
Nuclear burning destroys lithium (as some of you showed on your homework)
and increases the abundance of C and N, and in red giants we can observe these
altered compositions.

III. The Helium Burning Phase

We showed earlier that the temperature of the isothermal core of the star is given
approximately by
210 πα3 G3 µ4c Mc2 ρenv
Tc3 ≈ .
37 R3 µenv
As the star ascends the red giant branch, ρenv is dropping, but at the same time Mc
is rising as more and more mass is added to the core, and its 2nd power-dependence
beats the first power dependence on the dropping ρenv . Thus the core heats up with

7
time. Once it violates the Schönberg-Chandresekhar limit, and it becomes powered by
gravitational contraction, it heats up even more. Thus the core is always getting hotter
during the red giant phase. What happens next depends on the mass of the star.

A. Stars 1.8 − 10 M
First consider fairly massive stars, which turn out to be those larger than 1.8 M .
In such stars, the core temperature eventually reaches ∼ 108 K, which is sufficient
for He burning via the 3α process. At this point He burning provides a new source
of energy in the core, which halts its contraction. Burning of hydrogen continues
in the shell around the He core, but, since it is no longer being driven out of
equilibrium by the contraction of the He core, it slows down. This allows the star
to cease expanding and instead begin to contract, and the star’s luminosity to
decrease. The result is that the star comes back down from the red giant branch,
and moves down and to the left on the HR diagram – higher effective temperature,
lower luminosity.
[Slide 2 – HR diagram tracks]
After a short period the luminosity stabilizes, and since Lnuc < L, the star re-
sponds by having its envelope contract. That contraction leaves the luminosity
unchanged, but moves the star to higher effective temperature. The motion is
roughly horizontal in the HR diagram, so this is known as the horizontal branch –
it is shown by points 7-9 . The duration of this phase is roughly 108 yr, set by the
amount of energy that is produced by a combination of He burning in the core
and H burning in the shell. It ends when the core has been entirely transformed
into C and O.

B. Stars 1 − 1.8 M
For stars from 1 − 1.8 M , the helium core becomes degenerate before it violates
the Schönberg-Chandrasekhar limit. This does not stop it from heating up, but it
does change what happens once the He ignites. Recall our discussion of runaway
nuclear burning instability. In a degenerate gas, the pressure and density are
not connected to the temperature. As a result, once a nuclear reaction starts it
heats up the gas, but does not cause a corresponding expansion that pushes the
temperature back down. This tends to cause the reaction rate to increase, leading
to a runaway. This is exactly what happens in the He core of a low mass star.
Once helium burning starts, it runs away, in a process called the helium flash.
The helium flash ends once the nuclear reactions generate enough energy to lift
the degeneracy in the core, leading it to undergo rapid expansion. This only takes
a few seconds. Thereafter, the envelope responds in a way that is essentially the
opposite of what happens due to core collapse in the red giant phase: it contracts
and heats up. The star therefore moves down off the red giant branch and across
into the horizontal branch much like a more massive star, but it does so rapidly
and violently, on a KH timescale rather than something like a nuclear timescale.

8
 C. Stars Below 1 M
For an even smaller star, the core never heats up enough to reach He ignition, even
once much of the core mass has been converted to He. In this case the remainder
of the envelope is lost through processes that are not completely understood,
and what is left is a degenerate helium core. This core then sits there and cools
indefinitely. This is a helium white dwarf. Stars in this mass range therefore skip
the AGB and PN phases we will discuss in a moment, and go directly to white
dwarfs.

IV. The AGB and PN Phases

The He burning phase ends when the core has been completely converted to carbon
and oxygen. At that point, what happens is essentially a repeat of the red giant phase.
The core begins to contract, driving out-of-equilibrium He burning on its surface.
This forces the envelope to expand, so the star moves back to the right, and lower
temperature, on the HR diagram. Once the temperature drops to ∼ 4000 K at the
surface, the star is up against the Hayashi limit, and the envelope cannot cool any
further. Instead, the star’s radius expands, leading its luminosity to rise as well. The
result is that the star climbs another giant branch, this one called the asymptotic giant
branch, or AGB for short.
As in the red giant phase, the cool envelope becomes convective, and this convection
drags up to the surface material that has been processed by nuclear burning. This is
called second dredge-up, and it manifests in an increase in the helium and nitrogen
abundances at the surface.
While the core is contracting and the envelope is expanding, the hydrogen burning
shell goes out as its temperature drops. However, contraction of the core halts once it
becomes supported by degeneracy pressure. At that point the hydrogen shell reignites,
and this leads to a series of unstable thermal pulses. Thermal pulses work in the
following cycle. As hydrogen burns, it produces helium, which sinks into a thin layer
below the hydrogen burning shell. This layer has no source of energy, so it contracts
and heats up. Once it gets hot enough, it ignites, and, as we showed a few weeks
ago, nuclear burning in a thin shell is also unstable, because the shell can’t expand
fast enough to keep its temperature from rising. Thus all the accumulated He burns
explosively, driving the core of the star to expand and cool, just like in the helium
flash. This expansion also extinguishes the hydrogen burning. Once the He is gone,
however, the cycle can resume again.
This chain of reactions and explosive burning has two other noteworthy effects. First,
it temporarily produces neutron-rich environments, which synthesize elements heavier
than iron via the s process. Second, it briefly churns up carbon from the core and
convects it to the surface. The result is that carbon appears in significant quantities
on the stellar surface, producing what is known as a carbon star. This process is called
third dredge-up.
AGB stars also have significant stellar winds, which drive large amounts of mass loss

9
 from them. The details are not at all understood, but observationally we know that
mass loss rates can reach ∼ 10−4 M yr−1 . The mechanism responsible for carrying the
winds is likely radiation pressure, which is very significant in these stars due to their
high luminosities. These winds carry lots of carbon with them, which condenses as the
gas moves away from the stars and produces carbonaceous dust grains in interstellar
space. The winds also reduce the total mass of the star significantly. As a result all
stars with initial masses below roughly 8 M end up with cores that are below the
Chandrasekhar limit.
The winds eventually remove enough mass from the envelope that all nuclear burning
there ceases, and the star finally goes out. However, the core remains very hot, and,
once enough mass is removed, it is directly exposed and shines out the escaping gas.
The high energy photons produced by the hot core surface are sufficient to ionize this
gas, and the entire ejected shell of material lights up like a Christmas tree. This object
is known as a planetary nebula. (Even though it has nothing to do with planets, the
people who named it didn’t know that at the time, and through a very low resolution
telescope they look vaguely planetary.)
PN are some of the most visually spectacular objects in the sky, due to the variety of
colors produced by the ionized gas, and the complex shapes whose origins we do not
understand.
[Slides 3-5 – a gallery of PN]

V. White Dwarfs
The final state once the gas finishes escaping is a degenerate core of carbon and oxygen
with a typical mass of ∼ 0.6 M . Lower mass stars that cannot ignite helium end up
with masses of ∼ 0.2 − 0.4 M . We can understand the final evolution of these stars
with a simple model. The center of the star consists of a degenerate electron gas.
However, the pressure must go to zero at the stellar surface, so at some radius the
pressure and density must begin to drop, and the gas ceases to be degenerate. Thus
the star consists of a degenerate core containing most of the mass, and a non-degenerate
envelope on top of it. Within the degenerate part, thermal conductivity is extremely
high, so the gas is essentially isothermal – it turns out that a degenerate material acts
much like a metal, and conducts very well.
In the non-degenerate part of the star, the standard equations of hydrostatic balance
and radiative diffusion apply:
dP GM
= −ρ 2
dr r
dT 3 κρ L
= − .
dr 4ac T 3 4πr2
Note that we have M and not m in the numerator of the hydrostatic balance equation
because we’re approximating that all of the star’s mass is in the inner, degenerate
part. We also approximate that all the energy lost from the star comes from the
inner, degenerate part, so F = L = constant in the non-degenerate layer. Finally,

10
note that the energy conservation equation dF/dm = q does not apply, because we are
not assuming that the star is in thermal equilibrium – indeed, it cannot be without a
source of nuclear energy.
We assume that the opacity in the non-degenerate part of the star is a Kramer’s opacity
κ0 µ
κ = κ0 ρT −7/2 = P T −9/2 ,
R
where we have used the ideal gas law to set ρ = (µ/R)(P/T ). Substituting this into
the radiative diffusion equation gives
dT 3 1 κ0 µ L 3κ0 µ P ρ L
 
=− P T −9/2 ρ =− .
dr 4ac T 3 R 4πr 2 16πacR T 15/2 r2
If we now divide by the equation of hydrostatic balance, we obtain
dT 3κ0 µ P L
=
dP 16πacRG T 15/2 M
16πacRG M 15/2
P dP = T dT.
3κ0 µ L

We can integrate from the surface, where P = 0 and T = 0 to good approximation,

inward, and obtain the relationship between pressure and temperature
Z P !1/2  1/2
0 0 16πacRG M Z T 015/2 0 64πacRG M
P dP = T dT =⇒ P = T 17/4 .
0 3κ0 µ L 0 51κ0 µ L

Using the ideal gas law ρ = (µ/R)(P/T ) again, we can turn this into
1/2  1/2
64πacµG M

ρ= T 13/4 .
51κ0 R L

This relationship between density and temperature must hold everywhere in the ideal
gas region, and so we can apply it at the boundary between that region and the
degenerate region. The pressure in the non-degenerate region is just
R
Pnd = ρT,
µe
where we’ve used µ = µe because the electron pressure completely dominates. Just on
the other side of the boundary, in the degenerate region, the pressure is
!5/3
ρ
Pd = K10 .
µe

Pressure, density, and temperature must change continuously across the boundary, so
the ρ that appears in these two expressions is the same. Moreover, since the core is

11
isothermal, T = Tc , where Tc is the core temperature. Finally, since the pressures must
match across the boundary, we have
!5/3
R ρ
ρTc = K10
µe µe
0
K1 2/3
T = 2/3
ρ
Rµe
1/2  1/2 #2/3
K10
"
64πacµG M 13/4
= 2/3
T
Rµe 51κ0 R L
L 64πacGK103 µ 7/2
= T .
M 51R4 κ0 µ2e c

We have therefore derived the luminosity of a white dwarf in terms of the temperature
of its degenerate core. Plugging in typical values gives
7/2
L/L Tc

≈ 6.8 × 10−3 .
M/M 107 K

We can use this relation to infer how long white dwarfs will shine brightly enough for
us to see them. The internal energy of the white dwarf is just the thermal energy of
the gas. Since the electrons are degenerate they cannot lose energy – there are no lower
energy states available for them to occupy. The ions, however, are not degenerate, and
they can cool off. Since the ions are a non-degenerate ideal gas, their internal energy
is
3R
UI = M Tc ,
2 µI
and conservation of energy requires that
dUI 3 R dTc
L=− =− M .
dt 2 µI dt
It is convenient to recast this relation in terms of the luminosity. Using our temperature-
luminosity relationship we have
!2/7
51R4 κ0 µ2e L
Tc =
64πacGK103 µ M
!2/7
dTc 2 51R4 κ0 µ2e 1 dL
= L−5/7
dt 7 64πacGK103 µ M dt

Plugging this into the equation for L gives

!2/7
3 R15/7 5/7 51κ0 µ2e dL
L=− M L−5/7 .
7 µI 64πacGK103 µ dt

12
Separating the variables and integrating from an initial luminosity L0 to a luminosity
L at some later time, we have
!−2/7 Z
Z L
0−12/7 0 7 µI −5/7 51κ0 µ2e t
L dL = − M dt0
L0 3 R15/7 64πacGK103 µ 0
!−2/7
7 −5/7
 7 µI 51κ0 µ2e
− L−5/7 − L0 = − M −5/7
t
5 3 R15/7 64πacGK103 µ

5/7 !−2/7 −7/5
5 µI L0 51κ0 µ2e

L = L0 1 + t
3 R15/7 M 64πacGK103 µ

For long times t, we can drop the +1, and we find that L ∝ t−7/5 . Since the white
dwarf birthrate in the galaxy is about constant, this immediately yields an important
theoretical prediction. The number of white dwarfs we see with a given luminosity
should be proportional to the amount of time they spend with that luminosity, which
we have just shown varies as t ∝ L−5/7 . Thus luminous white dwarfs should be rare
because they cool quickly, while dimmer ones should be more common because they
cool more slowly, and the ratio of the number of white dwarfs with luminosity L1 to
the number with luminosity L2 should vary as (L1 /L2 )−5/7 . Observations confirm this
result.
We can also define a characteristic cooling time tcool as the time it takes a white dwarf’s
luminosity to change significantly. This is simply the time required for the second term
in parentheses to become of order unity, which is
!2/7  5/7 !5/7
3R15/7 51κ0 µ2e M 6 M/M
tcool ≈ ≈ 2.5 × 10 yr.
5µI 64πacGK103 µ L0 L/L

Thus we conclude that white dwarfs with luminosities of L ∼ 104 L , typical of the
planetary nebula phase, should last only a few thousand years, while those with much
lower luminosities ∼ L can remain at that brightness for of order a million years.

13