Vous êtes sur la page 1sur 17

1

Fermi-Dirac Gases

9.1 THE FERMI ENERGY

The behavior of fermions, indistinguishable particles of half-integer spin, is governed by
Fermi-Dirac statistics. Fermions obey the Pauli exclusion principle, which prohibits the
occupation of any quantum state by more than one particle. Consider an ideal gas comprising

N noninteracting fermions, each of mass

m, in a container of volume

V held at temperature

T. The Fermi-Dirac distribution is


f
j
=
N
j
g
j
=
1
e
("
j
#) kT
+1
(9.1)

or in the continuum approximation


f (") =
N(")
g(")
=
1
e
(" #) kT
+1
(9.2)

The function

f (") is the Fermi function; it gives the probability that a single particle state

"
will be occupied by a fermion. Clearly,

0 " f (#) "1.

At

" = , the Fermi-Dirac distribution

f (") = e
(" # ) kT
+1
( )
#1
= 1+1
( )
#1
=
1
2
, regardless of the
temperature. The chemical potential

= (T) is a function of temperature and its value at

T = 0, that is

(0), is called the Fermi energy

"
F
= (0) .

The behaviour of the exponent in the Fermi function as

T "0 depends on the sign of the
numerator as


(" #(0))
kT
=
#$ if " < (0)
$ if " > (0)
%
&
'
so

f (") =
1 if " < (0)
0 if " > (0)
#
$
%
(9.3)

Thus at

T = 0 all states with energy

" < (0) are occupied and all states with

" > (0) are
unoccupied. At absolute zero fermions occupy all the lowest energy states available. The
exclusion principle says that only one fermion is allowed per state, so all

N particles will be
crowded into the

N lowest energy levels.

2
Thus only one configuration (or microstate) is
possible for the whole assembly at

T = 0. Then
the thermodynamic probability

w =1 and

S = k lnw = 0. The vanishing of the entropy at
absolute zero is consistent with the third law of
thermodynamics. The Fermi function at

T = 0
is shown in Figure 9.1.

How does the Fermi energy

(0) depend on

m,

N, and

V ? It obviously doesnt depend on the
temperature since

T = 0. We need the density
of states function

g(") for particles of spin

1
2
,
such as electrons, where the spin factor is 2, so


g(")d" =
8 2#V
h
3
m
3 2
"
1 2
d" = 4#V
2m
h
2
$
%
&
'
(
)
3 2
"
1 2
d" . (9.4)

The number of particles is conserved, so

N
j
j
"
= N, or equivalently for the continuum case


N(")d"
0
#
$
= f (")g(")d"
0
#
$
= N. (9.5)

For energies greater than the Fermi energy the Fermi-Dirac function is zero, so combining
equations (9.3), (9.4) and (9.5), we get


N = g(")d"
0
(0)
#
= 4$V
2m
h
2
%
&
'
(
)
*
3 2
"
1 2
d"
0
(0)
#
=
8$V
3
2m
h
2
%
&
'
(
)
*
3 2
(0)
3 2
(9.6)

Solving equation (9.6) for the Fermi energy

(0) we obtain the result


(0) =
h
2
2m
3N
8"V
#
$
%
&
'
(
2 3
(9.7)

For convenience we introduce the Fermi temperature

T
F
such that

(0) ="
F
= kT
F
. This can
be written as


T
F
=
(0)
k
=
h
2
2"mk
3"
3 2
N
8"V
#
$
%
&
'
(
2 3
=
h
2
2"mk
3"
1 2
N
8V
#
$
%
&
'
(
2 3
=
h
2
2"mk
N
1.504V
#
$
%
&
'
(
2 3
(9.8)

Figure 9.1 The Fermi function at

T = 0.
3

where

3
8
"
1 2
=1 1.504. This is analogous to the Bose temperature given in eqn (8.38), but with
a different numerical factor.


9.2 THE CALCULATION OF

(T)

Taking our system of fermions and
increasing the temperature from zero to
some finite

T the chemical potential

(T) must change. In figure 9.2 is the
change in the Fermi function with
temperature. To obtain these curves, we
must determine the chemical potential at
the current temperature, a considerably
more complicated exercise than for

T = 0. The number of particles has not
changed so


N = f (")g(")d"
0
#
$
= 4%V
2m
h
2
&
'
(
)
*
+
3 2
"
1 2
d"
e
(" ,) kT
+1
0
#
$
(9.9)

Equating equation (9.9) with the

T = 0 result in equation (9.6) gives



N =
8"V
3
2m
h
2
#
$
%
&
'
(
3 2
(0)
3 2
for T= 0
! " # # # $ # # #
= 4"V
2m
h
2
#
$
%
&
'
(
3 2
)
1 2
d)
e
() *) kT
+1
0
+
,
for T-0
! " # # # # $ # # # #
.

Eliminating common factors


2
3
(0)
3 2
=
"
1 2
d"
e
(" #) kT
+1
0
$
%
(9.10)

0
0.2
0.4
0.6
0.8
1
1.2
0 0 5 1 1.5 2
f(!)
!/(0)
T=1.2T
F
="0.45(0)
T=0.2T
F
=0.96(0)

Figure 9.2 The Fermi function at

T = 0.2T
F
and at

T =1.2T
F
.
4
The plus sign in the denominator of the integrand makes all the difference in our ability to
evaluate the integral. It can be evaluated numerically. But we can find a very good infinite
series approximation for

(T) in the following way.

The function

"
1 2
is smooth and so is the Fermi function

f (") , but the derivative of the Fermi
function

df d" has a maximum near

" = and for low temperatures is strongly peaked at the
maximum. This suggests integration by parts where


"
1 2
f (")d"
0
#
$
=
2
3
"
3 2
f (")
[ ]
0
#
%
2
3
"
3 2
df (")
d"
d"
0
#
$
(9.11)

The boundary term


2
3
"
3 2
f (") =
2
3
"
3 2
e
(" #) kT
+1
$
0 " $0
0 " $%
&
'
(
so

2
3
"
3 2
f (")
[ ]
0
#
= 0

Therefore


"
1 2
f (")d"
0
#
$
= %
2
3
"
3 2
df (")
d"
d"
0
#
$


Looking at Figure 9.1 shows that at

T = 0, the function

f (") has a zero slope everywhere except at

" = (0) ,
where the slope is infinite and negative. Thus the
derivative

df (") d" is a Dirac delta function. At
temperatures less than

T
F
but greater than zero, we might
expect the derivative to be a smoothed delta function
peaked at

" = (see figure 9.3). This suggests that if we
expand

F(") =
2
3
"
3 2
in a Taylor series about its
maximum

, the only significant contributions to the
integral will be in the vicinity of

" = .
1
The expansion is


F(") = F() +
#
F () " $
( )
+
1
2!
# #
F () " $
( )
2
+
1
3!
F
(3)
() " $
( )
3
+
1
4!
F
(4)
() " $
( )
4
+...
=
2
3

3 2
+
1 2
" $
( )
+
1
4

$1 2
" $
( )
2
$
1
3!
1
4

$3 2
" $
( )
3
+
1
4!
3
8

$5 2
" $
( )
4
+...
(9.12)


1
This method of evaluating the chemical potential and the internal energy for low temperatures is known as the
Sommerfeld expansion.
0
0 2
0.4
0 6
0 8
1
1 2
0 0.5 1 1.5 2
eps/mu(0)
f(!)
-df(!)/d!
(-df(!)/d!)!
3/2

Figure 9.3
5
Changing to the dimensionless variable

y = (" #) kT , this becomes


F(y) =
2
3

3 2
+
1 2
ykT +
1
4

"1 2
ykT
( )
2
"
1
4!

"3 2
ykT
( )
3
+
1
64

"5 2
ykT
( )
4
+ ... (9.13)

Evaluating the derivative the Fermi function and changing variables we obtain


"
df (#)
d#
= "
d
d#
1
e
(# ") kT
+1
=
1
kT
e
(# " ) kT
e
(# " ) kT
+1
( )
2
=
1
kT
e
y
e
y
+1
( )
2
,

and as

dy = d" kT, the integral becomes


"
2
3
#
3 2
df (#)
d#
d#
0
$
%
= F(y)
e
y
e
y
+1
( )
2
dy
" kT
$
%
& F(y)
e
y
e
y
+1
( )
2
dy
"$
$
%
=
2
3

3 2
+
1 2
kTy +
1
4

"1 2
kTy
( )
2
"
1
4!

"3 2
kTy
( )
3
+
1
64

"5 2
kTy
( )
4
+ ...
{ }
e
y
e
y
+1
( )
2
dy
"$
$
%
.

(The integrand is approximately zero for

y < " kT so we can extend the integral to

"#
).
The factor

e
y
e
y
+1
( )
2
= e
y
e
2y
+ 2e
y
+1
( )
=1 e
y
+ 2 + e
"y
( )
is an even function of

y, and the
integral of an even function times an odd function from

"#
to

"
is zero, so the only non-zero
terms are


"
1 2
f (")d"
0
#
$
=
2
3

3 2
e
y
e
y
+1
( )
2
dy
%#
#
$
+
(kT)
2
4
1 2
y
2
e
y
e
y
+1
( )
2
dy
%#
#
$
+
(kT)
4
64
5 2
y
4
e
y
e
y
+1
( )
2
dy
%#
#
$
+ ...

The first integral has the value 1
2
, the second integral is

"
2
3, and the third
3
is

7"
4
15. Thus
we have


"
1 2
f (")d"
0
#
$
=
2
3

3 2
+
%
2
12
(kT)
2

1 2
+
7%
4
15
(kT)
4
64
5 2
=
2
3
(0)
3 2
&
3 2
1+
%
2
8
kT

'
(
)
*
+
,
2
+
7%
4
640
kT

'
(
)
*
+
,
4
'
(
)
)
*
+
,
,
= (0)
3 2



2
Show that

e
y
+ 2+e
"y
= e
y 2
+e
"y 2
( )
2
= 4 cosh
2
y 2
( )
and hence evaluate the integral as

1
4
sec h
2 y
2
( )
dy
"#
#
$
=
1
2
sec h
2
y dy
"#
#
$
=
1
2
tanh y [ ]
"#
#
=1.
3
See also Kardar, Statistical Physics of Particles, page 190.
6
Taking the

2
3
power of both sides of this equation and then expanding the bracketed term
gives


= (0) 1+
"
2
8
kT

#
$
%
&
'
(
2
+
7"
4
640
kT

#
$
%
&
'
(
4
#
$
%
%
&
'
(
(
)2 3
* (0) 1)
"
2
12
kT

#
$
%
&
'
(
2
+
"
4
720
kT

#
$
%
&
'
(
4
...
#
$
%
%
&
'
(
(


for

kT << . Finally, we replace

in the correction term (on the RHS) by

(0) = kT
F
, to
obtain


= (0) 1"
#
2
12
kT
(0)
$
%
&
'
(
)
2
$
%
&
&
'
(
)
)
* (0) 1"
#
2
12
T
T
F
$
%
&
'
(
)
2
$
%
&
&
'
(
)
)
T << T
F
( ) . (9.15)

A comparison of this approximate
expression with the exact numerical
evaluation is shown in Figure 9.4. At

= 0 (the intersection with the
temperature axis) the approximation
gives

T T
F
= 12 " =1.10 instead of

T T
F
= 0.999, and

=1"
3
16
#
2
= "0.85(0) instead of

= "1.1(0) at

T T
F
=1.5.

We especially note that

is positive for
temperatures below the Fermi
temperature and negative for higher
temperatures. As the temperature
increases above

T
F
, more and more of the fermions are in the excited states and the mean
occupancy of the ground state falls below

1
2
. In this region,


f (0) =
1
e
" kT
+1
<
1
2


which implies that

e
" kT
+1> 2 # e
" kT
>1 # " kT > 0

kT
< 0,


Figure 9.4 Exact and approximate calculations of
!
(0)
versus

T T
F
.
7
or

< 0. We also note that this situation is different for a boson gas, where

is negative at
all temperatures and is zero at absolute zero.

At high temperatures the fermion gas approximates the classical ideal gas. In the classical
limit,


= "kTln
Z
N
#
$
%
&
'
( (9.16)

with


Z
N
= 2
2"mkT
h
2
#
$
%
&
'
(
3 2
V
N
(9.17)

(The spin degeneracy factor is 2 for fermions.) For

T >> T
F
,

kT takes on a large negative
value and

exp " kT
( )
>>1. As an example, consider a kilomole of

3
He gas atoms (which are
fermions) at standard temperature and pressure (Problem 9.2). The Fermi temperature is

0.069 K, so that

T T
F
= 3900. Using equations (9.16) and (9.17), we find that

kT = "12.7
and

exp " kT
( )
= 3.3#10
5
. The average occupancy of single particle states is very small, as
in the case of an ideal dilute gas obeying the Maxwell-Boltzmann distribution.


8
9.3 FREE ELECTRONS IN A METAL

Statistical thermodynamics provides profound insights into the behavior of conduction
electrons in metals at moderate temperatures. Electrons are spin

1
2
fermions. Each atom in the
crystal lattice of the metal is assumed to part with some number of its outer valence electrons,
which can then move freely about in the metal. There is an electric field due to the positive
ions that varies widely from point to
point. However, the effect of the field
is canceled out except at the surface of
the metal where there is a strong
potential barrier, called the work
function that draws back into the
metal any electron that happens to
make a small excursion outside. The
free electrons are therefore confined to
the interior of the metal as gas
molecules are confined to the interior of a container. We speak of the electrons as an electron
gas.

In this model, the free electrons move in a potential box or well whose walls coincide with the
boundaries of the specimen. They occupy energy states up to the so-called Fermi level, which
is the chemical potential

(T) . The work function

" is the energy required to remove an
electron at the Fermi level from the metal surface. The depth of the potential well is equal to

(T) + " (Figure 9.5).

The Fermi level of the free electrons in most metals at room temperature is only fractionally
less than the Fermi energy

"
F
# (0) . It is often assumed that the two are equal, and this leads
to confusion. The Fermi level, strictly speaking, is

(T) , which is an approximation to the
Fermi energy valid for

T << T
F
.

A more realistic picture of the
potential well is given in Figure 9.6,
which shows how the potential varies
in the vicinity of the positive ions in
the crystal lattice. The periodicity
leads to a band structure in the density
of quantum states, which is the
foundation of semiconductor physics.

Figure 9.6 Sketch of the potential well showing periodicities
associated with the positive ions of the crystal.

Figure 9.5 Potential well for free electrons in a metal.
9

To get an idea of the magnitude of the quantities in the electron gas model, we consider the
free electrons in silver, which is monovalent (one free electron per atom). The density of
silver is

10.5 "10
3
kg m
#3
and its atomic weight is 107. The concentration is therefore


N
V
=10.5 "10
3
kg
m
3
"
1kilomole
107 kg
"
6.02 "10
26
atoms
kilomole
= 5.90 "10
28
m
#3


Since silver is monovalent, this is also the electron concentration. Fermi energy is given by
Equation (9.7):


"
F
= (0) =
h
2
2m
3N
8#V
$
%
&
'
(
)
2 3


Here

m is the electron mass, equal to

9.11"10
#31
kg. So


"
F
=
6.63#10
$34
( )
2
2 #9.11#10
$31
3#5.90 #10
28
8%
&
'
(
)
*
+
2 3
= 8.85 #10
$19
J #
1eV
1.6 #10
$19
J
= 5.6eV.

The Fermi temperature is


T
F
=
"
F
k
=
5.6 eV
8.62 #10
$5
eVK
$1
= 65000K.

The ratio

T T
F
at room temperature is


T
T
F
=
300
6.5 "10
4
= 0.00462

At room temperature, therefore, the electron gas is in the so-called degenerate region

T << T
F
.
The chemical potential (the Fermi level) can be found from Equation (9.15). The computation
gives


(T) = 0.999"
F
.

This shows why

(T) is often identified with

"
F
.

10
The work function

" depends on the metal and the condition of its surface and is typically of
the order of

3" 4eV. At very high temperatures, some of the free electrons may have
sufficient energies to leave the metal. This results in thermionic emission. The condition for
their escape is


p
x
2
2m
> " + (T) ,

where

p
x
is the component of the electrons momentum normal to the surface of the metal.


9.4 PROPERTIES OF A FERMION GAS

The function

N(")d" is the number of fermions in the single particle energy range

" to

" + d".
We know that


N(")d" = f (")g(")d",

where

f (") is the Fermi function and

g(") the degeneracy or density of states function which
behave as shown in Figure 9.7 for a temperature in the range

0 < T << T
F
. The product of
these two functions gives

N(") versus

" as shown in Figure 9.8. The electrons
crowd around the Fermi energy
because the degeneracy increases with
energy; there are therefore more
available quantum states to be
occupied with 0 or 1 electron(s) per
state. (It is the electrons in the tail of
this distribution that have the best
chance of escaping from the metal in
the free electron model.)

The internal energy of a gas of

N
fermions is
0
0.5
1
1.5
0 0.5 1 1.5 2
!
1/2
f(!)
!
!
1/2
f(!)

Figure 9.7 The variation with single particle energy of (a)
the Fermi function, and (b) the degeneracy (density of
states) function. The curves are sketched for

0 <T <<T
F
.

11


U = "N(")d"
0
#
$
= "f (")g(")d"
0
#
$
= 4%V
2m
h
2
&
'
(
)
*
+
3 2
"
3 2
d"
e
(" ,) kT
+1
0
#
$


An approximate evaluation of the integral can
be carried out in the same manner as we
evaluated the integral in section 9.2. If the term
of order

T
4
is included, the result is


U "
3
5
N#
F
1+
5$
2
12
T
T
F
%
&
'
(
)
*
2
+
$
4
16
T
T
F
%
&
'
(
)
*
4
+ ...
,
-
.
.
/
0
1
1
(9.18)

At

T = 0,

U =
3
5
N"
F
; this energy is large because all the electrons must occupy the lowest
energy states up to the Fermi level. The average energy of a free electron in silver at

T = 0 is


" (0) =
U(0)
N
=
3
5
"
F
= 3.4 eV

Note that the mean kinetic energy of an electron, even at absolute zero, is two orders of
magnitude greater than the mean kinetic energy of an ordinary gas molecule at room
temperature.

The electronic heat capacity

C
e
can be found by taking the derivative of Eq (9.18):


C
e
=
dU
dT
=
"
2
2
Nk
T
T
F
#
$
%
&
'
(
)
3"
2
10
T
T
F
#
$
%
&
'
(
3
+ ...
*
+
,
,
-
.
/
/
(9.19)

For temperatures that are small compared with the Fermi temperature, we can neglect the
second term in the expansion compared with the first and obtain


C
e
=
"
2
2
Nk
T
T
F
#
$
%
&
'
(
=
"
2
2
Nk
kT
)
F
#
$
%
&
'
(
(9.20)

For silver at room temperature,


Figure 9.8 The energy distribution of fermions
for

0 <T <<T
F

12

C
e
=
"
2
2
Nk
0.025 eV
5.6 eV
#
$
%
&
'
(
= 2.2 )10
*2
Nk .

Thus the electronic specific heat capacity is

2.2 "10
#2
R. This small value explains a puzzle.
Metals have a specific heat capacity of about

3R, the same as for other solids. It was
originally believed that their free electrons should contribute an additional

3
2
R associated with
their three translational degrees of freedom. Our last calculation shows that the contribution is
negligible.

Why is it so small? While the kinetic
energy of the electrons is much greater
than the thermal energy of electrons in
a gas, the energy of the electrons
changes only slightly with temperature
(

dU dT is small). Only those electrons
near the Fermi level can increase their
energies as the temperature is raised,
and there are very few of them.

At very low temperatures the picture is different. From the Debye theory,

C
V
"T
3
and so the
heat capacity of a metal takes the form


C
V
= AT + BT
3
.

where the first term is the electronic contribution and the second is associated with the crystal
lattice. At sufficiently low temperatures the electronic contribution can dominate, as Figure
9.9 indicates.

Whereas we have emphasized free electrons in metals, most of our results apply to any ideal
gas of fermions. Using the fact that the reversible heat flow into a gas at constant volume is
given by

TdS = C
V
dT, we can calculate the entropy from Equation (9.19):


S =
C
e
d " T
" T
0
T
#
=
$
2
2
Nk
T
T
F
%
&
'
(
)
*
+
$
2
10
T
T
F
%
&
'
(
)
*
3
+ ...
,
-
.
.
/
0
1
1
(9.21)

Therefore

S = 0 at

T = 0, as it must be. The Helmholtz function

F "U #TS is


Figure 9.9 Sketch of the heat capacity of a metal as a
function of temperature showing the electronic and lattice
contributions.
13

F = NkT
F
3
5
"
#
2
4
T
T
F
$
%
&
'
(
)
2
+
#
4
80
T
T
F
$
%
&
'
(
)
4
+ ...
*
+
,
,
-
.
/
/
(9.22)

The fermion gas pressure is found from


P = "
#F
#V
$
%
&
'
(
)
T,N
(9.23)

It is left as Problem 9.9 to prove that


P =
2
5
NkT
F
V
1+
5"
2
12
T
T
F
#
$
%
&
'
(
2
+ ...
)
*
+
+
,
-
.
.
(9.24)

Note that

T
F
depends on volume. Comparison with Equation (9.18) shows that

P =
2
3
(U V).

For silver we found that

N V = 5.9 "10
28
m
#3
and

T
F
= 65,000K. Thus


P "
2
5
5.90 #10
28
( )
1.38 #10
$23
( )
6.5 #10
4
( )
= 2.1#10
10
Pa = 2.1#10
5
atm.

Given this tremendous pressure, we can appreciate the role of the surface potential barrier in
keeping the electrons from evaporating from the metal.


9.5 APPLICATION TO WHITE DWARF STARS

Very high pressures, of the order of

10
17
atm, exist in the degenerate electron gas of a white
dwarf star. It is this pressure that prevents its gravitational collapse.

A typical star has a hot core with a temperature

T ~ 10
7
K. The heat energy is supplied by
thermonuclear reactions. The atoms are completely ionized (

kT is 900 eV at this
temperature), creating a huge electron gas. In young stars the electron pressure is sufficient
to withstand the weight of the material pressing on the center, thereby preventing collapse. In
old stars, the hydrogen at the core has run out, fusion has stopped, and the core cools.
However, the loss of gravitational energy results in an increase in the kinetic energy of the
electrons and ions and the cooling process is partially offset. In white dwarfs the electron gas
14
pressure prevents the collapse beyond a certain point. White dwarfs typically have the mass of
the Sun and the radius of the Earth.

The pressure of the electron gas in a white dwarf can be estimated using


P =
2
5
N
V
"
#
$
%
&
' (
F
, T << T
F
, (9.25)

Sirius B is believed to be a white dwarf. We need the following characteristics of Sirius B:

Mass

M = 2.09 "10
30
kg;
Radius

R = 5.57 "10
6
m which implies Volume

V = 7.23"10
20
m
3


Strictly we will use the observed values of

M and

V to predict

R.

We assume that nuclear fusion has ceased - that all the core hydrogen has been converted to
helium. The helium is completely ionized - so each He atom has lost two electrons. In
addition, we suppose that the number of nucleons in the star is equal to its mass divided by
the mass of a nucleon (proton or neutron). Thus

No. of nucleons

=
2.09 "10
30
1.66 "10
#27
=1.26 "10
57


Since there are four nucleons and two electrons in a helium atom, the number of electrons in
the fermion gas of Sirius B is

N = 0.63"10
57
. Then, with

m
e
= 9.11"10
#31
kg,


"
F
=
h
2
2m
e
3N
8#V
$
%
&
'
(
)
2 3
= 5.33*10
+14
J = 0.33 MeV.

Then the Fermi temperature is

T
F
="
F
k = 3.9 #10
9
K. If

T =10
7
K, then the condition

T << T
F
is satisfied. To a fair approximation, the electron gas pressure is given by Equation
(9.25). Thus


P =
2
5
0.63"10
57
7.23"10
20
#
$
%
&
'
(
5.33"10
)14
( )
=1.8 "10
22
Pa =1.8 "10
17
atm.

Wed like to show that this pressure is sufficient to prevent further collapse.
15

A white dwarf is stable when its total energy is a minimum. The energy is


U =U
e
+U
grav
(9.26)

where

U
e
is the energy of the electron gas and

U
grav
is the gravitational energy. We need to
express each in terms of the stars radius

R, then take the derivative of

U with respect to

R
and set it equal to zero. For

T << T
F
,


U
e
"U(0) =
3
5
N#
F
=
3
5
h
2
2m
e
3
8$V
%
&
'
(
)
*
2 3
N
5 3
.

But

V =
4
3
"R
3
, So

U
e
"
3
5
h
2
2m
e
3
8#
4
3
#R
3
$
%
&
'
(
)
2 3
N
5 3
=
3h
2
10m
e
9
32#
2
$
%
&
'
(
)
2 3
N
5 3
1
R
2
=
a
R
2



U
e
=
a
R
2
where

a =
3h
2
10m
e
9
32"
2
#
$
%
&
'
(
2 3
N
5 3
. (9.27)

The gravitational energy of a solid sphere of mass

M and radius

R is known from classical
mechanics to be


U
grav
= "
b
R
where

b =
3
5
GM
2
. (9.28)
Thus

U =
a
R
2
"
b
R
(9.29)

Minimizing this function, we find that


R
min
=
2a
b
(9.30)

Substituting the values of

a and

b from Equations (9.27) and (9.28), respectively, we find
that (see Problem 9.12)


R
min
" 7 #10
6
m. (9.31)

16
Since the observed value of the radius of Sirius B is roughly equal to this, we conclude that
the star has contracted down to its stable minimum size and is truly a white dwarf.

White dwarfs are at present viewed as comprising a core of degenerate electron gas
surrounded by a nondegenerate (

T > T
F
) outer layer. Nuclear reactions have stopped. The star
is cooling slowly. The time before it becomes invisible is estimated to be

10
10
years. Since the
age of the universe is believed to be

1.5 "10
10
years, it follows that very few white dwarfs
have had time to become invisible.

The stability of white dwarfs is but one example of the very extensive use of quantum
statistics in the study of stellar evolution.
4



9.6 NEUTRON STARS
5


Once a star is more massive than about 1.4 times the mass of the sun, electrons behave
relativistically and cannot prevent further collapse. However, the star will contain neutrons
and these will still be non-relativistic since the neutron mass is larger than the electron mass.
Neutrons are fermions and their pressure, although lower than the electron pressure, can
prevent gravitational collapse. A compact object composed mainly of neutrons is called a
neutron star and the first observation came with the discovery of pulsars in 1967.



4
The discussion of this section was motivated by Chapter 12 of Statistical Physics by WC.V Rosser, Ellis
Horwood, 1982. Rosser gives an extensive treatment of white dwarfs and neutron stars.
5
See section 36.3 of Concepts in Thermal Physics by S. J. Blundell and K. M. Blundell.
17
Appendix:

Following section 9.2 we calculate the average energy of the system, equation (9.18), as

!
U = "f (")g(")d"
0
#
$
= 4%V
2m
h
2
&
'
(
)
*
+
3 2
"
3 2
f (")d"
0
#
$
= 4%V
2m
h
2
&
'
(
)
*
+
3 2
2
5
"
5 2
f (")
[ ]
0
#
,
2
5
"
5 2 df
d"
d"
0
#
$
-
.
/
0
1
2
=
8
5
%V
2m
h
2
&
'
(
)
*
+
3 2
"
5 2
,
df
d"
( )
d"
0
#
$


Expanding
!
F "
( )
="
5 2
in a Taylor series about
!
" = gives

!
F "
( )
=
5 2
+
5
2

3 2
" #
( )
+
15
4

1 2 1
2
" #
( )
2
+
15
8

#1 2 1
6
" #
( )
3
#
15
16

#3 2 1
24
" #
( )
4
+...

Changing variables from
!
" to
!
y = " #
( )
kT gives

!
F y
( )
=
5 2
+
5
2

3 2
kTy
( )
+
15
8

1 2
kTy
( )
2
+
5
16

"1 2
kTy
( )
3
"
5
128

"3 2
kTy
( )
4
+...

Changing variables in the integral from
!
" to
!
y = " #
( )
kT gives
!
dy = d" kT and only the terms that are even
in
!
y contribute to the integral so that

!
U =
8
5
"V
2m
h
2
#
$
%
&
'
(
3 2

5 2 e
y
e
y
+1
( )
2
dy
0
)
*
+
15
8

1 2 (kTy)
2
e
y
e
y
+1
( )
2
dy
0
)
*
+
5
128

+3 2 (kTy)
4
e
y
e
y
+1
( )
2
dy +...
0
)
*
,
-
.
/
.
0
1
.
2
.
****lower -!

But the Fermi energy is
!
"
F
= (0) =
h
2
2m
3N
8#V
( )
2 3
so

!
U =
8
5
"V
2m
h
2
#
$
%
&
'
(
3 2

5 2 e
y
e
y
+1
( )
2
dy
0
)
*
+
15
8
kT

( )
2
y
2
e
y
e
y
+1
( )
2
dy
0
)
*
+
5
128
kT

( )
4
y
4
e
y
e
y
+1
( )
2
dy +...
0
)
*
,
-
.
/
.
0
1
.
2
.


The integrals take the same values as before so

!
U =
8
5
"V
2m
h
2
#
$
%
&
'
(
3 2

5 2
1+
15
8
kT

( )
2
"
2
3
)
5
128
kT

( )
4
7"
4
15
+...
*
+
,
-
.
/
=
8
5
"V
2m
h
2
#
$
%
&
'
(
3 2

5 2
1+
5"
2
8
kT

( )
2
)
7"
4
384
kT

( )
4
+...
*
+
,
-
.
/


Let
!
= (0) 1"
#
2
12
kT
(0)
( )
2
+
#
4
720
kT
(0)
( )
4
+...
$
%
&
'
(
)
and
!
(0) ="
F
on the RHS gives
!
kT
(0)
=
kT
"
F
=
kT
kT
F
=
T
T
F



!
U =
8
5
"V
2m
h
2
#
$
%
&
'
(
3 2
h
2
2m
( )
5 2
3N
8"V
#
$
%
&
'
(
5 3
1)
"
2
12
kT
(0)
( )
2
+...
*
+
,
-
.
/
5 2
1+
5"
2
8
kT

( )
2
)
7"
4
12803
kT

( )
4
+...
1
2
3
4
5
6


!
U =
3
5
N
h
2
2m
( )
3N
8"V
#
$
%
&
'
(
2 3
1)
5"
2
24
kT
(0)
( )
2
+
19"
4
1152
kT
(0)
( )
4
+...
*
+
,
-
.
/
1+
5"
2
8
kT
(0)
( )
2
+
11"
4
128
kT
(0)
( )
4
+...
0
1
2
3
4
5
=
3
5
N6
F
1+
5"
2
12
T
T
F
( )
2
)
"
4
36
T
T
F
( )
4
+...
0
1
2
3
4
5


The fourth order term needs to be done more carefully!

Vous aimerez peut-être aussi