PHY461: Statistical physics and thermodynamics

Lecture 08

QUANTUM STATISTICAL MECHANICS

The Fermi-Dirac distribution
In the previous section, we developed the statistics of a system of bosons and applied it to the
study of black body radiation and also computed the specific heat of solids. In this section, we
shall deal with a system composed of fermions. We will begin with equation 2.5:

gs !
W =Π (2.5)
s ns !(gs − ns )!

Using the Stirling approximation, we get:

X gs !
lnW = ln
s
ns !(gs − ns )!
X (12.1)
= [gs lngs − ns lnns − (gs − ns )ln(gs − ns )]
s

Similar to how we derived the M-B distribution and B-E distribution, we will now maximize
lnW .

X ∂lnW
dlnW = dns
s
∂ns
 
X ns gs − ns
= − − lnns + + ln(gs − ns ) dns (12.2)
s
ns gs − ns
X  gs − ns 
= ln dns
s
n s

We then get the most probable configuration of the system as:

 
X gs − ns
ln + α + βs dns = 0, (12.3)
s
ns

so we have:

1
  
gs − ns
ln + α + βs = 0
ns
gs − ns
= e−(α+βs )
ns (12.4)
gs
= e−(α+βs ) + 1
ns
gs
ns = −(α+β )
e s + 1

This is the distribution function for a system of fermions. Usually the term,

1
= f (s ) (12.4a)
e−(α+βs ) +1

is known as the Fermi function. Hence we can write:

ns = gs f (s ) (12.5)

That means that the Fermi function gives the probability that an energy level s be occupied
by fermions. In equation 12.4, replacing:
1
eα = F /(kT ), β = − kT , where F is called the Fermi energy. We get:

gs
ns = (s −F )
(12.6)
+1
e kT

The Fermi energy

The Fermi energy is an important concept in the solid state physics of metals and
superconductors. It is also a very important to nuclear physics and to understanding the stability
of white dwarf stars against gravitational collapse.
To find the ground state of the whole system, we start with an empty system, and add
particles one at a time, consecutively filling up the unoccupied stationary states with the lowest
energy. When all the particles have been put in, the Fermi energy is the kinetic energy of the
highest occupied state.
Examples of fermions are: electrons, neutrons and protons.
In order to appreciate the physical significance of the Fermi energy F , let us substitute for
α and β in the Fermi function given by equation 12.4a. That is:
2
 1
f (s ) = (12.7)
e−(α+βs ) +1

becomes:

1
f (s ) = (s −F )
(12.8)
e kT +1

1
f () = (−F )
(12.8a)
e kT +1
At absolute zero temperature, let the value of F be F (0). Then,

1
f () = [−F (0)]
(12.9)
e kT +1
Let us evaluate f () for:

(i)  > F (0). Then

1
f () = =0
e∞ +1

(ii)  < F (0). Then

1
f () = =1
e−∞ +1

This means that at T = 0, absolute zero temperature, the probability that an energy level
with energy less than F (0) be occupied is unity. That is , at T = 0 everytime all en-
ergy levels below F (0) must be occupied. Also, the probability of occupation is zero for
levels having energy more than F . That is, at T = 0 , we do not expect a particle
at an energy level higher than at F . This all means that at absolute zero temperature,
all the levels up to F (0) are occupied, those having energy more than F (0) are empty. As
the temperature is increased, some fermions gain enough energy to be excited into higher lev-
els. Most of the excitations occur to those partciles closest to F (0). Therefore, some levels
having energy slightly lower than, but close to F (0) get depleted of the particles, and some
unoccupied higher levels close to F (0) become occupied. The figure on the board shows the
Fermi function at T = 0 and at T > 0. Note that as the temperature increases, the probability
of occupation of a level with energy less than F (0) decreases while that of a level having en-
ergy more than F (0) increases. The probability of occupation of an energy level at F (0) is half.
3
 To understand the physical meaning of the Fermi level F (0), let us assume that there is N
fermions to be placed in s energy levels. Due to Pauli”s exclusion principle, not more than one
fermion can occupy an energy state. Thus if gs is the degeneracy of an energy level, that level
can be occupied by gs fermions only. The remaining ones should go to the higher levels. If we
continue placing all the N fermions this way, at absolute zero temperature, then the highest
level occupied is the Fermi energy and the corresponding energy level is called the Fermi level.
Clearly, all the levels above Fermi level must be empty. Thus Fermi energy is defined as the
energy of the highest occupied level of a system of fermions at absolute zero temperature. Bosons
do not have such a restriction. We may observe some discrepancy in our definition of Fermi
energy. In equation 12.6 we called F the Fermi energy, and later defined F (0) as the Fermi
energy. This is because at temperatures of interest, the Fermi energy F at temperature T is
nearly equal to the Fermi energy F (0) at absolute zero. Fermi energy varies with temperature
according to the relation:

2
π2

T
F = F (0) 1 −
12 TF

where TF is called the Fermi temperature. (For proof see A.J. Pointon, Introduction to
Statistical Physics, Longman, London, 1967, Appendix VII). T /TF is negligibly small so that
F ≈ F (0). For the same reason, we shall not distinguish between the two henceforth. Let us
now derive an expression for the Fermi energy.
Replacing summation by integration, equation 12.5 becomes:

n()d = f () g()d (13.0)

Since,

X Z ∞
N= ns = n()d
s 0

This implies that:

Z ∞
N= f () g()d (13.1)
0

4
 For fermions, the maximum value of  = F , and for  < F , f () = 1. Therefore:

Z F
N= g()d (13.2)
0

We consider the density of states, but fermions have two states per energy level corresponding
to spin + 21 and − 21 , so we will need to double it. Hence we get the density states for fermions
as:

2π 3 1
g()d = V (2m) 2  2 d (13.3)
h3

Thus:

3
Z F
4πV (2m) 2 1
N=  2 d
0 h3
3 3
4πV (2m) 2 F2 (13.4)
=
h3 3/2
3
8πV (2m) 2 23
= F
h3

Therefore:

2
h2

3N 3
F = (13.5)
2m 8πV

The magnitude of the Fermi energy is ≈ a few electron volts.

The Fermi temperature

It is a convemient way of comparing the Fermi energy with the thermal energy kT . The
relationship is given as:

F
TF = (13.6)
k

Equation 13.5 shows that the Fermi energy depends on the mass m and the particle density
N/V of a system. Hence, it is different for different systems. For metals, the Fermi energy is of
the order of a few electron volts. The average kinetic energy of electron at room temperature
(300 K) is (3/2) kT ≈ 0.04 eV, which is much smaller than the Fermi energy of the metal.
5
Therefore, all the electrons tend to go to the lowest energy states possible, filling each state
with electrons allowed by the Pauli exclusion principle. The metal is then said to be highly
degenerate. This condition can be written as (3/2) kT << F (0) or kT << F (0). This
is responsible for the negligible contribution to the specific heat by the conduction electrons in
metals. On the other hand, the situation differs for molecular gases. For example, in comparison
with an electron gas, a gas of helium isotopes, 32 He (they are fermions), has a very low particle
density and Fermi energy ≈ 0.94 × 10−3 eV . The average energy, 0.04 eV of the isotope is much
more tha the Fermi energy, and hence, they can occupy states having energy more than the
Fermi energy. This makes the exclusion principle invalid because the average occupancy of a
state becomes less than unity. This condition is said to be non-degenerate. if the average energy
of particles is much more than the Fermi energy, that is  >> F (0) or kT >> F (0), we can
write  − F (0) >> kT. Thus the Fermi function given by equation 12.8a approaches very close
to the classical expression given by equation 4.0a:

f (s ) = eα+βs (4.0a)

The energy distribution essentially approximates to the classical M-B distribution. Thus, we
see that at high temperatures or low density, the properties of the F-D gas approach those of a
classical perfect gas, and the quantum effects become negligible. Also, we note that for a metal
to become non-degenerate, kT should approach the Fermi energy F (0). Since F (0) ≈ a few
electron volts, T should be very high ≈ 104 − 105 K, a temperature at which a metal would
melt. Therefore at ordinary temperatures, metals are all degenerate.
Classroom Exercise
Potassium has a density of 8.6 × 102 kg/m3 and an atomic weight of 39. Each atom donates
one electron to the metal.

1. Show that the Fermi energy = 2.05 eV.

The Fermi energy of such a system of fermions is given by:

2
h2

3N 3
εF =
2m 8πV

6
 N (6.02 × 1023 atoms.mol−1 ) × (8.6 × 102 kg.m−3 )
= 39


−1
V 1000 kg.mol

= 1.33 × 1028 m−3

Therefore:

2
(6.63 × 10−34 j.s−1 )2

3 −3 3
εF = × × 1.33 × 1028m
2(9.11 × 10−31 kg) 8π

= 3.28 × 10−19 J = 2.05eV

2. Show that the Fermi temperature TF = 2.38 × 104 K

εF
TF =
k

For the metal in (a)(i):

3.28 × 10−19 J
TF = = 2.38 × 104 K
1.38 × 10−23 J.K −1

7
Application of F-D statistics to semi-conductors
We will now apply F-D statistics to study the electronic properties of semiconductors. This class
of materials are electronic conductors whose resistivity at room temperature lies between that
of conductors (≈ 10−4 to 109 Ω.m) and that of insulators (≈ 1012 to 1020 Ω.m). The resistivity
of semiconductors is strongly dependent on temperature. Resistivity refers to how stringly
a material resists electric current or how strongly a material opposes the flow of
current. At absolute zero temperature, a pure semiconductor acts as an insulator. This can be
explained on the basis of the band theory of solids according to which when a solid is formed
by bringing several atoms together, the electronic energy levels of individual atoms spread into
energy bands. For example, 1s electrons form the 1s band. This is because the Pauli exclusion
principle does not allow the electrons to be crowded into a single level, rather they are placed
in several closely spaced levels forming a band as shown in the figure on the board.
Electrons are arranged in these energy bands separated by gaps called energy gaps, in which
no electron energy states are allowed. If the total number of electrons in a solid is such that
the allowed bands are either completely filled or empty, that is to say that the electrons in the
atom fill the energy levels fully, leaving no free electrons responsible for electrical conduction,
the solid becomes an insulator. If one or more bands are partially full, the solid then becomes
a good conductor. However, in some insulators the energy gap is small enough such that at
ordinary temperatures a few electrons acquire sufficient heat energy to cross the gap between
the highest filled band called the valence band (V.B) and the lowest empty band called
the conduction band (C.B). These electrons are free to take part in electrical conduction
and such solids are called semiconductors. When an electron moves from the valence band to
the conduction band it leaves behind a vacant site called a hole. Electrons in the conduction
band and holes in the valence band equally contribute to conductivity of these solids. In a
pure semiconductor, this type of conductivity is known as intrinsic conductivity. A peculiar
characteristic of semiconductors is that their conductivity can be increased by deliberate addition
of impurities, a process called doping. Increase in conductivity due to doping is known as
extrinsic conductivity.

8
 Extra reading for interest
We shall now calculate the intrinsic conductivity of a pure semiconductor. Let Ne be the
number of electrons per unit volume in the conduction band and Nh the number of holes per
unit volume in the valence band at tenperature T respectively called the density of states in the
conduction bands. At equilibrium, these two are equal, i.e. Ne = Nh . The conductivity of a
semiconductor is given as:

σ = e(Ne µe + Nh µh ) (13.7)

where e is the electronic charge and µe and µh are the mobilities of electrons and holes. To
find the values of Ne and Nh we can use F-D statistics. Let c be the energy at the bottom of
the conduction band and v the energy of an electron at the top of the valence band.

Figure 1

9
 Figure 2

For easy calculation, let us assume that the energy difference ( − F ) >> kT , the thermal
energy. Therefore for electrons in the conduction band, the Fermi-Dirac distribution function of
equation 12.8a becomes:

1
fe () = (−F )
(13.8)
e kT +1
which reduces to

(F −)
fe () = e kT (13.9)

The density of states for a system of fermions per unit volume is given by:

4π 3 1
g()d = 3
(2m) 2  2 d (14.0)
h

In this analysis, let the energy at the top of the valence band be zero and the energy of an
electron in the conduction band be . From the figure the energy of this electron measured from
the bottom of the conduction band is ( − g ) and the energy of a hole in the valence band is
−. Therefore, for electrons in the conduction band  must be replaced by ( − g ) in equation
4.3. Thus we have the density of states for electrons,

4π 3 1
g()d = 3
(2m) 2 ( − g ) 2 d (14.1)
h
10
 Hence the number of electrons per unit volume in the conduction band,

Z ∞
Ne = fe () g()d
0
3
Z ∞
4π(2m) 2
(F −) 1
= e kT
3
( − g ) 2 d (14.2)
g h
3
4π(2m) 2 F ∞
Z
1 
= 3
e kT ( − g ) 2 e− kT d
h g

If we let x = ( − g )/kT , i.e. xkT =  − g ; dxkT = d. Therefore:

3 Z ∞
4π(2m) 2 F 1 xkT +g
Ne = e kT (xkT ) 2 e−( kT
)
kT dx
h3 0
3 Z ∞
4π(2m) 2 3  −
( FkT g ) 1
= (kT ) 2e x 2 e−x dx (14.3)
h3 0
 3
2πmkT 2 ( F −g )
=2 e kT
h2

The term before the exponential is called the effective density of states in the C.B.
Let us now take the holes Nh , in the valence band. If the probability of a level at the top of
the valence band being occupied by an electron is fe (), then the probability that it is occupied
by a hole, fh (), is 1 − fe (). thus for ( − F ) >> kT we have:

1
fh () = 1 − (−F )
e kT +1
1
= (−F )
(14.4)
e kT +1
(−F )
=e kT

The density of states per unit volume for holes is:

4π 3 1
gh ()d = 3
(2m) 2 (−) 2 d (14.5)
h

Therefore, the number of holes per unit volume in the valence band is given by:

11
 Z ∞
Nh = fh () gh ()d
0
 3 (14.6)
2πmkT 2 F )
−( kT
=2 e
h2

Verify this using the same method for electrons. in the expressions for Ne and Nh , equations
14.3 and 14.6, the mass (m) should be replaced by me and mh , the effective masses of emectrons
and holes, respectively. Effective mass is the mass of the electron or hole while isnide a lattice
and is different from the gravitational mass. In an intrinsic semiconductor, Ne = Nh . Hence we
get:

 
1 3 mh
F = g + kT ln
2 4 me

If we assume that the effective masses are equal, then, F = (1/2)g . that is, the Fermi level
is in the middle of the energy gap. Thus:

−g
Ne = Nh = constant.e 2kT

This shows that the electron and hole concentrations and hence the electrical conductivity
σ of a pure semiconductor depend exponentially on g /2kT . this can be used to find the band
gap of a semiconductor from the σ versus T data.

12
Table of physical constants

QUANTITY SYMBOL VALUE

Speed of light c 3.00 × 108 m.s−1

Plank’s constant h 6.63 × 10−34 J.s

Boltzmann constant k 1.38 × 10−23 J.K−1

Electronic charge e 1.61 × 10−19 C

Mass of electron me 9.11 × 10−31 kg

Mass of proton mp 1.67 × 10−27 kg

Gas constant R 8.31 J.mol−1 .K−1

Avogadro’s number NA 6.02 × 1023

Bohr magneton µB 9.27 × 10−24 J.T−1

Permeability of free space µ0 4π × 10−7 H.m−1

Stefan constant σ 5.67 × 10−8 W.m−2 .K−4

Atmospheric pressure 1.01 × 105 N.m−2

Mass of 42 He atom 6.65 × 10−27 kg

Mass of 32 He atom 5.11 × 10−27 kg

Volume of an ideal gas at STP 22.41 mol−1

13