Electronic states in confined systems and low

dimensions
Ion Errea

Abstract
In this topic we will analyze the properties of electrons in confined
systems and low dimensions mainly making use of the free electron ap-
proximation. We will first review the general properties of the electronic
system; second, the independent electron approximation in 1D, 2D, and
3D; third, the role of confinement in different directions; and, finally, the
role of interactions and the the effective mass approximation in semicon-
ductors.

Contents
1 Electrons in solids 2
1.1 Separable potentials and independent electrons . . . . . . . . . . 3

2 Free electrons 4
2.1 Free electrons in 1D . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 Free electrons in 2D . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.3 Free electrons in 3D . . . . . . . . . . . . . . . . . . . . . . . . . 9

3 Density of states of the free electron gas 10

3.1 3D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
3.2 2D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.3 1D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

4 Confinement 13
4.1 Electrons in a 1D box . . . . . . . . . . . . . . . . . . . . . . . . 13
4.2 2D electron gas confined in one direction . . . . . . . . . . . . . . 14
4.3 1D electron gas confined in two directions . . . . . . . . . . . . . 16
4.4 Electrons in 0D: confinement in three directions . . . . . . . . . . 17

5 Crystal structure and effective mass approximation in semicon-

ductors 17
5.1 Bloch’s theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
5.2 The effect of the lattice on the band structure . . . . . . . . . . . 19
5.3 Effective mass in semiconductors . . . . . . . . . . . . . . . . . . 20

1
6 Exercises 23

1 Electrons in solids
All the properties of solids are derived from the Coulomb interactions that couple
the set of ions and electrons that form a solid or, in general, any material. If
we forget about spin degrees of freedom, the Hamiltonian that couples all the
electrons and ions is the following:
X −~2 1X e2 X ZI e2 X −~2 1 X e2 ZI ZJ
H= ∇2i + − + ∇2I + .
i
2m 2 ij |ri − rj | |ri − RI | 2MI 2 |RI − RJ |
iI I IJ
(1)
Here lowercase indexes denote electrons and uppercase indexes ions, m is the
electron mass, MI the mass of ion I with charge eZI , e is the electron charge,
ri is the position vector of the electron i, and RI is the position vector of the
ion I. We are assuming that 4π0 = 1. Diagonalizing this many-body problem
is too complex.
If we assume that ions are static and we neglect the kinetic energy of the
ions, we can focus first on the electronic part of the problem. This is equivalent
to assuming that ions are static particles that stay fixed at their R = {RI }
positions. In this case, the Hamiltonian to be solved can be written as

H = Te + Vee + Vext , (2)

where
X −~2
Te = ∇2i (3)
i
2m
is the kinetic energy of the electrons,
1X e2
Vee = (4)
2 ij |ri − rj |

is the Coulomb interaction between the electrons, and

X ZI e2
Vext = − (5)
|ri − RI |
iI

is the external potential that acts upon the electrons, which in this case is the
Coulomb interaction between the electrons and the ions. We could also consider
any other external potential that affects the electronic system, for instance, the
potential imposed by an electromagnetic probe (Raman, Infrared, etc.) or the
one imposed by an impurity. Note that in Eq. (2) we have neglected the
1 e2
P
2 IJ |RI −RJ | Coulomb interaction between the ions. We can safely do that
because this term simply adds a constant energy to the electronic system. This
term cannot be disregarded, however, if we want to study the dynamics of the
ions, i.e., lattice vibrations (phonons).

2
 In order to calculate the properties of the electronic system we need to solve
the Schrödinger equation of the electronic Hamiltonian in Eq. (2):

H(r)ψA (r) = EA ψA (r). (6)

Here r = {ri } is a collective variable that describes the positions of all electrons.
This problem is unsolvable because the Vee potential is inseparable.

1.1 Separable potentials and independent electrons

The Vee potential is inseparable because it cannot be written as a sum of po-
tentials in which each addend only affects one electron. The denominator in
Eq. (4) makes it impossible as it involves the positions of two electrons. If we
neglect the Vee term in the potential and the external potential is separable, i.e
it can be written as X
Vext (r) = V (ri ), (7)
i

the problem becomes tractable. An example of a separable external potential is

indeed the electron-ion potential of Eq. (84), since it can be written as in Eq.
(7) with
X −ZI e2
V (ri ) = . (8)
|ri − RI |
I

In this situations in which Vee is neglected and Vext (r) is separable, we are
dealing what independent electrons.
In the independent electron case the Hamiltonian can be written as a sum
of independent Hamiltonians that act upon only one electron:
X X  −~2 
2
H(r) = H(ri ) = ∇ + V (ri ) . (9)
i i
2m i

Let’s assume now that we can write the total wave-function of the whole electron
system ψA (r) as a product of single-electron wave functions:

ψA≡α1 ,··· ,αNe (r) = ψα1 (r1 ) · · · ψαNe (rNe ), (10)

where Ne is the total number of electrons in the system. In this case, if we are
able to solve the single-electron Hamiltonian,
 2 
−~ 2
∇ + V (r) ψα (r) = Eα ψα (r), (11)
2m
it is straightforward to show that ψA≡α1 ,··· ,αNe (r) is an eigenfunciton of the
system and
EA≡α1 ,··· ,αNe = Eα1 + · · · + EαNe (12)
is the eigenenergy of the system. In this sense the independent electrons problem
is easy to tackle as long as the single-electron Schrödinger equation in Eq. (11)
is solvable.

3
 The fact that electrons are indistinguishable fermionic particles implies that
the total wave-function of the system must be antisymmetric to the exchange
of any two electrons:
ψα1 ,··· ,αNe (· · · , ri , · · · , rj , · · · ) = −ψα1 ,··· ,αNe (· · · , rj , · · · , ri , · · · ). (13)
Therefore, the product of single-particle wave-functions in Eq. (11) cannot be
the true wave-function of the electronic system as it is symmetric to the exchange
of electrons. An antisymmetric wave-function can be built out of single-particle
electronic wave-functions instead with the so-called Slater determinant:
ψα1 (r1 ) ··· ψα1 (rNe )
1 .. .. ..
ψα1 ,··· ,αNe (r) = √ . . . (14)
Ne !
ψαNe (r1 ) · · · ψαNe (rNe ).
A Slater determinant is thus the true wave-function of an independent electron
many-body system.
An important conclusion from the antisymmetric character of the electronic
wave-function is that all single-particle wave-functions must have a different
quantum number: αi 6= αj for any i and j. In other words, all electrons must
have a different quantum number. If this is not the case the wave-function
would not be antisymmetric for the exchange of all electron pairs. Therefore,
the ground state energy of the system is formed when electrons fill the lowest
energy states until we run out of electrons:
EGS = E0 + E1 + E2 + · · · + EF . (15)
The highest occupied electronic energy is called the Fermi energy EF . Here we
are assuming that E0 , E1 , . . . are, respectively, the lowest energy of the solution
of Eq. (11), the second lowest energy, etc.
It is important to note that electrons are spin 1/2 fermions and, thus, can be
on spin up (| ↑i) or spin down (| ↓i) states. As we will not consider spin-orbit
coupling nor the presence of magnetic fields, both spin up and down states will
be degenerate. Therefore, the Slater determinant above can be perfectly used
as long as it is assumed that each state is doubly degenerate. If that was not
the case each state should be divided in two different states, i.e., ψαi (ri ) into
ψαi (ri )| ↑i and ψαi (ri )| ↓i.

2 Free electrons
The free electron system is a particular version of the independent electron
system in which there is no external potential. Thus, in the free electron case
the Hamiltonian is simply H = Te . Obviously the Hamiltonian is separable and
everything described in Sec. 1.1 holds. The Schrödinger equation that needs to
be solved is thus the following:
−~2 2 −~2 ∂ 2 ∂2 ∂2
 
∇ ψα (r) = + 2 + 2 ψα (r) = Eα ψα (r). (16)
2m 2m ∂x2 ∂y ∂z

4
In the following we will solve this simple differential equation for the 1D, 2D,
and 3D cases.

2.1 Free electrons in 1D

In 1D, assuming that the one-dimensional electron gas expands in the x direc-
tion, Eq. (16) simply becomes:

−~2 d2 ψα (x)
= Eα ψα (x). (17)
2m dx2
This equation is trivially solved with plane waves:

ψα (x) → ψk (x) = eikx . (18)

Plugging in the plave-wave solution into the Schödinger equation we obtain the
dispersion relation of the 1D free electron gas:

~2 k 2
Ek = . (19)
2m
We see, therefore, that the wave number k is the good quantum number for the
free electron gas and it determines the wave function.
The dispersion relation, the relation between the wave number and the en-
ergy of the state, shows a parabolic dependence as shown in Fig. 2.1. The
ground state of the 1D free electron gas would therefore be formed by occupy-
ing all the electronic states up to the Fermi energy EF . The largest occupied
wave number, the quantum number associated to the states with energy EF , is
called the Fermi wave number kF . In the 1D case, the ground state of the free
electron gas is thus formed by occupying all the states from −kF to kF .
It is customary to assume that the free electron gas is formed by wire of
macroscopic length L, which is much larger than the Fermi wavelength λF =
2π/kF . The plane wave solutions are normalized to this length as
1
ψk (x) = hx|ki = √ eikx , (20)
L
where we are introducing Dirac’s notation. Then, the eigenfunctions are or-
thonormal,
Z L Z L Z L
0 0 ∗ 1 0
hk|k i = dxhk|xihx|k i = dxψk (x) ψ (x) =
k0 dxei(k−k )x = δkk0 ,
0 0 L 0
(21)
as it must be the case for any eigenfunctions of a Hamiltonian.
It is illustrative to say that the probability density defined by the plane-wave
solution of the free electron gas is homogeneous:

|ψk (x)|2 = 1/L. (22)

5
Ek

EF

0
kF 0 kF
k
Figure 1: Dispersion relation of the 1D free electron gas.

This means that the probability to find a free electron is the same at any point
of space.
The plane wave solutions of the free electron gas, interestingly, are eigenval-
d
ues of the momentum operator p = −i~ dx :

1
pψk (x) = −i~ik √ eikx = ~kψk (x). (23)
L
Thus, the momentum of a plane wave with wave number k is pk = ~k. This
allows to define the Fermi momentum as pF = ~kF as well as the Fermi velocity
with this classical analogy:
~kF
vF = . (24)
m
Even if this electron gas is finite in size, we do not want to consider surface
effects at the moment, but just focus on its bulk properties. In order to suppress
the surface, periodic boundary conditions are assumed:

ψk (x) = ψk (x + L). (25)

This implies that not all wave numbers are allowed: kL = 2πn, where n is an

6
integer. Therefore the allowed wave numbers are
2πn
k= . (26)
L
Considering that L is of a macroscopic size the states can be assumed to form a
continuous. We see that there is one state (doubly degenerate due to spin) per
2π/L wave number length.
The number of allowed states due to periodic boundary conditions deter-
mines the relation between the electronic density of the system (Ne /L in this
1D case) and the Fermi wave number. The total number of electrons can be
calculated simply counting the number of allowed states in the range going from
−kF to kF :
1
Ne = 2 × 2kF × . (27)
2π/L
The first 2 comes from the spin degeneracy. Thus, the electronic density of the
1D free electron gas is
N 2
n= = kF . (28)
L π
The Fermi energy is therefore also exclusively dependent on the density:

~2 π 2 n 2
EF = . (29)
8m
So far we have assumed that the electronic system is at 0 K. In this limit all
states are occupied up to the Fermi level and unoccupied above. However, at
finite temperature this is not exactly true as states are occupied following the
Fermi-Dirac distribution function
1
fF (ε) = ε−µ , (30)
e kB T
+1
where kB is Boltzmann’s constant and µ the chemical potential, whose limit at
0 K is EF . The Fermi-Dirac distribution function is shown in Fig. 2.1.

2.2 Free electrons in 2D

In the 2D case, a similar procedure (Exercise 1) as the one followed for the 1D
case in Sec. 2.1 yields the wave functions of the free electron gas of a macroscopic
area A:
1
ψk (r) = hr|ki = √ eik·r . (31)
A
Here, we assume that the 2D electron gas expands in the x and y directions,
thus, r = (x, y) and k = (kx , ky ). Again the wave number k is a good quantum
number. The eigenenergy for each state is

~2 k 2
Ek = , (32)
2m

7
 1 T=0
T 0
kB T
fF ( )

0
0

Figure 2: Fermi-Dirac distribution function at different temperatures.

where k 2 = kx2 + ky2 . The dispersion describes, thus, a paraboloid.

It is useful to define at this stage the Fermi line, which is the geometric line
that is defined by the states at the Fermi level:

kF2 = kx2 + ky2 . (33)

We can see trivially that in the 2D free electron gas the Fermi line is a circle.
Periodic boundary conditions in this case impose that

ψk (x, y) = ψk (x + Lx , y) (34)
ψk (x, y) = ψk (x, y + Ly ), (35)

where Lx and Ly are the macroscopic dimensions of the 2D electron gas, which
is assumed to be a square with area A = Lx Ly . Consequently, the allowed wave
vectors need to fulfill the following conditions:
2πnx
kx = (36)
Lx
2πny
ky = , (37)
Ly

8
with nx and ny integers. This means that for the area occupied by each k state
(doubly degenerate due to spin) in the (kx , ky ) space is (2π)2 /A. Considering
this feature it is trivial to show that for 2D electron gas with electronic density
n = Ne /A, the Fermi wave number depends on n as
√
kF = 2πn (38)
and the Fermi energy as
~2 πn
EF = . (39)
m

2.3 Free electrons in 3D

In the 3D case the results are obtained following a similar procedure again
(Exercise 2). The 3D electron gas is contained in a box of size Lx , Ly , and Lz
whose volume is V = Lx Ly Lz . Its orthonormal wavefunctions, normalized in
this volume, are:
1
ψk (r) = hr|ki = √ eik·r . (40)
V
The eigenenergy for each state is
~2 k 2
Ek = , (41)
2m
where k 2 = kx2 + ky2 + Kz2 . The dispersion describes again paraboloid but in
higher dimensions that cannot be visualized.
In 3D instead of the Fermi line, the Fermi surface is defined. The Fermi
surface is formed by all the k points that are at the Fermi level. In the case of
the 3D free electron gas it is given by
kF2 = kx2 + ky2 + kz2 , (42)
which is exactly the equation of a sphere. Thus, the Fermi surface of a 3D free
electron gas is a sphere.
Periodic boundary conditions in this case impose that
ψk (x, y, z) = ψk (x + Lx , y, z) (43)
ψk (x, y, z) = ψk (x, y + Ly , z) (44)
ψk (x, y, z) = ψk (x, y, z + Lz ), (45)
(46)
The allowed wave vectors need to fulfill the following conditions:
2πnx
kx = (47)
Lx
2πny
ky = (48)
Ly
2πnz
kz = (49)
Lz

9
with nx , ny , and nz integers. This means that for the volume occupied by each
k state (doubly degenerate due to spin) in the (kx , ky , kz ) space is (2π)3 /V .
Considering this feature it is trivial to show that for the 3D electron gas with
electronic density n = Ne /V , the Fermi wave number depends on n as

kF = (3π 2 n)1/3 (50)

and the Fermi energy as

~2
EF = (3π 2 n)2/3 . (51)
2m

3 Density of states of the free electron gas

In Sec. 2 we have studied the free electron gas and we have seen that indeed there
are some differences depending on the dimension. An important magnitude that
is useful to calculate properties of compounds is the density of state (DOS),
which, as we will see, has clear differences for the free electron gas depending
on the dimensionality. The DOS g(ε) measures the number of available states
for a given energy ε. The DOS is normally defined as the number of states per
volume, per area, per length, respectively, for the 3D case, 2D case, and 1D
cases.

3.1 3D
Following this definition, the DOS should be calculated in the 3D case as
1 X
g(ε) = 2δ(ε − Ek ), (52)
V
k

where the factor 2 comes from the spin degeneracy and Ek is the dispersion
relation. The calculation can be performed rather straightforwardly substituting
the sum with an integral:
Z
X 1
→ dk. (53)
(2π)3 /V
k

This is possible since the states are very close together in k space and each state
occupies a volume in this space of (2π)3 /V . Thus,
Z ∞
~2 k 2 ~2 k 2
Z    
1 1 2
g(ε) = dk2δ ε − = dkk δ ε − . (54)
8π 3 2m π2 0 2m

Making use of the property of the Dirac delta that states that
X δ(x − xi )
δ(f (x)) = , (55)
i
|f 0 (xi )|

10
where xi are the poles of f (x),
" r ! r !#
~2 k 2
  r
m 2mε 2mε
δ ε− = δ k− +δ k+ . (56)
2m 2~2 ε ~2 ~2

With this the DOS becomes of the 3D electron gas becomes

m √
g(ε) = 2mε. (57)
π 2 ~3
The DOS is useful to calculate several properties that depend on the energy.
For example, if we want to calculate the energy density of the 3D electron gas we
should simply sum all energies of the electrons and include the spin degeneracy
for all occupied states according to the Fermi-Dirac distribution function:
1 X
u= 2Ek fF (Ek ). (58)
V
k

This can be calculated rather easily with the DOS since

Z ∞ Z ∞
1 X
u= dε 2εfF (ε)δ(ε − Ek ) = dεεg(ε)fF (ε). (59)
0 V 0
k

For the 3D case at 0 K, limit in which the Fermi-Dirac distribution function

is a step function, substituting Eq. (57) into the equation above, it is easy to
obtain that the energy density is
√
m 2m 2 5/2
u= E . (60)
π 2 ~3 5 F

3.2 2D
In the 2D limit the DOS is calculated as the number of possible states per area:
1 X
g(ε) = 2δ(ε − Ek ). (61)
A
k

The calculation should be performed substituting the sum with an integral,

which in the 2D case should be performed as
Z
X 1
→ dk, (62)
(2π)2 /A
k

as the area occupied per state is (2π)2 /A. It is rather straightforward to show
that
m
g(ε) = (63)
π~2
in the 2D limit.

11
 1.0
1D
2D
3D
0.8
g( ) (arb. units)

0.6

0.4

0.2

0.0
0 2 4 6 8 10
(arb. units)
Figure 3: DOS of the free electron gas in different dimensions.

3.3 1D
In the 1D limit the DOS should be calculated as the number of possible states
per length:
1X
g(ε) = 2δ(ε − Ek ). (64)
L
k

The calculation should be performed substituting the sum with an integral,

which in the 1D case should be performed as
Z
X 1
→ dk, (65)
(2π)/L
k

as the length occupied per state is (2π)/L. It is rather straightforward to show

that r
1 2m
g(ε) = (66)
π~ ε
in the 1D limit.
Comparing Eqs. (57), (63), and (66) we can see that the dependence on
energy of the DOS is extremely different depending on the dimension. In the
3D case the number of allowed states increases with the energy, in the 2D limit

12
 16

2 2/(2mL 2)]
1(x) 9
2(x)
3(x)

En[
4

0 L 1 2 3 4
x k[ n/L]
Figure 4: φn (x) wave functions for the electrons in a 1D box of length L (left
panel). Discretized dispersion relation compared to the quadratic dispersion of
the free electron gas (right).

it is constant, and in the 1D limit it decreases with energy. The three different
curves are compared in Fig. 3.3.

4 Confinement
We have thus far studied the free electron gas in different dimensions. However,
even if the free electron gas can be realized in low dimensions, for instance in
interfaces, it is more common to find systems that are confined in one or several
of the Cartesian directions. We will study now the effect of confinement on the
electronic states, but first we will overview the simple example of 1D electrons
confined in a 1D box.

4.1 Electrons in a 1D box

Let’s assume we have a 1D system confined in the space that goes from x = 0
to x = L. This is obtained by setting the potential to be 0 in this region, and
infinity outside. The equation that we need to solve is again Eq. (17), but

13
now the boundary conditions are different: the wave functions need to vanish
at the edges of the box: φα (0) = φα (L) = 0. As the Schödinger equation is a
second-order equation, the most general solution can be written as
φα (x) = Aeikx + Be−ikx (67)
considering that k is positive. It is easy to show that in order to satisfy these
conditions the wave number must be quantized as
πn
kn = , (68)
L
where n is in this case a natural number (1,2,3,. . . ). This result immediately
yields to the wavefunctions
r
2  πn 
φn (x) = sin x (69)
L L
and the corresponding eigenvalues
~2 kn2 ~2 π 2 2
En = = n . (70)
2m 2mL2
The dispersion relation and the eigenfunctions are shown in Fig. 4.1. As we can
see confinement discretizes the energies and momentum available, contrary to
the free electron case, where both the energies and the available wave numbers
are continuous. The discretization is larger for smaller L. In fact, in the L → ∞
limit, the energy jumps vanish and the wave functions become a plane wave.

4.2 2D electron gas confined in one direction

Let’s assume now a set of electrons in 3D space, confined in the third direction.
Let’s assume that the electrons live in a box of dimension Lx × Ly × Lz , but
while Lx and Ly are large, i.e. of macroscopic size, Lz is smaller than the Fermi
wavelength. This means that while in the xy plane electrons are not confined,
they are strongly confined in the z direction. This is what we call a quantum
well. The simplest approximation we can do to analyze this system is to assume
that electrons are free, i.e. they satisfy Eq. (16), but with periodic boundary
conditions in the xy plane, while in the z direction the wave function must
vanish at z = 0 and z = Lz . Considering that the latter are the boundary
conditions of the electrons in a box, the solution of the Schrödinger equation
will be a product of the 2D free electron gas and the solution of the electrons
in a box: r  
1 i(kx x+ky y) 2 πn
φkx ,ky ,n (r) = √ e sin z (71)
A Lz Lz
where n is a natural number. The eigenvalues, or the dispersion relation, will
be the following in consequence:
~2 ~2 π 2 2
kx2 + ky2 +


Ekx ,ky ,n = n . (72)
2m 2mL2z

14
Figure 5: Ekx ,ky ,n energy dispersion of free electrons confined in the z direction.
The dispersion is formed by parabolloids.

We can see that the dispersion relation is formed by a parabolloid for each value
of the quantum number n associated with the confined direction.
With this dispersion relation, the DOS (per volume) of the 2D electron gas
confined in the z direction will be
1 X ~2 k 2
g(ε) = 2δ(ε − − En ), (73)
V 2m
kn

~2 π 2 2
where we have assumed that in this case k = (kx , ky ) and En = 2mL2z n . This
can be calculated as follows:
~2 k 2
Z
1 1 X
g(ε) = 2
dk2δ(ε − − En )
Lz (2π) n 2m
 q 
2m
1 X ∞ k δ k − ~2 (ε − En )
Z
= dk q , (74)
Lz n 0 π ~2 2m
(ε − E )
m ~ 2 n

15
which simplifies to
1 m X
g(ε) = θ(ε − En ). (75)
Lz π~2 n
As we can see the DOS is formed by different steps that increase the DOS each
time an energy En is reached.

4.3 1D electron gas confined in two directions

Now we assume that the confinement is both in y and z direction, i.e., Ly , Lz <
λF . We are talking about a quantum wire. Again the simplest approximation
is to assume that we have free electrons, with periodic boundary conditions
along x, but the box-like boundary conditions along y and z. In analogy, the
eigenfunctions will be
   
1 ikx 2 πn πm
φk,n,m (r) = e p sin y sin z , (76)
Lx Ly Lz Ly Lz

where both n and m are natural numbers. The eigenenergies, or the dispersion
relation, is thus
~2 k 2 ~2 π 2 n2 m2
 
Ek,n,m = + + 2 . (77)
2m 2m L2y Lz
The dispersion relation shows parabolas for each value of n and m.
Let’s calculate the DOS of the 1D electron gas confined in y and z:

1 X ~2 k 2
g(ε) = 2δ(ε − − En,m ), (78)
V 2m
k,n,m
 
~2 π 2 n2 m2
where En,m = 2m L2y + L2z . This is equal to

1 X ∞ dk
Z
1
g(ε) = q
Ly Lz n,m −∞ π ~ 2 (ε − E )
m n,m
" r ! r !#
2m 2m
× δ k− (ε − En,m + δ k + (ε − En,m , (79)
~2 ~2

which leads to the final result:

√
1 2m X 1
g(ε) = p θ(ε − En,m ). (80)
Ly Lz π~ n,m ε − En,m

As we can see, the DOS in this case will have singularities at ε − En,m , and will
decrease as ε−1/2 out of it.

16
4.4 Electrons in 0D: confinement in three directions
Let’s assume now that the confinement is in the three Cartesian directions so
that Lx , Ly , Lz < λF . This is what we call a quantum dot. In this case the
simplest approximation in which we can describe the electrons in this system
is imposing the 3D box boundary conditions on the 3 Cartesian directions to a
free electron gas. In analogy with the 1D box, the eigenfunctions of this system
are √      
2 2 πn πm πl
φn,m,l (r) = p sin x sin y sin z , (81)
Lx Ly Lz Lx Ly Lz
and the eigenenergies
~2 π 2 n2 m2 l2
 
En,m,l = 2
+ 2 + 2 . (82)
2m Lx Ly Lz
n, m, and l are natural numbers. As we can see the eigenenergies form a discrete
spectrum that recalls atomic states, clearly different to extended systems like
those described above. The DOS is therefore simply
1 X
g(ε) = 2δ(ε − En,m,l ), (83)
V
n,m,l

a collection of Dirac delta functions at the eigenenergies.

In Fig. 4.4 we compare the DOS calculated for the different levels of con-
finement studied.

5 Crystal structure and effective mass approxi-

mation in semiconductors
All the properties we have described so far are based on the free electron system.
We have just basically solved the Schrödinger equation Eq. (16) with different
boundary conditions, periodic in the extended case and box-like in the confined
case. However, electrons interact in solids with the periodic lattice potential
X ZI e2
Vext (r) = − . (84)
|ri − RI |
iI

Indeed, the potential is periodic with the lattice periodicity

Vext (r + R) = Vext (r), (85)
where R is a lattice vector:
R = n1 a1 + n2 a2 + n3 a3 . (86)
Here n1 , n2 , and n3 are integers, while a1 , a2 , and a3 are the unit cell vectors.
Obviously, this potential will affect the electrons and deviate the solution from
the free electron description we have thus far studied.

17
• DOS in 3, 2,1D

Quantum Confinement and Dimensionality

25

Figure 6: Comparison of the DOS of the 3D free electron gas with the one with
confinement to 2D (quantum well), to 1D (quantum wire), and to 0D (quantum
dot).

Before starting with the effects of the lattice potential on electrons, let’s
recall what the reciprocal lattice is. The basis vectors of reciprocal lattice are
a2 × a3
b1 = 2π (87)
a1 · a2 × a3
a3 × a1
b2 = 2π (88)
a1 · a2 × a3
a1 × a2
b3 = 2π . (89)
a1 · a2 × a3
Any reciprocal lattice vector G can be written as

G = m1 b1 + m2 b2 + m3 b3 , (90)

with m1 , m2 , and m3 integers. By construction,

bi · aj = 2πδij . (91)

18
Figure 7: Dispersion relation of the 1D free electron gas (black) and the opening
of the gaps due to the interaction with the lattice (red). The lattice has a
periodicity of a and the Brillouin zone goes from k = −π/a to k = π/a.

5.1 Bloch’s theorem

Bloch’s theorem states that for an external lattice-periodic potential that fulfills
Eq. (85), the solution of the
 2 
−~
Hψα (r) = ∇2 + Vext (r) ψα (r) = Eα ψα (r) (92)
2m
Schrödinger equation must be equal to

ψnk (r) = eik·r unk (r), (93)

where unk (r) is lattice periodic, i.e. unk (r+R) = unk (r). These are called Bloch
states and have a plane wave component plus a lattice-periodic part. Note that
the quantum number α has been split in two different quantum numbers, the
wave number k and n. The dependence of the Enk on the wave number creates
the so called band structure of solids, which as a consequence is usually only
plotted for states in the Brilluin zone.
Due to the shape of the Bloch states we see that the quantum number k
can be restricted to the Brillouin zone, as the eigenvalues and eigenfunctions of
k0 = k + G are the same:
Enk = Enk+G . (94)
Thus, it is enough to calculate the eigenenergies for the first Brillouin zone
only. For a 1D lattice with periodicity a we show the band structure of the free
electron gas in Fig. 5.1. As we can see the parabolic band appears folded into
the Brillouin zone (from k = −π/a to k = π/a) creating more than one state
for each k, which yields to a band structure like dispersion Enk .

5.2 The effect of the lattice on the band structure

Let’s assume we want to analyze the effect on the lattice potential on the band
structure. Let’s assume that the lattice potential is weak and can be treated

19
perturbatively. In this limit the first order correction to the the plane wave
quadratic dispersion will be
Z Z
1 −ik·r ik·r 1
hk|Vext |ki = dre Vext (r)e = drVext (r). (95)
V V
As we can see the correction is independent of k, thus, this simply implies a
rigid correction to the energies. This is a completely negligible effect.
However, in order to reach that conclusion we have assumed that there are
no degenerate states. In case, plane-wave states |ki and |k0 i are degenerate
we should diagonalize the hk0 |Vext |ki matrix to see what the correction on the
states is.
Z
1 0
hk0 |Vext |ki = drei(k−k )·r Vext (r)
V
Z
1 X i(k−k0 )·R 0
= e drei(k−k )·r Vext (r)
V unit cell
R
Z
N 0
= δk−k0 ,G drei(k−k )·R Vext (r)
V unit cell
= Vext (G)δk−k0 ,G , (96)

where Vext (G) is the Fourier transform for the reciprocal lattice G. So if the
difference between the wave numbers of two degenerate states equals a reciprocal
lattice vector, the degeneracy will be split and the opened energy gap will be of
the order of Vext (G). In the free electron system this happens at zone border.
At these points, as shown in Fig. 5.1, gaps will be opened.

5.3 Effective mass in semiconductors

Due to the interaction with the lattice (or potentially other type of interactions
as well) gaps are opened in the band structure. It could happen that electrons
occupy all states up to the gap. In this case we have an insulator, as there are no
free electrons to move. If the gap is small (∼1 eV) we have a semiconductor. In
a semiconductor a small number of electrons might be excited to the unoccupied
conduction band from the occupied valence band due to thermal effects or other
reasons. When electrons are excited to the conduction band they leave in top
of the valence bands empty states that effectively have positive charge. These
are called holes. See Fig. 5.3 for a simple sketch of the situation.
The simplest approximation to understand both electrons and holes is to
assume they are free electrons and they have a parabolic dispersion. However,
electrons or holes do not have the mass of electrons m because the width of the
parabola depends on the interaction that opens the gap. This is usually treated
assuming that electrons or holes have an effective mass m∗ , whose dispersion is
then
~2 k 2
Ek = E0 + (97)
2m∗e

20
Figure 8: Conduction and valence bands of a semicondugtor of gap Eg . Some
electrons are excited in the conduction, which leave holes in the valence band.

for electrons and

~2 k 2
Ek = E0 − (98)
2m∗h
for holes (note the dispersion is concave).
Once the effective mass is defined, the DOS for semiconductors with given
effective mass can also be calculated with the equations we have obtained for the
confined free electron gas before. These equations will be valid in the vicinity
of the gap. For the electrons in 3D:
m∗e p ∗
g(ε) = 2me (ε − Ec ), (99)
π 2 ~3
where Ec is the minimum energy of the conduction band. For holes in 3D:
m∗
q
g(ε) = 2 h3 2m∗h (Ev − ε), (100)
π ~
where Ev is the top energy of the valence band. In the case we have a 2D free
electron gas confined in one direction we will have for electrons
1 m∗e X
g(ε) = θ(ε − Ec − En ) (101)
Lz π~2 n

21
and
1 m∗h X
g(ε) = θ(−ε + Ev + En ) (102)
Lz π~2 n
for holes. For the 1D free electron gas confined in two directions we have for
electrons
√ ∗X
1 2me 1
g(ε) = p θ(ε − Ec − En,m ) (103)
Ly Lz π~ n,m ε − Ec − En,m

and for holes

2m∗h X
p
1 1
g(ε) = p θ(−ε + Ec + En,m ). (104)
Ly Lz π~ n,m −ε + Ev + En,m

In the equations above m∗e and m∗h are the effective masses for electrons and
holes, respectively.

22
6 Exercises
1. Show that the wave functions of the 2D free electron gas of macroscopic
area A (with A = Lx Ly , Lx and Ly being the dimensions of the square
that contain the 2D electron gas) are given by Eq. (31) and the dispersion
relation is ~2 k 2 /(2m). Show also the relations between the Fermi energy
and wave number and the electronic densities given in Eqs. (38) and (39).

2. Show that the wave functions of the 3D free electron gas of macroscopic
volume V (with V = Lx Ly Lz , and Lx , Ly , Lz being the dimensions of
the box that contains the 3D electron gas) are given by Eq. (40) and
the dispersion relation is ~2 k 2 /(2m). Show also the relations between the
Fermi energy and wave number and the electronic densities given in Eqs.
(50) and (51).

3. Calculate the energy density (total energy per length) of the 1D electron
gas.
4. Calculate the total number of electrons per unit area of a 2D free electron
gas confined in one direction (quantum well).

5. Calculate the total number of electrons per unit length of a 1D free electron
gas confines in two directions (quantum wire).

23