Band Theory of Solids: Brajesh Tiwari and R. E. Amritkar
Band Theory of Solids: Brajesh Tiwari and R. E. Amritkar
Band Theory of Solids: Brajesh Tiwari and R. E. Amritkar
In this short article we will consider band formation, Bloch theorem and Kronig-Penny Model.
where k is a wave vector and the free electron has the energy
~2 k 2
= (2)
2m
FIG. 1. (a) Plot of energy verses wavevector k for a free electron. (b) Plot of energy verses wavevector k for an electron in
a one dimensional lattice with lattice constant a. The figure shows the first energy gap at k = ±π/a.
In a crystal the above energy spectrum is modified and energy gaps open. One such energy gap at k = ±π/2 is
shown in Fig. 1. The region between −π/a and π/a is called the first Brillouin zone. At the Brillouin zone boundary
(k = ±π/a) the wavefunctions are not traveling waves of the type eikx but are stationary waves which are obtained
by superposing traveling waves to the right and left. Thus the two stationary waves are
ψ (+) = A(eiπx/a + e−iπx/a ) = 2A cos(πx/a)
ψ (−) = B(eiπx/a − e−iπx/a ) = 2Bi sin(πx/a)5 (3)
where A and B are normalization constants. The wavefunctions normalized over one period are (integrate over x = 0
to x = a)
r
(+) 2
ψ = cos(πx/a)
a
r
2
ψ (−) = sin(πx/a) (4)
a
Note that the two wavefunctions ψ (+) and ψ (−) are even and odd functions1 . The probability densities for the two
stationary waves are (see Fig. 2)
2
ρ(+) = |ψ (+) |2 = cos2 (πx/a)
a
2
ρ(−) = |ψ (−) |2 = sin2 (πx/a) (5)
a
From the probability density we see that ψ (+) has larger probability near the ion cores while ψ (−) has higher
probability away from the ion cores. There is an attractive force between the ion cores and electrons and hence we see
that ψ (−) will have a larger potential energy compared to ψ (+) . The difference in the energies of the two wavefunctions
gives the band gap. The wavefunction ψ (+) corresponds to the top of the first energy band while ψ (−) corresponds to
the bottom of the second energy band. The kinetic energy of the two standing waves is the same. (Why?)
1 An even function does not change sign as x goes to −x while an odd function changes sign, i.e. cos(−πx/a) = cos(πx/a) and
sin(−πx/a) = − sin(πx/a)
3
FIG. 2. Probability density of Eqs. (??) verses the position x. Standing wave 1 is ρ(−) and standing wave 2 is ρ(+) .
In the free electron approach we assume that the electrons are free to move inside the solid. However, these electrons
move in the potential produced by the core of the atoms (positive nuclei). Since the atoms in a solid are arranged in
a periodic structure, the potential produced by them will also have the same periodicity. A typical one dimensional
periodic potential in shown in Fig. 3.
Bloch theorem2 is an important theorem for our understanding of the wavefunction of an electron moving in a
periodic potential and the resulting band structure.
FIG. 3. A typical periodic potential produced by the core of atoms in a solid. The electrons move in this potential. Here, a
is the atomic spacing, d is the width of the solid and d/2 and −d/2 are the boundaries of the solid. The electrons can move
inside the solid but require a minimum energy V0 , called work function, to come out of the solid.
2 Felix Bloch (1905-1983) was born in Zurich, Switzerland. Bloch theorem is one of his important contributions. But, he received Nobel
prize in physics in 1952 together with Edward Purcell for their work on nuclear magnetism.
4
A. Periodic potential
Consider an array of N nuclei located at a, 2a, 3a, . . . N a, N a = L. The j-th atom produce a potential at x as
φ(x − ja). Hence the net potential at x is
N
X
V (x) = φ(x − ja) = φ(x − a) + φ(x − 2a) + · · · + φ(x − N a). (6)
j=1
V (x + a) = V (x) (8)
To remove the effect of the boundary terms, one must consider infinite lattice, i.e. take the limit N → ∞. But,
this makes the calculations difficult. To get around this difficulty, Bloch introduced the concept of periodic boundary
conditions (different from lattice periodicity). The wavefunction of an electron satisfies periodic boundary condition
given by
where N is the number of atoms in a chain. This is like putting the atoms on a circle (Fig 5). In a circle N -th atom
and the first atom are neighbors.
C. Bloch Theorem
d2 ψ(x) 2m
+ 2 [E − V (x)]ψ(x) = 0 (10)
dx2 ~
5
with periodic potential (Eq. (8)) and periodic boundary condition (Eq. (9)) has the form
where
uk (x + a) = uk (x)
with
Thus Bloch theorem tells that the wavefunction can be written as a product of two functions, a plane wave, exp(ikx),
and a periodic function, uk (x).
Note that the wavefunction does not have the periodicity of the potential.
Kronig-Penny model (KP model) is a simple approximation to the periodic potential of Fig. 3, which allows an
analytical treatment and leads to energy bands and band gaps.
The potential of KP is given by
0, 0 < x < a
U (x) = (13)
U0 , −b < x < 0
The wave function is a solution of the Schrödinger equation and in the two regions of one period of the potential
the solutions can be written as follows
• Region 1: 0 < x < a
Here U (x) = 0 and the Schrödinger equation is
d2 ψ(x) 2m
+ 2 ψ(x) = 0 (14)
dx2 ~
The general solution can be written as
where A and B are integration constants and K is related to the energy of the particle by (obtained by
substituting ψ in the Schrödinger equation)
~2 K 2
= (16)
2m
6
d2 ψ(x) 2m
− 2 (U0 − )ψ(x) = 0 (17)
dx2 ~
and we choose the energy such that < U0 . Hence the solution can be written as
where C and D are integration constants and Q is given by (obtained by substituting ψ in the Schrödinger
equation)
~2 Q2
U0 − = (19)
2m
To determine the four constants A, B, C, D we require four conditions (see Appendix II).The four equations that are
obtained from these conditions have a solution only if the following condition is satisfied.
where k is the index of the Bloch theorem. Note that p = ~k is the momentum of the particle.
The above condition (20) can be further simplified by taking the limit b → 0 and U0 → ∞ such that Q2 ba/2 = P
is finite3 . Since Q2 b is finite, Qb → 0, i.e. Qb 0 and sinh Qb ' Qb, cosh Qb ' 1. Thus, the condition (20) becomes
Fig. 7 shows the plot of left hand side of this equation (P/Ka) sin Ka + cos Ka as a function of Ka. Since we have
−1 ≤ cos ka ≤ 1, the shaded regions in the figure do not satisfy the condition (21) and those values of K are not
allowed. From Eq. (16), the corresponding values of energy are not allowed. This leads to the band gaps. From
condition (21), For each allowed K value we can calculate the corresponding ka value and thus obtain = ~2 K 2 /(2m)
as a function of ka.
FIG. 7. Plot of (P/Ka) sin Ka + cos Ka as a function of Ka for P = 3π/2. The shaded regions give the band gaps.
Fig. 8 shows the plot of energy vs ka. We see band gaps for ka = nπ, n = 1, 2, . . . . The bands and the band gaps
can be determined by solving
We get multiple solutions for Ka and thus different bands. (See the arrows in Fig. 7.)
(1) The bottom of the first band is given by ka = 0 and (P/Ka) sin Ka + cos Ka = 1, 0 < Ka < π.
3 This is same as the condition that U0 b is finite. In this limit Kronig-Penny potential takes a periodic delta function form.
7
A rough sketch of how bands are formed in a real solid is illustrated in Fig. 9(a) for sodium metal. A general case
of band formation is shown in Fig. 9(b).
(a) (b)
FIG. 9. (a) Energy bands in sodium metal. In sodium metal, 1S, 2S, and 3P bands are fully occupied while 3S band has only
one electron per atom and hence it is half full and half empty. (b) Typical energy bands in a solid. The bands with energies
in the potential wells near each atomic core, are connected across the atoms by tunneling through the potential barriers.
If the top of one band and the bottom of the next band occur at the same momentum or k value then the band
gap is known as a direct band gap. (See Fig. 10.) However, if they occur at different momemtum values then it is
known as an indirect band gap. The creation of electron hole pair, by exciting an electron from the valance band to
the conduction band is more difficult in indirect semiconductors since it involves both minimum energy input Eg and
change in momentum. Hence, in many applications such as solar cells, direct band gap semiconductors give better
efficiencies. Examples: Direct band gap - GaAs, InAs, CdTe. Indirect Band gap - Si, Ge.
8
If either or both valance and conduction bands have multiple maxima and minima, then the maxima of valance
band and minima of conduction band which give the lowest band gap are to be considered (see Fig. 11).
FIG. 11. Direct and indirect band gap in semiconductors. In the second figure the conduction band has two minima. Note that
there is a direct band gap, but the indirect band gap has a lower value and hence it is an indirect band gap semiconductor.
—————————————————————————————
V. APPENDIX
We look for the solution of the Schrödinger equation in a periodic potential with period a, of the form
where C is a complex constant. Due to the periodic boundary condition (9) we have
C = ei2πs/N , s = 0, 1, 2, . . . , N − 1 (25)
with
uk (x + a) = uk (x)
ψ = eikx uk (x)
where uk (x) = uk (x + a + b). Thus, we can write uk (−b) = uk (−b + a + b) = uk (a). Hence,
ψ(a) = eika uk (a)
= eik(a+b) e−ikb uk (−b)
= eik(a+b) ψ(−b)
Similar condition can be obtained for the derivative dψ/dx.
dψ(x) ik(a+b) dψ(x)
=e
dx x=a dx x=−b
Using the solutions from Eqs. (15) and (18) for ψ(a) and ψ(−b) respectively we get