Blochs Functions by the Fourier Development Method.

Version June 2003, preliminary.

Here we study in detail for one-dimensional model of the crystal structure the general approach to the Bloch states, based on the Fourier method. Though the problem is quite simple in principle, it includes many steps. Since the discussion would get too long if all details were discussed in detail, some of the details are left to the reader, but mentioned as questions . The method is shortly discussed in textbooks, but clearly not to sucient depth. Here we shall discuss the details of the matrix formulation, usually omitted in the textbooks. In Kittels book we nd this approach in Chapter 7, paragraph Restatement of Blochs Theorem, page 169. In Hemmers book (Norwegian), the relevant part is paragraph 23c, called Konstruktivt bevis, page 122. In the book by Ashcroft and Mermin this method is discussed in Chapter 9, paragragh named General Approach to the Schr odinger equation when the potential is weak, page 152. We look at a hypothetical one-dimensional atomic lattice, setting up a periodic potential. This potential is supposed to be a periodically repeated reection-symmetric potential well of shape U(x), so that the periodic potential can be written as

V (x) =

U (x xi ) =
j =

U (x j.a)

where the distance between the centres of two neighbouring potential wells is the lattice constant a. Question: What is the typical realistic size a that can have? Which methods can be used to nd experimentally the lattice parameters for real crystals? We consider the solution in a box of size L = N a, i.e. the (one-dimensional) box contains N atoms. As will be seen from the following, the space of the k-values (wave numbers) has a ne structure, decided by the size of the box, and a gross structure and size decided by the precision with which the potential U (x) will be described by the Fourier series in the cell of size a.
-2G -G 0 G 2G

~ k-2G

~ k-G

~ k

~ k+G

Figure 1: Example of Km and kn values We set up the Schr odinger equation for a one-electron problem in such periodical potential. We assume that the wave function (x) can be expanded in a Fourier series, i.e. written as (x) = and the potential U (x) = (Km ) eiKm x U d) The periodicity conditions on the wavefunction limit and determine the values of kn ( periodicity over N atoms ) (x) = (x + N a) aN kn = n . 2 (kn ) eikn x

while the possible values of Km are given by the so called reciprocal lattice. These are given by the condition Km a = m . 2 On g. 6. the allowed values of kn and Km are represented graphically. (Letters B and G are shothands for Brillouin and gitter -lattice in Norwegian.)
~ ~ k-G ~ ~ ~ l-2G k-2G l l-G
L L L

~ ~ ~ k ~ k+G ~ l+G l+2 G k+2 G ~ L L l-2G

K L L

K L

~ k-2G ~ l-G ~ k-G ~

K L L

K L

K L

K L

K L L

K L

K L

K L

~ k ~
l+G

K L L

K L

K L

K L

~ k+G ~

l+2 G

~ k+2 G

Figure 2: The matrix shown when the basis states are arranged as their k-values in g.1. The matrix elements which are non-zero and belong to k or l are indicated by K or L Using the orthogonality of the basis functions, one can transform the Schr odinger equation to a system of equations in our notation h 2 2 ki 2m (ki ) + (ki Km ) = 0 (Km ) U

In the textbooks we usually nd instead of the above a notation of the following type: h 2 2 ki 2m c (ki ) + U (Km ) c (ki Km ) = 0

since the convenient notation for the Fourier components denoted by tilda used by us here is not commonly accepted in Solid State texts. Question: Which quantities in these equations are unknown? In the gure 6. is the system of the k-value shown for N = 6 and only m = (1, 2, 3, 4) are shown. Use one of the attached copies of the gure to diskuss the meaning of the mathematical symbols in the transformed Schr odinger equation. It is very useful to represent this system of equations in matrix form, at least schematically. This is done in gures , where we have taken a very little nunber of k-values and 5 K-values. We have chosen to put the matrices in small cells of grids. The coecients tildek outside of the matrices should strictly speaking not be there, in fact only their indicies, i.e. the values of k . Figure 3. shows the same as the previous gure, but now we have replaced the K and L by the relevant number Km (from -4 to +4).

~ ~ ~ l-2G k-2G
D 1

l-G

~ k-G ~ l
2

~ ~ ~ k ~ k+G ~ l+G l+2 G k+2 G ~ 3 4 l-2G

D -1 D

1 1

~ k-2G ~ l-G ~ k-G ~

-1 -2 -1

D D

1 1

2 2

-2 -3 -2

-1 -1

D D

1 1

~ k ~
l+G

-3 -4 -3

-2 -2

-1 -1

D D

~ k+G ~

l+2 G

-4

-3

-2

-1

~ k+2 G

Figure 3: The same as the previous gure, but with number Km . Figure 4. is the crucial gure. Here we show that by rearranging the basis states in groups indicated, i.e. related by the k-number from the rst Brilouin zone, the matrix fragments into (here small) nonzero matrices on the diagonal,while all other matrix elements are zero. The pictures make it also possible to understand that only values of kn are in one of the dense submatrices, which are of the form k n kk + G where is an integer, positive or negative. Realizing this, we can see that (x) = (kn ) eikn x (kn ) ei(kk +
G) x

The last sum can be rewritten by taking out the common exp(ikk x) so that (x) = e(ikk x) (kk + G) ei
G x

where the sum is of exactly the same form as yhe sum for the potential U (x) = (Km ) eiKm x U

since Km and G run over the same values. This constitutes a direct and constructive proof of the the Bloch theorem.

~ ~ l+2 G ~ l+G ~ l l-G ~ l-2G

~ ~ ~ k-G k ~ ~ k-2G k+G k+2 G ~ ~ l-2G l-G ~ l ~ ~ l+G l+2 G

KK KKK K KK KK KK K K K K K KK K K K K KK

~ ~ k-2G k-G ~ ~ k+G k ~ k+2 G

Figure 4: The hamiltonian matrix for rearranged basis.

~ k-2G E k-2G ~ + U 0

~ k-G ~ UG E k-G ~ + U 0 ~ U-G ~ U-2G

~ k ~ U 2G ~ UG ~ E k +U 0 ~ U-G ~ U-2G

~ k+G

~ k+2 G ~ k-2G ~ k-G ~ k ~ k+G

3G

4G

~ U-G ~ U-2G

~ U 2G ~ UG E k+G ~ + U0 ~ U-G

3G

~ U 2G ~ UG

-3 G

-4 G

-3 G

Ek+2 G ~ ~ k+2 G U + 0

Figure 5: Here the matrix indicated in the previous gure by a group of letters K is shown in detail The U (K ) for values of K2 are left undined, they might be zero in the simplest models.

B G G

B G

B G

B 0

B G

B G

B G

B G

Figure 6: This gure shows the values of multiples of G - reciprocal lattice, together with the Brillouin zones borders, shifted by G/2. The lines show the possible Km and kn values.

Figure 7: The diagonal energies with the two chosen values l and k plotted. In the three versions of the picture we follow the use of periodicity condition. This is further demonstrated in the gures below, where the top and bottom drawings are combined with the matrices.

~ ~ ~ l-2G k-2G
D 1

l-G

~ k-G ~ l
2

~ k

~ l+G
3

~ k+G

~ k+2 G ~ l+2 G ~ 4 l-2G

D -1 D

1 1

2 2

~ k-2G ~ l-G ~ k-G ~

-1 -2 -1

D D

1 1

2 2

-2 -3 -2

-1 -1

D D

1 1

~ k ~ l+G ~ k+G ~ l+2 G ~ k+2 G

-3 -4 -3

-2 -2

-1 -1

D D

-4

-3

-2

-1

Figure 8: This gure shows the original diagonal energies on the parabola, together with the matrix and the special values k and l, with associated values displaced by multiples of G, i.e. the values Km .

~ ~ l+2 G ~ ~ l l+G ~ l-G l-2G

~ ~ k~ ~ ~ k-G k-2G k+G k+2 G ~ l-2G ~ l-G ~ l ~ l+G ~ l+2 G

KK KKK K KK KK KK K K K K K KK K K K K KK

~ k-2G ~ ~ ~ k-G k ~ k+G k+2 G

Figure 9: This gure shows that the periodicity-moving in the First Brillouin zone corresponds to the rearrangements of the matrix shown above.

Figure 10: This gure shows the First Brillouin zone before and after the matrices diagonalization. For each value of allowed k a matrix must be diagonalized. The eigenvalues originating from closely placed diagonal values get pushed away from each other, as indicated. The middle part shows the rearrangement compared with the top, the bottom gure the result.