Now, substituting Eqs. (4.61) and (4.62) into Eq. (4.53) and using the same procedure as described above for the first-order correction of electron wave functions, one obtains the second-order correction of energy, which is
( 220nmnmnomomnHEEE∞=≠′=−Σ (4.63)
Using Eq. (4.63), the expression for the electron energy corrected to the second order is given by
()20nmnnomnomomnHEEEE∞=≠′=+−Σ (4.64)
Equations (4.59) and (4.64) are the new wave functions and energies of electrons derived from the quantum mechanical stationary perturbation theory. The results may be used in the NFE approximation to find the wave functions and energies of the outer-shell electrons of a crystalline solid. As mentioned earlier, the valence electrons in a semiconductor are loosely bound to the atoms, and hence the periodic crystal potential seen by these valence electrons can be treated as a small perturbing Hamiltonian. The unperturbed one-electron Schrödinger equation is depicted by
222()()ooookkkmrEr−∇=φφh (4.65)
Which has the solutions of free electron wave functions and energies given respectively by
1()okikrrNVeφ⋅= (4.66)
222okokEm=h (4.67)
Where N is the total number of unit cells in the crystal, V is the volume of the unit cell, okφ(r) is the free electron wave functions, and is the free electron energy. The pre-exponential factor given by Eq. (4.66) is the normalization constant. The one-electron Schrödinger equation in the presence of a periodic crystal potential V(r) is given by okE
()22*()()()2kkkrVrrErm⎛⎞−∇+=⎜⎟⎜⎟⎝⎠φφh (4.68)
Where m* is the effective mass of electrons in the crystal. The crystal potential V(r) can be expressed in terms of the Fourier expansion in the reciprocal space, which is given by
()()jjKijKrKVrve−⋅=Σ (4.69)
Where Kj is the reciprocal lattice vector and v(Kj) is the Fourier coefficient of the periodic potential V(r).
The new electron wave functions and energies can be obtained by finding the matrix element Hk′k due to the periodic crystal potential V(r) using the stationary perturbation method described above. Now substituting Eq. (4.69) into Eq.(4.60), the matrix element due to the periodic potential V(r) is given by
3*''()|()|()kkkkHrVrrdr=∫φφ
()31()jjjKiKrikrikrvKdrNVeee−⋅′−⋅⋅⎛⎞=⎜⎟⎝⎠Σ∫ (4.70)
Note that the integral on the right-hand side of Eq. (4.70) will vanish unless k – k′ = Kj, where Kj, is the reciprocal lattice vector. Thus, by substituting k – k′ = Kj in Eq.(4.70) and carrying out the integration one obtains
()kkjHKv′= (4.71)
Now, substituting Eq. (4.71) into (4.59) yields the new electron wave functions, which is
()()1()1jjiKjkooKkkrikrvKereNVEE−⋅′⋅=+−⎡⎤⎢⎥⎢⎥⎣⎦Σφ (4.72)
It is interesting to note that the term inside the square bracket on the right-hand side of Eq. (4.72) has the periodicity of the crystal potential V(r), and may be designated as the Bloch function uk(r). Thus, the new electron wave functions given by Eq. (4.72) are indeed satisfied the Bloch type wave functions defined by Eq. (4.17).
The expression of electron energy can be derived in a similar manner by substituting Eq. (4.71) into Eq (4.64), and the result yields
()()2jjokkooKkkvKEEEE′=+−Σ (4.73)
It is seen that the expressions for the electron wave functions and energies given by Eqs. (4.72) and (4.73) become infinity if , and hence the perturbation approximation is no longer valid. This condition occurs at the zone boundaries, and the electron energy corresponding to this condition is given by okEE′=
()22222jokkKkEmm ′−===hh (4.74)
Solving Eq.(4.74) yields
22jjKkK⋅= (4.75)
Here the relation k′ = k – Kj is used in Eq. (4.74). Equation (4.75) represents exactly the Bragg diffraction condition in a crystalline solid, which occurs at the zone boundaries. Failure of the perturbation theory at the zone boundaries is due to the fact that the periodic crystal potential V(r) at zone boundaries is no longer small, and hence cannot be treated as a small perturbing potential. In fact, the Bragg diffraction condition results in a very severe perturbation of electron wave functions at the zone boundaries. Therefore, to find a proper solution for the electron energy and wave functions at the zone boundaries, it is necessary to reconstruct a new perturbed wave function, which is a linear combination of an incident- and a reflected- plane wave. Using a linear combination of the incident- and reflected- plane waves, one can construct a new electron wave function at the zone boundary, which is given by
o1()koikrikrrAeAeφ⋅′⋅=+ (4.76)
Where k′ = k – Kj. Substituting Eq. (4.76) into Eq. (4.65) yields
()()22221220ikrikrkokkkVrEAeVrEAemm′⋅⎧⎫⎧⎫′⎪⎪⎪⎪+−++−=⎡⎤⎡⎤⎨⎬⎨⎬⎣⎦⎣⎦⎪⎪⎪⎪⎩⎭⎩⎭hh (4.77)
Now, multiplying Eq. (4.77) by and integrating the equation over the entire space, one obtains ikre−⋅
()()*10ookkjAEEAvK−− (4.78)
Where22=2okokEmh, and v*(Kj), the conjugate of the Fourier coefficient, is given by
(4.79) ()()*0ikrikrjvKeVredr∞−⋅⋅=∫
Similarly, multiplying Eq. (4.77) by e–ik′· r and integrating over the entire space, one obtains
()()10oojkkAvKAEE′−−= (4.80)
Where 22'=2okokEm′h, and
(4.81) ()()30ikrikjrvKeVredr∞′−⋅⋅=∫
is the Fourier coefficient of the periodic crystal potential V(r). A nontrivial solution exists in Eqs. (4.78) and (4.80) only if the determinant of the coefficients of Ao and A1 is set equal to zero, namely,
()()()()*0okkjojkkKKEEvvEE′−−=−− (4.82)
Now, solving Eq. (4.82) for Ek, and the result yields
()()()()122*142ooookkkkkjjKEEEEEvKv′′⎧⎫⎪ ⎡⎤=+±−+⋅⎨ ⎢⎥⎣⎦⎪⎪⎩⎭ (4.83)
Equation (4.83) shows that a forbidden gap exists at the zone boundaries, and the width of the forbidden gap is determined by the value of 4v*(Kj).v(Kj) inside the square bracket of Eq.(4.83), which is determined by the Fourier coefficient of the periodic crystal potential. In general, the energy band gap will increase with increasing value of the Fourier coefficient ()jvK. Figure 4.7 shows the schematic energy band diagram in the reduced zone scheme derived from the NFE approximation. It is interesting to note that the energy band scheme derived from NFE approximation is similar to that obtained from the Kronig–Penney model for the 1-D periodic lattice. Furthermore, the electron wave functions derived from the NFE approximation are indeed satisfied the Bloch condition. The results show that, except at the zone boundaries where an energy discontinuity (or a band gap) occurs, the energy band scheme derived from the NFE approximation resembles that of the free-electron case (with v(Kj) = 0) discussed earlier.
The NFE approximation presented in this section provides a qualitative description of the electronic states for the outer-shell valence electrons of a 3-D crystal lattice. However, in order to obtain true energy band structures for a real crystal, a more rigorous and sophisticated method, such as the pseudopotential or the orthogonalized plane wave method, must be employed in the energy band calculations. Both methods have been widely used in the energy band calculations of semiconductors.
4.6. THE TIGHT-BINDING APPROXIMATION
In this section energy band calculation using the tight-binding approximation or the linear-combination-of-atomic-orbits (LCAO) method is depicted. The LCAO method, which was first proposed by Bloch, is often used to
calculate the electronic states of core electrons in a crystalline solid. It is generally known that core electrons are tightly bound to the individual atoms, which interact with one another within the crystal lattice. In this case, the construction of electron wave functions is achieved using the LCAO method, and the energy bands of electrons are calculated for the corresponding periodic crystal potential. The atomic orbitals are centered on one of the constituent atoms of the crystal. The resulting wave functions are then substituted into the Schrödinger equation, and the energy values are calculated by a procedure similar to that of the NFE approximation described in Section 4.5. In order to apply the LCAO method to core electrons in a crystalline solid, the solution for the free atomic orbital wave functions must be obtained first. This is discussed next.
If φn(r - Rj) represents the atomic orbital wave functions centered at the lattice site Rj, then the wave functions of the crystal orbits φk(r) corresponding to the wave vector krmay be represented by a Bloch sum, which is
()()()kjnjrCkrRφφ= Σ (4.84)
The summation in Eq. (4.84) extends over all the constituent atoms of the crystal. The coefficient Cj(K), which satisfies the Bloch condition, can be written as
()jikRjCke⋅= (4.85)
Now substituting Eq. (4.85) into Eq. (4.84) one obtains
()()()(),jikrRikrikrknjjreerReUφφ−⋅−⋅=−=Σ (4.86)
To satisfy the Bloch condition, the summation given by Eq. (4.86) must have the periodicity of the crystal lattice.
The LCAO method is clearly an approximation to the true crystal orbitals. This method is adequate when the interatomic spacing is large enough such that overlapping among the atomic orbital wave functions φn(r - Rj) is negligible. Thus, the LCAO method is most suitable for the tightly bound core electrons, and is frequently referred to as the tight-binding approximation. Using this method to derive the wave functions and energy band schemes for the core electrons of a crystalline solid is discussed next.
If φn(r -Rj) represents a set of atomic orbital wave functions that satisfy the free-atom Schrödinger equation, then one can write
()()()(22*2njnojnjnonrRVrRrRErRmφφ⎛⎞−∇−+−−=−⎜⎟⎜⎟⎝⎠h (4.87)