A superlattice structure is formed when thin layers (d ≤ 25 nm) of a larger-band gap semiconductor (e.g., AlGaAs) and a smaller-band gap semiconductor (e.g., GaAs) are grown alternatively on a conducting or a semi- insulating substrate. The periodic structure formed by alternate deposition of thin epitaxial layers of two different-band gap materials produces a periodic potential similar to the 1-D Kronig-Penney potential discussed in Section 4.3. A potential barrier is formed between a larger- band gap material (AlGaAs) and a smaller band gap material (GaAs), while a potential well is formed in the smaller band gap material sandwiched between two wide band gapmaterials. The energy band diagram for the superlattice is similar to that of free electrons exposed to a periodic crystal potential, except that now the periodic potential is imposed on Bloch electrons with an effective mass . Depending on the width of the superlattice, the energy states inside the quantum well could be discrete bound states or minibands. Figure 4.17 shows the calculated widths of minibands and intermittent gaps as a function of the period length (i.e., l = l*nm1 + l2) for a symmetrical barrier/quantum well structure with a barrier height of 0.4 eV.(4) It is noted that for an equal barrier/well width (i.e., l1 = l2 = 4 nm) superlattice, the lowest band is extremely narrow and lies 100 meV above the bottom of the quantum well. The second miniband extends from 320 to 380 meV, while higher bands overlap above the top of the potential barrier.Figure 4.18 shows (a) the first and second minibands inside the conduction band of a superlattice along the growth direction (i.e., the z- direction), (b) minibands and minigaps in the kz- direction inside the Brillouin zone, and (c) energy (E1 and E2) versus wave vector k in the kx- and ky- directions (i.e., in the plane of the superlattice). It is seen that within the conduction band, we observe a subband structure of minibands across the potential barrier and the quantum well; the higher minibands extend beyond the height of potential barriers. The lower minibands inside the well are separated by the minigaps in the direction of superlattice periodicity (i.e., the z -direction). Within the plane of the superlattice layers (i.e., the x-y plane), the electron wave functions experience only the regular periodic lattice potential. Therefore, the energy dispersion relations (i.e., E vs. kx and ky) are similar to that of the unperturbed crystal lattice except for mixing the states in the z -direction, which results in lifting the lowest-energy states at k = 0 above Ec of the bulk well material as shown in Figure 4.18b and c. The second miniband results in a second shifted parabola along the kx - and ky - directions. It is seen that the E versus k relation in the kx-ky plane is continuous, while a minigap between the first and second minibands appears in the direction perpendicular to the superlattice (kz). Formation of the miniband in a superlattice can be realized when the wave functions of carriers in the neighboring quantum wells of a multilayer heterostructure overlap significantly. The energy levels broaden into minibands with extended Bloch states. These minibands are expected to lead to the transport of carriers perpendicular to the superlattice layers, which include tunneling, resonant tunneling, ballistic and miniband transport.
Calculations of energy band structures in a superlattice can be carried out by several methods. These include the pseudopotential, tight-binding (LCAO), and envelop-function (i.e., kp⋅) methods. Among these methods, the envelop-function approach is most widely used due to its simplicity. With several refinements, this method can become quite effective in dealing with many problems such as band mixing, the effects of external
fields, impurities, and exciton states. A detailed description of the envelop-function approximation for calculating the energy bands in the superlattice heterostructure devices has been given by Altarelli.(5,6)
The density of states in the minibands of a superlattice is discussed next. It is shown in Figure 4.19 that the density-of-states function has a staircase character (dashed steps) for the isolated quantum wells (i.e., the barrier width is much larger than the well width).(4) In this case, each level can be occupied by the number of electrons given by its degeneracy multiplied by the number of atoms in the quantum well. Thus, the two-dimensional (2-D) density of states, g(E), in each discrete level can be described by
()*2nmgE=πh (4.119)
Where g(E) is measured in cm–2. Eq.(4.119) shows that g(E) for a 2-D system is a constant and independent of energy. When significant overlap occurs, tunneling becomes possible and each energy level splits into minibands, and the staircase behavior (dashed line) changes shape as shown by the solid curly line in Figure 4.19. For comparison, the density-of-states function for a 3-D system is also included in Figure 4.19 for a parabolic band. The density of states functions for other low dimensional systems have also been published in the literature. Figure 4.20 shows the plots of density of states functions versus energy for the 3-D, 2-D, 1-D, Q1-D (quantum wire), and Q2-D (quantum well) systems. The density of states functions for the low- dimensional systems are given respectively as follows:(7)
3*3/21/222(2)()()LmNEEπ=h (3-D) (4.120)
*222()LmNEπ=h (2-D) (4.121)
*1/221/2(2)()2LmNEEπ−=h (1-D) (4.122)
*1/222(2)()()nnLmNEEHEEπ=Σh (Q2-D) (4.123)
*1/221/2(2)()()2()lmlmlmLmNEEEHEEπ−=−Σh (Q1-D) (4.124)
(0-D) (4.125) ()()lmnNEEEδ=−
Where ()Hσis the Heaviside function (()1Hσ= for σ >0; 0()Hσ=for σ<0), and the energy levels, En, Eln, Elnm for the 2-D, 1-D, and 0-D systems are given respectively by (7)
*222nnEmL=⎡⎤⎢⎥⎣⎦hπ (2-D) (4.126)
*22,2lnlnEmLL=⎡⎤+⎢⎥⎣⎦hππ (1-D) (4.127)
*22,,2lmnlmnEmLLL=⎡⎤++⎢⎥⎣⎦hπππ (0-D) (4.128)
Equations (4.125) and (4.128) denote the density of states function and energy levels for the quantum dots (0-D system). The density of states function is a very important function for calculating many response functions and the transport parameters such as thermoelectric power, thermal conductivity, electrical conductivity, and Hall coefficients, which are all depenedent on the density of states at the Fermi energy (EF) or the energy derivative at EF. The concept of minibands and the density of states functions in a superlattice and the low-dimensional systems described in this section are very important for the design and understanding of the quantum devices using multi- quantum well and quantum dot heterostructures grown by the MBE and MOCVD techniques, as will be discussed further in Chapters 12, 13, 14, and 16.
PROBLEMS
4.1. Using the nearly-free electron approximation for a one-dimensional (1-D) crystal lattice and assuming that the only nonvanishing Fourier coefficients of the crystal potential are v(π/a) and v(–π/a) in Eq. (4.56), show that near the band edge at k = 0, the dependence of electron energy on the wave vector k is given by
22*2kokEEm=+h
Where m* = mo [l – (32m2oa4/h4π4)v(π/a)2]–1 is the effective mass of the electron at k = 0.
4.2. The E–k relation of a simple cubic lattice given by Eq. (4.79) is derived from the tight-binding approximation. Show that near k 0 this relation can be expressed by ≈
22*2knokEEm=+h
Where m* = ħ2/2βna2.
And for k ≈π/a, show that the E-k relation is given by
22*2knokEEm=+h
Where m*= – ħ2/2βna2.
4.3. If the conductivity and the density-of-states effective masses of electrons are defined respectively by
***1*2/3*132 (//) and ()cnltdnltmmmmmm−=+=ν
Where and denote the longitudinal and transverse effective masses, respectively, find the *lm*tm
conductivity effective mass and the density-of-states effective mass for Si and Ge crystals. *cnm*dnm
Given: = 0.19m*tm0, = 0.97m*tmo, v = 6 for silicon; and = 0.082m*tmo, = 1.64m*lmo, v = 4 for Ge.
4.4. Explain why most of the III-V compound semiconductors such as GaAs, InP, and InSb, have smaller electron effective masses than that of silicon and germanium.
4.5. Sketch the constant-energy contours for a two-dimensional (2-D) square lattice using the expression derived from the tight-binding approximation
E(k) = Eo + B cos(kx a/2) cos(kya/2)
4.6. Derive expressions for the group velocity (vg), acceleration (dvg/dt), and the effective mass (m*) of electrons using the E-k relation for the two-dimensional square lattice described in Problem 4.5. If cos(kya/2) = 1, plot E, vg, dvg/dt, and m* versus k for the one-dimensional (1-D) crystal lattice.
4.7. If the E–k relation for a simple cubic lattice corresponding to an atomic state derived by the tight-binding approximation is given by
E(k) = Eo – E′o – 2E′ (cos k1α + cos k2α + cos k3α)
Derive the expressions of (i) group velocity, (ii) acceleration, and (iii) the effective mass tensor.
4.8. Repeat Problem 4.7 for a body-centered cubic lattice (s-like states). (See Eq. (4.84).)
4.9. Using the tight-binding approximation, derive the E–k relation for the s-like states in a facecentered cubic lattice.
4.10. The E–k relation near the top of the valence band maximum for silicon and germanium is given by
()()12222422222221223312EkAkBkCkkkkkkm⎛⎞⎧⎫⎡⎤=−±+++⎜⎟⎨⎬⎜⎟⎣⎦⎩⎭⎝⎠h
Where E is measured from the top of the valence band edge. Plus refers to the heavy-hole band and minus is for the light-hole band.
A
B
C
Ge
13.1
8.3
12.5
Si
4.0
1.1
4.1
Using the values of A, B, and C for germanium and silicon given by the above table, plot the constant-energy contours for the heavy- and light-hole bands in silicon and germanium.
4.11. Plot the energy bandgap (Eg) versus temperature (T) for the EΓ, EL, and EX conduction minima of GaAs crystal for 0 < T < 1000 K. Given:
()()()()()()()4242L42X5.405101.5192046.05101.8152044.60101.981eV204TETTTETTTETT−Γ−−×=−+×=−+×=−+
การแปล กรุณารอสักครู่..
