on a simplified form of kane's dispersion relation for semiconductors
TRANSCRIPT
Short Notes K45
phys. stat, sol. (b) 2, K45 (1975)
Subject classification: 13.1; 22.2.1; 22.2.3
Institute of Radio physics and Electronics, University College of Science and Technology, Calcutta
On a Simplified Form of Kane’s Dispersion Relation for Semiconductors
By
B.R. NAG and A.N. CHAKRAVARTI
In recent years, Kane’s dispersion relation has been widely used for taking
into account the non-parabolic nature of the energy bands in calculations on many
of the semiconductors including some important III-V compounds. However, since
the dispersion relation of Kane in its original form is somewhat complicated and is
not very much convenient to be handled in most of the cases, an approximate simpli-
fied form of the relation has been mostly used in the literature in which the coeffi-
cient of non-parabolicity has been taken to be equal to the reciprocal of the band
gap. It may be noted that the e r ror introduced by this approximate form is within reasonable limits as long as the valence band spin-orbit splitting is either much
greater than o r is much less than the band gap. The e r ror increases very much
when the two quantities a re comparable to each other. Thus, with the growihg inter-
est in ternary semiconductars in which the valence band spin-orbit splitting and the
band gap are both functions of the alloy composition leading often to comparable
values, the necessity has been created in obtaining a different simplified form which
gives better accuracy in all the cases. In the present communication such a simpli-
f ied form is given.
For non-parabolic bands, the E(k) relation may be expressed (reference (1))
from Kane’s theory (2) in a simplified manner as mostly used in the literature by
1 5fik 2m = - E ( l + & E ) , a , = ~ , n is
2 2
where m is the effective mass at the band edge, E is the band gap, and the energy n g
E is measured from the band edge. This simplified form of Kane’s relation is widely
used in the literature and i s valid within tolerable limits when the valence band spin-
orbit splitting A is either much greater than or is much less than the band gap E g ‘
K46 phylica status solidi (b) 71
However, the er ror increases very much when A is comparable to E as is often
the case. In what follows, a different simplified form is first derived and is then
demonstrated to have better accuracy in at least three semiconductors than the
form given above.
g
According to Kane's theory (2), the E (k) relation is given by
2 2 g g g 3
( E - E o ) ( E - E o + E ) ( E - E o + E + A ) - P k ( E - E o + E + - A ) = O , (2)
2 2 where E =h k /2m0, m being the free electron mass, and P being the momentum
matrix element. Near the band edge (i.e. when k -. 0), E =fi k /2m and we can 2 2 0 0
n write from equation (2)
E l - z ) [ . ( 1 - $ ) + E g ] [ E ( 1 - 2 ) + E g + h ]
(3) 2 P k =
Now, a s k- 0 , E - 0 and from equation (3) we get
m E (E + A ) P k 2 =-(1- h2k2
2 m e) ' '2 . E + - A g 3
n (4)
Further, since the non-parabolicities in the energy bands a re significant when
m /m << 1, we can neglect (Eo/E)2 and higher order terms and write equation (2)
as n o
2 2 g ' g g 3
E ( E + E ) ( E + E + A ) - P k ( E + E + - A ) - E o
2 + E(E + E + A ) + (E + E )(E + E + A ) - P k ] = o . g g g
2 Replacing P k in the third term of equation(5) by its first order approximate value
given by equation (3) with mn/mo << 1, we get
g 2 2
E ( E + E ) ( E + E + A ) - P k ( E + E + - A ) - Eo ) + E ( E + E +A)+ g g g 3
E(E + E )(E + E + A )
E + E + - A + ( E + E ) ( E + E ) + A ) - 2 I=..
g g g 3
Short Notes K47 2
Substituting the value of P from equation (4) and neglecting (E/E ) compared to g
unity, we can further write from the above equation
Expanding binomially and keeping terms upto (E/E ) , we finally get from equation
(7) g
2 2
n = E(1+ dE) , (8)
fik 2 m
where
Further, if instead of taking a w 1/E , we write from equation (1) g
I% =(e E - 1)/E (9)
and determine this quantity from the original relation of Kane, then we can com-
pare the accuracy obtained from the two simplified forms with that obtained from
the original relation. For this, we equate the values of P k as obtained from
equations (2) and (4) and by putting (E - 5. k /2 m ) = El, we get
2
2 2 0
2 p f i k =
n
E1(E1 + E )(El + E + A ) (E + z A ) 2
m E (E + A )
m 2 n g g
2 m (El i- E + - A ) ( l - -
0 g 3
Thus, from equation (10) we can determine the diffemnt values of u, a8 given by
equation (9) for different values of E , since E = (El+ ( 5 k /2m n n o )(m /m ))-These 2 2
K48 physica status solidi (b) 71
Fig. 1. Plots of the energy dependence of a according to KBne’s dispersion relation
for three different semiconductors together with the values of a,’ and u, = 1/E
g
values of u as obtained from the original
relation of Kane tagether with the value of a
given by 1/E are compared in Fig. 1 for
three different semiconductors with the
value of a’ given under equation (8). The
parameters used for these semiconductors
a re a s follows (reference (3)): (i) InSb:
g
I I I I
0.1 0.2 I o(1 E =0.265eV, m n =0.013m 0’ and A = g =0 .9 eV, (ii) InAs: E =0.46 eV, m = E-
g n =0.022mo, and A =0.43eV,and(iii)GaAs:E =1.58eV, m =0.0655m and g n 0’
A = 0.35 eV . It is apparent from the figure that the simplified form derived here
is a better approximation of Kane’s dispersion relation when A is comparable to
E as in InAs and even when A is much greater than or is much less than E a s g g
in InSb and GaAs , respectively.
Acknowledgement
The authors a re indebted to Prof. J.N. Bhar for his keen interest in the work.
References
(1) E.M. CONWELL and M.O. VASSELL, Phys. Rev, 166, 797 (1968).
(2) E.O. KANE, J. Phys. Chem. Solids I, 249 (1957).
(3) M. NEUBERGER, III-V Semiconducting Compounds, Plenum Publ. Co., New
York 1971 (p. 77, 93, and 45).
(Received April 10, 1975)