
48 Effective Mass Theory
and
Transport Phenomena
•
F(r)
= Ce-
r1ao
where C
is
the normalization constant. Thus the solutions to equation for a
semiconductor are hydrogenic energy levels with the substitutions
m
~
111*,
e
2
~
(e
2
/£) and the ionization energy usually taken as the zero
of
energy for
the hydrogen atom now becomes
E
e
,
the conduction band extremum.
For a semiconductor like we have a very large dielectric constant,
£;
16.
The value for the effective mass
is
somewhat more difficult to specify
in
since
the constant energy surfaces for are located about the L-points
in
the Brillouin
zone and are ellipsoids
of
revolution. Since the constant energy surfaces for
such semiconductors are non-spherical, the effective mass tensor
is
anisotropic.
However we will write down
an
average effective mass value
m*lm;
0.12"
(Kittel
ISSP) so that we can estimate pertinent magnitudes for the donor levels
in
a
....
typical semiconductor. With these values for £ and
nz*
we obtain:
Ee
-
Ed::=:'
0.007eV
and the effective Bohr radius
ab::=:'
7oA.
These values are to
be
compared with the ionization energy
of
13.6 eV
for the hydrogen atom and with the hydrogenic Bohr orbit
of
a
o
=
/i
2
/me
2
=
o.sA.
Thus we see that
a6
is
indeed large enough to satisfy the requirement
that
F(r)
be
slowly varying over a unit cell. On the other hand,
if
a6
were to
be comparable to a lattice dimension, then
F(r)
could not be considered as a
slowly varying function
ofr
and generalizations
of
the above treatment would
have to be made. Such generalizations involve:
1.
Treating
E(k)
for a wider region
of
k -space, and
2.
Relaxing the condition that impurity levels are to be associated with
a single band. From the uncertainty principle, the localization
in
momentum space for the impurity state requires a delocalization in
real space; and likewise, the converse is true, that a localized
impurity in real space corresponds to a delocalized description
in
k -space. Thus "shallow" hydro genic donor levels can be attributed
to a specific band at a specific energy extremum at
ko
in the
Brillouin zone.
On the other hand, "deep" donor levels are not
hydrogenic and have a more complicated energy level structure. Deep
donor levels cannot be readily associated with a specific band or a
specific
k point
in
the Brillouin zone.
In
dealing with this impurity problem, it
is
very tempting to discuss the
donor levels
in
silicon
and.
For example
in
silicon where the conduction band
extrema are at the
~
point, the effective mass theorem requires us to