Page 157 - Electrical Properties of Materials

III–V and II–VI compounds 139

All the compounds mentioned so far have been binary compounds, that is
they consist of two elements. In principle there are no difficulties in growing
ternary crystals consisting of three elements. However, if the growth is carried
out from a melt, variations in vapour pressure between the elements make it
almost impossible to maintain their relative concentrations (or stoichiometry)
without additional precautions such as a pressurized, impermeable blanket. As
a result, binary compounds are often the practical limit for crystals (and hence
for substrates). However, the same difficulties are not encountered during epi-
taxy, and it is simple to grow ternary and quaternary compounds on binary
substrates. Al for example is quite happy to occupy a Ga site, so there are no
difficulties in producing a GaAlAs crystal. What will be the effect of adding
Al to GaAs? According to the argument given above, the energy gap should
increase. The more Al is added the larger will be the energy gap. Is that good
for something? Yes. The sources for optical communications depend crucially
on our ability to tailor the energy gap. Let me give you an example. The pre-
ferred material for blue-green semiconductor lasers is GaN, but its energy gap
puts it into the ultraviolet region. This can be adjusted by replacing some of
the gallium with indium, the heavier element reducing the gap. We shall say a
lot more about this in Section 12.7.
There are a few obvious things we can say about II–VI materials. They
have more ionic bonding than III–V materials and less ionic bonding than
I–VII materials. For more accurate figures of the contribution of ionic forces to
bonding see Table 8.3. What about the energy gaps? Just as GaAs has a higher
energy gap than Ge, we would also expect ZnSe (look it up in the periodic
table: Zn is to the left of Ga, and Se is to the right of As) to have a higher
energy gap than GaAs. It is actually 2.58 eV. The same rule, as for III–V ma-
terials, applies again for going up and down in the periodic table. ZnS (S is
above Se) has an energy gap of 3.54 eV, whereas that of ZnTe (Te is below Se)
is 2.26 eV.
Zinc blende is of course the name of the ore from which ZnS is obtained.
It is a little unfortunate that ZnS is one of the II–VI compounds which also
crystallizes in another form called wurtzite. Here the crystals have similar
bond lengths, the formal difference is that alternate (111) plane layers are
◦
rotated 180 about the 111 axis. This gives a hexagonal atomic arrangement.
It is reminiscent of the deviations of diamond to hexagonal planes in graphite
and C 60 . It is not too serious a problem having two versions of ZnS, as well as
ZnO, ZnSe, and CdS. The bond length and density are the same, and electrical
properties practically identical. The bonding is only slightly different because
of different distances of third nearest atoms. This is characterized by the
Madelung constant (Section 5.3.1) which is 1.638 for zinc blende and 1.641
for wurtzite. So we do not have to record different bandgaps and melting
points for the two crystal types. Most of the III–V compounds crystallize in
the zinc blende structure but the nitrides have the wurtzite form. A clue as to
why this happens is that the third nearest atoms are unlike hence attractive.
The nitrogen atom with its small size and high electronegativity is prone to
take a more ionic form of crystal.
The main problems with II–VI materials used to arise from the fact that
some compounds could be made p-type, some others n-type, but no compound
could be made both types. The change for the better came with the advent