Molecular beam epitaxial growth of III-V semiconductor ... - KOBRA
Molecular beam epitaxial growth of III-V semiconductor ... - KOBRA
Molecular beam epitaxial growth of III-V semiconductor ... - KOBRA
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
3.4 Challenges <strong>of</strong> Hetero<strong>epitaxial</strong> Growth <strong>of</strong> <strong>III</strong>-V on Silicon<br />
Figure 3.5: Growth <strong>of</strong> hetero<strong>epitaxial</strong> layer on a mismatched substrate (a) pseudomorphic<br />
(fully strained) layer below the critical thickness (b) Partially relaxed layer above<br />
the critical thickness with mist dislocations generation in red color. Figure modied<br />
according to reference [31].<br />
discribed in Eq. 3.9. The introduction <strong>of</strong> crystal dislocations and other defects<br />
is an important aspect <strong>of</strong> lattice-mismatched heteroepitaxy. The mist dislocations<br />
located at the heterointerface will degrade the performance <strong>of</strong> any device<br />
whose operation depends on it. On the other hand, any device fabricated in the<br />
hetero<strong>epitaxial</strong> layer will tend to be compromised by the presence <strong>of</strong> threading<br />
dislocations in this layer. The threading dislocations are associated with the mis-<br />
t dislocations and are introduced during the relaxation process. Whereas the<br />
mist dislocations are expected to be present in partially relaxed layers under<br />
the condition <strong>of</strong> thermal equilibrium, threading dislocations are nonequilibrium<br />
defects [42]. Threading dislocation are 1D crystallographic structures, they act as<br />
non-radiative recombination centers and their presence is not welcome for <strong>III</strong>-V<br />
hetero<strong>epitaxial</strong> structures on Si.<br />
There are important dierences between low-mismatch and high-mismatch<br />
hetero<strong>epitaxial</strong> systems, which are not simply a matter <strong>of</strong> degree. The actual<br />
mechanisms <strong>of</strong> strain relaxation and defect introduction have been found to be<br />
dierent. This is due, at least in part, to the three-dimensional nucleation mode<br />
<strong>of</strong> highly mismatched hetero<strong>epitaxial</strong> layers. It is <strong>of</strong>ten expected that a hetero<strong>epitaxial</strong><br />
layer will take on the same crystal orientation as its substrate. In practice,<br />
both pseudomorphic and partly relaxed layers <strong>of</strong>ten exhibit small misorientations<br />
35