Optical Detection of Deep Defects in Cu(In,Ga)Se2

The aim of this thesis is to shed light on the deep defect structure in Cu(In,Ga)Se2 by
photoluminescence measurements and to propose a possible conclusive defect model by
attributing experimental findings to a literature review of defect calculations from first
principles.
Epitaxial films are grown on GaAs by metal organic vapor phase epitaxy and characterized
by photoluminescence at room or low temperature. In CuGaSe2, deep defect
bands at ca. 1.1 eV and 1.23 eV are resolved. A model for the power law behavior in
excitation dependent measurements of the peak intensities is derived, which leads to the
experimental finding of two deep donor-like defects as a result.
In Cu(In,Ga)Se2, the deeper band at around 1.1 eV remains constant in energy when
more and more gallium is replaced by indium in the solid solution. For decreasing Ga-contents,
the band gap is mainly lowered by a decrease of the conduction band energy.
From fitting models for electron-phonon coupling, the dominating deep donor-like defect
is determined at 1.3 eV above the valence band maximum. This level is proposed to be
crucial for high Ga-contents when it is deep inside the band gap and most likely acts
as a recombination center. At low Ga-contents it is resonant with the conduction band.
The larger open circuit voltage deficits for high Ga-contents are proposed to stem at least
partly from this defect which is qualitatively supported by simulations.
Additionally another defect band at around 0.7 eV is observed for high Ga-contents
at low temperatures and at 0.8 eV for low Ga-contents. The intensity of the 0.8 eV band
seems to disappear in a sample with Cu-deficiency. In general, deep luminescence is
always observed with similar energies in all Cu-rich compositions, independent of the Ga-content.
The deep defect involved could explain inferior efficiencies of Cu-rich devices
which show increased non-radiative recombination in general. It is further discussed that
the same deep defect could be the origin of a level at 0.8 eV which is observed in several
photo-capacitance measurements in literature.
Based on the literature review for intrinsic defect calculations by hybrid-functionals, a
possible defect model for shallow and deep defects is derived with a focus on those results,
where different authors using different methods agree. By comparing the experimental
results in the scope of this thesis, the deep defect found at 1.3 eV above the valence band
is attributed to the GaCu antisites. The single (0/-1) charge transition of CuIn and CuGa is
proposed to be the main shallow acceptor in the near-band-edge luminescence of Cu-rich
compositions at 60 - 100 meV, whereas the second (-1/-2) charge transition is attributed
to the deep 0.8 eV defect band.
The present findings could be useful for the improvement of Cu(In,Ga)Se2 solar cells
with stochiometric absorber compositions (Cu-rich growth) or with high band gaps (high
Ga-content). Furthermore, the results show a very good agreement of experiment and recent
theoretical calculations of defects, which can be seen as a promising relation between
photoluminescence spectroscopy and predictions from theory for other complex materials.