dc.description.abstract | As a step towards reducing the CO2 emissions to the atmosphere, generating electricity
from renewable energy resources instead of fossil fuels is a good contributor. Solar cells
convert energy from sunlight directly into electricity, and they have long lifetimes and
require little maintenance as they contain no mechanically moving parts. Even though
efficiency has increased and prices have dropped significantly during the last decade, there
are still many steps that can be made towards making solar cells even more attractive.
Intermediate band solar cells (IBSCs) have a theoretical potential of being 50 % more
efficient than the theoretically most efficient single band-gap solar cells, which is what
is used in today s commercial cells, without any major increase in cost. Many steps are
yet to be taken to develop highly efficient IBSCs, and one way to increase the speed and
reduce the costs of this development is to perform simulations of the cells, which can lead
to a better understanding of how the cells work and which compositions provides the most
efficient cells.
To simulate a solar cell, a good optical model is needed for that cell. Quantum dot (QD)
IBSCs currently have no physically explainable, good optical models, and one way to create
a model is by using spectroscopic ellipsometry (SE). In this thesis, InAs/(Al)GaAs
QD-IBSCs are investigated using this technique. As a QD-IBSC has a relatively complex
structure, different parts of this structure are studied separately to reduce the amount of
unknowns in each measurement. This work looks into GaAs with varying doping, samples
with two and ten layers of InAs QDs with GaAs and AlGaAs spacers, as well as
full solar cell structures. Most emphasis is put into understanding the two-layer QD samples,
as these are the simplest QD structures and therefore contain the least number of
unknowns. Also, these samples have previously been measured using photoluminescence
spectroscopy, atomic force microscopy, scanning electron microscopy and transmission
electron microscopy, providing useful information for SE analysis.
Different models from literature are applied in the modeling of QDs and wetting layer.
The QD samples are first modeled only using bulk GaAs, oxide and surface roughness, before
a macroscopic model with bulk InAs values and an effective medium approximation
(EMA) is applied, and at last an oscillator model of Lorentz oscillators is tried. This thesis
finds that the simple model outperforms the more detailed models, indicating that the
assumptions made in the detailed models are not correct. Applying the oscillator model
gives some surprising results indicating a negative contribution to the pseudo-dielectric
function (PDF) of the sample although the oscillators have positive amplitude. No explanation
to this has been found, and it is not known if this is caused by a bug in the
software used for analysis or if it has a physical explanation. However, when combining
the two detailed models by making an EMA of InAs and an oscillator model, the oscillator
model contributes as expected to the PDF, outperforming the simple model for some of
the measurements.
Although the oscillator model showed some unexpected behaviour, it is still believed
to be the model with most potential to make a good and physically explainable model for
QDs. | |