Modelling of Perovskite Solar Cells
RONGSHENG WEI Master Degree
Submitted in fulfilment of the requirement for the degree of Master of
Engineering (Research)
FINAL THESIS
SCHOOL OF CHEMISTRY, PHYSICS AND MECHANICAL ENGINEERING
SCIENCE AND ENGINEERING FACULTY
Queensland University of Technology
2018
Queensland University of Technology
1
CONTENTS
STATEMENT OF ORIGINAL AUTHORSHIP ............................................................................ 3
ACKNOWLEDGEMENT ............................................................................................................... 4
ABSTRACT .................................................................................................................................... 5
THE SUPERVISORS AND THEIR CREDENTIALS .................................................................... 7
CHAPTER 1. INTRODUCTION .................................................................................................. 8
1.1 Energy Issues ................................................................................................................................... 8
1.2 Solar Cells ....................................................................................................................................... 10
1.2.1 Working Mechanism in Solar Cells .............................................................................................. 10
1.2.2 Main Parameters in Solar Cells ..................................................................................................... 12
1.3 Perovskite Solar Cells ................................................................................................................... 13
1.3.1 Perovskite Materials ......................................................................................................................... 15
1.3.2 Development of Perovskite Solar Cells ...................................................................................... 17
1.4 SCAPS-1D Based Numerical Simulation .................................................................................. 18
1.4.1 Basic Semiconductor Physics in SCAPS-‐1D .............................................................................. 20
1.4.2 Grading Models in SCAPS-‐1D ...................................................................................................... 23
1.4.3 Operation Theory ............................................................................................................................. 24
1.4.4 Strengths and Limitations .............................................................................................................. 25
1.5 Research Objective and Outline ................................................................................................. 26
CHAPTER 2. NUMERICAL MODEL FOR PSCs ..................................................................... 28
2.1 Introduction ................................................................................................................................... 28
2.2 Physics Model in PSCs ................................................................................................................. 30
2.2.1 Basic Equations .................................................................................................................................. 30
2.2.2 Generation and Recombination Mechanism ........................................................................... 32
2.3 Numerical Simulation Method .................................................................................................... 34
2.3.1 Discretization of Equations ............................................................................................................ 34
2.3.2 Linearization of Equations .............................................................................................................. 36
2.4 Results and Discussion ................................................................................................................. 39
2.4.1 The Effect of Effective Density of State ...................................................................................... 39
2.4.2 The Effect of Relative Dielectric Permittivity ............................................................................. 43
Queensland University of Technology
2
2.4.3 The Effect of Band Gap Energy .................................................................................................... 46
2.5 Conclusion ...................................................................................................................................... 49
CHAPTER 3. SCAPS SIMULATION FOR HTM LAYER IN PSCs ......................................... 50
3.1 Introduction ................................................................................................................................... 50
3.2 Device Model .................................................................................................................................. 52
3.3 Results and Discussion ................................................................................................................. 56
3.3.1 The Influence of HTM Layer Characteristics ............................................................................. 56
3.3.2 Comparison for Different Hole Transporting Materials ........................................................ 59
3.4 Conclusion ...................................................................................................................................... 62
CHAPTER 4. SUMMARY AND FUTURE WORKS ................................................................. 63
4.1 Summary ........................................................................................................................................ 63
4.2 Future Works ................................................................................................................................. 63
REFERENCES ............................................................................................................................... 65
APPENDIX ................................................................................................................................... 72
Queensland University of Technology
3
STATEMENT OF ORIGINAL AUTHORSHIP
The work contained in this report has not been previously submitted to meet requirements for
an award at this or any other higher education institution. To the best of my knowledge and
belief, the report contains no material previously published or written by another person except
where due reference is made.
Signature:
Date:
QUT Verified Signature
Queensland University of Technology
4
ACKNOWLEDGEMENT
First of all, I would like to thank my supervisor, Associate Professor Hongxia Wang. She is
very professional, and she has given me a considerable amount of help during my study at QUT.
I really appreciate her sharing her knowledge of perovskite solar cells and discussing questions
about physical models with me. She also gave me many valuable suggestions regarding my
project and thesis. I also would like to thank Professor Aijun Du and Fawang Liu for giving
me the guidance regarding my thesis and numerical modelling. As well, I would like to thank
my research group members, including Shengli Zhang, Teng Wang, Nima and Disheng Yao.
My sincere thanks also go to Libo Feng, who is a member of another group. All of them gave
me lots of assistance during my project and I have really appreciated their generosity. I
acknowledge the services of professional editor, Diane Kolomeitz, who provided copyediting
and proofreading services, according to the guidelines laid out in the university-endorsed
national ‘Guidelines for editing research theses’. Finally, I want to thank my family members
for their ongoing support, both financially and emotionally. Without them, I would not have
achieved what I have got today.
Queensland University of Technology
5
ABSTRACT
Perovskite solar cells have become a hot topic in the solar energy device area. With 10 years
of development, the energy conversion efficiency has seen a great improvement from 2.2% to
more than 22%, and it still has great potential for further enhancement. Numerical simulation
is a crucial technique in deeply understanding the operational mechanisms of solar cells and in
predicting the maximum value of a solar cell with controlled design. This technique can also
give guidance on the structure optimisation for different devices. In this project, two main areas
of research have been discussed. The first part of this thesis illustrates a numerical model of
perovskite solar cells (PSCs) by using the Matlab program. It also introduces a specific
computation process for this model. This model is used to study semiconductor physics in PSCs
and investigate the effect of three parameters on device performance: density of state, relative
dielectric permittivity and band gap energy. The simulation results reveal that a large value of
effective density state can decrease both short-circuit current (Isc) and open-circuit voltage (Voc)
in particular on Voc. A large value of relative dielectric permittivity can cause a large Voc and
Isc, but the effect is not significant. Larger band gap energy is beneficial for the Voc and Isc
without considering reduction of light absorption. The second part shows a model of a
perovskite solar cell with the structure of glass/ FTO/ TiO2/ CH3NH3PbI3/ HTM/ Au by using
SCAPS-1D software. The influence of hole transport material layer characteristics, including
hole mobility and band gap offset, on the performance of PSCs are investigated by using this
model with Spiro-MOeTAD. Besides Spiro-OMeTAD, two-hole transport materials based on
inorganic materials including CuO and Cu2O are also discussed. The simulation results show
that with the increase of hole mobility in the HTM layer, the value of PCE is enhanced. When
Queensland University of Technology
6
hole mobility reaches 5e-3 cm2/Vs, PCE closes to the maximum value and the change reaches
a plateau. The device can obtain high performance when the value of the valance band gap
offset is between -0.2 eV and 0.2 eV. Furthermore, the device shows a better performance by
using inorganic materials (CuO, Cu2O) as the HTM layer, than by using spiro-MOeTAD,
especially to Cu2O with a PCE of 21.87%.
Key Words: Perovskite solar cell, Matlab, SCAPS-1D, Modelling
Queensland University of Technology
7
THE SUPERVISORS AND THEIR CREDENTIALS
Principal Supervisor: Assoc. Prof. Hongxia Wang
Hongxia Wang is an Associate Professor of Energy and Process Engineering in the Faculty of
Science and Engineering of Queensland University of Technology (QUT). She is a member of
the Royal Australian Chemical Institute, MRACI, and a member of the Australian Photovoltaic
Institute. Her research areas focus on energy conversion and storage devices, semiconductor
materials, and the charge transport process.
Associate Supervisor: Prof. Aijun Du
Aijun Du is a Professor of Energy and Process Engineering in the Faculty of Science and
Engineering of Queensland University of Technology (QUT). He is a member of the Australian
Research Council Nanotechnology Network, American Nano Society and American Chemical
Society. His research areas are related to clean energy, nanoelectronics, and environment,
through the use of advanced theoretical modelling.
Queensland University of Technology
8
CHAPTER 1. INTRODUCTION
1.1 Energy Issues
With the rapid development of the economy, the demand for energy is increasing. An
investigation from the World Energy Resources has predicted that global population will reach
to 8.1 billion in 2020, and total primary energy supply is expected to reach 17208 Mtoe in 2020
[1]. Shafiee and Topal have reported that the world’s reserves of oil, gas and coal will be nearly
used up after 35 years, 37 years and 107 years respectively [2]. In addition, a growing number
of research works have shown that the issues, such as pollution and global warming, are the
consequence of burning of fossil fuels [3]. Along with the issue of depletion of fossil fuels,
environmental pollution and global warming caused by burning of fossil fuels have raised
significant concerns as well. Thus, an urgent mission is to find an alternative source, which is
clean, renewable and sustainable, to replace fossil fuels. Solar energy is an infinite, clean and
flexible energy, and it can be converted into many other categories of energy for different
demands. Compared with other renewable energy resources such as hydropower, wind energy,
bioenergy, geothermal and nuclear energy, solar energy is more accessible and rich [4]. Solar
technology has developed rapidly in the past decades. It successfully overcomes the technical
difficulties for improving efficiency of energy conversion, meanwhile, it reduces the cost that
meets the requirements of commercialisation [5]. Fig.1.01 illustrates different energy
consumptions from 2005 to 2015; from this, it is clear that the proportion of solar energy
consumption has increased more than 40 times from 2005 to 2015. Furthermore, the cost of
electricity generated from the solar PV and solar thermal has reduced significantly compared
with 2010 and 2015, as shown in Fig.1.02 [6]. It is clear that solar energy as a renewable energy
Queensland University of Technology
9
is promising to change the world’s energy pattern. It also has a lot of room to progress with the
continuous improvement of technology.
Fig.1.01The comparative energy consumption in 2005, 2010, and 2015 [1]
Fig.1.02 The tendency of global renewable energy cost of electricity in 2010 and 2015 [1]
Queensland University of Technology
10
1.2 Solar Cells
The solar cell is an important energy conversion device harnessing solar energy. The first solar
cell was fabricated by using single silicon crystal in Bell labs, which showed an energy
conversion efficiency of 6% in 1954 [7]. The first commercial silicon solar cell was reported
in 1955, with a 2% efficiency. Because of the high cost and price, very few people actually
used the solar cell for home application, but it gradually began to be widely used in daily life
after 1973 [8]. A new type of solar cell with low fabrication cost and 10% efficiency was made
at the Institute of Energy Conversion of University of Delaware in 1980; it was called thin film
solar cell [9]. In 1982, a polysilicon solar cell was widely produced using a casting method, by
the Kyocera Corporation [10]. In 1991, Michael Gratzel and Brian O’Regan reported a new
type of solar cell called dye-sensitized solar cells (DSSCs), and further reducd the cost of
fabrication by 50% compared to silicon solar cells, by using solution processes in 1988 [11].
Since then, research on solar cells has entered the third generation. Perovskite solar cells (PSCs)
are derived from the research concept of dye-sensitized solar cells.
1.2.1 Working Mechanism in Solar Cells
Solar cells are made by semiconductor materials, which can generate electricity from sunlight
directly by using a photovoltaic effect. When a solar cell is exposed to light, a portion of the
photon with the energy larger than the bandgap is absorbed by the semiconductor. The
absorbed photons with sufficient excitation energy (E>Egap) can cause the transport of electrons
and holes; electrons in the conduction band and holes in the valence band move in different
directions, as shown in Fig.1.03 [12].
Queensland University of Technology
11
Fig.1.03 Excitation and charge separation [12] Fig.1.04 Working principle of solar cell [13]
The carriers, which are generated around the P-N junction, reach the space charge region
without recombination. Due to the effect of an internal electric field, holes diffuse into the P-
type region, and electrons flow into the N-type region. As a result, there are excess holes in the
P-type region and electrons, which are stored in the N-type region, have the same situation.
Excess electrons and holes form an electrostatic field near the P-N junction, which has the
opposite direction to the potential energy barrier. The electrostatic field can not only offset
parts of the potential energy barrier effects, but also make the P-type region have positive
electricity and N-type region have negative electricity. As a consequence, the electromotive
force is created between these two regions [13]. If there is a circuit loop, current can be
produced. The complete process is shown in Fig.1.04.
Queensland University of Technology
12
1.2.2 Main Parameters in Solar Cells
The performance parameters of a solar cell mainly refer to the output characteristics of the
device (I-V) including short-circuit current, open-circuit voltage, fill factor and power
conversion efficiency [14], as shown in Fig.1.05.
Fig.1.05 Output characteristics of the device (I-V) [14]
l Open-circuit voltage: The open-circuit voltage is the maximum output voltage of solar
cells. It can be obtained when the value of the output current is zero. Open-circuit voltage
can be expressed by Eq.1.1
𝑉"# =%&'(𝐿𝑛[ ,
,-+ 1] (1.1)
where, 𝑘2 is the Boltzmann constant, T is temperature, q is elementary charge, I is the
light-generated current density, I0 is the saturation current density.
l Short circuit-current: Set a solar cell under a standard light source, when the output is in
short circuit state, which means the voltage value is 0, the current is the maximum output
current called short-circuit current. It is given as:
𝐼4# = 𝐼 − 𝐼6(𝑒9:;&< − 1) (1.2)
Queensland University of Technology
13
l Fill factor: This is the ratio of maximum output power to the open circuit voltage and short
circuit current. Fill factor is associated with maximum output power. A higher fill factor
leads to a greater output power. The value of fill factor depends on series resistance and
voltage. Fill factor is defined by using the equation below:
𝐹𝐹 = ?@ABC,DC
(1.3)
Where, Pm is the maximum output power. Voc is the open-circuit voltage. J is the short-
circuit current.
l Power conversion efficiency: Power conversion efficiency (PCE) is a significant
parameter. It is defined as the ratio of the maximum output power to the incident light
power, as shown in Eq.14.
𝜂 = ,DC∗GG∗ABC?HI
(1.4)
where, 𝑃KL is the incident light power. It can be easily seen from this equation that the
value of power conversion efficiency is determined by Isc, FF, Voc and Pin.
1.3 Perovskite Solar Cells
In 2012, Nature published an article of a perovskite solar cell with an efficiency over 10%.
Since then, perovskite solar cells have drawn a lot of attentions due to their high performance,
low cost of fabrication and great potential for commercialisation [15]. Generally, the structure
of perovskite solar cells consists of five layers, which are a metal-based cathode layer, hole
transporting material (HTM) layer, perovskite layer, electron transporting material (ETM)
layer and anode (FTO/ITO). A typical perovskite solar cell with planar structure is shown in
Queensland University of Technology
14
Fig.1.06. The perovskite layer is used as a light absorber in the device, where photon excitation
occurs. The generated electrons and holes are separated and transferred to the ETM layer and
HTM layer respectively. Due to its ambipolar characteristic, perovskite has a high mobility of
both electrons and holes. The function of the ETM layer is to extract and transfer electrons and
block holes. For the HTM layer, it is used to extract and transfer holes and block electrons. It
means high mobility for electrons in the ETM layer, high mobility for holes in the HTM layer,
and appropriate band offsets between ETM layer/perovskite layer/HTM layer that are
necessary for high efficiency [16-20]. The main processes of charge transport in perovskite
solar cells are shown in Fig.1.07. The green arrows represent the favourable processes for
energy conversion including photon excitation, transportation of electrons from perovskite
layer to ETM layer, and transportation of holes from perovskite layer to HTM layer. The red
arrows indicate the undesirable processes, which cause energy loss, in perovskite solar cells
involving the recombination of charge carriers, back electron flow from the ETM layer and
hole flow from the HTM layer to perovskite layer [21].
Fig.1.06. The configuration of perovskite solar cell with planar structure [40]. Fig.1.07
Schematics showing of charge transport in perovskite solar cells [40]
Queensland University of Technology
15
1.3.1 Perovskite Materials
Perovskite material is a type of material that shares similar crystal structure with calcium
titanate (CaTiO3) and can be expressed as ABX3 [22]. As shown in Fig.1.08, A is organic
ammonium, such as CH3- NH+3, B is metal cation, (e.g. Pb2+), X is halide anion (e.g. Cl, Br, I.
In the crystal structure, A ion is surrounded by eight three-dimensional structure of a corner
sharing octahedral BX6 units [23].
Fig.1.08 The Perovskite crystal structure [40]
The probable crystallographic structure of a perovskite material can be predicted by analysing
a tolerance factor t and an octahedral factor u. t can be expressed as Eq.1.01.
𝑡 = (𝑟O + 𝑟P)/( 2[𝑟2 + 𝑟P]) (1.01)
where, rA, rB and rX are the ionic radii of A, B and X respectively. The report shows that the
tolerance factors of most perovskite materials lie in the range from 0.75 to 1 [24]. However, it
is not enough to deduce the probable crystallographic structure of perovskite materials by only
Queensland University of Technology
16
considering tolerance factors. Therefore, the octahedral factor u is used as an additional
indicator to predict the formation of perovskite structure. U can be described by Eq.1.02.
𝑢 = 𝑟2/𝑟P (1.02)
Generally, the ABX3 perovskite structure can be formed under the conditions of 0.813< t
<1.107 and 0.442< u <0.895 [25]. The structure map by considering t and u for ABX3
perovskite materials is shown in Fig.1.09.
Fig.1.09 The structure map for ABX3 perovskite materials [25]
Compared with other materials, perovskite materials have their own unique characteristics such
as appropriate band gap energy (around 1.55eV), long carrier lifetime and diffusion length, and
high extinction coefficient [26-39]. Due to these favourable characteristics, perovskite
materials are widely applied as a light absorber in solar cell device. However, they also have
some limitations including low stability under moisture and ultraviolet radiation environments,
which are easily to result in degradation of performance, and the effect of toxicity problems,
caused from the toxicity ion of Pb2+ on both the environment and the human body during the
fabrication and disposal processes [30-31].
Queensland University of Technology
17
1.3.2 Development of Perovskite Solar Cells
The first perovskite solar cell was reported by Miyasaka and colleagues in 2006. They used
CH3NH3PbBr3 as the material and obtained a solar cell with an efficiency of 2.2% [32]. Three
years later in 2009, they used iodine instead of bromine, and the efficiency achieved by the
solar cell rose to 3.8% [33]. Park and co-workers furthered the increase in efficiency to 6.5%
through the TiO2 surface treatment in 2011. The perovskite quantum dot (QD) sensitized solar
cell has a better light absorption compared with the dye-sensitized solar cell. However, the
stability of the perovskite QD-sensitized solar cell is low, because QD dissolves into electrolyte
solution [34]. In order to avoid the effects from electrolytes, Park and teammates attempted to
use solid state organic molecules or polymers as HTM to replace the liquid electrolytes. They
used Spiro-MeOTAD based HTM with mesoscopic TiO2 as ETM to improve the device
stability. They achieved a solar cell with an efficiency of 9.7% in 2012 [35]. In the same year,
Snaith and colleagues reported perovskite solar cells using Spiro-MeOTAD based HTM and
Al2O3 as scaffold. The device presented an efficiency of 10.9%. In their report, they showed
that a better performance can be obtained by using mixed-halide (CH3NH3PbI3-xClx) due to
improved ability of charge transport. They also showed that perovskites had a bipolar charge
carrier transport of electron and hole [36]. In 2013, Seok, Grätzel and colleagues reported an
efficiency of 12.3% by using the structure of nanoporous TiO2 infiltrated by mixed-halide
perovskite [37]. Simultaneously, an efficiency over 15% was reported from Burschka and
group members by using TiO2 scaffold and iodide deposition [38]. Snaith removed scaffolding
from the device and applied planar structure, obtaining similar conversion efficiency compared
with the report by Burschka et al [39]. Subsequently, efficiencies of 16.2% and 17.9% were
Queensland University of Technology
18
reported by Seok et al. in early 2014 by using CH3NH3PbI3-xClx and a poly-triarylamine HTM
[40-41]. In 2016, Saliba introduced a perovskite solar cell with an efficiency of 21.1% by using
a mixture of triple cation (Cs/MA/FA). It showed high stability and reproducibility [42]. Seok
and co-workers introduced an approach to reduce defects in the perovskite layer by using an
intramolecular exchanging process in 2017, which is favorable in reducing the defects
concentration, and they obtained an efficiency of over 22% [43].
1.4 SCAPS-1D Based Numerical Simulation
Numerical simulation is a crucial and efficient way to investigate the physical mechanism in a
solar cell device without actually making the device. It can save both time and money in device
development. Much simulation software has been developed and applied in the research of
solar cell devices such as AMPS-1D, SCAPS-1D, PC1D, AFORS-HET and so on. SCAPS-1D
is a one-dimensional simulation software developed by the University of Gent, Belgium. It has
been applied to the study of different types of solar cells such as CZTS, CdTe, CIGS, etc. [43-
47]. Compared with other software, SCAPS has a very intuitive operation window and
diversified models for grading, defects and recombination. The main features of SCAPS
including [48] materials and defects properties can be defined in 7 semiconductor layers, as
shown in Fig.1.10 and Fig.1.11, where plentiful grading laws are provided for almost all
parameters of materials and defects, and the defect definition can be set in both bulk and
interface. There are five defect types and distributions available in the software and a variety
of properties related with solar cells, such as energy bands, concentrations, currents, I-V
characteristics, C-V, C-f, and QE can be determined by SCAPS. SCAPS can also provide
Queensland University of Technology
19
flexible calculation and record functions including single shot, batch calculation, curve fitting,
data and diagram recording.
Fig.1.10 Layer definition panel in SCAPS-1D
Fig.1.11 Material and defect definition panel in SCAPS-1D
Queensland University of Technology
20
1.4.1 Basic Semiconductor Physics in SCAPS-1D
l Basic Equations
Poisson’s equation is used to describe the relationship between potential and space charges, as
shown in Eq.1.03
TU
TVU𝜑 𝑥 = (
Y[𝑛 𝑥 − 𝑝 𝑥 − 𝑁\](𝑥) + 𝑁O^(𝑥) − 𝑝_ 𝑥 + 𝑛_(𝑥)] (1.03)
where, 𝜑 is the potential, q is the elementary charge, 𝜀 is the permittivity, n is the density of
free electron, p is the density of free hole, 𝑁\] is the ionised donor-like doping density, 𝑁O^
is the ionised acceptor-like doping density, 𝑝_ is the trapped hole density, 𝑛_ is the trapped
electron density.
Continuity equations Eq1.04 -1.05 define the transportation of carriers
𝑞 TLT_= TbI
TV+ 𝑞𝐺 − 𝑞𝑅
𝑞 TeT_= − Tbf
TV+ 𝑞𝐺 − 𝑞𝑅
(1.04) 𝐽L = 𝑞𝑛𝜇L
TjTV+ 𝑞𝐷L
TLTV
𝐽e = −𝑞𝑝𝜇eTjTV+ 𝑞𝐷e
TeTV
(1.05)
where, G is the optical generation rate, R is the recombination rate, 𝐷L is the electron diffusion
coefficient, 𝐷e is the hole diffusion coefficient, 𝜇L is the electron mobility, and 𝜇e is the
hole mobility.
l Concentration for Electrons and Holes
In thermal equilibrium, the free carrier concentrations are expressed by Eq.1.06-1.07
𝑛 = 𝑁#exp (op^oC%&'
) (1.6)
𝑝 = 𝑁qexp (or^op%&'
) (1.7)
where, Ef is Fermi level, 𝑘2 is Boltzmann constant, T is temperature, Ec and Ev are energy
levels under a steady state.
Queensland University of Technology
21
l Diffusion Length
Diffusion length describes the transport ability of carriers in a solar cell device. It depends on
diffusion coefficient and carrier lifetime, which is shown in Eq.1.09
𝐿 = 𝐷𝜏 (1.09)
where, L is diffusion length, 𝜏 is carrier lifetime.
l Recombination Mechanism
There are three recombination models used in SCAPS.
Band-to-band recombination is a reverse process of photon absorption. The electrons in the
conduction band drop back down to the empty valence band and recombine with holes. The
band-to-band recombination rate can be expressed as:
𝑅 = 𝛾(𝑛𝑝 − 𝑛Ku) (1.10)
The Shockley-Read-Hall (SRH) recombination is also called trap-assisted recombination. It
occurs due to the defects or impurities in the materials. The SRH recombination rate can be
given by:
𝑅 = Le^LHU
vf L]L- ]vI(e]e-) (1.11)
Auger recombination is a process when a pair of electron and hole recombination occurs during
the transition from high energy level to low energy level, with the resulting energy being given
to the third carrier. It can be described by:
𝑅 = (𝑐LO + 𝑐eO)(𝑛𝑝 − 𝑛Ku) (1.12)
where, R is recombination rate, 𝛾 is recombination coefficient, 𝜏L and 𝜏e are lifetimes for
electron and hole. 𝑛6 and 𝑝6 are equilibrium electron concentration and equilibrium hole
concentration, 𝑐LO and 𝑐eO are constants, which can be set in SCAPS.
Queensland University of Technology
22
l Work Function
Work function is the minimum energy required to move an electron from a solid to vacuum.
The value of work function is used to describe the strength of binding energy of an electron in
materials. In SCAPS, work function can be set by the user, or it can be calculated by using the
model in SCAPS, as shown in Eq.1.13-1.15
n-type contact:
𝜙y = 𝜒 + 𝑘2𝑇𝑙𝑛(}~
}�^}�) (1.13)
p-type contact:
𝜙y = 𝜒 + 𝐸��e − 𝑘2𝑇𝑙𝑛(}~
}�^}�) (1.14)
intrinsic contact:
𝜙y = 𝜒 + 𝑘2𝑇𝑙𝑛(}~LH) (1.15)
where, 𝜙y is work function, 𝜒 is electron affinity, Nc is effective density of state for
conduction band, NA and ND are acceptor and donor dopant densities. Ni is intrinsic carriers’
densities.
l Absorption Coefficient
The absorption coefficient can be defined as the extent to which a material absorbs energy. It
is determined by the nature of material. In SCAPS, the absorption coefficient is given by
Eq.1.16
𝛼 𝜆 = 𝐴 + 2��
ℎ𝜈 − 𝐸��e (1.16)
Queensland University of Technology
23
where, A and B are the absorption constants, h is Planck’s constant, v is light speed [49].
1.4.2 Grading Models in SCAPS-1D
In SCAPS, parameter grading is an important function which can make a model more flexible.
Most of the parameters can be graded in the material and defect definition panel with various
grading models provided in SCAPS. Each layer is assumed to have composition 𝐴�^�𝐵� . If
the layer is only made of pure compound A, it means y = 0, and y = 1when the layer is only
made of pure compound B. If the layer contains both compounds A and B, the value of y is
between 0 and 1. The grading y(x) is through the whole layer; y is required to be defined at
both sides of the layer. We can describe the distribution of composition though the layer by
using grading laws. The material properties P are composition dependence, which can be
expressed as P[y(x)]. SCAPS provides several grading laws, which are listed below [48], and
these laws can be defined in the grading selection.
Uniform:
𝑃O = 𝑃2
Linear:
𝑃O 𝑦2 − 𝑦 + 𝑃2 𝑦 − 𝑦O𝑦O − 𝑦2
Parabolic:
𝑃O 𝑦2 − 𝑦 + 𝑃2 𝑦 − 𝑦O𝑦2 − 𝑦O
− 𝑏(𝑦2 − 𝑦)(𝑦 − 𝑦O)
(𝑦2 − 𝑦O)u
Parabolic2:
𝑦 < 𝑦6: 𝑃6 + (𝑃O − 𝑃6)(𝑦 − 𝑦6𝑦O − 𝑦6
)u
𝑦 > 𝑦6: 𝑃6 + (𝑃2 − 𝑃6)(𝑦 − 𝑦6𝑦2 − 𝑦6
)u
Queensland University of Technology
24
Logarithmic:
𝑃O�&^��&^�O×𝑃2
�^���&^��
Exponential:
𝑃6 + 𝑃O − 𝑃6sinh 𝑦2 − 𝑦
𝐿Osinh 𝑦2 − 𝑦O
𝐿O
+ 𝑃2 − 𝑃6sinh 𝑦 − 𝑦O
𝐿2sinh 𝑦2 − 𝑦O
𝐿2
Beta function:
𝑃O + (𝑃O − 𝑃2)𝛽�,�(𝑦 − 𝑦O𝑦2 − 𝑦O
)
Power law:
(𝑃O�y×
𝑦2 − 𝑦𝑦2 − 𝑦𝐴
+ 𝑃2�/y×
𝑦 − 𝑦O𝑦2 − 𝑦O
)y
where, b is bowing factor, P0 is background value, LA,B is characteristic lengths, 𝛽�,� is the
incomplete beta-function, m is power.
1.4.3 Operation Theory
The main theory of SCAPS-1D is to solve Poisson’s equation and continuity equations. Fig.1.2
shows the working strategy of SCAPS-1D. Each calculation begins from the start point, and
uses the initial assumption, which is expressed by using quasi-Fermi levels, to obtain the
equilibrium situation. In this situation, no illumination and voltage are applied. When the work
point is set in a dark condition, the equilibrium condition is applied to calculate the solution.
When it is set under a light condition, the short circuit situation will be calculated, and this new
value will be used as the initial value of the next step [48].
Queensland University of Technology
25
Fig. 1.12. Working Strategy of SCAPS-1D [48]
The convergence of the Gummel type iteration scheme with Newton-Raphson algorithm is
used in SCAPS for numerical calculations. After a calculation point is set, SCAPS will follow
Newton-Raphson algorithm to undertake iteration calculation until obtaining the optimisation
value.
1.4.4 Strengths and Limitations
The strengths of SCAPS-1D include the following features [50]:
l It has a clear user interface, and it is a kind of interactive software. Almost all parameter
settings can be saved as external ASCI code files and read from external files.
l It allows users to write scripting languages and run them.
l In the aspect of the device parameter setting, almost all parameters can be graded in
SCAPS-1D, which is beneficial to the study of the composite layer formed by the
interdiffusion between different layers.
Queensland University of Technology
26
l It considers the defect in bulk and interfaces and provides a variety in types of selection
of defect charge. The energy distribution caused by defects can be single level, uniform
distribution, Gauss distribution, and Band tail.
However, there are three limitations in the SCAPS-1D. Firstly, the device model can only add
seven semiconductor layers. Due to this reason, SCAPS-1D is not suitable for a device with
graded junctions. Another limitation is that SCAPS-1D is developed for thin film solar cells,
so it cannot be applied for other device structures, such as a perovskite solar cell with
mesoporous structure. The last limitation is about the accuracy of calculation. Although the
resolution is already a small number (<10-15), it is usually insufficient for the simulation of
semiconductors.
1.5 Research Objective and Outline
There are numerous studies related to silicon solar cell, thin film solar cell and organic solar
cell. However, few studies and models have been understood in perovskite solar cells.
Numerical analysis approach can give an assistance to understand PSCs deeply, and it is a
significant way to improve the research efficiency in PSCs. This project aims to 1) understand
the operation mechanism and physical model of perovskite solar cells, and 2) study the
mathematical approaches related to numerical simulation. Furthermore, it seeks to 3)
investigate the effect of HTM layer characteristics on the performance of perovskite solar cells,
and 4) study the perovskite device performance by using the inorganic materials (CuO and
Cu2O) as the hole transport materials. This project gives two simplified models of PSCs in
order to study PSCs more clearly and easily. Numerical simulations by using Matlab and
Queensland University of Technology
27
SCAPS are employed to study the absorber layer and HTM layer of perovskite solar cells.
Chapter 1 gives an introduction regarding the current world energy issues, solar cell device,
perovskite solar cells and SCAPS simulation program. Chapter 2 presents a numerical
simulation with specific computation processes accordingly for perovskite solar cells by using
Matlab to investigate the influence of effective density state, relative dielectric permittivity and
band gap energy on the device performance. Chapter 3 shows a model of perovskite solar cells
with the structure of glass/ FTO/ TiO2/ CH3NH3PbI3/ HTM/ Au by using SCAPS-1D software.
The influence of hole transport material layer characteristics on the performance of PSCs are
investigated. Furthermore, three different hole transporting materials, including CuO, Cu2O
and Spiro-OMETAD are also discussed. As for Chapter 4, it is about the works which are
required to extend in the future research.
Queensland University of Technology
28
CHAPTER 2. NUMERICAL MODEL FOR PSCs
2.1 Introduction
Numerical simulations have been applied to study the performance of solar cells over a period
of 40 years [51]. Jerry provided an effective analysis approach to discuss the carrier transport
problem of silicon solar cells by solving Poisson’s equation and continuity equations for
electrons and holes in 1975 [52]. Several software programs are widely used for thin film solar
cells, such as SCAPS-1D, PC1D and AMPS-1D, to understand the mechanism of solar cells in
recent years [53]. The numerical analysis for dye sensitized solar cells (DSCs) is also studied
and developed by many researchers [54-56]. With the development of these numerical models,
an enhanced understanding of the mechanisms in solar cells has been gained, which in turn
contributes to the development of the device towards improved efficiency.
However, the operation mechanisms of silicon solar cells, DSCs and perovskite solar cells are
different. The model of silicon solar cells can be seen as a p-n junction with doping, and it uses
the dopant density and distribution to describe the performance of solar cells [57]. For DSCs,
the redox level in electrolytes affects the output voltage of a device, so the boundary conditions
and initial value are based on it [58]. In the model of perovskite, the output voltage is
determined by the difference between quasi-Fermi energy level for electrons and holes at the
two electrodes of device [59]. Therefore, in order to obtain accurate results, the density of
carriers and their potential need to be expressed by using the model of quasi-Fermi energy level.
Furthermore, the boundary conditions and initial value need to be created by using the model
of quasi-Fermi energy level.
Queensland University of Technology
29
In this chapter, a numerical model based on Poisson’s equation and continuity equations for
perovskite solar cells has been developed. This model has been simplified as much as possible
under some appropriate assumptions in order to reduce the complication for computation. In
addition, the details of computation approaches and procedures are also introduced in this
chapter. The simulation results are obtained by using Matlab, and these results are used to
understand how the effective density of state, relative dielectric permittivity and band gap
energy affect the performance of perovskite solar cells.
Queensland University of Technology
30
2.2 Physics Model in PSCs
2.2.1 Basic Equations
In this section, a one-dimensional model is established by using Poisson’s equation and
continuity equations to describe the performance of perovskite solar cells. The effects of defect
density and ionised doping density are not considered in this model, so the Poisson’s equation
can be given by:
TU
TVU𝜑 𝑥 = (
Y[𝑛 𝑥 − 𝑝(𝑥)] (2.01)
where 𝜑 is the potential, q is the elementary charge, 𝜀 is the permittivity, n and p are the
density of free electrons and the density of free holes respectively.
The continuity equations of electrons and holes can be described by:
𝑞 TLT_= TbI
TV+ 𝑞𝐺 − 𝑞𝑅 (2.02)
𝑞 TeT_= − Tbf
TV+ 𝑞𝐺 − 𝑞𝑅 (2.03)
where 𝐽L is the electron current density, 𝐽e is the hole current density, G is the optical
generation rate, R is the recombination rate.
Under a stead state condition, electron and hole densities do not change with time, so the
continuity equations can be simplified as: TbITV= −𝑞𝐺 + 𝑞𝑅 (2.04)
TbfTV= 𝑞𝐺 − 𝑞𝑅 (2.05)
Queensland University of Technology
31
In a perovskite solar cell, the output current includes diffusion current and drift current caused
by electrons and holes. It can be obtained from the following equations:
𝐽L = 𝐽�K���4K"L + 𝐽��K�_ = 𝑞𝑛𝜇LTjTV+ 𝑞𝐷L
TLTV
(2.06)
𝐽e = 𝐽�K���4K"L + 𝐽��K�_ = −𝑞𝑝𝜇eTjTV+ 𝑞𝐷e
TeTV
(2.07)
where 𝐷L is the electron diffusion coefficient, 𝐷e is the hole diffusion coefficient, 𝜇L is the
electron mobility, and 𝜇e is the hole mobility. In this model, it is assumed that 𝐷L,e and 𝜇L,e
meet the Einstein relation, which can be expressed as
𝐷L,e = 𝜇L,e%&'(= 𝑉_ (2.08)
where 𝑘2 is the Boltzmann constant, and T is the temperature.
There are three unknowns to be solved, which are n, p and 𝜑. In order to obtain these values,
appropriate boundary conditions are required. At the cathode and anode, density of electrons
and holes obey the Boltzmann distribution, and they can be expressed as:
𝑛 0 = 𝑁#𝑒 C¡ pI¡;&< (2.09)
𝑝 𝐿 = 𝑁#𝑒 r¡ pf;&< (2.10)
in which, 𝑁# is the density of state, 𝐸�L and 𝐸�e are the electron and hole Quasi-Fermi level.
In perovskite solar cells, the output voltage is equal to the difference between 𝐸�L and 𝐸�e
without considering the work function of the metal electrode, and it can be expressed as:
𝜑 = 𝐸�L − 𝐸�e = 𝐸��e + 𝑘2𝑇𝑙𝑛L 6}C
+ 𝑘2𝑇𝑙𝑛(e(¢)}C) (2.11)
Here, in order to simplify the device model, it assumes that the output voltage value is the
bandgap energy of the perovskite, and the corresponding boundary conditions can be given by:
Queensland University of Technology
32
𝜑(¢) − 𝜑 6 = 𝑉� − 𝐸��e (2.12) 𝑛 6 = 𝑁#𝑛 ¢ = 𝑁# exp − o£¤f
A¥
(2.13)
𝑝 ¢ = 𝑁#𝑝 6 = 𝑁# exp − o£¤f
A¥
(2.14)
where, 𝑉� is the applied voltage.
2.2.2 Generation and Recombination Mechanism
In 2004, Hoppe showed that if the thickness of a device is very thin (under 100nm), the
exponential relationship of light absorption and distance do not affect the performance of
device significantly, so the generation rate can be set as a constant [60]. Here, the generation
constant used for perovskite is difficult to find but the absorption coefficient can be easily
found. Therefore, we still use the exponential model, and the charge generation rate can be
given by:
𝐺(V) = 𝐺 𝜆, 𝑥 𝑑𝜆 = 𝐼𝑃𝐶𝐸 𝜆 ,(¨)ש(¨)ת¡«(¬)
�#/¨¨-6
¨-6 𝑑𝜆 (2.15)
where, IPCE is incident photon to current efficiency, 𝐼(¨) is the incident light density, 𝛼(¨) is
the absorption coefficient, h is the Plank constant, c is the light speed, 𝜆 is the photon’s
wavelength. Here, it assumes that IPCE is equal to 100%, which means all of the excitons can
be separated to pairs of hole and electron accordingly. So, the charge generation rate can be
rewritten as:
𝐺(V) = 𝐺 𝜆, 𝑥 𝑑𝜆 = ,(¨)ש(¨)ת¡«(¬)
�#/¨¨-6
¨-6 𝑑𝜆 = 𝛼𝑁6𝑒^©V (2.16)
Queensland University of Technology
33
where, 𝐼(¨) is the AM1.5 standard solar spectra, which is calculated by NREL, 𝑁6 =
,(¨)�#/¨
¨-6 𝑑𝜆. Generally, the bandgap energy of perovskite used for theoretical study is 1.55ev.
So, using the linear fitting result of incident photon density and bandgap energy, 𝑁6 becomes:
𝑁6 ≈ −2.2×10�°×1.55 + 5.12×10�° = 1.71×10�° [61].
In the device, all of the carriers will move to their corresponding electrode, and when they
move, electrons and holes will try to recombine. There are two kinds of common recombination
mechanism, which are direct (band to band) and indirect (RSH) recombination. The direct
recombination is caused by the electrons’ direct transition between the conduction band and
valence band. Generally, it is used to describe the device with low defect concentration. It can
be expressed as [62]:
𝑅 = 𝛾(𝑛𝑝 − 𝑛KL_u ) (2.17)
𝑛KL_ = 𝑁#exp (−o£¤fuA¥
) (2.18)
𝛾 = (³Y
(2.19)
where, 𝛾 is the recombination coefficient, and 𝑛KL_ is the intrinsic carrier concentration. The
indirect recombination is used to describe the device with high defect concentration, and the
recombination occurs in the traps and defect centers. It can be given by [62]:
𝑅 = Le^LHI¥U
vf L]L´ ]vI(e]e´) (2.20)
where, 𝜏L is the lifetime of the electron, 𝜏e is the lifetime of the hole. In this model, it ignores
the trap and defect concentrations, so only considers the direct recombination mechanism.
Queensland University of Technology
34
2.3 Numerical Simulation Method
Poisson’s equation and continuity equations are partial differential equations, which are
difficult to solve directly by using conventional methods. Therefore, numerical computation
approaches are used in the simulation in order to obtain the numerical solutions. In this
simulation, the device works under small current and voltage conditions; the coupling between
the equations is not significant, so finite difference method and Gummel iteration method are
chosen to process these equations.
2.3.1 Discretization of Equations
Finite difference method is a discrete approximation computation approach for differential
equations; its calculation results are not a continuous function in the domain but the
approximate value of the functions at each mesh point. The basic idea of the difference method
is to use different coefficients to replace the derivative part in the equations. In this simulation,
it assumes the thickness of perovskite is L nm, and discretises the domain [0, L] with the
distance of h. Here, in order to obtain convenient calculation and accurate results, nested grid
method is used to discretise the device. The meshing diagram of the device can be shown in
Fig.2.01.
Fig.2.01 Meshing diagram of device
Queensland University of Technology
35
Based on the discretisation model of the device, the Poisson’s equation and continuity
equations with the discrete form can be described by:
j K]� ^j K ^[j K ^j(K^�)]�U
= 𝑛 𝑖 − 𝑝(𝑖) (2.21)
bI K ^bI(K^�)�
+ 𝑞𝐺 − 𝑞𝑅 = 0 (2.22)
bf K ^bf(K^�)�
− 𝑞𝐺 + 𝑞𝑅 = 0 (2.23)
where, k=1,2, 3, …, N-1, i=1, 2, …, N, k is the middle point between i-1 and I, N is the number
of the mesh point. The Scharfetter Gummel approach assumes the change of electric field,
carrier mobility and current density can be ignored between i and i+1, so that Eq.2.06 and
Eq.2.07 become:
𝐽L 𝑖 = ³I(�
𝜑 𝑖 − 𝜑 𝑖 + 1 L K]� ª¶ H ¡¶ H·´
:¥ ^L(K)
[ª¶ H ¡¶ H·´
:¥ ^�]
(2.24)
𝐽e 𝑖 = ³f(�
𝜑 𝑖 − 𝜑 𝑖 + 1 e K ª¶ H ¡¶ H·´
:¥ ^e(K]�)
[ª¶ H ¡¶ H·´
:¥ ^�]
(2.25)
There are 3×(𝑁 − 2) unknowns (𝑛, 𝑝, 𝜑) after substituting Eq.2.24 and Eq.2.25 into the
Poison’s equation and continuity equations.
𝐹L 𝑛 𝑖 − 1 , 𝑛 𝑖 , 𝑛 𝑖 + 1 , 𝑝 𝑖 , 𝜑 𝑖 − 1 , 𝜑 𝑖 , 𝜑 𝑖 + 1 = 0 (2.26)𝐹e 𝑝 𝑖 − 1 , 𝑝 𝑖 , 𝑝 𝑖 + 1 , 𝑛 𝑖 , 𝜑 𝑖 − 1 , 𝜑 𝑖 , 𝜑 𝑖 + 1 = 0 (2.27)𝐹j 𝑛 𝑖 , 𝑝 𝑖 , 𝜑 𝑖 − 1 , 𝜑 𝑖 , 𝜑 𝑖 + 1 = 0 (2.28)
Here, we have 3×(𝑁 − 2) equations with the boundary conditions, so the numerical solutions
can be obtained.
Queensland University of Technology
36
2.3.2 Linearization of Equations
It can be easily seen that Jn, Jp and R are non-linear functions, so that Eq.2.26, Eq.2.27 and
Eq.2.28 are non-linear equations too. It is necessary to linearise these equations by using Taylor
expansion in order to simplify the computation. According to the theory of the Gummel
iteration method, when one variable is calculated we are assuming that other variables are
constants. So, when we calculate 𝜑, it needs to keep n and p as constants. Using the first order
Taylor expansion at 𝜑K^�6 , 𝜑K6, 𝜑K]�6 the Eq.2.28 can be rewritten as
𝐹j =TG¶-
Tj K^�∆𝜑 𝑖 − 1 + TG¶-
Tj K∆𝜑 𝑖 + TG¶-
Tj K]�∆𝜑 𝑖 + 1 + 𝐹j6 = 0 (2.29)
In order to ensure the convergence and the accuracy of results, the relationships between 𝜑
and n, p need to be considered:
𝑛 𝑖 = 𝑛KL_ exp𝜑 𝑖 − 𝐸�L 𝑖
𝑉_ (2.30)
𝑝 𝑖 = 𝑛KL_ exp𝐸�e 𝑖 − 𝜑 𝑖
𝑉_ (2.31)
Each coefficient in Eq.2.29 is shown below:
TG¶-
Tj K^�= �
�U
TG¶-
Tj K= − u
�U− (
Ye- K ]L- K
A¥TG¶-
Tj K]�= �
�U
𝐹j6 =��U𝜑6 𝑖 − 1 − u
�U𝜑6 𝑖 + �
�U𝜑6 𝑖 + 1 + (
Y𝑝6 𝑖 − 𝑛6 𝑖
(2.32)
Here, the chasing method is chosen to solve the equations. This is because the elements of
coefficient matrixes are very small, and some of them are approximately zero, where it is
possible to make the coefficient matrixes become a tri-diagonal matrix. In addition, these
matrixes meet the limitation conditions of the chasing method.
Queensland University of Technology
37
Solve the Poisson’s equation in an independent loop until 𝜑 meets the accuracy condition.
After getting new potential values, we substitute them into the continuity equations to calculate
n and p respectively. When n is calculated, maintain 𝜑 and p as constants. Using the first
order Taylor expansion at 𝑛K^�6 , 𝑛K6, 𝑛K]�6 the Eq.2.26 becomes:
𝐹L =TGI-
TL K^�∆𝑛 𝑖 − 1 + TGI-
TL K∆𝑛 𝑖 + TGI-
TL K]�∆𝑛 𝑖 + 1 + 𝐹L6 = 0 (2.33)
TGI-
TL K^�= ³I
�U[j K^� ^j K ]
¼½¾ ¶ H¡´ ¡¶ H:¥
^�
TGI-
TL K= − ³I
�U𝜑 𝑖 − 1 − 𝜑 𝑖
¼½¾ ¶ H¡´ ¡¶ H:¥
¼½¾ ¶ H¡´ ¡¶ H:¥
^�− ³I
�Uj K ^j K]�
¼½¾ ¶ H ¡¶ H·´:¥
^�− 𝛾𝑝6(𝑖)
TGI-
TL K]�= ³I
�U[𝜑 𝑖 − 𝜑(𝑖 + 1)]
¼½¾ [¶ H ¡¶(H·´):¥
]
¼½¾ ¶ H ¡¶ H·´:¥
^�
𝐹L6 =TGI-
TL K^�𝑛6 𝑖 − 1 + [ TGI
-
TL K+ 𝛾𝑝6(𝑖)]𝑛6 𝑖 + TGI-
TL K]�𝑛6 𝑖 + 1 + 𝐺 𝑖 − 𝑅(𝑖)
(2.34)
The first order Taylor expansion form of Eq.2.28 can be obtained by using the same approach,
as shown previously. The results are shown in Eq.2.35 and Eq. 2.36.
𝐹e =TGf-
Te K^�∆𝑝 𝑖 − 1 + TGf-
Te K∆𝑝 𝑖 + TGf-
Te K]�∆𝑝 𝑖 + 1 + 𝐹e6 = 0 (2.35)
TGf-
Te K^�= −³f
�U[𝜑 𝑖 − 1 − 𝜑(𝑖)]
¼½¾ [¶ H¡´ ¡¶(H):¥
]
¼½¾ ¶ H¡´ ¡¶ H:¥
^�
TGf-
Te K= ³f
�U𝜑 𝑖 − 𝜑 𝑖 + 1
¼½¾ ¶ H ¡¶ H·´:¥
¼½¾ ¶ H ¡¶ H·´:¥
^�− ³f
�Uj K^� ^j K
¼½¾ ¶ H¡´ ¡¶ H:¥
^�+ 𝛾𝑛6(𝑖)
TGf-
Te K]�= −³f
�U[j K ^j(K]�)]
¼½¾ ¶ H ¡¶ H·´:¥
^�
𝐹e6 =TGf-
Te K^�𝑝6 𝑖 − 1 + TGf-
Te K− 𝛾𝑛6 𝑖 𝑝6 𝑖 + TGf-
Te K]�𝑝6 𝑖 + 1 − 𝐺 𝑖 + 𝑅(𝑖)
(2.36)
Queensland University of Technology
38
After solving the continuity equations for electrons and holes, it is needed to be estimated
whether n and p meet the accuracy condition. If these solutions meet the accuracy condition,
the iteration loop is stopped and Jn and Jp are calculated. On the contrary, the new carrier
densities (n and p) are substituted into the Poisson’s equation and a new iteration is started until
all the solutions meet the accuracy condition. The Gummel iteration method flow diagram is
shown in Fig.2.02.
Fig.2.02 Gummel iteration method flow diagram
Queensland University of Technology
39
2.4 Results and Discussion
The parameters used in the simulation are summarised in Table.2.1. The parameters are
extracted from literature [63-66].
Table.2.1 Parameters used in the simulation
Parameters Symbol Value
Perovskite thickness L 100 nm
Band gap 𝐸��e 1.55 eV
Temperature T 300 K
Electron mobility 𝜇L 1 cm2/Vs
Hole mobility 𝜇e 2 cm2/Vs
Density of state Nc 2.2e18 cm-3
Relative dielectric permittivity 𝜀 30
Absorption coefficient 𝛼 5.7e4 cm-1
2.4.1 The Effect of Effective Density of State
Effective density of state is the number of states, which are available to be occupied, per energy
interval around the bottom of the conduction band and the top of the valance band. It assumes
that the effective density of states for conduction band and valance are the same in this
simulation. According to Eq.2.37, it is clear that the effective density of state depends on
effective mass, which is determined by the material, and the value of Nc can affect the density
of electrons and holes directly. So, in order to investigate the influence of effective density of
state for the device performance, 𝑁# is varied from 1018 cm-3 to 1021 cm-3, and other
Queensland University of Technology
40
parameters remain the same. The simulation result of I-V characteristics is shown in Fig.2.03,
and the values of Isc, Voc, and FF are summarised in Table.2.2
𝑁# = 2 (u¿y%&')ÀU
�-À (2.37)
where, 𝑁# is effective density of state, m is effective mass, 𝑘2 is the Boltzmann constant, T
is temperature, ℎ6 is the Planck constant.
Fig.2.03 The I-V curves for different value of Nc
Table.2.2 The values of Isc, Voc, and FF for different Nc
Nc/cm-3 Isc/mA/cm-2 Voc/V FF
1018 11.4 1.072 0.778
1019 11.2 0.939 0.761
1020 10.9 0.833 0.738
1021 10.7 0.701 0.747
It can be seen that Nc has a significant influence on the device performance in particular on Voc,
when it is changed. With the increase of Nc, the short-circuit current density slowly decreases,
Queensland University of Technology
41
the open-circuit voltage increases rapidly, and the fill factor changes irregularly. In order to see
the difference clearly and explain these simulation phenomena, the following discussions focus
on the simulation results of Nc=1018 cm-3 and Nc=1020 cm-3.
Since the carrier and potential distribution can reflect the electrical properties of a device,
therefore, the simulation based on these two aspects are carried out. Fig.2.04 and Fig.2.05 show
the carriers distribution in the device for Nc=1018 cm-3 and Nc=1020 cm-3 respectively. It can be
seen that the densities of minority carriers almost remain the same when Nc is changed from
Nc=1018 cm-3 to Nc=1020 cm-3 but the majority carriers increase significantly. It means that the
space charges go up with the increase of Nc, and the space charge effect becomes more
significant. This can be explained by the Poisson’s equation with the electric field form
TopHÁÂÃTV
= (Y(𝑛 − 𝑝).
Fig.2.04 Charge carriers’ distribution for Nc=1018 cm-3
Queensland University of Technology
42
Fig.2.05 Charge carriers’ distribution for Nc=1020 cm-3
Fig.2.06 illustrates the potential distribution for Nc=1018 cm-3 and Nc=1020 cm-3 respectively.
When Nc is small, the potential distribution is nearly linear. When Nc is large, the potential
changes rapidly at both sides of the device but it changes minimal in the middle of the device.
Because the electric field intensity is the derivative of the potential, so it can be easily seen that
Nc=1020 cm-3 has smaller electric field intensity than Nc=1018 cm-3 in the greater part of the
device. The voltage corresponding to the value of current is zero, and called open circuit
voltage. In this case, the value of the electric field intensity is zero. Therefore, carriers cannot
move in a controlled way in the same direction without the effect of the electric field.
Consequently, there is no current output. When the applied voltage is changed from 0 V to1.2
V with the same interval value, the smaller the electric field intensity, the easier it reduces to
zero. This is the reason why the open-circuit voltage goes down with the increase of Nc.
Queensland University of Technology
43
Fig.2.06 Potential distribution for Nc=1018 cm-3 and Nc=1020 cm-3
Although the electric field intensities are strong at both sides of device, it exists at a short
distance. The carriers move under a small electric field intensities in a long distance. It means
that the carriers cannot easily move through the device, and this is the reason why the short-
circuit current decreases with the increase of Nc.
2.4.2 The Effect of Relative Dielectric Permittivity
According to Poisson’s equation and continuity equations, it can be found that relative
dielectric permittivity (𝜀) is one of the crucial parameters affecting the potential and carrier
densities. Therefore, the simulation results based on relative dielectric permittivity dependent
performance are discussed in this section in order to understand the effect of relative dielectric
permittivity on the device performance. These results are obtained by varying 𝜀 from 5 to 9
Queensland University of Technology
44
without changing other parameters. Fig.2.07 shows the short-circuit current density increases
from 11.226 mA/cm2 to 11.289 mA/cm2 with the increase of 𝜀.
Fig.2.07 The effect of relative dielectric permittivity on current
This has the same tendency for open-circuit voltage and fill factor. Open-circuit voltage
increases gradually from 1.012 V to 1.0358 V, and fill factor increases from 0.7416 to 0.750,
as shown in Fig.2.08. But overall, the effect of relative dielectric permittivity on Isc, Voc and FF
is small.
Fig.2.08 The effect of relative dielectric permittivity on voltage and fill factor
Fig.2.09 depicts net generation rate for 𝜀=5 and 𝜀=9. It is clear that high relative dielectric
permittivity has a relatively large net generation rate. In this simulation, if the absorption
coefficient is fixed, the generation rate is the function of position. Therefore, a large net
Queensland University of Technology
45
generation rate can be obtained due to the reduction of recombination loss. This proves that
high relative dielectric permittivity can reduce the bimolecular recombination. It is also one
factor why short-circuit current goes up slightly with the increase of 𝜀.
Fig.2.09 Net generation rate for 𝜀=5 and 𝜀=9
The potential distribution for 𝜀=5 and 𝜀=9 is shown in Fig.2.10. The large value of 𝜀 has a
low rate of change for potential, compared with a small value of 𝜀 at both sides of the device,
which means that the electric field intensities are weak at these parts, but electric field
intensities are strong in the middle part of device. Therefore, the open-circuit voltage would
increase with the increase of 𝜀. Although the electric field is weak near the electrodes, it is
strong in the greater part of device, which is beneficial for the carriers moving though the
device, and this is another important factor to explain why short-circuit current goes up with
the increase of 𝜀.
Queensland University of Technology
46
Fig.2.10 Potential distribution for 𝜀=5 and 𝜀=9
2.4.3 The Effect of Band Gap Energy
Many reports show that the band gap energy of MAPbI3 is between 1.45 eV and 1.7 eV
[64,65,66]. As the material used as the absorber layer, the band gap energy of MAPbI3 has a
significant influence on device performance. Here, the effect of band gap energy on device
performance is investigated. The value of band gap energy is changed from 1.45 eV to 1.6 eV.
Fig.2.11 shows the I-V characteristic for different values of band gap energy. It is clear that
both short-circuit current and open-circuit voltage increases with the increment of band gap
energy, and the change of open-circuit voltage is more significant to compare with short-circuit
current. Open-circuit voltage increases from 0.97 V to 1.106 V, and the value of short-circuit
current rises from 11.34 mA/cm2 to 11.4 mA/cm2. The influence of band gap energy on fill
Queensland University of Technology
47
factor is displaced in Fig.2.12. With the increase of the value of band gap energy, the fill factor
decreases from 0.773 to 0.68.
Fig.2.11 The I-V curves for different value of band gap energy
Fig.2.12 The effect of band gap energy on fill factor
Queensland University of Technology
48
In the actual situation, short-circuit current should decrease slightly due to the reduction of
light absorption caused by the increase of the band gap energy. In this simulation, G is set as
constant following the exponential distribution, so it does not consider the reduction of light
absorption. Therefore, the enhancement of short-circuit current can be explained as the result
of the increase of the net generation rate. Broadening of the band gap energy can strengthen
the electric field, as shown in Fig2.13 It can be seen that the different potentials between the
electrodes increase with the increment of band gap energy, and the change rate of potential in
the device has the same tendency. This is the reason why open-circuit voltage goes up.
Fig.2.13 Potential distribution for different band gap energy
Queensland University of Technology
49
2.5 Conclusion
This chapter mainly introduces a numerical model for perovskite solar cell and a specific
computation process for Poisson’s equation and continuity equations. The simulation results
are obtained by using the Matlab program. The influence of effective density of state, relative
dielectric permittivity, and band gap energy on device performance is investigated. Through
the analysis of the simulation results, it is found that a large value of effective density state can
decrease open-circuit voltage significantly, meanwhile it can also decrease the short-circuit
current slightly. Thus, in order to obtain good device performance, it is necessary to choose a
small value of effective density state material or control the density of state of the material used
in the absorber layer. The device model with large value of relative dielectric permittivity
shows a high open-circuit voltage and short-circuit current performance, and the simulation
results also prove that high relative dielectric permittivity can reduce the bimolecular
recombination. In addition, with the increment of relative dielectric permittivity, the value of
fill factor increases slightly. Broader band gap energy is benefit for the open-circuit voltage
and short-circuit current, while it is not good for the fill factor.
Queensland University of Technology
50
CHAPTER 3. SCAPS SIMULATION FOR HTM LAYER IN PSCs
3.1 Introduction
The emergence of perovskite solar cells has attracted a lot of attention due to their high
efficiency. However, the unstable performance and expensive hole transport materials limit its
development [67]. Generally, organic materials (Spiro-OMETAD, PEDOT:PPS, MEH-PPV),
which have relatively high cost and low hole mobility, are used as HTM in perovskite solar
cells [68-69]. Some reports also showed that the HTM layer affects the device performance
and the degradation of device performance was observed by using organic material as HTM
[70-72]. Therefore, in order to commercialise the perovskite solar cells and dominate the
photovoltaic industry, it is necessary to find an alternative HTM layer for PSCs. Inorganic
materials are considered as promising materials for the HTM layer because of their favourable
characteristics, such as high mobility of hole, low cost of fabrication, better chemical stability,
and appropriate valance band position [73-75].
Numerical simulation can give assistance to studying solar cell devices. SCAPS-1D has been
developed by the University of Gent and has been widely used for thin film solar cells.
Perovskite solar cells with a planar structure have similar structure and excitation type
comparing with thin film solar cells. Therefore, many simulation-based PSCs have been also
built by using SCAPS. Hossain and co-workers built a simulation based on PSCs with ZnO as
the ETM layer by using SCAPS. They showed that the device with ZnO as the ETM layer can
obtain a high efficiency of 22.83% compared with TiO2 as the ETM layer [69]. The comparison
of different organic hole transport materials including NPB, MEH-PPV, PEDOTPSS, Spiro-
Queensland University of Technology
51
OMeTAD and P3HT was reported by Karimi and colleagues by using SCAPS. They showed
that the device with Spiro-OMeTAD as the HTM layer achieves the best performance [66]. A
simulation of lead-free CH3NH3SnI3 perovskite solar cells was provided by Du and colleagues.
They showed that the device based on lead-free CH3NH3SnI3 as the absorber layer obtained an
efficiency of 23.36% [76]. Minemoto and Murata investigated the effect of work function on
device performance by using SCAPS. Their simulation results illustrated that work function
matching is necessary to improve the device performance especially to Voc [77].
In this chapter, a numerical model based on a perovskite solar cell with the structure of
glass/FTO/TiO2/CH3NH3PbI3/HTM/Au is built by using SCAPS-1D software. The influences
of HTM layer characteristics on perovskite solar cells based on this model are discussed here.
Furthermore, the simulation results of two inorganic materials, Cu2O and CuO, as candidates
for the HTM layer, are also shown here to compare with Spiro-OMETAD, a standard HTM
used in PSCs.
Queensland University of Technology
52
3.2 Device Model
The device model is built by SCAPS-1D software. The configuration of the perovskite solar
cell used in this simulation is showed in Fig.3.1. It is a planar heterojunction structure
consisting of four layers. Following the direction of incident light, the device configurations
are front contact(TCO), SnO2:F(FTO), n type electron transportation layer(TiO2), perovskite
material layer(CH3NH3PbI3), p type hole transportation layer(Spiro-OMETAD,CuO,Cu2O),
and back contact (Au).
Fig.3.1 The structure of perovskite solar cell in the simulation
The main idea of SCAPS-1D is to solve Poisson’s equation and continuity equations for
electrons and holes:
𝜕u
𝜕𝑥u 𝜑 𝑥 =𝑞𝜀 [𝑛 𝑥 − 𝑝 𝑥 − 𝑁\](𝑥) + 𝑁O^(𝑥) − 𝑝_ 𝑥 + 𝑛_(𝑥)]
𝑞 TLT_= TbI
TV+ 𝑞𝐺 − 𝑞𝑅
𝑞 TeT_= − Tbf
TV+ 𝑞𝐺 − 𝑞𝑅
Queensland University of Technology
53
𝐽L = 𝑞𝑛𝜇L𝜕𝜑𝜕𝑥 + 𝑞𝐷L
𝜕𝑛𝜕𝑥
𝐽e = −𝑞𝑝𝜇e𝜕𝜑𝜕𝑥 + 𝑞𝐷e
𝜕𝑝𝜕𝑥
where, 𝜑 is the potential, q is the elementary charge, 𝜀 is the permittivity, n is the density of
free electron, p is the density of free hole, 𝑁\] is the ionised donor-like doping density, 𝑁O^
is the ionized acceptor-like doping density, 𝑝_ is the trapped hole density, 𝑛_ is the trapped
electron density, G is the optical generation rate, R is the recombination rate, 𝐷L is the electron
diffusion coefficient, 𝐷e is the hole diffusion coefficient, 𝜇L is the electron mobility, and 𝜇e
is the hole mobility.
Here, it needs to set characteristics’ parameters for each layer and device parameters in SCAPS.
The parameters used in the simulation are adopted from literatures, experimental work and
theoretical study. The main parameters of materials and defects are summarised in Table.3.1(a)
and Table.3.1(b) [63,65,78,79,80]:
Table.3.1(a) Materials parameters used in the numerical analysis
Parameters FTO TiO2 CH3NH3PbI3
Thickness(nm) 400 50 450
Acceptor density(𝑐𝑚^Æ) 0 0 0
Donor density(𝑐𝑚^Æ) 1019 1017 1013
Bandgap energy(eV) 3.5 3.26 1.55
Electron affinity(eV) 4 4.2 3.9
Relative dielectric permittivity 9 10 6.5
Mobility of electron 20 20 2
Mobility of hole 10 10 2
Defect density(𝑐𝑚^Æ) 1015 1015 2.5*1013
Queensland University of Technology
54
Table.3.1(b) Materials parameters used in the numerical analysis
Parameters Spiro-OMETAD CuO Cu2O
Thickness(nm) 150 150 150
Acceptor density(𝑐𝑚^Æ) 1018 1018 1018
Donor density(𝑐𝑚^Æ) 0 0 0
Bandgap energy(eV) 2.9 1.3 2.17
Electron affinity(eV) 2.2 4.07 3.2
Relative dielectric permittivity 3 18.1 7.11
Mobility of electron 10-4 0.1 80
Mobility of hole 10-4 0.1 80
Defect density(𝑐𝑚^Æ) 1015 1015 1015
Other parameters of materials such as effective density of state of conduction band and
effective density of state of valance band are set as 2.2*1018 cm-3, electron thermal velocity and
hole thermal velocity are set as 1.8*1018 cm-3, 107cm/s, respectively. In order to simplify the
device model, the absorption coefficient for each layer is set as 105 cm-1, and the defect type is
set as neutral, which means that the defect can lead to the Shockley-Read-Hall recombination
but cannot contribute to the space charge. The energetic distribution uses the Gaussian model,
and its characteristic energy uses the default value.
The main device parameters are shown in Table.3.2. The standard AM1.5G spectrum is used
in this simulation. The values of working point and numerical setting use the default value. The
scanning voltage is set from 0 V to 1.2 V. All the simulations will operate under these
conditions.
Queensland University of Technology
55
Table.3.2 Device parameters used in the numerical analysis
Back contact electrical properties
Thermionic emission/surface recombination
velocity of electron(cm/s)
105
Thermionic emission/surface recombination
velocity of hole(cm/s)
107
Metal(Au) work function(eV) 5.1
Front contact electrical properties
Thermionic emission/surface recombination
velocity of electron(cm/s)
107
Thermionic emission/surface recombination
velocity of hole(cm/s)
105
Work function of TCO(eV) 4.4
Operation temperature(k) 300
Queensland University of Technology
56
3.3 Results and Discussion
3.3.1 The Influence of HTM Layer Characteristics
The investigation of the influence of HTM layer characteristics on the device performance is
based on Spiro-MeOTAD, which is widely used in current perovskite solar cells as a standard
material. Hole mobility and valance band gap offset are two crucial parameters, which need to
be considered as criteria in the selection of hole transport materials. Here, these two parameters
are investigated in order to understand how they affect device performance.
3.3.1.1 Hole mobility
The influence of defect density in the HTM layer on device performance is not significant, so
a HTM layer without defect is applied here in order to see the effect of performance only caused
by hole mobility.
Queensland University of Technology
57
Fig.3.2 The device performance for different hole mobility
Fig.3.2 shows the simulation results by changing the value of hole mobility from 1e-4 cm2/Vs
to 1e-2 cm2/Vs. The value of Isc, FF and PCE gradually rises with the increase of hole mobility,
while the value of Voc decreases slightly. When the hole mobility reaches to 5e-3 cm2/Vs, the
change of Isc, Voc, FF and PCE tends to become stable, at this time, Isc is nearly 23.5 mA/cm2,
Voc is approximately to1.3141 V, FF closes to 65% and PCE is around 17%. A high hole
mobility can lead to the improvement of the conductivity and performance of the device. The
hole mobility of Spiro-MeOTAD is low, which is not good for the device performance, and
this is the reason why p-type doping to enhance hole mobility in the HTM layer is necessary.
Therefore, materials with large hole mobility should be considered for the selection of hole
transport material.
3.3.1.2 Band gap offset
As we know, the HTM layer is used to transfer holes from the perovskite layer to the HTM
layer while blocking the back charges flow from HTM layer to perovskite layer, so an
appropriate valance band position is important in the device design. The influences of valance
band offset on the device performance are investigated in this section. It is assumed that if
Eabsorber_vc-EHTM_vc>0, the band offset is called a positive band gap offset. Otherwise, the band
offset is called a negative band gap offset. Different values of band gap offset can be obtained
by changing electron affinity. Both positive and negative valance band offsets are varied from
0.1 eV to 0.5 eV by every 0.1 eV. The simulation results are showed in Fig.3.3 and Fig.3.4
Queensland University of Technology
58
Fig.3.3 Device performance for different values of positive valance band offset
Fig.3.3 shows the simulation results from different positive valance band gap offsets. In this
case, the valance band position of the HTM layer is lower than the perovskite layer. With the
increase of band gap offset, the higher potential barrier is formed between these two layers.
The high potential barrier means that the holes generated in the perovskite layer are difficult to
move to the HTM layer. From Fig.3.4, it can be seen that when the value of the positive valance
band gap offset goes up, the values of Isc and Voc are almost unchanged, while the value of FF
decreases, which causes the reduction of PCE. Furthermore, the reduction of PCE becomes
significant when the positive band gap offset is larger than 0.2 eV.
Fig.3.4 Device performance for different values of negative valance band offset
Queensland University of Technology
59
Fig.3.4 illustrates the simulation results from different negative valance band gap offsets. In
this situation, the valance band position of the HTM layer is higher than the perovskite layer.
It seems that the device can achieve a good performance with the enhancement of negative
valance band gap offset, because of the contribution from the electric field to the holes.
However, the results present an opposite phenomenon. It can be seen that the Isc is nearly kept
the same, but PCE, FF and Voc of the device go down with the increase of negative valance
band gap offsets. Moreover, the effect of negative valance band gap offset on Voc and PCE are
significant when the negative valance band gap offset is larger than 0.2eV.
3.3.2 Comparison for Different Hole Transporting Materials
Copper oxide (CuO) and Cuprous oxide (Cu2O) are P-type semiconductors. Both of them have
a relatively wide band gap. The band gap energy of CuO is 1.3 eV [81]. The band gap energy
of Cu2O is 2.17 eV [82]. The energy level diagram of CuO and Cu2O in the device is shown in
Fig.3.5. From Chapter 3.3.1, we know that in order to reduce energy loss and obtain high device
performance, it is expected to have a small valance band offset (-0.2 eV to 0.2 eV) between the
HTM layer and the perovskite layer. Thus, it can be seen from Fig.3.5, that both CuO and Cu2O
have an appropriate valance band position, and the values of valance band offset of CuO and
Cu2O are less than 0.1 eV. In addition, Cu2O exhibits a high hole mobility up to 80 cm2V-1s-1,
and CuO also has a relative high hole mobility 0.1 cm2V-1s-1 compared with organic materials
(Spiro-MeOTAD 0.0001 cm2V-1s-1). Therefore, for all these reasons that have been mentioned,
it is believed that CuO and Cu2O are the probable alternative materials that can be utilised in
Queensland University of Technology
60
HTM layers. The simulations are carried out by using three types of hole transporting materials
(CuO, Cu2O, and Spiro-MeOTAD) here.
Fig.3.5 Energy level diagram of CuO and Cu2O in the device
The simulation results are shown in Fig.3.6 and Table.3.3. Fig.3.6 illustrates the I-V
characteristics for the perovskite solar cell with different HTM layers including CuO, Cu2O
and Spiro-MeOTAD. Table.3.3 presents the results of device performance including Isc, Voc,
FF and PCE for different HTM layers. From the simulation results, it can be easily seen that
the device with CuO or Cu2O as HTM shows a better performance than Spiro-MeOTAD.
Especially using Cu2O as HTM, the PCE of the device reaches 21.87%. Furthermore, relative
high Isc 24.01 mA/cm3 and Voc 1.128 V are obtained by using Cu2O, because of its appropriate
band gap energy, valance band position and high hole mobility. These simulation results
provide evidence to prove that Cu2O and CuO are probable materials to be used as hole
transport material.
TiO2
CH3NH3PbI3 CuO
Cu2O
-4.4
-4.2
-7.46
-3.9
-5.45 -5.37
-4.07
-5.37
-3.2
-5.1
E(eV)
Queensland University of Technology
61
Fig.3.6 I-V characteristics for different HTMs
Table.3.3 Device performance for different HTMs
HTMs Isc mA/cm3 Voc V FF % PCE %
Spiro-MeOTAD 23.93 1.13 71.24 19.32
CuO 24.04 1.07 77.05 19.83
Cu2O 24.01 1.128 80.67 21.87
Queensland University of Technology
62
3.4 Conclusion
In this chapter, a device model with the structure of glass/FTO/TiO2/CH3NH3PbI3/HTM/Au
has been created by using SCAPS-1D software. The effects of HTM layer characteristics,
including hole mobility and valance band offset, have been discussed. Furthermore, the device
performance by using CuO, Cu2O and Spiro-MeOTAD as hole transport materials has also
been illustrated. The simulation results show that high value of hole mobility in the HTM layer
is good for the device performance. For Spiro-MeOTAD, when hole mobility reaches 5e-3
cm2/Vs, PCE closes to the maximum value, and the change of it tends to become stable with
the continuing increase of hole mobility. The value of valance band gap offset shows a crucial
influence on the device performance. Both positive band gap offset and negative band gap
offset can affect the device performance significantly. The results reveal that the band gap
offset between -0.2 eV and 0.2 eV can obtain a good device performance. Through the
comparison of CuO, Cu2O and Spiro-MeOTAD, it is found that the inorganic materials CuO
and Cu2O used as HTM, exhibit a better performance than Spiro-MeOTAD, and Cu2O achieves
a superior performance with a PCE of 21.87%.
Queensland University of Technology
63
CHAPTER 4. SUMMARY AND FUTURE WORKS
4.1 Summary
In this thesis, I illustrate the influence of effective density state, relative dielectric permittivity,
band gap energy on the device performance by using matlab model firstly. The simulation
results show that a large value of effective density state can decrease open-circuit voltage
significantly, meanwhile it can also decrease the short-circuit current slightly. A large value of
relative dielectric permittivity can lead to a large Voc and Isc, but the effect is not significant.
Broader band gap energy is benefit for the open-circuit voltage and short-circuit current, while
it decreases the fill factor. Then I show the influence of hole transport material layer
characteristics, including hole mobility and band gap offset, on the performance of PSCs and
comparing device performance with different HTM including CuO, Cu2O and Spiro-
MeOTAD by using SCAPS model. The simulation results show that high value of hole
mobility in the HTM layer is good for the device performance. The value of valance band gap
offset shows a crucial influence on the device performance. The device can obtain high
performance when the value of the valance band gap offset is between -0.2 eV and 0.2 eV. In
addition, the device shows a better performance by using inorganic materials (CuO, Cu2O) as
the HTM layer, than by using spiro-MOeTAD, especially to Cu2O with a PCE of 21.87%.
4.2 Future Works
This report gives two numerical models based on perovskite solar cells. One was built and
solved by myself, another was created by using SCAPS-1D directly. The performance effect
factors in the perovskite layer and HTM layer including effective density state, relative
Queensland University of Technology
64
dielectric permittivity, band gap energy, hole mobility, and valance band gap offset have been
discussed. However, there are still some problems that need to be solved. Firstly, the perovskite
solar cell model built by myself only considers perovskite layers with some reasonable
assumptions to simplify the physical equations. This model can only get a qualitative
explanation for limited effect factors. Thus, in order to investigate more factors with a
quantitative relation, the physical model needs to consider the effects from HTM layer and
ETM layer. Through the optimisation of the device model, the simulation results that can be
obtained are more accurate and valuable. Secondly, there are only two factors in the HTM layer
that have been investigated; other factors such as the thickness of device, doping density and
carriers’ lifetime are also worthy to be studied. Finally, this report only presents two kinds of
inorganic materials (CuO and Cu2O) used as HTM layer, but there are lots of inorganic
materials in the world that could be considered as likely materials to be used as the HTM layer,
such as NiO, CuI, and CuSCN. Consequently, there needs to be some further study and
researches into this area.
Queensland University of Technology
65
REFERENCES
[1] World Energy Council, editor. World Energy Resources 2016 Survey. World Energy
Council, 2016.
[2] Shafiee, S., & Topal, E. (2009). When will fossil fuel reserves be diminished? Energy
policy, 37(1), 181-189.
[3] Barbir, F., Veziroǧlu, T. N., & Plass Jr, H. J. (1990). Environmental damage due to fossil
fuels use. International journal of hydrogen energy, 15(10), 739-749.
[4] Armaroli, N., & Balzani, V. (2007). The future of energy supply: challenges and
opportunities. Angewandte Chemie International Edition, 46(1-‐2), 52-66.
[5] Halls, J. J. M., Walsh, C. A., Greenham, N. C., Marseglia, E. A., Friend, R. H., Moratti, S.
C. & Holmes, A. B. Efficient photodiodes from interpenetrating polymer networks. Nature
376, 498 – 500 (1995).
[6] World Energy Council, editor. World Energy Resources 2016exective summary. World
Energy Council, 2016.
[7] D. M. Chapin, C. S. Fuller, and G. L. Pearson. A New Silicon p-n Junction Photocell for
Converting Solar Radiation into Electrical Power. J. Appl. Phys., 25(5):676, 1954.
[8] https://janaloka.com/sejarah-teknologi-panel-surya/
[9] Bragagnolo, J. A., Barnett, A. M., Phillips, J. E., Hall, R. B., Rothwarf, A. L. L. E. N., &
Meakin, J. D. (1980). The design and fabrication of thin-film CdS/Cu 2 S cells of 9.15-
percent conversion efficiency. IEEE Transactions on Electron Devices, 27(4), 645-651.
[10] https://en.wikipedia.org/wiki/Timeline_of_solar_cells
[11] O’regan, B., & Grätzel, M. (1991). A low-cost, high-efficiency solar cell based on dye-
sensitized colloidal TiO2 films. Nature, 353(6346), 737.
[12] Würfel, P. (2005). Physics of solar cells (Vol. 1, p. 139). Weinheim: Wiley-vch.
[13] Villalva, M. G., & Gazoli, J. R. (2009). Comprehensive approach to modeling and
simulation of photovoltaic arrays. Power Electronics, IEEE Transactions on,24(5), 1198-
1208.
[14] Lun, S. X., Du, C. J., Yang, G. H., Wang, S., Guo, T. T., Sang, J. S., & Li, J. P. (2013).
An explicit approximate I–V characteristic model of a solar cell based on padé
approximants. Solar energy, 92, 147-159.
Queensland University of Technology
66
[15] Snaith, H. J. Perovskites: the emergence of a new era for low-cost, high- efficiency solar
cells. J. Phys. Chem. Lett. 4, 3623–3630 (2013).
[16] Zhicai He, Chengmei Zhong, Shijian Su, Miao Xu, Hongbin Wu, and Yong Cao.
Enhanced power-conversion eciency in polymer solar cells using an inverted device
structure. Nat. Photonics, 6(9):593–597, aug 2012.
[17] Lichen Zhao, Deying Luo, Jiang Wu, Qin Hu, Wei Zhang, Ke Chen, Tang- hao Liu, Yi
Liu, Yifei Zhang, Feng Liu, Thomas P. Russell, Henry J. Snaith, Rui Zhu, and Qihuang
Gong. High-Performance Inverted Planar Heterojunc- tion Perovskite Solar Cells Based on
Lead Acetate Precursor with Eciency Exceeding 18%. Adv. Funct. Mater., pages 3508–3514,
2016.
[18] Thomas Kietzke. Recent Advances in Organic Solar Cells. Adv. Optoelectron., 2007:1–
15, 2007.
[19] Kim, H.-S., Im, S. H. & Park, N.-G. Organolead halide perovskite: new horizons in solar
cell research. J. Phys. Chem. C 118, 5615–5625 (2014).
[20] Potscavage Jr, W. J. (2010). Physics and engineering of organic solar cells.
[21] Marchioro, A. et al. Unravelling the mechanism of photoinduced charge transfer
processes in lead iodide perovskite solar cells. Nature Photon. 8, 250–255 (2014).
[22] Perovskite wikipedia. https://en.wikipedia.org/wiki/Perovskite, Ac- cessed: 2016-06-21.
[23] Wang, Z. L. & Kang, Z. C. (1998). Functional and Smart Materials. New York: Plenum
Press.
[24] Li, C., Soh, K. C. K., & Wu, P. (2004). Formability of ABO3 perovskites. Journal of
Alloys and Compounds, 372(1-2), 40-48.
[25] Li, C., Lu, X., Ding, W., Feng, L., Gao, Y., & Guo, Z. (2008). Formability of ABX3 (X=
F, Cl, Br, I) Halide Perovskites. Acta Crystallographica Section B: Structural Science, 64(6),
702-707.
[26] Xing, G., Mathews, N., Sun, S., Lim, S. S., Lam, Y. M., Grätzel, M., ... & Sum, T. C.
(2013). Long-range balanced electron-and hole-transport lengths in organic-inorganic
CH3NH3PbI3. Science, 342(6156), 344-347.
[27] Burschka, J., Pellet, N., Moon, S. J., Humphry-Baker, R., Gao, P., Nazeeruddin, M. K.,
& Grätzel, M. (2013). Sequential deposition as a route to high-performance perovskite-
sensitized solar cells. Nature, 499(7458), 316.
Queensland University of Technology
67
[28] Kim, H. S., Lee, C. R., Im, J. H., Lee, K. B., Moehl, T., Marchioro, A., ... & Grätzel, M.
(2012). Lead iodide perovskite sensitized all-solid-state submicron thin film mesoscopic solar
cell with efficiency exceeding 9%. Scientific reports, 2, 591.
[29] Toshniwal, A., & Kheraj, V. (2017). Development of organic-inorganic tin halide
perovskites: A review. Solar Energy, 149, 54-59.
[30] Berhe, T. A., Su, W. N., Chen, C. H., Pan, C. J., Cheng, J. H., Chen, H. M., ... & Hwang,
B. J. (2016). Organometal halide perovskite solar cells: degradation and stability. Energy &
Environmental Science, 9(2), 323-356.
[31] Service, R. F. Perovskite solar cells keep on surging. Science 344, 458 (2014).
[32] Kojima A, Teshima K, Miyasaka T, et al. Novel Photoelectrochemical cell with
mesoscopic electrodes sensitized by lead-halide compounds. Meeting Abstracts: The
Electrochemical Society, 2006: 397-397.
[33] Kojima, A., Teshima, K., Shirai, Y., & Miyasaka, T. (2009). Organometal halide
perovskites as visible-light sensitizers for photovoltaic cells. Journal of the American
Chemical Society, 131(17), 6050-6051.
[34] Im, J. H., Lee, C. R., Lee, J. W., Park, S. W., & Park, N. G. (2011). 6.5% efficient
perovskite quantum-dot-sensitized solar cell. Nanoscale, 3(10), 4088-4093.
[35] Kim, H. S., Lee, C. R., Im, J. H., Lee, K. B., Moehl, T., Marchioro, A., ... & Grätzel, M.
(2012). Lead iodide perovskite sensitized all-solid-state submicron thin film mesoscopic solar
cell with efficiency exceeding 9%. Scientific reports, 2, 591.
[36] Lee, M. M., Teuscher, J., Miyasaka, T., Murakami, T. N., & Snaith, H. J. (2012).
Efficient hybrid solar cells based on meso-superstructured organometal halide
perovskites. Science, 1228604.
[37] Malinkiewicz, O. et al. Perovskite solar cells employing organic charge- transport
layers. Nature Photon. 8, 128–132 (2014).
[38] Burschka, J. et al. Sequential deposition as a route to high-performance perovskite-
sensitized solar cells. Nature 499, 316–319 (2013).
[39] Liu, M., Johnston, M. B., & Snaith, H. J. (2013). Efficient planar heterojunction
perovskite solar cells by vapour deposition. Nature, 501(7467), 395.
[40] Green, M. A., Ho-Baillie, A., & Snaith, H. J. (2014). The emergence of perovskite solar
cells. Nature Photonics, 8(7), nphoton-2014.
[41] Saliba, M., Matsui, T., Seo, J. Y., Domanski, K., Correa-Baena, J. P., Nazeeruddin, M.
K., ... & Grätzel, M. (2016). Cesium-containing triple cation perovskite solar cells: improved
Queensland University of Technology
68
stability, reproducibility and high efficiency. Energy & Environmental Science, 9(6), 1989-
1997.
[42] Yang, W. S., Park, B. W., Jung, E. H., Jeon, N. J., Kim, Y. C., Lee, D. U., ... & Seok, S.
I. (2017). Iodide management in formamidinium-lead-halide–based perovskite layers for
efficient solar cells. Science, 356(6345), 1376-1379.
[43] Simya OK, Mahaboobbatcha A, Balachander K. Compositional grading of CZTSSe
alloy using exponential and uniform grading laws in SCAPS-ID simulation. Superlattices and
Microstructures. 2016 Apr 1;92:285-93.
[44] Houshmand M, Zandi MH, Gorji NE. SCAPS modeling for degradation of ultrathin
CdTe films: materials interdiffusion. JOM. 2015 Sep 1;67(9):2062-70.
[45] Gretener C, Perrenoud J, Kranz L, Cheah E, Dietrich M, Buecheler S, Tiwari AN. New
perspective on the performance stability of CdTe solar cells. Solar Energy Materials and
Solar Cells. 2016 Mar 31;146:51-7.
[46] Chihi A, Boujmil MF, Bessais B. Investigation on the Performance of CIGS/TiO2
Heterojunction Using SCAPS Software for Highly Efficient Solar Cells. Journal of Electronic
Materials. 2017 Aug 1;46(8):5270-7.
[47] Khoshsirat N, Yunus NA, Hamidon MN, Shafie S, Amin N. Analysis of absorber layer
properties effect on CIGS solar cell performance using SCAPS. Optik-International Journal
for Light and Electron Optics. 2015 Apr 1;126(7-8):681-6.
[48] Niemegeers, A., Burgelman, M., Decock, K., Verschraegen, J., & Degrave, S. (2014).
SCAPS manual. University of Gent.
[49] Soucase, B. M., Pradas, I. G., & Adhikari, K. R. (2016). Numerical Simulations on
Perovskite Photovoltaic Devices. In Perovskite Materials-Synthesis, Characterisation,
Properties, and Applications. InTech.
[50] Burgelman, M., Verschraegen, J., Degrave, S., & Nollet, P. (2004). Modeling thin-‐film
PV devices. Progress in Photovoltaics: Research and Applications, 12(2-‐3), 143-153.
[51] Basore, P. A. (1990). Numerical modeling of textured silicon solar cells using PC-
1D. IEEE Transactions on electron devices, 37(2), 337-343.
[52] Jerry G. Fossum. Computer-aided numerical analysis of silicon solar cells. Solid. State.
Electron., 19(4):269–277, Apr 1976.
[53] Lee, C., Efstathiadis, H., Raynolds, J. E., & Haldar, P. (2009, June). Two-dimensional
computer modeling of single junction a-Si: H solar cells. In Photovoltaic Specialists
Conference (PVSC), 2009 34th IEEE (pp. 001118-001122). IEEE.
Queensland University of Technology
69
[54] Kei Ogiya, Chen Lv, Ai Suzuki, Riadh Sahnoun, Michihisa Koyama, Hideyuki Tsuboi,
Nozomu Hatakeyama, Akira Endou, Hiromitsu Takaba, Momoji Kubo, Carlos A Del Carpio,
and Akira Miyamoto. Development of Multiscale Simula- tor for Dye-Sensitized
TiO2Nanoporous Electrode Based on Quantum Chemical Calculation. Jpn. J. Appl. Phys.,
47(4):3010–3014, 2008.
[55] Julio Villanueva, Juan A Anta, Elena Guillen, and Gerko Oskam. Numerical Simulation
of the CurrentVoltage Curve in Dye-Sensitized Solar Cells. J. Phys. Chem. C, 113(45):3–4,
November 2009.
[56] Jinhua Cai and Liyuan Han. Theoretical Investigation on Interfacial-Potential- Limited
Di↵usion and Recombination in Dye-Sensitized Solar Cells. J. Phys. Chem. C,
115(34):17154–17162, September 2011.
[57] Pietro P. Altermatt. Models for numerical device simulations of crystalline silicon solar
cellsa review. J. Comput. Electron., 10(3):314–330, July 2011.
[58] Mari Onodera, Kei Ogiya, Ai Suzuki, Hideyuki Tsuboi, Nozomu Hatakeyama, Akira
Endou, Hiromitsu Takaba, Momoji Kubo, and Akira Miyamoto. Mod- eling of Dye-
Sensitized Solar Cells Based on TiO2Electrode Structure Model. Jpn. J. Appl. Phys.,
49(4):04DP10, 2010.
[59] Zhou, Y., & Gray-Weale, A. (2016). A numerical model for charge transport and energy
conversion of perovskite solar cells. Physical Chemistry Chemical Physics, 18(6), 4476-
4486.
[60] Hoppe, H., Arnold, N., Meissner, D., & Sariciftci, N. S. (2004). Modeling of optical
absorption in conjugated polymer/fullerene bulk-heterojunction plastic solar cells. Thin Solid
Films, 451, 589-592.
[61] Zhou, Y., Huang, F., Cheng, Y. B., & Gray-Weale, A. (2017). Numerical analysis of a
hysteresis model in perovskite solar cells. Computational Materials Science, 126, 22-28.
[62] Fonash, S. (2012). Solar cell device physics. Elsevier.
[63] Tan, K., Lin, P., Wang, G., Liu, Y., Xu, Z., & Lin, Y. (2016). Controllable design of
solid-state perovskite solar cells by SCAPS device simulation. Solid-State Electronics, 126,
75-80.
[64] Lin, L., Jiang, L., Qiu, Y., & Yu, Y. (2017). Modeling and analysis of HTM-free
perovskite solar cells based on ZnO electron transport layer. Superlattices and
Microstructures, 104, 167-177.
Queensland University of Technology
70
[65] Casas, G. A., Cappelletti, M. A., Cédola, A. P., Soucase, B. M., & y Blancá, E. P.
(2017). Analysis of the power conversion efficiency of perovskite solar cells with different
materials as Hole-Transport Layer by numerical simulations. Superlattices and
Microstructures, 107, 136-143.
[66] Karimi, E., & Ghorashi, S. M. B. (2017). Investigation of the influence of different hole-
transporting materials on the performance of perovskite solar cells. Optik-International
Journal for Light and Electron Optics, 130, 650-658.
[67] Qin, P., Tanaka, S., Ito, S., Tetreault, N., Manabe, K., Nishino, H., ... & Grätzel, M.
(2014). Inorganic hole conductor-based lead halide perovskite solar cells with 12.4%
conversion efficiency. Nature communications, 5, 3834.
[68] Leng, C., Qin, H., Si, Y., & Zhao, Y. (2014). Theoretical prediction of the rate constants
for exciton dissociation and charge recombination to a triplet state in pcpdtbt with different
fullerene derivatives. The Journal of Physical Chemistry C, 118(4), 1843-1855.
[69] Hossain, M. F., Faisal, M., & Okada, H. (2016, December). Device modeling and
performance analysis of perovskite solar cells based on similarity with inorganic thin film
solar cells structure. In Electrical, Computer & Telecommunication Engineering (ICECTE),
International Conference on (pp. 1-4). IEEE.
[70] Malinauska, T., Tomkute-Luksiene, D., Sens, R., Daskeviciene, M., Send, R.,
Wonneberger, H., Jankauskas, V., Bruder, I., Getautis, V., 2015. Enhancing thermal stability
and lifetime of solid-state dyesensitized solar cells via molecular engineering of the hole
transporting material spiro-OMeTAD. ACS Appl. Mater. Interfaces 7, 11107–11116.
[71] Nguyen, W.H., Bailie, C.D., Unger, E.L., McGehee, M.D., 2014. Enhancing the hole-
conductivity of spiro-OMeTAD without oxygen. J. Am. Chem. Soc. 136, 10996–11001.
[72] Bailie, C.D., Unger, E.L., Zakeeruddin, S.M., McGehee, M.D., 2014. Melt-infiltration of
spiro-OMeTAD and thermal instability of solid- state dye-sensitized solar cells. Phys. Chem.
Chem. Phys. 16, 4864– 4870.
[73] Li, M. H., Shen, P. S., Wang, K. C., Guo, T. F., & Chen, P. (2015). Inorganic p-type
contact materials for perovskite-based solar cells. Journal of Materials Chemistry A, 3(17),
9011-9019.
[74] Liu, L., Xi, Q., Gao, G., Yang, W., Zhou, H., Zhao, Y., ... & Xu, J. (2016). Cu2O
particles mediated growth of perovskite for high efficient hole-transporting-layer free solar
cells in ambient conditions. Solar Energy Materials and Solar Cells, 157, 937-942.
Queensland University of Technology
71
[75] Niu, G., Li, W., Meng, F., Wang, L., Dong, H., & Qiu, Y. (2014). Study on the stability
of CH 3 NH 3 PbI 3 films and the effect of post-modification by aluminum oxide in all-solid-
state hybrid solar cells. Journal of Materials Chemistry A, 2(3), 705-710.
[76] Du HJ, Wang WC, Zhu JZ. Device simulation of lead-free CH3NH3SnI3 perovskite
solar cells with high efficiency. Chinese Physics B. 2016 Aug 25;25(10):108802.
[77] Minemoto T, Murata M. Impact of work function of back contact of perovskite solar
cells without hole transport material analyzed by device simulation. Current Applied Physics.
2014 Nov 1;14(11):1428-33.
[78] Zhu L, Shao G, Luo JK. Numerical study of metal oxide heterojunction solar cells.
Semiconductor Science and Technology. 2011 Jun 8;26(8):085026.
[79] Goudarzi M, Banihashemi M. Simulation of an inverted perovskite solar cell with
inorganic electron and hole transfer layers. Journal of Photonics for Energy. 2017
Apr;7(2):022001.
[80] Minemoto T, Murata M. Impact of work function of back contact of perovskite solar
cells without hole transport material analyzed by device simulation. Curr Appl Phys
2014;14:1428–33.
[81] Sahdan MZ, Nurfazliana MF, Kamaruddin SA, Embong Z, Ahmad Z, Saim H.
Fabrication and characterization of crystalline cupric oxide (CuO) films by simple immersion
method. Procedia Manufacturing. 2015 Jan 1;2:379-84.
[82] Scanlon, D.O., Morgan, B.J., Watson, G.W., 2009. Modelling the polaronic nature of p-
type defects in Cu2O: the failure of GGA and GGA+U. J. Chem. Phys. 131, 124703.
Queensland University of Technology
72
APPENDIX
Matlab Code (main function) %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%% %%
%% Device Model Solver--Non-coupled algorithm %%
%% %%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
clc;
clear;
format long e
%Fundamental and Material Constants%
E_gap=1.55; % E_gap
mu_n=1; % mu_n
mu_p=2; % mu_p
Nc=2.2e+18; % N_c
q=1.602e-19;
k=1.38e-23;
T=300;
V_t=k*T/q;
V_a=0;
n_int=Nc*exp(-E_gap/(V_t*2));
e0=30*8.85*10^-14;
L=100e-7;
delta=0.0001;
n_max=399;
%Grid Size%
N=401;
delx=L/(N-1);
%Initial Value%
ga=q*(mu_n+mu_p)/e0;
x=0:delx:L;
n0=Nc*exp(-E_gap/(V_t)*x/L);
Queensland University of Technology
73
p0=Nc*exp(-E_gap/(V_t)*(1-x/L));
% fei0=(V_a-E_gap)*(x/L-1/2);
fei0=(V_a-E_gap)*x/L;
G0=(9.71e21)*exp(-5.7e4*x);
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%% %%
%% Calculation Procedure - at Equilibrium %%
%% %%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Solve fi %
for i=1:N-2
a(i)=1/delx/delx;
b(i)=-2/delx/delx-(q/e0)*(n0(i+1)+p0(i+1))/(V_t);
c(i)=1/delx/delx;
f(i)=-(fei0(i)/delx/delx-2*fei0(i+1)/delx/delx+fei0(i+2)/delx/delx+q/e0*(p0(i+1)-n0(i+1)));
end
a(1)=0;
c(N-2)=0;
con_v=0;
while(~con_v)
fi=chase1(a,b,c,f);
fi0=zeros(1,N);
fi0(2:N-1)=fi;
X1=max(abs(fi));
if X1<delta
fei0=fei0+fi0;
con_v=1;
else
fei0=fei0+fi0;
for i=1:N-2
f(i)=-(fei0(i)/delx/delx-2*fei0(i+1)/delx/delx+fei0(i+2)/delx/delx+q/e0*(p0(i+1)-n0(i+1)));
end
end
Queensland University of Technology
74
end
con_n=0;
while(~con_n)
% Solve n %
R=ga*(n0.*p0-n_int^2);
for i=1:N-2
b1(i)=-mu_n/delx/delx*(fei0(i)-fei0(i+1))*exp((fei0(i)-fei0(i+1))/V_t)/(exp((fei0(i)-
fei0(i+1))/V_t)-1)-mu_n/delx/delx*(fei0(i+1)-fei0(i+2))/(exp((fei0(i+1)-fei0(i+2))/V_t)-1)-
ga*p0(i+1);
a1(i)=mu_n/delx/delx*(fei0(i)-fei0(i+1))/(exp((fei0(i)-fei0(i+1))/V_t)-1);
c1(i)=mu_n/delx/delx*(fei0(i+1)-fei0(i+2))*exp((fei0(i+1)-fei0(i+2))/V_t)/(exp((fei0(i+1)-
fei0(i+2))/V_t)-1);
n1(i)=-((mu_n/delx/delx*(fei0(i)-fei0(i+1))/(exp((fei0(i)-fei0(i+1))/V_t)-1))*n0(i)+(-
mu_n/delx/delx*(fei0(i)-fei0(i+1))*exp((fei0(i)-fei0(i+1))/V_t)/(exp((fei0(i)-fei0(i+1))/V_t)-1)-
mu_n/delx/delx*(fei0(i+1)-fei0(i+2))/(exp((fei0(i+1)-fei0(i+2))/V_t)-
1))*n0(i+1)+(mu_n/delx/delx*(fei0(i+1)-fei0(i+2))*exp((fei0(i+1)-fei0(i+2))/V_t)/(exp((fei0(i+1)-
fei0(i+2))/V_t)-1))*n0(i+2)+G0(i+1)-R(i+1));
end
a1(1)=0;
c1(N-2)=0;
n=chase1(a1,b1,c1,n1);
n10=zeros(1,N);
n10(2:N-1)=n;
% Solve p %
for i=1:N-2
b2(i)=mu_p/delx/delx*(fei0(i+1)-fei0(i+2))*exp((fei0(i+1)-fei0(i+2))/V_t)/(exp((fei0(i+1)-
fei0(i+2))/V_t)-1)+mu_p/delx/delx*(fei0(i)-fei0(i+1))/(exp((fei0(i)-fei0(i+1))/V_t)-1)+ga*n0(i+1);
p1(i)=-((-mu_p/delx/delx*(fei0(i)-fei0(i+1))*exp((fei0(i)-fei0(i+1))/V_t)/(exp((fei0(i)-
fei0(i+1))/V_t)-1))*p0(i)+(mu_p/delx/delx*(fei0(i+1)-fei0(i+2))*exp((fei0(i+1)-
fei0(i+2))/V_t)/(exp((fei0(i+1)-fei0(i+2))/V_t)-1)+mu_p/delx/delx*(fei0(i)-fei0(i+1))/(exp((fei0(i)-
fei0(i+1))/V_t)-1))*p0(i+1)+(-mu_p/delx/delx*(fei0(i+1)-fei0(i+2))/(exp((fei0(i+1)-fei0(i+2))/V_t)-
1))*p0(i+2)-(G0(i+1)-R(i+1)));
Queensland University of Technology
75
a2(i)=-mu_p/delx/delx*(fei0(i)-fei0(i+1))*exp((fei0(i)-fei0(i+1))/V_t)/(exp((fei0(i)-
fei0(i+1))/V_t)-1);
c2(i)=-mu_p/delx/delx*(fei0(i+1)-fei0(i+2))/(exp((fei0(i+1)-fei0(i+2))/V_t)-1);
end
a2(1)=0;
c2(N-2)=0;
p=chase1(a2,b2,c2,p1);
p10=zeros(1,N);
p10(2:N-1)=p;
X2=max((abs(n10)+abs(p10))./(n0+p0));
if X2<delta
n0=n0+n10;
p0=p0+p10;
for k=1:N-1
Jn(k)=mu_n*q/delx*(fei0(k)-fei0(k+1))*(n0(k+1)*exp((fei0(k)-fei0(k+1))/V_t)-
n0(k))/(exp((fei0(k)-fei0(k+1))/V_t)-1);
Jp(k)=mu_p*q/delx*(fei0(k)-fei0(k+1))*(p0(k)*exp((fei0(k)-fei0(k+1))/V_t)-
p0(k+1))/(exp((fei0(k)-fei0(k+1))/V_t)-1);
end
AA=G0-R;
J=mean(Jn+Jp);
con_n=1;
else
n0=n0+n10;
p0=p0+p10;
for i=1:N-2
b(i)=-2/delx/delx-(q/e0)*(n0(i+1)+p0(i+1))/(V_t);
f(i)=-(fei0(i)/delx/delx-2*fei0(i+1)/delx/delx+fei0(i+2)/delx/delx+q/e0*(p0(i+1)-n0(i+1)));
end
con_v1=0;
while(~con_v1)
fi=chase1(a,b,c,f);
fi0=zeros(1,N);
Queensland University of Technology
76
fi0(2:N-1)=fi;
X3=max(abs(fi));
if X3<delta
fei0=fei0+fi0;
con_v1=1;
else
fei0=fei0+fi0;
for i=1:N-2
f(i)=-(fei0(i)/delx/delx-2*fei0(i+1)/delx/delx+fei0(i+2)/delx/delx+q/e0*(p0(i+1)-
n0(i+1)));
end
end
end
end
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%% %%
%% Calculation Procedure - at Non-Equilibrium %%
%% %%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Vscan=1.1;
interv=60;
Vinterval=Vscan/interv;
for j=1:interv+1
V_a=Vinterval*(j-1);
%Initial Value%
x=0:delx:L;
n0=Nc*exp(-E_gap/(V_t)*x/L);
p0=Nc*exp(-E_gap/(V_t)*(1-x/L));
fei0=(V_a-E_gap)*(x/L-1/2);
%Coefficient Matrix%
for i=1:N-2
Queensland University of Technology
77
a(i)=1/delx/delx;
b(i)=-2/delx/delx-(q/e0)*(n0(i+1)+p0(i+1))/(V_t);
c(i)=1/delx/delx;
f(i)=-(fei0(i)/delx/delx-2*fei0(i+1)/delx/delx+fei0(i+2)/delx/delx+q/e0*(p0(i+1)-n0(i+1)));
end
a(1)=0;
c(N-2)=0;
con_v=0;
while(~con_v)
% Solve fi %
fi=chase1(a,b,c,f);
fi0=zeros(1,N);
fi0(2:N-1)=fi;
X1=max(abs(fi));
if X1<delta
fei0=fei0+fi0;
con_v=1;
else
fei0=fei0+fi0;
for i=1:N-2
f(i)=-(fei0(i)/delx/delx-2*fei0(i+1)/delx/delx+fei0(i+2)/delx/delx+q/e0*(p0(i+1)-n0(i+1)));
end
end
end
con_n=0;
while(~con_n)
% Solve n %
R=ga*(n0.*p0-n_int^2);
for i=1:N-2
b1(i)=-mu_n/delx/delx*(fei0(i)-fei0(i+1))*exp((fei0(i)-fei0(i+1))/V_t)/(exp((fei0(i)-
fei0(i+1))/V_t)-1)-mu_n/delx/delx*(fei0(i+1)-fei0(i+2))/(exp((fei0(i+1)-fei0(i+2))/V_t)-1)-
ga*p0(i+1);
a1(i)=mu_n/delx/delx*(fei0(i)-fei0(i+1))/(exp((fei0(i)-fei0(i+1))/V_t)-1);
Queensland University of Technology
78
c1(i)=mu_n/delx/delx*(fei0(i+1)-fei0(i+2))*exp((fei0(i+1)-fei0(i+2))/V_t)/(exp((fei0(i+1)-
fei0(i+2))/V_t)-1);
n1(i)=-((mu_n/delx/delx*(fei0(i)-fei0(i+1))/(exp((fei0(i)-fei0(i+1))/V_t)-1))*n0(i)+(-
mu_n/delx/delx*(fei0(i)-fei0(i+1))*exp((fei0(i)-fei0(i+1))/V_t)/(exp((fei0(i)-fei0(i+1))/V_t)-1)-
mu_n/delx/delx*(fei0(i+1)-fei0(i+2))/(exp((fei0(i+1)-fei0(i+2))/V_t)-
1))*n0(i+1)+(mu_n/delx/delx*(fei0(i+1)-fei0(i+2))*exp((fei0(i+1)-fei0(i+2))/V_t)/(exp((fei0(i+1)-
fei0(i+2))/V_t)-1))*n0(i+2)+G0(i+1)-R(i+1));
end
a1(1)=0;
c1(N-2)=0;
n=chase1(a1,b1,c1,n1);
n10=zeros(1,N);
n10(2:N-1)=n;
% Solve p %
for i=1:N-2
b2(i)=mu_p/delx/delx*(fei0(i+1)-fei0(i+2))*exp((fei0(i+1)-fei0(i+2))/V_t)/(exp((fei0(i+1)-
fei0(i+2))/V_t)-1)+mu_p/delx/delx*(fei0(i)-fei0(i+1))/(exp((fei0(i)-fei0(i+1))/V_t)-1)+ga*n0(i+1);
p1(i)=-((-mu_p/delx/delx*(fei0(i)-fei0(i+1))*exp((fei0(i)-fei0(i+1))/V_t)/(exp((fei0(i)-
fei0(i+1))/V_t)-1))*p0(i)+(mu_p/delx/delx*(fei0(i+1)-fei0(i+2))*exp((fei0(i+1)-
fei0(i+2))/V_t)/(exp((fei0(i+1)-fei0(i+2))/V_t)-1)+mu_p/delx/delx*(fei0(i)-fei0(i+1))/(exp((fei0(i)-
fei0(i+1))/V_t)-1))*p0(i+1)+(-mu_p/delx/delx*(fei0(i+1)-fei0(i+2))/(exp((fei0(i+1)-fei0(i+2))/V_t)-
1))*p0(i+2)-(G0(i+1)-R(i+1)));
a2(i)=-mu_p/delx/delx*(fei0(i)-fei0(i+1))*exp((fei0(i)-fei0(i+1))/V_t)/(exp((fei0(i)-
fei0(i+1))/V_t)-1);
c2(i)=-mu_p/delx/delx*(fei0(i+1)-fei0(i+2))/(exp((fei0(i+1)-fei0(i+2))/V_t)-1);
end
a2(1)=0;
c2(N-2)=0;
p=chase1(a2,b2,c2,p1);
p10=zeros(1,N);
p10(2:N-1)=p;
X2=max((abs(n10)+abs(p10))./(n0+p0));
if X2<delta
Queensland University of Technology
79
n0=n0+n10;
p0=p0+p10;
for k=1:N-1
Jn(k)=mu_n*q/delx*(fei0(k)-fei0(k+1))*(n0(k+1)*exp((fei0(k)-fei0(k+1))/V_t)-
n0(k))/(exp((fei0(k)-fei0(k+1))/V_t)-1);
Jp(k)=mu_p*q/delx*(fei0(k)-fei0(k+1))*(p0(k)*exp((fei0(k)-fei0(k+1))/V_t)-
p0(k+1))/(exp((fei0(k)-fei0(k+1))/V_t)-1);
end
J(j)=mean(Jn(j)+Jp(j));
con_n=1;
else
n0=n0+n10;
p0=p0+p10;
for i=1:N-2
b(i)=-2/delx/delx-(q/e0)*(n0(i+1)+p0(i+1))/(V_t);
f(i)=-(fei0(i)/delx/delx-2*fei0(i+1)/delx/delx+fei0(i+2)/delx/delx+q/e0*(p0(i+1)-
n0(i+1)));
end
con_v1=0;
while(~con_v1)
fi=chase1(a,b,c,f);
fi0=zeros(1,N);
fi0(2:N-1)=fi;
X3=max(abs(fi));
if X3<delta
fei0=fei0+fi0;
con_v1=1;
else
fei0=fei0+fi0;
for i=1:N-2
f(i)=-(fei0(i)/delx/delx-2*fei0(i+1)/delx/delx+fei0(i+2)/delx/delx+q/e0*(p0(i+1)-
n0(i+1)));
end
Queensland University of Technology
80
end
end
end
end
end
V_a=0:1.1/60:1.1;
figure(4)
plot(V_a,J*1000,'LineWidth',2);
axis([0 1.1 0 15]);
xlabel('Voltage(V)');
ylabel('Current(mA/cm2)');
title('IV Characteristics');