Formatted: Font: 12 pt

A numerical study of charge transport properties and an influence of buffer nano-layers on J-V

characteristics of organic solar cells

(Thesis abstract)

Candidate: Tran Van Hung (email: hungtran.vnu@gmail.com)

Supervisor: Dr. Dang Dinh Long (email: longdd@vnu.edu.vn)

I Introduction
Over the past few decades, organic solar cells (OSCs) based on donor:acceptor blends have been attracted
tremendous interest attention because of their potentials such as low-cost, flexible, lightweight, portable and large-area
energy-harvesting devices. Two main approaches are appliedcan be applied in order to improve the overall power
conversion efficiency (PCE) includign are either modification of photoactive layer itself or introduction of
transport/blocking (buffer) layers. At the present, the factIt is empirically proved that an the considerable riseenhancement
of PCE is observed by insertingthanks to the insertion of an interfacial layer between active layer and electrodes [1].
Whereas, on the other sideBeside of that, works have also been done with modifying the active layer and gave the
risetrivial increase of PCE is still limited in the case of modification of active layer. In this work, we performed studying
inuse theoritical simulation aiming at proving the effectiveness of both cases by simulation methods. Regarding buffer
layer, we applied single diode model for simulating working principle of in multilayer organic solar cells and simulate its
working principle as well as provided proposed a method to extract parameters of interest of organic solar cells from
experimental J-V curves with high accuracy. We also investigated the conversion from the power factor to efficiency of
organic solar cells under variation ofeffected by buffer layer thickness. The influence of thickness on device properties is
figured out to be very clear and in good accordance with experimental data. From our calculations, an optimal thickness
is found in each type of organic solar cells. Besides, the work also focus typically on analyzing charge transport of organic
materials by Monte Carlo method [2]. By this way, we give influence of parameters such as temperature, electric field
and lattice size on electric properties of materials. Commented [DT1]: Viet lai doan nay cho ro rang hon di
I.1II Organic solar cells Hung, viet cho logic vao nhe, tranh trung lap voi doan tren.
Traditionally, organic solar cells have a multilayer structure in which each layer could be fabricated by an
individual method. The photoactive layer consists of two constituents including a donor material which is usually a Commented [DT2]: Cau nay ko ro nghia!
conjugated polymer , conjugated pigments or oligomers and in conjunction with an acceptor material which is often
fullerene derivative. The photoactive layer is inserted between anode and cathode electrodes in which at least one
electrode is transparent. In addition, in order to enhance the performance of solar cells, the interfacial layers called hole
conducting layer and electron conducting layer are located between anode-photoactive and cathode-photoactive interfaces,
respectively. Recently inverted architecture has been applied for organic solar cells where . In this device, the cathode is
considered as transparent electrode while the other is anode. The fact that efficiency of inverted devices is comparable
with the conventional devices but the stability is higher much better in environment. Both conventional and inverted solar
cells are illustrated in Fig. 1.

Figure 1: Two fundamental organic solar cell structures

1
 Figure 2: Working mechanism of organic solar cells [3]
The working mechanism can divide intobe seen in four fundamental steps as shown in Fig. 2;. (1)Firstly, when
light is absorbed by donor material, electrons is excited from highest occupied molecular orbital (HOMO) to lowest
occupied molecular orbital (LUMO). As a result, the electron-hole pairs with biding energies in range of 0.1 𝑒𝑉 and
1.4 𝑒𝑉 are generated because of low mobility of organic materials. (2) Tthe excitons diffuse to the donor-acceptor
interface, (3) where these excitons they are separated into free carriers (electrons and holes) due to offset potential energy
at donor-acceptor interface. (4) the After that free carriers diffuse to respective electrodes through continuous pathway
and avoid recombination or traps.
I.2 III The role of buffer layers
Conventionally, an organic solar cell has three main el[4]ements including photoactive layer and two electrodes. Commented [DT3]: This sentence is repeated again from
However, dDue to the poor photoactive-electrode interfaces, interfacial layers must be have been established to adjust the the in organic solar cell introduction. Avoid repetition
band alignment at between electrodes and active layerthe interfaces, select one type of carriers, prevent chemical reaction
between electrodes and photoactive layer, improve stability in environment, and act as optical spacer. There are exist two
types of buffer layers commonly used inin organic solar cells including anode and cathode buffer layers. They correspond
to the layers inserted at interfacial layer between anode/ and photoactive layer interface is called anode buffer layer (hole
blocking layer) while the interfacial layer between and cathode/ and photoactive layer interface is call cathode buffer
layer (electron blocking layer). More importantly, buffer layers can play an important role in the improvement of photo
current. In particular, because of the low mobility of carriers in organic materials, careers are limited in diffusion length
and lifetime. Commented [DT4]: This sentence does not make sense in
IV II. Simulation methods this part
Numerical simulations are imapplied to analyze problems from theoretical mechanisms to experimental
phenomena. Regarding to semiconductors and solar cells simulation methods can be appliedy with different difficulty Commented [DT5]: This sentence is not clear, therefore
levels such as quantum mechanism, Monte Carlo technique, classical analysisze, and equivalent circuit. In order to cannot make sense
determine which level is implied for examining problem, it is dependent on type of problem and knowledge of technician.
In this work, I would like to apply both two different levels. In the first case, We first start with single diode model the Commented [DT6]: Which two levels?
to interpret the working principle of organic solar cell is interpreted by single diode model which is widely used. This
model reflects the non-liner relationship between 𝐽 − 𝑉 characteristic and five electrical parameters series resistance,
shunt resistance, photocurrent, saturation current and diode ideality factor. In the other caseFinally, the charge transport
mechanism in organic semiconductor was studied by Monte Carlo technique. Commented [DT7]: Hung viet them vai cau nua
II.1 Single diode model
In general, the simple approach to evaluate performance of an organic solar cell based on analyzing 𝐽 − 𝑉 under
given temperature and irradiation condition. Analytical model as diode model is used to study and evaluate behavior of
organic solar cell including non-linear electrical elements. To understand the origin of single diode model we should
notice that a solar cell consists of two different layers of semiconductor materials. Commented [DT8]: This graph is not clear, rewrite and
In the dark region, a single-junction solar cell behaves similarly to a semiconductor diode and 𝐽 − 𝑉 characteristic make it easy to understand!
is described by following Shockley equation.
𝑞𝑉
𝐽𝑠 = 𝐽0 (𝑒 𝑛𝑘𝑇 − 1)
Where: 𝐽𝑠 is dark current, 𝐽0 is the reverse saturation current density, 𝑉 is applied voltage, 𝑞 is the electron
charge (1.602 × 10−19 C), 𝑘 is the Boltzmann constant, 𝑇 is the absolute temperature and 𝑛 is diode factor depend
on the kind of semiconductor.

2
 Figure 3: Ideal diode model
Under the illumination, the cell generates electron-hole pairs that yield a current density 𝐽𝑝ℎ - photocurrent
density. The solar cell can be represented by a basic electric circuit, which consists of a current source with intensity 𝐽𝑝ℎ ,
and a diode as in figure1. The solar cell has the same behavior as a diode. The J-V characteristic is described by:
𝐽 = 𝐽𝑝ℎ − 𝐽𝑠
𝑞𝑉
𝐽 = 𝐽𝑝ℎ − 𝐽0 (𝑒 𝑛𝑘𝑇 − 1)

In particular, solar cells behave actually in a complex way, especially organic solar cell, therefore, in which leak
current cannot be overlooked. The most efficient model that is applied to describe the working principle of solar cell is
the single diode model as shown in Fig. 4, and the relation between current and voltage is mathematically defined by the
flowing equation:
𝑞𝑉 𝑉 + 𝐽𝑅𝑠
𝐽 = 𝐽𝑝ℎ − 𝐽0 (𝑒 𝑛𝑘𝑇 − 1) −
𝑅𝑠ℎ
Where 𝐽 and 𝑉 represent the current density and voltage of solar cell, respectively. 𝐽𝑝ℎ is the photo-generated
current, 𝐽0 is the reverse bias saturation current density, 𝑅𝑠 and 𝑅𝑠ℎ represent the series and shunt resistances
respectively, n is the diode ideality factor of cell, and Vth is thermal voltage of the cell is expressed by 𝑉𝑡ℎ = 𝑘𝐵 𝑇/𝑞. Here,
𝑘𝐵 is the Boltzmann’s constant (1.3806 × 10−23 𝐽/𝐾), 𝑇 is the absolute temperature (𝐾), and 𝑞 is the electron charge
(1.6022 × 10−19 𝐶).

Figure 4: The single diode for solar cell

By using the single diode model, this work successfully described again the working principle of the multilayers
organic solar cell, and examined the relation between buffer layer thickness and power conversion efficiency through
analyzingses series and shunt resistances from 𝐽 − 𝑉 characteristics under given irradiation condition.
II.2 The hopping transport model
It is clear that the single diode model cannot reflect intensively electrical influence on the efficiency of the solar
cell since the losing energy and carrier transport take place in nanoscale with many different ways. Herein the Monte Commented [DT9]: Hung xem lai doan nay nhe
Carlo technique-based model is given to study charge transport through organic materials as well as organic
semiconductors.
The Monte Carlo is a technique for analyzing phenomena by means of computer algorithms based on the
generation of random numbers. Monte Carlo method is maned from the gambling city in Monaco because of using
stochastic process to accept or reject events during solving. Monte Carlo simulation is implied applied for a wide range
of applications ofwith different fields such as science, commerce, and finance. In physics, Monte Carlo technique is
suitable to examine systems which have large degree of freedoms including gases, fluids, disordered materials, optical
systems, many body particles, quantum physics. Remarkably, the accuracy of the model is higher when system is larger
and more complex.
a) Modeling organic molecule and polymer layers

3
 Figure 5: a) Polymer chains. b) Long chains would break into small conjugation units (sites-yellow circles)
To examine charge transport in organic materials the readers should identify how organic layer should beis
modeledtreated? It is well known that a solid polymer film contains a series of polymer chains which distribute over space.
It is assumed that the polymer chains are perfectly continuous without kinks and twists, charge carriers are not localized
over the whole chains. In practical, the polymer chains in electric device are broken into subconjugation units by chemical
or physical defects. As a result of this, there are many small electronic units that can consider as small molecules forming
the film. In this case, the charge carriers are localized at small units which are called sites. Fig 5b shows illustration of
how polymer chains would be broken into subunits. Therefore, to simplify the model we only consider the appearance of
distribution of electrical sites that can host charge and enable them to move through the film. In summary, what we need
now is to examine charge transport across the site matrix in real space as well as in energy space shown in Fig.6.

Figure 6: Distribution of sites in real space (a), energy space (b), energy landscape in space (c). Red line is the
path electron must take while hopping from one side to the other (under applied electric field)
b) The hopping rate
As we discussed above, the property of charge transport is studied through a distribution of sites that haveof slightly
different energies with each other owing to differences in their environments as well as the lengths of segments. In the
present model, a carrier is able to tunnel to another that is its vicinity at the same time asbetween sites as it emits or
absorbs a phonon. The tunneling process are named as the “hops” in hopping transport across localized states. The hopping
events from site 𝑖 to site 𝑗 are controlled by Miller-Abrahams hopping rate that gives ability to tunneling from site 𝑖
to site 𝑗 with a spatial separation 𝑟𝑖𝑗 .

4
 Figure 7: The hop from site to the other site. Where 𝑎 is cut-off radius and 𝛤𝑖𝑗 is hopping rate
(∆𝐸𝑖𝑗 )
2𝑟𝑖𝑗 exp (− ), 𝐸𝑗 −𝐸𝑖 > 0
Γ𝑖𝑗 = 𝑣0 exp(− ){ 𝑘𝐵 𝑇
𝑎
1, 𝐸𝑗 − 𝐸𝑖 ≤ 0
Where 𝑣0 is the attempt-to-escape frequency which depends on the phonon frequencies, 𝑎 is the localization
length which describes spatial size of electron wave functions of the localized states, and ∆𝐸𝑖𝑗 = 𝐸𝑗 − 𝐸𝑖 is the energy
difference of two sites. The energy difference would be ∆𝐸𝑖𝑗 = 𝐸𝑗 − 𝐸𝑖 − 𝐹𝑒(𝑥𝑗 − 𝑥𝑖 ) under applied electric field F
parallel to x direction. It is clear obvious that the hopping rate between site 𝑖 and site 𝑗 depends exponentially on both
(∆𝐸𝑖𝑗 )
the distance and the energy difference as well. The exponential function exp (− ) for a transition upward in energy
𝑘𝐵 𝑇

from a localized state 𝑖 to a higher energy state 𝑗 describes the probability of absorbing a phonon with the required
energy. Whereas, for the reversea transition downward to a lower energy state, a process emits excess energy hold would
be taken place with 1 value of a probability of 1.
By using the Miller-Abraham rate, it is assumed that the sites have no structure, so characterized only by their
positions and energies. The hopping rate therefore depends on the distance between sites and their energy differenceies.
In Tthe case of polymer as showncalculated in Fig. 5. seems to be a very rough approximation, and it is known that
because the hopping rate between two units sites also depend on their orientations. In order to simplify, we omitted the
orientation factor However, it is necessary to avoid complicated tasks, therefore, the relative orientation of the molecules
and site structure are overlooked. MoreoverEven omitting that feature, there are many of works were successful to that
acquired the conducting features of polymers with using simple model of Miller-Abrahams hopping rate. The last but not
least, we also assumed that the charge carriers are only electrons in the model since holes behave in the same way with
electron except for the sign of charge.
c) Distribution of sites in space and energy
In order to calculate the mobility, the positions and energies of all sites are principle inputs that have tomust be
known in the model. The energy distribution of the sites is described by the density of state 𝑔(𝜀). This distribution is and
chosen according to the material which being used in the model. The obtained final results are influenced by how we
determineing that distribution because the hopping rate depends strongly not only on the distance between the sites but
also on the energy difference between them. There are two common energy distributions including exponential
distribution for inorganic materials and Gauss distribution for organic materials. These distributions were established by
measuring the optical absorption spectrum and the absorption bands of materials. Hence, for organic materials, the Gauss
distribution is selected.
𝑁 (𝜀)
𝑔(𝜀) = 𝑒 𝜀0
𝜀0
Here 𝑁 is the spatial density of sites, 𝜀0 is the energy scale of the distribution. Formatted: Indent: First line: 0"

𝑁 𝜀2
(− )
𝑔(𝜀) = 𝑒 2𝜎2
𝜎√2𝜋
Where, 𝜎 is the standard deviation of the distribution. Formatted: Indent: First line: 0"
The other one is the site distribution which are randomly placed in space with a uniform distribution. Nevertheless, another

5
more effective approach more effectively is that the sites are placed on a square lattice in two dimensions or a cubic lattice
in three dimensions for computation.
d) Parameters

e) Monte Carlo simulation

The simplest approach is to simulate one single electron at a time. In this case the electron density is low and the
interaction between electrons can be ignored. The electron performs a random walk in the system of sites, and then we
collect that stochastic walks. As we mentioned above, the destination site is chosen randomly, but it depends on the
hopping rates to each other possible destination site. The time that charge carriers spendsabide on each site is also Commented [DT10]: Hung viet lai cau nay nhe
generated randomly and depends on the total hopping rate. After setting thea set of jumps, a statistic of electron motion
is collected, and the transport properties of the material can be determined.
More specific, how a carrier performs a hopping process to a destination site destination in a group of sites will be
described below:

Figure 8:
It is known thatWhere, the pink circles are described asdenotes the possible destination sites of destination and the
electron are on grey circular shade. In the very short period of time, the electron will hop to one of the possible destination
sites, so which one will be chosen. When the charge carrier is located on site 𝑖, and the value of the total hopping rate
(Γ𝑖 ) is given by:

Γ𝑖 = ∑ Γ𝑖𝑗
𝑗

In the case of Fig. 8. Γ𝑖 = Γ𝑖1 + Γ𝑖2 + Γ𝑖3 + Γ𝑖4 + Γ𝑖5 + Γ𝑖6 with Γ𝑖𝑗 is the hopping rate is calculated by Miller-
Abraham formula. The probability that a carrier jumps from a site 𝑖 to any site 𝑗 is given by:
Γ𝑖𝑗
𝑝𝑖𝑗 =
Γ𝑖
And ∑𝑗 𝑝𝑖𝑗 = 1
The time 𝜏 for a carrier spending on the site 𝑖 before hopping is calculated as
𝜏 = 𝑇⁄Γ
𝑖

Where 𝑇 is taken from another exponential distribution of random numbers with unit expectation value and unit
variance.
The destination site that the carrier jumps to is determined by picking a random number 𝑥 between 0 and 1
from a uniform distribution and satisfy the following condition:
𝑗−1 𝑗

∑ 𝑝𝑖𝑘 ≤ 𝑥 < ∑ 𝑝𝑖𝑘

𝑘=1 𝑘=1
The program places an electron at an arbitrary starting site and simulate its trajectory during a fixed simulation
time.
III.V Result and discussion
III.1 The influence of buffer layer on performance of organic solar cells
In this section the results of influence of buffer layer’s thickness on 𝐽 − 𝑉 characteristics as well as performance
of organic solar cells such as efficiency, fill factor, series and shunt resistances. In order to examine the influence from
experiment curves, the simulation technique and the numerical method are implied. Here the Ortiz-Conde method based

6
on Lambert function is used to compute parameters of organic solar cells.
For the first, the accuracy of method will be verified and compared with other methods in Fig. 9. It is clear that
Table 1: Rs and Rsh are calculated from three different methods
Resistances Present method Optimization method Five-point method
Rs (Ω) 5.56 6.21 0.804
Rsh (Ω) 206.5 203.25 221.73

Figure 9:
In order to examine the influence of buffer layer on parameters, we use Irwin’s data about organic solar cell with
structure of glass/ITO/NiO/PEDOT:PSS/-P3HT:PCBM/LiF/Al in which NiO plays anode buffer layer.

Figure 10: J-V characteristics vary in thickness of NiO thực nghiệm (a) và dữ liệu mô phỏng (b)

7
 Commented [DT11]: This figure gives the same meaning
as Fig. 10 (Delete Fig. 11)
The captions, translate them into English.

Figure 11: Đường mô phỏng và dữ liệu thực nghiệm với độ dày lớp đệm (a) 5 nm; (b) 20 nm; (c) 43 nm; (d) 77
nm
From experimental curves we calculated the parameters of solar cell by using numerical method as shown table 2.
Fig. 10b. shown that simulated results behave similarly with experimental curves. More specific, Fig. 11. shows the
comparison relatively between simulated curve and experimental curve vary in thickness of NiO.
The calculated values from experimental data are quite accurate with measurable values, especially fill factor (FF)
and power conversion efficiencies (PCE) (table 2 and table 3). Clearly, when comparing extracted values and experimental
values, the maximum of FF and PCE reach at 10 nm thickness of NiO. For the fill factor, the extracted value and
experimental value are 0.678 and 0.693 respectively. For the power conversion efficiency, extracted value and
experimental value are 4.9 and 5.16 respectively. Moreover, this extracted method also carries out electrical parameters
including series resistance (Rs) and shunt resistance (Rsh). At the thickness of maximum power conversion efficiency,
Table 2: Giá trị tham số sau khi tính toán với PMT có lớp đệm NiO
Thickness of Rs Rsh FF PCE Sai số thực nghiệm
NiO (%)
(Ω) (Ω) (%)
FF PCE

05 nm 1.100 584.948 0.636 4.5 0.47 5.3

10 nm 1.400 3960.108 0.678 4.9 2.1 5

20 nm 5.771 588.501 0.588 2.8 6.5 5.4

43 nm 8.585 486.750 0.544 2.4 3.8 5.9

77 nm 9.341 554.866 0.542 2.2 8.8 3

Table 3: Giá trị thông số của bài báo với PMT có lớp đệm NiO Commented [DT12]: Hung, plot a figure to compare PCE
Thickness of NiO VOC JSC and FF between calculation and experimental data if
FF PCE
possible. Because this is the most important factor for the
(V) (mA/cm2) (%) solar cell. Now one wants to see raw data putted in table, but
05 nm 0.634 11.5 0.633 4.75 colorful figures.

10 nm 0.638 11.3 0.693 5.16

8
 20 nm 0.591 8.83 0.552 2.96
43 nm 0.586 8.09 0.524 2.55
77 nm 0.581 7.49 0.498 2.23

Figure 12: Đồ thị Rs (a) và Rsh (b) thay đổi theo độ dày lớp đệm
III.2 charge transport in organic materials

VIIV. Conclusion

[1] D. Chi, S. Huang, S. Yue, K. Liu, S. Lu, Z. Wang, et al., "Ultra-thin ZnO film as an electron transport layer
for realizing the high efficiency of organic solar cells," RSC Advances, vol. 7, pp. 14694-14700, 2017.
[2] S. D. Baranovskii, "Theoretical description of charge transport in disordered organic semiconductors,"
physica status solidi (b), vol. 251, pp. 487-525, 2014.
[3] C. Yan, S. Barlow, Z. Wang, H. Yan, A. K. Y. Jen, S. R. Marder, et al., "Non-fullerene acceptors for organic
solar cells," Nature Reviews Materials, vol. 3, p. 18003, 02/13/online 2018.
[4] K. A. Mazzio and C. K. Luscombe, "The future of organic photovoltaics," Chemical Society Reviews, vol. 44,
pp. 78-90, 2015.