Explorar E-books

Categorias

Explorar Audiolivros

Categorias

Explorar Revistas

Categorias

Explorar Documentos

Categorias

0 notas0% acharam este documento útil (0 voto)

16 visualizações113 páginasDissertation - Nícolas Matheus da Fonseca Tinoco de Souza Araújo

© © All Rights Reserved

0 notas0% acharam este documento útil (0 voto)

16 visualizações113 páginasDissertation - Nícolas Matheus da Fonseca Tinoco de Souza Araújo

Você está na página 1de 113

TECHNOLOGY CENTER

POST-GRADUATION PROGRAM IN MECHANICAL ENGINEERING

CELL TECHNOLOGIES

NATAL- RN, 2019

FEDERAL UNIVERSITY OF RIO GRANDE DO NORTE

TECHNOLOGY CENTER

POST-GRADUATION PROGRAM IN MECHANICAL ENGINEERING

CELL TECHNOLOGIES

Program in Mechanical Engineering (PPGEM) of

the Federal University of Rio Grande do Norte as

part of the requirements for obtaining the title of

MASTER IN MECHANICAL

ENGINEERING, supervised by the Prof. Dr.

Fábio José Pinheiro Sousa and co- supervised by

the Prof. Dr. Flavio Bezerra Costa.

NATAL - RN

2019

INFLUENCE OF TEMPERATURE ON POLYCRYSTALLINE SILICON SOLAR

CELL TECHNOLOGIES

(PPGEM) of the Federal University of Rio Grande do Norte

Federal University of Rio Grande do Norte – Supervisor

Prof. Dr. Flavio Bezerra Costa ___________________________

Federal University of Rio Grande do Norte – Co-Supervisor

Prof. Dr. Gabriel Ivan Medina Tapia ___________________________

Federal University of Rio Grande do Norte – External Evaluator

Profa. Dra. Juliana Ricardo de Souza ___________________________

Federal Institue of Rio Grande do Norte – External Evaluator

Dedication

who help make the world a better place for

unknowns, but have never had a worthy

recognition.

Acknowledgments

A Deus.

As duas pessoas mais importantes da minha vida: meus pais, Sérgio Roberto de

Araújo e Juraneide da Fonseca Tinoco de Souza Araújo, os quais, juntos, mostraram-me a

importância dos estudos e da perseverança na vida, além de propiciar todos os possíveis e

impossíveis meios para que eu pudesse estudar da melhor forma.

Aos meus irmãos, Nickson Sérgio da Fonseca Tinoco de Souza Araújo e Nilberth

Roberto da Fonseca Tinoco de Souza Araújo, os quais me deram todo tipo de apoio sempre

que necessitei.

Aos meus avôs (Diniz Araújo e Jurandir Fonseca) e às minhas avós (Mirtô Tinoco

e Maria das Neves - Sassá) por terem propiciado um ambiente ímpar aos meus pais e, portanto,

a mim. Minha eterna gratidão a vocês.

A minha companheira, Lílyan Manásseias Romeiro Dantas, com quem tenho a

oportunidade de crescer cada dia mais. Uma mulher que admiro pela inteligência, simplicidade

e garra, a qual não permitiu que obstáculos a impedissem de buscar sempre o melhor.

Aos professores Dr. Fábio José Pinheiro Sousa e Dr. Flavio Bezerra Costa por me

aceitarem como seu aluno orientado e co-orientado, respectivamente, e por incentivarem a

pesquisa científica sempre nivelada por cima. Além de, juntos comigo, terem aceitado a dar

início a uma nova empreitada acadêmica, uma nova ramificação na árvore do conhecimento.

Ao Professor Dr. Gabriel Ivan Medina Tapia por ceder espaço físico no Laboratório

de Sistemas Térmicos e Energias Alternativas da UFRN para a realização dos ensaios

experimentais importantíssimos para o trabalho. Além das incontáveis discussões agregadoras

de conhecimento acadêmico e pessoal.

A todos os integrantes do Grupo de Estudos de Tribologia e Integridade Estrutural da

UFRN e do Laboratório de Sistemas Térmicos e Energias Alternativas com quem tive a

oportunidade de conviver e aprender. Grato em especial aos meus orientadores diretos de

Iniciação Científica João Andrade Lopes de Sousa, Gracilene dos Santos Aquino Pontes e

Fernando Nunes da Silva, e ao amigo produtivo Felipe Pinheiro Maia. Grato também ao meu

orientador indireto de Iniciação Científica o professor Dr. João Telésforo Nóbrega de

Medeiros o qual admiro enormemente por seu incentivo às pesquisas científicas, assim como

os debates de cunho sociais e pessoais ‘haut niveau’.

Ao Programa de PPGEM-UFRN. E, por fim, à CAPES, pelo apoio financeiro.

Thank you so much!

“Everything should become as simple as possible, but not simplified.”

Albert Einstein.

ix

policristalina de silício. 2019. 113 p. Dissertação de Mestrado em Engenharia Mecânica

(Programa de Pós-Graduação em Engenharia Mecânica) - Universidade Federal do Rio Grande

do Norte, Natal-RN, 2019.

Resumo

consciente está continuamente crescendo, enquanto seus custos, reduzindo. Nas tecnologias

fotovoltaicas (PV), a temperatura da célula solar e a irradiação solar incidente são os principais

fatores os quais afetam seu desempenho. O foco desse trabalho é a análise da influência da

temperatura sobre as células fotovoltaicas, tendo em vista seu impacto nos parâmetros físicos

desta tecnologia, na capacidade de simulação dos modelos equivalentes existentes e na

efetividade dos simuladores solares. Por conseguinte, a fim de compreender esses efeitos, fez-

se um estudo acerca do impacto da temperatura nos parâmetros de desempenho de uma célula

solar de silício, uma busca por qual o modelo mais adequado para simular células fotovoltaicas

com a temperatura de operação variando e uma análise do efeito da temperatura nas condições

de simulação ao utilizar um simulador de baixo custo, o SOLSIM. O experimento foi realizado

utilizando células fotovoltaicas policristalinas de silício com um intervalo de temperatura de 20

a 60 °C sobre uma intensidade luminosa média de 1.000 W/m², empregando um simulador solar

(SOLSIM) desenvolvido pelo INPE e adaptado pelo autor. Neste trabalho, uma breve revisão

das diversas relações matemáticas envolvendo a influência da temperatura sobre as células

solares foi desenvolvida. Mostrou-se que no intervalo entre 20 e 60 °C os parâmetros da célula

solar de silício policristalino variam linearmente com a temperatura, e a potência é reduzida

com o aumento da temperatura. Na literatura, identificaram-se sete equações as quais

determinam a dependência entre os parâmetros da célula solar e a temperatura. As principais

pesquisas acerca do problema de estimação dos parâmetros dos modelos equivalentes de células

solares são classificadas de acordo com o número de parâmetros, a forma de extração dos

parâmetros, as equações de translação e o material da célula. Uma classificação qualitativa foi

realizada e, assim, encontraram-se quatro modelos como sendo os melhores para simular o

comportamento de uma célula PV: Boutana et al. (2017), Mahmoud and Xiao (2013), Orioli

and Di Gangi (2013) and Ishaque (2011). A temperatura é determinante do valor teórico

máximo para a corrente gerada em função do simulador solar utilizado.

x

technologies. 2019. 113 p. Master’s Dissertation in Mechanical Engineering (Post-Graduation

Program in Mechanical Engineering) - Federal University of Rio Grande do Norte, Natal-RN,

2019.

Abstract

Over the years, the contribution of photovoltaics energy to an eco-friendly world is continually

increasing and its costs reducing. In photovoltaic (PV) technologies, the solar cell temperature

and incident solar radiation are the main factors that affect their behaviors. The focus of this

work is the analysis of temperature influence over the solar cells, in view of its impact on

physical parameters of this technology, on simulation capacity of the existent equivalent models

an on effectiveness of solar simulators. Therefore, to comprehend these effects, it was

performed a study on the impact of cell temperature on the performance parameters of silicon

solar cell, a search for the most suitable model to emulate a solar cell when its operation

temperature varies and a analysis of the temperature effect on conditions of simulation while

using a low cost solar simulator, the SOLSIM. The experiment was carried out using a

polycrystalline silicon solar cell with the temperature in the range 20–60 °C at constant light

intensity 1,000 W/m2 employing a solar cell simulator (SOLSIM) developed by the INPE and

adapted by the author. In this work, it was made a brief review of the various existing

mathematical relationships involving the influence of temperature over solar cells. In the range

20-60 °C, the polycrystalline solar cell parameters varied linearly with temperature, and the

power is reduced with temperature increasing. In literature, it was identified seven equations

which determine the dependence between the solar cell parameters and the temperature. The

main existing research works on PV cell model parameter estimation problem are classified

according to number of parameters, parameters’ extraction, translation equations and PV

material technology. A qualitative comparative ranking was made and four models was found

to be the best ones to emulating PV cell: Boutana et al. (2017), Mahmoud and Xiao (2013),

Orioli and Di Gangi (2013) and Ishaque (2011). The temperature is determinant of the

maximum theoretical value to photogenerated current in function of the used solar simulator.

xi

List of figures

Figure 1.1 World energy consumption by energy source (1990-2040). ..................................... 1

Figure 1.5 Current performance and price of different PV module technologies. ..................... 4

Figure 2.2 Power spectrum of a blackbody sun at 5760 K, and power available to its optimum

bandgap cell. ............................................................................................................................. 14

Figure 2.6 Spectral distribution of solar radiation for different atmospheric conditions. ........ 18

Figure 2.9 Illuminated J–V characteristics at various temperatures under illumination intensities

of (a) 10 suns, (b) 15 suns. ....................................................................................................... 21

Figure 2.10 Solar cells efficiency versus temperature for various materials............................ 22

Figure 2.11 Charge carriers’ mobility for silicon doped at two different donor concentrations in

relation to temperature. ............................................................................................................. 23

Figure 2.13 Evolution of bandgap energy of silicon with temperature. A) Temperature range (0

K to 300 K) according to the limitations of the relations; B) Temperature range (288.15 K to

xii

343.15 K) according to the usual operation temperature of solar cells, out of relations ranges.

.................................................................................................................................................. 27

Figure 2.14 Photon flux (AM1.5G) and the integrated short circuit current density as a function

of wavelength (bandgap – top x axis)....................................................................................... 28

Figure 3.1 The low cost solar simulator (SOLSIM) developed by INPE in partnership with the

company Orbital Eng. Ltda. ..................................................................................................... 40

Figure 3.2 Spectral irradiance curves for AM1.5G standard and SOLSIM. ............................ 40

Figure 3.3 Instruments used to measure the solar cells temperature. A) K-type Thermocouple

and B) NI 9211 module. ........................................................................................................... 42

Figure 3.6 Polycrystalline silicon solar cell used in this work. The solar cell is of 15.6 cm length.

A) Specimen before cutting process; B) Specimen after cutting process................................. 44

Figure 3.7 Materials used in the cutting procedure. A) DREMEL 4000 and B) Diamond Wheel.

.................................................................................................................................................. 44

Figure 3.8 Instruments used to measure the solar cells voltage and current output. A) NI 9205

module and B) Digital Multimeter model VC8045-II. ............................................................. 45

Figure 3.9 Components of the extraction circuit. A) terminal universal connector (5 ways) and

resistors, B) Protoboard (EPB 0057) and C) jumpers (20 cm length). ..................................... 45

Figure 3.10 Electrical circuit assembled to enable the extraction of the electrical parameters of

the solar cells (current and output voltage). ............................................................................. 47

Figure 4.1 Photon flux (SOLSIM) and the integrated short circuit current density as a function

of wavelength (bandgap – top x axis) at temperature 298 K. ................................................... 52

xiii

Figure 4.2 Maximum theoretical achievable (A) integrated short circuit current density, open-

circuit voltage and (B) efficiencies for a single p–n junction solar cell at 298 K as a function of

the wavelength for the SOLSIM spectra. ................................................................................. 53

Figure 4.3 Theoretical variations in solar cell performance parameters: (A) 𝐽𝑠𝑐 and (B) 𝑉𝑜𝑐 for

Si, InP, GaAs, CdTe and CdS with temperature for solar cells in the temperature range 298 K–

528 K at SOLSIM spectra. ....................................................................................................... 54

Figure 4.4 The theoretical efficiency of solar cells as a function of (A) material, (B) wavelength

and temperature for SOLSIM spectra. ...................................................................................... 54

Figure 4.6 Temperature controlling using the ultra-thermostatic bath model MA – 185 of

Marconi. .................................................................................................................................... 56

Figure 4.7 Temperature effect on the IxV curve of the solar cell. ........................................... 57

xiv

List of tables

Table 1.1 Advantages and disadvantages of solar photovoltaics during operation. ................... 5

Table 2.1 Varshni equation constants for silicon semiconductor material taken from different

works. ....................................................................................................................................... 26

Table 2.2 The main goal of 10 different equivalent circuit models for PV cell. ...................... 34

Table 2.4 Spectral distribution of the reference radiation, where percentage is relative to the

potential share of total radiation. .............................................................................................. 36

Table 3.1 The nine different conditions used in the present work. .......................................... 43

Table 4.1 Load configuration used to extract the whole I x V curve of silicon solar cell. The

uncertain is also available. ........................................................................................................ 55

Table 4.5 General ranking of the equivalent models for PV cell. ............................................ 65

xv

ABNT Brazilian Association of Technical Standards

AM Air Mass

AM0 Solar spectrum incident in outer space (1,353 W/m2)

AM1.5D Solar spectrum incident on the earth (1,000 W/m2; the light is incident directly

without any contribution from diffuse rays)

AM1.5G Solar spectrum incident on the earth (1,000 W/m2; includes direct and diffuse

rays from the sun)

a-SiC:H Hydrogenated amorphous silicon carbide

a-SiGe Amorphous silicon germanium

ASTM American Society for Testing and Materials

CB Conduction band

CdS Cadmium sulfide

CdTe Cadmium telluride

CIGS Copper indium gallium selenide

c-Si Crystalline silicon

DCA Diffusion-Current Dominant Area

EIA U.S. Energy Information Administration

FF Fill Factor

GaAs Gallium arsenide

HIT Heterojunction with intrinsic thin layer

IEA International Energy Agency

IEC International Electrotechnical Commission

InP Indium phosphide

INPE National Institute of Space Research (Instituto Nacional de Pesquisas Espaciais

- Brazil)

LabMetrol Metrology Laboratory of the Federal University of Rio Grande do Norte

LED Light Emitting Diode

mc-Si Mono-crystalline silicon

MPPT Maximum Power Point Tracking

NBR ABNT standard designation

NI National Instruments

xvi

pc-Si Poly-crystalline silicon

PPGEM Post-Graduation Program in Mechanical Engineering

PV Photovoltaic

RCA Recombination-Current Dominant Area

SCLC Space-Charge Limited Current

Si Silicon

SOLSIM Solar Simulator developed by INPE

STC Standard Test Conditions

TiO2 Titanium dioxide

UFRN Federal University of Rio Grande do Norte

VB Valence band

xvii

List of symbols

𝑎 Ideality factor

𝐴 Constant in 𝐽𝑠 expression

𝑐 Velocity of light in vacuum (299 792 458 m/s)

𝐶 Constant in 𝐽𝑠 expression

∆𝜇 Chemical potential (eV)

∆𝑡 Time variation (s)

𝐷𝑛 , 𝐷𝑝 Diffusion constants of minority carriers in 𝑛 and 𝑝 region

𝐸𝑐 Conduction band level (eV)

𝐸𝑔 Bandgap (eV)

𝐸𝑔,𝑟𝑒𝑓 Bandgap at STC (eV)

𝐸𝑖 Fermi level in equilibrium

𝐸𝑣 Valence band level

𝐺 Solar irradiation (W/m2)

𝐺0 Solar constant (=1,367 W/m2)

𝐺𝑟𝑒𝑓 Reference solar irradiation (=1,000 W/m2)

ℎ Planck's constant (6.62607004×10−34 J∙s)

𝐼 Cell output current (A)

𝐼𝑚𝑝 Maximum power current (A)

𝐼𝑑 Diode current (A)

𝐼𝑠 Cell saturation current (A)

𝐼𝑟𝑠 Cell reverse saturation current (A)

𝐼𝑠𝑐 Short-circuit current (A)

𝐼𝑝𝑣 Light-generated current (A)

𝐼𝑝𝑣,𝑟𝑒𝑓 Light-generated current at STC (A)

𝐽 Cell output current density (A/cm2)

𝐽𝑚𝑝 Maximum power current density (A/cm2)

𝐽𝑑 Diode current density (A/cm2)

𝐽𝑠 Cell saturation current density (A/cm2)

𝐽𝑟𝑠 Cell reverse saturation current density (A/cm2)

xviii

𝐽𝑝𝑣 Light-generated current density (A/cm2)

𝐽𝑝𝑣,𝑟𝑒𝑓 Light-generated current at STC density (A/cm2)

𝑘 Boltzmann’s constant (1.380649 x 10−23 𝐽/𝐾)

k Wavevector

𝐿𝑛 , 𝐿𝑝 Diffusion lengths of minority carriers in 𝑛 and 𝑝 region

𝜆𝑔 Cut-off wavelength

M Air mass modifier

µ𝑛 , µ𝑝 Mobility of the electrons and holes

𝜇𝐼𝑠𝑐 Temperature coefficient of 𝐼𝑠𝑐 (A/K)

𝜇𝑉𝑜𝑐 Temperature coefficient of 𝑉𝑜𝑐 (V/K)

𝑛 Electron density

𝑛𝑖 Intrinsic carrier density

𝜂 Efficiency (%)

𝑁𝐴 , 𝑁𝐷 Density of acceptor and donor atoms

𝑝 Hole density

𝑃𝑚𝑝 Maximum power (W)

𝑞 Electron charge (1.602176634×10−19 C)

𝑅𝑠 Series resistance (Ω)

𝑅𝑠0 Reciprocal of the slope of the I–V curve for (0, 𝑉𝑜𝑐 ) (Ω)

𝑅𝑠,𝑟𝑒𝑓 Series resistance at STC (Ω)

𝑅𝑠ℎ Shunt resistance (Ω)

𝑅𝑠ℎ,𝑟𝑒𝑓 Shunt resistance at STC (Ω)

𝑅𝑠ℎ0 Reciprocal of the slope of the I–V curve for (𝐼𝑠𝑐 , 0) (Ω)

𝑆 Dimensionless coupling constant

𝜎 Conductivity (𝛺 −1 𝑐𝑚−1)

𝑇 Cell temperature (K)

𝑇𝑚 Module temperature (K)

𝑇𝑗 Solar cell junction temperature (K)

𝑇𝑟𝑒𝑓 Reference temperature (= 298.15 K)

𝑉 Cell output voltage (V)

xix

𝑉𝑡 Thermal voltage (K/eV)

𝑉𝑛 Ionization energy for electrons (meV)

𝑉𝑝 Ionization energy for holes (meV)

𝑉𝑜𝑐 Open-circuit voltage (V)

〈ћ𝜔〉 Average phonon energy (meV)

Content

RESUMO ................................................................................................................................. IX

ABSTRACT .............................................................................................................................. X

1. INTRODUCTION ............................................................................................................... 1

1.2. OBJECTIVES ....................................................................................................................... 7

1.3. WORK STRUCTURE ............................................................................................................. 7

2.2. SEMICONDUCTORS AND DOPING .......................................................................................... 9

2.2.1. N TYPE DOPING ......................................................................................................................... 11

2.2.2. P TYPE DOPING .......................................................................................................................... 12

2.2.3. EFFECTS OF HEAVY DOPING ..................................................................................................... 13

2.3. PHOTOVOLTAIC EFFECT ................................................................................................... 13

2.4. PHOTOVOLTAIC SOLAR CELL ............................................................................................ 14

2.4.1. THE CHARACTERISTIC CURVE .................................................................................................. 15

2.5. PERFORMANCE FACTORS .................................................................................................. 17

2.5.1. SPECTRAL DISTRIBUTION AND MASS OF AIR ............................................................................ 17

2.5.2. IRRADIANCE ............................................................................................................................. 19

2.5.3. TEMPERATURE.......................................................................................................................... 21

2.5.4. RELATIONS ............................................................................................................................... 24

xxi

2.7. SOLAR SIMULATORS ......................................................................................................... 34

2.8. CALCULATION OF UNCERTAINTIES ................................................................................... 36

2.8.1. SPREAD OF UNCERTAINTIES ..................................................................................................... 37

3.2. SOLAR CELL CHARACTERIZATION .................................................................................... 39

3.2.1. IRRADIATION ............................................................................................................................ 39

3.2.2. TEMPERATURE.......................................................................................................................... 41

3.2.3. ELECTRICAL PARAMETERS ....................................................................................................... 43

3.3. EQUIVALENT MODELS ANALYSIS ....................................................................................... 47

3.3.1. PV OPERATION CONDITIONS..................................................................................................... 48

3.3.2. TRANSLATION EQUATIONS ....................................................................................................... 48

3.3.3. PV TECHNOLOGY...................................................................................................................... 49

3.3.4. ASSUMPTIONS........................................................................................................................... 49

3.3.5. OVERVIEW ................................................................................................................................ 49

4.2. SOLAR CELL CHARACTERIZATION .................................................................................... 51

4.2.1. SOLSIM ANALYSIS .................................................................................................................. 51

4.2.2. RESISTORS ASSOCIATION ......................................................................................................... 55

4.2.3. TEMPERATURE.......................................................................................................................... 55

4.2.4. POLYCRYSTALLINE SILICON SOLAR CELL ANALYSIS ............................................................... 57

4.3. EQUIVALENT MODELS ANALYSIS ....................................................................................... 58

4.3.1. PV OPERATION CONDITIONS..................................................................................................... 59

4.3.2. TRANSLATION EQUATIONS ....................................................................................................... 61

4.3.3. PV TECHNOLOGY...................................................................................................................... 61

4.3.4. ASSUMPTIONS........................................................................................................................... 62

4.3.5. OVERVIEW ................................................................................................................................ 63

xxii

5. CONCLUSIONS ............................................................................................................... 66

REFERENCES ......................................................................................................................... 70

APPENDICES .......................................................................................................................... 78

ANNEXES ............................................................................................................................... 89

ANNEX B – CALIBRATION CERTIFICATE: THERMOCOUPLE A1 ................................................... 91

ANNEX C – CALIBRATION CERTIFICATE: THERMOCOUPLE A2 ................................................... 92

ANNEX D – CALIBRATION CERTIFICATE: THERMOCOUPLE A3 ................................................... 93

1

1. Introduction

Energy demand is rising due to rapid population growth and rising living standards.

Between 2015 and 2040, it was projected that world energy consumption will grow by 28%,

according to the latest International Energy Outlook 2017 of the U.S. Energy Information

Administration (EIA, 2017; See Figure 1.1). This worldwide increasing energy utilization is

one of the greatest challenges that the world is currently dealing with, since there are both the

increasing accumulation of greenhouse gases and the decreasing reserves of fossil fuels (Khan

et al., 2016).

From Figure 1.2, the waste of energy is brutal. To produce electricity, 66% is lost

(25.3% is lost of a total of 38.2%). There is a lot to do in physics, chemistry, thermodynamics,

efficiency, emissions, etc.

the years, and critical environmental issues that have increased the awareness to reduce the

climate change and global warming. Among the up-to-date energy scenarios, renewable energy

is predicted to be a notable part of energy production in the close future (Cuce et al., 2014).

There is a vast range of green technologies accessible for clean energy generation, and the

utilization of solar energy through photovoltaic (PV) cells has emerged as an auspicious source

of green energy since it is one of the most efficient, with large availability, reliable, and eco-

friendly solution (Slimani et al., 2017) for satiating the global power demand (Cuce et al., 2015)

Chapter 1. Introduction 2

and for dealing with fossil fuel-oriented environmental concerns. PV systems are free of moving

parts and present low noise level (Ishaque et al., 2011) and among renewables, solar PV's

provide the highest power density (University of Manitoba, 2017). In this respect, operation

and maintenance costs of such systems are notably low in comparison with other clean energy

generation technologies (Bianchini et al., 2016). Nowadays, over than 100 countries around the

globe are using solar PV's (Green Technology, 2015).

It is well-known that the PV power research has been progressing on two fronts, in the

effort to diminish the cost of solar energy: producing lower cost PV cells/devices and improving

their performances. The unceasing research and development in the solar cells area has

introduced novel design and concepts, new materials and processing methods, and cost

strategies (Yin et al., 2013) to diminish the cost of PV technology.

the first 15 years of the century XXI the growth rate of photovoltaic installations was of 41%

(Sampaio et al., 2017). Because of intensive attempts to reduce the cost of clean energy

generation from PV modules, current production cost of crystalline silicon PV modules is in

the range of 0.29–0.35 US$/W as reported by Louwen et al. (2016). The market for photovoltaic

Chapter 1. Introduction 3

systems will probably remain growing in the future as strongly as so far, due to the thrust of

subsidies, tax breaks and other financial incentives (Devabhaktuni et al., 2013).

A strategy to reduce costs is to obtain economies of scale. This was evident with the

development of crystalline silicon cells and is likely to be true for other technologies when their

production volumes increase. In addition to economies of scale a combination of technological

innovation, research in this field and improvement in learning are likely to reduce costs

significantly. This is shown by the learning curve of the PV modules in Figure 1.3.

According to Figure 1.3, in the last 35 years, the module price decreased by about

19.1% at every duplication of cumulative production modules. Many scientists and engineers

familiar with the variety of materials and PV technology concluded that photovoltaic materials

of thin film and the third-generation ones are the most likely candidates to continue the 80%

price reduction (Sampaio et al., 2017).

Over the years, the solar cells performance is increasing for the different available

technologies. Figure 1.4 presents the growing trend of the PV technologies efficiency. There

were already, in the laboratory, multi-junction solar cells with high solar concentration with a

46% efficiency in 2016 (Sampaio et al., 2017).

Chapter 1. Introduction 4

To evaluate the performance and price of different PV module technologies, one can

look into the studies conducted by the International Energy Agency (IEA) exemplified in Figure

1.5 which is also presented the market share for each technology.

Figure 1.5 Current performance and price of different PV module technologies.

Source: Sampaio et al. (2017)

technologies, are crucial aspects in accelerating the widespread adoption of photovoltaic

Chapter 1. Introduction 5

systems. These two aspects play a key role in climate policy (Torani et al., 2016). Many

countries, such as Germany, Denmark, Spain, China, Taiwan, United States, United Kingdom,

Japan, Sweden and South Korea have been using different mechanisms to encourage the

renewable energy’s use (Sampaio et al., 2017). According to the reports, the distance between

non-renewable and renewable energy resources is narrowed steadily (Cuce et al., 2017), and

the task of PV technology in this bias is of significant relevance (Cuce et al., 2014).

Even in the face of all this development in energy costs and efficiency, the feasibility

of using a PV plant, however, no mattering the PV technology chosen, depends on the

environmental parameters such as irradiance and cell temperature. Sampaio et al. (2017)

compiled the main advantages and disadvantages of PV solar energy in Table 1.1.

Table 1.1 Advantages and disadvantages of solar photovoltaics during operation.

Advantages Disadvantages

• Low cost of operation and maintenance systems on the market

• Low maintenance • High initial cost

• Free energy source • Needs a relatively large area of

• Clean Energy installation

• High Availability • High dependence on technology

• The generation can be made closer to development

the consumer • Geographical conditions (solar

• Does not cause environmental impacts irradiation, temperature)

/Environmental friendly

• Potential to mitigate emissions of

greenhouse gases

• Noiseless

Source: Sampaio et al. (2017)

The temperature variation caused by different operational conditions of a PV module

was found to affect its electrical efficiency, and to have a considerable value, thus it cannot be

neglected (Ayman et al., 2019). In PV technologies, the PV temperature and incident solar

radiation are the main factors that affect their behaviors and therefore should be considered

during their respective modelling. Others factors, such as ambient temperature, wind speed and

direction, dusty, humidity and mounting structure will have a more indirect impact, by

modifying how the main ones will behave (Santhakumari and Sagar, 2019).

Chapter 1. Introduction 6

at Standard Test Conditions (STC) – an irradiance level of 1,000 W/m2, a cell temperature of

25 °C and an air mass (AM) = 1.5 spectrum – or at the normal operating cell temperature

(NOCT) – this condition is defined by IEC 61215 (2016) standard, which is measured on an

open rack-mounted module with an inclination of 45°, an irradiance level of 800 W/m 2, an

ambient temperature of 20 °C and a wind speed of 1 m/s. However, in view of the

interdependent behavior described above, such specifications are not enough to describe all

situations that one PV module can operate. In fact, these conditions occur rarely on site since

solar radiation of 800-1,000 W/m2 makes it difficult to have a cell temperature of 20-25 °C

(DGS, 2017). This leads to the necessity of translation equations that can relate the STC or

NOCT to the real ones that are experimented by the PV cell. The focus of this work will be the

temperature, in view of the few number of investigations currently available in literature when

leading with the interference over the equivalent model selection and solar simulator analysis.

In this work, the analyses will depart from the final behavior of cell temperature after

being influenced by its secondary factors (i.e. wind speed, environment temperature, solar cell

geometry, etc.). The relation between the main and secondary factors can better understood at

the works of Rawat et al. (2017) and Akhsassi et al. (2018).

Aiming to bridge this gap on understanding the temperature influence over the solar

cells, a study on the impact of cell temperature on the performance of polycrystalline silicon

solar cell is undertaken in this work. The experiment will be carried out using a polycrystalline

silicon solar cell with the temperature in the range 20 – 60 °C at constant light intensity 1,000

W/m2 employing a solar simulator.

I) They include single crystal, multi-crystalline and the hetero-structure which have

more than 80% of the market share (Yeh and Yeh, 2013).

of 800 W/m2 and an ambient temperature of 20 °C, the common module

temperature is around 42 °C (Schwingshackl et al., 2013), while during summer

days in Central Europe, it can easily reach 60 °C;

Chapter 1. Introduction 7

1.2. Objectives

The main purpose of this research is identifying and evaluating the temperature

influence over the silicon polycrystalline solar cell technology, highlighting the most important

physical concepts associated.

I. Evaluate theoretical and empirically the temperature influence over the silicon

polycrystalline solar cell technology focusing on the performance parameters (open-

circuit voltage, short-circuit current, maximum extracted voltage, maximum extracted

current and efficiency);

II. Review the various existing mathematical relationships involving the influence of

temperature over solar cells;

III. Identify robust mathematical equivalent electric circuit models that emulate the solar

cell parameters (open-circuit voltage, short-circuit current, reverse saturation current

and efficiency) as well as its material property (bandgap) against the temperature

variation;

As mentioned at the end of the introduction, this work comprises a study on the impact

of cell temperature on the performance of polycrystalline silicon solar cell. In this way, all

subsequent chapters are divided, at a certain moment, into the study of each performance

parameter: as open-circuit voltage, short-circuit current, or efficiency; or material property:

bandgap.

Chapter two provides a bibliographical review of the main concepts used throughout

the work, starting from the principles of semiconductors and operation of a photovoltaic cell.

Shortly thereafter, it is presented the solar cell performance factors, the equivalent electric

models and it is discussed about the solar simulators.

Chapter three describes the method adopted for the execution of each stage of this

study. The instruments used for simulation and measurement are specified, as well as the

assumptions.

The results obtained are presented and discussed in chapter four. This chapter was

divided in two fronts: solar cell characterization and the equivalent model analysis.

Chapter 1. Introduction 8

Finally, chapter five presents the conclusions obtained in this work, from the

experiments carried out following the methodology defined in chapter three, and presents some

directions for future researches in the subject discussed in this dissertation.

9

2. Literature review

2.1. Introduction

Sampaio et al. (2017) made a review on photovoltaic solar energy and showed that its

potential is continuously increasing mainly due to efficiency’s increasing and costs’ reducing.

When the center of attention is the efficiency, there is many parameters that have its

own influence either direct (temperature and irradiance) or indirect (wind speed and direction,

environment temperature, solar cell geometry, mounting structure, plane of the array,

transmittance of the glazing-cover, absorbance of the plate and other factors).

In this bibliographic review, it will be presented the photovoltaic effect, the basics

principles of a solar cell and the main equivalent models for solar cells since it will be used to

the extraction of the solar cells performance parameters. It was discussed the importance and

some methods for temperature measurements, and finally, the influence of the temperature on

the performance parameters of solar cells.

When a pair of atoms are brought together into a molecule, their atomic orbitals

combine to form pairs of molecular orbitals arranged slightly higher and slightly lower in

energy than each original level. It is sad that the energy levels have split. When a very large

number of atoms come together in a solid, each atomic orbital split into a very large number of

levels, so close together in energy that they effectively form a continuum, or band, of allowed

levels. The bands due to different molecular orbitals may or may not overlap. The energy

distribution of the bands depends upon the electronic properties of the atoms and the strength

of the bonding between them.

Bands are occupied or not depending upon whether the original molecular orbitals

were occupied. The highest occupied band, which contains the valence electrons, is normally

called the valence band (VB). The lowest unoccupied band is called the conduction band (CB).

If the valence band is partly full, or if it overlaps in energy with the lowest unoccupied

band, the solid is a metal. In a metal, the availability of empty states at similar energies makes

it easy for a valence electron to be excited, or scattered, into a neighboring state. These electrons

can readily act as transporters of heat or charge, and so the solid conducts heat and electric

current.

Chapter 2. Literature review 10

If the valence band is completely full and separated from the next band by an energy

gap, then the solid is a semiconductor or an insulator. The electrons in the valence band are all

completely involved in bonding and cannot be easily removed. They require an energy

equivalent to the bandgap to be removed to the nearest available unoccupied level. These

materials therefore do not conduct heat or electricity easily.

Semiconductors are distinguished, roughly, as the group of materials with the bandgap

in the range 0.5 to 3 eV. Semiconductors have a small conductivity in the dark because only a

small number of valence electrons will have enough kinetic energy at room temperature to be

excited across the bandgap at room temperature. This intrinsic conductivity decreases with

increasing bandgap. Insulators are wider bandgap materials whose conductivity is negligible at

room temperature. Materials of bandgap < 0.5 eV have reasonably high conductivity and are

usually known as semimetals.

When the solid forms a regular crystal, then the energies of the bands, or the band

structure can be predicted exactly. Exactly which crystal structure a solid will adopt depends

upon the number of valence electrons and other factors. It will prefer a configuration that

minimizes the total energy. A bandgap is likely to arise in a crystal structure where all valence

electrons are used in bonding.

The semiconductor described so far is intrinsic – it is a perfect crystal containing no

impurities. The only energy levels permitted are the levels of well-defined wavevector k which

arise from the overlap of the atomic orbitals into crystal bands (See Nelson, 2008). The

properties of those bands determine the position of the Fermi level in equilibrium, 𝐸𝑖 , and the

intrinsic carrier density, 𝑛𝑖 , which have enough thermal energy to cross the bandgap and

conduct electricity. They also determine the conductivity since intrinsic semiconductor has 𝑛

electrons and 𝑝 holes available for conduction and conductivity 𝜎 is given by

where µ𝑛 , and µ𝑝 are the mobilities of the electrons and holes, respectively, and 𝑞 is the

electronic charge. In equilibrium, 𝑛, electron density, and 𝑝, hole density, are replaced with 𝑛𝑖

in Equation 2.1.

At room temperature, the conductivity of an intrinsic semiconductor is generally very

small. For example, for intrinsic silicon, 𝜎 = 3 × 10−6 𝛺 −1 𝑐𝑚−1, at 300 K. Conductivity

increases with increasing temperature, and with decreasing bandgap since 𝑛𝑖 varies as

𝐸𝑔

𝑒𝑥𝑝 (− 2𝑘𝑇). For example, germanium with a bandgap of 0.74 eV has a conductivity of 𝜎 =

Chapter 2. Literature review 11

2 × 10−2 𝛺 −1 𝑐𝑚−1 while gallium arsenide with a bandgap of 1.42 eV has a conductivity of

𝜎 = 10−8 𝛺 −1 𝑐𝑚−1, many orders of magnitude smaller.

Now suppose that the crystal is altered by introducing an impurity atom or a structural

defect. The impurity or defect introduces bonds of different strength to those which make up

the perfect crystal, and therefore changes the local distribution of electronic energy levels. The

altered energy levels are localized (unless the density of defects is very high), unlike the

extended k states which make up intrinsic conduction and valence band states.

If the impurity energy levels occur within the bandgap they can affect the electronic

properties of the semiconductor. Introducing occupied impurity levels above 𝐸𝑖 increases the

Fermi level, which increases the density of electrons relative to holes in equilibrium. In the

same way, unoccupied levels below 𝐸𝑖 reduce the Fermi level and increase the density of holes

relative to electrons. The density and the nature of the carriers in the semiconductor can thus be

controlled by adding definite amounts of impurities which energy levels close to the conduction

or valence band edge. This is called doping.

2.2.1. n type doping

A semiconductor which has been doped to increase the density of electrons relative to

holes is called n type. The principal charge carriers are negative. Occupied levels between 𝐸𝑖

and 𝐸𝑐 are introduced by replacing some of the atoms in the crystal lattice with impurity atoms

which possess one too many valence electrons for the number of crystals bonds. Such impurities

are called donor atoms because they donate an extra electron to the lattice. An example would

be an atom of phosphorus, which has five valence electrons, in the tetravalent silicon lattice.

The extra electron is not needed in the strong directional covalent bonds which hold each atom

to its neighbors. Therefore, it is much less well bound than the other valence electrons, and is

instead held rather loosely in a Coulombic bound state with Phosphorus atom. The donor can

be ionized relatively easily, leaving the extra electron free to move and the fixed donor atom

positively charged. For typical donors, the ionization energy, 𝑉𝑛 , is some meV or tens of meV.

If the impurity is chosen so that 𝑉𝑛 is small enough, virtually all the donor atoms will be ionized

at room temperature.

In the band picture, the donor electrons have been promoted from a level in the

bandgap, 𝑉𝑛 below 𝐸𝑐 , into the conduction band. Since the donor states are all filled at 𝑇 = 0,

the Fermi level must ow lie between the donor level and 𝐸𝑐 .

The density of carriers can be controlled by varying the density of dopants, 𝑁𝐷 . If

𝑁𝐷 ≫ 𝑛𝑖 and the donors are fully ionized at room temperature, then

Chapter 2. Literature review 12

𝑛 ≈ 𝑁𝐷 (2.2)

and

𝑛𝑖 2

𝑝= (2.3)

𝑁𝐷

since 𝑛𝑝 = 𝑛𝑖 2 at equilibrium. The electron density is now greatly increased over its

equilibrium value, while the hole density is greatly reduced. The electrons are here called the

majority carriers and the holes the minority carriers. Relative to the intrinsic case, the total

density of carries is increased, and so therefore is the conductivity. Conductions in n type

semiconductors is mainly by electrons.

2.2.2. p type doping

A semiconductor which is doped to increase the density of positive charge carriers

relative to negative is called p type. It is produced by replacing some of the atoms in the crystal

with acceptor impurity atoms, which contribute too few valence electrons for the bonds they

need to form. An example would be the trivalent element boron in silicon. The acceptor

becomes ionized by removing a valence electron from another bond to complete the bonding

between it and its four neighbors. This releases a hole into the valence band. The energy of this

ionized state is higher, by a small amount of energy 𝑉𝑝 , than the highest energy of a valence

band electron, and so the acceptor energy level appears in the bandgap close to the valence band

edge. The Fermi level now lies between the acceptor level and 𝐸𝑣 .

These p type semiconductors contain an excess of positive carriers, holes. For a doping

density of 𝑁𝐴 ≫ 𝑛𝑖 ionisable acceptors, it is found

𝑛 ≈ 𝑁𝐴 (2.4)

and

𝑛𝑖 2

𝑛= (2.5)

𝑁𝐴

Again, the total carrier density and hence the conductivity is greatly increased, but now

conduction is mainly by holes, which are the majority carriers in this case.

Notice how the changes in n and p are consistent with the changes in the Fermi level.

In an intrinsic (i.e. pure) semiconductor 𝐸𝐹 lies roughly in the middle of the bandgap. The effect

of doping is to shift the Fermi level away from the center of the bandgap, towards 𝐸𝑐 in n type

material, and towards 𝐸𝑣 in p type material.

Chapter 2. Literature review 13

Note also that, unlike heating and illumination, doping is a way of increasing the

conductivity of the semiconductor at equilibrium without requiring a constant input of energy.

2.2.3. Effects of heavy doping

Heavy doping causes imperfections in the crystal structure which appear as states in

the bandgap close to the CB and VB edges. These states have the effect of adding a tail to the

CB and VB density of states functions, and so effectively reducing the bandgap. The intrinsic

carrier density will increase, because thermal excitation across the bandgap is easier, and this

has consequences for solar cell performance. Heavy doping is also likely to increase the density

of defect states which act as centers for carrier recombination or trapping.

2.3. Photovoltaic effect

The conversion of solar irradiation into electricity occurs due to the photovoltaic

effect, which was observed by the first time by Becquerel in 1839. This effect occurs in

materials known as semiconductors, which present two energy bands, in one of them the

presence of electrons is allowed (valence bad) and in the other there is no presence of them,

i.e., the band in completely “empty” (conduction band).

energy in the form of photons (electromagnetic waves). According to this energy amount,

different phenomena will occur: Photons with energy 𝐸 < 𝐸𝑔 cannot promote an electron to the

excited state; Photons with 𝐸 ≥ 𝐸𝑔 can raise the electron, thereby creating mobile electron/hole

pairs (Sze, 1981) but any excess energy is quickly lost as heat as the carriers relax to the band

edges (Figure 2.1), delivering only ∆𝜇(= 𝑞𝑉𝑚 ) of electrical energy to the load, so only ∆𝜇/𝐸

of their power is available (Nelson, 2008). An absorbed photon with energy 𝐸 ≫ 𝐸𝑔 achieves

the same result as photon with 𝐸 = 𝐸𝑔 , all the extra energy becomes heat.

When the available data is the light spectral distribution, it is interesting make a

relation between the wavelength and photon energy. The cut-off wavelength of photons of

energy useful for carrier generation depends on 𝐸𝑔 . The cut-off wavelength is given by (Singh

and Ravindra, 2012)

1240

𝜆𝑔 = (𝑛𝑚) (2.6)

𝐸𝑔 (𝑒𝑉)

Chapter 2. Literature review 14

The Figure 2.2 shows how usable energy fraction falls as E increases. Even at 𝐸 = 𝐸𝑔

only a fraction of the incident power is available, since 𝑞𝑉𝑚 < 𝐸𝑔 . For this reason, it is the

incident photon flux and not the photon energy density which determines the photogeneration.

Figure 2.2 Power spectrum of a blackbody sun at 5760 K, and power available to its optimum bandgap

cell.

Virtually all photovoltaic devices incorporate a p-n junction in a semiconductor, which

through a photovoltage is developed. These devices are also known as solar cells or

photovoltaic cells. A typical solar cell is shown in Figure 2.3.

The PV solar cell has fundamentally two layers of individually doped semiconductor

material with its p-n junction exposed to incident irradiation (Jordehi, 2016; Villalva and

Gazoli, 2009). With the purpose of reducing the blockage of incident light, the metal grid on

the top side is constructed with thin and discontinuous structure with finger-like metal elements

Chapter 2. Literature review 15

ingrained into the silicon (Ciulla et al., 2014). It is designed to reduce the contact resistances

and to maximize the absorbing area (Jordehi, 2016).

In the presence of irradiation, as aforementioned, the p-n junction absorbs the photons

with energy greater than the bandgap of the semiconductor from incident light and create

carriers, namely electron-hole pairs. All the rest of the photons becomes heat.

These carriers are swept away, under the influence of the internal electric fields of the

p-n junction and create a current which is proportional to the incident radiation, to opposite

sides of a built-in voltage barrier between the dissimilar materials, where the charges are

collected at low resistance metallic contacts. The resulting electrical current is named light-

generated current and denoted by 𝐼𝑝𝑣 . It can be extracted if the two oppositely charged contacts

are then connected through an external load circuit. Desired currents and voltages are obtained

by connecting cells in series and parallel to form modules, and by aggregating modules in series

and parallel groupings to form arrays, both following the normal electrical aspects. Though the

magnitudes of currents and voltages differ, the electrical characteristics are similar for cells,

modules, or arrays.

equation, can only be achieved in certain configurations, such as a p-n junction. Otherwise, the

electron and hole gradients that are similar tend to cancel each other out. The net diffusion

currents usually arise only when the electron and hole carrier gradients are very different. In

this PV cell’s case, currents are dominated by minority carrier diffusion – See Nelson (2008).

The PV solar module are usually described by its characteristic curve (Figure 2.4)

which has three remarkable points, the 2-tuple: A - (0, 𝐼𝑠𝑐 ), B - (𝑉𝑚𝑝 , 𝐼𝑚𝑝 ) and C - (𝑉𝑜𝑐 , 0).

The 2-tuple A is the condition of short-circuit, that is, the voltage is equal to 0 and the current

Chapter 2. Literature review 16

is the short-circuit current (maximum value achievable); The 2-tuple B is the condition of

maximum power extraction, that is, both voltage and current are the ones that produce the

maximum power output; The 2-tuple C is the condition of open-circuit, that is, the current is

equal to 0 and the voltage is the open-circuit voltage (maximum value achievable).

In a typical solar cell, it can be concluded from (𝐼, 𝑉) = (𝐼𝑠𝑐 , 0) condition that the

Equation 2.7 is valid, i.e., the value of light-generated current (𝐼𝑝𝑣 ) can be assumed equal to the

value of short-circuit current (𝐼𝑠𝑐 ). (Mares et al., 2015)

The light-generated current density (𝐽𝑝𝑣 ) depends on the solar cell material and on the

given solar spectral irradiance and it is given by Equation 2.8. Where F(𝜆) represents the

incident photon flux in irradiance wavelength 𝜆, q is the electron charge and 𝐴 is the solar cell

superficial area.

𝜆𝑔

𝐽𝑝𝑣 = 𝑞 ∫ F(𝜆) 𝑑𝜆 (2.8)

0

The short-circuit current density 𝐽𝑠𝑐 , was calculated directly from the irradiance data. These

data are integrated by rectangle rule integration. The photon flux F(𝜆); is calculated from the

irradiance 𝐺(𝜆) and the wavelength 𝜆.

𝜆 ∙ 𝐺(𝜆) (2.9)

F(𝜆) =

ℎ∙𝑐

The current density at quantum efficiency of unity, when every incident photon creates an

electron, can be found by multiplying the photon flux with the electric charge 𝑞

Chapter 2. Literature review 17

𝐽𝑠𝑐 = 𝐹 ∙ 𝑞

The optical absorption is considered to be complete for energies higher than the bandgap. For

lower energies, no absorption takes place, no carriers are generated. For smaller quantum

efficiencies, a model for the optical absorption must be assumed, respectively.

Photovoltaic cells can have their operating performance changed according to the

current climatic conditions, impacting on their power generation capacity. These magnitudes

are directly linked to the geographical location of the operating environment, as well as other

local characteristics. These influencer factors can be divide in two groups in terms of how this

influence exists: direct and indirectly (Figure 2.5). In this work, as the focus is the temperature

influence, the other influencer factors will not be discussed in detail, just a un passant

explanation about spectral distribution, mass of air and irradiation since these aspects are

indispensable for understand completely the STC Condition.

The atmosphere changes the spectral distribution of light passing through it,

attenuating the intensity of its components (Figure 2.6). The spectrum also has a direct impact

on the photogenerated current, according to the spectral response of the cell (which depends on

the response of the semiconductor material from which it is made). Silicon, for example, has its

Chapter 2. Literature review 18

peak response near to the infrared region (900 nm), reaching red (740 nm) in the region of the

visible. The spectral distribution can also be found as spectral irradiance (𝐺λ ).

Figure 2.6 Spectral distribution of solar radiation for different atmospheric conditions.

The parameter AM (air mass) is associated with the angle presented by the light beam

when crossing a plane normal to the surface of the planet Earth. The AM 0 is the maximum

altitude, located above the atmosphere (free space); AM 1.5 is measured in tropical regions - in

Ecuador AM 1 is used, while AM 2 and AM 3 are used in regions of high latitudes.

𝐿 1

𝐴𝑀 = ≈ (2.10)

𝐿0 cos 𝑧

Where 𝐿 is the path length through the atmosphere, 𝐿0 is the path length at the zenith angle

(normal to the earth's surface) and 𝑧 is the zenith angle (in degrees). Figure 2.7 shows the values

of Air Mass according to the angle of the light beam relative to the zenith and the position of

entry into the Earth's atmosphere.

Chapter 2. Literature review 19

2.5.2. Irradiance

It is a unit of magnitude used to describe the incident power per unit of surface area,

the integral of the spectral distribution (in λ). As reference, there is the solar constant (𝐺0 ),

irradiance measured outside the atmosphere (imagining a flat horizontal surface perpendicular

to the sun's rays, oriented towards the zenith, at a mean Earth-Sun distance), whose value is

equal to 1,367 W/m2. According to the region of the planet, there are several different values

of irradiance (the highest values are in the equatorial regions and the lowest are in the poles).

The irradiance (𝐺) and temperature (T) is related to the 𝐼𝑠𝑐 of a cell by the approximate relation:

𝐺

𝐼𝑝𝑣 (𝐺) = [𝐼 + 𝜇𝐼𝑝𝑣 (𝑇 − 𝑇𝑟𝑒𝑓 )] (2.11)

𝐺𝑟𝑒𝑓 𝑝𝑣,𝑟𝑒𝑓

Where 𝐼𝑝𝑣,𝑟𝑒𝑓 is the short-circuit current, 𝑇𝑟𝑒𝑓 is the cell temperature, 𝐺𝑟𝑒𝑓 is the

irradiance in the STC condition and 𝜇𝐼𝑝𝑣 is the temperature coefficient of 𝐼𝑝𝑣 . This equation try

to describe the behaviour presented in Figure 2.8. From a certain value, irradiance increments

do not promote the growth of photogenerated current; this threshold value is called the

maximum current or saturation current of the cell (Carvalho ALC, 2014).

The air mass is related to the irradiance with the approximate relation (based on Laue's

observations):

Chapter 2. Literature review 20

where 𝐺0 is the radiance outside the atmosphere and 𝑧 is the distance from the zenith. The

constants 𝑐 and 𝑠 are empirical and are respectively 0.357 and 0.678.

Concentrated irradiance

Khan et al. (2016) carried out illuminated J–V measurements with temperatures in the

298–353 K interval, under a 𝐺 range of 2–18 suns, for a silicon solar cell. They observed

variations in the performance parameters under illumination conditions of up to 6 suns are like

those observed under normal illumination conditions. It should be noted that the methods

(analytical, curve fitting, and numerical) based on single J–V characteristics sometimes result—

at high temperatures—in negative values of 𝑅𝑠 and excessive values of 𝑎, owing to limitations

of these methods. The parameter variation trends at 18 suns are like the variation trends at 15

suns (Khan et al., 2016). Therefore, they performed detailed studies only for 𝐺 values of 10 and

15 suns. The experimentally measured illuminated J–V curves of the PV cell at various values

Chapter 2. Literature review 21

of T (298, 308, 318, 333, 343, and 353 K) are shown in Figure 2.9(a) and (b), for 𝐺 values of

10 and 15 suns, respectively.

Figure 2.9 Illuminated J–V characteristics at various temperatures under illumination intensities of (a)

10 suns, (b) 15 suns.

2.5.3. Temperature

Solar cells work best at certain temperatures, according to their material properties:

Some materials are more appropriate for use in orbit around the earth, some for terrestrial uses,

and others for high-temperature applications either under concentrated sunlight or in space near

the sun. Figure 2.10 shows how different cell materials lose efficiency with increasing

temperature. Note that at normal terrestrial temperatures, 25 °C, silicon's efficiency compares

favorably with other materials; but at high temperatures, 200 °C for instance, silicon's efficiency

has dropped to 5%, whereas the other materials are near 12%. Silicon is a good material for

ambient temperature terrestrial uses; it fails in high-temperature applications.

Although Figure 2.10 does not show it, there is a similar drop-off of efficiency below

a certain low temperature for each of the materials. Note that all materials lose efficiency in the

range shown.

High-Temperature Losses

The physical effects that determine efficiency's relationship with temperature are quite

complex. For the most part, two predominate in causing efficiency to drop as temperature rises:

As thermal energy increases, (1) lattice vibrations interfere with the free passage of charge

carriers and (2) the junction begins to lose its power to separate charges.

Chapter 2. Literature review 22

The first effect severely degrades silicon's performance even at room temperatures

(Figure 2.11). The ability of charge carriers to move without losing their energy is measured by

their mobility. Note that silicon is already below its maximum efficiency at room temperature.

The high-temperature losses shown here are mostly due to collisions with thermally excited

atoms.

Figure 2.10 Solar cells efficiency versus temperature for various materials.

The second effect does not occur until temperatures of about 300 °C for silicon are

reached. At such temperatures, a huge number of electrons in normal silicon bonds are

thermally jostled from their positions; on the n-type side, they join and greatly outnumber the

free electrons donated by the n-type dopant. At the same time holes are formed on the n-type

side (left behind by the thermally freed electrons); the n-type silicon begins to lose its n-type

character as the number of free electrons and holes become similar. The same process is

working on the p-type side, which is losing its p-type character. This leads to two effects: (1)

The thermally agitated charge carriers have so much energy, they cross over the junction in

both directions almost as if the barrier field were not there. (2) Ultimately the junction itself

disappears because there is no longer n- and p-type sides to induce it. All of these effects

accumulate to erode the activity of the cell, and efficiency diminishes to nearly zero.

Since solar cells are sensitive to temperature increases, and since a PV cell converts a

small portion of the irradiance into electrical energy while the remaining is converted into heat

(Fouad et al., 2017) which contributes to increase the cell temperature, it is frequently necessary

to either match the cell material to the temperature of operation or to continually cool it,

removing the extra, unwanted heat. Sometimes this latter method can lead to positive results,

Chapter 2. Literature review 23

raising a solar installation's overall efficiency if the heat is applied to useful purposes. The

overheating of the cell mainly occurs due to excessive solar radiation and high ambient

temperatures (Abd-Elhady et al., 2016).

Figure 2.11 Charge carriers’ mobility for silicon doped at two different donor concentrations in

relation to temperature.

Low-Temperature Losses

Low-temperature losses are, if anything, more complex and less understood. They are

important, however, only for deep-space PV applications. Two effects are thought to play roles:

(1) As temperature falls, thermal energy is less able to free charge carriers from either dopant

atoms or intrinsic silicon. Mobility for light-generated charge carriers drops because they

collide more frequently with ionized donors or acceptors in n- and p-type regions, respectively.

The donors and acceptors are not screened as much by clouds of thermally aroused charge

carriers. Also (2) at very low temperature, there is so little thermal energy that even dopants

behave as if they were normal silicon atoms. For instance, in the n-type material, donor atoms

retain their extra electrons; in the p-type material, holes remain fixed in place because electrons

are less likely to pop out of their normal positions to fill them. Since the n- and p-type sides no

longer exhibit their doped character, the junction disappears.

Chapter 2. Literature review 24

2.5.4. Relations

There are some proposed correlations in the literature that express the module

temperature as a function of variables such as the weather variables (depends on the location)

especially the ambient temperature, the local wind speed and direction, mounting structure

(Santhakumari and Sagar, 2019), as well as the solar or irradiance incidence on the plane of the

array (Fouad et al., 2017). Not only this but also the temperature depends on the material and

system-dependent properties such as the transmittance of the glazing-cover, absorbance of the

plate and other factors. The effect of different operational conditions on the temperature of a

PV module was found not to be negligible and it is also dependent on the actual electrical

efficiency due to the solar energy converted to heat (Ayman et al., 2019).

The study of the behavior of solar cells with temperature (𝑇) is important as, in

terrestrial applications, they are generally exposed to temperatures ranging from 15 °C (288 K)

to 50 °C (323 K) (Singh and Ravindra, 2012) and to even higher temperatures in space and

concentrator-systems (Zeitouny et al., 2018).

The cell temperature is the key environmental parameter that has great influence on

the behavior of a PV system, as it greatly affects the system efficiency and energy output. It is

the key to decide the quality and performance of a solar cell (Chander et al., 2015b).

Earlier studies (Singh and Ravindra, 2012) have pointed out that the performance of

solar cells degrades with increase in temperature. The effect of the temperature on the cell

performance is mainly reflected in short-circuit current which increase slightly, in open circuit

voltage which decrease substantially with temperature (Ayman et al., 2019), and in an overall

reduction on the maximum power point. These effects can be visualized in Figure 2.12.

Bandgap

The bandgap energy (𝐸𝑔 ) of semiconductors is the primordial property that renders

them a veritable source of fundamental and technological interest (Bensalem et al., 2017). The

gap energy is a fundamental concept, according to the quantum theory, in semiconductor

crystals; the lowest point of the conduction band and the highest point of the valence band are

separated by a forbidden region called bandgap. The behavior of the gap under external

disturbances such as pressure and temperature is a determinant factor of semiconductor-based

device performance. In fact, the study of the gap dependence with temperature of

Chapter 2. Literature review 25

silicon, the estimated gap value at the vicinity of the ambient temperature (300 K) is: 1.12 eV

for Collings (1980), 1.124 eV for Madelung (1991) and 1.123 eV for Bensalem et al. (2017).

Bludau et al. (1974) reevaluated experimentally the 𝐸𝑔 of silicon with good precision

between 2 and 300 K through absorption spectroscopy method. Based on the experimental

results, the authors performed a quadratic fit using the following approximation:

(2.13)

𝐸𝑔 = 𝐴 + 𝐵𝑇 + 𝐶𝑇 2

where,

0 𝐾 < 𝑇 ≤ 190 𝐾 150 𝐾 ≤ 𝑇 ≤ 300 𝐾

𝐴 = 1.170 eV A = 1.1785 eV

𝐵 = 1.059 × 10−5 eV/𝐾 𝐵 = −9.025 × 10−5 eV/𝐾

𝐶 = −6.05 × 10−7 eV/𝐾 2 𝐶 = −3.05 × 10−7 eV/𝐾 2

Chapter 2. Literature review 26

here it is possible to note that the intervals for each constants group (A, B and C) are conflicting

due to an interpolating technique that just aims the precision.

Another empirical equation well-known in the literature that describes the temperature

dependence of the bandgap is the empirical relationship of Varshni (1967; Equation 2.14). The

constants in this expression are different according to the work, some examples are presented

in Table 2.1.

𝛼𝑇 2

𝐸𝑔 (𝑇) = 𝐸𝑔 (0) − (2.14)

𝑇+𝛽

Table 2.1 Varshni equation constants for silicon semiconductor material taken from different works.

𝐸𝑔 (0) (𝑒𝑉) 𝛼×10−4 (𝑒𝑉/𝐾) 𝛽 (𝐾) Source of data

1.170 ± 0.001 4.73 ± 0.25 636 ± 50 Thurmond (1975)

1.1692 4.9 ± 0.2 655 ± 40 Alex et al. (1996)

O'Donnell and Chen (1991) proposed a new three-parameter fit to the temperature

dependence of semiconductor bandgaps. Their fitting is given by Equation 2.15.

〈ћ𝜔〉 (2.15)

𝐸𝑔 (𝑇) = 𝐸𝑔 (0) − 𝑆〈ћ𝜔〉 [coth ( ) − 1]

2𝑘𝑇

here, 𝑆 is a dimensionless coupling constant, 𝑘 is the Boltzmann’s constant

(1.380649 x 10−23 𝐽/𝐾; 8.6173324 x 10−5 𝑒𝑉/𝐾) and 〈ћ𝜔〉 is an average phonon energy. For

silicon in the temperature range 0 − 300𝐾, 𝑆 = 1.49, 〈ћ𝜔〉 = 25.2 𝑚𝑒𝑉 and 𝐸𝑔 (0) =

1.170 𝑒𝑉.

Bensalem et al. (2017) performed a research and proved that the provided values using

the parameters reported by Varshni (1967) disagree somehow with both Thurmond (1975) and

Alex et al. (1996) which tried to improve the values of Varshni Eq. (2.13) parameters, and this

is the object of both contributions. Bensalem et al. (2017) noted that, despite the simplifications

used in their analysis, the results agree well with the known values of silicon bandgap until 300

K. They guessed that the deviations compared to the results of Thurmond (1975) and O'Donnell

and Chen (1991) are since the coefficients of both contributions are very sensitive to the

temperature range, which is limited to 300 K. However, a good agreement is observed between

Chapter 2. Literature review 27

their results and those of Alex et al. (1996), which brought out the temperature dependence for

silicon bandgap up to 750 K. Thus, extrapolation is not recommended if high accuracy is

required.

A brief visual comparison among the aforementioned relations for bandgap and

temperature is presented in Figure 2.13. It is possible to note that in the temperature range of

288.15 K and 343.15 K presents a linear variation. Bensalem et al. (2017) proposed a linear

relationship and they observed that it has a high degree of confidence (regression coefficient is

close to one).

Figure 2.13 Evolution of bandgap energy of silicon with temperature. A) Temperature range (0 K to

300 K) according to the limitations of the relations; B) Temperature range (288.15 K to 343.15 K)

according to the usual operation temperature of solar cells, out of relations ranges.

Short-circuit current

The PV module temperature affects positively its current (Figure 2.12). It is observed

that the short circuit current (𝐼𝑠𝑐 ) showed only slight increase with the increase of the

temperature cell (Ayman et al., 2019).

Variation in short circuit current with temperature is primarily due to the change in

bandgap with temperature (Equations 2.13 to 2.15). 𝐼𝑠𝑐 increases with increasing temperature

due to decreasing bandgap; 𝐼𝑠𝑐 decreases with decreasing the temperature due to increasing

Chapter 2. Literature review 28

bandgap. A simple and commonly used relation that express the 𝐼𝑠𝑐 (𝐺, 𝑇) is the Equation 2.16

- a reformulation of Equation 2.11. The short-circuit current (𝐼𝑠𝑐 ) is proportional to the number

of generated charge carriers (intimately related to the bandgap) and mobility as well as it

depends strongly on the generation rate and the diffusion length (Chander et al., 2015b). The

rate of generation of charge carrier increases with cell temperature which causes an increment

in the short circuit current (Chander et al., 2015a).

𝐺 (2.16)

𝐼𝑠𝑐 (𝐺, 𝑇) = [𝐼 + 𝜇𝐼𝑆𝐶 (𝑇 − 𝑇𝑟𝑒𝑓 )]

𝐺𝑟𝑒𝑓 𝑠𝑐,𝑟𝑒𝑓

Generally, for most semiconductors, as the temperature increases, the bandgap

decreases. Consequently, the solar cell responds to longer wavelength regions in the solar

spectrum and 𝐼𝑠𝑐 increases (Equation 2.8), as well as the short-circuit current density (𝐽𝑠𝑐 ). Thus,

𝐼𝑠𝑐 is roughly proportional to the incident spectral intensity at wavelengths near the band edge

(Figure 2.2).

In Figure 2.14, it can be observed the short circuit current density for different

materials with different bandgaps under the AM1.5G spectral distribution.

Figure 2.14 Photon flux (AM1.5G) and the integrated short circuit current density as a function of

wavelength (bandgap – top x axis).

Chapter 2. Literature review 29

existed beyond the higher which may be attributed to the increase in rate of charge carrier

generation with cell temperature reveals the rapid increment in the reverse saturation current.

Reverse saturation current density (𝐽𝑠 ; the reverse saturation current divided by the

solar cell area) is a measure of the leakage (or recombination) of minority carriers across the p–

n junction in reverse bias (Singh and Ravindra, 2012). This leakage is a result of carrier

recombination in the neutral regions on either side of the junction and, therefore 𝐽𝑠 , primarily

controls the value of 𝑉𝑜𝑐 in the solar cells; higher 𝐽𝑠 corresponds to more leakage or

recombination of carriers whereas lower 𝐽𝑠 corresponds to less recombination. The minority

carriers are thermally generated; therefore, 𝐽𝑠 is highly sensitive to temperature changes.

Reverse saturation current density, 𝐽𝑠 , for a p–n junction solar cell, has been modeled (Hu, 1983)

as:

𝐷𝑛 𝐷𝑝

𝐽𝑠 = 𝑞 ( − )𝑛 2 (2.17)

𝐿𝑛 𝑁𝐴 𝐿𝑝 𝑁𝐷 𝑖

where, 𝑛𝑖 is the intrinsic carrier density, 𝑁𝐴 and 𝑁𝐷 are densities of acceptor and donor atoms,

𝐷𝑛 and 𝐷𝑝 are diffusion constants of minority carriers in 𝑝 and 𝑛 regions, 𝐿𝑛 and 𝐿𝑝 are

diffusion lengths of minority carriers in n and p regions, respectively. As from Equation 2.17,

𝐽𝑠 is strongly determined by the proportionality to ~𝑛𝑖 2 and 𝑛𝑖 can be represented as (Nelson,

2008):

2

𝐸𝑔 2𝜋𝑘𝑇 3 ∗3/2 ∗3/2 𝐸𝑔

𝑛𝑖 = 𝑁𝑐 𝑁𝑣 𝑒𝑥𝑝 (− ) = 4 ( 2 ) 𝑚𝑒 𝑚ℎ 𝑒𝑥𝑝 (− ) (2.18)

𝑎𝑘𝑇 ℎ 𝑎𝑘𝑇

where, 𝑁𝑐 , 𝑁𝑣 are effective density of states in conduction band, valance band and 𝑚𝑒 , 𝑚ℎ are

effective mass of electron, hole respectively.

Combining Equations 2.17 and 2.18, the expression for 𝐽𝑠 can be written in terms of

temperature and bandgap energy (Nell and Barnett, 1987) as:

𝐸𝑔

𝐽𝑠 = 𝐶𝑇 3 𝑒𝑥𝑝 (− ) (2.19)

𝑎𝑘𝑇

Chapter 2. Literature review 30

In Equation 2.19, doping and the material parameters of solar cells are combined in

this one constant 𝐶 (Nell and Barnett, 1987). The important solar cell parameters for the model

calculations are the temperature and bandgap. The higher the bandgap, lower will be the

saturation current density; Ge has very high 𝐽𝑠 values which can be attributed to the low bandgap

of Ge. The higher the temperature, higher will be the saturation current density.

An optimized value of 𝐶 = 17.90 𝑚𝐴 𝑐𝑚−2 𝐾 3 has been used by Nell and Barnet

(1987) for the calculation of performance parameters of solar cells made from a variety of

materials (Nell and Barnet, 1987). Therefore, for an optimized solar cell design, a minimum

value of 𝐶 is required and in reality, a single value is not applicable to all materials. Fan et al.

(1982) chose a value of 𝐶 = 50 𝑚𝐴 𝑐𝑚−2 𝐾 3 . Nell and Barnet (1987) investigated the effect of

various values of 𝐶 on the performance parameters of solar cells and found that the constant 𝐶

affects the efficiency logarithmically and a reasonable variation in the value of 𝐶, around 𝐶 =

17.90 𝑚𝐴 𝑐𝑚−2 𝐾 3 , will only have a modest effect on the overall system efficiency.

Green (1982) has proposed a simple empirical relation for 𝐽𝑠 where the product 𝐶𝑇 3 is

replaced by a constant 𝐴 = 1.5×108 𝑚𝐴 𝑐𝑚−2 :

𝐸𝑔

𝐽𝑠 = 𝐴 𝑒𝑥𝑝 (− ) (2.20)

𝑎𝑘𝑇

Loferski (1956) has discussed the cases where 𝐴 = 1×108 𝑚𝐴 𝑐𝑚−2 and 𝐴 =

1.44×108 𝑚𝐴 𝑐𝑚−2 .

Singh and Ravindra (2012) showed that the performance of cells for case 𝐴 =

1.5×108 𝑚𝐴 𝑐𝑚−2 gives the best agreement between the calculated and available theoretical

and experimental data for solar cells made from Si, Ge and GaAs whereas, for InP, CdTe and

CdS based solar cells, case 𝐶 = 17.90 𝑚𝐴 𝑐𝑚−2 𝐾 3 seems to be more appropriate at 298 K.

Moreover, as temperature changes, cases 𝐶 = 17.90 𝑚𝐴 𝑐𝑚−2 𝐾 3 and 𝐶 = 50 𝑚𝐴 𝑐𝑚−2 𝐾 3 are

more suitable to describe the performance of solar cells.

Open-circuit voltage

The PV cell temperature affects negatively its voltage. The open circuit voltage is

observed to decrease with cell temperature and is strongly depended on it (Ayman et al., 2019).

The most used relation for 𝑉𝑂𝐶 (𝑇) is

Chapter 2. Literature review 31

𝐺 (2.21)

𝑉𝑂𝐶 (𝑇) = 𝑉𝑂𝐶,𝑟𝑒𝑓 + 𝑎𝑉𝑡 ln ( ) + 𝜇𝑉𝑜𝑐 (𝑇 − 𝑇𝑟𝑒𝑓 )

𝐺𝑟𝑒𝑓

According to Orioli and Di Gangi (2013), the above expression of 𝑉𝑂𝐶 (Equation 2.16),

is quite imprecise because it was obtained from model 1D2R on the basis of the simplified

hypotheses of the 1D1R model, in which it is 𝑅𝑠ℎ = ∞. Moreover, when the irradiance tends

to zero, Equation 2.21 yields an unrealistic value of the open circuit voltage (𝑉𝑜𝑐 → −∞.). Thus,

these authors proposed the Equation 2.22 based on the I–V characteristics of 108 models of PV

panels (among heterojunction with intrinsic thin layer, polycrystalline and monocrystalline

silicon) issued on the Internet by 23 manufacturers.

2 3

𝐺 𝐺 𝐺 (2.22)

𝑉𝑜𝑐 = 𝑉𝑜𝑐,𝑟𝑒𝑓 + {𝐶1 𝑙𝑛 ( ) + 𝐶2 𝑙𝑛 ( ) + 𝐶3 𝑙𝑛 ( ) } + 𝜇𝑉𝑜𝑐 (𝑇 − 𝑇𝑟𝑒𝑓 )

𝐺𝑟𝑒𝑓 𝐺𝑟𝑒𝑓 𝐺𝑟𝑒𝑓

From the 1D2R model at the open-circuit voltage (0, 𝑉𝑜𝑐 ), it can be extracted that:

𝑞𝑉𝑜𝑐 𝑉𝑜𝑐

𝐼𝑝𝑣 = 𝐼𝑠 [exp ( ) − 1] + (2.23)

𝑎𝑘𝑇 𝑅𝑠ℎ

Assuming

𝑉𝑜𝑐 𝑞𝑉𝑜𝑐

≪1 exp ( )≫1 𝐼𝑝𝑣 ≈ 𝐼𝑠𝑐

𝑅𝑠ℎ 𝑎𝑘𝑇

𝑎𝑘𝑇 𝐽𝑠𝑐

𝑉𝑜𝑐 = 𝑙𝑛 ( ) (2.24)

𝑞 𝐽𝑠

𝐸𝑔 𝑎𝑘𝑇 𝐶𝑇 3

𝑉𝑜𝑐 = − (𝑙𝑛 ) (2.25)

𝑞 𝑞 𝐽𝑠𝑐

Therefore, 𝑉𝑜𝑐 decreases with an increasing temperature. All terms in Equation 2.25

affects negatively the voltage value. Including the first: the energy bandgap decreases with cell

temperature and consequently open circuit voltage is decreased. Thus, it strongly depended on

temperature.

Chapter 2. Literature review 32

The temperature influence over the open circuit voltage (𝑉𝑜𝑐 ) can be explained since it

is proportional to the contact potential difference (𝑉𝐷 ) which is given by relation concerned

(Wen et al., 2012).

𝑘𝑇 𝑁𝐷 𝑁𝐴

𝑉𝐷 = ( ) 𝑙𝑛 ( 2 ) (2.26)

𝑞 𝑛𝑖

intrinsic concentration. As cell temperature increases, 𝑛𝑖 increases rapidly and consequently

both 𝑉𝐷 and 𝑉𝑜𝑐 decreases.

The temperature dependence of 𝑉𝑜𝑐 can be obtained both from Equation 2.24 and 2.25,

in case of derivate from the first one, it is obtained as

= + ( − ) (2.27)

𝑑𝑇 𝑇 𝑞 𝐽𝑠𝑐 𝑑𝑇 𝐽0 𝑑𝑇

The mathematical modelling of PV cells is crucial for purposes of design, simulation,

assessment, manage and optimisation of solar PV systems (Hansen, 2015). Furthermore, it is

also decisive for proper computations and maximum power point tracking (MPPT) of PV

systems. It is completely necessary to extract maximum power of cells because of economic

reasons and growing demand for energy. So many methods have been presented for tracking

the maximum power point of PV cells and the most common of them have been reviewed by

Danandeh and Mousavi (2018).

The equivalent circuit models are the well-known ways for modelling PV cells

(Jordehi, 2016), however, there are other approaches for modelling PV cells. Furthermore,

proper modelling of PV cells encompasses not just proper circuit model, but precise circuit

model parameters (Jordehi, 2016). A challenging problem in the field of renewable energies is

achieving the circuit model parameters of PV cells which is a nonlinear optimisation problem

since the I–V curve of PV cells is nonlinear. A proper parameter estimation method for PV cells

should:

• Present repeatability, i.e., when it is applied to a specific condition for multiple times,

similar results are achieved;

Chapter 2. Literature review 33

• Be robust, i.e., be stable while delivering accurate model parameters, in other words,

model parameters which lead to I–V data or remarkable I–V points as close as possible

to the manufacturer information and/or experimental data even with variations in entries

(Jordehi, 2016);

• Low time-consuming, especially when it is applied for Maximum Power Point Tracking

(MPPT) (Ram et al., 2017).

A huge deal of study has already been performed to solve “PV cell equivalent circuit

model parameter estimation problem”, however, efforts in research are still being put to

effectively solve this problem. (Jordehi, 2016)

During the past decades, there were many efforts to model the solar cell. The

conception of a circuit model usually starts from the basic principles of physics of

semiconductors, taking into account the influence of ambient conditions that is analysed from

the thermodynamics. There is no general agreement on which singular equation which can be

used for modelling the I–V characteristic of the PV cell. However, most models of the solar

cells have as starting point the Shockley theory of illuminated p-n junction (Mares et al., 2015).

Thus, the equivalent circuit of the solar cell is described at different levels of approximation

(Mares et al., 2015). These distinct models, i.e., models with different levels of emulation

capability are described next, together with some comments on their particularities. It is

important to note that the PV module’s I–V characteristic results from the combination of the

I–V characteristics of the solar cells that constitute it.

Resuming the main characteristics of each model available (Table 2.2) is pertinent to

allow the interested people to choose the best option at specific applications since the models

have different levels of emulation capability, and some particularities, as above mentioned.

Chapter 2. Literature review 34

Table 2.2 The main goal of 10 different equivalent circuit models for PV cell.

Model Main goal

1D Explain fundamental concepts of PV cells. It cannot emulate the behavior of

physical PV cells.

1D1R Take into account dissipative effects, and the sum of several contact

resistance and the resistance of the materials which compose the module and

causes a reduction on the power converted by this device.

1D2R Be more robust in relation to temperature influence. Emulates the behavior of

construction defects which cause leakage currents within the PV cell.

2D2R Improve the accuracy deteriorates at low irradiances. Does not assume that

the recombination loss in the depletion region is absent.

1D2R1C Emulate transient processes of the voltage on PV cell terminals in the

instantaneous load connection/disconnection.

2D4R Proportionate the division of the current into two components associated to

power factor and other as a loss factor, so it is possible that solar cells can be

compared with each other, and the way to improve solar cell performance

may be easily found.

3D2R Consider: 1) the polaron-pair generation rate as constant for a given light

intensity, rather than the current; 2) the current depends on competition

between polaron recombination and its dissociation/collection by the

electrodes.

3D5R Model small size solar cells since it present leakage current through

peripheries.

xD2R Model different PV cell technologies to achieve low modelling error.

2D1R1Rv Emulate other conduction mechanisms which are described by a nonlinear

dependence and most of the classical analysis does not considered. It was used

an SCLC mechanism.

2.7. Solar simulators

Outdoor experimental studies play an extremely important role in the research and

development of solar energy technologies. Natural sunlight is unstable and unrepeatable, and is

always contingent upon the availability of the sun, which increases the difficulty and cost of

experiments and reduces the reproducibility of experimental results. In contrast, indoor solar

simulators, which can provide radiation approximating natural sunlight in both intensity and

spectrum, offer numerous advantages in terms of size, cost, operational flexibility and weather

Chapter 2. Literature review 35

solar energy, and experiments using solar simulators are controllable, reliable, and repeatable.

Solar simulators fall into two major categories: non-concentrating solar simulators that

produce uniformly distributed quasi-collimated light, and concentrating solar simulators that

produce high-flux radiation. The former is mainly used for testing photovoltaic (PV) cells, solar

water heater concentrators and spacecrafts. The latter is mainly employed for high-temperature

solar thermal and solar thermochemical research. (Jian et al., 2019)

A non-concentrating solar simulator generates uniform and quasi-collimated light with

both the spectral distribution and radiation intensity mimicking sunlight.

In the extraterrestrial solar spectrum found in orbit around the planet Earth, known as

AM0 (air mass null), the intensity is 1,367 W/m2 and the terrestrial spectrum found on the

surface after the AM0 has crossed the atmosphere layer is known as AM1.5G and has an

intensity of 1,000 W/m2. The uniformity of natural sunlight is much better than 1% due to the

Earth-Sun distance. The temporal stability, too, is better than 1% for open skies with suitable

climatic conditions for a short period of a few minutes (Veissid and Vaz, 2011).

The ABNT NBR 11879 Standard (1991) establishes performance requirements for

solar simulators, in terms of the parameters discussed above. Table 2.3 shows the requirements

needed to classify a solar simulator.

Table 2.3 Classification of a solar simulator relative to its characteristics.

Attribute Class A Class B Class C

Uniformity of radiation ≤ ± 2% ≤ ± 5% ≤ ± 10%

Temporal stability ≤ ± 2% ≤ ± 5% ≤ ± 10%

Spectral match ≤ ± 25% ≤ ± 40% ≤ +100 e – 60%

The spectral match corresponds to the percentage ratio (value measured by the default

value), specified in Table 2.4, for each wavelength range. The wavelength ranges in this table

are suitable for characterization of silicon solar cells. Therefore, solar simulators for use with

cells from other materials need current standards covering a wider range of wavelengths.

A typical solar simulator consists of a discharge lamp in high pressure Xenon gas, 10

MPa when lit, power supply, ellipsoidal reflector, 45 ° mirrors, optical integrator, shutter, filter

holder and collimating lens. Normally, with a 1,000 W lamp, it illuminates an area of 10 cm x

10 cm. The price of an equipment, with this description, is in the order of US $ 50,000. There

are also solar pulsed light simulators used to illuminate large areas. For example, Spectrolab's

Large Area Pulsed Solar Simulator (LAPSS) costs around $ 300,000 and illuminates an area of

Chapter 2. Literature review 36

two meters in diameter during the 10 ms light pulse. This type of equipment allows to obtain

the IxV characteristic curve of a solar panel without changing its temperature (Veissid and Vaz,

2011).

Table 2.4 Spectral distribution of the reference radiation, where percentage is relative to the potential

share of total radiation.

Percentage of total radiation in Percentage of total radiation in

Wavelength range (µm)

AM1,5G (%) AM0 (%)

0,4 a 0,5 18,5 20,5

0,5 a 0,6 20,1 20,5

0,6 a 0,7 18,3 17,5

0,7 a 0,8 14,8 14,0

0,8 a 0,9 12,2 11,2

0,9 a 1,1 16,1 16,3

The analysis of a physics experiment requires the evaluation of the reliability of the

measured values through the instruments. It is not possible to find the exact value of a physical

variable, since it can only be measured with an instrument, which always has limitations.

However, many variables, such as the electron charge and the universal constant of the gases,

have a very determined value, that we will denominate here of true value. Thus, in a measure,

we obtain a value close to the true value. Although it is impossible to determine the difference

between the measured value and the true value, since the latter is unknown, we can define

quantities that reflect this difference in some way, the so-called uncertainties, directly related

to the concept of precision, which is larger in the measure the smaller the associated uncertainty.

Unfortunately, we cannot quantify the uncertainty simply by averaging the difference between

the measured value and the true value, because the difference, which can be both positive and

negative, has no null average value, nothing informative. This forces us to resort to more

complex mathematical elaborations.

Physics makes use of probability and statistical theories in the representation of its

measurements. For the majority of experiments, the expression form of the value of a variable

x will be done by means of the mean of the values measured in a set of N measures of it, that

is, {𝑥𝑖 | 𝑖 = 1, 2, . . . , 𝑁}. The average value of the variable x is given by:

𝑁

1

𝑥̅ = ∑ 𝑥𝑖 (2.28)

𝑁

𝑖=1

Chapter 2. Literature review 37

The value of the uncertainty associated with the measurements of the magnitude 𝑥, in

turn, can be taken from the variance (𝜎 2 ) of the set of N measures:

𝑁

1

𝜎𝑥2 = ∑(𝑥𝑖 − 𝑥̅ )2 (2.29)

𝑁−1

𝑖=1

Note that the variance does not have the same physical dimension as the magnitude

([𝜎𝑥2 ] = [𝑥 2 ] ≠ [𝑥]), so that the uncertainty associated with the measure of magnitude 𝑥 is

defined by the standard deviation of the set of N measures:

𝜎 = √𝜎𝑥2 (2.30)

The standard deviation, 𝜎, is, therefore, a definite positive quantity because of the

radiciation operation, and represents the uncertainty related to each of the data, that is, it

evaluates the distance between each measurement and the true value. Considering that the mean

was calculated on N measurements, the standard representation of the magnitude x is given by:

𝑥 = 𝑥̅ ± 𝜎𝑥 (2.31)

Where the uncertainty of the mean value is described by Equation 2.32, by the so-

called mean standard deviation, 𝜎𝑥 :

𝜎

𝜎𝑥 = (2.32)

√𝑁

Let 𝑓 be a quantity dependent on other quantities 𝑎, 𝑏, ⋯ , 𝑧, independent of each other.

Thus, 𝑓 can be represented as a function of several variables: 𝑓 = 𝑓(𝑎, 𝑏, . . . , 𝑧).

The uncertainty in 𝑓 is given by the uncertainty propagation formula, stated as follows:

the standard deviation of 𝑓, 𝜎𝑓 , is the square root of the variance of 𝑓, calculated as

2 2 2

𝜕𝑓 𝜕𝑓 𝜕𝑓

𝜎𝑓2 2

= ( | ) 𝜎𝑎 + ( | ) 𝜎𝑏 + ⋯ + ( | ) 𝜎𝑧2

2 (2.33)

𝜕𝑎 𝑎=𝑎̅ 𝜕𝑏 𝑏=𝑏̅ 𝜕𝑧 𝑧=𝑧̅

Validity conditions

Equation 2.33 has general validity provided since two conditions are obeyed:

i) The set variables are statistically independent. This means that none of the variables can

be calculated from one or some of the variables in the set, nor does their measurement

interfere with the measurement of another. If this happens, the equation must be

expanded in the expression of 𝑓 before applying the propagation formula. What happens

is that each term in Equation 2.33 must represent all the dependence of the function with

Chapter 2. Literature review 38

respect to the variable on which the partial derivative was calculated; there can be no

underlying dependencies on the other variables. Note that if any of the quantities are

measured directly from other quantities that are not in this set, we can apply the

uncertainty propagation formula in isolation to them and use the calculated variances in

the propagation formula.

ii) For each variable x, we need the inequality 𝑥̅ ≫ 𝜎𝑥 . In many of the propagation

formulas we will see, the variables appear in the denominator, which causes problems

if the variable can be null, since its inverse is not defined. Even if it is not null, when

𝑥̅ ≈ 𝜎𝑥 , there is a probability that a given observation of x gives zero result. Equation

2.33 for the propagation of uncertainties assumes that observations can be repeated

infinitely; the result is nothing more than the expected value for the average of infinite

observations, so we can never use an undefined value when using equation.

Chapter 3. Materials and methods 39

3.1. General considerations

The photovoltaic solar cells in this work is made of silicon as is used in PV plants all

over the world. It is the most common type employed.

polycrystalline solar cell technology focusing on the performance parameters (open-circuit

voltage, short-circuit current, it is important to determine the maximum extracted voltage,

maximum extracted current, maximum extracted power and efficiency for a specific spectral

distribution.

This work also reviewed the various existing mathematical relationships involving the

influence of temperature over solar cells and identify robust mathematical equivalent electric

circuit models that emulate the solar cell parameters (open-circuit voltage, short-circuit current,

reverse saturation current and efficiency) as well as its material property (bandgap) against the

temperature variation.

All necessary materials and the used methods are described below.

3.2. Solar cell characterization

3.2.1. Irradiation

The low cost solar simulator (SOLSIM; Figure 3.1) developed by INPE in partnership

with the company Orbital Eng. Ltda was used to supply the irradiation. It consists of a set of

two different types of lamps and the AM1.5G spectrum (class A; Figure 3.2) is obtained by the

superposition of the individual spectra of each type of lamp.

Veissid and Vaz (2011) made a comparison between the values of the light-generated

current of SOLSIM and other solar simulators in AM1,5G irradiation following the standards

NBR 11879 (1991), IEC 60904-9 (2007) and ASTM E 927-85. The simulation was done

considering spectral irradiance curves of different solar simulators and spectral response of

different solar cells. SOLSIM and X25 were the equipment which presented the lowest average

standard deviation, around 1%, in relation to the values of light-generated current calculated for

the AM1.5G standard. In addition to this fact, SOLSIM is the equipment that shows the best

trade-off between accuracy and capital feasibility since its price (fifteen thousand dollars) is

one third order of the other equipment.

Chapter 3. Materials and methods 40

Figure 3.1 The low cost solar simulator (SOLSIM) developed by INPE in partnership with the

company Orbital Eng. Ltda.

Technical characteristics

Non-uniformity Class A

Temporal instability Class B

Spectral match ratio Class B

Figure 3.2 Spectral irradiance curves for AM1.5G standard and SOLSIM.

SOLSIM characterization

It is important to know the maximum theoretic values that are achievable by any solar

cell using a specific solar simulator – in this research case is the SOLSIM – since becomes

possible to create a new scale for comparisons.

From the SOLSIM solar spectral irradiance (Figure 3.2) and using Equations 2.7-2.9

could be found the relation between the 𝐽𝑠𝑐 and the λ. Also from the SOLSIM solar spectral

irradiance, but now using 𝐽𝑠𝑐 and Equations 2.19 and 2.24, could be found the relation between

Chapter 3. Materials and methods 41

the 𝑉𝑜𝑐 and the λ. Furthermore, using the relations between the 𝑉𝑜𝑐 and the λ, the 𝐽𝑠𝑐 and the λ,

could be found the relation between the 𝜂 and the λ.

The efficiency of a solar cell is the ratio of the power output corresponding to the

maximum power point to the power input and is represented as:

𝜂= = (3.1)

𝐺∙𝐴 𝐺

excellent accuracy.

𝑉𝑜𝑐 𝑉

⁄𝑉 − 𝑙𝑛 ( 𝑜𝑐⁄𝑉 + 0.72)

𝑡ℎ 𝑡ℎ

𝐹𝐹 = (3.2)

𝑉𝑜𝑐

⁄𝑉 + 1

𝑡ℎ

3.2.2. Temperature

Monitoring

Temperature monitoring was achieved by using K-type thermocouples connected to a

NI 9211 module from National Instruments (Figure 3.3). All thermocouples were calibrated at

the Metrology Laboratory of the Federal University of Rio Grande do Norte

(LabMetrol/UFRN). The calibration certificates are in the annexes (Annexes A-D).

Although the thermocouples had a large measurement range (-270 to 1200 ° C) that

could lead to inaccuracy during measurements when calibrating them by comparison, the

thermocouples showed an uncertainty of ± 0.5 °C. The uncertainties presented have a

confidence level of approximately 95% and coverage factor k = 2.1, according to EA-4/02.

Besides a level of uncertainty appropriate to the purpose of this paper, they are easy to acquire

and use.

Chapter 3. Materials and methods 42

Figure 3.3 Instruments used to measure the solar cells temperature. A) K-type Thermocouple and B)

NI 9211 module.

Controlling

The temperature is a parameter that is necessary to control due to the scope of the

present work. It is required to the system a temperature range of 20 to 60 °C. This variation

become possible by using an ultra-thermostatic bath (from Marconi and model MA - 185)

integrated to the SOLSIM simulator. This equipment works as a heat exchanger that uses the

water as fluid; It has a cool/heat system, besides the pump that enables the water circulation.

Aiming the objective of this work of evaluate empirically the temperature influence

over the silicon polycrystalline solar cell technology focusing on the performance parameters,

it was chosen nine conditions presented in Table 3.1. These conditions were chosen based on

the range adopted to these work (temperature varying from approximately 20 to 60 °C) and an

increment near to 5 °C. The irradiation was maintained constant to avoid its undesirable

influences.

Chapter 3. Materials and methods 43

Table 3.1 The nine different conditions used in the present work.

Condition Irradiation (W/m²) Temperature (°C)

#1 1,000 19.1

#2 1,000 24.0

#3 1,000 28.7

#4 1,000 34.1

#5 1,000 38.8

#6 1,000 44.9

#7 1,000 49.1

#8 1,000 54.2

#9 1,000 59.0

Specimen conception

There are any necessary adaptions to become possible use the SOLSIM due to its light

incident area. The local where the solar cells are positioned in the SOLSIM is of circle shape

of a 19.4 cm diameter (Figure 3.5). This dimension is not adequate to support the whole solar

cell used in this work which presents a squares shape of a 15.6 cm length (Figure 3.6).

Given this limitation, a cutting process was performed to provide an appropriate size

to the specimen. This procedure was resumed to a removal of an isosceles triangle with 2.5 cm

side length in the specimen corns (superficial area reduction of 5,14%). Following this

proceeding, the initial form (Figure 3.6A) was modified to an adequate shape (Figure 3.6B).

The cutting process was performed using a DREMEL 4000 and a Diamond wheel (Figure 3.7)

refrigerated by water in order to proportionate a regular edge with negligible shunt resistance.

Voltage monitoring was achieved by using metal wires connected to a NI 9205 module

from National Instruments (Figure 3.8A; considered as a voltmeter). The current monitoring

was achieved by also using metal wires but connected to a multimeter from Electronics

Instruments (Figure 3.8B; considered as an ammeter).

Chapter 3. Materials and methods 44

Figure 3.6 Polycrystalline silicon solar cell used in this work. The solar cell is of 15.6 cm length. A)

Specimen before cutting process; B) Specimen after cutting process.

Figure 3.7 Materials used in the cutting procedure. A) DREMEL 4000 and B) Diamond Wheel.

Chapter 3. Materials and methods 45

Figure 3.8 Instruments used to measure the solar cells voltage and current output. A) NI 9205 module

and B) Digital Multimeter model VC8045-II.

In order to simulate the load, proportionate the connections and, thus, extract the

electrical parameters, the following items (Figure 3.9) were used:

i. Protoboard (EPB0057);

ii. Jumpers (20 cm length);

iii. Resistors (1.0, 1.5, 10 and 150 Ω);

iv. Quick splicing connectors (5 ways).

Figure 3.9 Components of the extraction circuit. A) terminal universal connector (5 ways) and

resistors, B) Protoboard (EPB 0057) and C) jumpers (20 cm length).

Moreover, the electrical circuit used to determine the electrical parameters of the solar

cell must have a variable load form that enables the extraction of the entire IxV curve. This

variable load was obtained from associations (in series and / or parallel) of the resistors.

However, using this extraction method, the phenomenon of uncertainty propagation (see

Subsection 2.8) arises, which must and were calculated from Equation 2.33. The propagation

of uncertainty generally occurred in two ways:

Chapter 3. Materials and methods 46

From literature:

𝑅𝑒𝑞 = 𝑅1 + 𝑅2 (3.3)

Then,

2 2

𝜕𝑅𝑒𝑞 𝜕𝑅𝑒𝑞

𝜎𝑅2𝑒𝑞 =( | ) 𝜎𝑅21 + ( | ) 𝜎𝑅22

𝜕𝑅1 𝑅 ̅̅̅̅ 𝜕𝑅2 𝑅 ̅̅̅̅

1 =𝑅 1 2 =𝑅 2

From literature:

𝑅1 ∙ 𝑅2

𝑅𝑒𝑞 = 𝑅1 | 𝑅2 = (3.5)

𝑅1 + 𝑅2

Then,

2 2

2

𝜕𝑅𝑒𝑞 𝜕𝑅𝑒𝑞

𝜎𝑅𝑒𝑞 =( | ) 𝜎𝑅21 + ( | ) 𝜎𝑅22

𝜕𝑅1 𝑅 ̅̅̅̅ 𝜕𝑅2 𝑅 ̅̅̅̅

1 =𝑅 1 2 =𝑅 2

2 2

𝑅2 2 2

𝑅1 2

=( ) 𝜎𝑅1 + ( ) 𝜎𝑅22

(𝑅1 + 𝑅2 )2 (𝑅1 + 𝑅2 )2

2 2

𝑅2 2 𝑅1 2

∴ 𝜎𝑅𝑒𝑞 = √( 2

) 𝜎𝑅1 + ( ) 𝜎𝑅22 (3.6)

(𝑅1 + 𝑅2 )2 (𝑅1 + 𝑅2 )2

A schematic drawn of the used general electrical circuit is presented in Figure 3.10.

The voltmeter is used in parallel to the load while the ammeter is used in series. The internal

resistance associated with both the voltmeter and the ammeter was considered negligible due

to the accuracy of the respective apparatus. In addition, all the results of this work will be

expressed as percentages, so that possible errors lose their significance.

Chapter 3. Materials and methods 47

Figure 3.10 Electrical circuit assembled to enable the extraction of the electrical parameters of the

solar cells (current and output voltage).

In the used PC-based measurement and control system, the NI hardware products were

connected to a laptop through USB. This kind of system has two main architectures: directly

connect a multifunction I/O device to the PC or connect a CompactDAQ Chassis to the PC, and

populate this chassis with conditioned I/O modules, this last one was the option used in the

present work, since the CompactDAQ provides the most customizable solution with direct

sensor connectivity. It was used the CompactDAQ NI cDAQ-9172 from National Instruments

(Figure 3.11).

Arguably, equivalent models have several applicability. However, each model has its

peculiarities and are designed to be applied under certain solar cells and specific operating

conditions. Thus, it is necessary to define robust mathematical equivalent electric circuit models

Chapter 3. Materials and methods 48

that emulate polycrystalline silicon solar cell which have their operating temperature varying

while operating. However, it is not feasible to analyze all equivalent models due to its

abundance, so it is necessary to choose a group of models and make a comparison between

them.

In order to know what models should be compared, an intensive research on the

literature was carried out aiming to made a list of the available equivalent circuit models, and

to establish relations based on the available information. The process of ranking the models

analyzed by this work was divided in two parts: 1) literature research; and 2) comparative and

qualitative analysis according to an error analysis, operation conditions – focusing on

temperature –, Translation equations and PV technology.

In the literature, it is possible to realize that there are many relations among the existent

models. These relations were separated in three groups: 1) Worse than: it was used when during

the article the author(s) prove by error data that his (their) model is more efficient than the other

in question; 2) Indecisive: it was used when the model was compare to others, but no

quantitative results were given in order to affirmed which is more efficient. This second group

also involves the models that were just compare to datasheet or experimental data; and 3) Base

for: it was used when the model is, evidently, a base for another one and there are not relations

that can be classified in the first two groups.

As already discussed, the operation conditions of a PV module affect directly on its

performance. These distinct circumstances are due to locality, seasons or yet to Earth’s position

to the sun and other stars. Thus, it is important a ranking where the criterion is the “PV operation

conditions” to better emulate in different conditions more accurately.

The ranking on “PV operation conditions” criteria was made based on the following

aspects of the models: 1) Different operation conditions – focusing on temperature; 2)

Experimental data from literature and not just datasheet information. These criteria were chosen

since the model would better emulate at different operation conditions, using also experimental

data. When necessary, once more, the tie-breaking criterion was to be the most recent model.

The presence of translation equations is important since they reduce the necessity of

large data since in possession of a unique condition data, it is possible to calculate the

Chapter 3. Materials and methods 49

parameters in any other. Besides, using this bias it is possible to relate the operation conditions

directly to each parameter. The ranking on “Translation equation” criteria was made based on

the following aspects of the models: 1) Number of parameters having translation equations; 2)

Equations recently developed; 3) Number of parameters considered. These criteria were chosen

since the model would better emulate the real condition if the maximum of parameters is

considered using recent well-established equations. When necessary, once more, the tie-

breaking criterion was to be the most recent model.

3.3.3. PV technology

The models are usually said to be useful in many different PV technologies. However,

it is worthy to highlight that the performance generally goes down when a model is used in a

situation other than that used to validate itself. Thus, it is pertinent to classify the selected

models based on the cell material.

The ranking on “PV technology” criteria was made based on the following aspects of

the models: 1) Number of PV module/cells; 2) Different PV technologies. These criteria were

chosen since the model which has the largest and varied sample, it could represent more

accurately the entire PV technology population, i.e., it would better emulate different PV

module technologies with less dispersion. When necessary, the tie-breaking criterion was to be

the most recent model. It was supposed that be better than the previous models under the same

aspects is a condition to be accepted by the scientific community.

3.3.4. Assumptions

The use of assumptions is common for all models. It is used aiming the simplification

of some calculus. Its presence is greater mainly in models that present transcendental equations.

However, its easiness results in loss of accuracy, thus, it is important to identify and discuss the

main assumptions made by the models.

3.3.5. Overview

For conceiving a general ranking involving all criteria, some aspects were established.

Firstly, it was considered the models that present comparisons to any work in the literature, in

specific, the relation “worse than”. This was the main aspect to rank the models since the

superiority was declared by the authors themselves. Secondly, it was considered the different

operation conditions followed by different PV technology. This sequence was chosen because

all models use silicon technology, which represents 80% of the PV all market (Sudhakar et al.,

Chapter 3. Materials and methods 50

2016), i.e., the models that use others technologies, despite emulate in a broader variety of

technology, the difference is balanced due to the percentage of each technology

commercialized. Finally, it was computed the ranking of translation equations.

Chapter 4. Results and discussions 51

4.1. Initial considerations

The brief results obtained during the accomplishment of this work are separated in: the

analysis of the simulator SOLSIM considering the variation of temperature and solar cell

material; the errors associated to the extraction circuit develop in this work; and the output of

the system of temperature monitoring and controlling

Then, it is presented the main results of this research: the influence of the temperature

in the range of 20 to 60 °C on the polycrystalline silicon solar cells; and the equivalent model

selection considering different aspects.

4.2.1. SOLSIM analysis

The maximum achievable short circuit current density, 𝐽𝑠𝑐 , dependent on the bandgap,

is calculated directly from SOLSIM spectral data. The units of spectral data are typically given

as spectral irradiance (Wm -2 nm -1), which can be easily converted to photon flux (m -2 s -1 nm

-1

). Equations 2.8 and 2.9 can be used to calculate the maximum attainable 𝐽𝑠𝑐 for a solar cell,

which is the integral of the SOLSIM photon flux curve up to the bandgap of the absorber

materials (Figure 4.1). The variation in photon flux and 𝐽𝑠𝑐 with wavelength (or bandgap-top x

axis) is illustrated in Figure 4.1. The maximum obtainable 𝐽𝑠𝑐 for solar cells based on materials

Si, InP, GaAs, CdTe and CdS are denoted by dashed lines in Figure 4.1; The bandgap energy

values for these materials were the same of Singh and Ravindra (2012). The maximum 𝐽𝑠𝑐 at

-2

temperature 298 K for solar cells based on these materials in mA cm are: Si - 42.72, InP -

35.28, GaAs - 31.14, CdTe - 27.17 and CdS - 5.12. These values are compatible to the ones

reported by other authors (Sze, 1981; Singh and Ravindra, 2012) as well as to the reported

experimental values (Green et al., 2011).

Furthermore, it is important to discuss the variation in maximum achievable efficiency

with wavelength when using the simulator SOLSIM. Figure 4.2A shows the maximum

achievable integrated short circuit current density, open-circuit voltage and Figure 4.2B shows

the maximum achievable efficiencies for a single p–n junction solar cell at 298 K as a function

of the wavelength for the SOLSIM spectra. It can be seen from the values that they are in

accordance to the literature (Green et al., 2011; Singh and Ravindra, 2012). As can be seen

from Figure 4.2B, the materials with the optimum bandgap are GaAs, InP and CdTe with a

bandgap of 1.43 eV, 1.31 eV, and 1.54 eV respectively having highest efficiencies of 32.66,

Chapter 4. Results and discussions 52

32.64 and 31.48% respectively at 298 K for SOLSIM spectra. Therefore, in Figure 4.2B, at the

maximum theoretical efficiency, an exact choice of the corresponding wavelength and so the

band gap or material can be chosen. However, comparing Figure 4.2B to the literature, it is

observer that there is difference between the spectra SOLSIM and AM1.5G and therefore it

must be weighed in any study using this simulator.

Figure 4.1 Photon flux (SOLSIM) and the integrated short circuit current density as a function of

wavelength (bandgap – top x axis) at temperature 298 K.

Figure 4.3A presents the variation in 𝐽𝑠𝑐 with temperature using SOLSIM spectra for

solar cells based on Si, InP, GaAs, CdTe and CdS. 𝐽𝑠𝑐 increases with increasing temperature

whereas it decreases with increasing bandgap. For CdS based cells, the difference in 𝑑𝐽𝑠𝑐 /𝑑𝑇

values in comparison to Singh and Ravindra (2012) values is high. Variation in short circuit

current density with temperature is primarily due to the change in bandgap with temperature

(Sze, 1981). Generally, for most semiconductors, as the temperature increases, the bandgap

decreases (Sze, 1981). Consequently, the solar cell responds to longer wavelength regions in

the solar spectrum and 𝐽𝑠𝑐 increases. Thus, 𝐽𝑠𝑐 is roughly proportional to the incident spectral

intensity at wavelengths near the band edge.

The variation in the open circuit voltage for solar cells based on Si, InP, GaAs, CdTe

and CdS in the temperature range 273–523 K using SOLSIM spectra is shown in Figure 4.3B.

Open-circuit voltages shown in Figure 4.3B are determined from the calculated 𝐽𝑠𝑐 and 𝐽𝑠 using

Equation 2.19 for 𝐶 = 17.90 𝑚𝐴 𝑐𝑚−2 𝐾 3 since it was observed by Nell and Barnet (1987) that

Chapter 4. Results and discussions 53

will only have a modest effect on the overall system efficiency, besides, it was observed by

Singh and Ravindra (2012) that this value is more suitable to describe the performance of solar

cells as temperature changes. According to Equation 2.24, the voltage will increase if the light

intensity is increased. For solar cells based on InP and CdTe, the 𝑉𝑜𝑐 calculated are very close

to the experimentally achieved 𝑉𝑜𝑐 by Singh and Ravindra (2012). Nonetheless, the values

calculated in this work are close to the values reported by Wysocki and Rappaport (1960).

According to Figure 4.3B, 𝑉𝑜𝑐 decreases with increasing temperature whereas it increases with

increasing bandgap. For solar cells based on Si, at 298 K, 𝑉𝑜𝑐 value is 0.6948 V and decreases

further to 0.2239 V at 528 K. The decrease in bandgap with increasing temperature results in

lower 𝑉𝑜𝑐 . The slight difference in both reported in literature (Fan et al. 1982; Wysocki and

Rappaport, 1960) and calculated 𝑉𝑜𝑐 in this work may be attributed to different spectra as well

as to the 𝐽𝑠 values.

Figure 4.2 Maximum theoretical achievable (A) integrated short circuit current density, open-circuit

voltage and (B) efficiencies for a single p–n junction solar cell at 298 K as a function of the

wavelength for the SOLSIM spectra.

temperature changes is shown in Figure 4.4 for solar cells based on Si, InP, GaAs, CdTe and

CdS in the temperature range 298–528 K. The efficiencies have been calculated for SOLSIM

spectra. As temperature increases, efficiency decreases and the maximum achievable efficiency

shifts towards lower wavelengths, that is, higher bandgap materials. According to Figure 4.4B,

Chapter 4. Results and discussions 54

the optimum cut-off wavelength shifts from 907.53 nm (bandgap of 1.37 eV) at 298 K to 825,54

nm (bandgap of 1.50 eV) at 528 K.

Figure 4.3 Theoretical variations in solar cell performance parameters: (A) 𝐽𝑠𝑐 and (B) 𝑉𝑜𝑐 for Si, InP,

GaAs, CdTe and CdS with temperature for solar cells in the temperature range 298 K–528 K at

SOLSIM spectra.

Figure 4.4 The theoretical efficiency of solar cells as a function of (A) material, (B) wavelength and

temperature for SOLSIM spectra.

Chapter 4. Results and discussions 55

After testing 17 solar cells, it was possible to define 20 values, in ohms (Ω), for the

extraction electric circuit load, and to define how to associate resistors of 1, 1.5, 10 and 150 Ω

to reach these values. The uncertainty associated with each of these configurations has been

determined (Table 4.1).

Table 4.1 Load configuration used to extract the whole I x V curve of silicon solar cell. The uncertain

is also available.

Associated

Equivalent uncertain

# Configuration

resistance (Ω)

(Ω) (%)

1 1R0 | 1R0 | 1R0 0.33 0.1 30

2 1R0 | 1R0 0.50 0.1 20

3 1R0 1.00 0.05 5

4 1R0 + (1R0 | 1R0) 1.50 0.15 10

5 1R0 + 1R5 2.50 0.18 7.07

6 (1R5 + 1R5) + (1R0 | 1R0 | 1R0) 3.33 0.249 7.47

7 10R0 | 10R0 5.00 1 20

8 (10R0 | 10R0) + 1R0 6.00 1.05 17.5

9 (10R0 | 10R0) + 1R0 + 1R0 7.00 1.1 15.7

10 (10R0 | 10R0) + 1R0 + 1R0 + 1R0 8.00 1.15 14.4

11 10R0 10.00 0.5 5

12 10R0 + 10R0 20.00 1.41 7.07

13 10R0 + 10R0 + 10R0 30.00 2.56 8.66

14 10R0 + 10R0 + 10R0 + 10R0 40.00 4 10

15 10R0 + 10R0 + 10R0 + 10R0 + 10R0 50.00 5.60 11.18

16 10R0 + 10R0 + 10R0 + 10R0 + 10R0 + 10R0 60.00 7.35 12.24

17 10R0 + 10R0 + 10R0 + 10R0 + 10R0 + 10R0 + 10R0 70.00 9.26 13.23

18 150R0 150.00 7.5 5

19 150R0 + 150R0 300.00 21.21 7.07

20 20 * (150R0) 3000.00 670.82 22.36

4.2.3. Temperature

Monitoring

The temperature monitoring was successfully registered (Figure 4.5) for all nine

conditions. All conditions (different temperatures) presented a steady-state, without any

significant variation.

Chapter 4. Results and discussions 56

Controlling

The initial procedures to guarantee the temperature controlling was successfully

performed (Figure 4.6). The input temperature that was registered in the ultra-thermostatic bath

and the output temperature of the solar cell surface present a linear relation. This relation type

is the easiest one when the goal is controlling.

Figure 4.6 Temperature controlling using the ultra-thermostatic bath model MA – 185 of Marconi.

Solar cell temperature (°C)

R² = 0.9994

60,0

50,0

40,0

30,0

20,0

10,0

10,0 15,0 20,0 25,0 30,0 35,0 40,0 45,0 50,0 55,0 60,0

Setting temperature (°C)

In order to know how linear is the relation, it was made a linear fitting. As expected,

the linear fitting is adequate because the R-square was 0.9994.

From the Equation (4.1) obtained from the linear fitting and knowing the smallest

possible variation of temperature sourced by ultra-thermostatic bath (0.1 °C), the minimum

interval for temperature variation, it is approximately 0.101 °C.

Chapter 4. Results and discussions 57

Figure 4.7 Temperature effect on the IxV curve of the solar cell.

150

Current (mA)

100

50

0

0,00 0,20 0,40 0,60

Voltage (V)

100 100

Current (mA)

Current (mA)

80 80

60 60

40 40

20 20

0 0

0,00 0,20 0,40 0,60 0,00 0,20 0,40 0,60

Voltage (V) Voltage (V)

80 80

Current (mA)

Current (mA)

60 60

40 40

20 20

0 0

0,00 0,20 0,40 0,60 0,00 0,10 0,20 0,30 0,40 0,50

Voltage (V) Voltage (V)

50 40

Current (mA)

Current (mA)

40 30

30

20

20

10 10

0 0

0,00 0,10 0,20 0,30 0,40 0,50 0,00 0,20 0,40 0,60

Voltage (V) Voltage (V)

Chapter 4. Results and discussions 58

25 20

Current (mA)

Current (mA)

20 15

15

10

10

5 5

0 0

0,00 0,10 0,20 0,30 0,40 0,50 0,00 0,10 0,20 0,30 0,40 0,50

Voltage (V) Voltage (V)

The intensive research on the literature results in a list of the available equivalent

circuit models (Figure 4.8), it establishes the three relations types (‘Worse than’, ‘Indecisive’

and ‘Base for’; See subsection 3.3) based on the available information with regard to the error

analysis. When a flowchart ends in a certain model, it means that the comparison was limited

to datasheet or experimental data. Not all the possible blocks were included in order to avoid

overfloat.

terms of operation conditions, translation equations, PV technology and assumptions and time-

consuming level; Their box are highlighted on green background. The corresponding literature

are: Chaibi et al. (2018); Boutana et al. (2017); Jaimes and Sousa (2017); Mares et al. (2015);

Cibira and Koscová (2014); Lineykin et al. (2014); Peng et al. (2014); Mahmoud and Xiao

(2013); Orioli and Gangi (2013); and Ishaque et al. (2011). This prior selection method was

based on the most recent models and on their accuracy.

Chapter 4. Results and discussions 59

Following the ranking on “PV operation conditions” aspects: 1) Different operation

conditions – focusing on temperature; 2) Experimental data from literature and not just

datasheet information. Finally, considering the tie-breaking as the most recent model, the

resulting rank is presented in Table 4.2.

Chapter 4. Results and discussions 60

Temperature (°C) Radiation (W/m²) Experimental

N Literature

-25 0 20 30 40 50 60 75 200 400 500 600 800 1,000 (G;T)

pc-Si pc-Si

Isaque et al. pc-Si pc-Si pc-Si pc-Si mc- pc-Si pc-Si mc-

1 mc-Si mc-Si mc-Si - mc-Si mc-Si pc-Si - -

(2011) HIT mc-Si mc-Si Si mc-Si Si HIT

HIT HIT

pc-Si

Chaibi et al. mc- pc-Si pc-Si pc-Si pc-Si mc- pc-Si pc-Si mc-

2 - - mc-Si (55) mc-Si - - -

(2018) Si mc-Si mc-Si mc-Si Si mc-Si Si

mc-Si

pc-Si

Lineykin et al. pc-Si mc- pc-Si mc- pc-Si pc-Si pc-Si mc-

3 mc-Si mc-Si - - - - mc-Si mc-Si pc-Si -

(2014) Si Si mc-Si mc-Si Si

(300)

pc-Si pc-Si pc-Si

Orioli and Di pc-Si mc- pc-Si mc- pc-Si mc- pc-Si mc-

4 - - - - - - mc-Si mc-Si - mc-Si -

Gangi (2013) Si HIT Si HIT Si HIT Si HIT

HIT HIT HIT

(776.23;31.57)

pc-Si

Boutana et al. mc-Si pc-Si mc- (800;35)

5 - - - - - pc-Si - pc-Si mc-Si - pc-Si -

(2017) CIGS Si CIGS (997;53.47)

CIGS

(738.52;29,84)

Mahmoud and pc-Si mc-

6 - - - - - pc-Si - pc-Si - pc-Si - pc-Si pc-Si -

Xiao (2013) Si HIT

mc-

Peng et al.

7 - - - Si - - - - mc-Si - - - mc-Si mc-Si -

(2014) (33)

Cibira and (930;68)

8 - - - - - - - - - - - - - -

Koscová (2014) (570;40)

Mares et al. pc-Si

9 - - - - - - - - - - - - - -

(2015) mc-Si

10 Jaimes and

- - - - - - - - - - - - - - -

0 Sousa (2017)

*Use lm/m² and not W/m²

The values between parentheses is the exact one that was used.

In the temperature variation, it is implicit the value of the irradiation (1,000 W/m²). In the radiation variation, it is implicit the value of the temperature (25 °C).

Chapter 4. Results and discussions 61

Following the ranking on “Translation equation” aspects: 1) Number of parameters

having translation equations; 2) Equations recently developed; 3) Number of parameters

considered. Finally, considering the tie-breaking as the most recent model the resulting rank

can be seen in Table 4.3.

Translation Equations

#N Literature

𝐼𝑝𝑣 𝐼𝑠𝑐 𝑉𝑜𝑐 𝐼𝑠 𝐸𝑔 𝑎

1 Orioli and Di Gangi (2013) (2.10) - (2.17) - x -

2 Peng et al. (2014) - (2.15) b (2.16) - x cte

3 Boutana et al. (2017) x (2.15) (2.16) x x x

d

4 Chaibi et al. (2018) (2.10) - - (2.14) cte -

5 Cibira and Koscová (2014) (2.10) - - (2.14) d cte cte

6 Lineykin et al. (2014) (2.10) a - - (2.14) d cte 1<a<2

c

7 Mahmoud and Xiao (2013) (2.10) - (2.16) - x -

8 Isaque et al. (2011) (2.10) - (2.16) c - x cte

9 Jaimes and Sousa (2017) x

10 Mares et al. (2015) x

- : the relation is implicit according to the others relations established. cte: it is considered constant.

𝜇𝐼𝑝𝑣 𝐺

a b

: 𝜇𝐼𝑝𝑣,𝑛𝑒𝑤 = : 𝜇𝐼𝑝𝑣 ,𝑛𝑒𝑤 = 𝜇𝐼𝑝𝑣

𝐼𝑝𝑣,𝑟𝑒𝑓 𝐺𝑟𝑒𝑓

c d

: The variation on irradiance was neglected. : It is achieved by the relation "𝐴 ∙ (𝐽𝑠 /𝐽𝑠,𝑟𝑒𝑓 )".

x: This parameter it is not used in the model.

4.3.3. PV technology

Following the ranking on “PV technology” aspects: 1) Number of PV module/cells; 2)

Different PV technologies. Besides, considering the tie-breaking as the most recent model, it

was achieved the ranking in Table 4.4.

Chapter 4. Results and discussions 62

#N Literature mc-Si pc-Si CdTe CIGS HIT

1 Orioli and Di Gangi (2013) 55(2) 76(1) 14(1)

2 Mahmoud and Xiao (2013) 4(4) 7(6) 2(2)

3 Isaque et al. (2011) 3(3) 2(2) 1(1)

4 Mares et al. (2015) 10(2) 8(4)

5 Boutana et al. (2017) 1(1) 1(1) 1(1) 1(1)

6 Lineykin et al. (2014) 2(2) 2(2)

7 Chaibi et al. (2018) 1(1) 1(1)

8 Peng et al. (2014) 2(2)

9 Jaimes and Sousa (2017) 1(1)

10 Cibira and Koscová (2014) 1(1)

Number of PV modules used in the work (number of PV modules used to evaluate the model)

4.3.4. Assumptions

Chaibi et al. (2018), Cibira and Koscová (2014) and Lineykin et al. (2014) assumed

that the bandgap is constant in function of temperature variation. This assumption brings not

just loss of accuracy, but it does not represent the reality properly. It is recommended the use

of some equation to relate the bandgap energy and the temperature since it gives back to the

model the physical meaning and it do not increase the time-consuming because it is a simple

equation.

(2014), Lineykin et al. (2014), Peng et al. (2014). This method reduces the time-consuming

since it was developed to solve exponential equations since it can be seen as the inverse function

of 𝑓(𝑊).

The assumption of 𝐼𝑝𝑣,𝑟𝑒𝑓 = 𝐼𝑠𝑐,𝑟𝑒𝑓 used by Chaibi et al. (2018), and Cibira and

Koscová (2014) is widely accepted, however, the use of 𝑅𝑠ℎ ≫ 𝑅𝑠 used by Peng et al. (2014)

and Chaibi et al. (2018) and 𝑅𝑠 = 𝐼𝑠𝑐 /40 used by Cibira and Koscová (2014) are not. The

acceptation or rejection by the scientific community is related, respectively, to the reduction at

time-consuming and to the loss of accuracy.

discuss ways to estimate the correct value of this constant (Walker, 2001; Carrero et al., 2007).

Usually, 1 ≤ 𝑎 ≤ 1.5 and the choice depends on other parameters of the I–V model. As is given

in Carrero et al. (2007), there are different opinions about the best way to choose 𝑎. According

to Lineykin et al. (2014), the correct value of 𝑎 corresponds to the modelled curve with

Chapter 4. Results and discussions 63

minimum deviation from the datasheet/measured I–V curve at STC. Peng et al. (2014) assumed

𝑎 is unit; Ishaque et al. (2011) also used 𝑎1 = 1 and 𝑎2 ≥ 1.2. In this work, it is defended that

the 𝑎 value has also to be related to the PV technology, and not just a parameter to reduce the

deviation of the curve from the datasheet/measured data. Once it is constant and related to the

PV technology, the time-consuming is reduced and the model presents physical meaning.

4.3.5. Overview

Regarding evaluation on different temperatures, the model proposed by Isaque et al.

(2011) was verified that it has the greatest temperature range, however, the models of Chaibi et

al. (2018), Lineykin et al. (2014), Orioli and Di Gangi (2013), Boutana et al. (2017) and

Mahmoud and Xiao (2013) are considered well-evaluated since they were validated in the

temperature range (from 50 to 75 °C) that is usual in operation conditions for most of the solar

cells.

Ishaque et al. (2011), despite proposing a two-diode modeling method for PV cell,

achieved a method that requires the computation of only four parameters and computes the

series and shunt resistances using a simple and rapid iterative approach. It presented excellent

precision at lower irradiance conditions and it was superior when exposed to temperature and

irradiance variations using datasheet information from the manufacturers of six PV modules of

different technologies.

expression where only three parameters are required. This model presented the relatively

smallest errors among other four models. Besides, for experimental data, this model achieved

results very close to the ones extracted by a powerful optimization tool (FODPSO algorithm).

A correlation coefficient value greater than 0.996 for the four PV module technologies was

achieved by the model. Furthermore, it can give a good estimation of the maximum power

point.

Cibira and Koscová (2014) do not compare themselves against any other work. The

error analysis mentioned in the work just comments that differences occur at bending and tail

areas (highest difference point at V = 1.6 V); but, they do not exceed 2.3% of measured level.

The analysis could be showed at all curve. In the graphics with varying temperature, it was

observed an anomalous behavior that deserves more attention, since it is not in accordance with

general literature.

Chapter 4. Results and discussions 64

The model presented in Chaibi et al. (2018) shows good agreement between the

proposed method and datasheet, except for irradiance lower than 400 W/m². The curve presents

a great inclination, and so, a remarkable discrepancy can be observed at vicinities of 𝑉𝑜𝑐 , it was

also observed at Saloux et al. (2011) and Das (2011), different of Villalva and Gazoli (2009)

and Mermoud (2012). The present work made comparisons among his model, well established

models and datasheet separately. It is recommended a simultaneous comparison to confirm its

accuracy since it is easier to visualize. This model usually underestimate the MPP, thus, it is

not the best choice when the aim is the prediction of maximum energy production to study the

feasibility of a power plant. It is ideal to conservative predictions.

Jaimes and Sousa (2017) developed a new model which is very effective in harvesting

applications because it forecasts the power constrain introduced by the indoor PV solar cells

apparatus to the payload. In addition, this model forecasts precisely the operation of the indoor

PV solar cells under warm LED spectrum – with illuminance from 177 lm/m2 to 33.3×103

(0.67–107 W/m2) – at ambient temperature. However, it is not suitable to be compared to the

others models presented here due to its radiation source and application field.

Mares et al. (2015) presented a five-parameter model which was tested on six different

commercially available crystalline silicon PV modules. The values of the determination

coefficient r² are in the range 0.976–0.998 demonstrating a very good agreement between the

experimental I–V characteristic and the estimated I–V characteristics. However, there are minor

differences in two curves. These discrepancies can be attributed to the PV technology; these

two modules are mc-Si while the other four are pc-Si, which contributed to the point already

discussed: it should be a criterion during the extraction of parameters that differentiate the PV

technologies, i.e., the 𝑎 value. Lineykin et al. (2014) also discussed how the value of the s

affects the curve. Despite this work affirms that its accuracy and the one reported by Lo Brano

et al. (2010) is comparable, it was not proved by error analysis. However, for running the

algorithm of Mares et al. (2015), the graphical presentation of the I–V characteristic in the PV

module datasheet is not required, which is a major advantage.

Lineykin et al. (2014) used the minimization of the divergence between the modeled

and experimental I–V curves of several off the shelf panels of leading manufacturers to obtain

the fifth parameter. The proposed method indicates feasibility and high accuracy through the

results of an average deviation of 0.1 – 0.5%.

Chapter 4. Results and discussions 65

The general ranking is presented in Table 4.5. Contrasting to the others rankings, this

final one is divided in 3 groups. This classification manner was chosen since it was not feasible

to precisely weight each aspect because of lacking quantitative analysis. However, this division

is enough and reliable to elect the most appropriate equivalent models for PV cells (group 1):

Ishaque et al. (2011), Orioli and Di Gangi (2013), Mahmoud and Xiao (2013) and Boutana et

al. (2017).

Among these models, it is selected the one developed by Ishaque et al. (2011) as the

most robust mathematical equivalent electric circuit models that emulate the solar cell

parameters in the respect of “Operation conditions” and “Error analysis”.

Literature Error analysis Operation PV Translation Final

conditions technology equations

66

5. Conclusions

This work aiming to study the influence of temperature on polycrystalline silicon solar

cell technology was divided in three fronts: IxV curve, simulator spectral distribution and

equivalent circuit model.

IxV curve

The I-V curve for a polycrystalline silicon solar cell was carefully measured to observe

how the solar cells parameters depend on temperature. This work permitted to analyze the

development of a simple system capable of extract all desired parameters using only resistors,

an ultra-thermostatic bath, a low-cost solar simulator, a voltmeter and an amperemeter.

This work also allowed to confirm some previously reported results, and provided new

ones. It was observed that both 𝑉𝑜𝑐 and 𝐽𝑠𝑐 are sensitive to temperature, almost inversely and

directly proportional, respectively. With increasing temperature, reverse saturation current

increases, and therefore, 𝑉𝑜𝑐 decreases which decreases the fill factor and hence the efficiency

of the solar cell. At the same time, the bandgap also decreases with increasing temperature and

this results in an increase in 𝐽𝑠𝑐 which acts to improve the efficiency of the cell. Therefore, the

tendency of 𝑉𝑜𝑐 to decrease and 𝐽𝑠𝑐 to increase with increasing temperature in the solar cells

results in a decrease in the efficiency with increasing temperature. It was observed a

considerable loss of power by the solar cell indicated by the shape of the IxV curve.

The method used to perform the analysis of SOLSIM spectral distribution was found

to be very robust and of easy implementation, providing reliable results for all single-diode

parameters with low difference to literature. In the SOLSIM analysis, the temperature

dependence of performance parameters, 𝑉𝑜𝑐 , 𝐽𝑠𝑐 , and 𝜂 of solar cells based on Si, GaAs, InP,

CdTe and CdS has been investigated in the temperature range 298–528 K.

The maximum achievable 𝑉𝑜𝑐 , 𝐽𝑠𝑐 , and 𝜂 of solar cells are nearly the same as in the

literature, the differences are explained due to the existent difference in spectral distribution

considered in the works analyzed. Furthermore, using such method was possible to conclude

that the spectral distribution of the simulator used in a research is of fundamental significance

since it will influence the maximum achievable value for a determined parameter. For

Chapter 5. Conclusions 67

researchers, relative parameters applying the AM1.5G and the used simulator spectra should be

employed, since the results required to be given in a much universal comparable way.

Besides, in this work, the existing research works on PV cell model parameter

estimation problem are classified per number of parameters, parameters’ extraction, translation

equations and PV technology. The existent models were discussed pointing out its different

levels of approximation. According to the qualitatively comparison in terms of PV technology,

operation conditions, translation equations, and assumptions and time-consuming level

performed in this work, four models were classified as the most appropriate to be used to

emulate the solar cell behaviour: Boutana et al. (2017), Mahmoud and Xiao (2013), Orioli and

Di Gangi (2013) and Ishaque (2011). Among these models, it is selected the one developed by

Ishaque et al. (2011) as the most robust mathematical equivalent electric circuit models that

emulate the solar cell parameters in the respect of “Operation conditions” and “Error analysis”.

The error analysis is considered the most important, however, it was observed that is

common it not be complete and adequate. It is suggested that this investigation should be like

that performed by Boutana et al. (2017) and Mahmoud and Xiao (2013) that presented

graphically the error at all curve I-V, or at least, as was done by Ishaque et al. (2011) that

computed the error to each different condition. Since it was observed that the most recent is not

the best one necessarily, it is indispensable the comparison to other existent models, mainly the

ones appointed here as Group 1, and under the same conditions that the first one was developed.

There are few different operation conditions and there are also a few PV technologies

considered in almost all PV models which limits its accuracy. Some PV modules can be used

to overcome these limitations, such as: KC 200GT, S36, SW255, SQ150-PC, SP-70, SM 55

and ST40. It is also advised the use of experimental data to guarantee the effectiveness of the

model when both temperature and radiation are acting at different ranges simultaneously. The

implementation of more data, improve the capacity of energy production’s prediction which is

more adequate than using direct expressions that relate just the efficiency of the PV cells to

ambient conditions since it is not possible to describe the entire I-V curve, fundamental to

design the power plant.

Chapter 5. Conclusions 68

already considered outdated. Parameters such as ideality factor presents lack of clearness, it is

necessary detailed study to related it to PV technologies. The direct relation between ideality

factor and the PV technology deserve special attention in future research. The findings of this

brief review indicate that although a great deal of research effort has been put for solving

parameter estimation problem in PV cells, there is still room for improvement.

Final considerations

From the results of this work and the literature review, the temperature then influences

mainly the material of the solar cell. In the literature, this dependence has been represented by

three distinct equations. Consequently, temperature influences the photogenerated current as

well as the potential difference generated among the cell poles. These last two dependencies

are represented in the literature by one and two relations, respectively, having, in the case of

voltage, a more precise equation.

Some gaps have been observed in the literature under the theme “influence of

temperature on polycrystalline silicon solar cell technologies” and, therefore, they can be

studied in future work. They are:

i) Analyze whether the temperature dependence of the bandgap energy can be used

for efficiency gain at any operation condition;

ii) Develop a method that compares qualitative and quantitatively the existing

equivalent circuit models of solar cells. The methods of Boutana et al. (2017) and/or

Mahmoud and Xiao (2013) and/or Ishaque et al. (2011) can be used as references. It is

recommended the using of new experimental data and those already well established in

the literature (acquired from KC 200GT, S36, SW255, SQ150-PC, SP-70, SM 55 and

ST40 module datasheets);

iii) Develop a method in order to provide a more reliable comparison between data

extracted from different solar simulators without suffer significant influence of

temperature and its spectral distribution;

iv) Analyze whether the ideality factor can be related to solar cell technology as

already suggested by Cuce et al. (2017), and also determine its temperature dependence;

Chapter 5. Conclusions 69

and disadvantages. Especially regarding physical simplifications and computational

gains.

70

References

Abd-Elhady MS, Fouad MM, Khalil T. Improving the efficiency of photovoltaic (PV)

panels by oil coating. Energy Convers Manag 2016; 115:1–7.

https://doi.org/10.1016/j.enconman.2016.02.040

Experimental investigation and modeling of the thermal behavior of a solar PV module. Sol

Energy Mater Sol Cells 2018; 180:271–9. https://doi.org/10.1016/j.solmat.2017.06.052

crystalline silicon, J. Appl. Phys. 79 (1996) 6943-6946, http:// dx.doi.org/10.1063/1.362447.

American Society for Testing and Materials 10(2015). ASTM E927. Standard

Specification for Solar Simulation for Photovoltaic Testing.

El-Sayed. Enhancement of Photovoltaic System Performance via Passive Cooling: Theory

versus Experiment, Renewable Energy, http://dx.doi.org/10.1016/j.renene.2019.03.048

semiconductors from solar cells electrical parameters. Current Applied Physics. 2017; 17: 55-

59.

economic assessment of different photovoltaic technologies based on experimental

measurements. Renew Energy 2016; 85:1–11. https://doi.org/10.1016/j.renene.2015.06.017

Appl. Phys. 45 (1974) 1846-1848, http://dx.doi.org/10.1063/ 1.1663501

Boutana N, Mellit A, Haddada S, Rabhi A, Massi Pavan A. An explicit I-V model for

photovoltaic module technologies. Energy Convers Manage 2017; 138:400-12.

https://doi.org/10.1016/j.enconman.2017.02.016

fotovoltaicos – Simulador solar – Requisitos de desempenho.

References 71

silicon PV modules and evaluate the loss resistances. Renew Energy 2007; 32:2579–89.

https://doi.org/10.1016/j.renene.2007.01.001

circuit parameters of a photovoltaic panel. Sol. Energy 2018; 163, 376–86.

https://doi.org/10.1016/j.solener.2018.02.017

Chander S, Purohit A, Sharma A, Nehra SP, Dhaka MS, 2015a. Impact of temperature

on performance of series and parallel connected mono-crystalline silicon solar cells. Energy

Reports 1, 175-180

Chander, S., Purohit, A., Sharma, A., Arvind, Nehra, S.P., Dhaka, M.S., 2015b. A

study on the photovoltaic parameters of mono-crystalline Silicon solar cell with cell

temperature. Energy Reports 1, 104–109. https://doi.org/10.1016/j.egyr.2015.03.004

Surface Science journal 2014; 312:74-80. https://doi.org/10.1016/j.apsusc.2014.05.080

models for the representation of I–V characteristic of a PV cell. Renew Sustain Energy Rev

2014; 32:684–96. https://doi.org/10.1016/j.rser.2014.01.027

Collings PJ, Simple measurement of the bandgap in silicon and germanium, Am. J.

Phys. 48 (1980) 197-199, http://dx.doi.org/10.1119/1.12172

Cuce E, Cuce PM, Karakas IH, Bali T. An accurate model for photovoltaic (PV)

modules to determine electrical characteristics and thermodynamic performance parameters.

Energy Convers Manage 2017, 146:205-16. https://doi.org/10.1016/j.enconman.2017.05.022

Cuce E, Cuce PM, Wood CJ, Riffat SB. Toward aerogel based thermal superinsulation

in buildings: a comprehensive review. Renew Sustain Energy Rev 2014; 34:273–99.

https://doi.org/10.1016/j.rser.2014.03.017

Cuce E, Young CH, Riffat SB. Thermal insulation, power generation, lighting and

energy saving performance of heat insulation solar glass as a curtain wall application in Taiwan:

a comparative experimental study. Energy Convers Manage 2015; 96:31–8.

https://doi.org/10.1016/j.enconman.2015.02.062

References 72

power point tracking methods for PV cells. Renewable and Sustainable Energy Reviews 2018;

82:2743–67. https://doi.org/10.1016/j.rser.2017.10.009

Das AK. An explicit J–V model of a solar cell for simple fill factor calculation. Solar

Energy 2011; 85:1906–9. https://doi.org/10.1016/j.solener.2011.04.030

Devabhaktuni V, Alam M, Depuru SSSR, Green RC, Nims D, Near C. Solar energy:

trends and enabling technologies. Renew Sustain Energy Rev 2013; 19:555–64.

https://doi.org/10.1016/j.rser.2012.11.024

DGS, Planning and installing photovoltiac systems: a guide for installers, architects,

and engineers, Deutsche Gesellschaft fur Sonnenenergie (DGS), 5th ed., Landesverband Berlin

Brandenburg e.V, Berlin. 〈WWW.epices-energie.fr〉.

https://www.eia.gov/todayinenergy/detail.php?id=32912; 2017. (accessed nov 29, 2018)

Fan JCC, Tsaur BY, Palm BJ, Optimal design of high efficiency tandem cells, in

Conference Record 16th IEEE Photovoltaic Specialists Conference, 1982, pp. 692–701.

Fouad MM, Shihatab LA, Morgana EI. An integrated review of factors influencing the

performance of photovoltaic panels. Renew Energy 2017; 10:1499–511.

https://doi.org/10.1016/j.rser.2017.05.141

Green MA, Keith Emery, Yoshihiro Hishikawa, Wilhelm Warta, Solar cell efficiency

tables (version 37), Progress in Photovoltaics: Research and Applications 19 (2011) 84–92.

Green MA, Solar Cells, Prentice-Hall, Englewood Cliffs, NJ, 1982, p. 88.

2013. (accessed march 19, 2018).

http://www.greentechmedia.com/articles; 2015. (accessed march 13, 2018)

Sandia National Laboratories, Albuquerque, NM, Forthcoming; 2015

https://doi.org/10.13140/rg.2.1.4336.7842

References 73

Hu C, White RM, Solar Cells, McGraw-Hill, New York, 1983, pp. 21.

Photovoltaic Devices – Part 5: Determination of the Equivalent Cell Temperature (ECT) of

Photovoltaic (PV) Devices by the Open-Circuit Voltage Method, pp. 1–11.

– Part 9: Solar Simulator Performance Requirements, Photovoltaic Devices.

Terrestrial Photovoltaic (PV) modules - Design qualification and type approval.

Ishaque K, Salam Z, Taheri H. Simple, fast and accurate two-diode model for

photovoltaic modules. Sol Energy Mater Sol Cells 2011, 95:586-94.

https://doi.org/10.1016/j.solmat.2010.09.023

Jaimes AF, Sousa FR. Simple modeling of photovoltaic solar cells for indoor

harvesting applications. Sol. Energy 2017; 157: 792–802.

https://doi.org/10.1016/j.solener.2017.08.077

Jian Jin, Yong Hao, Hongguang Jin, A universal solar simulator for focused and quasi-

collimated beams, Applied Energy 2019; 235: 1266-1276.

https://doi.org/10.1016/j.apenergy.2018.09.223.

Jordehi AR. Parameter estimation of solar photovoltaic (PV) cells: A review. Renew

Sustain Energy Rev 2016; 61: 354–71. https://doi.org/10.1016/j.rser.2016.03.049

Joya Zeitouny, Noémie Lalau, Jeffrey M. Gordon, Eugene A. Katz, Gilles Flamant ,

Alain Dollet , Alexis Vossier. Assessing high-temperature photovoltaic performance for solar

hybrid power plants. Solar Energy Materials and Solar Cells 2018; 182:61-6.

https://doi.org/10.1016/j.solmat.2018.03.004.

Khan F, Baek SH, Kim JH. Wide range temperature dependence of analytical

photovoltaic cell parameters for silicon solar cells under high illumination conditions. Appl

Energy 2016; 183:715-24. https://doi.org/10.1016/j.apenergy.2016.09.020

References 74

diode equivalent circuit parameters of a photovoltaic cell/panel. Renew Sustain Energy Rev

2014; 30: 282–289. https://doi.org/10.1016/j.rser.2013.10.015

photovoltaic module. Sol Energy Mater Sol Cells 2010; 94:1358–70.

https://doi.org/10.1016/j.solmat.2010.04.003

Louwen A, van Sark W, Schropp R, Faaij A. A cost roadmap for silicon heterojunction

solar cells. Sol Energy Mater Sol Cells 2016; 147:295–314.

https://doi.org/10.1016/j.solmat.2015.12.026

Springer, Berlin, 1991.

accuracy. Trans Ind Electron IEEE 2013; 60:5708–16.

Mares, O., Paulescu, M., Badescu, V., 2015. A simple but accurate procedure for

solving the five-parameter model. Energy Convers. Manage. 105, 139–48.

https://doi.org/10.1016/j.enconman.2015.07.046

Mermoud A. Pvsyst: Software for the Study and Simulation of Photovoltaic Systems

2012. ISE, University of Geneva. <www.pvsyst.com>.

Nell ME, Barnett AM, The spectral p–n junction model for tandem solar-cell design,

IEEE Transactions on Electron Devices 24 (1987) 257–266

Nelson J. The physics of solar cells. 6th ed. Singapore: Imperial College Press; 2008.

Phys. Lett. 58 (1991) 2924-2926, http://dx.doi.org/10.1063/ 1.104723.

silicon photovoltaic modules on the basis of the tabular performance data. Appl Energy 2013;

102:1160–77. https://doi.org/10.1016/j.apenergy.2012.06.036

References 75

photovoltaic cells using only three state points at standard test condition. J Power Sour 2014;

248:621–31. https://doi.org/10.1016/j.jpowsour.2013.07.058

Radziemska E. Thermal performance of Si and GaAs based solar cells and modules: a

review. Prog. Energy Combust. Sci. 2003; 29(5):407–24. https://doi.org/10.1016/S0360-

1285(03)00032-7

Ram JP, Babu TS, Rajasekar N. A comprehensive review on solar PV maximum power

point tracking techniques. Renew Sustain Energy Rev 2017; 67: 826–47.

https://doi.org/10.1016/j.rser.2016.09.076

conversion: An overview. Renew Sustain Energy Rev 2017; 71:630–8.

https://doi.org/10.1016/j.rser.2016.12.089

voltage and currents at the maximum power point. Sol. Energy 2011; 85:713–22.

https://doi.org/10.1016/j.solener.2010.12.022

Renew Sustain Energy Rev 2017; 74:590-601. https://doi.org/10.1016/j.rser.2017.02.081

performance of silicon wafer-based photovoltaic modules: Failure detection methods and

essential mitigation techniques. Renewable and Sustainable Energy Reviews 2019; 110:83-110.

https://doi.org/10.1016/j.rser.2019.04.024

Wind effect on PV module temperature: analysis of different techniques for an accurate

estimation. Energy Procedia 2013; 40:77–86. http://dx.doi.org/10.1016/j.egypro.2013.08.010

SERI. Solar Energy Research Institute. Basic photovoltaic principles and methods. SP-

290-1448. Washington, DC 20402. Published February 1982.

Singh, P., Ravindra, N.M., 2012. Temperature dependence of solar cell performance -

an analysis. Sol. Energy Mater. Sol. Cells 101, 36–45.

https://doi.org/10.1016/j.solmat.2012.02.019

References 76

thermal-electrical model of three photovoltaic/thermal (PV/T) hybrid air collectors and

photovoltaic (PV) module: comparative study under Algiers climatic conditions. Energy

Convers Manage 2017; 133:458–76. https://doi.org/10.1016/j.enconman.2016.10.066

extraction of two diode solar PV model using Fireworks algorithm. Sol. Energy 2016; 140:265–

76. https://doi.org/10.1016/j.solener.2016.10.044

Thurmond CD, The standard thermodynamic functions for the formation of electrons

and holes in Ge, Si, GaAs, and GaP, J. Electrochem. Soc. 122 (1975) 1133-1141,

http://dx.doi.org/10.1149/1.2134410.

real options analysis of solar energy. Energy Policy 2016; 92:255–69.

https://doi.org/10.1016/j.enpol.2015.07.010

http://home.cc.umanitoba.ca/?vsmil/pdf_ pubs/oecd.pdf; 2017 (accessed 28 march 2018).

34 (1967) 149-154, http://dx.doi.org/10.1016/0031-8914(67)90062- 6.

Veissid N, Vaz CC. Simulador solar de baixo custo (SOLSIM): simulação da corrente

fotogerada em diferentes células solares. Revista Brasileira de Energia Solar 2011; 2: 71-7.

photovoltaic arrays. Power Electron IEEE Trans 2009; 24:1198–208.

Elect. Electron. Eng. Aust. 2001; 21(1):49–55. Availability:

https://search.informit.com.au/documentSummary;dn=537020271845747;res=IELENG.

ISSN: 0725-2986.

References 77

Wen C, Fu C, Tang JL, Liu DX, Hu SF., Xing ZG, 2012. The influence of environment

temperatures on single crystalline and polycrystalline silicon solar cell performance. Sci.

China—Phys. Mech. Astron. 55, 235–241.

conversion, Journal of Applied Physics 31 (1960) 571–578.

Xiao W, Dunford WG, and Capel A. A novel modeling method for photovoltaic cells.

In: IEEE 35th Annu. Power Electron. Spec. Conf. (PESC), 2004; 3:1950–6.

https://doi.org/10.1109/pesc.2004.1355416.

Yeh N, Yeh P. Organic solar cells: their developments and potentials. Renew Sustain

Energy Rev 2013; 21:421–31. https://doi.org/10.1016/j.rser.2012.12.046.

Yin WW, Wang XS, Zhang F, Zhang LJ. 19.6% Cast Mono-MWT Solar Cells and 268

W Modules. IEEE J Photovolt 2013; 3:697–701. https://doi.org/10.1109/jphotov.2013.2239357

78

Appendices

Appendices A: A brief review of equivalent models

Appendices 79

Ideal model

If the solar irradiation does not exist, the cell acts as a simple p-n junction diode. Its

attributes are governed by the well-known Shockley diode equation, which expresses diode

current 𝐼𝑑 as (Chin et al., 2015):

𝑞𝑉

𝐼𝑑 = 𝐼𝑠 [exp ( ) − 1] (Ap.0)

𝑎𝑘𝑇

where, 𝐼𝑠 represents the saturation current and 𝑎 is the ideality (or quality) factor of the diode.

Constant 𝑘 is the Boltzmann’s constant, 𝑞 is the absolute value of electron’s charge

(1.60217646 x 10−19 C), 𝑉 is the cell output voltage, while 𝑇 is the temperature of the junction

(K). This temperature is generally assumed to be close enough to the temperature of the cell

itself (Villalva and Gazoli, 2009). However, it can be accurately determined by transfer heat as

done by Akhsassi et al. (2018). The ratio 𝑘𝑇/𝑞 is known as the thermal voltage (𝑉𝑡 ).

illuminated PV cell (Chin et al., 2015). The resultant circuit is referred to as ideal PV cell model

and is illustrated in Figure Ap.1. So, 𝐼 is the superposition of 𝐼𝑝𝑣 and 𝐼𝐷 , thus, 𝐼𝐷 determines its

shape while 𝐼𝑝𝑣 defines the translation on the ordinate axis of the curve.

It is evident that the mentioned ideal PV cell model has three parameters: 𝐼𝑝𝑣 , 𝑎 and 𝐼𝑠 . Its

I–V curve characteristic is given by:

𝑞𝑉

𝐼 = 𝐼𝑝𝑣 − 𝐼𝑠 [exp ( ) − 1] (Ap.1)

𝑎𝑘𝑇

Appendices 80

According to Khan et al. (2013), the 𝐼𝑠 value is also indicative of the recombination in

the bulk of semiconductor materials, while 𝑎 indicates the recombination at the surfaces of the

solar cells and also in the bulk space charge regions. Accordingly, the value of ideality factor

notably depends on the PV technology (Cuce et al., 2017).

Bätzner et al. (2001) support that 𝑎 value depends on the current transport mechanism.

A unit value indicates ideal charge transport through the p-n junction, while a value of two

corresponds to the superposition of recombination mechanisms and diffusion. When multi-

recombination or multi-tunnelling steps occur, values higher than two can be obtained.

1D1R model

The ideal PV cell model is not commonly used for simulation of PV cells, but it is only

used to explain fundamental concepts of PV cells since it cannot emulate the behaviour of

physical PV cells (Lim et al., 2015). Therefore, to improve this emulation (Xiao et al., 2004),

the 1D1R model, which is also known as single diode 𝑅𝑠 model, presents a new element, the

series resistance R s as can be observed in Figure Ap.2.

According to literature, the presence of the series resistance 𝑅𝑠 represents the sum of

several structural resistances of the device (Boutana et al., 2017), including the parasitic series

resistance (Mazhari, 2006) and dissipative effects (Cuce et al., 2017), the contact resistance of

the metal base with the p semiconductor layer, the resistances of the p and n bodies, the contact

resistance of the n layer with the top metal grid (Ciulla et al., 2014; Cibira and Koscová, 2014),

and the resistance of the grid (Khan et al., 2013), and finally, the resistance of the materials

which compose the module and causes a reduction on the power converted by this device

(Ruschel et al., 2016).

Appendices 81

This model has four parameters: 𝐼𝑝𝑣 , 𝑎, 𝐼𝑠 and 𝑅𝑠 , and its I–V curve characteristic is

given by:

𝑞(𝑉 + 𝑅𝑠 𝐼)

𝐼 = 𝐼𝑝𝑣 − 𝐼𝑠 [exp ( ) − 1] (Ap.2)

𝑎𝑘𝑇

1D2R model

Although 1D1R model imitates the behaviour of physical PV cells better than ideal

PV cell model, it can also lack accuracy, especially in the situations where the PV cell presents

many defects and/or important temperature variation (Mares et al., 2015). Therefore, to take

into account this phenomenon and improve the similarity between model and real PV cell, the

1D2R model, also known as single diode 𝑅𝑠ℎ model, presents a new element (Figure Ap.3): the

shunt resistance 𝑅𝑠ℎ . Despite the improved performance, the accuracy deteriorates at low

irradiances, especially in the proximities of the open-circuit voltage Voc (Salam et al., 2010).

According to the existing literature, the presence of this shunt resistance represents the

construction defects (Orioli and Gangi, 2013) which cause leakage currents within the PV cell

(Jordehi, 2016; Boutana et al., 2017), i.e., any parallel high-conductivity paths (shunts) for free

carriers produced by the solar irradiation across the PV cell p-n junction or on the PV cell edges

(Mares et al., 2015). A high shunt resistance means that the clear majority of these carriers

generate power, whereas a low resistance indicates large losses (Ruschel et al., 2016). The

magnitude of the shunt resistance varies with different fabrication methods since it is intimately

related to the construction defects. This model has five parameters: 𝐼𝑝𝑣 , 𝑎, 𝐼𝑠 , 𝑅𝑠 , and 𝑅𝑠ℎ , and

its I–V curve characteristic is given by:

𝑞(𝑉 + 𝑅𝑠 𝐼) 𝑉 + 𝑅𝑠 𝐼

𝐼 = 𝐼𝑝𝑣 − 𝐼𝑠 [exp ( ) − 1] − (Ap.3)

𝑎𝑘𝑇 𝑅𝑠ℎ

Appendices 82

Comparing Equations (Ap.1) and (Ap.3), the series resistance affects the output

voltage while the shunt resistance reduces the available electrical current.

According to Bai et al. (2014), the shunt resistance is the key parameter to analyse

more complex situations of PV cells, PV modules or arrays, such as mismatch and hot-spot

phenomena. The five-parameter model has been confirmed to be more accurate than the four-

parameter model (Lo Brano et al., 2012). It is arguably the most popular used PV cell model

thanks to its relatively appropriate trade-off between accuracy and simplicity (Jordehi, 2016).

2D2R model

Despite the improved performance of the 1D2R model, its accuracy deteriorates at low

irradiances since the single diode models assumed that the recombination loss in the depletion

region is absent. In a real solar cell, the recombination represents a substantial loss, which

cannot be adequately modelled using a single diode. Consideration of this loss leads to a more

precise model, especially in the proximities of 𝑉𝑜𝑐 . Therefore, according to Gupta et al. (2012),

to overcome this limitation and to take into account the recombination phenomenon, the 2D2R

model, which is also known as two-diode model, presents a new element, the second diode 𝐷2

as shown in Figure Ap.4.

According to the existing literature, the presence of this second diode represents a

global effect due to the presence of a plurality of adjacent and uniformly distributed elementary

diodes along the surface that separates the two layers of the semiconductor (Wolf and

Rauschenbach, 1963), i.e., the effect of recombination current loss in the depletion region.

This model has seven parameters: 𝐼𝑝𝑣 , 𝑎1 , 𝑎2 , 𝐼𝑠1 , 𝐼𝑠2 , 𝑅𝑠 , and 𝑅𝑠ℎ , and its I–V curve

characteristic is given by:

Appendices 83

𝑞(𝑉 + 𝑅𝑠 𝐼) 𝑞(𝑉 + 𝑅𝑠 𝐼) 𝑉 + 𝑅𝑠 𝐼

𝐼 = 𝐼𝑝𝑣 − 𝐼𝑠1 [exp ( ) − 1] − 𝐼𝑠2 [exp ( ) − 1] − (Ap.4)

𝑎1 𝑘𝑇 𝑎2 𝑘𝑇 𝑅𝑠ℎ

In the scientific literature, there are few entirely elucidated models that allow

implementing of the algorithm to determine the seven parameters. The transcendental equation

and the existence of two exponential terms make the determination of the seven parameters an

arduous task. The procedures demand to be correctly lead during the primary estimation of the

parameters to avoid contradictory outcomes; some researchers admit: the primary conditions

heavily affect the resolution (Ciulla et al., 2014).

Bail et al. (2003) affirms that 1D1R and 1D2R model adequately emulate the operating

of the solar cells in virtue of insignificant recombination in the space charge region since they

are under normal illumination conditions.

electron diffusion coefficient, intrinsic carrier density and other semiconductor parameters is a

substitute approach to characterize the 2D2R model. These models are useful to comprehend

the behaviour of the cell, however, knowledge about these parameters is not always accessible

in commercial PV letter.

1D2R1C model

This model is similar to the single diode model but with the addition of a capacitance.

It is depicted in Figure Ap.5.

Suskis and Galkin (2013) verified that the voltage on the terminals of the PV panel

remains at almost the same value after the instantaneous load connection for 188 microseconds

(∆𝑡). According to them, this can be explained by p-n junction capacitance that is maintaining

the bias at almost the same level, and that after discharge of this characteristics capacitance, the

Appendices 84

bias drops to the load steady-state value. Thus, a model with included p-n junction capacitance

can provide more precise and closer to the reality simulations of transient processes.

2D4R model

with each other. It can be affirmed that an option that enables this comparison is the separation

of the diffusion current and recombination current.

structure’’ (Figure Ap.6), where DCA is a ‘‘diffusion-current dominant area’’, which has an

ideal diode of ideality factor equals to unit value, and RCA is a ‘‘recombination-current

dominant area’’, which has an ideal diode of ideality factor of two. The analysis introduced by

Kurobe and Matsunami (2005) can precisely separate the two current components by using the

parameters described in their work, one of these parameters can be associated as a power factor

and other as a loss factor, so it is possible that solar cells can be compared with each other, and

the way to improve solar cell performance may be easily found.

3D2R model

Mazhari (2006) develops a new simplified model, which is depicted in Figure Ap.7,

based on the assumption that: a) electron-hole pairs generation rate is steady for a given

radiation intensity; and b) the energy flux relies upon competition between electron-hole

recombination and its collection by the finger-like metal elements.

Appendices 85

It is known that the magnetic field contributes to the current in the solar cells and the

electric field is influenced by the voltage, thus, it is expected that current would, in general,

depends on the voltage across the solar cell indirectly since the charge extraction efficiency

depends on internal electric field.

The difference of this model to previous ones is that the internal series and shunt

resistances (it is used the index in lowscript ‘int’) come into effect only under the presence of

light. Furthermore, this model considers the shunt resistance as constant and series resistance

as a variable depending on the magnitude of current across it, although both internal resistances

can, in general, be functions of externally applied voltage, and non-linear in nature (Mazhari,

2006).

3D5R model

It was perceived by Nishioka et al. (2007) that when the modelling was leading with

small size solar cells, it was difficult to perform precise fitting, because the leakage current

through peripheries considerably affects the I–V characteristics of solar cells. With the

objective of solve this limitation, the model depicted in was proposed.

The model in Figure Ap.8 overcomes the limitation of the cells size. Besides, it enables

to model small size cells, which is very important, mainly, when it comes to developing cells.

Appendices 86

xD2R model

Soon et al. (2014) demonstrates that different PV models is required to model different

PV cell technologies to achieve low modelling error. In the present work, the physics meaning

of each diode is not explained, but the quantity varies according to the necessity of better data’s

fitting. The model is depicted in Figure Ap.9.

Appendices 87

2D1R1Rv model

dependence and most of the classical analysis does not take into account (Pallarès et al., 2006).

In literature, it has been reported in CIGS solar cells, GaAs concentrator solar cells and organic

solar cells (Mazhari, 2006). According to Pallarès et al. (2006), among all the possible

conduction mechanisms with a nonlinear current-voltage dependence, the space-charge limited

current (SCLC) mechanism (Rose, 1955) has been reported not only in organic solar cells, but

in different cases (Goldenblum et al., 2005). Thus, Pallarès et al. (2006) proposed an 2D2R

adapted equivalent circuit model whose has the addition of an SCLC mechanism to emulate the

nonlinear current-voltage dependence. This circuit is illustrated in Figure Ap.10.

References

Bai J, Liu S, Hao Y, Zhang Z, Jiang M, Zhang Y. Development of a new com- pound method to extract

the five parameters of PV modules. Energy Convers Manag 2014; 79:294–303.

https://doi.org/10.1016/j.enconman.2013.12.041

lifetime Si wafers. Appl Phys Lett 2003; 82:757–9. https://doi.org/10.1063/1.1541115

Bätzner DL, Romeo A, Zogg H, Tiwari AN. CdTe/CdS solar cell performance under low irradiance. In:

17th EC PV Solar Energy Conference, Munich, Germany, October 22–26, 200: VB 1.40.

Chin VJ, Salam Z, Ishaque K. Cell modelling and model parameters estimation techniques for

photovoltaic simulator application: a review. Appl Energy 2015; 154:500–19.

https://doi.org/10.1016/j.apenergy.2015.05.035

involving carrier injection into impurity bands of high-k insulators. Appl. Phys. Lett. 2005; 86:203506.

https://doi.org/10.1063/1.1935045

Gupta S, Tiwari H, Fozdar M, Chandna V. Development of a two-diode model for photovoltaic modules

suitable for use in simulation studies. In: proceedings of Asia-Pacific power and energy engineering conference

(APPEEC 2012) IEEE; 2012: 1–4. https://doi.org/10.1109/appeec.2012.6307201

Appendices 88

Khan F, Baek SH, Park Y, Kim JH. Extraction of diode parameters of silicon solar cells under high

illumination conditions. Energy Converse Manage 2013; 76:421–9.

https://doi.org/10.1016/j.enconman.2013.07.054

Kurobe K, Matsunami H. New two-diode model for detailed analysis of polycrystalline silicon solar

cells. Jpn J Appl Phys 2005; 44:8314. https://doi.org/10.1143/JJAP.44.8314

Lim LHI, Ye Z, Ye J, Yang D, Du H. A linear method to extract diode model parameters of solar panels

from a single I–V curve. Renew Energy 2015; 76:135–42. https://doi.org/10.1016/j.renene.2014.11.018

for silicon photovoltaic modules. Sol Energy Mat Sol C 2012; 105:27–39.

https://doi.org/10.1016/j.solmat.2012.05.028

Mazhari B. An improved solar cell circuit model for organic solar cells. Sol Energy Mater Sol Cells

2006; 90:1021–33. https://doi.org/10.1016/j.solmat.2005.05.017

Nishioka K, Sakitani N, Uraoka Y, Fuyuki T. Analysis of polycrystalline silicon solar cells by modified

3-diode equivalent circuit model taking leakage current through periphery into consideration. Sol Energy Mater

Sol Cells 2007; 91:1222–7. https://doi.org/10.1016/j.solmat.2007.04.009

Pallarès J, Cabré R, and Marsal LF. A compact equivalent circuit for the dark current-voltage

characteristics of non-ideal solar cells. J Appl Phys 2006; 100:84513. https://doi.org/10.1063/1.2357641

https://doi.org/10.1103/PhysRev.97.1538

dependence on solar irradiance, Sol Energy 2016. 133:35-43. https://doi.org/10.1016/j.solener.2016.03.047

Salam Z, Ishaque K, Taheri H. An improved two-diode photovoltaic (PV) model for PV system. In:

Joint international conference on power electronics, drives and energy systems (PEDES) & 2010 Power India,

2010: 1–5. https://doi.org/10.1109/pedes.2010.5712374

Soon JJ, Goh ST, Low KS. Multi-dimension diode photovoltaic model for different PV cell

technologies. In: 2014 International Symposium on Industrial Electronics at Istanbul, Turkey.

https://doi.org/10.1109/ISIE.2014.6865012

considering solar cell junction capacitance. In: 39th Annual Conf. IEEE Ind Elect. Society (IECON 2013). IEEE;

2013; 1613–18. https://doi.org/10.1109/iecon.2013.6699374/

Wolf M, Rauschenbach H. Series resistance effects on solar cell measurements. Adv Energy Convers

1963; 3:455–79. https://doi.org/10.1016/0365-1789(63)90063-8

89

Annexes

Annex A – Calibration certificate: Thermocouple A0

Annex B – Calibration certificate: Thermocouple A1

Annex C – Calibration certificate: Thermocouple A2

Annex D – Calibration certificate: Thermocouple A3

Annexes 90

Annexes 91

Annexes 92

Annexes 93