Optimisation of alumina coating by atomic layer deposition (ALD) process as a protective scheme for CsPbBr3 perovskite quantum dots

(1)

Master’s Programme in Chemical and Process Engineering

Master’s Thesis

Sanduni Thiyaga Pathiraja Mudalige Dona

OPTIMISATION OF ALUMINA COATING BY ATOMIC LAYER DEPOSITION (ALD) PROCESS AS A PROTECTIVE SCHEME FOR CsPbBr

₃

PEROVSKITE QUANTUM DOTS

Examiners: Professor Mika Sillanpää Supervisors: Professor Mika Sillanpää

Professor Raffaella Buonsanti Dr. Anna Loiudice

(2)

ABSTRACT

Lappeenranta University of Technology LUT School of Engineering Science

Master’s Programme in Chemical and Process Engineering

OPTIMISATION OF ALUMINA COATING BY ATOMIC LAYER DEPOSITION (ALD) PROCESS AS A PROTECTIVE SCHEME FOR CsPbBr₃ PEROVSKITE QUANTUM DOTS

Master’s Thesis 2018

92 pages, 56 figures, 11 tables.

Examiners: Professor Mika Sillanpää

Keywords: perovskites, qauntum dots, CsPbBr₃, semiconductor nanocrystals, photovoltaics, atomic layer deposition

All-inorganic perovskite quantum dots (PeQDs) have emerged as a new class of semiconductor nanocrystals with outstanding optical characteristics. Two main practical prob- lems have to be addressed before implementing PeQDs in energy harvesting architectures: anion-exchange and stability under various stringent conditions. Nanocomposites of QD/AlOx assembled using an atomic layer deposition (ALD) process has been developed in the recent years for the growth of amorphous alumina (AlO_x) as a new protection technique for these semiconductor nanocrystals, that acts as a gas and ion diffusion barrier. This thesis work has optimized the ALD process using different ALD modes of operation. Each deposition cycle of the optimized ALD process comprised of a TMA pulse (t₁), diffusion (t₂), purging (t₃), a H₂O pulse (t₄), diffusion (t₅) and purging (t₆). Op- timal conditions were found by varying the ALD process parameters that resulted in an uniform coating, which protects the QDs from degradation. The optimal reaction times were found to be t1=0.010 s, t2=5 s, t3= 10 s, t4=0.010 s, t5=5 s, and t6=60 s.

(3)

ACKNOWLEDGEMENTS

I would first like to convey my sincere gratitude to Prof. Raffaella Buonsanti for granting me this great opportunity to carry out my master’s thesis related research in her research group of Laboratory of Nanochemistry for Energy at École Polytechnique Fédéral de Lau- sanne (EPFL), Switzerland. I would like to thank Dr. Anna Loiudice, my thesis supervi- sor, for her motivation, immense knowledge, constructive and valuable recommendations throughout the planning, development and writing of this research work as well as for her guidance in order to keep my progress on schedule.

I would like to acknowledge Prof. Mika Sillanpää of Lappeenranta University of Tech- nology (LUT), Finland as the first examiner of this thesis. I am extremely indebted to him for his immense support and guidance as well as encouragements, since the moment I chose this master’s thesis opportunity.

Besides my supervisors and examiners, a very special and a humble thanks goes to doc- toral student Seryio Saris for his insightful comments and worthwhile discussions, which incented me to widen my research from various perspectives. My appreciation to all the other members of the research group for making the working atmosphere pleasant and friendly.

Last but not the least, I would like to express my heartfelt gratitude to my parents and my siblings for the continuous encouragement and unfailing support throughout my years of study and during the process of completing this thesis successfully. This accomplishment would not have been possible without them. Thank you.

Lappeenranta, January 5, 2019

(4)

List of Abbreviations

AFM Atomic Force Microscopy ALD Atomic Layer Deposition CQDs Colloidal Quantum Dots EQE External Quantum Efficiency GUI Graphical User Interface ICs Integrated Circuits

ICP-MS Inductively Plasma Coupling Mass Spectrometer LEDs Light Emitting Diodes

LHPs Lead Halide Perovskites NCs Nanocrystals

ODE Octadecene OLAC Oleic Acid OLAM Oleylamine

PL Photoluminescence

PLQY Photoluminescence Quantum Yields PMT Photomultiplier

PeQDs Perovskite Quantum Dots PV Photovoltaics

QDs Quantum Dots

QY Quantum Yield

TCSPC Time-correlated Single Photon Counting TEM Transmission Electron Microscopy TMA Trimethylaluminum

TOP Trioctylphosphine TOPO Trioctylphosphine Oxide TOPSe Tri-n-octyl-phosphine Selenide TOPTe Tri-n-octylphosphine Telluride TR-PL Time-Resolved Photoluminescence SEC Spectro-electrochemical

SEM Scanning Electron Microscopy

(7)

List of Figures

Figure 1 Schematic illustration of the CsPbBr₃/AlO_xnanocomposite [Loiu- dice et al., 2017]. . . 14 Figure 2 (a) Energy scheme of a semiconductor [Weller, 1993] (b) The

expression for the size dependent quantum confinement energy obtained with the spherical “quantum box” model. [I. Klimov, 2003] . . . 20 Figure 3 Quantum confinement of CdSe colloidal QDs(a)CdSe QDs with

their core diameters ranging from 1.8 nm to 6.9 nm. (b)Schematic illustration of band structures and bandgaps in bulk semiconductors as well as QDs of different sizes. (c)Absorption and emission spectra with particle sizes from 1.8 nm to 20 nm. [Chou and Dennis, 2015] . . . 21 Figure 4 Schematic illustration of LaMer’s model for the monodisperse col-

loidal particles growth [Lee et al., 2014] [Hollingsworth, 2006]. . . 23 Figure 5 Illustration of the experimental apparatus used during the synthesis

of CQDs by hot-injection method [Kim et al., 2013] [Hollingsworth, 2006] 25 Figure 6 Three-dimensional (3D) structures of colloidal lead halide per-

ovskite nanocrystals. Two typical structures; cubic (MAPbX₃ where MA is methylammonium, FAPbX₃where FA is formamidinium) and orthorhombic (CsPbX₃) [Akkerman et al., 2018] . . . 27 Figure 7 CsPbBr₃ nanocrystals(a)Schematic of CsPbBr₃ perovskites syn-

thesized by Protesescu et al. (b & c)Images of CsPbBr₃ NCs captured through Transmission Electron Microscopy (TEM) [Protesescu et al., 2015] 29 Figure 8 Colloidal CsPbX3perovskite nanocrystals exhibit size- and composition-

tunability. (a)Colloidal perovskite nanocrystals in solution with toluene.

(b)Photoluminescence spectra for different halide compositions (c)Ab- sorption and photoluminescence spectra for different halide compositions.

(d) Quantum size-dependent absorption spectra for CsPbBr₃ NCs in the size range of 4 - 15nm [Protesescu et al., 2015] . . . 30 Figure 9 PL spectra dependent on temperature for(a)QDs (273-393 K).(b)

Bulk material (50-300 K) [Yang and Zhong, 2016] . . . 30 Figure 10 Defect tolerant beahviour of conventional semiconductors and Lead

Halide Perovskites [Akkerman et al., 2018] . . . 31 Figure 11 Colloidal LHP optoelectronic device architectures. (a)LHP quan-

tum dots embedded in PL down conversion such as LCD dispalys and lighting(b)LEDs; electroluminescent devices(c)Solar cells [Akkerman et al., 2018] [Kagan et al., 2016] . . . 32

(8)

Figure 12 CIE color coordinates corresponding to the CsPbX3 QDs compared to most common colour standards [Protesescu et al., 2015] [Yang and Zhong, 2016] . . . 33 Figure 13 Scheamtic of ligand desorption of LHPs that leads to losing the

colloidal stability [Akkerman et al., 2018] . . . 34 Figure 14 (a)Anion exchange between different different halide perovskites

(b)PL spectra of mixed halide perovskite NCs (CsPb(Br:X)3(X = Cl, I)).

Synthesised from CsPbBr3 NCs using anion exchange [Akkerman et al., 2015]. . . 35 Figure 15 Schematic of a multi-layer device application using different LHPs 36 Figure 16 Schematic of the ALD process. (a) Precursor introduction. (b)

Purge using an inert gas for the removal of excess reactants.(c)Introduc- ing the co-reactant. (d)Excess co-reactants and by-products removed by a second purge [Palmstrom et al., 2015]. . . 37 Figure 17 Schematic representation of Al₂O₃ deposition on QD films . . . . 37 Figure 18 CsPbBr₃Synthesis.(a)Experimental setup.(b)Synthesized CsPbBr₃

nano-crystals in solution. . . 39 Figure 19 Transmission Electron Microscopy (TEM) image of CsPbBr3quan-

tum dots synthesized by hot injection method . . . 40 Figure 20 Spin Coating Technique. (a) An image of the spin-coater. (b)

Schematic representation of the spin-coating technique [Yang et al., 2015].

. . . 42 Figure 21 Four stages of spin coating [Tyona, 2013] . . . 43 Figure 22 Schematic representation of dip-coating technique [Yang et al., 2015] 44 Figure 23 Savannah S200 ALD system [Cambridge NanoTech, 2004] . . . . 45 Figure 24 Savannah S200 front view [Cambridge NanoTech, 2004] . . . 46 Figure 25 Savannah S200 GUI software overview [Cambridge NanoTech,

2004] . . . 47 Figure 26 ALD Mechanism. Schematic diagram of a single ALD cycle with

TMA and water precursors on Si substrate. . . 48 Figure 27 Recipe table (left), Recipe status (right) . . . 50 Figure 28 PL Measurement set-up for the Flurolog spectrometer showing the

excitation port and the emission port . . . 51 Figure 29 Schematic image of cell with a simulated sample (green) placed

inside . . . 52 Figure 30 TCSPC Principle(a)Measurement of fluorescence photons, a cer-

tain amount of time following the excitation pulse. (b)Binned star-stop times forming a fluorescence decay histogram. . . 52

(9)

Figure 31 Integrating Sphere (a) External structure (b)Measurement set-up for a generic QY measurement . . . 53 Figure 32 Experimental set-up for measuring the stability of QDs in oxygen. 54 Figure 33 Schematic illustration of alumina infilling during ALD in QD film,

(a)-(c)multilayer QD and(d) monolayer QD.(a)Cross-sectional view of the multilayer QD before alumina deposition, ligands are excluded.

(b)-(c) Alumina infilling near the top layers of the multilayer film. (d) Alumina infilling on thin QD film. [Palmstrom et al., 2015] . . . 55 Figure 34 SEM images of thin films fabricated at different CsPbBr33 QD

concentrations . . . 57 Figure 35 SEM images for thin films fabricated at different hexane volume

ratios, sample 1,2,3 respectively (from left to right) . . . 58 Figure 36 SEM images of thin films fabricated at different dip coating pa-

rameters, refer table 2 . . . 59 Figure 37 Contact angle measurement of p-doped silicon after surface treat-

ment with 3-mercaptopropyl trimethoxysilane . . . 60 Figure 38 CsPbBr3 film fabrication on treated silicon. (a)Dip coating (con-

centration - 2 mg/ml with 1:1 heaxne/octane, withdrawal speed - 1.9 mm/s, waiting time - 60 sec)(b)Spin coating (concentration - 1.5 mg/ml with 1:1 hexane/octane) . . . 61 Figure 39 AFM measurement of dip coated sample in figure 38 (a)(a)Topog-

raphy of the sample, including line where a profile was measured (dark area is the scratch). (b)Line profile. . . 62 Figure 40 SEM images of dip coated samples as in table 3(a)Sample 1(b)

Sample 2(c)Sample 3(d)Sample 4(e)Sample 5(e)Sample 6 . . . 63 Figure 41 SEM images of dip coated sample 5 towards the edge of the 2x2

substrate . . . 64 Figure 42 SEM images of spin coated samples as in table 4. (a)Sample 1(b)

Sample 2(c)Sample 3(d)Sample 4(e)Sample 5(e)Sample 6 . . . 64 Figure 43 CsPbBr₃stability test in oxygen environment. (a)Schematic illus-

tration of the interaction between Oxygen and QDs. (b)Normalized PL intensity and QY in oxygen/vacuum conditions [Lorenzon et al., 2017] . . 66 Figure 44 Schematic illustration of continuous mode operation. . . 67 Figure 45 Overview of PL measurement to check stability in oxygen . . . 69 Figure 46 PL scans obtained using Fluorolog. . . 69 Figure 47 Evolution of PL intensity for QD films fabricated on glass wafers

with different number of ALD cycles, upon switching from vacuum to oxygen atmosphere (samples with concentration of 1.5 mg/ml) . . . 70

(10)

Figure 48 Evolution of PL intensity for QD films (monolayers) fabricated on silicon wafers with different number of ALD cycles, upon switching from vacuum to oxygen atmosphere . . . 71 Figure 49 (a)Emission spectrum in nitrogen for the nano-composites in fig-

ure 48. (b)Maximum PL intensity plotted against the samples with different ALD cycles. . . 72 Figure 50 Schematic illustration of continuous mode operation at higher tem-

perature . . . 72 Figure 51 (a)Evolution of PL intensity upon switching from vacuum to oxy-

gen. (b) Emission spectrum in nitrogen. (c)PL decay dynamics in oxygen. (d)Summary of biexponential fitting results of PL decay in (c). . . . 73 Figure 52 Schematic representation of exposure mode operation [Cambridge Nan-

oTech, 2004] . . . 75 Figure 53 Exposure mode ALD operation at different diffusion waiting times

for stabilizing QD/AlO_xnano-composites in oxygen atmosphere(a)Recipe 1 (samples Expo 1,2,3) and Recipe 2 (samples Expo 4,5,6)(b)Recipe 3 (samples Expo 7,8,9) and Recipe 4 (samples Expo 10,11,12) . . . 77 Figure 54 Effect of H2O purge on stabilizing QD/AlOx nano-composites in

oxygen atmosphere. (a) Evolution of PL intensity upon switching from vacuum to nitrogen. (b)Emission spectrum in nitrogen. . . 79 Figure 55 Effect of N₂flow during H₂O purge on stabilizing QD/AlO_xnano-

composites in oxygen atmosphere. (a) Evolution of PL intensity upon switching from vacuum to nitrogen. (b)Emission spectrum in nitrogen. . 80 Figure 56 Overview of different ALD recipes on stabilizing QD/AlO_xnano-

composites on p-doped silicon in oxygen atmosphere.(a)Evolution of PL intensity upon switching from vacuum to nitrogen. (b)Emission spectrum in nitrogen. . . 81

(11)

List of Tables

1 Overview of hexane and octane volume ratios used for QD stock solution

dilution . . . 57

2 Overview of dip coating . . . 59

3 Overview of dip coating on treated silicon at different concentrations of QD solution as well as with different hexane/octane volume ratios . . . . 62

4 Overview of spin coating on treated silicon at different concentrations of QD solution as well as with different hexane/octane volume ratios . . . . 64

5 Standard ALD recipe . . . 67

6 Initial sample preparation on glass substrate . . . 68

7 Calculation of%quenching for 100 ALD cycle samples . . . 71

8 Initial Exposure mode ALD Recipe . . . 76

9 Different recipes to study the effect of TMA/H2O diffusion time on alumina growth . . . 77

10 Summary of PL intensities for the samples in figure 53 . . . 78

11 Optimized Exposure mode ALD Recipe . . . 81

(12)

1 INTRODUCTION AND BACKGROUND

1.1 Introduction to the Research Topic

Semiconductors can be defined as a group of materials having conductivities and properties lying between those of insulators and conductive metals [Neamen, 2003, p.1-3]

[Brand et al., 2009]. They are generally classified as elemental semiconductor materials and compound semiconductor materials [Neamen, 2003, p.1-3]. Elemental semiconductor materials are composed of single species of atoms, found in group IV of the periodic table and special combination elements from group II and group IV, III and V or elements from the group IV of the periodic table are used to form compound semiconductor materials [Neamen, 2003, p.1-3] [Brand et al., 2009] [Colinge and Colinge, 2006, p.363-370].

In the 20th century, semiconductors have proven to have a vital role for the advancement of technological devices [Ozcan and Dincer, 2018]. A wide range of semiconductor devices can be identified, such as transistors, diodes and integrated circuits (ICs), as well as photodetectors and solar cells [Sze, 2002, p.1-15]. They made their ways into our every day lives through various electronic devices that make our lives agile and more produc- tive [Ozcan and Dincer, 2018]. Majority of the semiconductor devices are made of silicon (Si), however germanium (Ge) and compound semiconductors such as gallium arsenide (GaAs) have demonstrated to be extremely efficient as well [Neamen, 2003, p.1-3] [Oz- can and Dincer, 2018]. They have been widely employed in the field of energy production such as photovoltaics (PV) and light emission such as light emitting diodes (LEDs) [Oz- can and Dincer, 2018] [University of Maryland, 2018] [Mishra and Singh, 2007].

The global semiconductor industry is highly competitive as they are crucial components of electronic devices [Statista, 2018]. There has been year-on-year growth rate in the global semiconductor industry sales over the past twenty years, with an annual sales amounting to over 400 billion U.S. dollars worldwide in 2017 [Semiconductor Industry Associa- tion (SIA), 2017].

Semiconductor nanocrystals also familiar as colloidal quantum dots (CQDs) are a pi- oneering discovery of the intensive fundamental research by scientists, that have caught enormous attention due to their distinctive absorption properties dependent on nanocrystal size, absorption cross-sections over a extensive spectral range, and the feasibility to gen- erate multiple excitons by single photons [Jun et al., 2006] [Kim et al., 2013] [Bera et al., 2010]. CQDs have gained popularity as a next generation of high-efficiency photovoltaics,

(13)

with low cost materials and processes [Kramer and Sargent, 2011] [Song and Jeong, 2017]. The history of quantum dots (QDs) dates back to 1980, where Russian physicist Ekimov discovered QDs in glass matrix [Navillum Nanotechnologies, 2016] [Zhu et al., 2013]. However, it was in 1981 when Loius E. Brus actually discovered the colloidal semi-conductor nanocrystals [Navillum Nanotechnologies, 2016] [Zhu et al., 2013]. CdX (X = S, Se, Te) are the most investigated QDs, as they have great optical and electrochemical properties, since their synthesis by Murray et.al after nearly a decade of the last CQDs discovery [Zhu et al., 2013] [Murray et al., 1993].

Whilst these initial CQDs discoveries mark a milestone in the next-generation photovoltaics, perovskite quantum dots (PeQDs) have grabbed significant interest in the recent years [Wang et al., 2017b]. One of them, CsPbX3 (X = Cl, Br, I) has shown outstanding unique optical properties such as colour tunable narrow emission wavelength that covers the entire visible region as well as high photoluminescence quantum efficiency [Wang et al., 2017b] [Shi et al., 2017] [Yu et al., 2018] [Protesescu et al., 2015]. Despite the excellent optical properties they maintain when they are in organic solutions such as toluene, their instability towards air, moisture, light irradiation and temperature as solids remains a ubiquitous impediment [Loiudice et al., 2017] [Ren et al., 2017] [Pan et al., 2015]. The instability of PeQDs has thus restricted the use of optoelectronics based on perovskites in practical as well as commercial applications, especially in energy harvesting applications [Pan et al., 2015]. The exploration of their properties has also been restrained, that could lead to highly sanitized conditions for the nanocrystals [Pan et al., 2015].

In this study, an atomic layer deposition (ALD) process, which is recognised as a thin film deposition technique has been developed and optimized with regards to system conditions such as reaction time, temperature and pressure. The ALD process encapsulates the QD surfaces with an amorphous alumina (AlO_x) matrix thus, acting as a gas and an ion diffusion barrier to stabilize the CsPbBr₃QDs. Steady-state photluminescence (PL) of QD/AlO_xnanocomposites developed this way has been studied to determine the stability and PL efficiency of QDs upon exposure to oxygen.

(14)

1.2 Scientific Background

This study has been adopted from the works of Loiudice et al. and they were exploited for further optimization of the atomic layer deposition process for stabilizing the CsPbBr3

QDs based on the established data of developed CsPbBr3 QD/AlOx inorganic nanocomposites on glass substrates. The authors used ALD process, that has been developed as a novel protection scheme for these semiconductor NCs for the growth of amorphous alumina matrix (AlO_x) on their surfaces, thus to study the stability of CsPbBr₃ QD/AlO_x nanocomposites [Loiudice et al., 2017]. Aforementioned procedure has been followed to fabricate CsPbBr₃QD/AlO_xinorganic nanocomposites on p-doped silicon substrates during this thesis work. A comparison study was conducted to evaluate the nanocomposite’s photoluminescence and stability in oxygen before and after the ALD process optimization.

As a precursor of alumina, Trimethylaluminium (TMA) has been used, whereas H2O acts as a co-reactant. ALD process was developed at a low temperature of 50^◦C, where each deposition cycle comprised of a TMA pulse (t1) and a H2O pulse (t2) with purging (t3 &

t₄) to remove excess precursor and co-reactant as well as the methane produced during the process [Loiudice et al., 2017]. Optimal conditions were found by varying the ALD process parameters that resulted in an uniform coating, which protects the QDs from degradation. The optimal reaction times for the aforementioned steps of each deposition cycle were found to be t₁= 0.015s, t₂= 10s, t₃= 0.01s and t₄= 180s [Loiudice et al., 2017].

Avoiding QD degradation requires maintaining a low temperature as well as use of short H2O pulses during the ALD process [Loiudice et al., 2017]. The nanocomposites showed exceptional stability during their exposure in air, heat, irradiation and upon immersion in water, whilst preserving over 50% of the original photoluminescence [Loiudice et al., 2017]. A schematic illustration of the model is given in figure 1.

Figure 1.Schematic illustration of the CsPbBr₃/AlO_xnanocomposite [Loiudice et al., 2017].

(15)

Panet al. have pursued a surface defects passivation strategy in nanosized organometal halide perovskites. An inorganic-organic hybrid ion pair of didodecyl dimethylammonium sulfide that acts as a capping ligand was used to coat the QD surfaces, which led to the development of extremely stable perovskite QDs in air [Pan et al., 2015]. This passivation approach has led to a considerable enhancement in the photoluminescence qauntum yields (PLQY) with outstanding operational stability in ambient conditions as well as high optical fluences [Pan et al., 2015]. An enhancement of the quantum yield was observed from 49% to 70% upon the addition of didodecyl dimethylammonium sulfide precursor to a solution of CsPbBr3QDs [Pan et al., 2015].

Li et al. have also applied a TMA vapour based crosslinking using an ALD system on PeQD thin films with the intention to render the nanocrystal films insoluble in organic solvents that is typically caused by the aliphatic ligands present on them. Thin films of perovskites have been exposed to short TMA vapour pulses in a confined vacuum chamber for achieving crosslinking and a well connected, hydroxide restricted aluminum oxide network that holds the nanocrystals together, thereby making them insoluble [Li et al., 2016a]. As this technique does not lead to ligand exchange, the original crystal arrangement, electronic properties as well as the photoluminescence of the film remain unchanged.

Many other previous research studies have revealed that metal oxide infilling with Al₂O₃ and ZnO using a low-temperature ALD process enhances the oxidative and photothermal stability of various other QDs such as CdSe,CdTe, PbSe and PbS [Loiudice et al., 2017]

[Lambert et al., 2011] [Liu et al., 2013] [Yin et al., 2016] [Liu et al., 2011] [Pourret et al., 2009] [Devloo-Casier et al., 2016] [Valdesueiro et al., 2016a].

This class of inorganic perovskite QDs are highly susceptible to temperature, moisture and light, which makes it crucial to develop and optimize the ALD technique. Afore- mentioned studies have served as a background for the development and optimization of an ideal ALD process that can stabilize PeQDs to a higher standard for long term device performance.

(16)

1.3 Problem Statement and Motivation

CsPbBr3 perovskite QDs are well studied in solution and have gained popularity as the next generation photovoltaics due to their defect tolerance, that leads to high quantum yields. However, practical application of QDs in photodetectors and solar cells require them to be in the form of thin films rather than solutions. Films of perovskite QDs have already been studied by many researchers recently in terms of their stability and interaction under different environmental conditions. While these initial reports are very en- couraging, a great amount of additional work must be done, particularly regarding their stability if they are to be implemented for practical and commercial applications.

The focus of this work therefore lies on investigating the behaviour of CsPbBr₃perovskite QDs deposited in thin films with the objective of improving their stability in different environmental conditions using the ALD process. To reach this goal the following objectives were made and studied;

• Film fabrication using dip and spin coating for obtaining a uniform monolayer on p-doped silicon wafers.

• Develop and optimize the alumina deposition process using different ALD operation methods.

• Investigate the environmental impact, in particular the consequences of oxygen on photophysical properties of thin films of QD/AlOxnanocomposites.

(17)

1.4 Structure of the Thesis

The thesis starts with an introduction to give a general overview about the research topic that has been studied, accompanied by a scientific background that summarises the most recent research that has been done along with their respective outcomes. This is further accompanied by a problem statement and a motivation section, which addresses the void in the knowledge and previous research work that the thesis has endeavoured to fulfil, accomplish and improve.

The theoretical framework and literature review, gives a deeper insight into the more specific background information on this topic of interest, as well as it evaluates the related previous research work that has been done.

The materials and methods chapter outlines all the technologies, measurement and analy- sis techniques that have been used in order to conduct the laboratory experiments.

Results and discussion section summarizes all the experimental data qualitatively, that have been found out in relation to the research questions and hypotheses. The achieved results have been discussed to interpret and provide explanations to the results obtained in a wider context, while relating the specific results to previous research or theory.

At last conclusions have been drawn to discuss the major outcomes from the research and if the research aims, objectives have been achieved and further suggestions as future directions on the research topic.

(18)

2 THEORETICAL FRAMEWORK AND LITERATURE REVIEW

2.1 Colloidal Quantum Dots

A nanostructure can be defined as a solid with a particle size variation of <100 nm, that manifests discrete optical and electronic properties, and is categorized as [Bera et al., 2010] [Semonin et al., 2012] [Balaguru and Jeyaprakash, 2013],

1. One dimensional confinement - Quantization of the particle motion arises in two directions, that leads to free movement only in a single direction. ex.: quantum wires

2. Two dimensional confinement - Quantization arises in one direction and the particle is able to move in both the other directions. ex.: quantum wells

3. Zero dimensional or three dimensional confinement - Quantization takes place in all three directions. ex.: quantum dots

Colloidal quantum dots are thus chemically prepared nanometer-sized semiconductor nanocrystals synthesized and suspended in solution phase [Song and Jeong, 2017] [Yu et al., 2018] [Kim et al., 2013]. A significant growth related to CQDs research domain has been witnessed worldwide in the past few decades, since their discovery [Kim et al., 2013]. Even though QDs are zero dimensional to bulk, they are interpreted as cubes in quantum mechanics [Bera et al., 2010]. The electronic structure of these materials is tailored by changing the local material composition and by confining the electron wave functions in nanometer-sized foils or grains [Boxberg and Tulkki, 2004]. They are often called quantum structures due to this quantization and confinement of electron energies that are determined by their particle size in the nanometer scale. By confining the electrons in all three directions by a potential barrier, the nanocrystals are called quantum dots [Boxberg and Tulkki, 2004] [Bakkers, 2000].

QDs offer great benefits through their easy solution based synthesis from inexpensive materials, and distinctive chemical and physical properties that facilitate colour tunable functionality as a result of their size dependent fundamental properties in the nanometer range, a phenomenon known as quantum size effect [Kim et al., 2013] [Hollingsworth, 2006]. Unique electronic and physical properties of CQDs are significant for both individual nanocrystals as well as packed nanocrystal films, which allow to investigate for novel physical and chemical phenomena [Kim et al., 2013]. A considerable attention

(19)

has been focused on quantum dots in the last three dacades, as a significant new class of semiconductor materials in rapid development of a various novel emerging technologies, where they are used as the major constituent for fabricating electronic and optoelectronic devices, such as sensitive photodetectors, lighting solutions and thin film photovoltaics [Bakkers, 2000] [Bera et al., 2010] [Kim et al., 2013] [Song and Jeong, 2017].

Commercialization is underway, where the commercial sector has started to implement the already reported laboratory scale applications of colloidal semiconductor quantum dots for real applications in photodetection and in light emission [Smyder and Krauss, 2011] [Kim et al., 2013].

2.1.1 From Bulk Semiconductors to Quantum Dots

The semiconductor properties observed in many inorganic substances are not the properties of the individual molecules or atoms themselves, but of an ordered crystal lattice of the constitutive elements in a specific array [Weller, 1993]. Therefore, the photophyiscs of semicondcutor quantum dots can be described by the band theory [Smyder and Krauss, 2011]. In general, an isolated atom consists of discrete energy levels where electrons orbit. When many atoms are in close proximity such as in crystalline solids, the atomic orbitals overlap leading to splitting of the original discrete levels to many separated levels [Johansen, 2005] [Weller, 1993]. They however may be considered as a continuous band of energy states because the levels are closely separated. The two highest energy levels are known as the electron rich valence band and the empty conduction band. These two levels are parted by a region known as the forbidden gap, or the bandgap, E_g and defined as energies that cannot be held by electrons in a solid [Johansen, 2005]. Bandgap can therefore be defined as the difference in energy that separates the highest occupied valence band energy and the lowest unoccupied conduction band energy.

If a semiconductor absorbs a photon with an energy that is eqaul to or above the bandgap energy, electrons (e^-) receive energy to move from valence band to the empty conduction band, which leave holes (h⁺) in the valence band [Smyder and Krauss, 2011]. A particle known as an exciton is formed by the combination of the two charge carriers, electrons and holes, while coulombic forces hold them together similar to that in a hydrogen atom [Smyder and Krauss, 2011] [Saris, 2017]. The formed exciton recombines either through fluorescence and re-emits a photon with an energy equal to E_g, or through non-radiative processes such as creating heat [Smyder and Krauss, 2011]. The distance that separates the electron and the hole is known as the exciton Bohr radius,a_B, which varies depending on the material and is given as,

(20)

a_B = h² e² ( 1

m_e∗+ 1

m_h∗) (1)

whereis known as the dielectric constant of the semiconductor,eis the electron charge, meis the effective electron mass andmhis the hole mass [Saris, 2017] [Bakkers, 2000].

Figure 2. (a) Energy scheme of a semiconductor [Weller, 1993] (b) The expression for the size dependent quantum confinement energy obtained with the spherical “quantum box” model.

[I. Klimov, 2003]

Bandgap and the associated energy is determined by the interconnection between the crystal size and the exciton Bohr radius, a defining feature of a semiconductor, which dictates the emission energy and the colour of light emitted in the visible region of the spectra [Mishra et al., 2012] [I. Klimov, 2003] [Chou and Dennis, 2015]. In semiconductors of macroscopic size, the bandgap is a established parameter depending on the identity of the material [I. Klimov, 2003] [Hollingsworth, 2006]. However, for semiconductor nanocrystals with sizes below about 25 nm, the particle size is identical or relatively smaller in comparison to the size of excitons observed in semiconductors of macroscopic size [Weller, 1993] [Hollingsworth, 2006] [I. Klimov, 2003]. This leads to an excited electron and a hole to be physically confined into a geometry smaller than their natural Bohr radius, in a regime of strong confinement, where a real materials system experiences the qauntum mechanical ’particle-in-a-box or quantum box’ potential energy function model [Weller, 1993] [Smyder and Krauss, 2011] [Norris and Bawendi, 1996]. Three dimensional quantum confinement is experienced by the exciton wave functions due to these geometrical constraints arising from the particle boundaries. As a result of this confinement, continuous energy bands of a bulk material are quantized such that quantum

(21)

dots have discrete, atomic-like energy levels and electronic transitions that shift to higher energies with decreasing nanocrystal size [Norris and Bawendi, 1996] [I. Klimov, 2003].

The quantum box model envisages that for a spherical QD having a radius R, bandgap energy is proportional to 1/R² as shown by the expression in figure 2b), which implies that with decreasing QD size the gap increases [I. Klimov, 2003]. This phenomenon is recognized as the quantum-size effect as illustrated in figure 3, and it has a significant role in QDs [Kim et al., 2013] [Hollingsworth, 2006].

Figure 3. Quantum confinement of CdSe colloidal QDs(a)CdSe QDs with their core diameters ranging from 1.8 nm to 6.9 nm. (b) Schematic illustration of band structures and bandgaps in bulk semiconductors as well as QDs of different sizes. (c)Absorption and emission spectra with particle sizes from 1.8 nm to 20 nm. [Chou and Dennis, 2015]

As illustrated in figure 3(b), a bulk semiconductor of macroscopic size has a fixed energy gap,E_gthat separates its continuous conduction and valence energy bands. All states of a valence band up to the edge are typically completely occupied by the electrons, whereas those of a conduction band are empty. Figure 3(a) shows size dependent emission colour tunability of QDs [Chou and Dennis, 2015] [I. Klimov, 2003]. The decreasing QD size leads to a larger bandgap indicating a higher energy requirement to excite the electrons that produces higher frequency, shorter wavelengths corresponding to blue light in the spectra, whereas a small bandgap with lower energy corresponds to a red shift in the emission spectra with longer wavelengths as illustrated in figure 3(c) [Hollingsworth, 2006].

(22)

The unique quantum confinement of semiconductor nanocrystals thus facilitates new approaches for applications, particularly in electronics and optoelectronics due to their size-tunable emission, which makes it feasible to evaluate the electronic behaviour in a size regime transitional to molecular and bulk limits of matter [Norris and Bawendi, 1996] [Wang et al., 2017b] [Song and Jeong, 2017] [Yu et al., 2018] [Hollingsworth, 2006]. Optoelectronics have been developed based on QDs capable of functioning at wavelengths that were previously unfeasible or have access to, for a given semiconductor category [Bimberg and Pohl, 2011]. As fluorophores, high quality QDs are eminently bright with narrow emissions, whilst quantum yields (QYs) or fluorescence efficiencies have values close to unity, which is an indication of emitted fluorescence photons for each photon absorbed [Smyder and Krauss, 2011]. Photostability of QDs is remarkably longer, up to hours under continuous excitation in comparison to individual organic molecules that are prone to photobleach within seconds or minutes [Smyder and Krauss, 2011].

QDs further display minor Stokes shifts, which corresponds to the red shift observed in the emission spectra with regards to the absorption spectra, resulting due to vibrational re- laxation and solvent reorganization [Smyder and Krauss, 2011] [Britannica, 2018] [Bagga et al., 2007]. When a bulk semiconductor absorbs photons with energy greater than their bandgap, charge carriers having surplus kinetic energy are produced that dissipate via phonon emission, which are associated with lattice vibrations of a solid [Semonin et al., 2012]. However, the quantum confinement of semiconductor nanostructures has the distinctive potential to provide novel routes that control the energy flow in optoelectronics when excited at energies far above their bandgap, that enhance the efficiency of the pri- mary photoconversion step [Smyder and Krauss, 2011] [Semonin et al., 2012]. Most significantly, a fixed excitation wavelength enables the collection of several QD emission wavelengths [Smyder and Krauss, 2011].

These outstanding properties of CQDs over bulk semiconductors make them much more versatile for new applications in electronics and optoelectronics.

2.1.2 Synthesis of Colloidal Quantum Dots

Technological as well as fundamental scientific importance of colloidal nanocrystals have led to the development of versatile and successful synthetic routes in the last three decades [Hyeon, 2003] [Kim et al., 2013]. The synthetic techniques of colloidal QDs are typically classified as ’top down’, and ’bottom up’ methods [Sytnyk, 2015] [Bera et al., 2010].

1. Top down method - Involves thinning or milling of a bulk semiconductor mechan- ically to a fine powder, which is eventually dispersed in a surfactant solution. Syn-

(23)

thesis of III-V and II-IV QDs having a particle size of approximately 30 nm has been achieved with this method [Sytnyk, 2015] [Bera et al., 2010]. Major draw- back of this method are,

• Impurities are incorporated in the QDs during milling.

• Shapes and sizes of nano-crystals are uncontrolled.

2. Bottom up method - Nanoparticles are synthesized from metal precursors where ligands are present to control the nanocrystal growth and prevent coagulation [Sytnyk, 2015]. Monodisperse samples synthesized this way can self assemble and form ordered superstructures [Sytnyk, 2015] [Bera et al., 2010]. Bottom up methods can also be categorized as wet-chemical and vapour-phase methods [Bera et al., 2010].

The use of wet chemical methods have proven to be successful due to their uncomplicated nature as well as their potential for producing large batches of the product [Hyeon, 2003].

The foundation for the colloidal synthesis was laid by Victor K. LaMer and Robert H.

Dinegar in the early 40s. The kinetics of ’burst nucleation’ was described quantitatively by the classical nucleation theory while working with oil aerosols and sulfur hydrosols [Robb and Privman, 2008] [LaMer and Dinegar, 1950]. Lamer and Dinegar suggested as shown in figure 4, the driving force of a rapid nucleation is supersaturation, followed by the diffusing atomic matter absorption, known as Ostwald Ripening to the already nucleated particles that subsequently grow into bigger crystals [Robb and Privman, 2008].

Figure 4.Schematic illustration of LaMer’s model for the monodisperse colloidal particles growth [Lee et al., 2014] [Hollingsworth, 2006].

(24)

Synthesising colloidal nanocrystals using wet chemical methods require three components: precursors, organic surfactants and solvents. It is possible that surfactants may serve as solvents [Yin and Paul Alivisatos, 2005]. Typically, alkyl phosphonic acids, alkyl phosphine oxides, alkyl phosphines, some nitrogen-containing aromatics, fatty acids and amines are used as organic surfactants that are capable of solvating nanocrystals dynam- ically [Yin and Paul Alivisatos, 2005]. Various wet chemical synthesis techniques have been in use known as sol-gel process, microemulsion process and hot-solution decomposition method [Bera et al., 2010].

Sol-gel Process

• Metal precursors such as acetates, alkoxides or nitrates in a medium which is acidic or basic are used to prepare asol, where nanoparticles are dispersed in a solvent by brownian motion[Bera et al., 2010].

• Hydrolysis, condensation where the sol formation takes place and growth during which the gel formation occurs are the three main steps in this method [Bera et al., 2010].

• II-VI and IV-VI QDs such as PbS, ZnO and CdS have been synthesized this way [Bera et al., 2010].

• Its application is limited by wide particle size distribution as well as high number of defects [Bera et al., 2010].

Micro-emulsion Process

• Synthesis can be done at room temperature [Bera et al., 2010].

• Classified as normal microemuslsions (i.e. oil-in-water) and reverse microemul- sions (i.e. water-in-oil) [Bera et al., 2010].

• Core and core-shell QDs of group II-VI such as CdS, CdSe/ZnSe, ZnSe, ZnS/CdSe as well as QDs of group IV-VI are prepared this way [Bera et al., 2010].

Hot solution decomposition method

Hot solution decomposition method also known as the hot injection method is a well established method that revolutionized the research in the field of nanotechnology and it has become the widely used method for preparing QDs [Bera et al., 2010] [Yin et al., 2016]. A precursor solution usually kept at room temperature is rapidly injected into a warm surfactant solution in this method. The high temperature facilitates chemical transformation of precursors into monomers [Yin et al., 2016]. This subsequently leads to the formation of monodisperse QDs (standard deviation of 5% about the average particle size) and the presence of surfactant molecules affects the nanocrystal growth [Lim et al., 2012] [Yin et al., 2016] [Bera et al., 2010].

(25)

The two steps of nanocrystal formation are,

• Nucleation- Precursor decomposition at high temperature leads to instantaneous monomer supersaturation that occurs at the critical point of nucleation within a small time span followed by a burst of nanocrystal nucleation [Lim et al., 2012] [Yin et al., 2016]. The temperature is decreased immediately after hot injection and nucleation is seized as a result, which prevents the consumption of all the precursor that could drive fast aggregation and ripening process [Murray et al., 1993].

• Growth- Additional monomers present in the solution drives the nuclei growth to bigger crystals at the reduced temperature [Yin et al., 2016] [Murray et al., 1993].

Figure 5. Illustration of the experimental apparatus used during the synthesis of CQDs by hot- injection method [Kim et al., 2013] [Hollingsworth, 2006]

This hot injection method generates nanocrystals based on the condition that monomers are able to anneal and rearrange during growth [Yin et al., 2016]. It is crucial to know the cohesive energy per atom that coordinates with the solid melting temperature for deter- mining the optimal conditions for nanocrystal growth [Yin et al., 2016]. Thus, deciding the ideal temperature that allows monomer rearrangement and annealing within a growing nanocrystal is crucial [Yin et al., 2016].

Identification of ideal surfactants as well as precursors is also pivotal. Choosing organic surfactant is based on their tendency to adhere to a growing nanocrystal [Yin et al., 2016]. The metal coordinating and solvophilic groups inhibit the growth and aggregation of nanocrystals due to their electron donating nature that allow coordination to electron- deficient metal atoms at the nanocrystal surface. Nanocrystal solubility varies as the second end of the surfactant extends to the solvent leaving nanocrystals with a hydrophobic

(26)

surface [Yin et al., 2016]. Accordingly, the presence of organic surfactants also deter- mines the size, spatial arrangement as well as shape of some of the inorganic solids during the colloidal nanocrystal growth [Yin et al., 2016]. Precursors such as organometal- lic compounds are chosen for their propensity to react or disintegrate rapidly, yielding monomers at the nanocrystal growth temperature that drives nanocrystal nucleation and growth [Yin et al., 2016].

This approach was first experimented by Bawendi and co-workers in 1993 for synthesizing nanocrystals of cadmium chalcogenides (CdS, CdSe, CdTe) [Bera et al., 2010] [Mur- ray et al., 1993] [Hyeon, 2003]. Cadmium precursor has been chosen as dimethyl cadmium, ME2Cd, while chalcogen precursors were chosen to be tri-n-octyl-phosphine telluride (TOPTe) and tri-n-octyl-phosphine selenide (TOPSe) [Murray et al., 1993] [Kim et al., 2013]. Coordinating solvent is a mixture of trioctylphosphine oxide (TOPO) and trioctylphosphine (TOP), which is added into a three-neck round bottomed flask followed by degassing. The precursor solution is made with ME₂Cd and TOPSe. This is then injected at a temperature of 300^oC into a hot TOPO placed in the flask. CdSe nuclei are instantly formed followed by QDs growth throughOstwald Ripening [Murray et al., 1993] [Bera et al., 2010] [Mahajan et al., 2013]. As a coordinating solvent, TOPO stabilizes the dispersion of QDs, enhances surface passivation as well as slows the growth of QDs by providing an adsorption barrier [Bera et al., 2010]. Precursor reactivity as well as reaction variables such as temperature, concentration, time and coordinating solvents/surfactants strongly affect the nucleation and growth kinetics [Lim et al., 2012] [Bera et al., 2010].

This facilitates the synthesis of crystalline, monodisperse and highly luminescent QDs across a broad range of well controlled and narrowly distributed sizes [Kim et al., 2013].

2.2 All-Inorganic Cesium Lead Halide (CsPbX

₃

, X = Cl, Br, I) Per- ovskite Quantum Dots

A new category of semiconductor nanocrystals have emanated over the recent years known as trihalide perovskite quantum dots and received enormous attention as one of the most promising semiconductor materials [Yang and Zhong, 2016] [Pan et al., 2015] [Pan et al., 2016] [Shi et al., 2017] [Wang et al., 2017b] [Huang et al., 2017] [Lorenzon et al., 2017] [Zhang et al., 2017] [Akkerman et al., 2018]. These layered trihalide perovskites have a crystal structure of AMX₃, where A is an organic (CH₃NH₃⁺, CH(NH₂)₂⁺) or an inorganic (Cs⁺) monovalent cation, M is a divalent metal cation such as Sn²⁺, Pb²⁺, and X, a halide (Cl^–, Br^–, l^–) [Lorenzon et al., 2017] [Akkerman et al., 2018] [Bekenstein et al., 2015].

(27)

Figure 6. Three-dimensional (3D) structures of colloidal lead halide perovskite nanocrystals.

Two typical structures; cubic (MAPbX3 where MA is methylammonium, FAPbX3 where FA is formamidinium) and orthorhombic (CsPbX3) [Akkerman et al., 2018]

These two structures have the unique 3D coordination, where M metal cation, Pb²⁺ is in coordination with six halide ions that are arranged in an octahedral configuration as shown by figure 6 [Shi et al., 2017]. The octahedra share corners with the organic or inorganic A cation, MA⁺, FA⁺or Cs⁺that is positioned in the large voids intermediate to the octahedra [Shi et al., 2017] [Akkerman et al., 2018].

Lead halide perovskites (LHPs) appear to be newcomers in the semiconductor nanocrystal family, however the first discovery of LHPs dates back to 1890s, where bulk cesium LHPs were reported [Akkerman et al., 2018]. The most established research and the eventual discovery of CsPbX₃ NCs are reported in 2015 [Akkerman et al., 2018].

Since then, they have gained a lot of popularity as the future promising generation of photovoltaics, due to their remarkable electronic and optical properties that results from their colour tunability over the entire visible spectra, as well as their remarkably high near-unity photoluminescence, quantum yields and conversion efficiencies [Akkerman et al., 2018] [Bekenstein et al., 2015] [Pan et al., 2015] [Huang et al., 2017]. Further- more, the solution-based processing makes the synthesis straightforward making them more promising [Huang et al., 2017] [Wang et al., 2017b]. Colloidal QDs are also easier to process into devices with inexpensive techniques such as spin-coating, drop-casting and dip-coating. Moreover, this preparation technique offers the possibility to functional- ize the QD surface and to tune its chemistry according to the application in need [Saris, 2017].

(28)

2.2.1 Synthesis of CsPbX₃ PeQDs

The superior chemical stability and excellent opto-electronic properties of all-inorganic CsPbX3 perovskite quantum dots in comparison to organic perovskite quantum dots has motivated scientific research to develop disparate colloidal synthesis routes [Shi et al., 2017] [Akkerman et al., 2018]. Several synthesis routes have been realized over the years that involve expedient solution-based processes, as mentioned in section 2.1.2 [Shi et al., 2017]. Protesescu and co-workers’ research in 2015 has proven to be the most influential and successful, in discovering the ultimate colloidal CsPbX₃ perovskite quantum dots [Akkerman et al., 2018] [Pan et al., 2016] [Shi et al., 2017] [Wang et al., 2017a] [Bekenstein et al., 2015]. Protesescuet al. adopted the previously developed hot- injection and fast-cooling approach by Bawendi and co-workers, who synthesized metal chalcogenide quantum dots, for synthesizing the ultimate nearly monodisperse colloidal CsPbX3 that exhibit both size- and compositional-bandgap tunability as in figure 8 [Pan et al., 2016] [Shi et al., 2017] [Protesescu et al., 2015].

Hereby, the cesium precursor of Cs2CO3 was allowed to react at a temperature of120^◦C with oleic acid (OLAC) and octadecene (ODE) that serve as surfactants and solvents forming a solution of Cs-oleate that is kept at a temperature of100^◦C. Cs-oleate is subsequently injected into a hot solution (140^◦C -200^◦C) of Pb(II)-halide solubilized in OLAC and oleylamine (OLAM) that colloidally stabilize the CsPbBr₃ nanocrystals [Protesescu et al., 2015] [Shi et al., 2017]. In situ PL measurements indicated fast nucleation and growth kinetics soon after the hot injection, where the first 1-3 seconds correspond to majority of nanocrystal growth [Protesescu et al., 2015] [Shi et al., 2017]. This in fact is faster in the presence of heavier halides. Altering the reaction temperature in the range of 140^◦C -200^◦C, expediently modulates the nanocrystal size in the 4-15 nm range, where the QD size increases with increasing temperature [Protesescu et al., 2015] [Shi et al., 2017]. Reaction for CsPbBr₃ formation can be written as follows [De Roo et al., 2016],

2 Cs(OOCR) + 3 PbBr2→2 CsPbBr3 + Pb(OOCR)2

where, OOCR = oleate

CsPbX3are typically crystallized in tetragonal or orthorhombic polymorphs as in figure 6, however a cubic phase was observed (figure 7(a)) for all perovskite lattices at high temperature synthesis state accompanied by contributions from the surface energy [Protesescu et al., 2015] [Shi et al., 2017].

(29)

Figure 7.CsPbBr3nanocrystals(a)Schematic of CsPbBr3perovskites synthesized by Protesescu et al.(b & c)Images of CsPbBr₃NCs captured through Transmission Electron Microscopy (TEM) [Protesescu et al., 2015]

2.2.2 Optical Properties of CsPbX₃ PeQDs

The organic surfactants or the commonly called ligands play a crucial role during colloidal synthesis, which subsequently affect the optical as well as electronic properties of the quantum dots [Akkerman et al., 2018] [Pan et al., 2015] [Alivisatos, 1996]. In addition, the optical properties of perovskites are determined by the constituent halide ions as well as their size [Shi et al., 2017].

Absorption, photoluminescence (PL), qauntum yield and size tunable emission wavelength are the most remarkable optical properties of colloidal QDs [Shi et al., 2017].

Photoluminescence and Absorption Properties

Contrary to conventional chalcogenide quantum dots, photoluminescence emission spectra and absorption spectra of LHPs are tunable by size through quantum confinement effect as well as by their halide composition that includes mixed halide compositions covering the entire visible region of spectra (400-700 nm) [Protesescu et al., 2015].

A couple of unique features of LHPs contribute to the enhanced PL properties, such as their excitonic properties and defect tolerant behaviour [Yang and Zhong, 2016] [Akker- man et al., 2018] [Shi et al., 2017].

(30)

Figure 8.Colloidal CsPbX₃perovskite nanocrystals exhibit size- and composition-tunability.(a) Colloidal perovskite nanocrystals in solution with toluene. (b) Photoluminescence spectra for different halide compositions(c)Absorption and photoluminescence spectra for different halide compositions. (d)Quantum size-dependent absorption spectra for CsPbBr₃ NCs in the size range of 4 - 15nm [Protesescu et al., 2015]

Binding energy of exciton or the localization of exciton is a crucial parameter that charac- terizes the recombination dynamics of photo-generated charges [Yang and Zhong, 2016]

[Shi et al., 2017]. Efficient separation of photocarriers is a prerequisite for photovoltaics, whereas high recombination is a prerequisite for emissive materials [Shi et al., 2017].

Free carriers (electrons and holes) are dominant in perovskites, and they recombine emitting photons equivalent to the respective bandgap energy of the perovskites [Shi et al., 2017] [Yang and Zhong, 2016]. Temperature-dependent PL measurements have been conducted by many researchers for both bulk and QD semiconductors as shown in figure 9. It showed that the exciton binding energy for the bulk is 65 meV whereas 375 meV for the QDs, which confirms the origination of PL from exciton recombination ascribed by the dominating excitonic stability with respect to thermal stability [Shi et al., 2017] [Yang and Zhong, 2016].

Figure 9. PL spectra dependent on temperature for(a)QDs (273-393 K).(b)Bulk material (50- 300 K) [Yang and Zhong, 2016]

(31)

Another prerequisite for highly luminesce photovoltaic material is a minimal concentration of crystalline defects and impurities that tend to act as electronic traps or dopants, influencing the exciton recombination as well as other optical properties [Akkerman et al., 2018]. In conventional semiconductor quantum dots, cadmium chalcogenides (CdSe) for instance, the bandgap exists intermediate to bonding (σ) and antibonding (σ*) orbitals as shown in figure 10, thus defects or dangling bonds are present within the bandgap [Akker- man et al., 2018] [Huang et al., 2017]. Bandgap in LHPs however exists between two antibonding orbitals and defects are consequently enclosed within the energy bands, which minimizes the impact on radiative recombination unlike in cadmium chalcogenides, thus making them more luminesce [Akkerman et al., 2018] [Huang et al., 2017].

Figure 10.Defect tolerant beahviour of conventional semiconductors and Lead Halide Perovskites [Akkerman et al., 2018]

Photoluminescence Quantum Yield, PLQY

Quantum Yield shows the effectiveness of a fluorophore as an emitter and is defined as in equation 2.

QY_{P L} = EmittedP hotons

AbsorbedP hotons (2)

Surface properties are crucial in addition to the exciton binding energy to obtain highly luminescence QDs. A defect-free or a well passivated QD is expected to have QYs close to uniformity [Hollingsworth, 2006] [Akkerman et al., 2018] [Pan et al., 2015] [Yang and Zhong, 2016]. LHPs typically have the highest QYs of up to 99+% among all types of QDs discovered over the years. The LHPs are not entirely defect free, but the halogen rich surface and the well passivated surface defects cause excited electrons to be confined leading to enhanced photoluminescence Quantum Yields [Akkerman et al., 2018] [Pan et al., 2015] [Yang and Zhong, 2016].

(32)

These intuitions explain the observation of high-photoluminescence and quantum yields for LHPs in comparison to previously used semiconducting nanocrystals.

2.2.3 Applications of CsPbX₃PeQDs

The high-photoluminescence, quantum yields, size as well as compositional tunable emission wavelength of LHPs as discussed in section 3.5.2 has made them ideal for high performance, low cost and light weight optoelectronic device applications. [Kagan et al., 2016] [Yang and Zhong, 2016] [Shi et al., 2017] [Kim et al., 2013].

Figure 11. Colloidal LHP optoelectronic device architectures. (a)LHP quantum dots embedded in PL down conversion such as LCD dispalys and lighting(b)LEDs; electroluminescent devices (c)Solar cells [Akkerman et al., 2018] [Kagan et al., 2016]

It has been reported by many researchers, that the wide colour gamut PL of CsPbX₃ nano-crystals meet the existing colour standards for display applications [Akkerman et al., 2018].

A wider colour gamut was shown by CsPbX₃ QDs synthesized by Protesescu and co- workers (figure 12) in comparison to the common colour standards for LCD and NTSC TVs, that demonstrates the enhanced probability of using all-inorganic perovskite QDs in LCD display applications [Protesescu et al., 2015] [Yang and Zhong, 2016]. Song and co-workers were the first to experiment LEDs based on LHP nanocrystals, where orange, green, and blue LEDs were reported to have luminescences of 528, 946, and 742 cdm⁻², with external quantum efficiencies (EQE) of 0.09, 0.12, and 0.07%, respectively [Shi et al., 2017] [Song et al., 2015]. EQE is a performance metric, that refers to the ratio of

(33)

Figure 12. CIE color coordinates corresponding to the CsPbX3 QDs compared to most common colour standards [Protesescu et al., 2015] [Yang and Zhong, 2016]

rate of emission of photons into the final medium and rate of injection of current expressed in electrons [Kagan et al., 2016]. Zhang and co-workers developed inorganic perovskite quantum dots, which showed quantum yields of 70%, 95%, 80%for blue, green, and red emissions respectively [Li et al., 2016b]. Luther and co-workers were the first to discover all-inorganic perovskite (with CsPbI3 QDs) solar cells. The CsPbI3 QD films embedded in a solar cell, exhibited power conversions of 10.77%at a remarkably high open circuit voltage of 1.23 volts, making them promising for light harvesting applications or for LEDs [Luther et al., 2016]. Absorption characteristics were maintained even after a long storage time period of 60 days in ambient conditions [Luther et al., 2016]. The great potential of QDs in various applications attract them as the next generation of optoelectronics.

2.2.4 Environmental impact on photophysics of CsPbX₃PeQDs

Despite the potential applications of QDs, the practical applications are restricted by two major issues; their stability or structural lability in moisture, light, heat and oxygen and anion exchange between different halide compositions [Chen et al., 2016] [Akkerman et al., 2015] [Akkerman et al., 2018].

Structural Lability

LHPs are highly ionic compounds and this ionic nature has made the synthesis of LHPs a straightforward process, as they readily arrange into forming crystalline nanocrystals

(34)

[Huang et al., 2017]. However, this also means that they tend to lose their colloidal and structural integrity easily, when they are isolated, purified and handled [Akkerman et al., 2018].

Figure 13. Scheamtic of ligand desorption of LHPs that leads to losing the colloidal stability [Akkerman et al., 2018]

The long alkyl chain ligands are ionically bonded to the surface, which tend to quickly desorb as shown in figure 13 and conditions such as moisture, light, oxygen and heat stim- ulate this desorption [Akkerman et al., 2018]. Chen and co-workers carried out a research to study the stability and photo-degradation of CsPbBr₃QDs suspended in toluene [Chen et al., 2016]. Stability tests of CsPbBr₃ QDs were carried out at different temperature, atmosphere and light irradiation conditions. Light irradiation was found to be the most influential towards degradation of QDs, where PL shows a drastic quenching in intensity along with a red shift in the emission peaks. PLQY showed a decrease of up to 40%. The effect of light towards the QD degradation is further explained by two different phenomena;

• Thermal-degradation: Heat is generated in QDs in case of non-radiative recombination (charge carriers recombine without releasing photons in case of using light having photon energy lower than the energy associated with bandgap) of charge carriers or the vibration excitation caused by the incident photon. Perovskite QDs are liable to undergo structural change, as heat dissipation is quite significant that leads to more trap states, which further quenches PL.

• Photo-degradation: Cubic structure of QDs changed to elongated nano-rods due to aggregation upon light irradiation.

To overcome these instability issues, Chen and co-workers proposed a surface passivation technique with tightly bonded agents such as SiO2 or other polymer layers.

(35)

Anion Exchange

Perovskites with different halide compositions tend to exchange anions when they are in solution or when they come into close proximity (figure 14)(a). This is beneficial as it results in NCs that have mixed properties to the parent NCs, as well as this gives access to a narrow PL spectra intermediate to those of the original parent nanocrystals as shown in figure 14(b) [Akkerman et al., 2015].

Figure 14. (a)Anion exchange between different different halide perovskites(b) PL spectra of mixed halide perovskite NCs (CsPb(Br:X)3 (X = Cl, I)). Synthesised from CsPbBr3 NCs using anion exchange [Akkerman et al., 2015].

Akkerman and co-workers reported a method capable of enhancing the optical properties of LHPs using different halide compositions that exchanged anions, while maintaining the original crystal shape and size [Akkerman et al., 2015].

However, if devices are to be designed with multi-layers of different halide compositions as shown in 15, anion exchange of perovskites becomes a limitation.

(36)

Figure 15.Schematic of a multi-layer device application using different LHPs

2.3 Atomic Layer Deposition (ALD)

Atomic layer deposition, also known as atomic layer epitaxy has gained enormous popularity as a thin film deposition method, that was first developed by Suntola and co-workers in 1970s for depositing ZnS for display devices [Palmstrom et al., 2015] [Bakke et al., 2011] [Miikkulainen et al., 2013]. It is a gas phase deposition technique that results in uniform films on nanostructures [Yin et al., 2016].

With regards to the issues of PeQDs mentioned in section 2.2.4, that leads to a degradation or limitation to device performance, developing effective encapsulation methods are crucial and ALD technique is ideally suited for addressing these challenges through surface and interface modification as well as functioning as a passivation layer [Palmstrom et al., 2015] [Ip et al., 2013] [Bakke et al., 2011].

This deposition technique is self-limiting in nature, where rate of adsorption of precursor reaches zero upon formation of a self-saturating surface monolayer [Palmstrom et al., 2015] [Bakke et al., 2011]. Introducing additional precursor or increasing the precursor exposure time thus have no longer any effect on the amount of precursor adsorption.

Figure 16 illustrates the general mechanism of the ALD technique. Operational temperature of the ALD process has a pivotal role and generally a moderate temperature is preferred that provides sufficient thermal energy for a complete reaction between the precursor and the co-reactant [Palmstrom et al., 2015]. Higher and lower temperatures lead to either higher or lower growth rates per cycle than the anticipated [Palmstrom et al., 2015] [Valdesueiro et al., 2016b].

(37)

Figure 16. Schematic of the ALD process. (a)Precursor introduction. (b)Purge using an inert gas for the removal of excess reactants. (c)Introducing the co-reactant. (d)Excess co-reactants and by-products removed by a second purge [Palmstrom et al., 2015].

Al2O3 is a widely used deposition material grown by ALD [Groner et al., 2004]. The initial cycles of ALD process predominantly results in infilling, where alumina grows between the QDs until the voids between the QDs are filled. This is then followed by over- coating of the films above QD film by the additional cycles [Valdesueiro et al., 2016b].

This is represented in figure 17.

Figure 17.Schematic representation of Al₂O₃deposition on QD films

Liu and co-workers reported ALD alumina treatment on PdSe nanocrystal solar cells that achieved improvements in both efficiency as well as stability. They reported 95% efficiency even after a month of storage in air for devices made of PdSe nanocrystals with alumina infilled and overcoated. Unprotected devices were shown to degrade about 30%

of their original efficiency within hours of exposure to air [Liu et al., 2011].

(38)

3 MATERIALS AND METHODS

3.1 Chemicals and Other Materials

All chemicals required for the CsPbBr₃synthesis were purchased from Sigma Aldrich and used without purification, unless otherwise noted. Cesium carbonate (Cs2CO3, 99.9%), lead (II) bromide (PbBr2, 99.99%, Alfa Aesar), oleic acid (OLAC, technical grade 90%), oleylamine (OLAM, technical grade 70%), 1-octadecene (ODE, technical grade 90%), octane (anhydrous, >99%) toluene-d8 (99.95%) and hexane (99%) have been used during the synthesis.

3-mercaptopropyl trimethoxysilane (95%) was supplied by Sigma Aldrich to be used for the surface treatment of silicon wafers.

3.2 CsPbBr

3

PeQDs Synthesis

The CsPbBr₃ QDs have been synthesized in accordance with a process developed by Protesescuet al.. This synthesis requires reacting precursors in solvents in the presence of surfactants. Cs₂CO₃ and PbBr₂ have been used as precursors, whereas oleic acid (OLAC) and oleylamine (OLAM) have been used as surfactants. The synthesis involves two steps as described below.

Synthesis of Cs-oleate precursor

Cs₂CO₃ (0.8 g), OLAC (2.5 ml) and octadecene, ODE (80 ml) were added to a 100 ml three-necked round bottom flask followed by stirring under vacuum for 1 hour at120^◦C.

The flask was then purged with N2 for 10 min and then switched back to vacuum. This process of switching to vacuum and N2 alternately, was carried out thrice for the removal of moisture as well as oxygen. The temperature was raised to150^◦C for the reaction to proceed further, until a clear solution was obtained that indicates the complete consumption of Cs₂CO₃with OLAC. The prepared solution of Cs-oleate in ODE was stored in N₂, until it was used for the QD synthesis in the next step.

Synthesis of CsPbBr₃PeQDs

CsPbBr₃ QDs synthesis was carried out as reported by Protesescuet al.using a modified hot injection method. At a temperature of 120^◦C, 0.42 g of PbBr₂ and 30 ml of ODE were mixed in a 100 ml three-necked round bottom flask followed by degassing under