Morphological Studies of Organometal Halide Thin Films for Perovskite Solar Cells

Transcription

1 Morphological Studies of Organometal Halide Thin Films for Perovskite Solar Cells by Donghan Chen A thesis submitted in conformity with the requirements for the degree of Master of Applied Science Department of Materials Science and Engineering University of Toronto Copyright by Donghan Chen 2014

2 Morphological Studies of Organometal Halide Thin Films for Perovskite Solar Cells Abstract Donghan Chen Master of Applied Science Department of Materials Science and Engineering University of Toronto 2014 Thin film solar cells are important for making photovoltaic technologies affordable and easily fabricated. The main objective of research into thin film solar cells is to achieve highenergy convert efficiency with easy manufacturing methods and accessible elements. Thin film solar cells based on organometal halide perovskites have demonstrated outstanding efficiency among photovoltaics. The morphologies of solution-processed perovskite thin films, such as uniformity of thickness and surface coverage, have been shown to be important factors for device performance. Perovskite thin films were prepared with uniform thickness and full coverage by using vapour deposition methods. The morphology features and crystal quality of the perovskite thin films were examined to optimize the preparation conditions. Influence of the annealing temperatures was investigated to shed light on the stability of perovskite films in postpreparation treatment. Atomic force microscopy (AFM) was used to capture high-resolution morphology changes during annealing treatment and film formation. ii

3 Acknowledgements First and foremost, I would like to express my gratitude to my supervisor, Dr. M. Cynthia Goh, for her continuous guidance and patience through this journey. This thesis would not be possible without her encouragement and consideration. Also, I would like to thank Dr. Zheng- Hong Lu for his generous advices and help on conducting the experiments in this study. I have been very much honored to work with Professor Goh and Professor Lu. I am very grateful to my colleagues in Goh group who I have worked with for last two years. Special thanks go to Dr. Richard Loo, Dr. Jane Goh, Dr. Cheng Lu and Dr. Stanley Wong for their advices and discussions in research. And it is my honor to have Dr. Alon Eisenstein, Nari Kim, Calvin Cheng and Dr. Zhe She as my lab mates. I also appreciate all the help provided by Emmanuel Thibau and Robin White from Lu group. It was great pleasure to work with them. Last but not least, I would like to thank my friends and family for their supports and love. My parents, my brother and my dear wife have always been there for me. There are no words to express my truly gratitude for their love and support. iii

4 Table of Contents Acknowledgements... iii Table of Contents... iv List of Figures... vi List of Abbreviations... ix Chapter 1 Introduction Introduction Perovskites for High Performance Solar Cells Crystal Structure of Perovskites Perovskite Distortions Dye Sensitized Solar Cells (DSSCs) and Perovskite Solar Cells Device Structure of DSSC Working Principles Evolution from DSSCs to Perovskite Solar Cells Achievements in Morphological Studies of Perovskite Thin Film Motivations and Thesis Overview Chapter 2 Materials and Experimental Methods Materials Methods for Perovskite Thin Film Preparation Solution Processed Technique Vapour Deposition Technique Characterization Powder X-Ray Diffraction (XRD) Scanning Electron Microscopy (SEM) Atomic Force Microscopy (AFM) iv

5 Chapter 3 Results: Morphological Studies of Perovskite Thin Films Introduction Experimental Details Results and Discussion Characterization of Perovskite and Related Materials Annealing Effects for Perovskite Thin Films The Effects of Annealing Temperature on Perovskite Thin Films Degradation Study of the Perovskite Thin Films (in Air) High Resolution Study of Perovskite Thin Film with AFM Summary Chapter 4 Conclusions and Future Work References: v

6 List of Figures Figure 1.1 Comparison of the efficiency of several third-generation photovoltaic technologies [3]... 2 Figure 1.2 Illustration of ideal ABX 3 cubic perovskite crystal structure Figure 1.3 Illustrative sketches of possible distortions in perovskite structure: (a) ideal cubic structure of perovskite SrTiO 3 ; (b) rotated octahedra structure of perovskite GdFeO 3 (Side view); (c) rotated octahedra structure of perovskite BaNiO 3 ; (d) Jahn-Teller distorted octahedra structure of perovskite LaMnO Figure 1.4 Illustrative sketch of a typical DSSC structure Figure 1.5 Sketch of a perovskite solar cell with a simplified planar structure Figure 1.6 Thin-film topology characterization: (a-d) SEM images of top-view and cross sectional view of perovskite thin film and perovskite solar cells made by solution processed technique and vapour deposition technique. (e-f) SEM images of large cross sectional view images of perovskite make by solution processed technique and vapour deposition technique [11] Figure 2.1 Illustrative sketches of organic chamber used for vapor deposition. (a) Side view of the organic chamber (b) top view of the organic chamber. The label of each part a h is described in page Figure 2.2 Illustrative sketch of vapour deposition process Figure 2.3 (a) Illustrative diagram of Bragg s Law; (b) Principle of the XRD measurement [46] Figure 2.4 Scheme of a typical AFM system Figure 2.5 The dependence of force on the probe distance vi

7 Figure 3.1 Photos of 100-nm thin film perovskite samples made by the vapour deposition technique: right sample, a typical failure deposited sample film on glass; left sample, a typical film with deposited with organometal halide perovskite Figure 3.2 XRD patterns: (a) Pure perovskite thin film of CH 3 NH 3 PbI 3 ; (b) PbI 2 powder used for film preparation; (c) PbCl 2 powder used for film preparation Figure 3.3 XRD patterns of (a) perovskite sample before annealing treatment and (b) perovskite ample after annealing treatment in 140 ºC for 20 minutes Figure 3.4 XRD patterns of perovskite thin films annealed at temperature ranging from 100 ºC to 400 ºC. Peaks labeled with (*) are characteristic peaks of perovskite; peaks labeled with ( ) are PbI 2 peaks; peaks label with ( ) are PbCl 2 peaks Figure 3.5 SEM micrographs of perovskite samples: (a) Sample without annealing treatment; (b) Sample annealed at 100 ºC Figure 3.6 SEM micrographs of perovskite samples: (c) Sample annealed at 140 ºC; (d) Sample annealed at 180 ºC Figure 3.7 SEM micrographs of perovskite samples: (e) Sample annealed at 260 ºC; (f) Sample annealed at 300 ºC Figure 3.8 Perovskite surface coverage as a function of annealing temperatures Figure 3.9 XRD patterns of perovskite thin films: (a) Perovskite film made by vapour deposition technique and annealed at 140 ºC; (b) The same sample from (a) exposure in air for 14 days Figure 3.10 SEM images of perovskite films: (a) Perovskite film made by vapour deposition technique and annealed at 140 ºC; (b) The same sample from (a) exposure in air for 14 days Figure 3.11 AFM images of perovskite films annealed at different temperature from 85 ºC to 140 ºC vii

8 Figure 3.12 AFM images of perovskite films with10 nm thickness (left) and 100 nm thickness (right) viii

9 List of Abbreviations AFM a-si CIGS CIS DSSC OPV PV PVD QCM SEM Atomic force microscopy Amorphous Silicon Copper indium gallium selenide Copper indium selenide Dye sensitized solar cells Organic photovoltaics Photovoltaic Physical Vapour Deposition Quantz crystal microbalance Scanning electron microscopy ix

10 1 Chapter 1 Introduction 1.1 Introduction For the last two decades, renewable energy has drawn great attention from research to investments. As the most easily accessible energy, solar energy can be used at the most place of the world. Therefore the use of solar energy would be an ideal solution for the increasing energy demands. Specifically, photovoltaics (PVs), or solar cells, are ideal solutions since they have many advantages. First of all, solar energy is an abundant energy form that does not rely on the geographic conditions like other renewable energy. Second, solar cell devices are usually easy to install and they require relatively less maintaining operation compared to others. Third, solar cells convert solar energy into electricity so they can directly supply energy for most of our needs today. Photovoltaic industry has been one of the fastest growing industries in renewable energy development. For effective photovoltaic production, it is very important to develop efficient and affordable techniques. Traditional silicon based photovoltaic techniques cannot decrease production costs due to the high price of materials and complex manufacturing requirements. Second-generation photovoltaics, based on thin film technologies, reduce the cost significantly by simplifying the device manufacturing procedures. Even though this generation s efficiency is not as high as silicon based solar cells, the ultra-low manufacturing costs make it a much more cost efficient technology. Therefore, thin film solar cells have become a rapidly growing, and increasingly important photovoltaic production type in industry [1, 2]. In terms of categorization, thin film solar cells can be classified into different types according to their photovoltaic materials, such as cadmium telluride solar cells (CdTe), copper indium gallium selenide solar cells (CIS or CIGS), and amorphous silicon (a-si) solar cells. Several relatively new types of solar cells have emerged in recent years, including quantum dots solar cells, dye sensitized solar cells (DSSC), organic photovoltaics (OPV, organic molecules or conjugated polymers), and perovskite solar cells, all of which can also be classified as thin film solar cells these are referred to as the third generation of solar cells.

11 2 Figure 1.1 Comparison of the efficiency of several third-generation photovoltaic technologies [3]. Despite being relatively recent, the popularity of perovskite solar cells has soared since its efficiency has increased dramatically in a very short time. Figure 1.1 compares the efficiency development of several new solar cell types (third generation solar cells). The power conversion efficiency (PCE) of perovskite solar cells reached 15% by the end of 2013, and 19% by 2014 [3], becoming the highest among third generation solar cells. Perovskites are considered to be great candidates for solar cell production because of they use inexpensive materials and the same manufacturing techniques used for other thin film solar cells [4].

12 3 1.2 Perovskites for High Performance Solar Cells Today, the term perovskite refers to crystal species with structures of ABX 3. However, the term initially only referred to CaTiO 3, and was discovered by German mineralogist Gustav Rose in the Ural Mountains in 1839, and named after Russian mineralogist Count Lev Perovskite ( ). Along with many other compounds found with the same ABX 3 crystal structure, Perovskite became one of most common class of minerals on earth. In 1926, Victor Goldschmidt described the crystal structure of perovskite for the first time, and illustrated its tolerance factors [5]. An accurate crystal structure was later published in 1945 using a X-ray diffraction study by Irish crystallographers [6]. Materials in the perovskites family already presented a wide range of applications, such as conductors, semiconductors, insulators, and even superconductors [7]. Other physical properties of perovskite also drew substantial interest, especially in magneto-resistance, ionic conductivity, and a multitude of dielectric properties [7-9]. One can achieve these properties, which are of great importance in microelectronics and telecommunications, by either keeping or modifying the ideal perovskite structure. Due to the flexibility of bond angles inherent in the perovskite structure, many different types of distortions can occur in the ideal structure, including tilting of the octahedra, displacements of the cations out of the centers of their coordination polyhedra, and distortions of the octahedra driven by electronic factors [9, 10]. A group of organic-metal halide perovskites has been found with ideal photovoltaic properties. Particularly in recent years, methylammonium lead halides (CH 3 NH 3 PbI 3 and CH 3 NH 3 PbI 3-X Cl X ) have shown extremely high power conversion efficiency. Within just two years, the device efficiency tested in laboratory increased from 8% to 19.3%, which is the highest among the third generation thin film solar cells [3]. The organometal halide perovskites have high charge carrier mobility and charge carrier lifetime, which allows light-generated electrons and holes to move far enough to be extracted as current, instead of creating heat and losing energy. The effective diffusion lengths of CH 3 NH 3 PbI 3 are several hundred for both electrons and holes [3]. For photovoltaic performance of CH 3 NH 3 PbI 3, open-circuit voltage (V OC ) can approach 1 V, while for CH 3 NH 3 PbI 3-x Cl x, V OC > 1.1 V has been reported [11]. Since the band gaps (E g ) for both materials are 1.55 ev, the ratios of V OC to E g are higher than what is

13 4 usually obtained from similar thin film cells. By tuning the band-gap of perovskite, V OC can reach up to 1.3V, which is the highest performing among thin film photovoltaic devices [3, 12]. Another attention-catching advantage of perovskite solar cells is that the device can be fabricated in simple planar structure, which drastically simplifies the production process. The most common spin-coating technique has been shown to be a sufficient method of efficiently preparing perovskite solar cells. On the other hand, the vapor deposition technique is also considered as a potential fabrication method that simplifies the production and enhances the degree of thin film quality control. For example, simple planar heterojunction perovskite solar cells have already been fabricated without complex nanostructures using the vapor deposition technique [14]. In addition, compared with other thin film solar cells, perovskite is free from rare elements requirement, which result in relatively low manufacturing costs. There remain two areas of concerns for perovskite use in solar cells. The first is the use of heavy metal (lead) in cell fabrication. Since other substitutions with low toxicity (such as tin [13]) can potentially be used, this may not be a major concern. Also, compared with the large amount of lead used in lead-related industries every year, the amount used in perovskite solar cells is relatively small. A second concern for perovskite solar cell application is the stability of the organic-inorganic hybrid materials. Since the hybrid materials are very sensitive, the perovskite films degrade quickly in exposure of ambient environment, and the cell durability is currently insufficient for commercial use [14]. Since the length of device s lifetime depends on the stability of the perovskite thin film, studies of the stability of perovskite are of great value for understanding the device s durability, and controlling the production conditions. This research focuses attention mainly on the investigations of stability in the perovskite materials excluded from the solar devices and photovoltaic performance. The preparation conditions of pure perovskites have been optimized and the material tolerances for annealing temperature and air exposure time have been studied. The discoveries in materials properties will benefit by controlling the production and post-treatment conditions. To better understand the fundamentals of perovskite, the following section introduces the perovskite crystal structure and its distortion features.

14 Crystal Structure of Perovskites A general perovskite structure in ideal cubic model is shown in Fig The crystalline architecture consists of three elements A, B and X. Each A atom sits at the central position of the unit cell coordinating to 12 X atoms, where X atoms are usually O 2, F, Cl or other large ions. B cations and X ions are coordinated and form [BX 6 ] octahedra. There are 8 [BX 6 ] octahedra in each unit cell. Therefore, A and X atoms are close-packed with B is occupying the centre of octahedra. In the organometal halide perovskites studied in this paper, A represents the methylammonium cation (Ch 3 NH + 3 ), while B stands for the metal cation (Pb 2+ ), and X is the halide ions (I or Cl ). In ideal perovskite structure, the unit cell axis, a, can be described by the ionic radii of A, B and X (r A, r B and r x ) by the following equation: a 2( r r ) 2( r r ). (1) A X B X Figure 1.2 Illustration of ideal ABX 3 cubic perovskite crystal structure.

15 Perovskite Distortions Because of the flexibility of bond angles inherent in the perovskite structure, many different types of distortions can occur from the ideal model [15, 16]. Only a few compounds, such as CaRbF 3 and SrTiO 3, have ideal perovskite structures, and even mineral perovskite CaTiO 3 itself is also distorted [16]. There are three major causes for the distortions in perovskite: 1) the size effect of ions and cations; 2) non-ideal stoichiometric deviation; 3) geometrical distortion caused by electron configuration, also known as Jahn-Teller effect [15]. First, the perovskite structure easily gets distorted by varying the sizes of A, B or X. To estimate the degree of distortion in a particular ionic perovskite, the Goldschmidt Tolerance Factor [5] has been defined as t : r r r r 1 A X t (2) 2 B X The ideal cubic perovskite structure has t = 1, due to the high symmetry in cubic system. The factor t gets smaller with the decrease in the cation A size. When the size of cation A dropped below a certain value, t will be smaller than 1. In this case the octahedra will tilt to make space for the large ions, such as the cases in CaTiO 3 and GdFeO 3 (t = 0.81). But t cannot be smaller than 0.81, since the structure will then be assigned to an ilmenite structure in that range. For a system with large A or small B ions than in ideal cubic system, the tolerance factor is larger than 1. The close-packing structure will be stable as a varied hexagonal perovskite, such as BaNiO 3. However, the tolerance factor assumes only ionic bonds existing in the structure so it is limited to distortion from the ideal perovskite structure. Second, the non-ideal stoichiometric deviation can cause distortion in perovskite structures. Taking SrFeO x as an example, the valency of the Fe can be different depending on either a sample heated in an oxidizing or a reducing environment, whereby the oxygen content varies from 2.5 to 3 in the perovskite structure. Both +3 and +4 oxidation states can be assigned to Fe in SrFeO 2.875: thus, the FeO 5 pyramid structure can be formed instead of octahedra. The original structure distortion takes place because of the deviations from ideal structure [16].

16 7 The third distortion type is Jahn-Teller effect, which is a geometrical distortion caused by a certain electron configuration in order to lower the structure s overall energy. Particularly, some perovskite systems hava Jahn-Teller active ions at B position [17]. For example, in the LnMnO 3 in which Ln is La, Pr or Nd structure, Jahn-Teller active Mn 3+ ions result in the elongation of the [MnO 6 ] octahedra [16]. All three types of distortions are illustrated as sketches in Fig 1.3. The distortions in perovskite structure have influences on the stability of properties. For organometal halide perovskites, their hybrid composition enlarges the flexibility of the crystal structure, which could be one of the causes of their low stability. Figure 1.3 Illustrative sketches of possible distortions in perovskite structure: (a) ideal cubic structure of perovskite SrTiO 3 ; (b) rotated octahedra structure of perovskite GdFeO 3 (Side view); (c) rotated octahedra structure of perovskite BaNiO 3 ; (d) Jahn-Teller distorted octahedra structure of perovskite LaMnO 3.

17 8 1.3 Dye Sensitized Solar Cells (DSSCs) and Perovskite Solar Cells As a new type of thin film photovoltaics, perovskite solar cells emerged from the previous generation, DSSCs [18]. It is important to understand the basics of thin film DSSCs and their connections with perovskite solar cells. Here, we simply demonstrated the typical device structure of DSSC and its work principle. Also, a historic evolution from the DSSCs to perovskite solar cells is reviewed in order to illustrate the development of perovskite application in solar cells, the present device designs, and possible future directions Device Structure of DSSC A typical DSSC is commonly built with a series of layers of materials serving for specific functionalities for transforming light into electrical energy. Figure 1.4 shows a typical DSSC structure [19]. The substrate that supports the whole cell structure is usually transparent glass. A layer of transparent conductive material is coated on glass as anode, which is usually indium tin oxide (ITO) or fluorine doped tin oxide (FTO). The transparency of the anode allows the light injecting into the device to provide the energy. On top of the anode material, n-type semiconductor, TiO 2 is typically built for electron transporting. Commonly, two layers of TiO2 are coated in different morphologies [20, 21]. The compact TiO 2 layer works as electron transport layer in solar cells, and the mesoporous TiO 2 layer is usually made into nano-scale structure to provide necessary morphology for dye sensitizer exhibiting the photovoltaic into its pores. Mixing with the organic dye molecules [22-26], quantum dots [27-29], or other light sensitizers [30, 31], the porous layer is the essential layer for absorbing light and generating electron-hole pairs in the device. On top of the TiO 2 there is a layer of p-type semiconductor, typically Spiro-OMeTAD, which transports holes from the photovoltaic material structure. A highly conductive cathode layer finalizes the cell preparation.

18 9 Figure 1.4 Illustrative sketch of a typical DSSC structure Working Principles Thin film solar cells are connected to conventional solar cells in their working principle. For traditional solar cells, both n-type and p-type materials are silicon based layers. The new generation of thin films solar cells, however, use thin film light absorbing materials. For traditional solar cells, the performance of photovoltaic cells depends on its core material, semiconductor, which performs as insulator in their pure form but is able to conduct electricity in high temperature or combine with other materials. A host semiconductor combined with electron donor materials develops an excess of free electrons, known as an n-type semiconductor. A host semiconductor combined with acceptor materials develops excess of holes (or equivalently the removal of electrons), known as a p-type semiconductor. A photovoltaic cell contains adjacent n-type and p-type materials, and the interface between is known as a P-N junction. Resulting from the nature of two types of semiconductor, a small number of electrons always move across the junction from the n-type to the p-type semiconductor, producing a small voltage output, even in dark. With the presence of light, a great number of electrons can be activated and flow across the junction thus creating an electric potential difference at each side. Photovoltaic devices are designed to apply this principle, and convert solar energy into electrical energy.

19 10 For the new generation of solar cells, which is the focus of this research, the process of electricity generation is different from P-N junction-based photovoltaics. After passing through the transparent glass substrate and anode, the sunlight striking onto the absorbing layer excites electrons from the materials valence band to the conduction band. Accordingly, the excitation of the electrons creates a free electron in conduction band and a hole in the valence band, which is referred to as excitons. After generating excitons, electrons attract into the TiO 2 layer and conducted into metallic anode. The hole will be replaced by an electron provided by the p-type material layer, and thus be conducted to the cathode. A potential difference, V OC, for an open circuit formed from anode to cathode and a current density J SC can be generated by continuous power supplied form light.

20 Evolution from DSSCs to Perovskite Solar Cells In thin film solar cells, a variety of organic dyes and inorganic quantum dots were used as photon sensitizers. Perovskite was also considered as a sensitizer in DSSCs but then discovered with electron transporting property. This section briefly introduces how perovskites become photovoltaic material that enables simple planar structured solar cells. Mitzi and co-workers discovered organic-inorganic halide perovskite as a possible candidate for thin film transistor and light-emitting diode (LEDs) early 1990s [4, 32]. Since then, properties of photovoltaic are anticipated but have not sufficiently studied due to concerns of lead toxicity and low stability [4]. Miyasaka [33] was the first to report the photovoltaic performance of perovskite in the nanoporous TiO 2 layer of dye-sensitized cells. The method used for their study was to spin coat the sensitized layer of perovskite with solution of CH 3 NH 3 I and PbI 2 on top of the TiO 2 film. The initial efficiency of their CH 3 NH 3 PbI 3 cell was 2.2%, and they were then able to increase efficiency to 3.8% by replacing bromine with iodine. This result was not a groundbreaking achievement in terms of efficiency; however, it demonstrated the potential for a new light absorbing material in photovoltaic cells. With the further investigation, perovskite was proven to be an ideal candidate of light sensitizer, which could achieve appreciable efficiency. Subsequently, Park [34] implemented perovskite in similar structure by depositing sparsely spaced hemispherical nanoparticles that were approximately 2.5 nm in diameter. Along with surface treatment on TiO2, they achieved an efficiency of 6.5% in 2011 [34]. The performance of perovskite became comparable to that of organic dyes at this time, but degraded rapidly since perovskite material can easily dissolve in its electrolyte cell. This lead Park, Gratzal and his co-works to consider replacing the electrolyte with a solid-state hole transport material, spiro-meotad (2,2,7,7 -tetrakis(n,n-di-p-methoxyphenylamine)-9,9 -spirobifluorene) [19, 35]. The perovskite material penetrates the nanoporous structure of the TiO 2 layers in all their structures, and is thus considered a sensitizing layer rather than a fully functioned photovoltaic layer. It not only improved the stability, as expected, but also it improved efficiency to 9.7% at 2012 [35].

21 12 Around the same time, Snaith and his workers [19] also reported success in using spiro- MeOTAD in perovskite structure, and developed photovoltaic cells in several ways. First, they employed a mixed-halide material CH 3 NH 3 PbI 3-x Cl x and improved both the stability and performance-efficiency of the cell, compared to the pure iodine equivalent. Furthermore, they capped the TiO 2 porous structure with a thin perovskite layer and the device was still functional. By replacing the TiO 2 layer with a similar but non-conducting Al 2 O 3 network, V OC of the device increased and efficiency boosted to 10.9%. This research demonstrated that perovskite has the potential to transport both electrons and holes between cell electrodes. The single-step perovskite deposition process created large morphological inconsistency, and low stability in device performance. With the method using both perovskite in TiO 2 scaffold structure and a capping layer of pure perovskite overlaying the scaffold, the efficiency jumped to 12.0% [36]. Same efficiency was observed in a similar structure with Br content in perovskite, and this structure was proven with high humidity stability [37]. There exists a structural transition from tetragonal to pseudo-cubic mainly due to a higher t factor, which is caused by the smaller ionic radius of Br. In 2013, Gratzel s group used TiO 2 scaffold along with two-step iodide deposition, and improved the efficiency up to 14%. Snaith s group creatively used a two-source deposition method, but avoided the previous scaffold structure, and achieved an increased efficiency of 15.4%. This method greatly enhanced the morphology by physical vapor deposition (PVD) and encouraged our work in this research. A similar structured solar cell with an efficiency of 19.3% was reported in May 2014 [38]. In summary of the evolution path of perovskite solar cells, perovskite was initially used as an organic-inorganic hybrid dye in DSSC and also as a solid state DSSC (ssdssc). It was substituted for the organic dye molecules, and acted as active material extremely thin absorbers (ETAs) in DSSC [39]. The discovery of perovskite s ability to transport charge carriers enabled it to act as the photovoltaic materials, without depending on the TiO 2 electron transporting materials. Perovskite meso-superstructured solar cells (MSSC) exhibited a high photovoltaic efficiency on insulation AlO 3 [40]. Therefore, the possible future directions of perovskite solar cells could be MSSC, simple planar p-i-n thin film solar cell, or p-n heterojunction cells [4, 41, 42]. Figure 1.5 shows the perovskite solar cell with a simplified planar structure, which is a p-in thin film solar cell.

22 Figure 1.5 Sketch of a perovskite solar cell with simplified planar structure. 13

23 Achievements in Morphological Studies of Perovskite Thin Film The most progressive morphological study in surface uniformity and consistency is the co-vapour deposited perovskite thin film, conducted by Snaith s group in 2013 [43]. This study discovered significant differences regarding film morphology between solution-processed and vapour deposited perovskite films. In this work, Snaith s research team achieved a 5% increase in efficiency by removing scaffold structure, and simply building the solar cell structure with a planar heterojunction thin film. The vapour deposition method produced extremely homogeneous perovskite film with highly uniform thickness[44], which provided contrasts to the inconsistent thickness and uneven morphology of the solution-processed films (shown in Fig. 1.6 [43]). Figure 1.6 shows SEM images comparing perovskite solar cells prepared by vapour deposition and solution processed techniques. The top row photos show that vapour deposition technique achieved full surface coverage of the film, and the solution processed material unevenly distributed. The vapour deposited sample is fully covered with perovskite materials and no vacancies or voids could be observed, whereas the solution processed sample showed a significant portion of area with substrate exposure. The cross-sectional figure also indicates that some part of the solution processed film may suffer from short circuits by the varying thickness from 0 nm to 410 nm, which hinders the performance of the device because of the pinhole formation. But the vapour deposited film with evenly covered layer of perovskite material is of no concern in this problem. This morphological study stressed the advantage of the vapour deposition, in that this method allows precise control of the thickness of the perovskite film. It is important to optimize the film thickness since a thicker film is needed to absorb enough light for generating excitons. However, one needs to maintain the thickness at a thin enough level to allow for the transport of electrons and holes. Last, optimized perovskite with film thickness of 330 nm indicates that electron-hole diffusion length exceeds such a length in perovskite material. The investigation of stability of the perovskite films is also discussed in this report regarding solution processed perovskite thin film [44]. Using a scanning electron microscope

24 15 (SEM) is the proper technique for examining the morphology change of the perovskites surface. Particularly, the surface coverage rate is an important parameter that can be used to describe the film s formation, or the surface loss in post-treatment. SEM has thus been used in this research, and changes of surface morphology and coverage are used to evaluate the degree of degradation. Figure 1.6 Thin-film topology characterization: (a-d) SEM images of top-view and cross sectional view of perovskite thin film and perovskite solar cells made by solution processed technique and vapour deposition technique. (e-f) SEM images of large cross sectional view images of perovskite make by solution processed technique and vapour deposition technique [11].

25 Motivations and Thesis Overview Thin film solar cell technologies are important in making photovoltaics more affordable. One goal of studies in thin film solar cells is to achieve of high efficiency devices with easy processing methods and accessible element resources. Organometal halide perovskites based solar cells have been demonstrated to have significantly high efficiency. The morphology conditions of solution-processed perovskite, such as uniformity of thickness and surface coverage, are key factors to control film quality and device performance. Using the vapour deposition method in high vacuum system, we prepared perovskite thin films with uniform thickness and full surface coverage. In this study, we aimed to optimize the vapor deposition conditions for preparing perovskite thin films that have been used in a number of solar cells. Perovskite samples were prepared and annealed under a series of temperatures. By checking the crystal structures with X- ray diffraction (XRD), and surface morphologies with scanning electron microscope (SEM), the effect of different annealing temperatures was investigated. The stability of the perovskite thin film was studied by examining the changes of the material exposing in air. AFM was used to check high-resolution surface information that cannot be captured by SEM. In sum, this chapter introduced the fundamentals of perovskite structures, and provided a brief recount of the evolution from DSSC to perovskite solar cells. This chapter also reviewed a study of morphological control of the thin film perovskite. Next, chapter 2 illustrates the materials used in preparing the perovskite thin film, as well as the experimental details of our study. Chapter 3 shows the experimental results and the conclusions of our investigation. Last, Chapter 4 is a brief summary of the entire thesis and several suggestions on future research.

26 17 Chapter 2 Materials and Experimental Methods 2.1 Materials Our organometal halide perovskite is made from organic and inorganic sources through the vapour deposition technique. The organic source is methylammonium iodide (CH 3 NH 3 I), and the inorganic source is lead chloride (PbCl 2 ). CH 3 NH 3 I was synthesized by reacting methylamine with hydroiodic acid at 0 C for two hours, while stirring. PbCl 2 was purchased from sigma- Aldrich. This chapter will describe materials and experimental methods, including both solution processing and vapour deposition. Principles of characterization method used in this study will also be introduced. 2.2 Methods for Perovskite Thin Film Preparation There are two main techniques for preparing the perovskite thin films, solution processed technique and vapour deposition. The solution processed method is more commonly used because of the simplicity of equipment set-up. Vapour-deposition requires complex experimental conditions, but it pays off by providing the thin film with highly uniformed thickness. In this study, we mainly focused on perovskite thin films that were made by vapour deposition in high vacuum system Solution Processed Technique For perovskite thin film preparation, the most commonly used technique is spin coating the precursor solution on substrate, followed with an annealing treatment. In early studies of thin film perovskite solar cells, perovskite precursor solution was prepared and spin coated directly onto substrate to form thin films [19]. The precursor solution is usually the methylammonium lead halide (CH3NH3PbI2Cl or CH3NH3PbI3),) in N, N-dimethylformaide. Spin coating is usually conducted at ambient conditions, and the annealing procedure is carried out on sample immediately after spin coating. Before annealing, the thin film has lower crystallinity (smaller crystal size) and contains an excess of PbI 2 component mixed in the thin film. With the annealing process, the perovskite film forms larger crystal grains, and pure perovskite crystal structures [3]. Some other methods, based on the solution processed technique, were developed by varying part of the original procedures in order to obtain perovskite thin films more easily, or

27 18 with higher performance. A sequential solution deposition method was carried out with high efficiency in device performance, based on the previous study [14]. Instead of spin coating perovskite solution directly onto the substrate, the inorganic component PbI 2 solution was first spin coated, and then the substrate was dipped into a CH 3 NH 3 I solution. Further modification was made to introduce the organic component by vapour treatment on the substrate, spin coated with an inorganic source [11]. CH 3 NH 3 I powder was then spread out around the PbI 2 coated substrates in a covered petri dish. By heating the petri dish for desired time, perovskite formed from the CH 3 NH 3 I vapor-treated PbI2 substrate. Although preparation methods seem to differ from each other in details, all have been proven to be effective ways for perovskite film preparation. High performance devices with efficiency higher than 12% can be obtained by all the methods discussed above Vapour Deposition Technique A novel method is to use the vapour deposition technique to produce highly pure perovskite thin films with high device efficiency. For the vapour deposition technique, thin films are prepared onto substrates through condensation, or the reaction of vaporized materials. The preparation procedure is conducted in a multi-technique vacuum system, which has a centraldistribution chamber (CDC) with several sub-chambers located around it. All chambers are connected and kept in a vacuum at a pressure between to Torr. Perovskite thin films were deposited in an organic deposition chamber, which allows for relatively low temperature deposition compared to the oxides deposition chamber and the metal deposition chamber. In the organic deposition chamber, chemical sources are evaporated from the permanently mounted Knudsen cells (K-cells) whose heating temperature can be precisely controlled. The two source materials for perovskite preparation are vaporized in the vacuum system as the gas phase. They interact on the surface of substrate located on the top of K-cells, and where the chemical reaction takes place forming the thin film. The illustrative diagrams of organic deposition chamber are shown in Fig Each part of the chamber is labeled with a letter. Part (a) is the chamber body where the evaporation deposition takes place. Part (b) is a transfer-arm evaporator cell (TAE-cell) that allows substrate moving into and out of the chamber body. A sample holder, part (c), is connected at the end of the transferring arm and it can carry four pieces of substrate for each deposition. Substrates

28 19 prepared for perovskite deposition are placed at the sample holder so that the substrate can be delivered into the chamber. Part (d) is a quartz crystal microbalance (QCM) used for monitoring the films thickness. Part (e) is the shutter that controls the exposure of substrate to the chemical vapour. Part (f) is the connector to the central chamber, while part (g) is the K-cell. These K- cells are isolated from one another and the temperature of K-calls can be individually controlled. Each K-cell has its own shutter that controls the evaporation for coming out from the K-cell. Part (h) is the ion pump used to maintain the required vacuum. The lower panel of Fig. 2.1b illustrates the top view of the organic chamber. Figure 2.2 describes the vapour deposition process for perovskite thin film preparation. During the deposition process, organic source CH 3 NH 3 I and inorganic source PbCl 2 are placed in different K-cells. Once the K-cells are heated to the preset temperature, the chemicals start evaporating and moving upward to the substrate. When both PbCl 2 and CH 3 NH 3 I vapor reaches onto the substrate, a chemical reaction is initiated, and begins to form perovskite films on the substrate s surface. At the same height level of the substrate holder, there is a QCM device serving as the monitor for film thickness. QCM measures the mass per unit area by measuring the change in frequency of a quartz crystal resonator. The resonance changes with increases of a small mass due to the film deposition at the surface of the acoustic resonator. Film thickness can be monitored based on the mass addition on unit area. Usually, there is some difference from the real deposited thickness to the QCM readings. The ratio of the real deposition rate over the QCM reading is defined as the tooling factor of the QCM. Each QCM has a specific tooling factor in a certain deposition system. During the process, the system is maintained at pressure of Torr. As such, only chemical sources evaporated upon the film participate in the chemical reaction, which ensures the purity of the reagents and uniformity of the film thickness. For solution processed perovskite, it is known that the annealing procedure assists film formation and material crystallization[14, 34]. Most of the films after spin coating need an annealing process for better crystallization and film formation. To prepare an ideal perovskite thin film with solution processed method, it is very important to control the annealing temperature and treatment time. For vapour deposited perovskite, however, there is little research that addresses the annealing effect from previous studies. This dearth of information provided motivation for us to conduct an investigation on the annealing effects for perovskite over vapour deposited thin films.

29 Figure 2.1 Illustrative sketches of organic chamber used for vapor deposition: (a) Side view of the organic chamber (b) top view of the organic chamber. The label of each part a h is described in page

30 Figure 2.2 Illustrative sketch of vapour deposition process. 21

31 Characterization Powder X-Ray Diffraction (XRD) XRD, a traditional technique for crystal characterization, examines long-range ordering and phase purity. The periodicity of the electron density in a crystal structure generates X-ray scattering from electrons, and causes coherent diffraction pattern. The intensity of the diffracted X-ray can be plotted as the function of angle 2θ (θ is the angle of electron and the crystal plane) and thus depict the powder diffraction pattern. Peaks appear in the powder diffraction pattern at angles with maximum interference that satisfy Bragg s Law [45]: n 2d sin (5) Each crystal structure has a unique diffraction pattern with characteristic peaks and relative intensity. The XRD patterns of known structures has been tested and archived in databases. Therefore, one can compare the experimental XRD patterns with the standard patterns in order to identify the sample structure. In this paper, all X-ray diffraction patterns of the perovskite thin films were obtained on an X-ray diffractometer (Panalytical X Pert Pro), with Cu-Kα radiation (λ= å). Figure 2.3 shows an illustrative diagram of Bragg s Law, as well as sketches of the working principle of a XRD measurement. To verify the crystal structure of our perovskite thin film samples, we referred to the database and labeled the theoretical peak positions of perovskite structure. We can thus compare our XRD patterns with standard XRD patterns, and check if the thin film prepared by vapour deposition also has perovskite crystal structure. To investigate the effect of other impurity crystal structure, XRD patterns of possible lead halide (PbI 2 and PbCl 2 ) are also obtained. As noted, XRD is an ideal tool for checking the crystal structure evolution in annealing process; results of XRD tests provide strong support for the investigation of our thin film and thin film changes.

32 Figure 2.3 (a) Illustrative diagram of Bragg s Law; (b) Principle of the XRD measurement [46]. 23

33 Scanning Electron Microscopy (SEM) SEM is an imaging tool that uses a beam of focused electrons, scanning on sample surfaces, to produce images. The scanning electrons interact with atoms on sample surface producing signals of surface topography. There are several types of signals produced by SEM, such as secondary electrons (SE), back-scattered electrons (BSE), characteristic X-rays, cathodeluminescence (CL), specimen current and transmitted electrons. Secondary electrons are the most commonly used signal type. During the measurement, the electron beam scans over the surface of the sample dot-by-dot and line-by-line, producing many types of signals. Secondary electrons that are dislodged from the surface atoms have unique patterns at each dot. A detector that counts the secondary electrons scattered from the sample surface receives information of secondary electrons of the sample. Other sensors also detect BSE, X-ray, and other signals. By detecting the information along with the scanning beam by different sensors, a great deal of information is processed by the computer, and displayed by different levels of brightness on a monitor. SEM images were collected at the Centre for Nanostructure Imaging at Chemistry Department of University of Toronto using the Quanta FEG 250 environment SEM with both a bright field and dark field detector. The crossbeam combines a high resolution SEM for imaging, with a focused ion beam (FIB) for micromachining by sputter milling with a sub-100 nm lateral resolution. The spatial resolution of the images can focus down to 1 nm, depending on the material s conductivity. SEM is a powerful imaging tool that can be used in conductive and semi-conductive materials. For our perovskite sample, it is the proper choice with which to image surface topography. By detecting the surface of perovskite sample, we are able to understand the surface uniformity, thickness, and even chemical components in a rough scale. In addition, the change of the surface morphology observed from SEM benefits to explain the possible material affect.

34 Atomic Force Microscopy (AFM) Distinguished from its predecessor Scanning Tunneling Microscopy (STM), which relies on tunneling current between scanning tip and sample surface atoms AFM mechanically contacts with materials directly. Therefore, AFM affords the possibility of observing on almost any type of materials. Working principle of AFM is also relatively simple among advanced characterization techniques. Typically, AFM contains three main parts, as shown in Figure 2.4: the scanning system, controller, and computer. Samples are located on a piezoelectric scanner that provides precise movement on x, y and z direction. A very sharp tip attached to a soft cantilever scans on the sample surface line-by-line with the motion of the piezoelectric scanner. During scanning, the controller adjusts the z canner to maintain a constant tip-sample force value to make sure the tip-sample distance has a constant value. At the backside of the tip, a beam comes out of the laser source and reflects to a photodiode detector. With the monitoring of the laser beam signal, height change and deflection information of tips can be precisely captured by the controller. Thus, morphology images can be constructed in three dimensions by recording height change during lateral dimensional scanning. Figure 2.4 Illustrative sketch of a typical AFM system.

35 26 AFM achieves extremely high resolution on sample surface morphology, depending on sharpness of the probe tip, precision of the scanner, and the optical detection system. The probe tip usually has a radius of a few nanometers (~10nm), which only has a small amount of atoms at the end. High sensitivity piezoelectric ceramic provides accurate three-dimensional displacement. The long path of the laser beam amplifies slight changes of the tip, and the laser signal is precisely captured by photodiode. Tip-sample interaction also helps AFM achieve atomic level resolution. Tip-sample interaction may be described by the Lennard-Jones potential [ω(r)] [47], which considers the ideal interaction between two atoms: A B ω(r) (6) r 6 Where r is the distance between two bodies and A, B are constants for fixed atoms. The force interaction is then: r r 12 dω 6 A 12 B F (7) 7 12 dr r Figure 2.5 The dependence of force on the probe distance.

36 27 As shown in the above equation, as well as in Figure 2.5, the Lennard-Jones force depends directly on the probe distance. The curve increases steeply when separation distance is less than a certain benchmark, which provides the sensitivity to detect slight height changes from repulsive force. Different scanning modes are available, depending on several sample-tip relative motion types. In contact mode, the tip scans under an applied force, and a feedback system keeps the tip-sample interaction force in a constant value. In dynamic mode, oscillating cantilever beats on the sample surface to minimize the tip-sample contact effect, and improves the ability to detect soft materials. The oscillation amplitude and phase change are taken as parameters to report surface information. Therefore, AFM not only depicts morphology information, but also presents material surface property by phase shift information. Although AFM is a powerful tool for characterizing a variety of materials in different environments, it still has some limitations. As a result of the sharpness of the tip, the AFM used in this research can only scan within 100μm 2 in image size, and the scanning speed is much slower than other characterization methods, such as electron microscopy. Unsuitable tip shape introduces image artifacts in the morphology image. Artifacts are mainly due to changes in tip shape, such as sticking impurities from sample surface, or a small piece knocked off from the tip end. As a detector originally designed for flat materials, AFM works poorly on morphology with steep walls or overhangs.

37 28 Chapter 3 Results: Morphological Studies of Perovskite Thin Films 3.1 Introduction Organo-metal halide perovskites have garnered great attention over the last three years[48, 49] because they provide the highest efficiency among third-generation photovoltaic materials. The vapour deposition technique allows for the making of perovskite thin film, which can potentially be assembled into photovoltaic devices through easy fabrication. Compared with solution-processed methods that were previously common, vapour deposition possesses several advantages over controlling the film thickness and morphology. First, by controlling the deposition rate of two chemical sources, vapour deposition precisely controls thickness at the nanometer level [43]. Second, vapour deposited perovskite thin films have uniform morphology suitable for further device assembly [36, 44]. For an ideal perovskite thin film, uniform morphology effectively avoids the possible short circuit in the device, and provides homogeneous performance across the entire device. As such, the uniformity of morphology contributes to enhanced efficiency of such devices. Lastly, vapour deposition techniques are easier for large-scale production, which is a key factor when one considers the potential for future manufacturing. The primary goal of this research is to make perovskite structures with organic and inorganic source chemicals in a PVD system, which has been described in the experimental section. There are several parameters that need to be optimized in order to obtain the perovskite thin film. Two major factors for making highly crystalized perovskite are: deposition rate, and ratio of two source chemicals. By varying the temperature of the K-cells, we controlled the deposition rate of the film as well as the organic-inorganic ratio. Optimum temperature of each K-cell was decided by checking the crystal structure of the thin film with XRD. Besides preparing perovskite thin films, another important issue is perovskite stability. The stability of perovskite is the main factor determining device performance and lifetime. It is also decisive in choosing fabrication conditions (e.g. vacuum requirement). Therefore, investigations on stability of perovskite thin film will benefit from understanding the thin film properties, and predicting device performance and lifetime. Instead of checking the perovskite

38 29 performance and stability in solar devices, two studies were conducted directly on the perovskite thin film. The first study focused on the effects of annealing temperature on film formation and degradation. Annealing is one of the key procedures aiming to improve perovskite performance. By heating the film above a certain temperature, materials become more homogeneous and exhibit better properties. In this study, we observed the effect of annealing temperatures, and examined the temperature tolerance of perovskite thin film. To further investigate the stability of the perovskite film, we studied the degradation behavior of perovskite thin film exposed to air. Morphological information such as film thickness, surface coverage, and crystal grain sizes can assist in evaluating the rate of film formation, consistency of the preparation conditions, and changes upon treatments. Solution processed thin film with rough surface morphology is not suitable for monitoring the change of film surface [43]. However, vapour deposited film can achieve very uniform surfaces as a baseline for morphology studies, and enables capturing morphology changes. With a uniform surface, we clearly observed thin film morphology changes with varying annealing temperature, air exposure time, and other conditions that may cause the perovskite degradation. It is important to note that the vapour deposition method confronts several obstacles during experimental procedures. First, methylammonium iodide (CH 3 NH 3 I) does not stick on the substrate alone, so it is very hard to monitor its deposition rate, and precisely control the organic component independently. Second, the CH 3 NH 3 I powder easily gets clustered upon heating. The ideal solid material should disperse freely as powder; thus, the clustering makes it difficult to maintain a stable evaporation rate. Last, the deposition system needs to be calibrated often so as to keep the same experimental condition from batch to batch, especially the tooling factor of the QCM. In all, preparation of perovskite sample is time consuming, and a great number of samples are needed to obtain significant data points. In this chapter, we described conditions for preparing perovskite thin film with uniform thickness by the vapour deposition method. By controlling the heating temperature of K-cells, we successfully optimized the deposition rate of the CH 3 NH 3 I and PbCl 2 in order to obtain perovskite thin film. The relationship of the deposition rate and the temperature was also studied. The crystal structure was examined using XRD, while the surface morphology was studied using SEM and AFM. Surface coverage changes were also investigated.

39 Experimental Details Materials for vapour deposition of perovskite thin films are prepared separately. CH3NH3I was usually synthesized with following method: 24 ml of CH3NH2 reacted with 10 ml of HI in a 250 ml round-bottom flask at 0 C for two hours, with constant stirring. The precipitate of the reaction was then collected using a rotary evaporator, by carefully removing the solvents at 50 C. The obtained product was re-dissolved in 80 ml absolute ethanol, and precipitated with the addition of 300 ml diethyl ether; this procedure was repeated twice for purification. The final product of CH3NH3I was collected and dried at 60 C in a vacuum oven for 24 hours. The Ted Sargent group from the Department of Electrical and Computer Engineering provided the CH 3 NH 3 I, and PbCl 2 was purchased from Sigma-Aldrich. Substrate cleaning is a key procedure in guaranteeing final thin film uniformity in thickness. A substrate with insufficient cleaning will cause uneven film patterns, which lack uniformity. As we aimed to image the morphology of the perovskite film with SEM and AFM, very flat substrates were desired. As such, we chose a silicon wafer as the substrate that would provide flat surface, and would also allow the film to form on the surface. The cleaning procedure of the silicon wafer consists of three steps. The first step is to thoroughly wash the substrates with water, acetone, and methanol, in that order; extra wiping or sonication may be required depending on the surface condition. The use of a different solvent is to ensure that the substrate was cleaned with solvents of different polarity. The next step is to dry the substrate with a flow of nitrogen gas right after the solvent wash. Finally, the substrate is treated with UV-Ozone radiation. The UV-Ozone treatment not only cleans the surface with high-energy radiation, but also it oxidizes the surface in order to gain higher work function and surface potential, which helps the deposition process and enhances the adhesive property. The whole cleaning procedure was conducted immediately before the deposition, ensuring the least air exposure after cleaning. In addition, the color of the film is also a key indication of film quality. So glass slides were used as substrate to prepare sample for colour check.

40 31 Vapour deposition process is carried out on substrates following sufficient clean. Varying the heating temperature of the K-cells that contain source materials CH 3 NH 3 I and PbCl2 controlled the evaporation rate. The common temperature used for heating PbCl 2 is around 330 ºC, and for CH 3 NH 3 I it is 125 ºC in the chamber, with the pressure of Torr. These optimal temperatures were decided by varying the temperatures of K-cells and recording the deposition rate of each source. The device performance and the thin film quality depend greatly on the component of the CH 3 NH 3 I and PbCl 2. To optimize the component ratio of CH 3 NH 3 I to PbCl 2, the ratio of 1:1 to 10:1 were conducted, and the optimum ratio was found to be around 8:1 for making perovskite thin film in this system. Since it is difficult to deposit CH 3 NH 3 I on a silicon substrate when deposited alone, we set the QCM as the set of PbCl 2, and the deposition rate of CH 3 NH 3 I was calculated accordingly. The optimum perovskite film thickness is around 100 nm, and each deposition takes around two hours. This thickness is proper for subsequent characterizations [43]. After the deposition process, the sample was taken out from the chamber for annealing. As discussed, the perovskite is very sensitive especially to moisture; as such, we kept the sample in a glove box except when annealing or characterizing. The samples are annealed in a vacuum oven to separate the sample from air and humidity during heating. Some reported studies did not keep the sample in vacuum during annealing [44, 50]. Instead, the device was annealed in air with the relative humidity at around 30%. Our study kept the same from exposure to air, since moisture in air was considered to be a major factor in degradation of perovskite film [50]. All annealed samples were heated in the determined temperature for 20 minutes. To anneal samples at different temperatures, certain number of choice sample were taken out of the glove box. During the annealing process of the first sample, other samples were exposed to air. As such, the second sample actually stayed in air for 20 minutes more than the first sample. X-ray diffraction (XRD) is a powerful technique with which to study the long range ordering of perovskite crystal. It is used to examine the crystallinity of perovskite material, which indicates the purity of the thin film and the grain size of individual crystal. The tolerance of annealing temperature was detected by the XRD, and showed how the structure changes along

41 32 with increases to the annealing temperature. Taking advantage of the morphology uniformity provided by vapour deposition, it is convenient to study perovskite thin film morphology with a series of characterization methods. Using SEM, we can observe morphology of the perovskite thin film. AFM is a great tool with which to present morphological changes in very high resolution. Characterization: X-ray diffraction pattern (2θ scans) of the perovskite thin film were obtained on an X-ray diffractometer (Panalytical X Pert Pro), with Cu-Kα radiation (λ= å). SEM analysis was performed on a Quant FEG 250 environmental SEM. The AFM used is a JPK NanoWizard II AFM. A cantilever with nominal spring constants between 40 and 50 N/m (NCH probes, Nanoworld Innovative Technologies) was operated under the dynamic force mode. In this mode, the cantilever is vibrated at around the resonant frequency, and its amplitude reduces when the tip is in proximity to the sample surface, which is caused by tip sample interaction. Reduced amplitude is set as the feedback parameter (set point) so that the AFM system scans the surface contour of the sample with minimized error signals (the difference between the set point and the amplitude measured) by adjusting the distance between the tip and the sample surface. Mapping of this distance constructs a topographic image of the surface morphology. Mapping the error signal resulted in an image removing the height contribution, and stressing only the shape of surface features. When the height range is large, surface features with small height differences are obscured. In this case, it is advantageous to use the error signal image in order to show the shapes of surface features, while using the topographic image to estimate the height distribution. The scan rate for obtaining images is 1 Hz. The experiment was conducted in air with a relative humidity of ~40%.

42 Results and Discussion For each deposition, four pieces of substrates were aligned in the sample holder so that four identical thin films could be obtained. Usually, one of the four substrates is a glass slide that allows for a quick check of the film colour. The samples of 100 nm-thick perovskite film made by vapour deposition technique are shown in Fig Only samples deposited on glass are shown here because of the significant contrast between film and substrate. Samples deposited on silicon wafers are not well depicted on a photo with clear contrast. The left sample in Fig. 3.1 with light yellow colour is a typical failure deposition due to insufficient perovskite formation on the substrate surface. The yellow colour is mainly due to PbI 2 mixed in the film, indicating that PbI 2 is the intermediate in the reaction between CH 3 NH 3 I and PbCl 2. A successfully-deposited perovskite film deposition should have dark color, indicating it absorbed most of the visible light (Fig. 3.1 sample on the right). The color of the film also gets darker with the increase of the thickness. From the colour of the sample, we can quickly check if the deposition constitutes a desired perovskite film. Figure 3.1 Photos of 100-nm thin film perovskite samples made by the vapour deposition technique: left photo, a typical failure-deposition on glass; right photo, a typical successful deposition of the organometal halide perovskite.

43 34 The ideal temperatures for making a good perovskite CH 3 NH 3 PbI 3 are around 330 ºC for PbCl 2 and 125 º C for CH 3 NH 3 I. To set the deposition condition, we fixed the deposition rate of PbCl 2, and adjust the deposition rate of CH 3 NH 3 I. Controlling the temperature at these conditions provides the ratio of deposition rate of CH 3 NH 3 I:PbCl 2 is 8:1. Under this ratio the deposited thin film easily shows good perovskite structure as examined with XRD. For another previous study using vapour deposition, the temperature of PbCl 2 and CH 3 NH 3 I are 320 ºC and 116 ºC [43]. The possible reasons for the different deposition temperatures could be the result of many factors. Different chamber geometry, especially the distance from the K-cells to the substrate, has an influence on the deposition rate. Also, if the vacuum conditions are not exactly the same the deposition temperature may change form one system to another Characterization of Perovskite and Related Materials According to previous studies of perovskites, the main XRD peaks assigned to the (110), (220) and (330) crystal planes are at 14.1º, 28.4º, and 43.2º, respectively [19, 36]. Based on theoretical calculations (100), (200), (300) peaks are also closely aligned at these three respective positions with lower intensity. At 2θ = 14.1º, (100) the peak is usually covered by the (110) peak, so a small peak should would be expected. And similar situations happen at 28.4º for (200) and (220) peaks, and 43.2º for (300) and (330) peaks. Fig. 3.2a shows an XRD pattern of a typical pure CH 3 NH 3 PbI 3 perovskite thin film. In this pattern, three main diffraction peaks at 14.1º, 28.4º, and 43.2º presents with strong intensity. But each peak s width is broader than a typical single peak, indicating that the signal is a doublet of two peaks. Solution processed perovskite have shown identical peaks in previous studies, which demonstrate that both techniques are able to produce perovskite with same structure [19]. In order to interpret the XRD results of perovskite made by vapour deposition, several standard patterns of related materials are also examined, including the inorganic source PbCl 2 and the possible side product PbI 2. PbI 2 has a very high intensity of (110) at 12.65º, as shown in Fig. 3.2b; with two other notable peaks at 39º and 52.5º. PbCl 2 has a more complex pattern (Fig. 3.2c), two broad peaks at around 14º and 27.8º, five sharp peaks from 22.5º to 30º, and some other minor peaks around 40º - 50º. With the XRD patterns of PbI 2 and PbCl 2, the deposited perovskite structure can be explained if there are mixed components of PbI 2 and PbCl 2 in resultant thin films.

44 35. Figure 3.2 XRD patterns: (a) Pure perovskite thin film of CH 3 NH 3 PbI 3 ; (b) PbI 2 powder purchased from Sigma-Aldrich; (c) PbCl 2 powder used for film preparation purchased from Sigma-Aldrich.

45 Annealing Effects for Perovskite Thin Films The annealing process is one of the key procedures during solar cell fabrication. In materials science, annealing is a heat treatment that alters physical and sometimes chemical properties in order to improve the working properties. This heat treatment usually consists of heating a material to above a certain temperature, maintaining that for certain time, and then cooling to room temperature. Commonly, annealing is applied to soften material, relieve internal stresses, and refine material structure for making it homogeneous, which eventually makes the material workable with certain ideal properties. This study investigated the annealing effects on perovskite thin films. As described in the experimental section (chapter 2), the annealing procedure is conducted in vacuum oven under 260 ºC, and on a hot plate over the same temperature this is done because of the heating limitations of the vacuum oven. Perovskite preparation requires precise control of the evaporation system, including its K- cells temperature and QCM readings. Only several batches of samples are presenting signature peaks of perovskite, so the study is limited to discuss only a few samples with significant perovskite structures. Powder X-ray diffraction was used to study the effect of annealing, and selected patterns are shown in Fig Pattern (a) is the as-made sample (having been just removed from the deposition chamber), and (b) is a sample annealed at 140 ºC in a vacuum oven. The pattern of as-made sample shows (110), (220), and (330) peaks of perovskite at 14.1º, 28.5º, and 43.2º, respectively, demonstrating that perovskite material has been synthesized by the vapour deposition method. These sharp peaks are dominant signals in the pattern, which means that CH 3 NH 3 PbI 3 perovskite has been prepared as the main composition of the thin film. And as discussed in last section, the widths of all three characteristic peaks are broadening by some shoulder peaks. These peaks are (100), (200), and (300) peaks, covered under (110), (220), and (330) peaks. However, this pattern also presents a PbI 2 (110) peak at 12.6º, 27º, and 39º, which reveals that the thin films are actually a mixed composite of CH 3 NH 3 PbI 3 perovskite and PbI 2. Previous XRD studies have observed peaks of CH 3 NH 3 PbCl 3, indicating the existence of a mixed halide perovskite of CH 3 NH 3 Pb 3-x Cl x. In this study, however, a peak of CH 3 NH 3 PbCl 3 was not noticeable. This is mainly because the x is usually much smaller than 0.1, so XRD is not necessarily able to detect its existence.

46 37 According to the observation of PbI 2 peaks in perovskite thin films, the chemical process of the film formation reaction of organic and inorganic source chemicals can be explained. During the vapour deposition, the reactions taking place on the substrate can be described through the equation below: 2CH 3 NH 3 I + PbCl 2 = 2CH 3 NH 3 Cl + PbI 2 (1) PbI 2 + CH 3 NH 3 I =CH 3 NH 3 PbI 3 (2) Pattern (b) is the sample annealed at 140 ºC in a vacuum oven. The pattern has the same peak positions with a slight enhancement in the signal intensity. This enhancement attributes to two changes in perovskite thin film. First, the annealing process helps the crystallinity. Second, the perovskite crystal grains tend to gain size during the annealing process, because it provides a condition for recrystallization of small grains into bigger crystals. In all, the study supports the finding that annealing assists the crystallinity of perovskite thin film. This conclusion was confirmed by further study on the annealing process, and will be explained in detail in the next section.

47 Figure 3.3 XRD patterns of: (a) perovskite sample before annealing treatment and (b) perovskite ample after annealing treatment in 140 ºC for 20 minutes. 38

48 The Effects of Annealing Temperature on Perovskite Thin Films XRD Results To further examine the influence of annealing temperatures, we used an XRD to examine the crystal structure of the perovskite film after annealing at various temperatures. In order to guarantee that the thin films are identical before annealing, four pieces of film samples were deposited in one batch so that each sample could be regarded as one part of the same film. After each deposition, one of the film samples was used to check the purity of the perovskite with XRD. To make deposition conditions the same, all parameters were maintained between each deposition, and depositions in this section were conducted continuously without pauses or sources evaporating. The selected patterns of perovskite thin films after annealing at different temperatures are shown in Fig The 100 ºC annealed sample has identical peak alignments with the 140 ºC annealed sample, as well as the non-annealed sample shown in Fig Moreover, its peak intensities are also comparable with the 140 ºC annealed sample, indicating improvement of crystallinity after annealing at 100 ºC. This study found that for temperature ranges from 90 ºC to 160 ºC, 140 ºC is the optimal temperature for annealing treatment because the crystal structure detected by the XRD has the greatest-intensity peaks, whereas the surface coverage still stays at ideal level for devices. The sample annealed at 180 ºC exhibits the disappearance of the main (110), (220), and (330) peaks of perovskite at 14.1º, 28.5º, and Comparably, the peaks of PbI 2 become dominant at 12.8º, indicating the PbI 2 is the major remaining crystal structure at this stage. This is indicative that perovskite has completely disappeared after annealing at 180 ºC After annealing at 220 ºC, the XRD pattern only shows the peak of PbI 2 at 12.5º. However, samples heated at temperature beyond 260 ºC led to the emergence of a very broad peak at 27.0º, and a noticeable sharp peak at around 33.5º. These two peaks are identical to those of PbCl 2. The sample annealed at 400 ºC only shows weak signal of PbCl 2.

49 40 Figure 3.4 XRD patterns of perovskite thin films annealed at temperature ranging from 100 ºC to 400 ºC. Peaks labeled with (*) are characteristic peaks of perovskite; peaks labeled with ( ) are PbI 2 peaks; peaks label with ( ) are PbCl 2 peaks. The XRD results support the perovskite thermal dynamic property observed in our daily work. As mentioned, prepared thin films with darker color show identical perovskite peaks in their XRD patterns. Along with the increase of annealing temperatures, the sample thin films tend to become a lighter yellowish colour, and almost transparent after high temperature annealing ( ºC). This suggests that the dark color of the sample film is mainly attributable to the perovskite material, rather than PbI 2. In addition, the study clearly indicates that perovskite material CH 3 NH 3 PbI 3 is eliminated from the film sample as the temperature increases. The threshold temperature is between ºC, and the disappearing happens in a small temperature range without involving any new structural evolution in the 180 ºC sample. The main concern of CH 3 NH 3 PbI 3 dissolving concerns their reaction with moisture in air, as noted in previous studies [3, 51]. However, the samples are

50 41 annealed in a vacuum oven, which is a condition lacking of moisture. The possible dissolving process can be described as: CH 3 NH 3 PbI 3 = CH 3 NH 3 I + PbI 2 (3) Since the dissolving process happens between ºC, the mechanism is that CH 3 NH 3 I is actively reacting to other form, and thus accelerating the dissolving of perovskite. Limited by a lack of information on how CH 3 NH 3 I further reacts, or behaves in high temperatures, here we cannot exhibit the whole dissolving process. However, it can be concluded that the CH 3 NH 3 PbI 3 perovskite is dissolving at a specific temperature range. As shown in the XRD results, there are characteristic peaks of PbCl 2 appearing in the XRD of samples annealed 260 ºC, 300 ºC and 400 ºC. However, there is no Cl source contacting the samples after the deposition, and no Cl related structure has been observed in previous XRD studies. As mentioned, these samples are actually annealed on hot plates in ambient conditions, instead of in a vacuum oven. As such, one possible sources of the Cl is the dirt chemicals on the surface of the hot plate. The as-made samples were stored in a glove box that maintained a low humidity and oxygen environment, to ensure that samples remained free from degradation. For measuring the XRD patterns, a certain number of samples were taken out of the glove box and transferred into the XRD test room. The samples were exposed to air during the transferring process, and all pieces not annealed at once. During the annealing of the first sample, the rest of samples are also exposed in air. Therefore, the annealed sample in the second order stayed in air for around 20 minutes. The influence of this time gap may affect the crystal structure since the perovskite is very sensitive to moisture in the air. But this possible influence is not avoidable due to the limitations of the experimental procedure. This hypothesis would not be valid, however, if all of the PbI 2 comes from the original deposition, and not the dissolving process. In order to confirm whether the dissolving of the perovskite produces more PbI 2, quantitative examination needs to be carried out. A very easy and straightforward method to understand the material change is to monitor the morphology of the sample surface. Morphological quantitative information, such as coverage and material volume, can thus be used as tools with which to evaluate quantitative information. We therefore used

51 42 SEM to further examine the morphological changes of the perovskite thin films, along with the temperature increases.

52 SEM Results XRD results show the influence of annealing temperatures on perovskite thin films. It can be determined that the highest tolerant temperature of perovskite is about ºC. A series of crystal structural changes became clear from the XRD results. This would be helpful to predict tolerant conditions of perovskite during the fabrication of electronic devices. Based on the crystal structural changes observed in the XRD study, thin film morphology is very likely to shed some light on the annealing temperature change. Combining knowledge of thermal dynamic behavior with the morphological information, a solid prediction of the perovskite thin film thermal dynamic property could be made. In the SEM study on the influence of annealing temperatures, we were still using the same samples examined under XRD for section We observed perovskite thin films closely under SEM with different resolutions. In Fig. 3.5 we only show selected images with present significant features on sample surfaces. Figure 3.5a is the SEM image of the perovskite thin film without annealing treatment. The thin film shows a full coverage on top of the substrate surface. Unlike other thin films made by vapour deposition, our perovskite thin film does not show extremely uniform and flat film surface; this is possibly because this thin film is a mixture of aimed perovskite and its intermediate PbI 2. For samples without annealing, we noticed that the film degrades rapidly in a moist atmosphere, possibly due to the hygroscopicity of CH 3 NH + 3 cation. We discovered 140 ºC to be the ideal temperature for annealing, but the sample still suffered from reduction of surface coverage at this temperature. Therefore, the as-made sample without annealing is the only thin film that has a full coverage. For solution processed perovskite films [44], the film morphology has different behavior from sample made in air and inert gas environment. Low temperature annealing in an inert gas environment also prevents surface coverage reduction [3, 14]. The micrograph of 100 ºC annealed thin film is shown in Fig.3.5b. The thin film started losing surface coverage, and a portion area of silicon substrate exposure can be observed. During the heating process, many small pores formed rapidly, then either disappearing or joining together to form a void area that can be observed from the image. In addition, the thin film annealed at 100 ºC was homogenous, since the film was likely pieced together by crystal grains. As XRD verified two mixed crystal structure, perovskite and PbI 2, the film displayed in the SEM

53 44 image seems not clear enough to distinguish one crystal from the other. Nevertheless, it is notable that the surface coverage is slightly dropped to 94% from the fully covered thin film. Figure 3.5 SEM micrographs of perovskite samples: (a) Sample without annealing treatment; (b) Sample annealed at 100 ºC. Upon heating the thin film to 140 ºC, the pore kept enlarging in size, and surface coverage reduced to around 87% (Fig. 3.6c). Besides this main coverage change, the image shows two types of crystal morphology: one kind is big crystal grains similar to those observed

54 45 in 100 ºC annealed sample; the other is the smaller-grain groups gathering at the gap area between the big grains. According to the XRD results, this film still consists of two crystal structures, perovskite CH 3 NH 3 PbI 3 and PbI 2, mixing together. The emergence of the smallergrain groups suggests that one kind of the crystal may not mix as well as in previous samples. But further research is needed to determine and confirm which crystal is separating from the mixture. Figure 3.6 SEM micrographs of perovskite samples: (c) Sample annealed at 140 ºC; (d) Sample annealed at 180 ºC.

55 46 The thin film sample that annealed at 180 ºC is shown in Fig 3.6d. The materials lost half of their surface coverage over the whole substrate surface, and 51 % of the area is still covered with the remaining materials. The crystal morphology of this sample is small crystal grain groups that look similar to those small crystal grain groups between big crystal grains (Fig 3.6c). If the small grain groups observed in Fig. 3.6c are one pure crystal, then this morphology verifies it as PbI 2. However, it is hard to identify the chemical composition from the morphological observation. Figure 3.7 SEM micrographs of perovskite samples: (e) Sample annealed at 260 ºC; (f) Sample annealed at 300 ºC.

56 47 Surface coverage kept decreasing as annealing temperatures increases; the thin film sample annealed at 260 ºC has a surface coverage of 44%. The surface morphology is mainly small crystal particles with sizes around 1 μm (Fig 3.7e). XRD results support the idea of only PbI 2 crystal left in this sample. At very high temperature, the material only remains spherical crystals with sizes smaller than 500 nm. Shown in Fig 3.7f, the particles only cover 7 % of the surface. According to the XRD result, the remaining crystal is mainly PbI 2 at high annealing temperatures. Again, we were not able to detect exact chemical composition, so more study is needed to address and explain the particle composites. As shown in the images of samples annealed over 260 ºC, no significant morphology appears as a new crystal form. So the PbCl 2 cannot be assigned as any morphology features. Surface coverage is one of the parameters we can use to describe the thin film formation over a surface area. To calculate the coverage, we enlarged the film, and divided part of the image in to small squares. By counting the squares covered with thin film, we can calculate the thin film morphology coverage. Table 1 shows the detailed numbers of squares filled with materials, and also the coverage percentage of thin films. The data were then plotted in a diagram that shows the decreasing trend of the surface coverage (Fig. 3.8). Combining both the XRD and SEM results, we determined that the annealing temperature greatly affects the crystallinity and morphology of perovskite. The ideal annealing temperature assists a thin film sample for better crystallinity. In XRD results, perovskite completely disappeared beyond 180 ºC and the major component of the remaining film is PbI 2. Since the PbI 2 has been presented in the film before the perovskite is gone, it is hard to conclude if the PbI 2 is from the film preparation or is from the dissolution of perovskite. From the SEM results, even chemical components cannot be detected; here, a study of surface coverage sheds light on the quantitative information. The surface still had more than 50% coverage at 180 ºC, so that at least half of the thin film s volume was PbI 2. Along with temperature increases, most of the perovskite disappears from the sample. Our study can be referenced for the future fabrication of mixed perovskite annealing process, although interesting questions remain open for study..

57 48 Table 1: The surface coverage analysis of samples annealed at different temperatures. Annealing Surface area Occupied area Surface temperature (Counts of squares) (Counts of squares) coverage (ºC) (%) Non-annealed

58 49 Figure 3.8 Perovskite surface coverage as a function of annealing temperatures Degradation Study of the Perovskite Thin Films (in Air) Perovskite is known as an unstable material that can degrade in air [44]. During the fabrication of a solar cell, the exposure of perovskite to air can hardly be avoided, so understanding the change of the material is very important. Here we studied the degradation behavior of perovskite sample by exposing it in air, and monitoring the changes in crystal structure and morphology. Figure 3.9 shows the XRD patterns of the same sample in two stages. The first stage is right after annealing, and the second stage is after exposure in air for 14 days. The relative humidity is around 30% on average. Fig. 3.9a shows the XRD pattern of the sample after annealing at 140 ºC. The characteristic peaks whose position at 14.1º, 28.5º, and 43.2º correspond to (110), (220) and (330) reflections belong to perovskite structure. However, peaks of PbI 2 are also shown at 14.1º, 27.0º and 39.0º, specifying that samples made from our technique were mixture of CH 3 NH 3 PbI 3 perovskite, and PbI 2. After the XRD measurement, the samples were kept in a container that was not airtight for 14 days, which allowed the sample to

59 50 gain some exposure to air, but would not be contaminated by dust. Degradation took place during the 14-day process, and we measured the XRD pattern after 14 days, which is shown in Fig 3.9b. Comparing the XRD pattern of 14-day degraded sample with the one just after annealing, we found that three characteristic peaks of perovskite all completely disappeared. The remaining peaks were all due to PbI 2, indicating that perovskite degraded in air, and the remaining material is PbI 2 that exists in the film from beginning. Therefore, the result demonstrates that perovskite CH 3 NH 3 PbI 3 degrades in air spontaneously. Figure 3.9 XRD patterns of perovskite thin films: (a) Perovskite film made by vapour deposition technique and annealed at 140 ºC; (b) The same sample of (a) exposed to air for 14 days.

60 51 According to the XRD results, the only degraded component of the annealed film was the perovskite. Through observation of morphology changes during the process, it is possible to identify the morphology composition of perovskite and other components of the thin film. In this study, SEM was used to examine the morphology in the nano scale level. The information provided by SEM is more straightforward than a structural analysis of XRD in determining surface morphology. The SEM images of these two stages are as shown in Fig Both images were taken immediately after the XRD measurement. At the first stage (Fig. 3.10a), the majority of the image is the material-covered film, while only a small portion of area is the substrate exposure. By analyzing the area of the covered surface and uncovered surface, we determined that the thin film covered area is 87.3%. In this image, we observed that the thin film is not totally uniform with a certain degree of coverage. Moreover, there are two kinds of morphology features on the film. One morphological feature, circled with red on the left, has solid thin film fully covered in a certain area. This feature is identical with those of samples that had not been annealed. On the other hand, morphology as circled in green, on the right, has a structure of connected small grains sitting at the edge of a big solid film. Correlated to the XRD results, these two morphological features may strongly relate to crystal of perovskite and PbI2. Figure 3.10b shows the second stage of the degradation process. It highlights that after exposure in air for a 14-day period, the thin film coverage drops to 33.8%, which is less than half of its original coverage. Also, the thin film morphology consists of small grain sized crystals with disappearance of the bigger area solid thin film. This is indicative of the fact that the solid thin film pieces degraded in air. We may be able to confirm that the perovskite structure that changed in XRD and the bigger solid thin film in morphology are closely related. It is possible that the disappeared solid crystal pieces are perovskite, and the remaining smaller sized grains are PbI 2. Based on the XRD and SEM results, we are able to calculate the degradation rate of perovskite in air over time. Over 14 days, the surface coverage percentage dropped from 87.5% to 33.8%, equivalent to a daily average drop of 5.37% on the surface area. We believe that the degradation rate is related to the specific humidity condition and thus may vary each day. Nevertheless, we conducted an efficient method to quantitatively evaluate the degradation rate of perovskite thin film, which may contribute to the manufacturing control in large scale production.

61 Figure 3.10 SEM images of perovskite films: (a) Perovskite film made by vapour deposition technique and annealed at 140 ºC; (b) The same sample from (a) exposure in air for 14 days. 52

62 High Resolution Study of Perovskite Thin Film with AFM To investigate the surface features in high resolution, we used AFM to gain surface information that explained the annealing process and growth mechanisms. AFM is a powerful characterization method used to investigate surface morphology, and is designed for scanning on small-scale surfaces with height distribution within its capable z-direction adjustment. The image quality heavily relies on the sample surface conditions and the lateral limitation is very small (100 μm 100 μm). For our study, only suitable samples were selected to analyze surface morphology with AFM. As discussed in the XRD and SEM results sections (chapter 2), we only observed a surface coverage loss of 12 % during the heating process, but the crystal morphology remained almost the same. In addition, the film that showed grain size change in XRD cannot be characterized using SEM. The detail morphological information on the crystal is studied by one of high resolution scanning probe microscopy, AFM. Figure 3.11 AFM images of perovskite films annealed at different temperature from 85 ºC to 140 ºC. Here, AFM provided additional high-resolution information of surface features. Figure 3.11 is a collection of AFM images of perovskite, showing a significant small region (1 μm 1 μm) of perovskite surface, with different annealing temperatures. The samples annealed at 85 ºC, 100 ºC and 140ºC, as well as the sample that was not annealed, present morphological differences. The non-annealed and 85 º C annealed samples are both fully covered by materials. The thin film surface consists of uniform spherical grains with an average grain size of around 100 nm. The 100 ºC annealed sample shows a more flat shape in each grain, tending to have a sheet-like shape. The 140 ºC annealed sample tends to increase this effect. The grains are even flatter and exhibit a trend of amalgamating all grains into bigger crystal sheets. This phenomenon

63 54 indicates a recrystallization of the thin film during the annealing treatment. Grain sizes of the 140 ºC annealed sample become larger due to the recrystallization. This observation is consistent with the XRD results. The recrystallization effects of annealing can only be observed in high resolution AFM images, whereas SEM cannot differentiate morphological change at this scale. We can also use AFM morphological images to study the growth process by comparing images for films with different thickness, as shown in Fi The left panel shows the topdown-view of the surface of a 10-nm thick sample, while the right panel depicts a 100 nm thin film. The enlarged images (1 μm 1 μm) are inserted for a detailed observation in high image resolution. The image shows the grain sizes of the 10-nm thick sample range from 10 nm to 200 nm. At the same magnification, the surface of the 100-nm thick sample shows much more uniform thin film structures, and all the grains have a sized of 50 nm (± 5 nm). The difference between the two images indicates that the film formation is not always uniform along the deposition process. The grain sizes change along the deposition, but eventually become uniform when the film is deposited with enough thickness.. Figure 3.12 AFM images of perovskite films with10 nm thickness (left) and 100 nm thickness (right).