Synthesis and Application of Few-Layered Hexagonal Boron Nitrade

University of Wollongong Research Online

University of Wollongong Thesis Collection 1954-2016 University of Wollongong Thesis Collections

2016

Synthesis and application of few-layered hexagonal boron nitrade

Majharul Haque Khan University of Wollongong

Follow this and additional works at: https://ro.uow.edu.au/theses

University of Wollongong Copyright Warning You may print or download ONE copy of this document for the purpose of your own research or study. The University does not authorise you to copy, communicate or otherwise make available electronically to any other person any copyright material contained on this site. You are reminded of the following: This work is copyright. Apart from any use permitted under the Copyright Act 1968, no part of this work may be reproduced by any process, nor may any other exclusive right be exercised, without the permission of the author. Copyright owners are entitled to take legal action against persons who infringe their copyright. A reproduction of material that is protected by copyright may be a copyright infringement. A court may impose penalties and award damages in relation to offences and infringements relating to copyright material. Higher penalties may apply, and higher damages may be awarded, for offences and infringements involving the conversion of material into digital or electronic form. Unless otherwise indicated, the views expressed in this thesis are those of the author and do not necessarily represent the views of the University of Wollongong.

Recommended Citation Khan, Majharul Haque, Synthesis and application of few-layered hexagonal boron nitrade, Doctor of Philosophy thesis, Institute for Superconducting and Electronic Materials, University of Wollongong, 2016. https://ro.uow.edu.au/theses/4813

Research Online is the open access institutional repository for the University of Wollongong. For further information contact the UOW Library: [email protected]

Institute for Superconducting and Electronic Materials

SYNTHESIS AND APPLICATION OF FEW- LAYERED HEXAGONAL BORON NITRIDE

MAJHARUL HAQUE KHAN

Presented as part of the requirements for the award of the Degree of ‘Doctor of Philosophy’ of the University of Wollongong

July, 2016

ABSTRACT

Hexagonal boron nitride nanosheets (BNNS) is one of the most widely studied two- dimensional (2D) materials these days, due to their extraordinary properties and potential applications. For example, BNNS has been proposed as an ideal substrate for graphene based electronics because of its ultra-flatness and structural similarity to graphene. The synthesis of large-area, homogeneous, and few-layered BNNS, however, remains challenging. Among the various synthetic routes, atmospheric pressure – chemical vapour deposition (AP-CVD) is preferred on the grounds of its cost effectiveness and its ability to yield large-area BNNS from a single run. Maintaining the homogeneity and crystallinity of the nanosheets over a large surface area requires fine-tuning of variables in the AP-CVD procedure to their ideal levels.

Copper has been widely used as a catalyst for the growth of BNNS. A comparative study of BNNS growth on solid and melted copper confirms the advantages of melted copper. On the solid copper, the BNNS is largely inhomogeneous and tends to adopt a multilayered tetrahedral shape along the defects, while on the melted copper surface it is single crystalline over several microns, with the majority of nanosheets being mono- or bi-layer thick. This difference is likely caused by the significantly reduced and uniformly distributed nucleation sites on the smooth melted surface, in contrast to the large amounts of unevenly distributed nucleation sites that are associated with grain boundaries and other defects on the solid surface.

Pretreatment to remove the residual hydrocarbon from ammonia borane, the most commonly used precursor, is necessary to remove the carbon contamination found during the growth. In addition, flattening the copper and tungsten (chosen due to good wettability of the liquid copper) substrates prior to growth and slow cooling around the copper melting point facilitate the uniform growth of BNNS. This is the result of a minimal temperature gradient across the copper substrate. Confining the growth inside alumina boats effectively minimizes etching of the nanosheets by silica nanoparticles originating from the commonly used quartz tube. The CVD grown h- BNNS on solid Cu surfaces adopts AB, ABA, AC´, and AC´B stacking orders, which are known to have higher energies than the most stable AA´ configuration. Electron

energy loss spectroscopy (EELS) was found to be effective in determining the number of layers by means of the energy loss near edge structure (ELNES) of the N edge and by core loss edge quantification. These findings are instrumental for the fabrication of high-quality BNNS via CVD and would motivate studies on the stacking order dependent properties and performance of BNNS.

Compact nanoscale devices require ultra-thin coatings for protection. Hexagonal boron nitride has high thermal and chemical stability, is electrically insulating, and the hexagons are considered to be impermeable to water and oxygen. The reactivity of Cu in different environments with the presence of few-layered BNNS was therefore examined. Accelerated reaction of Cu directly below BNNS towards water was observed compared with oxygen, which was unexpected considering the chemical stability and insulating properties of BNNS. Density functional theory (DFT) calculations demonstrate that water dissociates spontaneously between the single BNNS layer and the Cu(111) support promoting the formation of Cu oxide underneath the BNNS. X-ray photoelectron spectroscopy (XPS) and EELS analysis studies reveal that there is no change in the chemical states of B and N atoms, indicating that h-BN catalyzes the reaction between Cu and water. Once fully covered by BNNS, Cu experiences a dramatically slowed reaction under salt water compared with bare Cu, which is likely due to the impermeability of the hexagons with respect to the ions. The efficacy of protection is closely related to the quality of the BNNS, where defects such as domain boundaries and wrinkles are detrimental. Future improvement in protection by BNNS lies in the minimization of defects or their patching up with protective atomic filler.

ACKNOWLEDGEMENTS

All praise and gratitude to Allah, the Almighty, for giving me the strength and patience to complete this thesis work.

During my PhD work, I received encouragement and assistance from a wide range of people. I wish to express my deep appreciation to my supervisors, Prof. Hua Kun Liu and Dr. Zhenguo Huang, for their encouragement, understanding, invaluable advice, and constant support during my study at the University of Wollongong.

Gratitude is owed to Dr. Gilberto Casillas, who helped to perform the aberration- corrected TEM studies. His expertise and knowledge prove invaluable to my research. I am also indebted to Dr. Sina S. Jamali for his help with the corrosion analysis. I am thankful for technical assistance from Dr. Tony Romeo (SEM and Sputter Coater), Dr. Mitchell Nancarrow (Leica DM 6000 Optical Microscope and Struers Durascan-70 Hardness), Dr. Patricia Hayes (Raman and UV-vis Spectrophotometer), Dr. Germanas Peleckis (XRD), Dr. Michael Higgins, Dr. Yi Du (AFM), Dr. David R. G. Mitchell (EELS analysis), and Dr. Dongqi Shi (XPS). I am also grateful to Mr. Robert Morgan and Mr. John Wilton for their unfailing assistance to solve the mechanical and electrical problems of the CVD furnace that I encountered several times during my doctoral study. My special thanks go to Dr. Tania Silver for editing this thesis as well as publications.

I am thankful for my International Postgraduate Tuition Award and University Postgraduate Award (3.5 years) from the University of Wollongong, Australia and Dr. Zhenguo Huang’s 6-months of financial support towards finishing my Ph.D.

This thesis could not have been completed without the contribution from my parents and my brothers and sister. They continuously encouraged me to obtain this higher degree. Their unfailing support and advice helped me to accomplish this study.

iii

PUBLICATIONS

1. Majharul Haque Khan, Zhenguo Huang, Feng Xiao, Gilberto Casillas, Zhixin Chen, Paul J. Molino, & Hua Kun Liu, “Synthesis of Large and Few Atomic Layers of Hexagonal Boron Nitride on Melted Copper”, Scientific Reports, vol. 5, pp 1- 8, 2015 (doi:10.1038/srep07743). – (one of the top 100 read Scientific Reports articles of 2015)

2. Majharul Haque Khan, Gilberto Casillas, David R.G. Mitchell, Hua Kun Liu, Lei Jiang, & Zhenguo Huang, “Carbon- and Crack-free Growth of Hexagonal Boron Nitride Nanosheets and Their Uncommon Stacking Order”, Nanoscale, 8, 15926-15923, 2016 (doi: 10.1039/C6NR04734C).

3. Majharul Haque Khan, Sina S. Jamali, Andrey Lyalin, Gilberto Casillas, Paul J. Molino, Lei Jiang, Hua Kun Liu, Tetsuya Taketsugu, & Zhenguo Huang, “Atomically thin hexagonal boron nitride nanofilm for Cu protection: the importance of film perfection”, Advanced Materials, Accepted, 2016.

4. Feng Xiao, Sina Naficy, Gilberto Casillas, Majharul Haque Khan, Tomas Katkus, Lei Jiang, Hua Kun Liu, Huijun Li, and Zhenguo Huang, “Edge- Hydroxylated Boron Nitride Nanosheets as an Effective Additive to Improve the Thermal Response of Hydrogels”, Advanced Materials, 27, 7247-7247, 2015 (doi: 10.1002/adma.201502803).

ABBREVIATIONS

1D One-Dimensional 2D Two-Dimensional AC-SECM Alternating Current - Scanning Electrochemical Microscopy ADF-STEM Annular Dark Field - Scanning Transmission Electron Microscope AFM Atomic Force Microscopy AP-CVD Atmospheric Pressure - Chemical Vapor Deposition BF-STEM Bright Field - Scanning Transmission Electron Microscope BNNS Hexagonal-Boron Nitride Nanosheet BNNT Boron Nitride Nanotube CVD Chemical Vapour Deposition DFT Density Functional Theory DF-TEM Dark Field - Transmission Electron Microscopy EDS Energy Dispersive Spectroscopy EELS Electron Energy Loss Spectroscopy EIS Electrochemical Impedance Spectroscopy FT-IR Fourier Transform - Infrared Spectroscopy h-BN Hexagonal Boron Nitride HR-TEM High Resolution - Transmission Electron Microscope LP-CVD Low Pressure - Chemical Vapour Deposition LPR Linear Polarization Resistance PMMA Poly (methyl methacrylate) PVA Polyvinyl Acetate SAED Selected Area Electron Diffraction SECM Scanning Electrochemical Microscopy SEM Scanning Electron Microscopy STEM Scanning Transmission Electron Microscopy TEM Transmission Electron Microscopy UHV Ultra-High Vacuum UV-Vis Ultraviolet-Visible Spectroscopy XPS X-ray Photoelectron Spectroscopy

TABLE OF CONTENTS ABSTRACT ...... i LIST OF FIGURES ...... x LIST OF TABLES ...... xxii Chapter 1 ...... 1 Introduction ...... 1 1.1 Motivation, prospective applications, and challenges ...... 2 1.2 Thesis structure ...... 3 Chapter 2 ...... 5 Literature review ...... 5 2.1 Structural information on boron nitride ...... 5 2.1.1 Structure of hexagonal boron nitride nanosheet...... 5 2.1.2 Defects in BNNS ...... 8 2.2 Properties and applications ...... 11 2.2.1 Bulk h-BN ...... 11 2.2.2 BNNS ...... 12 2.2.2.1 Substrate for nanodevices ...... 12 2.2.2.2 Mechanical properties ...... 13 2.2.2.3 Thermal conductivity ...... 14 2.2.2.4 Optical Properties ...... 14 2.2.2.5 Stability in air ...... 14 2.2.2.6 Chemical properties ...... 15 2.2.2.7 Sliding and frictional characteristics ...... 15 2.2.2.8 Other properties ...... 16 2.3 Synthesis of BNNS ...... 17 2.3.1 Top-down approaches ...... 18 2.3.1.1 Micromechanical cleavage ...... 18 2.3.1.2 Sonication-assisted exfoliation ...... 19 2.3.1.3 Electron beam irradiation ...... 20 2.3.2 Bottom-up approaches ...... 21 2.3.2.1 Chemical vapour deposition ...... 21 2.3.2.1.1 Precursors for BNNS synthesis ...... 23 2.3.2.1.2 Substrates for BNNS synthesis ...... 24

2.3.2.1.3 Theoretical studies of BNNS growth on transition metal substrates...... 25 2.3.2.1.4 BNNS on polycrystalline transition metal substrates ...... 27 2.3.2.1.5 Pre-treatment of the substrate ...... 28 2.3.2.1.6 Mechanism of BNNS growth ...... 28 2.3.2.1.7 Transferring BNNS to arbitrary substrates ...... 30 2.3.2.2 Physical vapour deposition (PVD) ...... 31 2.3.3 Other processes ...... 32 2.3.4 Limitation of the synthesis routes and challenges for application ...... 33 2.4 Characterization tools...... 35 2.4.1 Optical microscope...... 35 2.4.2 Raman microscope ...... 37 2.4.3 Fourier transform infrared spectroscopy (FTIR)...... 38 2.4.4 Scanning electron microscopy (SEM) ...... 38 2.4.5 Electron backscatter diffraction (EBSD) ...... 39 2.4.6 Transmission electron microscopy (TEM) ...... 39 2.4.7 Atomic force microscopy (AFM)...... 41 2.4.8 X-ray photoelectron spectroscopy (XPS)...... 43 2.4.9 UV-vis spectroscopy ...... 43 2.4.10 X-ray diffraction (XRD) ...... 44 2.4.11 Other techniques...... 44 Chapter 3 ...... 46 Experimental Approaches ...... 46 3.1 BNNS growth on copper substrate ...... 46 3.1.1 Growth of partially and fully covered BNNS ...... 47 3.1.2 Preparation of bare annealed solid and melted Cu substrates ...... 48 3.2 Transferring BNNS to different substrates ...... 49 3.3 Characterization ...... 51 3.4 Electrochemical analysis ...... 52 3.4.1 Electrochemical impedance spectroscopy (EIS) ...... 52 3.4.2 Scanning electrochemical microscopy (SECM) ...... 53 3.4.3 Linear Polarization Resistance (LPR) ...... 54 3.5 Simulation ...... 54

vii

3.5.1 HAADF-STEM image simulations ...... 54 3.5.2 Density Functional Theory (DFT) calculation ...... 55 Chapter 4 ...... 57 Synthesis of large and few atomic layers of h-BN by CVD method ...... 57 4.1 Introduction ...... 57 4.2 Results ...... 58 4.3 Discussion ...... 68 4.4 Conclusion ...... 73 Chapter 5 ...... 74 Carbon- and crack-free growth of BNNS, its uncommon stacking orders, and EELS assisted thickness analysis ...... 74 5.1 Introduction ...... 74 5.2 Results ...... 75 5.2.1 Impact of precursor purity ...... 75 5.2.2 Impact of carrier gas and tube material ...... 78 5.2.3 Crack free growth of BNNS ...... 79 5.2.4 DF-TEM analysis of the crystallinity of BNNS ...... 83 5.2.5 Stacking order of CVD grown BNNS ...... 85 5.2.6 EELS analysis to identify monolayer BNNS from the few-layer ...... 92 5.3 Conclusion ...... 94 Chapter 6 ...... 96 Reactivity of Cu with the presence of BNNS: from the perspective of atomically thin passivation ...... 96 6.1 Introduction ...... 96 6.2 Results ...... 97 6.3 Conclusion ...... 120 Chapter 7 ...... 121 Conclusion and future outlook ...... 121 7.1 Summary ...... 121 7.1.1 Growth of large and homogenous BNNS ...... 121 7.1.2 Crack- and carbon-free growth of BNNS ...... 121 7.1.3 Observation of unusual stacking in BNNS ...... 122 7.1.4 EELS assisted thickness analysis ...... 122

viii

7.1.5 BNNS as an ultrathin passivation layer on Cu substrate ...... 123 7.2 Suggestions for future work ...... 123 7.2.1 BNNS growth ...... 123 7.2.2 Origin of h-BN stacking orders and their characterization ...... 124 REFERENCES ...... 125

LIST OF FIGURES

Figure 1.1: (a) Illustration of contemporary 2D materials. (b) Potential applications of 2D materials in flexible smart systems. Adapted from Ref17...... 3 Figure 2.1: Schematic representation of 2D h-BN nanostructure. Adapted from Ref45 ...... 7 Figure 2.2: Six possible types of bi-layer h-BN stacking...... 7 Figure 2.3: (a) Schematic illustration of three types of topological defects, taking the example of graphene: disclination, dislocation, and domain/grain boundaries. Adapted from Ref56. (b) Proposed 5/7 and 4/8 type dislocations in BNNS, consisting of B- and N-rich charged dislocation cores in the 5/7 types and neutral dislocation cores in the 4/8 types. Adapted from Ref53. (c) Proposed domain boundaries in BNNS. Types I and II are made up of 4/8 type defects. Types III- V domain boundaries consist of 5/7 type defects having various stable arrangements of B and N atoms in those defects. Adapted from Ref49. (d) Experimental evidence of domain boundaries in BNNS, consisting of 5/7 type defects, which has been confirmed by HRTEM analysis. Adapted from Ref54. (e) 4/8 and (f) 5/7 type defects in the domain boundaries of BNNS confirmed by STM analysis. Adapted from Ref49...... 9 Figure 2.4: (a) Point defects in BNNS. N- and B-rich atomic defects, are expressed as

NB and BN, respectively, while N and B atom vacancy defects are expressed as 57 VN and VB, respectively. Adapted from Ref . (b) HR-TEM image of vacancy defects in BNNS. (c) The modeling of B and N vacancy defects. According to this model, all the vacancies in (b) turn out to be B-rich vacancy defects. (d) Exit wave reconstructed HR-TEM image of a boron monovacancy defect in BNNS. (e) Intensity profile along the line in (d). Adapted from Ref58...... 10

Figure 2.5: Schematic illustration showing a graphene sheet (red) supported by SiO2 substrate (left) and by BNNS (right). The corrugations, dangling bonds (inside

dashed circle), and charge inhomogeneities that are inherent to SiO2 surfaces are absent in BNNS. Adapted from Ref19...... 13 Figure 2.6: The proposed puckering effect in 2D materials (using graphene as an example), where the adhesion to the sliding AFM tip causes out-of-plane bending of the nanosheet, leading to increasing contact area and friction. Adapted from Ref101...... 16

Figure 2.7: Commonly used methods for the synthesis of BNNS...... 17 Figure 2.8: Schematic illustration of two exfoliation mechanism in the ball-milling process, (a) peeling off layer-by-layer, and (b) crumpling of the sheets under a shear force in the ball-milling process. Adapted from Ref106...... 19 Figure 2.9: Sequence of events during CVD growth process.121,122 ...... 21 Figure 2.10: Schematic diagram of the interrelationships among experimental parameters, process variables, and coating properties.123 ...... 22 Figure 2.11: (a) Structure of borazine, (b) structure of β-trichloroborazine, and (c) synthesis of BNNS by three step of boration, oxidation, and nitration process.145 ...... 24 Figure 2.12: (a) Atomic positions of boron and nitrogen in a stable h-BN monolayer sheet on Co (0001), Ni (111), Cu (111) (Left image, 1×1 h-BN unit cell on top of a 1×1 metal surface cell), and Pd, Pt, Ag, and Au (111) (Right image, 2×2 h-BN unit cell on top of a √3×√3 metal surface cell). Adapted from Ref175. (b) Energy vs lattice parameter plot, comparing equilibrium h-BN energy with h-BN sitting on other (111) oriented transition metals. Adapted from Ref174. (c) B-rich and N- rich triangles of size L = 10. (d) Total energy in a series of triangles shown as a function of their size L, red for B-rich and blue for N-rich. Adapted from Ref177...... 26

Figure 2.13: Schematic illustration of BNNS growth on Cu by the CVD process. Ji is

the impingement flux of borazine, JB is the diffusion flux of B into the Cu

substrate, JN is the diffusion flux of N out of the Cu substrate, and JB´ and JN´ are the flux of B and N atoms to form h-BN on Cu. Adapted from Ref190. (b) Three widely used thin film growth models, as a function of coverage θ in monolayer. (1) Volmer-Weber, (2) Frank-van der Merwe, and (3) Stranski-Krastanov growth models. Adapted from Ref192...... 29 Figure 2.14: Two commonly adopted techniques for graphene or BNNS transfer to TEM grids. Adapted from Ref196...... 31 Figure 2.15: (a) Optical contrast change of mono- (blue line) and bi-layer (red-line) region with different wavelengths of light. The contrast is positively maximum at ~ 570 nm wavelength. The figures below (a) shows the contrast variation of mono- and bi-layer h-BN under different wavelengths of light. Adapted from Ref219. (b) and (c) Schematics of liquid crystal method and oxidation methods

respectively, for the direct identification of graphene on Cu substrates. Similar techniques have been developed for identification of BNNS. Adapted from Ref223,225...... 36 Figure 2.16: (a) HRTEM image of few-layer graphene, which reveals that the few layer graphene has been creased. (b) An atomistic model of the creased few- layered graphene which gives rise to the HRTEM image in (a). Adapted from Ref237. (c) Atomic number, Z vs. ADF intensity plot, giving a perfect match between the experimental and theoretical profiles. (d) ADF-STEM image of monolayer BNNS, and (e) the corresponding intensity profile along the XX´ and YY´ directions. Impurity elements, i.e., C and O, in the nanosheet can be identified from their relative intensities. Adapted from Ref238...... 40 Figure 2.17: Determination of mechanical strength of BNNS by the AFM indentation

technique. (a) SEM image of large BNNS on a SiO2/Si substrate, with a patterned array of circular holes in the substrate. Areas C and E are fully covered by BNNS, and area F is the region fractured by indentation. (b) Schematic illustration of mechanical strength determination of BNNS by nanoindentation. Adapted from Ref50...... 42 Figure 3.1: Schematic illustration of an APCVD system. Volatile thermal decomposition products of ammonia borane were carried over the Cu substrates

by a 10% H2/Ar gas mixture. For the melted Cu, tungsten was used as the support due to its better wettability compared to the quartz slide...... 48 Figure 3.2: Time-temperature profiles for the CVD growth of BNNS. (a) Fast cooling process and (b) slow cooling process after the growth, while all the other parameters were kept fixed...... 49 Figure 3.3: SECM (a) experimental set-up and (b) electrical model explaining the data

collected in the AC mode. RCu, RSL, RSol, and RT represent the resistance associated with Cu, surface layer, solution and SECM tip, respectively. C

elements (Ccu, CSL, and CT) represent the associated capacitance elements...... 54 Figure 4.1: Discontinuous and inhomogeneous BNNS grown on solid Cu with a 200

ml/min 10% H2/Ar carrier gas flow. The corresponding line profile shows fragmented BNNS of about 1 nm in thickness...... 59

xii

Figure 4.2: (a), (b) Optical images; (c), (d) AFM height images with corresponding

thickness profiles on SiO2/Si wafer for the S-BNNS and M-BNNS samples, respectively...... 60 Figure 4.3: (a) Representative XPS survey spectrum of BNNS on melted Cu substrate. From the integrated peak area analysis of B 1s (b) and N 1s (c) peaks, a nearly 1:1 B/N atomic ratio is obtained, indicating sp2 h-BN formation. The Shirley background function was used for B 1s and N 1s peak background correction, and a 70% Gaussian, 30% Lorentzian combined function was used for peak-fitting analysis. XPS of C 1s, O 1s and Cu 2p peaks are shown in (d), (e) and (f). The small C 1s (c) and O 1s (d) peaks might have come from surface contamination by air in between the synthesis and the XPS measurements. The C 1s and O 1s peaks were also observed in a bare annealed Cu sample...... 61 Figure 4.4: Comparative Raman spectra of BNNS grown on solid and melted Cu

substrates and then transferred to SiO2/Si substrate...... 62 Figure 4.5: (a) HAADF-STEM image of a bi-layer BNNS. The dark triangular region is monolayer h-BN, which was exposed after electron beam radiation removed the top layer. The FFT pattern in the top inset shows a distinct six-fold hexagonal pattern, characteristic of h-BN. The inset in the red square shows a FFT filtered B mono-vacancy zone. (b) Raw HAADF-STEM image from a monolayer BNNS region. The heavier N atoms are slightly brighter than the B atoms. B and N atoms have been tagged with red and blue circles, respectively. (c) FFT filtered HAADF-STEM image of a defect-free h-BN mono-layer region and Line profile along the trace...... 63 Figure 4.6: Electron knock-on damage and layer opening in bi-layer BNNS, observed in ADF-STEM imaging mode at 80 kV: (a-d) Both B and N atoms were gradually knocked out of the top layer under electron beam irradiation, forming a triangle. HAADF-STEM image (d) can be an important tool to identify elements and the atomic layer thickness, by taking advantage of the Z-contrast.92 e) Direct layer identification from the intensity line profile analysis across the HAADF-STEM image in (d) clearly shows the vacancies, and the monolayer, and bi-layer BNNS regions. (d) was smoothed to lower the background noise in the line profile image in (e)...... 64

xiii

Figure 4.7: (a) A low magnification BF-TEM image of large area BNNS. (b) EELS spectrum acquired from the circled region in (a), showing the K-shell ionization edges of B and N, which confirms the sp2 bonded h-BN formation. The C peak possibly comes from the PMMA residues after the transfer...... 65 Figure 4.8: HAADF-STEM characterization of M-BNNS. (a) Low magnification HAADF-STEM image of M-BNNS. SAED patterns (bottom) were obtained from the four circled areas. The distinctive set of hexagonal spots in the SAED patterns indicates the single crystalline nature of the nanosheet within that region. The difference in rotational orientation in the four SAED patterns is less than 1°. (b) The statistical distribution of BNNS atomic layers based upon 40 BF-STEM images. BF-STEM images of (c) mono- and (d) bi-layer BNNS...... 66 Figure 4.9: (a) HAADF-STEM image of BNNS grown on melted Cu and (b) intensity profile along the line in (a). The similar intensity in the three zones indicates homogeneous BNNS in that region. The relatively bright regions are the PMMA residues...... 67 Figure 4.10: (a) Low magnification TEM image of BNNS grown on solid Cu. SAED patterns obtained from the four areas marked in (a) show the polycrystalline nature of the BNNS. (b-d) BF-STEM images of S-BNNS. (b) 3-layered, (c) 7- layered, and (d) 10-layered BNNS...... 68

Figure 4.11: SEM images S-BNNS and M-BNNS at different growth stage, with the insets showing corresponding images at higher magnification. For solid Cu, (a), (b) h-BN triangular domains with adlayers at the nucleation centers start to grow preferentially along the defect lines and grain boundaries. (c), (d) more ammonia borane results in new h-BN domains while the initial triangular domains grow in the lateral direction and coalesce with their neighbors to form a continuous sheet. For melted Cu, (e), (f) initial h-BN domains grow randomly on the surface due to the absence of defect lines and grain boundaries. Increasing the amount of ammonia borane leads to a complete coverage. For the whole process, no adlayer growth is observed. The scale bars are 5 μm in the images and 1 μm in the insets...... 70 Figure 4.12: Optical microscope images and Raman spectra of different Cu substrates.

(a) as-received, (b) 10 % HNO3 acid etched, (c) 1000 °C annealed, and (d) 1100 °C melted Cu substrate. Scale bar is 50 μm. The inset in each of the images shows

xiv

the three-dimensional (3D) AFM height profile across a 10 x 10 μm2 area. The color bar is within the 0 to 1 μm range in all the insets...... 72 Figure 4.13: Raman spectra of Cu foil surfaces...... 72 Figure 4.14: SEM images of BNNS grown at 1050 °C. (a) No preferential nucleation sites are observed for the initial h-BN domains. (b) Large and mixed multilayer BNNS with small adlayers on the top are found when more ammonia borane is used. Scale bar is 5 μm...... 73 Figure 5.1: Impact of impurities in the precursor on the growth. SEM images (a) and (b) show that as-received ammonia borane leads to the formation of graphite-like structures (appearing as dark spots) on top of BNNS. (c) Optical microscopy image of the BNNS grown on solid Cu substrate using as-received ammonia

borane. (d) Raman spectrum from the circled spots in (c) shows the ID (1334.04 -1 -1 -1 cm ), IG (1585.11 cm ), and I2D (2657.45 cm ) peaks of graphite. (e) and (f) High-resolution secondary electron image with corresponding EDX mapping (scanning TEM mode, 80 kV operating voltage) of C, O and Si. The presence of carbon in the spotted regions is clearly noticeable. (g) BF-TEM image of a dark spot region. (h) HR-TEM image of the dark spot reveals a layered graphitic structure. (i) SEM image of BNNS on top of solid Cu substrates without any noticeable dark spots when pentane-treated ammonia borane was used. Inset is the SEM image of BNNS on melted Cu substrate. The absence of dark regions (compare with (a)) indicates that treating ammonia borane with pentane followed by evacuation can effectively remove the residual organic contaminants...... 77 Figure 5.2: Effect of carrier gas on the BNNS growth. (a) Using UHP Hydrogen gas (99.999 %) as a carrier gas can produce BNNS on melted Cu substrate. Inset is the high magnification image of the BNNS islands, showing a 3D growth of BNNS in those regions. (b) Traces of BNNS are observed if only UHP Argon gas (99.999 %) was used as a carrier gas. Large particles (black under SEM contrast) are also found on top of Cu substrate. Inset is the SEM image of a portion of BNNS on Cu substrate. (c) SEM image of three particles which have joined each other. (d) and (e) EDX mappings show that silicon and oxygen clearly overlap

with the region of the particles, indicating the particles are SiOx...... 79 Figure 5.3: Digital camera images of Cu foils on W after 1100 °C melting and then solidification. (a) With no flattening of the Cu and W foils, an elliptical shaped

Cu surface is formed on the W, while (b) flattening the foils can produce a uniformly wetted Cu surface on the W foil...... 80 Figure 5.4: Schematic illustration of growth of BNNS on melted Cu substrate: (a) When the Cu and W foils are not flattened prior to growth, large cracks are visible. (b) SEM image of large cracks, hundreds of microns wide, around the center of an elliptical shaped Cu substrate, which is caused by the different shrinkage rates of the Cu on solidification due to the large thermal gradient. Inset is a high magnification SEM image from the highlighted region showing large crumples. (c) SEM image of relatively small cracks (identified by yellow arrows) near the edges of the Cu substrate. (d) When the Cu and W foils are flattened and the cooling rate is slow through the melting point, only small cracks are observed. (e) On the flattened foils, crumples and large cracks are largely minimized. (f) Enlargement of the image in (e), with the EDX spectrum (inset) indicating the

presence of SiOx which is highlighted by red circles in (c), (e), and (f). The presence of nanoscale cracks (indicated by green arrows in (c), (e), and (f)) is

possibly initiated by SiOx nanoparticle etching...... 81 Figure 5.5: When the Cu and W foils were not flattened prior to the growth, cracks generated at the centre of an elliptical shaped Cu surface cannot be avoided, even with a slow cooling through the Cu melting point...... 82 Figure 5.6: (a) Schematic illustration of the BNNS grown in a confined space. Cu/W

foils are kept inside two alumina boats to minimize direct exposure to SiOx nanoparticles. A slim opening was maintained to allow the precursor gas (mainly borazine gas from the decomposed ammonia borane) to make contact with the Cu

surface. (b) SEM image of large and nearly crack free BNNS. Very few SiOx nanoparticles are observed. The striations are probably due to small crumples.83 Figure 5.7: Grain size analysis using the DF-TEM technique. (a) BF-TEM image of two merged BNNS triangles. (b) SAED pattern of the overlapping region (marked by a dotted circle in (a)) shows two sets of spots with a hexagonal symmetry. (c) Schematic diagram of the overlapping triangular pyramids and the corresponding SAED patterns from the marked circular areas. (d) and (e) Color-coded DF-TEM images, derived by selecting the ringed (red and green) 1-210 spots in (b). (f) Combined DF-TEM image, obtained by merging the images in (d) and (e). (g) DF-TEM images show that BNNS grown on solid Cu is polycrystalline, and the

xvi

grains are mostly a few hundred nanometers in size. Some overlapping grains are identifiable from their mixed colors. (h) Large single-crystalline BNNS of several microns in size is obtained on melted Cu substrate. Inset is the red green blue (RGB) color model image...... 84 Figure 5.8: (a) BF-TEM image of BNNS on a lacey carbon TEM grid. (b) SAED pattern taken from the red-circled spot in (a) reveals three differently oriented h- BN crystals in that region. 1, 2, and 3 are DF-TEM images obtained along the red, green, and blue aperture filters marked in (b). (c) Combined DF-TEM color image (from 1, 2, and 3) reveals the crystalline structure in each region. Overlapping is observed along grain boundaries. (d) Raw ADF-STEM image from the overlapped region shown in (c). (e) FFT pattern obtained from (d), showing two different sets of hexagonal spots, rotated about 9.6° with respect to each other. (f) FFT filtered image of (d) showing a clear Moiré pattern developed due to the absence of definite stacking orders...... 86 Figure 5.9: (a) Raw HAADF-STEM image where a hole (dark) is visible. (b) Intensity profile along the rectangle in (a) shows a stepwise increment in intensity corresponding to vacuum, 1 layer, and 2 layers of BN...... 87 Figure 5.10: Identification of stacking orders in BNNS. (a) FFT filtered HAADF- STEM image of BNNS showing bi-layer surrounding a monolayer region. Simulated images together with normalized experimental (black) and calculated (red) intensity profiles along the arrows are shown. The center positions of the hexagonal cells are marked by black arrows. The experimental bi-layer structure matches that calculated for AB stacking. (b) Deconvoluted HAADF-STEM image of the 3/4-layer region, where the trilayer is ABA stacked. (c) FFT filtered HAADF-STEM image of bi-/tri- layer region. The bi-layer adopts AC´, and the tri-layer is AC´B stacked...... 89 Figure 5.11: Simulated annular ADF-STEM images of the six different types of h-BN stacking possible for the tri-layered BNNS in Figure 5.10b. The first two layers of this tri-layered structure have been found to be AB (Figure 5.10a). Therefore, six other stacking orders are possible in the third layer...... 90 Figure 5.12: Simulated ADF-STEM images of bi- and tri-layers of BNNS, having various stacking configurations...... 91

xvii

Figure 5.13: Using EELS to identify the number of layers of BNNS: (a) HAADF image of 8-layered BNNS in a tetrahedrally-shaped island. EELS spectral imaging was acquired from the marked rectangular region. The N K-edge color intensity map of the rectangular region in (a) with layer numbers is shown. (b) Intensity profile across the map showing 8 distinct layers. The spike at the first step is due to contamination along the base of the tetrahedrally-shaped BNNS island. (c) EELS spectra of the nitrogen K-edge for 1, 2, and 8 layer BNNS. The peak at 418 eV appears from regions 2 or more layers thick but is absent in the spectrum of the monolayer...... 94 Figure 6.1: (a) A SEM image of the completely covered solid Cu substrates by BNNS.

(b) and (c) AFM height profile images of the transferred BNNS on 285 nm SiO2 coated Si substrates, grown on solid Cu substrates. Height profile analysis along the red line in (b) indicating that the thickness of the BNNS is 0.8 nm, while multilayered islands of BNNS are clearly seen on the top. ~1 nm thickness from AFM, often indicates the presence of monolayer due to water and contaminants trapped in between contributing to the increased height of the monolayer sheet219...... 98 Figure 6.2: Oxidation of Cu surface in the presence of BNNS. When fully covered, (a) Cu surface experiences a negligible change in contrast when kept in oxygen or UHV; (b) Red spots and stripes show up after being kept in air for 3 days; (c) Dark red stripes were observed after being kept in water for 3 days. The Reddish region in the optical image is associated with oxidized Cu...... 99 Figure 6.3: (a) A SEM image of Cu substrate partially covered by BNNS. Black regions are BNNS. (b) Optical image of the corresponding sample shown in (a), after 3 days of air exposure. Black and blue arrows indicating more Cu oxidation along the edges and domain boundaries of BNNS, respectively...... 99 Figure 6.4: When partially covered, (a) Cu shows negligible change in contrast after being kept in oxygen/UHV; (b) strong oxidation of underlying Cu was observed after 3 days in water exposure. (c) Raman mapping from the marked rectangular region in (b) reveals a close correlation between the oxidation and the BNNS, where more oxidation is seen underneath the BNNS. (d) XPS analysis reveals more oxidation of the BNNS coated Cu compared with bare Cu...... 100

xviii

Figure 6.5: (a) A SEM image of partially grown BNNS on melted Cu substrate. (b) Optical microscopy image of the sample after 3 days of air exposure. Strong Cu oxidation underneath the BNNS is clearly identifiable from the reddish contrast. Blank Cu substrate is seen in white contrast under the optical microscope. .... 101

Figure 6.6: Energy diagram for H2O dissociation on Cu(111) surface (black line), on BNNS/Cu(111) (red line), between the BNNS and Cu(111) (blue line), and along the edges of OH-terminated BNNS (green line)...... 103

Figure 6.7: Optimized configurations of (a) H*, (b) H2O*, (c) O*, and (d) OH* between monolayer h-BN and Cu(111). B: blue; N: grey; O: red; H: electric blue (small sphere); Cu: celeste (large sphere)...... 104

Figure 6.8: Optimized configurations of (a) H2O* (b), H* (c), OH* and (d) O*; adsorbed in the vicinity of an OH-terminated zigzag edge of BNNS on a Cu(111) surface. B: blue; N: grey; O: red; H: electric blue (small sphere); Cu: celeste (large sphere)...... 105 Figure 6.9: XPS analysis of (a) B 1s and (b) N 1s peaks of the BNNS on Cu after 6 months of air exposure. B-O or N-O peaks are not detected from the peak deconvolution analyzes. (c) ADF-STEM image of the 5-layer h-BN triangular pyramid. The intensity profile indicates that the layer numbers is correlated with the ADF-STEM intensity. (d, e) EELS analysis of the red rectangle (along the edge) and blue rectangle (along the plane) in (c), respectively. No oxygen was detected in both places...... 107 Figure 6.10: Optical microscopy image of fully coated solid Cu substrate with BNNS, (a) after 1 day (b) after 1 week and (c) 1 year of air aging. The degree of Cu oxidation can be correlated from the optical contrast variation315,316 (dark red for heavily oxidized Cu surface). The optical contrast variations indicate that the Cu oxidation in the presence of BNNS slowed down significantly after 1 week. It is assumed that after the formation of initial thin layer of copper oxide (favoured by B-O bonds, as discussed in the DFT calculation section above), further oxidation is slowed down due to the retardation of electron transport by the insulating BNNS...... 108 Figure 6.11: Comparison of EIS test results between the bare and BNNS coated Cu substrates: (a) Nyquist and (b) Bode plots, acquired by EIS in 0.5 M NaCl solution after 5 minutes of immersion. (c) Nyquist and (d) Bode plots after 1 hour

xix

immersion in 0.5 M NaCl solution. Impedance values are the highest for BNNS coated melted Cu substrate. Note that both the Nyquist and Bode plots are presented as acquired for a surface of area of 0.785 cm2 and repeated for at least 3 times on each sample to check reproducibility...... 109 Figure 6.12: Electrical equivalent circuit used to fit the EIS data...... 110 Figure 6.13: Electron backscattered diffraction (EBSD) analysis of the (a) solid and (b) melted Cu substrates. Very large grains are seen on the melted Cu substrate. Inset (top right in b) is the EBSD map of a large area including the grain shown in b. The scale bars in the images are 500 μm...... 111 Figure 6.14: (a-f) Optical micrograph images of the bare and BNNS coated Cu substrates before and after the EIS tests. The BNNS coated melted Cu is found to be the least corroded after the EIS test. Scale bar is 100 μm...... 111 Figure 6.15: SEM images of (a) bare solid and (b) bare melted Cu substrates, and (c- f) BNNS coated Cu substrates before and after the EIS test. (c) and (e) are the SEM images of BNNS coated solid and melted Cu substrates respectively. Cu oxide layers and pits are visible after the test, on solid and melted Cu substrate (b). Large pits associated with the corrosion products are highly visible for BNNS coated solid Cu (d), while pitting is hardly visible for BNNS coated melted Cu (f). Inset of (c) is a high magnification image of a four-layered tetrahedral shaped BNNS island. SAED pattern obtained from this BNNS tetrahedron identifies it as single crystalline (inset of (c)). However, after corrosion, large numbers of pits are detected on these multilayer tetrahedrons (inset of (d)). Scale bars in blue and green correspond to 5 and 1 μm respectively...... 113 Figure 6.16: (a) A five-layered triangular h-BN pyramid and (b) two inverted triangles, imaged under the SEM mode inside TEM at 80 kV. Small holes in the center of the pyramids are apparent in both images, which formed during the growth...... 114 Figure 6.17: XPS analysis of (a) Cu 2p peaks and (b) Cl 2p peaks indicates that BNNS coated melted Cu suffers the least corrosion...... 114 Figure 6.18: Optical images of (a) bare solid, (b) bare melted Cu sample before the EIS test, (c) bare solid, (f) bare melted Cu, (e) BNNS coated solid, and (f) BNNS melted Cu after the EIS test. (g) Raman spectra from the circled regions in (a-f).

Bare Cu samples are seen heavily oxidized while the BNNS coated melted Cu sample suffered the least oxidation. Scale bar is 10μm...... 115 Figure 6.19: Oxidation of Cu substrates partially covered by BNNS at 200 °C in air for 10 minutes and 10 days. (a) After 10 min, BNNS coated and bare Cu regions are observed as whitish and red regions, respectively, in the optical image. (b) Raman mapping of the marked area in (a) taken in the range of 450-700 cm-1. Images marked with 1 and 2 represents the Raman spectra from the circled areas of (a). (c) Optical image of the sample after 10 days. Grayish areas are BNNS covered and the greenish regions are uncovered. (d) Raman mapping from the marked rectangular region in (c) taken in the range of 450-700 cm-1. (3) and (4) represents the Raman spectra from the highlighted regions in (c). BNNS can protect the Cu substrate for a short term. In the long run, oxygen ion migration occurs along the BNNS edges and defects and gradually the whole Cu surface underneath the BNNS is oxidized...... 117 Figure 6.20: Linear polarization resistance (LPR) curves of the bare and BNNS coated Cu substrates (collected on an area of 0.785 cm2). Resistance to corrosion was determined from the inverse slope value of the trendline of each curve. The BNNS coating has doubled the charge transfer resistance...... 118 Figure 6.21: Normalized AC-SECM approach curves from 1 kHz to 100 kHz in 1 mM NaCl solution, obtained using a Pt microelectrode with 25 μm diameter. (a) bare solid and melted Cu (inset), and (b) BNNS coated solid and melted Cu substrates (inset), respectively. The positive feedback indicates the insulating nature of the BNNS coating...... 119

xxi

LIST OF TABLES

Table 2.1: Comparison of graphene and BNNS structural parameters...... 6 Table 2.2: Factors and properties in the CVD growth process123 ...... 22 Table 2.3: List of substrates used as catalysts for BNNS growth.7,8,21,22,50,90,125,130,143– 171 ...... 25

xxii

Chapter 1

Introduction

Dimensionality is an important feature for many materials. The material properties can be altered if one of its dimensions is significantly reduced, down to the scale of a few atomic layers. Upon nanosizing, quantum confinement comes into effect, and some exceptional properties are displayed, which are quite contradictory to those in the bulk form. For example, graphite, when thinned down to its physical limit (1 atomic layer 1 thick), can act as ballistic charge carriers. MoS2 turns from an indirect to a direct bandgap semiconductor in its monolayer form2. Boron nitride nanoribbons with zigzag edge termination can show metallic behavior, while bulk boron nitride is insulating.3

The first reduced dimensional material (zero-dimensional) is fullerene synthesized in 1985.4 Later on, in 1991, the one-dimensional (1D) carbon nanotube was synthesized, which is one of the most studied materials in this research arena for the last two decades.5 Following the invention of 1D carbon nanotube and 0D fullerene, the only missing part in the reduced dimensional family was two-dimensional (2D) materials. The research in this field was limited, possibly due to the assumption of the environmental instability of 2D materials. Although there were several reports on the synthesis of 2D materials from the early 1960s6–8, the field did not gain enough attention until the discovery of graphene and other 2D materials in 2004 that were produced by micromechanical cleavage.9

Over the last decade, there has been an exponential increase in 2D materials research. Graphene occupies the largest portion of research output in the world of 2D materials (Figure 1a). This is attributed to the exceptional properties of graphene, which primarily include ballistic charge carrier mobility (> 200,000 cm2V-1s-1), excellent mechanical strength, remarkable thermal conductivity, etc.1 The other 2D layered materials include hexagonal boron nitride (h-BN), transition metal dichalcogenides 10,11 (TMD, including MoS2, WS2, MoSe, WSe, etc.), transition metal oxides, and the

buckled atomic crystals denoted as ‘Xenes’ (silicene12, phosphorene,13,14 and germanene15). In addition, there may be another 500 2D materials, yet to be discovered.16 Although there is vibrant research going on in the field of graphene, the research on the other 2D layered materials is still in its nascent stage, and significant developments are necessary to explore their properties and to utilize their potentials in practical forms.

1.1 Motivation, prospective applications, and challenges 2D materials have shown outstanding properties that make them the materials of choice for future semiconductor and flexible nanoelectronics. Examples of such anticipated flexible nanoelectronics applications are shown in Figure 1b.17 The necessity of ultra-thinness, transparency, flexibility, and energy efficiency in these applications have put the use of 2D materials at the forefront. Nevertheless, 2D materials are unable to support themselves due to their atomic scale thickness, while the use of conventional bulk substrates as supports drastically reduces their performance. For example, the simple act of putting graphene on top of SiO2/Si substrate causes scattering of the electrons due to the dangling bonds and roughness of 18 the SiO2 surface, which, consequently, considerably reduces its carrier mobility. Using h-BN as a support, which is sp2 bonded, atomically flat, devoid of dangling bonds, and an insulator has been proven to improve the transport properties of 18,19 graphene, to at least one order of magnitude better than on SiO2/Si substrate. A significant number of publications followed later on, using h-BN as a dielectric layer 20–23 in graphene-based transistors. In a recent study, a flexible transistor made of MoS2, h-BN, and graphene heterostructures has been demonstrated.24 Therefore, to realize the future graphene-based nanoelectronics applications, development of all the related 2D materials at the same time is indispensable.

In this thesis, the synthesis of h-BN nanosheet (BNNS) and its structural characterization have been carried out. The major challenges in 2D h-BN research are as follows: developing a large-scale production method, growing impurity- and defect- free nanosheet, maintaining large-scale homogeneity and single crystallinity, and controlling the stacking orders during the growth. The properties will vary considerably if any of these imperfections is not avoided. Also, defect, impurity, and 2

crack free transfer of the grown material to other substrates needs significant attention for successful integration of BNNS into heterostructures with other 2D materials. The chemical vapour deposition (CVD) method has proven to be the only effective technique for large-scale synthesis of BNNS. The necessity of optimizing a large number of CVD growth parameters often makes the synthesis challenging, however.

Figure 1.1: (a) Illustration of contemporary 2D materials. (b) Potential applications of 2D materials in flexible smart systems. Adapted from Ref17.

1.2 Thesis structure In this chapter (Chapter 1), the significance of 2D materials research and the standing of BNNS in relation to this context are introduced. The major challenges in BNNS growth are also summarized.

In the next chapter (Chapter 2), the relevant information on BNNS structure, growth, and characterization are summarized. Most up-to-date top-down and bottom-up BNNS growth processes are discussed, while importance is given to outlining the various substrate treatments, choice of substrates, CVD growth mechanisms, and transfer processes in the CVD route of BNNS growth.

In the experimental section (Chapter 3), the BNNS growth process, conditions, transfer techniques, and characterization instruments and parameters adopted in this work are described. The electrochemical characterizations that were carried out to determine the corrosion resistance of Cu substrates with and without the presence of 3

BNNS coating are explained. The parameters that were used for high angle annular dark field − scanning transmission electron microscopy (HAADF-STEM) image simulations and density functional theory calculations to determine the oxidation mechanisms of Cu substrate in the presence of BNNS are also presented.

In Chapter 4, melted Cu facilitated atmospheric pressure (AP) CVD growth of large and few-layer BNNS is reported. Large micron-sized single-crystalline BNNS growth was observed on melted Cu substrate compared to the nanocrystalline domains on solid Cu substrate. Also, the substrate and temperature dependence of the growth is explained in detail.

Chapter 5 is a continuation of the previous chapter, where the BNNS growth on melted Cu substrate is improved by flattening the Cu and W substrate foils prior to furnace insertion, slow cooling through the Cu melting point temperature, and pretreating the commonly used precursor. In addition, the stacking orders of the CVD grown BNNS are analysed and a method to identify monolayer and few-layered BNNS based upon electron energy loss spectroscopy (EELS) near-edge fine structure analysis of the N edge is discussed.

Chapter 6 provides details of systematic studies on the reactivity of Cu with the presence of BNNS in water, oxygen, and air. The mechanism of Cu surface oxidation in the presence of BNNS is explained by density functional theory (DFT) calculations. Moreover, the anti-corrosion performance of Cu substrate in the presence of BNNS is examined.

The summary of this thesis work and potential future work directions are presented in Chapter 7.

Chapter 2

Literature review

2.1 Structural information on boron nitride Boron nitride atomic orbitals can be either sp2 or sp3 hybridized, thus producing several crystalline forms. It forms hexagonal (h) and rhombohedral (r) lattices via sp2 hybridized bonding and forms cubic (c) or wurtzite (w) lattices via the sp3 hybridized bonding configuration. Among all of these, h-BN and c-BN are the most stable and widely available. c-BN has been synthesized under high pressure (> 5 GPa) and high temperature using h-BN powders. c-BN is considered the second hardest material after diamond.25 In addition, both p and n-type doping can be carried out on c-BN, which is very difficult to achieve in diamond.26,27

Apart from sp2 hybridized crystalline h-BN and r-BN, two other less ordered sp2 configured allotropes, amorphous (a)-BN and turbostratic (t) BN, can form during the synthesis of h-BN and r-BN. a-BN appears due to structural disorder introduced at the atomic level. Therefore, a-BN is completely disordered h-BN and is very unstable in air, reacting with water (moisture) in the environment to form boric acid, boric oxides, and hydroxides.28 t-BN, on the other hand, forms due to a fault in the c axis orientation of the h-BN layers (less or no ordering between the basal planes in subsequent layers).29 Most of the time, the interlayer spacing in t-BN is larger than those of h-BN and r-BN, and thus, only in-plane ordering of atoms is possible in t-BN. This type of BN has also been reported to be unstable after long storage in air, forming ammonium 30 borate compounds (NH4B5O84H2O).

2.1.1 Structure of hexagonal boron nitride nanosheet Hexagonal boron nitride nanosheet (BNNS) has an atomic-scale honeycomb lattice with in-plane hexagonal arrangements of boron and nitrogen atoms. Due to the remarkable structural similarity with graphene (bond length and interlayer distance), single layer hexagonal boron nitride is also called ‘white graphene’. The structural data 5

for graphene and BNNS are shown in Table 2.1 or comparison. There is a significant difference in the electronegativity between the boron (2.04) and nitrogen (3.04) atoms. Therefore, the boron nitride bond is slightly ionic in nature, and as a result, there is a lip-lip interaction between the neighboring layers of boron and nitrogen atoms in addition to the van der Waals forces between the layers.

Table 2.1: Comparison of graphene and BNNS structural parameters. Graphene BNNS In-plane lattice a = 0.246 nm a = 0.250 nm parameter (In-plane alternative C-C bond) (In-plane B-B or N-N bond) Nearest neighbour 0.144 nm distance/bond 0.142 nm (C-C bond) (In-plane B-N bond) length Interlayer spacing 0.335 nm 0.333 nm

Like the carbon allotropes of graphene, carbon nanotube, and fullerene, h-BN can form BNNS10, BN nanotube31, and octahedral BN fullerene32. In additions, various h-BN morphologies, such as nanoribbons3,33, nanoplates29,34, nanowires35, nanofibers36, nanoropes37, nanocups38, nanofunnels39, nanomikes39, microbelts40, and BN foams41, have been reported.

The 2D h-BN can be either zigzag or armchair edge terminated (Figure 2.1). From first principles and density functional theory (DFT) calculations, the armchair edge termination for h-BN nanoribbon has been found to be more energetically favorable than the zigzag edge termination.42–44 On the other hand, the nitrogen terminated zigzag edge is more favorable than the boron terminated one grown on Cu substrate, thus producing equilateral triangles during the growth.

Figure 2.1: Schematic representation of 2D h-BN nanostructure. Adapted from Ref45

Although h-BN is isostructural to graphite, there is a stacking order difference between these two materials. While graphite is a Bernal stacked (AB) structure, h-BN favors AA´ stacking (boron atoms on top of nitrogen atoms and vice versa). Nevertheless, h- BN can have 5 more bi-layer stacking configurations, which are shown in Figure 2.2. In the AC stacking of h-BN, nitrogen atom sits in the middle of the hexagon in the second layer instead of the boron atom as in the AB configured h-BN (Figure 2.2).46 Among these various orientations of h-BN, Bernal stacked AB/AC, similar to graphite, have been reported to be energetically favorable structures, in addition to the AA´ stacking for h-BN.47,48

Figure 2.2: Six possible types of bi-layer h-BN stacking. 7

2.1.2 Defects in BNNS Defects in a 2D system can have more far-reaching effects than when present in its bulk counterpart. This is because the majority of atoms are in the surface in 2D materials, and perturbation of the surface atomic arrangements due to defects can change the overall properties of the 2D crystal. For example, the presence of Stone- Wales (pentagon-heptagon pairs, 5/7) type line defects in BNNS can significantly lower the band gap.49 Also, defects have been assigned as the possible cause for the very low elastic and bending modulus in chemical-vapour-deposition (CVD) grown BNNS.50

In 2D materials where mono- to few-layer cases are mostly considered, the term ‘grain boundary’ which is a three-dimensional entity, is often replaced with ‘domain boundary’. The line defects in a bulk crystal thus correspond to the domain boundary in 2D materials. A polycrystalline 2D material is thus made up of various domains connected along the domain boundaries. The main topological defects in the 2D material can be divided into three types: disclination, dislocation, and domain boundary. An example of these three types of defects is shown in Figure 2.3a.

The Stone-Wales type defect, which is very common for graphene, was predicted to be energetically less favorable than the square-octagonal pair (4/8) line defects, due to the unfavorable homo-elemental bonding formations (B-B or N-N bonds) in the former case.51,52 In a recent study, however, through the density functional theory (DFT) calculations, both the 4/8 and 5/7 defects were predicted to be possible, depending on the tilt angles of the domain boundaries.53 The polar domain boundaries, involving the 4/8 and 5/7 type defects, neutralize the net charge, being positive at B-rich edges and negative at the N-rich edges.53 The predicted 4/8 or 5/7 type defects have been experimentally confirmed by scanning tunneling microscopy (STM)49 and high- resolution transmission electron microscopy (HRTEM) analysis54 (Figure 2.3d and Figure 2.3e-f respectively). The exact natures of the bonding elements in the domain boundaries, however, were unresolved. Nevertheless, the possible atomic configurations in the 4/8 and 5/7 type domain boundaries have been theoretically predicted for future experimental reference (Figure 2.3c).49 Among these various configurations, 4/8 type polar domain boundaries, including the less favorable B-B or

N-N bonds (Figure 2.3c, Type II), have been reported experimentally55, while the other configurations are yet to be confirmed.

Figure 2.3: (a) Schematic illustration of three types of topological defects, taking the example of graphene: disclination, dislocation, and domain/grain boundaries. Adapted from Ref56. (b) Proposed 5/7 and 4/8 type dislocations in BNNS, consisting of B- and N-rich charged dislocation cores in the 5/7 types and neutral dislocation cores in the 4/8 types. Adapted from Ref53. (c) Proposed domain boundaries in BNNS. Types I and II are made up of 4/8 type defects. Types III-V domain boundaries consist of 5/7 type defects having various stable arrangements of B and N atoms in those defects. Adapted from Ref49. (d) Experimental evidence of domain boundaries in BNNS, consisting of 5/7 type defects, which has been confirmed by HRTEM analysis. Adapted from Ref54. (e) 4/8 and (f) 5/7 type defects in the domain boundaries of BNNS confirmed by STM analysis. Adapted from Ref49. 9

Other common defects in 2D material are the point defects. Especially in the case of BNNS, due to the presence of heteroatoms, the point defects can be either B/N vacancy defects or B/N-rich point defects (Figure 2.4).57 Predominantly, boron monovacancy defect formation has been observed in high-resolution electron microscopy (HREM) imaging of BNNS (Figure 2.4b-d).58 Synthesis induced atomic vacancy formation in the nanosheet has not been reported.

Figure 2.4: (a) Point defects in BNNS. N- and B-rich atomic defects, are expressed as

NB and BN, respectively, while N and B atom vacancy defects are expressed as VN and 57 VB, respectively. Adapted from Ref . (b) HR-TEM image of vacancy defects in BNNS. (c) The modeling of B and N vacancy defects. According to this model, all the vacancies in (b) turn out to be B-rich vacancy defects. (d) Exit wave reconstructed HR- TEM image of a boron monovacancy defect in BNNS. (e) Intensity profile along the line in (d). Adapted from Ref58.

2.2 Properties and applications

2.2.1 Bulk h-BN Hexagonal boron nitride is a low-density material (2.1 g/cm3) with superior mechanical strength and elasticity. h-BN in its bulk form, has been widely used in many industrial applications. Some of its properties and applications in the bulk form are described below. h-BN has superb oxidation resistance up to 800 °C in air and is fairly stable up to 2400 °C in nitrogen atmosphere.59 This is in stark contrast to graphite, which is essentially burned away beyond 600 °C in air. h-BN is not wettable by and does not mix with molten metals as an impurity. Due to its high-temperature oxidation resistance and non-wettability by molten metals, h-BN is widely used in high-temperature refractory applications, such as furnace and ladle linings, coatings, etc.60

Due to its layered structure, h-BN has sliding properties and has a low coefficient of friction. This property has been utilized in the foundry industry, where it is used as a mold releasing powder and parting agents.60 h-BN powder can be used as a ‘dry lubricant’ because of its ability to maintain lubrication in both sub-zero and high- temperature environments. This property makes it an essential dry lubricating agent in spaceships and satellites.

Moreover, h-BN has found widespread application in the cosmetics industry. Relying on its good dispersion, sliding behavior, oil absorption capability, hiding characteristics (transparent under visible light), and good thermal conductivity, h-BN is used in cosmetics such as various forms of makeups: mascara, lipstick, and makeup pencils.60

A neutron detector made of h-BN epitaxial layers has been demonstrated on the 61 10 laboratory scale. B, an isotope of boron, has high neutron detection efficiency.

Compounds containing B4C and h-BN, mixed with a binder, have proven to have good potential for use as a solid-state thermal neutron detector.62 The results obtained using h-BN are better than with the widely used 3He gas detectors.61,62

h-BN is an insulator and has direct-band-gap energy of around 5.97 eV in its bulk form.63 The band gap of BNNS has been reported to be within 3.6-7.1 eV, however.63– 65 The discrepancies in the band-gap values in different reports are likely caused by the composition and the crystalline quality of the film. Carbon impurities66 and defects67 in the h-BN film can reduce the band gap, while a highly pure single- crystalline h-BN sample showed a band gap value of 5.97 eV. h-BN has been used as an ultraviolet light emitter due to its large absorption of light in the far ultraviolet region.63 A handheld device was demonstrated utilizing the ultraviolet absorption property of boron nitride.68

2.2.2 BNNS 2.2.2.1 Substrate for nanodevices The sp2 bonded BNNS is atomically flat, insulating, and isostructural to graphene. The charge impurities and dangling bonds that are inherent to a silica substrate are completely absent in BNNS19 (Figure 2.5). In addition, only weak van der Waals forces are active when graphene sits on BNNS. Therefore, by using BNNS as a substrate for graphene instead of the conventional SiO2, the charge transport through graphene was improved significantly.19 This fact has emphasised the necessity of making large-area crystalline BNNS and improving the transfer steps needed to build BNNS-supported 2D hybrid structures for ultrafast graphene-based electronics. In recent reports, BNNS has been demonstrated as a dielectric layer for the fabrication of graphene-based field effect transistors (FETs)21,69 and explored as a tunneling barrier between graphene layers.70 Moreover, BNNS supported flexible transistors have been reported, which hold promises for commercial applications of future nanoelectronics.24,71

BNNS has been also used as an effective release layer for GaN-based devices.72 Single-crystal h-BN can act as a buffer layer for the growth of nitride compounds. Later, the h-BN layer attached to the sapphire substrate was easily removed by mechanical force due to the weak van der Waals interaction between BNNS and the synthesized nitride compounds.72

Figure 2.5: Schematic illustration showing a graphene sheet (red) supported by SiO2 substrate (left) and by BNNS (right). The corrugations, dangling bonds (inside dashed circle), and charge inhomogeneities that are inherent to SiO2 surfaces are absent in BNNS. Adapted from Ref19.

2.2.2.2 Mechanical properties The elastic modulus of 1-2 nm thick BNNS has been theoretically and experimentally calculated (through the atomic force microscope (AFM) nanoindentation technique) to lie in the range of 184 to 510 N/m50,73, which is comparable to that for high strength steel, although much less than that of graphene (~1000 N/m).74 The bending of the nanosheet reached up to 70 nm in height for a 1 nm thick sheet, prior to final breakage, confirming the large out-of-plane flexibility of the BNNS.50 The experimental bending modulus, however, was much lower (8.8-16 N/m for 1-2 nm thick BNNS50), compared to the theoretical estimation (31.2 N/m)75, possibly because of the presence of a large number of defects in the CVD grown nanosheet.

The low density, high strength, and large elastic modulus of BNNS help to improve the mechanical performance of polymer derived composites. Several polymers, including poly (methylmethacrylate) (PMMA), polyvinyl acetate (PVA), and polybenzimidazole, have been studied separately as matrices, with exfoliated BNNS acting as filler to improve the strength of the composites.76–80 Good dispersion, nanosheet strength and wrinkles, and partial polarity in the B-N bonds were suggested as the possible reasons for the improvement in mehcanical properties.78

2.2.2.3 Thermal conductivity h-BN has a highly anisotropic thermal conductivity, with its in-plane thermal conductivity calculated to lie within the 100-2000 W/mK range; while along the longitudinal direction, the value falls to only a few W/mK.81–83 The experimental value for 11-layer (11 L) h-BN was reported to be 360 W/mK (390 W/mK for bulk h-BN) and 250 W/mK in 5 layer h-BN due to the higher polymeric residue contamination.84 The high in-plane thermal conductivity and insulating nature of BNNS have been used to prepare thermally conductive nano-oils containing BNNS nanofiller.85,86 The thermal conductivity of BNNS has also been utilized in polymeric composites to make highly thermally conductive composites.76,78,80 A 14-fold increase in thermal conductivity was observed in an electrically insulating epoxy/BNNS composite.87,88

2.2.2.4 Optical Properties Few-layered BNNS is highly transparent, with transparency over 99% from the near ultraviolet to the visible light range (250 – 900 nm).50 Highly transparent, high strength polymer composite was fabricated using exfoliated few-layered transparent BNNS as a filler.76 At a very low concentration of functionalized h-BN (0.01 mg/ml) in tetrahydrofuran (THF) solution, the Tyndall effect was observed from scattering of a laser beam by BNNS.89

Sharp absorption of BNNS is observed below 250 nm. For monolayer BNNS, the absorption occurs at ~200 nm wavelength, while for the few-layer BNNS, the absorption wavelength shifts to more positive direction.50 An exact correlation based on mono- to few-layer/bulk BNNS and its ultraviolet absorption phenomenon has not been reported, however. Monolayer h-BN has exhibited a relatively higher band gap (6.07 eV) than multilayer/bulk h-BN due to the absence of interlayer interactions.90

2.2.2.5 Stability in air Similar to its bulk form, few-layered h-BN has been found to be stable in an air atmosphere at very high temperatures (up to 850 °C in air).91 Oxidation of monolayer h-BN starts at 700 °C, and it is totally burned out when the temperature reaches 800 °C.91 In one report, 5 nm thick BNNS was reported to resist oxidation up to 1100 °C on Ni substrate and up to 500 °C on a Cu substrate in 300 mTorr oxygen atmosphere

for 30 minutes.92 Considering its honeycomb structure that is impermeable to small molecules, chemical inertness, high-temperature stability in air, and insulating nature, BNNS could be very stable at room temperature in air. However, oxygen reduction reaction at room temperature on top of nitrogen rich defects of the monolayer h-BN has been theoretically predicted.57 The large number of defects formed during the growth process thus makes it worthwhile to study the long-term stability of BNNS in a room temperature air environment.

2.2.2.6 Chemical properties h-BN is inert to both strong acidic and strong basic solutions.93 Due to its inertness, it is difficult to functionalize94, while its counterpart graphite can be easily converted to graphene oxide through the Hummers95 and its modified methods.96 BNNS, having a pore diameter of only 1.2 Å in the hexagons97, has been demonstrated to be impermeable to argon atoms.98 Therefore, BNNS has a good potential to be utilized for anti-corrosion, where the thickness, aesthetics, and stability of the underlying substrate are important in a harsh environment.

2.2.2.7 Sliding and frictional characteristics Its lamellar structure, weak van der Waals interaction between the layers, strong in- layer bonding, high mechanical strength, chemical inertness, and stability from sub- zero to very high temperatures have made h-BN a popular lubricating agent on the bulk scale.60 On the 2D scale, however, where the layer numbers are reduced from billions to a few tens, the tribological characteristics are yet to be fully discovered. With a view to incorporation in the microelectromechanical and nanoelectromechanical systems (MEMS and NEMS, respectively), BNNS has been studied both as a solid and as a liquid lubricant additive. As a solid lubricant on top of Cu substrate, mostly monolayer of h-BN is found to reduce the friction of the Cu substrate by a factor of 40.99 As a liquid lubricant, dispersed mono- to few-layer thick BNNS in water was used to reduce the friction and wear of SiC ball bearings sliding on silicon wafer.100 Formation of tribofilms containing BNNS on the wear tracks was responsible for the improvement of the tribological performance of the dispersion.100 Moreover, in-depth AFM studies considering mono- to few-layer 2D materials

(graphene, h-BN, MoS2, NbSe2) have revealed an increase in friction at the tip-

nanosheet interface when the layer number decreases.101 The low bending stiffness of the 2D material compared to its in-plane stiffness can cause the sheet to pucker when the tip approaches the sheet (Figure 2.6). These out of plane puckered regions contribute an additional barrier energy against tip sliding, contributing to increased friction for 2D materials with reduced layer numbers.101

Figure 2.6: The proposed puckering effect in 2D materials (using graphene as an example), where the adhesion to the sliding AFM tip causes out-of-plane bending of the nanosheet, leading to increasing contact area and friction. Adapted from Ref101.

2.2.2.8 Other properties With vertically aligned surface morphology, BNNS was found to be superhydrophobic, reaching a contact angle of 150° with water droplets.19 The ultra- high surface area and aspect ratio of porous h-BN nano-belts have been utilized in hydrogen storage application.40 Recently, BNNS has been used as an effective water cleaning agent (taking advantage of its large surface area) to adsorb oil from an oil- water solvent mixture.102

Summarizing all the properties of BNNS, such as low density, high strength, large direct-band-gap energy, low dielectric constant (4.6), high thermal conductivity, resistance to high temperature oxidation, ability to act as a high ultraviolet emitter, and high resistance to wet chemical attacks in both acid and basic environments, more utilization of this material is expected in the near future; and it might be the next-

generation material for electronic device fabrication and applications in harsh environments.

2.3 Synthesis of BNNS Boron nitride cannot be found in nature and is a synthetic compound with some remarkable properties. h-BN was first synthesized in 1842 by the reaction of boric 103 oxide and potassium cyanide. The commercialization of h-BN powders was initiated almost 100 years after its first synthesis. Nowadays, the most popular commercial method of synthesizing h-BN powders is by heating boric acid/boric oxide and an ammonia/melamine/urea mixture at 900 °C, followed by annealing at 1500 °C in N2 atmosphere to increase the crystallinity of the powders:59,104

B(OH)3/B2O3 + NH3 → BN + H2O………………………….……………….. (2.1)

B2O3 + NH2 − CO − NH2 → BN + CO2 + H2O……………………………… (2.2)

Interest in the synthesis of BNNS started immediately after the discovery of graphene in 2004. Nowadays, the most common synthesis route for BNNS is by chemical vapour deposition (CVD). A summary of the BNNS synthesis routes that have been generally adopted is provided in Figure 2.7.

Figure 2.7: Commonly used methods for the synthesis of BNNS.

2.3.1 Top-down approaches For the top-down approaches, the nanosheet size and cleanliness is limited by the pristine powder size and purity.

2.3.1.1 Micromechanical cleavage A simple piece of scotch tape was used to repeatedly peel off layered materials

(graphite, BN, MoS2, NbSe2, and Bi2Sr2CaCu2Ox) down to mono- to few-layer nanosheets10. Novoselov and Geim first published the paper in 200510 after their report of the discovery of graphene in 2004 by the same method.9 The peeled-off 2D layered nanosheets were transferred to 300 nm SiO2 coated silicon wafers for visualization under an optical microscope.

One of the primary advantages of this technique is the ease of stacking other layered materials with h-BN. Taking advantage of the ultra-flatness and high purity of the mechanically transferred BNNS, BNNS-graphene heterostructures were prepared, where obstruction and scattering of the conduction electrons through graphene was minimal at heterostructure interfaces. As a result, improved mobility was found in graphene lying on boron nitride substrate compared to that of graphene on SiO2-coated silicon substrate.19 Flexible field effect transistors (FFT) prepared using mechanically exfoliated BNNS and graphene have shown exceptional room temperature carrier mobility18,71,105 and flexibility.71 The cut-off frequency ratio approached near-unity (0.9) in these devices.71

Instead of the direct peeling applied in the scotch-tape method, shearing forces were employed in the ball-milling process to exfoliate BNNS from bulk h-BN powders (Figure 2.8).106,107 Direct contact of the h-BN powders with the grinding balls can be avoided by carrying out the process in a solvent medium, which improves the efficiency of the process and the quality of the nanosheets.106 Ball milling without using a solvent generates a large numbers of defects and impurities in the nanosheets.108–110

Figure 2.8: Schematic illustration of two exfoliation mechanism in the ball-milling process, (a) peeling off layer-by-layer, and (b) crumpling of the sheets under a shear force in the ball-milling process. Adapted from Ref106.

2.3.1.2 Sonication-assisted exfoliation Sonication in common organic solvents was found to be an effective route to overcome the van der Waals forces between the layers in highly oriented layered materials. An extensive study of sonication-assisted exfoliation was conducted by the Coleman et al., where they exfoliated h-BN, MoS2, WS2, MoSe2, MoTe2, TaSe2, NbSe2, NiTe2, and Bi2Te3 in a wide range of solvents, and the best 20 solvents for h-BN, MoS2, and 11 WS2 exfoliation were identified. The amount of nanosheet exfoliation and retention in the solvents was found to be dependent on the solvents’ surface tension value, and better exfoliation was seen for 40 mJ/m2. Several reports were published afterward on exfoliating BNNS by intercalating ions and compounds inside the layers.111–114 Molten hydroxides of NaOH and KOH were used to exfoliate h-BN powders down to mono- to few-layers BNNS by reaction, curling, cutting, and peeling mechanisms.115 Even water was found to be suitable for obtaining mono- and few-layer BNNS by sonication assisted hydrolysis of boron nitride bulk powders.116 Besides conventional solvents, BNNS were exfoliated in 1,2 dichloroethane solution of poly[(m-phenyl-enevinylene)- co-(2,5-dictoxy-p-phenylenevinylene)]117 (1.2 mg polymer in 10 ml solution) by ultrasonication.

Although BNNS are very inert towards acid and basic attacks, functionalization of the BNNS has been attempted to fully utilize its properties in polymer composites. Liu et al. used block copolymers [(P(S-b-MMA)), where PS is polystyrene, PMMA is methylmethacrylate and b represents the double block polymer], to interact with the exfoliated BNNS.118 The different solubility of polystyrene and PMMA in the block copolymer broadened the dispersion range of the nanosheets even where the Hansen solubility parameters were mismatched. For example, BNNS can be dispersed in

acetone and toluene using the block copolymers, although acetone and toluene are considered unsuitable for BNNS dispersion. In another work, BNNS were ball milled at high energy to incorporate mechanical defects. These defects were utilized to effectively exfoliate and functionalize the nanosheets by gold nanoparticles.119 Amine functionalized exfoliated stable BNNS dispersion in organic solvents and water was achieved using octadecylamine (ODA) for the functionalized molecules.89 Functionalization was achieved due to the strong Lewis acid-base interaction between electron deficient boron and electron rich amine molecules. In a recent work, oxygen radical functionalization of BNNS was also reported.94 Firstly, the exfoliated BNNS in N-methyl-2-pyrrolidone (NMP) were tertbutoxy functionalized using di-tert-butylperoxide reagent in a high- pressure autoclave at 120 °C for 12 hours. In the second stage, tertbutoxy functionalized

BNNS was reacted with piranha solution (H2SO4: H2O2, 3:1) to make hydroxyl functionalized BNNS. Nanocomposites prepared by the addition of 0.1% hydroxyl- functionalized BNNS to PVA showed very high elastic modulus, strength, elongation, and toughness values compared to bare PVA and non-functionalized BNNS incorporated in PVA. All the properties were enhanced due to effective load transfer in functionalized nanosheets by cross-linking of hydroxide ions with PVA and nanosheets.

2.3.1.3 Electron beam irradiation The micromechanical cleavage technique was used to create few-layer BNNS. Further thinning down to monolayer BNNS was accomplished under a high energy electron beam inside a TEM, where selective sputtering of B atoms was detected at 80 kV and 120 kV operating voltages.58,120 After the initial boron monovacancy formation in the nanosheet, more nitrogen and boron atoms vacancies are created gradually, and the whole layer under the electron beam is gradually removed. Through this process, in- situ monolayer BNNS can be achieved. With an HRTEM and exit wave (EW) reconstruction, it was possible to identify individual B and N atoms in HRTEM images.

2.3.2 Bottom-up approaches

2.3.2.1 Chemical vapour deposition Chemical vapour deposition (CVD) is a synthesis technique used to grow a thin solid film from gas source precursors. The deposition occurs when the precursor in the form of atoms or molecules, or a combination of the two is adsorbed and then coalesces on a substrate. The sequential events in the CVD process are shown in Figure 2.9.121,122

Figure 2.9: Sequence of events during CVD growth process.121,122 CVD is a complex system to optimize for the desired outcome, as interrelationships exist among the large number of experimental parameters, process variables, and coating properties. A summary of all of these parameters is shown in Table 2.2. Schematic interrelationships among these parameters are shown in Figure 2.10.

Table 2.2: Factors and properties in the CVD growth process123 Experimental Parameters Process Variables Coating Properties 1. Raw material 1. Gas temperature 1. Morphology 2. Input gas composition 2. Flow behaviour 2. Composition 3. Total flow rate 3. Mean gas velocity 3. Thickness 4. Pressure 4. Gas chemistry 4. Density 5. Temperature 5. Gas homogeneity 5. Uniformity 6. Selection of substrate 6. Deposition rate 6. Adhesion 7. Substrate surface and 7. Nucleation density 7. Crystallinity environment 8. Stress 8. Duration of deposition 9. Precursor purity and decomposition temperature 10. Precursor flow rate 11. Cooling rate 12. Experimental procedure

Figure 2.10: Schematic diagram of the interrelationships among experimental parameters, process variables, and coating properties.123

The synthesis of a thin film of h-BN by the CVD process can be dated back to the early 1970s93 after the CVD process became popular for the preparation of thin film samples.122 Several reports were published afterwards, where micron-thick h-BN was prepared on various substrates.124–128 Single layer h-BN was first reported to be synthesized by CVD in an ultra-high vacuum (UHV) system in 1995.7,8 Due to the limitations of the UHV system and the complexity of the system, however, the synthesis of atomic-layer-thin h-BN did not gain enough attention until 2005.10 The CVD process became an important synthesis method later on due to the ease of

fabrication and capability to produce large thin films through low pressure (LP) - and atmospheric pressure (AP)-CVD synthesis routes. 2.3.2.1.1 Precursors for BNNS synthesis One typical precursor for the CVD synthesis of BNNS is borazine (Figure 2.11a). Borazine is a toxic liquid, however, unstable in air at room temperature because it reacts with the moisture and often polymerizes to form solid polyborazylene.129 Therefore, it needs to be stored in cold conditions under an inert atmosphere, which limits its use as a direct precursor for synthesis. One of the indirect precursors, which gives off borazine when heated and is more stable than borazine, is β-trichloroborazine (Figure 2.11b).130,131 The major drawback of this precursor, however, is the release of HCl gas when it is exposed to humid air, which is corrosive to the CVD fittings and equipment. Ammonia borane is another popular indirect precursor. Ammonia borane is air-stable, non-toxic, and easy to handle. Ammonia borane decomposes to borazine, along with hydrogen, polyaminoborane, polyiminoborane, diborane, and ammonia, at around 110 °C.132,133 The solid decomposition products from ammonia borane can be prevented from reaching the reaction zone by decomposing ammonia borane outside the CVD furnace with a valve-controlled gas flow and by using a filter.22,50,90,134–136 Wang et al. developed an interesting chemical blowing process, where ammonia borane in a multistage heating process was converted into bubbles.137,138 At high temperature (> 1200 °C), these bubbles collapsed due to the release of the hydrogen trapped inside and formed BNNS.

Apart from using a single source precursor, multi-gas source precursors also have been used for the synthesis of BNNS. For example, BF3, N2, and H2 have been used to synthesize vertically aligned BNNS by microwave plasma (MP)-CVD process.139–141 The vertical alignment of the nanosheets on silicon wafers was likely induced by the electric field generated in the plasma sheath. The nanosheet thickness was controlled by varying the gas ratios. Due to the etching of the resultant BNNS by fluorine, the flow of hydrogen gas was adjusted to form HF in the system. Following a similar 142 mechanism, BCl3, NH3, N2, and H2 gasses were used to synthesize BNNS. Diborane and ammonia as two gas source precursors have been commonly studied for the synthesis of thin films of h-BN.93,124–128 These precursors were recently used in an LP- CVD system to synthesize few-layer BNNS on Ni and Cu foil.143 BNNS grown on Cu

foil was found to be turbostratic, while crystalline BNNS with controlled thickness was found on Ni foil. As two-step precursors, decaborane and ammonia have been also used to synthesize few-layer BNNS.144 Decaborane was used to take advantage of its stability in air and easy sublimation between room temperature to 90°C, with no other decomposition products. In a three-step boration, oxidation, and nitration process, where BNNS was synthesized on Rh/yttrium stabilized zirconia (YSZ)/Si (111) substrate, B(OCH3)3 and NH3 were used as the boron and nitrogen source, 145 respectively. Boration by introducing B(OCH3)3 gas, followed by oxidation (by introducing oxygen into the UHV system) brought the inner diffused boron to the surface and formed boron oxide on the surface of the Rh substrate. Flowing ammonia converted the thus-formed boron oxide into BNNS (Figure 2.11c).

Figure 2.11: (a) Structure of borazine, (b) structure of β-trichloroborazine, and (c) synthesis of BNNS by three step of boration, oxidation, and nitration process.145

2.3.2.1.2 Substrates for BNNS synthesis The growth of BNNS has been primarily carried out on transition metal substrates, due to their catalytic influence at high temperatures for easy adsorption of borazine. Beside the transition metals, BNNS growth has been also reported on Ge146, sapphire125,147, 148 149 Si and SiO2 coated Si substrate. A list of substrates that have been employed to grow BNNS are summarized in Table 2.3. 24

Table 2.3: List of substrates used as catalysts for BNNS growth.7,8,21,22,50,90,125,130,143–171

Substrates for BNNS growth Period Group 7 8 9 10 11 12 13 14 6 Others

3 Si148,156 Cu-Ni alloy171, 4 Fe165– Ni7,8,130,1 Cu21,22,50,90,144,15 Ge146 Cr170 Co160 Sapphire125 168 43,150–153 1,157–159 ,147 , SiO2 5 Rh145, Ru161 Pd172 Ag154 coated 169 –163 6 Pt131,135,1 Si149 Ir164 Au155 36

2.3.2.1.3 Theoretical studies of BNNS growth on transition metal substrates Like graphene, the most widely used catalysts for the BNNS growth are Cu and Ni. BNNS are reported to be strongly chemisorbed on Ni (111) and weakly chemisorbed/physisorbed on Cu (111).173–175 This is due to the strong hybridization of 2 the Ni dz orbital with N-pz and B-pz orbitals, while in the case of Cu, the hybridization energy is much less due to the filled d orbitals in Cu.173,174 Thus, the binding energy of BNNS grown on Cu/Ag/Au substrates is much smaller than that grown on Co/Ru/Ir. Theoretical calculations confirm that the binding energy decreases with the filling of the d band orbitals in the transition metals and is highest for the 4 d elements (i.e., Period 5 elements).174

The positions of boron and nitrogen atoms on the Ni (111) surface were theoretically analysed.176 Boron atoms preferentially enter into the face centred cubic (f.c.c)/hexagonal close packed (h.c.p) hollow sites of Ni (111), and nitrogen sits on top of each Ni atom for the formation of stable corrugated BNNS on top of Ni (111).174,176 The f.c.c position is slightly energetically more favourable (9 ± 2 meV/boron atom) for the boron atoms than the h.c.p hollow sites.176 In a later publication, the stable boron and nitrogen atomic positions on Co (0001), Ni (111), Pd (111), Pt (111), Cu (111), Ag (111), and Au (111) were established. B and N atoms on Co and Cu occupy the same atomic positions as was observed on Ni. On the other transitions metals, three boron and nitrogen atoms sit on the bridge sites and one of each atom sits on top of nitrogen alternatively to form stable h-BN (Figure 2.12a). 25

Figure 2.12: (a) Atomic positions of boron and nitrogen in a stable h-BN monolayer sheet on Co (0001), Ni (111), Cu (111) (Left image, 1×1 h-BN unit cell on top of a 1×1 metal surface cell), and Pd, Pt, Ag, and Au (111) (Right image, 2×2 h-BN unit cell on top of a √3×√3 metal surface cell). Adapted from Ref175. (b) Energy vs lattice parameter plot, comparing equilibrium h-BN energy with h-BN sitting on other (111) oriented transition metals. Adapted from Ref174. (c) B-rich and N-rich triangles of size L = 10. (d) Total energy in a series of triangles shown as a function of their size L, red for B-rich and blue for N-rich. Adapted from Ref177.

The lattice mismatch of Co (0001), Ni (111), and Cu (111) with h-BN is only 0.8%, - 0.4%, and 2.4% respectively [Lattice parameter of Co (0001) = 2.51 Å, Ni (111) = 2.49 Å, Cu (111) = 2.56 Å, and h-BN = 2.50 Å]. This facilitates commensurate epitaxial BNNS growth on Co (0001), Ni (111), and Cu (111) surfaces. In contrast to these metals, there is a considerable lattice mismatch of h-BN with all the other transition metal elements, thereby forming non-commensurate corrugated nanomesh patterns on those elements. The Moiré periodicity generated by the lattice mismatch between the substrate and BNNS is calculated according to this formula174:

푁BN × 퐴BN = 푁subs × 퐴subs……………………………………………………...…………………...... (2.3)

Where, NBN = No. of unit cell repetitions in h-BN

Nsubs = No. of unit cell repetitions in the substrate

ABN = Lattice parameter of h-BN

and Asubs = Lattice parameter of the substrate

As for example, Ru (111) and Rh (111) have almost the same in-plane lattice parameter, 0.271 nm and 0.270 nm, respectively. From Equation (2.4), it can be observed that 13×13 unit cells of BNNS coincide with 12×12 unit cells of Ru (111) / Rh (111). Therefore, the Moiré periodicity in the Ru (111) / Rh (111) lattice will be the length of the 13×13 unit cells of BNNS. Similarly on Pt (111), 10×10 unit cells of BNNS coincide with 9×9 unit cells of Pt (111). Therefore, h-BN will be more highly strained on the non-commensurate surfaces than on the commensurate ones (Figure 2.12b).

The growth of BNNS generally forms a triangular pattern on the transition metal substrates. This triangular nature of the BNNS can result in a chemically unbalanced state along the edges, i.e., either boron terminated or nitrogen terminated zigzag edges. From the theoretical calculations, it was found that nitrogen termination of the triangular BNNS is more favorable on Cu substrate than boron termination due to the lower edge energy formation by nitrogen terminated edges (Figure 2.12c).130 With increasing boron concentration, nitrogen-terminated triangular BNNS changes to truncated triangles or hexagons, having both boron and nitrogen terminated edges.159,178 In case of BNNS on Ni substrate, boron terminated edge structure can be energetically favorable.178 Experimental evidence, however, is yet to be reported for this case.

2.3.2.1.4 BNNS on polycrystalline transition metal substrates Although much importance was given to commensurate and non-commensurate growth of BNNS on single crystal transition metals in the earlier studies, highly crystalline mono- and few-layer BNNS have been synthesized in recent years on mostly polycrystalline Cu, Ni, Fe, and Pt foils.135,136,144,150,179,180 As a result of the weak chemisorption and the polycrystalline nature of the foils, triangular BNNS islands having random orientations are often found during the initial growth stage on polycrystalline Cu and Pt.90,136,181 Even different orientations of the triangular BNNS

islands within the same grain of Cu have been observed, which is likely due to the weak interaction of the BNNS with the Cu substrate.90 In contrast, the BNNS islands grown on Ni and Ru substrates maintain specific orientations with respect to the substrates due to the strong chemisorption of BNNS on those substrates.130,163

2.3.2.1.5 Pre-treatment of the substrate Commercially available rolled polycrystalline thin film (0.025 mm to 0.25 mm) foils are generally used to synthesize BNNS. These as-received foils are pre-polished prior to their introduction inside the CVD furnace to remove the oxide layer and organic contaminants from the surface. A short acetic acid wash has been found to be effective to remove the oxidized layer from the Cu surface.158 Acetone, isopropyl alcohol (IPA), and water rinsing are used to remove any organic substance contaminations.182 The most widely used technique, however, is to treat the Cu or Ni foils in dilute nitric acid for a very short period to remove the oxidized layer.50,183 Electrochemical polishing and Ar ion etching are the other substitutes for pre-polishing the foils before the synthesis step.184–188

2.3.2.1.6 Mechanism of BNNS growth Ammonia borane currently is the most widely used precursor for the growth of BNNS.50,90,136,159,171,181 It is usually assumed that during the decomposition of ammonia borane at around 110 °C (or by the reaction of NH3 and B2H6/B10H14 inside the furnace at high temperature), borazine is formed and flows into the reaction zone, where it is adsorbed on the catalyst surface, diffuses and polymerizes on the surface, and is transformed to BNNS on the substrate surface by the dehydrogenation reaction at high temperatures.21,152,189 In addition, borazine itself can decompose into boron and nitrogen when it comes into the vicinity of the substrate.190 Boron has sufficient solubility in Cu at high temperatures (0.29 at% at 1013 °C191) to diffuse towards the Cu, while nitrogen atoms are expelled from the surface, considering the very low solubility of N in Cu. The rest of the boron and nitrogen flux in a 1:1 ratio forms BNNS on the substrate (Figure 2.13a). B atoms can be diffused back towards the substrate at high temperatures which in presence of nitrogen, form multilayer islands of 168,190 BNNS. By pre-feeding the substrate (Fe) with nitrogen atoms (flowing NH3 at the beginning of the growth process), the diffusion of the boron atoms from the

decomposed Borazine can be restricted and island free, uniform monolayer growth of BNNS has been reported though this technique.168

Figure 2.13: Schematic illustration of BNNS growth on Cu by the CVD process. Ji is the impingement flux of borazine, JB is the diffusion flux of B into the Cu substrate,

JN is the diffusion flux of N out of the Cu substrate, and JB´ and JN´ are the flux of B and N atoms to form h-BN on Cu. Adapted from Ref190. (b) Three widely used thin film growth models, as a function of coverage θ in monolayer. (1) Volmer-Weber, (2) Frank-van der Merwe, and (3) Stranski-Krastanov growth models. Adapted from Ref192.

The multilayer growth of BNNS is often explained by the three established thin film growth models (Figure 2.13b).8,90,158,162,192,193 If the atoms of the deposited material are much more strongly bounded to each other than to the substrate, island or Volmer- Weber growth results (case 1: Figure 2.13b). In contrast, if the atoms are more attracted towards the substrate than among themselves, the growth occurs in a layer-by-layer manner, i.e., the growth follows the Frank-van der Merwe model (case 3: Figure 2.13b). According to the Stranski-Krastanov growth model, the growth follows the 29

Frank-van der Merwe model first (layer-by-layer growth), and after that, the growth switches towards the island-like growth (case 2: Figure 2.13b). Most of the BNNS growth models have been explained with the Stranski-Krastanov model.8,90,158,162

2.3.2.1.7 Transferring BNNS to arbitrary substrates Transfer of the CVD grown BNNS onto arbitrary metal substrates has been achieved with a support of PMMA/polydimethylsiloxane (PDMS)/thermal release tape/water soluble polymers.50,90,152,194–199 PMMA is the most widely used polymer, used to transfer the nanosheets (Figure 2.14).21,50,152,200 As the deposition occurs on both sides of the substrate, one side of the substrate is either cleaned using acid etching or with Ar ion sputtering. Then, PMMA is spin-coated on the upper side of the deposited 21,144,159,201 BNNS on Cu/Ni and the underlying metal is etched away in FeCl3 / 202 21,180 203 Fe(NO3)3 / transene / ammonium persulfate solution. The floating BNNS with PMMA is then scooped out to a silicon wafer for subsequent optical, Raman, and AFM analysis. PMMA can be removed by washing with hot acetone followed by

Ar/H2 annealing at 350-450 °C for 2-4 hours. This transfer process, however, often leaves PMMA residue all over the sample. Heating at 500 °C under Ar/O2 gas atmosphere has been found effective for complete removal of PMMA residues from the BNNS.204 Also, the acetone washing step can be avoided by burning the PMMA coated sample in air at 350 °C for 3-4 hours, just after scooping from the solution.205 Burning away the PMMA layers save several wet processing or cleaning steps and can avoid any unwanted contamination from washing with acetone or IPA. In a very recent work, electrostatic force has been used to transfer graphene from the Cu substrate using an electrostatic generator.199 Similar methods can be adopted to transfer the nanosheet to a TEM grid. Another widely used technique is the method used for direct graphene transfer to a TEM grid (Figure 2.14). The TEM grid is kept on top of the graphene/Cu foil and a few drops of IPA is added on the TEM grid to adhere it with the graphene/Cu foil interface. After etching the underlying Cu, the graphene sticks to the TEM grid and floats on the etching solution.196 This method can be also used for BNNS transfer to TEM grids.206

Figure 2.14: Two commonly adopted techniques for graphene or BNNS transfer to TEM grids. Adapted from Ref196.

2.3.2.2 Physical vapour deposition (PVD) PVD is another bottom-up synthesis process, where materials are vaporized from a solid or liquid source in the form of atoms or molecules and are transported towards a substrate under a UHV/low-pressure gas (plasma) environment for deposition.207 The vapor generated from the source material is generally carried out by two techniques: thermal evaporation and sputtering. In the thermal evaporation process, the source material is heated up by a focused high energy electron beam to vaporize the material. Molecular beam epitaxy, a PVD process, is an advanced form of thermal evaporation. Sputtering is generally carried out by bombardment using accelerated gas ions (typically Ar+).

Radio frequency magnetron sputtering was employed on a boron target at 850 °C in

Ar/N2 atmosphere to synthesize monolayer BNNS on Ru (0001). Single crystalline 161 100 nm thick Ru (0001) film was prepared in situ on α-Al2O3 (0001). Also, the pulsed laser deposition208 and molecular beam epitaxy209 techniques were employed for synthesizing BNNS on various substrates, including, Ni, Mo, Si, and AlN.

2.3.3 Other processes Apart from the conventional bottom-up or top-down approaches, BNNS was synthesized by substitutional reaction of graphene.210–213 To synthesize the nanosheets, boron trioxide powders were covered with molybdenum oxide (to act as a promoter), and graphene sheets, and all those materials were put into a graphite furnace at 1650 210 °C under flowing N2 gas. BN and BCN nanosheets were synthesized. Oxidizing the as-obtained product at 600 °C lowered the carbon content in the sample, and high purity BNNS could be obtained. Further modification of the process by using diverse carbon sources (including any vegetation, fleabane flowers, pine needles, wiper papers, etc.) resulted in gram scale fabrication of BNNS.88 Epoxy composite using this gram scale BNNS filler demonstrated a 14-fold increase in the thermal conductivity. Alternatively, BNNS to graphene conversion was also possible by flowing methane gas over BNNS deposited on Pt at high temperatures.214 B and N atoms are believed to be etched away gradually to form BH3 and NH3 gases, while the vacant positions are immediately occupied by C atoms to make graphene.

In another process, multiwalled boron nitride nanotube (BNNT) was unwrapped to make boron nitride nanoribbons.215 For the unwrapping process, previously synthesized multiwalled BNNT was first transferred to a silicon wafer. The top side was then spin-coated with PMMA. The PMMA/BNNT surface was then peeled off from the substrate in KOH solution. Under these conditions, the bottoms and sides of BNNT were coated with PMMA with one surface exposed, which was then plasma etched with Ar ions to open up the nanotubes. After removing the PMMA layer by acetone and exposure to 600 °C in oxidizing atmosphere, clean boron nitride nanoribbons can be obtained.

Synthesis of BNNS by chemical reactions in solution has been tried by several researchers. The first paper reported that the synthesis of BNNS by chemical routes involves the reaction of boric acid and urea in water at 65 °C, which forms a complex in the stirred solution. After evaporating the water, the formed solid powder was heated at 900 °C to obtain few-layer BNNS.216 A mixture of boron trioxide and guanidine chloride in methanol was used to produce porous BNNS at 1100 °C, which was utilized for the absorption of oil, organic solvents, and dyes in water.102

BNNS, 4 nm in thickness on average was synthesized by mixing B2O3, zinc powders, 217 and N2H4∙2HCl in a sealed autoclave at 500 °C. In another process, NaBF4, NaN3,

NH4Cl, and S were mixed under 450 MPa pressure to form a pellet. This was then kept in the autoclave at 300 °C for 1 hour to synthesize h-BN of a few nanometer thick.218 Nevertheless, all the attempts to prepare BNNS by wet chemical routes encounter difficulty in terms of controlling the nanosheet thickness and purity.

2.3.4 Limitation of the synthesis routes and challenges for application Although several synthesis routes have been identified to synthesize mono- and few- layer BNNS, methods for production of large-scale and high-quality single crystalline nanosheets for practical use are still not available. With the top-down approaches, the nanosheet size and cleanliness are limited by the pristine powder size and purity. In the micromechanical cleavage technique, nano- to micro-meter size nanosheets can be peeled off from the pristine highly oriented pyrolytic boron nitride samples. This process has extremely low output, however, and both mono- and few-layer nanosheets are formed at the same time. This method cannot be scaled up for industrial production. In the sonication assisted exfoliation, defective and small-sized nanosheets are often found. Functionalization of the nanosheet prior to exfoliation can increase the efficiency of exfoliation and dispersion; although it generally induces defects in the nanosheet. Also, using the ball-milling process to exfoliate nanosheets creates defects and possible foreign-element contamination in the nanosheets. Small-size nanosheets are obtained, which also depend on the pristine h-BN powders.

Electron beam irradiation and PVD methods have been utilized to synthesize BNNS, although these synthesis routes are very expensive and only laboratory scale 33

production is possible with them. Other processes, such as BNNT unwrapping and the substitutional reaction of graphene, do not produce very high purity or large nanosheets.

There have also been several reports on chemical synthesis routes for BNNS. Leftover impurities in the nanosheets are the main concern with these wet chemical routes. The as-synthesized nanosheets often require very high-temperature annealing (1000-1500 °C) to improve the crystallinity and sometimes require oxidation to remove the carbon contamination from the nanosheet surface.29,210

Among all the synthesis routes, the CVD method has been established as a popular route to synthesize large BNNS. Several years back, 30-inch long graphene was synthesized by a CVD method through a roll-to-roll transfer technique, which demonstrated strong evidence of the reliability of this technique for large-scale production.198 Although BNNS has not been synthesized on such a large scale as yet, by utilizing the suitable gas flow of the precursor and carrier gases, as well as the right temperature, catalyst and CVD design, large-scale production of BNNS is possible. The major bottleneck to overcome with this technique, however, is to reduce the grain boundaries (so to form large single grains) and control the mono- or few-layer characteristics (homogeneity) all over the synthesized product in the CVD system. Also, suitable techniques need to be developed to avoid surface contamination, which is formed during the synthesis and/or during the transfer process.

2.4 Characterization tools

2.4.1 Optical microscope Optical microscopy analysis can be a very straightforward route to initially examine the number of layers of BNNS by visual observation. However, mono- to few-layer BNNS is not easily to differentiate by an optical microscope, as it hardly absorbs any visible light (nearly transparent) due to its large direct-band-gap (~5.97 eV) properties. Even using a very smooth conventional 300 nm silica coating layer on top of the silicon wafer to increase the interference of reflected light in the presence of the nanosheet does not add sufficient contrast to detect a monolayer nanosheet. Monolayer BNNS show 1.5% contrast (which comes from the difference in optical path between the incident and reflected light) under white light in 300 nm SiO2 coated silicon, which is undetectable by the human eye.219 The contrast difference increased to 2.5% for monolayer h-BN on top of 80 nm SiO2/Si wafer under white light, and 3.0% using a green filter in the white light path (Figure 2.15a).219 With an increasing number of layers, the contrast also increases, thereby facilitating the differentiation of different layers under an optical microscope. In fact, in one study, 1-80 layers of h-BN showed a gradual change in contrast under green light illumination (516 nm wavelength) in an optical microscope.220

Spin coating liquid crystals (4-pentyl-4′-cyanobiphenyl (5CB); 4-octyl-4- cyanobiphenyl (8CB), etc.) on top of BNNS has proven to be another successful method to directly identify the grains, grain boundaries, and cracks of BNNS by polarized optical microscopy imaging (Figure 2.15b).180,221,222 The method was first adopted for graphene223,224 and later on, applied for BNNS.180,221,222 Liquid crystals have been found to be aligned anisotropically with the domain boundaries of BNNS and exhibit distinct birefringence when imaged under polarized light.223,224 Identification of the thickness of BNNS however, is difficult through this technique. Also, the grain boundaries of the growth substrate (Cu) were visualized after transfer180,223, correlating with false BNNS or graphene grain boundaries, which may limit its further use.

Figure 2.15: (a) Optical contrast change of mono- (blue line) and bi-layer (red-line) region with different wavelengths of light. The contrast is positively maximum at ~ 570 nm wavelength. The figures below (a) shows the contrast variation of mono- and bi-layer h-BN under different wavelengths of light. Adapted from Ref219. (b) and (c) Schematics of liquid crystal method and oxidation methods respectively, for the direct identification of graphene on Cu substrates. Similar techniques have been developed for identification of BNNS. Adapted from Ref223,225.

Exposing the graphene coated Cu substrate to ultraviolet (UV) light to a moisture rich environment can oxidize the underlying Cu substrate along the graphene grain boundaries and defects (Figure 2.15c).226 Hence, direct optical identification of polycrystalline graphene on Cu substrate is possible through the UV treatment. In a parallel timeframe, a more convenient method was demonstrated for direct identification of graphene by simply oxidizing the graphene deposited on a Cu substrate at 160 °C for 6 min225, which oxidized the underlying Cu substrate along with the graphene defects. This increased the interference contrast between the oxidized and non-oxidized Cu substrate, which facilitated the direct identification of graphene areas under an optical microscope. Samples immersed in 15 % H2O2 aqueous solution or a saturated solution of sulfur in ethanol also yielded a similar result. The oxidation treatment has been generally applied towards BNNS or graphene for their direct identification on Cu substrate immediately after synthesis.

2.4.2 Raman microscope Raman spectroscopy has been widely used to identify compounds, layer numbers, 227–230 crystallinity, and defects. Crystalline h-BN shows two Raman active E2g peaks at 1366 cm-1 and 49-52.5 cm-1. Both the peaks are due to in-place atomic vibrations. The peak at 1366 cm-1 is due to the B and N atom stretching vibrations against each other, while the other peak appears when the h-BN planes slide against each other.230

The appearance of the high frequency E2g Raman vibration mode is the most straightforward route to identify the h-BN. Monolayer h-BN gives a very weak Raman band. With an increasing number of layers, the Raman peak intensity also increases proportionally. Gorbachev et al. identified mono-, bi- and few-layer BNNS with a 514 nm micro-Raman spectrometer.219 The Raman peak of monolayer BNNS was blue shifted up to 4 cm-1 and was observed at 1370 cm-1. It was inferred that the monolayer h-BN was stretched by the underlying substrate, which develops strain in the nanosheet and consequently hardens the E2g phonons. In the bi-layer and few-layer cases, a red shift of 1-2 cm-1 was found. This could be due to the interaction between neighbouring 219 sheets, which can soften the E2g phonons. Heat generated from the high-intensity laser source can also soften the phonons in the presence of defects in the nanosheet and produce red shifting of the Raman peak. Defects (vacancy/impurity/c-BN/BN soot/BxCyNz compounds) restrict the heat flow and therefore, in its presence, the temperature can rise in the defective areas, which softens the E2g peak. In the absence of defects in the sheet, however, the temperature does not play a major role in peak shifting, as boron nitride itself is highly thermally conductive. No overtone of the E2g mode at higher frequency was detected.

th The low-frequency E2g mode is hard to distinguish, as the intensity is only 1/40 of that of the high-frequency vibration mode.227 Also, due to high Rayleigh background scattering and the detection limitations of most of the Raman systems231, this feature is mostly neglected, and only the intensity and peak position of the high-frequency

Raman mode are analysed. For the monolayer h-BN, the inter-plane E2g shear vibration mode should be absent, and this could be a direct indication of a monolayer.231 Another mode of vibration called the layer breathing mode has been observed in the graphene case at 128 cm-1 and the overtone observed at 256 cm-1 after careful set-up of the

Raman instrument.231 No layer breathing mode (LBM) has been observed for BNNS, however. This is possibly due to the different frequency of vibration of B and N atoms in a hexagonal ring.

2.4.3 Fourier transform infrared spectroscopy (FTIR) FTIR is an important tool to analyse boron nitride powder samples. An attenuated total reflectance (ATR)-FTIR arrangement is often used to analyse thick BNNS samples directly or in liquid solvents.94,116,132,208,232 FTIR is not very sensitive, however, for detection of mono- to few-layered BNNS on 300 nm SiO2/Si substrate. h-BN has two characteristic infrared (IR) modes. One is the in-plane B-N stretching vibration, E1u within the range of 1366-1392 cm-1, and the other is the out of plane B-N-B bending -1 116,232 vibration, E1u at 770-820 cm wavenumbers.

2.4.4 Scanning electron microscopy (SEM) Due to its minimal sample preparation requirements and its capability for micro- to nano-scale area imaging with high resolution, rapidity, and non-invasiveness, SEM is another commonly used characterization tool. The identification of BNNS, however, necessitates a conducting support substrate and operating the SEM under low kV and low working distance conditions.

Mono- to few-layer BNNS is considered transparent under high kV electron beam operations in SEM, as the interaction volume and consequently, the secondary electron generation from the nanosheet are very low compared to the substrate. Thus, the surface underneath the BNNS is imaged under high kV SEM operation. Also, BNNS is difficult to identify on top of insulating substrates (i.e., SiO2/Si). This is due to the charge build up on the insulating surface under electron beam irradiation, which repels further incidental electrons and restricts the secondary electron generation.

With an increasing number of layers of BNNS, SEM contrast also increases proportionally; which enables the rapid identification of BNNS layer numbers. The edges of the crumples in the nanosheet are seen as white contrast. For thick BNNS samples (boron nitride nanoplates on stainless steel), backscattered electron field 38

emission SEM (BSE-FESEM), combined with energy dispersive spectrometry (EDS), has been used for identification.34

2.4.5 Electron backscatter diffraction (EBSD) EBSD is a quantitative/statistical SEM technique for orientation measurement and phase analysis of microstructures on a macro to sub-micron or nanometer scale. EBSD analysis has been carried out on BNNS deposited on Cu or Ni substrate to understand the substrate grain orientation effect on the growth of BNNS.150,151,221,233,234 Both the Cu and Ni substrates undergo a significant recrystallization process when annealed at very high temperatures (at or above 1000 °C) and are found to be predominantly Cu{100} and Ni{100} surface oriented after the annealing and BNNS growth.150,233,235 In all the reported studies, the BNNS growth rate was found to be higher on the Cu{111} oriented regions than on the Cu{100} or Cu{110} regions.233 The growth of BNNS on Ni, however, follows the opposite trend. The BNNS growth rate was found to be higher on the Ni{100}- or Ni{100}-like surfaces.150,234 It should be noted that Cu(111) and Ni(111) have lower surface energies than their corresponding {100} and {110} surface planes.

2.4.6 Transmission electron microscopy (TEM) TEM is a state-of-the-art characterization tool which can provide information down to atomic resolution. The crystallinity of the BNNS is generally examined by selected area electron diffraction (SAED) analysis in TEM.50,152,157,165,167 The appearance of six-fold symmetrical spots from SAED confirms the presence of crystalline BNNS. Various domains of the nanosheet can be sequentially imaged under dark field (DF)- TEM mode along the specific SAED spots (using an aperture filter along that direction) arising from that crystal.205,236 The overall domain boundary mapped image is obtained by stitching all the DF images together.

Figure 2.16: (a) HRTEM image of few-layer graphene, which reveals that the few layer graphene has been creased. (b) An atomistic model of the creased few-layered graphene which gives rise to the HRTEM image in (a). Adapted from Ref237. (c) Atomic number, Z vs. ADF intensity plot, giving a perfect match between the experimental and theoretical profiles. (d) ADF-STEM image of monolayer BNNS, and (e) the corresponding intensity profile along the XX´ and YY´ directions. Impurity elements, i.e., C and O, in the nanosheet can be identified from their relative intensities. Adapted from Ref238.

The thickness of the BNNS is conventionally determined by analysing the folded edges of the sheets under HRTEM mode. The atomically thin sheets are generally folded along the edges during the transfer step or during the electron beam exposure under HRTEM analysis. The creases in the sheet can be also detected under similar HRTEM analysis (Figure 2.16a, b)237. The serendipity of getting a folded edge and the presence of creases however, often makes it challenging for accurate identification of thickness across all the regions of the nanosheet. Annular dark field (ADF) imaging in a scanning TEM (STEM) is another technique to identify the individual atoms and the layer thickness of BNNS in terms of ADF intensity, which varies with the atomic number (Z)239, as about Z1.64 (Figure 2.16c). The precise identification of thickness by ADF- STEM imaging, however, is a tedious process as it requires the presence of a 40

background (a hole, created by electron beam irradiation damage to BNNS) as a reference. A multislice simulation has been carried out to examine the changes in the SAED patterns, and the ADF-STEM and BF-STEM images for mono- to few-layer BNNS as the sample is tilted from 0 to 500 mrad.240 For the monolayer BNNS, the SAED pattern and ADF-STEM images were found to be nearly insensitive to tilt angle, while the images changed significantly with tilting for multilayer BNNS. This technique of identifying monolayer BNNS from its bulk, however, still remains to be confirmed by experimental analysis.

TEM is a powerful tool to probe individual atoms, defects, and grain boundaries of BNNS. The individual B and N atoms and the impurity atoms inside the BNNS have been successfully detected by ADF-STEM analysis (Figure 2.16d, e)238. The grain boundaries54 and vacancy defects58 in BNNS have been explored by HRTEM analysis. Moreover, the interlayer distance and bond length can be determined by the HRTEM analysis.

The bonding nature (sp2 or sp3 configurations) and the chemical constituents can be analysed by electron energy loss spectroscopy (EELS) in TEM. EELS is a surface- sensitive tool, which is more effective for identification of low atomic weight elements (from C to 3d transition metals)241, while EDS analysis in TEM is more sensitive for high atomic weight elements. EELS and EDS mapping are frequently done to analyse the spatial distribution of the elements. High resolution (HR)-EELS analysis has been also employed to identify the defects in the BNNS.242

2.4.7 Atomic force microscopy (AFM) AFM is a versatile characterization instrument for 2D materials. The thickness and roughness of the BNNS sample can be accurately measured, generally under tapping mode in AFM. For the mono- and few-layer cases, precise thickness determination under AFM can be sometimes challenging, however. It has been reported that often a thickness of 1 nm can correspond to monolayer BNNS (since the interlayer distance of h-BN is 0.333 nm), due to the possibility of trapped moisture/water/impurities 243,244 between the nanosheet and the SiO2/Si substrate. Therefore, other characterization techniques, such as optical, Raman, SEM, and TEM analysis of the 41

samples is often resorted (by counting the number of layers) to compare the AFM derived thickness.

The mechanical strength of few-layer BNNS has been determined using the AFM indentation technique (Figure 2.17).50 A circular hole of 1 μm in diameter was first created in a SiO2/Si surface by e-beam lithography and reactive ion etching techniques.

The nanosheet was then transferred to the SiO2/Si interface, and a diamond tip was used to indent the free-standing nanosheets to determine the fracture strength.

Figure 2.17: Determination of mechanical strength of BNNS by the AFM indentation technique. (a) SEM image of large BNNS on a SiO2/Si substrate, with a patterned array of circular holes in the substrate. Areas C and E are fully covered by BNNS, and area F is the region fractured by indentation. (b) Schematic illustration of mechanical strength determination of BNNS by nanoindentation. Adapted from Ref50.

The quality of BNNS has also been determined by conductive AFM analysis in contact mode.245 BNNS is generally considered as an insulator due to its wide band gap. The presence of impurities and defect in the nanosheet, however, will increase its conductivity, which helps to identify the quality of the sheet under conductive AFM. Frictional characteristics of mono- to multiple-layer 2D nanosheets were examined in friction force microscopy and contact mode AFM analysis.101 Friction increased with a decreasing number of layers, which was ascribed to the puckering characteristics of the reduced nanosheet thickness areas under AFM tip sliding.

2.4.8 X-ray photoelectron spectroscopy (XPS) XPS analysis is generally done directly on the growth substrates without any transfer steps. Identification of elements, and the quantity, ratio, and bonding nature of each element can be carried out on the XPS spectra. In BNNS samples, the presence of B and N has been identified at ~190.3eV and ~398.8 eV, respectively.50,114,135,142,150– 152,154,246–248 Two wide small intensity peaks also can appear at 9 eV and 25 eV higher energy positions ( plasmon and bulk plasmon loss peaks) than the B 1s and N 1s core level peaks, which is a fingerprint signature of the sp2 bonding nature between B and N atoms.34,249 For the amorphous and cubic BN samples, these peaks do not appear. Analysis of the elements from the XPS spectra (peak deconvolution, bonding nature, ratio of the elements, etc.) can be done in CasaXPS software. During the experiment, charge build-up can occur in the sample and can shift the peak positions of the elements. Calibration of the obtained spectrum can be done using the carbon peak as a reference and, therefore, eliminating the charge compensation related shift of energies over the whole spectrum.

2.4.9 UV-vis spectroscopy BNNS has a poor light absorption capability (< 1.5%) in the visible range and therefore appears transparent under optical microscopy.219,250 BNNS has an intense absorption edge in the ultraviolet-visible (UV-vis) range.68 Utilizing this property, the band-gap energy of few-layer BNNS has been determined by analyzing the absorption data from the visible to the UV-vis regime. The band gap has been determined according to Tauc’s formula251,

2 2 휔 휀 = (ℎ휔 − E푔) ……………………………………………………………. (2.4)

Or, through this reaction252:

1/2 훼 = C(E − E푔) /E …………………………………………………………... (2.5)

Here, ω = Angular frequency = 2π/λ α = Absorption coefficient; it can be obtained from 휀 = αL, where 휀 is absorption value and L is the film thickness. h = Planck’s constant 43

Eg = Band gap energy E = Incident photon energy

BNNS, grown on catalyst substrates by the CVD process, needs to be transferred to quartz plates for analyzing the UV-vis spectrum. Significant absorption of light does not occur until the wavelength drops to 350 nm from the visible range. A sharp absorption peak is observed at ~200 nm in the presence of BNNS.50 The h-BN bulk powder samples, however, show the peak in the 215-227 nm range.68

2.4.10 X-ray diffraction (XRD) Structural information about the material, including crystallinity, strain, and interlayer distance, can be obtained from XRD analysis. It has been a challenge to directly analyze few-layer BNNS, sitting generally on either the catalyzed substrate or 300 nm

SiO2/Si wafer (after transfer from the substrate) by XRD, however. Shi et al. obtained the (0002) reflection peak at 26.70° (2θ angle) for their 50 nm thick BNNS sample.152 XRD analysis is, therefore, generally considered for the h-BN powder samples. The 2θ peaks appear at 26.70°, 41.60°, and 55.00°, among which, the peak at 26.70° is the most intense and well-defined peak, confirming the presence of crystalline h-BN.253 The peak becomes broader from 20° to 35° for semicrystalline h-BN samples.124

2.4.11 Other techniques Scanning tunneling microscopy/spectroscopy (STM/STS) analysis gives atomic scale information where individual atoms can be detected by atom-to-atom raster scanning. Square octagon pairs (4ǀ8) and pentagon heptagon pairs (5ǀ7) were detected along the domain boundaries of BNNS by STM investigation.49 Moreover, in the in situ growth process163, defects generated during the growth of BNNS163 and the periodicity of BNNS163,175,254 growth on different substrates have been analyzed by STM. Auger electron spectroscopy (AES) can be employed to determine the boron and nitrogen concentration in BNNS as a substitute/addition to XPS analysis.171,179,247,255 In the study of in situ BN doping of graphene, Chang et al. reported that they did not find any boron signal in XPS for 2% and 6% BN doping in graphene layers due to the low atomic sensitivity to boron in XPS.179 In AES analysis, however, both boron and 44

nitrogen peaks were clearly apparent.179 High energy X-ray absorption spectroscopy (XAS) and X-ray emission spectroscopy (XES) obtained from synchrotron radiation can be also used to obtain a better understanding of the fine details of the boron and nitrogen bonding, the interaction with the substrate, and corresponding properties.166,179,256 Moreover, secondary ion mass spectroscopy (SIMS) have been used for depth profiling and elemental identification.34,165 The crystallinity of the nanosheets, in addition to SAED studies under TEM, can be also confirmed by low energy electron diffraction (LEED) analysis in a photoemission/low energy electron microscope (PEEM/LEEM).257166258

Chapter 3

Experimental Approaches

3.1 BNNS growth on copper substrate

BNNS growth was carried out through atmospheric pressure CVD (APCVD) process which was first developed by Li et al. for the growth of graphene.259 Figure 3.1 shows a schematic illustration of an APCVD system used for the BNNS growth on Cu substrates. A piece of Cu (0.25 mm thick, 99.98 % purity, Sigma-Aldrich) was pretreated with 30 % HNO3 solution (70 % concentration, Sigma-Aldrich) for about 5 minutes to remove the oxide layers from the surface, and then, blow dried in N2 steam (>99.99 % purity, BOC, Australia) before being transferred into a quartz tube (44 mm inner diameter, 50 mm outer diameter, 1200 mm length, MTI Corporation, China). Prior to heating, the whole system was evacuated and backfilled three times with a

10% H2/Ar (volume percentage, >99.999 % purity, BOC, Australia) gas mixture to remove air and moisture from the system. For the BNNS growth on melted Cu substrate, Tungsten (W) foil (0.127 mm, ≥ 99.9% purity, Sigma-Aldrich) was used as a support for the melted Cu, since it facilitates sufficient wetting of the melted Cu on it.

Ammonia borane (97 % purity, Sigma-Aldrich) was primarily used as the precursor for the synthesis of BNNS. Ammonia borane is decomposed at 110°C, which leads to the formation of borazine (B3N3H6), hydrogen, polyaminoborane [(BH2NH2)n], polyiminoborane [(BHNH)n], polborazylene [(B3N3H6)n], diborane (B2H6), and 129,132,133,260,261 ammonia (NH3). These polymeric compounds have been reported to degrade the quality of the grown BNNS.22,90 In this study, ammonia borane was decomposed outside the CVD furnace in a Schlenk reaction tube fitted with a glass filter (Figure 3.1) that prevents these polymeric compounds from entering the reaction zone. Therefore, the volatiles reaching the substrate are H2 and borazine, since trace amounts of B2H6 and NH3 will form solid amine complexes along the pipe before reaching the substrate.262 Borazine tends to polymerize into solid polyborazylene at 46

room temperature, so it requires low temperature for long-term storage. It reacts with moisture to form boric acid, ammonia, and hydrogen.90,260 It is also toxic. The thermal decomposition of ammonia borane in this experiment serves as the in-situ source of borazine. The decomposition products of ammonia borane were carried into the reaction zone by a flow of 100 ml/min 10% H2/Ar through the Schlenk reaction tube.

The whole synthesis process can be divided into four steps, i.e., ramping up the temperature, annealing the substrate, growing BNNS, and cooling. Ramping was done at 10 °C/min under a 10% H2/Ar gas mixture flow. An annealing treatment was then carried out at 1000 °C and 1100 °C for the growth on solid and melted Cu substrates, respectively, for 40 min under 500 ml/min 10% H2/Ar gas mixture flow to remove residual surface oxides and contamination. After the initial annealing period, the growth of BNNS was continued for one hour by flowing the gaseous thermal decomposition products of ammonia borane over the Cu substrate. At the end of the growth, the furnace was cooled down to room temperature by lifting the top cover of the furnace. The cooling rate was varied to avoid large crumples and cracks in the BNNS. The conditions of fast and slow cooling rates are shown in the time- temperature diagrams (Figure 3.2).

3.1.1 Growth of partially and fully covered BNNS Initially, 100 mg of ammonia borane was used to compare the BNNS morphology, thickness, and crystallinity on solid and melted Cu substrates. Growth processes of BNNS on those two substrates were then studied by varying the amounts of ammonia borane. 40 mg of ammonia borane and 30 minutes of growth were found to be sufficient for the complete coverage of the solid and melted Cu substrate under the applied conditions (ammonia borane was decomposed at 110 °C and 100 ml/min of

10% H2/Ar flow was used during the synthesis and cooling periods). Partial coverage by BNNS was tuned by varying the ammonia borane content, i.e., using 10 to 30 mg of ammonia borane.

Figure 3.1: Schematic illustration of an APCVD system. Volatile thermal decomposition products of ammonia borane were carried over the Cu substrates by a

10% H2/Ar gas mixture. For the melted Cu, tungsten was used as the support due to its better wettability compared to the quartz slide.

3.1.2 Preparation of bare annealed solid and melted Cu substrates Bare solid and melted Cu substrates were prepared using the same procedure for BNNS growth on solid and melted Cu substrates, except the ammonia borane decomposition step, to compare the bare and BNNS deposited substrates under similar conditions.

(a)

(b)

Figure 3.2: Time-temperature profiles for the CVD growth of BNNS. (a) Fast cooling process and (b) slow cooling process after the growth, while all the other parameters were kept fixed.

3.2 Transferring BNNS to different substrates

BNNS was transferred to 285 nm SiO2/Si wafers (p doped, Graphene Supermarket) using the conventional poly(methyl methacrylate) (PMMA, average Mw~996,000, Sigma-Aldrich) assisted transfer process.194 3 % PMMA in chlorobenzene (ACS reagent, ≥99.5 % purity, Sigma-Aldrich) solution was spin coated at 4000 rpm for 30 seconds on the BNNS grown on the substrates. The samples were then dried in air for few hours before being kept in a 0.1 M ammonium persulfate (ACS reagent, ≥98 %

purity, Sigma-Aldrich) etching solution floating overnight, with the PMMA layer facing up. As the deposition of BNNS can occur on both front and back sides of the foils, the backside BNNS is removed for clean transfer. The commonly adopted technique is to etch the backside Cu surface in 0.1M ammonium persulfate solution for 1 min before transferring the PMMA coated substrate to another etching solution. After etching, the BNNS/PMMA layer was liberated and floated on the solution. The BNNS/PMMA was washed with deionized water several times before it was fished out on the SiO2/Si wafers. The sample was then kept inside a desiccator overnight to dry and improve the adhesion of the BNNS to the substrate. After the drying process, the adhesion was further strengthened by heating at 180 °C for a few minutes. The PMMA layer was then removed by soaking in warm acetone (40 °C) for one hour, and followed by isopropanol (IPA, ACS reagent, ≥ 99.5 % purity, Sigma-Aldrich) washing to obtain the final transferred sample. After that, the samples were oven dried at 80 °C for a couple of hours to remove the residual IPA. An additional heat treatment at 500

°C for two hours in 10 % O2/Ar flow was found to lower the PMMA residue significantly.204

BNNS grown on solid Cu of Ni substrates was transferred to Au-coated TEM grids by a direct transfer process.196 Holey carbon, lacey carbon and quantifoil TEM grids (TED PELLA, Inc.) were used for the analysis under TEM. The TEM grid was kept on top of the BNNS coated solid Cu or Ni substrates and some ethanol was then dropped on top of the grid. After drying in an oven at 100°C for a few minutes, the sample with the TEM grid was kept floating on 0.1M ammonium persulfate Cu etching solution. After overnight etching, the TEM grid with the BNNS sample was scooped up and transferred to deionized water solution for cleaning the etchant residues. The scooped up sample from the water was then dried in a desiccator overnight, and subsequently, in the oven at 80°C for a few hours.

The solidified Cu surface from melted stage was also spin coated with PMMA. As the underlying region of solidified Cu was attached to tungsten foil and upper surface was covered by PMMA, the surrounding edge of the solidified Cu was scratched with a sharp needle for letting the etching solution to come in contact with the underlying Cu through the scratch. Complete dissolution of underlying Cu took about one day using

0.1 M ammonium persulfate solution. After that, the etching solution was replaced by deionized water. PMMA attached BNNS floated on top of deionized water and was ready to be transferred to different substrates. BNNS was transferred to 285 nm SiO2/Si wafer for AFM and Raman analysis and to the TEM grids for TEM analysis. PMMA was removed by hot acetone washing and further heating at 500 °C in 10 % O2/Ar flow

(samples transferred to SiO2/Si substrate), as has been described in the previous paragraph.

3.3 Characterization Optical microscope images were collected using a Leica DM6000 microscope.

The surface roughness of the as-received and treated Cu foils, together with the thickness of the BNNS, was acquired using an Asylum AFM MFP-3D in the tapping mode.

X-ray photoelectron spectra (XPS) were collected using a PHOIBOS 100 hemispherical analyzer. The X-ray excitation wavelength was provided by Al Kα radiation with photon energy of 1486.6 eV. The XPS data were analyzed using CasaXPS software. The Shirley background function was used for B 1s and N 1s peak background correction. A 70% Gaussian and 30% Lorentzian combined function was used for peak-fitting analysis.

Raman spectra were acquired directly on the Cu foil substrates using a Horiba Raman JY HR 800 spectrometer. A 632.81 nm laser with a He-Ne source was used for the characterization. Raman spectra of BNNS on silicon wafers were obtained via a Renishaw inVia Raman spectrometer using a 514.0 nm Ar+ laser. Raman spectra of the graphite-like structure on top of BNNS were collected using a Horiba Raman JY HR 800 spectrometer with an internal HeNe laser source (632.81 nm). Raman spectra on top of the Cu substrates were acquired by a 632.81 nm He-Ne laser source in Horiba Raman JY HR 800 spectrometer.

Field emission SEM images of BNNS on solid Cu substrate were obtained on a scanning electron microscope (JEOL JSM-7500FA). SEM images of the BNNS on top 51

of the Cu substrate were obtained at 1 kV working voltage, using a 4 mm working distance in a JEOL-JSM-7500 fitted with a field emission gun. SEM-based EDX analysis of the SiOx particles was performed at 10 kV operating voltage.

TEM/STEM examination was carried out using a probe-corrected JEOL ARM200F equipped with a cold field emission gun operating at an acceleration voltage of 80 kV. HAADF images were acquired with inner and outer collection angles of 50 to 180 mrad, respectively, ensuring Z-contrast, while BF images were acquired with a maximum collection angle of 17 mrad. All images were acquired with a 20 μs dwell time and convergence angle of 25 mrad, corresponding to a probe size of ~1 A and a current of 40 pA. Images were filtered by two different methods: Fast Fourier transform (FFT) filtering involved selecting all the periodic reflections in the FFT by masking and carrying out the inverse transform. The second method involved deconvoluting the probe effects in the images via the maximum entropy method. {1120} reflections were clearly observed in the FFTs, which represents a minimum resolution of 0.145 nm. EELS was performed using a GIF Quantum equipped in the same microscope. Spectrum images (SI) were acquired in STEM mode using a convergence and collection semi-angle of 25 and 28 mrads, respectively. All spectra in the SI were aligned to the leading edge of the boron K-edge. Optimized dark corrections were carried out on all SIs using the standard Gatan procedures. The B and N edges were extracted by fitting an exponential background window from 163 eV to 183 eV and 376 to 396 eV, respectively. B and N signal integration windows were positioned at 190 eV to 220 eV and 401 to 431 eV, respectively.

3.4 Electrochemical analysis

3.4.1 Electrochemical impedance spectroscopy (EIS) EIS analysis was performed using an automated CH Instruments electrochemical workstation model 660D with CHI integrated software version 11.17. Bare Cu and BNNS coated Cu surface areas of 0.785 cm2 (circular areas having a diameter of 1 cm) were exposed to 0.5 M NaCl solution for 30 min prior to measurements. A standard 3- electrode set-up was employed with Pt mesh serving as the counter electrode and Ag/AgCl (3.0 M NaCl) serving as the reference electrode. The frequency range was 52

100 kHz to 10 mHz with a ±10 mV perturbation around the open circuit potential (OCP).

3.4.2 Scanning electrochemical microscopy (SECM) SECM was performed using CH Instruments SECM model 920D utilizing CHI integrated software version 12.26. Pt disk ultra-micro-electrodes (UME) with 25 µm diameter, purchased from CH Instruments Inc., was used as working electrode. Pt mesh and Ag/AgCl (3.0 M NaCl) were used as a counter electrode (CE) and a reference electrode (RE), respectively. Figure 3.3 shows the experimental setup and an electrical model explaining the SECM data collection in AC mode. Alternating current SECM (AC-SECM) approach curves were obtained in 1 mM NaCl solution in order to maximize the sensitivity of the AC response. Approach curves were collected by moving the Pt UME from a distance of 75 µm above the Cu substrate down to < 1 µm from the substrate surface. Approach curves were generated at the range of frequency from 100 kHz to 1 kHz. The UME was held at OCP with an excitation AC signal amplitude of ±200 mV. A positive feedback is generally detected when the frequency (f) of the applied AC signal to the UME is smaller than the reciprocal of the time constant of the interfacial impedance (f < ½ πRC; here R is resistance and C the capacitance).263 Due to the insulating nature of BNNS, the time constant associated with the surface film increases, resulting in a positive feedback at low frequencies. The negative feedback at high-frequency domain indicates a higher concentration of ionic species and lower solution resistance in close proximity to the surface.264,265

Figure 3.3: SECM (a) experimental set-up and (b) electrical model explaining the data collected in the AC mode. RCu, RSL, RSol, and RT represent the resistance associated with Cu, surface layer, solution and SECM tip, respectively. C elements (Ccu, CSL, and

CT) represent the associated capacitance elements.

3.4.3 Linear Polarization Resistance (LPR) Linear polarization Resistance (LPR) experiments were conducted in a standard 3- electrode arrangement with Pt mesh serving as the counter electrode and Ag/AgCl (3.0 M NaCl) serving as the reference electrode. In linear polarization, a bias potential of ±20 mV range around OCP was applied and linear polarization resistance (LPR) was calculated from the linear behavior of E-I curve in close proximity of OCP. PDP measurements were made within the potential range of -0.25 V to -0.05 V (vs. Ag/AgCl) and Tafel plots were produced using the integrated CH instruments software.

3.5 Simulation

3.5.1 HAADF-STEM image simulations QSTEM software was used to simulate the HAADF-STEM images of various h-BN stacking orders along the [0001] zone axis.266 QSTEM is a suite of software for quantitative image simulation, including model building and TEM/STEM/CBED image simulation. In my thesis, the specific objective for using this software was to match the experimental ADF-STEM images with the simulated ones generated by the QSTEM software. QSTEM was written to do HAADF-/ADF-/ABF-STEM simulations. To use this software, the cif files of each of the possibly configured h-BN

had to be prepared first. The ADF-STEM image simulation was later carried out by keeping the conditions close to the experimental values.

The supercell structure consisted of 147, 98, and 49 atoms for 3, 2, and 1 layer of h- BN stacking simulation. The scanning window in the XY plane was 15 ×15 Å, and a pixel size of 50 × 50 was used. The probe resolution was 0.05 Å, and the probe array was 400 × 400 pixels. The microscope parameters used for the simulation are as follows: 80 kV voltage, 0 nm defocus, 0 nm astigmatism, 0.001 mm spherical aberration, 300 Kelvin temperature, 30 thermal diffuse scattering (TDS), 1.0 mm chromatic aberration, 0.4 eV dE, 25 mrad convergence angle, 0° beam tilt, 5 × 108 A/cm2 Sr brightness, and 20 μs dwell time. The detector inner and outer angle positions were set at 68 and 280 mrad, respectively. The offset in the X and Y directions was set at 0 mrad.

3.5.2 Density Functional Theory (DFT) calculation The calculations were performed using the DFT method with the gradient-corrected exchange-correlation functional of Wu and Cohen (WC)267 as implemented in the SIESTA package.268 The WC functional gives a realistic description of BN−transition- metal interaction and has been successfully used in previous works to describe various physical and chemical properties of BN/metal systems.174,269–271 The double-ζ plus polarization function (DZP) basis set was used to treat valence electrons of H, B, N, O, and Cu atoms. The core electrons are represented by the Troullier–Martins norm- conserving pseudopotentials. Periodic boundary conditions were used for all systems, including free molecules. The Cu fcc lattice was optimized using the Monkhorst-Pack 14x14x14 k-point mesh for Brillouin zone sampling. The calculated Cu lattice parameter is in excellent agreement with the experimental value.272 The optimized lattice of bulk Cu was used to construct a four-layer 8x8 slab for Cu(111), containing 256 Cu atoms. The calculated lattice parameter of the h-BN monolayer has a close value to the surface lattice constants of Cu(111), resulting in a good match of the h- BN and Cu(111) lattices. The periodically replicated slabs were separated by a vacuum region of 20 Å. In calculations, the bottom two layers of the Cu(111) slab was fixed, while all other atoms were fully relaxed. Only the Gamma-point was used for sampling the Brillouin zone of the slabs because of the large size of the supercell. An energy 55

cut-off of 200 Ry was chosen to guarantee convergence of the total energies and forces. A common energy shift of 10 meV was applied. Self-consistency of the density matrix was achieved with a tolerance of 10-4. For geometry optimization, the conjugate- gradient approach was used with a threshold of 0.02 eV Å-1. To obtain the most stable configurations of H2O, OH, H, and O on the pure and BN modified Cu(111) surfaces, a large number of starting geometries in different non-equivalent positions and orientations were generated. The starting structures were optimized without any geometry constraints. All adsorption energies are corrected for the basis set superposition error (BSSE). A similar approach and set of parameters have been used in a recent work.273

Chapter 4

Synthesis of large and few atomic layers of h-BN by CVD method

4.1 Introduction Atomic layered h-BN, i.e., nanosheet (BNNS), is also called ‘white graphene’, since it is white with a large band gap of 5.0 to 6.0 eV63,274–276, in contrast to graphene’s zero band gap that results in a black color. BNNS has shown some exceptional properties that lend themselves to applications different from those of graphene. For instance, BNNS is oxidation resistant up to 1000 °C in air92, while graphene deteriorates around 600 °C.277 BNNS is highly resistant to wet chemical attack and is difficult to functionalize93 while graphite readily turns into graphene oxide in a mixed sulfuric acid, sodium nitrate, and potassium permanganate solution.95 Using BNNS as a substrate for graphene has opened up a large potential for application in graphene- based devices.19 BNNS has been utilized as a dielectric layer for the fabrication of graphene-based field-effect transistors21,69,278 and explored as a tunneling barrier within graphene layers.70 For these applications, high-quality BNNS that is large in size, uniform in thickness, and free of defects, is highly critical. The synthesis of such high-quality BNNS, however, remains challenging.

Among the various synthesis techniques, the CVD method has been established as a popular route to grow large nanosheet within a single run. The first CVD route of monolayer BNNS was produced as early as 1995 using an ultra-high vacuum (UHV) CVD system.8 Recently, simple atmospheric pressure CVD (APCVD)50,136,152,181 and low-pressure CVD (LPCVD)90,135,143 systems have been developed to grow large BNNS. Polycrystalline Cu and Ni foils have been routinely used as substrates/catalysts in CVD systems because of their wide availability, low cost, and ease of etching to facilitate the transfer of BNNS onto different substrates. As far as conventional 57

polycrystalline foils are concerned, however, the CVD method is not free from limitations. Poor controllability over the BNNS thickness is one of the primary drawbacks.19,25 In the CVD method, nucleation of h-BN has been observed along the crystallographic defects of the polycrystalline catalyst surface, which consequently reduces the quality of the final product.90,181 Single crystal foil can be an option to remove the grain boundary induced nucleation sites8,130,154; however, the high cost and small size of single crystal foils limit their use. It has been demonstrated that improving the surface smoothness by employing techniques such as electro-polishing is conducive to the formation of large domains of BNNS.181 Melting can lead to an ultra- flat liquid surface that is free of grain boundaries and crystallographic defects, and has been found to facilitate the growth of large uniform graphene single crystal.279,280 In this chapter, melted-Cu facilitated APCVD growth of large and few-layer BNNS has been reported. For comparison, the growth was also carried out on solid Cu foil at 1000 °C under similar conditions. The BNNS samples were characterized using X-ray photoelectron spectroscopy (XPS), atomic force microscopy (AFM), Raman spectroscopy, scanning electron microscopy (SEM), and transmission electron microscopy (TEM).

4.2 Results BNNS was grown on solid and melted Cu substrates in a split type tube furnace (see details in the experimental section). For simplicity, BNNS grown on solid and melted Cu substrates are denoted hereafter as S-BNNS and M-BNNS, respectively. 10 %

H2/Ar gas mixture flow was maintained at 100 ml/min for all the experiments, during the growth period of BNNS. Large flow rates such as 200 ml/min during the growth resulted in fragmented BNNS on solid Cu substrate (Figure 4.1).

Figure 4.1: Discontinuous and inhomogeneous BNNS grown on solid Cu with a 200 ml/min 10% H2/Ar carrier gas flow. The corresponding line profile shows fragmented BNNS of about 1 nm in thickness.

To determine the crystallinity, size, and thickness of the nanosheets, both S-BNNS and

M-BNNS were transferred to 285 nm SiO2 coated Si wafers and holey carbon TEM grids (see details in the experimental section). Clear optical contrast was observed between S-BNNS and M-BNNS on the SiO2/Si wafers. The clear blue region in Figure 4.2a and the nearly transparent area indicated by the dotted line in Figure 4.2b represent thick and thin BNNS grown on the solid and melted Cu substrates, 219 respectively. The thickness of BNNS on SiO2/Si wafer was measured by AFM (Figure 4.2c and d). A thickness of 4.2 nm was observed for the S-BNNS, indicating ~ 12-layer stacking. In contrast, M-BNNS is only 1.0 nm thick, corresponding to 1-3 atomic layers. It should be noted that underlying trapped water or contamination between the BNNS and the SiO2 substrate can often increase the thickness under AFM, and therefore, 1 nm corresponds to the height of a mono-layer of h-BN.45,219 The AFM thickness analysis results were further corroborated by high resolution (HR)-TEM studies, which will be discussed later.

Figure 4.2: (a), (b) Optical images; (c), (d) AFM height images with corresponding thickness profiles on SiO2/Si wafer for the S-BNNS and M-BNNS samples, respectively.

XPS was employed to assess the stoichiometry and nature of the bonding between B and N atoms (Figure 4.3). For both samples, the B 1s, and N 1s peaks are situated at 189.8-190.9 eV and 397.1-398.7 eV, respectively, which match the published results.50,135,143,181 The B/N atomic ratio was estimated based upon the integrated peak areas and the peak sensitivity factors (Figure 4.3b, c). The B/N atomic ratio is found to be within the 1:1.04−1:1.12 range, close to the 1:1 stoichiometric ratio for sp2 bonded h-BN. The calibration of the peak positions for all the elements were done using C 1s peak at 284.8 eV as a reference (Figure 4.3d).

Figure 4.3: (a) Representative XPS survey spectrum of BNNS on melted Cu substrate. From the integrated peak area analysis of B 1s (b) and N 1s (c) peaks, a nearly 1:1 B/N atomic ratio is obtained, indicating sp2 h-BN formation. The Shirley background function was used for B 1s and N 1s peak background correction, and a 70% Gaussian, 30% Lorentzian combined function was used for peak-fitting analysis. XPS of C 1s, O 1s and Cu 2p peaks are shown in (d), (e) and (f). The small C 1s (c) and O 1s (d) peaks might have come from surface contamination by air in between the synthesis and the XPS measurements. The C 1s and O 1s peaks were also observed in a bare annealed Cu sample.

Raman spectra of BNNS were obtained on a Renishaw InVia Raman spectrophotometer using a 514.5 nm Ar+ laser source. Figure 4.4 shows the typical B- 219,281 N stretching vibration mode (E2g) of h-BN on a SiO2/Si substrate. The lower intensity of the E2g peak of M-BNNS reflects the fact of fewer atomic layers in comparison to S-BNNS.50,136,219,281 A peak shift of 3 cm-1 was also observed for M-

BNNS. In general, shifting of E2g peak can occur under different strain conditions 61

within the layers: a compressive strain results in a blue shift and a tensile strain in a red shift.50,135,219 For monolayer BNNS, several reasons have been proposed for the large blue shift, including hardening of the E2g phonon due to the shorter B-N bonds in the monolayer219, intrinsic wrinkles due to the substrate surface50, doping effects20,219, and lack of interaction between neighbouring layers.219 The blue shift in the BNNS case is likely to be associated with compressive strain and wrinkle formation. Thermal expansion mismatch between the BNNS and the Cu occurs during cooling, where Cu contracts and BNNS expands, giving rise to a compressive strain 282 on the sheets. For the S-BNNS, the shift in E2g is negligible. It is reasonable to assume that less compressive strain exists in S-BNNS since the thermal contraction of the solid Cu substrate is much less than that of the melted Cu. In addition, thick S- BNNS is less susceptible to the formation of wrinkles.

Figure 4.4: Comparative Raman spectra of BNNS grown on solid and melted Cu substrates and then transferred to SiO2/Si substrate.

Figure 4.5a shows a high angle annular dark field scanning TEM (HAADF-STEM) image of mono- and bi-layer regions of BNNS obtained at 80 kV. The fast Fourier transform (FFT) obtained from the whole image shows one group of distinct hexagonal symmetric spots, which indicates a highly periodic h-BN structure with no differences in rotational orientation between the layers. The darker and brighter regions in Figure 4.5a have been identified as mono- and bi-layer BNNS, respectively.92 The FFT

filtered B mono-vacancy image (inset enclosed by the red square in Figure 4.5a) clearly indicates a missing B atom in the hexagonal network.

A raw HAADF-STEM image of a monolayer h-BN region is shown in Figure 4.5b, where the relative darker and brighter regions in the hexagonal network can be assigned to boron and nitrogen atoms respectively. This is due to slightly higher atomic number of N than B atoms that causes Z-contrast intensity variation in the HAADF images238. An FFT filtered HAADF-STEM image of a monolayer h-BN is shown in Figure 4.5c. The line profile along the trace in Figure 4.5c confirms that the sp2 bonded B and N form a hexagonal honeycomb structure in the monolayer h-BN, while the presence of other elements would have been identified from the relative intensity difference.238

Figure 4.5: (a) HAADF-STEM image of a bi-layer BNNS. The dark triangular region is monolayer h-BN, which was exposed after electron beam radiation removed the top layer. The FFT pattern in the top inset shows a distinct six-fold hexagonal pattern, characteristic of h-BN. The inset in the red square shows a FFT filtered B monovacancy zone. (b) Raw HAADF-STEM image from a monolayer BNNS region. The heavier N atoms are slightly brighter than the B atoms. B and N atoms have been tagged with red and blue circles, respectively. (c) FFT filtered HAADF-STEM image of a defect-free h-BN mono-layer region and Line profile along the trace.

The monolayer triangular region in Figure 4.5a was created by removing the B and N atoms from the top layer with an electron beam. The triangular geometry is probably associated with the more energetically favourable nitrogen edge termination in h- BN.58,177 It is to be noted that the operation voltage (80 kV), used in this study is below 63

the critical knock-on energy of nitrogen of 84 kV. Nevertheless, both B and N have been found to be knocked out of the second layer. It can be deduced that B monovacancy defects were formed initially due to its lower knock-on energy of 74 kV, and consequently, both B and N atoms can be removed below their critical knock-on energy due to the presence of these defects.283 B monovacancy creation by electron beam has been also reported at higher operating voltage (120 kV).58

Figure 4.6: Electron knock-on damage and layer opening in bi-layer BNNS, observed in ADF-STEM imaging mode at 80 kV: (a-d) Both B and N atoms were gradually knocked out of the top layer under electron beam irradiation, forming a triangle. HAADF-STEM image (d) can be an important tool to identify elements and the atomic layer thickness, by taking advantage of the Z-contrast.92 e) Direct layer identification from the intensity line profile analysis across the HAADF-STEM image in (d) clearly shows the vacancies, and the monolayer, and bi-layer BNNS regions. (d) was smoothed to lower the background noise in the line profile image in (e).

EELS was conducted to investigate the elemental composition and nature of the bonding in the BNNS. Figure 4.7a is the bright field (BF)-TEM image of a large BNNS region from where the EELS was taken (indicated in circle). The EELS spectrum (Figure 4.7b) confirms the presence of characteristic B K-edge and N K-edge peaks of h-BN, starting at 195 eV and 406 eV, respectively.31,50 The atomic ratio of B/N derived from the EELS spectrum is 1:1.16, close to the number based on the XPS results. It should be noted that this ratio can have an error of up to 12.2% due to uncertainty in the baseline determination31. The presence of 1s→π* and 1s→σ* anti-bonding orbitals next to B K- and N K-shell ionization edges confirms the formation of sp2 bonded h- BN.21,31,50,281,284 The small C K-edge near 289 eV is from the residual poly(methyl methacrylate) (PMMA) on BNNS.21,50

An aberration-corrected STEM study was conducted to confirm the formation of large single crystalline BNNS grown on the melted Cu. Figure 4.8a shows a low magnification HAADF-STEM image of the M-BNNS. Selected area electron diffraction (SAED) patterns were acquired from the specified regions circled in red in Figure 4.8a. One set of discrete six-fold symmetry patterns was obtained, with less than 1° rotational orientation difference between the patterns, suggesting ~ 4 μm2 65

single crystalline BNNS in that region. The dark areas in the image are voids, where the nanosheet was broken during the transfer process. A scrolled up nanosheet edge is identified in (a) by the white arrow. Some of the sheets were folded, which facilitates the quantification of the BNNS layers in the BF-STEM imaging mode. A statistical distribution of the BNNS layers from the analysis of 40 STEM images from different regions is shown in Figure 4.8c. Around 90 % of the BNNS are found to be mono- and bi-layer on melted Cu substrate. Figure 4.8c and d show the edges of mono- and bi- layer BNNS, respectively.

Figure 4.8: HAADF-STEM characterization of M-BNNS. (a) Low magnification HAADF-STEM image of M-BNNS. SAED patterns (bottom) were obtained from the four circled areas. The distinctive set of hexagonal spots in the SAED patterns indicates the single crystalline nature of the nanosheet within that region. The difference in rotational orientation in the four SAED patterns is less than 1°. (b) The statistical distribution of BNNS atomic layers based upon 40 BF-STEM images. BF- STEM images of (c) mono- and (d) bi-layer BNNS.

The homogeneity of the BNNS grown on melted Cu substrate can also be checked from the relative contrast in different areas of the nanosheet in a HAADF-STEM image

(Figure 4.9). The similar ADF-STEM intensity in the 400 nm × 300 nm region confirms the large scale homogeneity of the BNNS grown on melted Cu substrate.

Figure 4.9: (a) HAADF-STEM image of BNNS grown on melted Cu and (b) intensity profile along the line in (a). The similar intensity in the three zones indicates homogeneous BNNS in that region. The relatively bright regions are the PMMA residues.

The crystallinity of the BNNS grown on solid Cu substrate was also checked by SAED analysis from different regions, similar to what has been described above. Figure 4.10a shows a low magnification BF-TEM image of a large area BNNS on holey carbon TEM grid. For S-BNNS, the SAED patterns (Figure 4.10b) obtained from four adjacent areas marked in the BF-TEM image (Figure 4.10a) does not indicate any fixed angular orientation between the layers. Moreover, each SAED pattern shows several sets of hexagonal spots, which is indicative of polycrystalline BNNS in that region. 1- 10 layers of BNNS were observed within few micron-sized BNNS, which indicates largely inhomogeneous BNNS growth on the solid Cu substrate (Figure 4.10). Figure 4.10 (b-d) shows the BF-STEM images along the edges, indicating 3-, 7- and 10- layered BNNS having 0.3344 nm interlayer spacing.

Figure 4.10: (a) Low magnification TEM image of BNNS grown on solid Cu. SAED patterns obtained from the four areas marked in (a) show the polycrystalline nature of the BNNS. (b-d) BF-STEM images of S-BNNS. (b) 3-layered, (c) 7-layered, and (d) 10-layered BNNS.

4.3 Discussion There are several factors proposed as responsible for the observed differences in BNNS morphology and crystallinity when the nanosheets are grown on the solid and melted Cu. One key factor in growing large single crystalline BNNS is reducing the number of nucleation sites at the beginning of growth181, which has also been found to be critical for the CVD synthesis of large crystalline graphene.285,286 Due to the high interfacial energy along the defects (grain boundaries, rough edges, defect lines, impurities) on the solid Cu surface, the activation energy of h-BN nucleation is lower at those sites than within the smooth regions. This phenomenon can be clearly observed when the growth is carried out using a limited amount of ammonia borane.

Triangular h-BN domains are seen to grow preferentially along the manufacturing- induced parallel defect lines and grain boundaries on the solid Cu surface (Figure 4.11a, b). At the nucleation center of each triangular domain, minute triangular h-BN adlayers are clearly noticeable (inset in Figure 4.11a) (Numbers of layers are identifiable based upon the contrast in the SEM images.255) h-BN adlayers are also observed along the merging areas (inset in Figure 4.11c). Adlayers have been observed during the CVD synthesis of graphene.287 It has been proposed that carbon atoms diffuse between the Cu and the existing monolayer, reach the center, and then form the secondary adlayers. This is unlikely the case for BNNS, however, since the diffusion of the h-BN monomers would encounter much resistance due to their size. The formation of adlayers is associated with the defects, both at the center and at grain boundaries, of existing h-BN monolayer. The nucleation sites associated with substrate defects would cause strain in the existing monolayer, particularly prominent at the center, which favors further nucleation on top of the existing layer, thus leading to multiple BNNS on a solid Cu surface. When more ammonia borane (10-20 mg) is used, these triangular domains grow in the lateral direction and join up with the neighboring ones (Figure 4.11c). Along the joining lines, secondary h-BN nucleation is observed. Further increasing the amount of ammonia borane (i.e., 100 mg) results in complete coverage of the Cu surface, but with various small triangular h-BN islands on the top (Figure 4.11d). The coalescence of these domains occurs by the edge attachment of new diffused h-BN monomers. The diffusion rate of these monomers is limited on the solid Cu surface at 1000 °C in comparison to 1100 °C for melted Cu. The movement of these monomers is also inhibited by the defect lines on the surface. As a result, the crystal size of the BNNS on solid Cu is found to be much smaller than that on the melted surface (Figure 4.8 and Figure 4.10). Due to the effective catalyst contact, the h-BN growth rate of the first layer exceeds that of the adlayers. The island- shaped domains grow in the lateral direction in the presence of sufficient borazine and ultimately join up with the neighboring domains. This leads to a large BNNS with a large number of adlayers attached to the sheet. Accordingly, S-BNNS growth follows the Volmer-Weber island model.192

Figure 4.11: SEM images S-BNNS and M-BNNS at different growth stage, with the insets showing corresponding images at higher magnification. For solid Cu, (a), (b) h- BN triangular domains with adlayers at the nucleation centers start to grow preferentially along the defect lines and grain boundaries. (c), (d) more ammonia borane results in new h-BN domains while the initial triangular domains grow in the lateral direction and coalesce with their neighbors to form a continuous sheet. For melted Cu, (e), (f) initial h-BN domains grow randomly on the surface due to the absence of defect lines and grain boundaries. Increasing the amount of ammonia borane leads to a complete coverage. For the whole process, no adlayer growth is observed. The scale bars are 5 μm in the images and 1 μm in the insets. 70

Melting the Cu eliminates grain boundaries and other high energy defects, which then effectively reduces the number of nucleation sites. The h-BN nucleation starts with the supersaturation of the borazine molecules288 over random locations on the melted surface. The initial h-BN domains are monolayer in nature, with no observable adlayers on top (Figure 4.11e). After the initial h-BN nucleation, the smooth defect- free melted Cu surface provides a lower kinetic energy barrier and longer diffusion distance for the h-BN monomers to travel, adjust, and attach themselves to the existing nuclei in an oriented way. Such a growth process can, therefore, lead to the formation of large single crystalline BNNS. Increasing the amount of ammonia borane (100 mg) causes the single layers to join together, and finally, a large continuous and wrinkled nanosheet is obtained (Figure 4.11f). The presence of up to 3 layers (< 10 %) of BNNS on the melted Cu suggests that the growth is not surface-limited. The growth pattern can be described using the Stranski-Krastanov model.192

The Cu foil surfaces of as-received, HNO3 acid etched, 1000 °C annealed, and 1100 °C treated (melted) samples were analyzed. Defect lines developed during the manufacturing process and after HNO3 acid treatment are clearly visible from the optical images (Figure 4.12a, b). After annealing at 1000 °C, grain boundaries are more noticeable, while the intrinsic defect lines are less identifiable (Figure 4.12c). A very smooth surface with almost no defects is observed for the melted Cu (Figure 4.12d). Root-mean-square (RMS) roughness values of the as-received Cu range from 112 to 2 258 nm in five 10 × 10 μm sized areas. After the HNO3 acid etching treatment, the roughness value decreases slightly to 90-194 nm. Annealing the HNO3 acid treated Cu foil at 1000 °C under H2/Ar flow effectively lowered the roughness value to 7.023- 10.21 nm, with a further improvement to 4.681-4.761 nm for the melted Cu.

-1 -1 289–291 Surface Cu oxides (Raman peaks at 526 cm and 617 cm assigned to Cu2O ) in the as-received Cu can be largely eliminated by the HNO3 acid etching treatment (Figure 4.13). Raman analysis of the surfaces of the annealed solid and melted Cu picked up negligible amounts of Cu oxides (Figure 4.13).

Figure 4.12: Optical microscope images and Raman spectra of different Cu substrates.

(a) as-received, (b) 10 % HNO3 acid etched, (c) 1000 °C annealed, and (d) 1100 °C melted Cu substrate. Scale bar is 50 μm. The inset in each of the images shows the three-dimensional (3D) AFM height profile across a 10 x 10 μm2 area. The color bar is within the 0 to 1 μm range in all the insets.

Figure 4.13: Raman spectra of Cu foil surfaces. 72

An interesting phenomenon was observed when the BNNS growth was carried out on solid Cu at 1050 °C. At this temperature, initial h-BN domain formation is not found to occur preferentially on the original Cu grain boundaries (Figure 4.14). Furthermore, no adlayers on the initial domains (as at 1000 °C) were observed under SEM. With more ammonia borane (50 mg), multilayer nanosheet with small h-BN adlayers on the top are observed. It is understood that the Cu surface undergoes a significant recrystallization process as the temperature approaches the melting point (1083 °C), resulting in a smoother surface and larger grains. At this temperature, the diffusion kinetic energy of h-BN monomers is expected to be higher than that at 1000 °C, which is conducive to the formation of large h-BN domains. The insensitivity to high energy grain boundaries of the initial h-BN nucleation is not fully understood, however. Detailed theoretical and experimental investigations are required for this, which is beyond the scope of this study.

Figure 4.14: SEM images of BNNS grown at 1050 °C. (a) No preferential nucleation sites are observed for the initial h-BN domains. (b) Large and mixed multilayer BNNS with small adlayers on the top are found when more ammonia borane is used. Scale bar is 5 μm.

4.4 Conclusion Micron-sized single crystalline and mostly mono- and bi-layer BNNS can be grown on melted Cu substrate, in contrast to largely inhomogeneous, 1–10 layer thick, nanocrystalline BNNS on the solid Cu substrate. Characterization of the BNNS grown on solid and melted Cu substrates in XPS and EELS confirmed close to stoichiometric 1:1 B/N ratio and sp2 bonded h-BN. Possible BNNS growth processes on both solid and melted Cu surfaces have been discussed. The smooth and nearly defect-free surface of the melted Cu is believed to play the key role in the growth of high quality BNNS. 73

Chapter 5

Carbon- and crack-free growth of BNNS, its uncommon stacking orders, and EELS assisted thickness analysis

5.1 Introduction The highly electrically insulating nature, structural similarity to graphene, and atomic flatness (sp2 bonding) make BNNS a superior substrate for graphene-based electronics.19 Moreover, the high-temperature stability in air91, chemical inertness93, transparency in its 2D form219, and deep ultraviolet (UV) luminescence63 make BNNS a promising candidate for anti-corrosion, and nano- and optoelectronics applications. These applications require large BNNS with controlled and uniform thickness, and enormous efforts have been devoted to achieving these characteristics. CVD is the best method so far with the potential to produce BNNS with these properties, especially a large lateral size. Various CVD growth parameters require optimization since they can have dramatic impacts on the quality of the nanosheet. These parameters include, but are not limited to: choice of substrate and its treatment; choice of precursor and its concentration; growth temperature, pressure, and duration; heating and/or cooling rate; type of carrier gas and its flow rate; and the tube material. For example, high- temperature growth159 and substrate treatments (electropolishing181, melting157, extended high-temperature annealing23) can improve the crystal size of BNNS. Ultra- high vacuum (UHV) systems enable the formation of monolayer BNNS on Ni (111) substrates.7,8 Also, the precursor decomposition temperature90,292 and growth time90 can be modulated to grow mono- to few-layer BNNS. In view of the importance of the CVD method, in the first section of this chapter, the effects of the treatments of precursor and Cu substrate, and the influence of the quartz tube on the quality of BNNS are discussed.

The most stable structure of h-BN adopts an AA´ stacking, where B atoms sit on top of N atoms and vice versa in the adjacent layers. According to theoretical studies, AA´ and AB (Bernal) stacking have a minimal energy difference.47,48 Both stacking orders have been detected using DF-TEM46 in CVD grown BNNS. AB stacking has also been reported in chemically exfoliated BNNS, and this is attributed to the presence of impurities in between the layers, which modify the AA´ stacking towards AB stacking.293 In high-temperature CVD growth, the electrostatic forces that dictate the optimal stacking mode in h-BN294 are disturbed by the high kinetic energies of the atoms, and the film may adopt various stacking orders.48 Taking advantage of the direct elemental identification achievable through annular ADF-STEM Z-contrast imaging295, the CVD grown BNNS on solid Cu surfaces and their stacking orders have been analyzed. In the second section of this chapter, the observations of AB, ABA, AC´, and AC´B stacked BNNS using aberration-corrected STEM are discussed.

Raman analysis, SEM, AFM, and HR-TEM have been used to identify the thickness of BNNS. In the Raman spectrum, however, differences in the intensity and position of peaks caused by thickness are minute and therefore do not permit reliable thickness determination.219 Contrast differences observed in SEM imaging are indicative of the differences in thickness, but other techniques have to be combined with SEM to quantitatively determine the exact layer numbers.181,296 For AFM analysis, the presence of water or impurities underneath the BNNS often distorts the height profile of h-BN.45,219 In the third section of this chapter, a method to identify monolayer and few-layered BNNS based upon EELS and on EELS near-edge fine structure (ELNES) of the N edge is discussed.

5.2 Results

5.2.1 Impact of precursor purity BNNS was respectively grown on solid and melted Cu substrates in a quartz tube under atmospheric pressure CVD (AP-CVD) process, using ammonia borane as the precursor157 (See the experimental section). Ammonia borane has been the most common precursor to the CVD growth of BNNS. In the literature, ammonia borane is used in the as-received condition and there has been no report on the impact of

impurities on the growth. Commercial ammonia borane from Sigma-Aldrich is 97 % pure. Although the exact manufacturing method is unclear, a wet chemistry route is likely297–299 and trace amounts of hydrocarbon solvents would remain trapped inside the solid ammonia borane. The amount of impurities is believed to be very low since they were not detected using 11B nuclear magnetic resonance (NMR) and 1H NMR spectroscopy (not shown).

In this work, using the as-received ammonia borane (100 mg) resulted in micron-sized dark spots on top of the BNNS, grown on both melted and solid Cu substrates, as shown in the SEM images (Figure 5.1a, b). Note that this has been observed in several batches of ammonia borane from Sigma-Aldrich. In a control experiment without ammonia borane precursor, no such dark spots on Cu substrates are observed, confirming that they arise from the precursor. Raman analysis of these regions (marked in circle in the optical image of Figure 5.1c) revealed the presence of graphite (Figure 5.1d). The first images of Figure 5.1e and f, are the high-resolution secondary electron images of a representative dark spot regions that were observed in the SEM images in Figure 5.1a and b respectively. The corresponding EDX mapping analysis (in TEM at 80 kV operating voltage) from these spots clearly reveals the prominent presence of carbon (Figure 5.1e, f). Presences of many SiOx nanoparticles on and around these graphitic carbon spots are also detected, which probably originated from the quartz tube that was used during the CVD growth of BNNS. Figure 5.1g is the BF-TEM image of a dark spot region. HR-TEM imaging reveals an entangled graphite-like structure (Figure 5.1h). To remove the residual organic impurities, the ammonia borane was washed with pentane four times. The pentane-treated ammonia borane was then heated at 40 °C overnight under dynamic evacuation. Since pentane can dissolve most organic solvents, it is an effective cleaning agent, and since it has a very low boiling point (36 °C), any pentane residue can be readily removed under vacuum at temperatures below the ammonia borane decomposition temperature (55-130 °C).50,90,300 This treatment effectively removed the dark spots (Figure 5.1i). Keeping the as-received ammonia borane under vacuum for around one week at room temperature can also yield BNNS without dark spots.

Figure 5.1: Impact of impurities in the precursor on the growth. SEM images (a) and (b) show that as-received ammonia borane leads to the formation of graphite-like structures (appearing as dark spots) on top of BNNS. (c) Optical microscopy image of the BNNS grown on solid Cu substrate using as-received ammonia borane. (d) Raman -1 -1 spectrum from the circled spots in (c) shows the ID (1334.04 cm ), IG (1585.11 cm ), -1 and I2D (2657.45 cm ) peaks of graphite. (e) and (f) High-resolution secondary electron image with corresponding EDX mapping (scanning TEM mode, 80 kV operating voltage) of C, O and Si. The presence of carbon in the spotted regions is clearly noticeable. (g) BF-TEM image of a dark spot region. (h) HR-TEM image of the dark spot reveals a layered graphitic structure. (i) SEM image of BNNS on top of

solid Cu substrates without any noticeable dark spots when pentane-treated ammonia borane was used. Inset is the SEM image of BNNS on melted Cu substrate. The absence of dark regions (compare with (a)) indicates that treating ammonia borane with pentane followed by evacuation can effectively remove the residual organic contaminants.

5.2.2 Impact of carrier gas and tube material To understand the carrier gas effects on the BNNS growth, the BNNS growth on melted Cu substrates was carried out under ultra-high pure (UHP, >99.999 % purity)

H2, Ar, and 10% H2/Ar mixture, respectively. H2 is believed to remove the oxide layers from the Cu substrate at high temperatures and, therefore, help to control the nucleation. Although, there have been reports suggesting H2 can etch the 162,221,292,301 BNNS ; using UHP H2 as a carrier gas, BNNS on melted Cu substrate can be readily obtained (Figure 5.2a), where islands of BNNS are highly visible (inset of

Figure 5.2a). However, SiOx nanoparticles and crack lines are still detected inside the nanosheet (denoted by green arrows in Figure 5.2a). Several reports are available where BNNS was synthesized under pure hydrogen gas atmosphere149,180,302, thus corroborating the experimental results. Under UHP Ar atmosphere, large micron-sized particles on top of the Cu substrate are found (Figure 5.2b) and only traces of BNNS are observed after the growth (Inset of Figure 5.2b). EDX mapping analysis (Figure 5.2d, e) along those particles (dark under SEM contrast, Figure 5.2c) indicates those as SiOx. The growth under pure Ar gas atmosphere cannot remove the residual oxide layers from the Cu substrate. Also, the presence of any oxygen impurities inside the quartz tube (even after evacuation/argon gas backfilling three times at the start of the experiment), can readily etch the BNNS at 1100 °C, likely forming boron oxides and ammonia gas. SiOx may initially start as tiny particles over the Cu substrate, and during the nanosheet etching at high temperatures under Ar environment, they start to agglomerate together forming such large particles. Thus, to minimize extensive etching of the BNNS, exposure of SiOx nanoparticles towards the nanosheet should be kept at a minimum, if not can be avoided completely. A mixture of hydrogen and argon gas looks essential to avoid three-dimensional BNNS growth and to suppress the formation of large SiOx particles.

Figure 5.2: Effect of carrier gas on the BNNS growth. (a) Using UHP Hydrogen gas (99.999 %) as a carrier gas can produce BNNS on melted Cu substrate. Inset is the high magnification image of the BNNS islands, showing a 3D growth of BNNS in those regions. (b) Traces of BNNS are observed if only UHP Argon gas (99.999 %) was used as a carrier gas. Large particles (black under SEM contrast) are also found on top of Cu substrate. Inset is the SEM image of a portion of BNNS on Cu substrate. (c) SEM image of three particles which have joined each other. (d) and (e) EDX mappings show that silicon and oxygen clearly overlap with the region of the particles, indicating the particles are SiOx.

5.2.3 Crack free growth of BNNS As has been mentioned in the previous chapter, BNNS on solid Cu tends to be multilayered, with a grain size of a few hundred nanometers while on melted Cu, it is mostly 1-2 layered with a much larger grain size of several microns. Since melted Cu is beneficial towards the control of thickness and grain size, it is further optimized to control the growth of BNNS. The flatness of the substrate and cooling rate are found to significantly affect the quality of the final product. W foil is used as a support since melted Cu has a better wettability on it. When Cu and W foils are used without flattening, an ellipse-shaped Cu surface on the W foil is formed after solidification

(Figure 5.3a), and this is likely due to the poor contact between Cu and W. Flattening both the W and the Cu foils by mechanical press prior to growth creates good contact between them and helps to create uniform wetting of the Cu on the W foil (Figure 5.3b).

Figure 5.3: Digital camera images of Cu foils on W after 1100 °C melting and then solidification. (a) With no flattening of the Cu and W foils, an elliptical shaped Cu surface is formed on the W, while (b) flattening the foils can produce a uniformly wetted Cu surface on the W foil.

This ellipse shape causes a large temperature gradient from the edge towards the center when the melted Cu is rapidly solidified (Figure 5.4a). Due to the latent heat of fusion, the central region stays molten after the thin edges have solidified. Therefore, BNNS on the edges of the substrate contracts, while BNNS at the center floats on the molten Cu. This generates a tensile stress along the interface between the two regions, which causes large cracks hundreds of microns wide (Figure 5.4b) and extensive crumpling of the sheet around the center (Figure 5.4b). Along the edges (Figure 5.4c), micron- wide cracks (identified by yellow arrows) are observed among the nanoscale cracks that spread out in the nanosheet (identified by green arrows).

Flattening both the W and the Cu foils prior to growth creates good contact between them and helps create uniform wetting of the Cu on the W foil. The thermal gradient in the Cu substrate during cooling is therefore largely minimized. With slow cooling near the melting point, BNNS that is free of large cracks and wrinkles, thus, can be made reproducibly (Figure 5.4d). Nanoscale cracks with nanoparticles inside, however, are still observed (Figure 5.4e). These nanoparticles are found to be silicon

oxides (SiOx) (EDX analysis, inset in Figure 5.4f), which are likely to have originated from the quartz tube used for CVD. These nanoparticles are believed to etch h-BN and consequently generate minor cracks along the etching points (discussed further in a later section).

Figure 5.4: Schematic illustration of growth of BNNS on melted Cu substrate: (a) When the Cu and W foils are not flattened prior to growth, large cracks are visible. (b) SEM image of large cracks, hundreds of microns wide, around the center of an elliptical shaped Cu substrate, which is caused by the different shrinkage rates of the Cu on solidification due to the large thermal gradient. Inset is a high magnification SEM image from the highlighted region showing large crumples. (c) SEM image of relatively small cracks (identified by yellow arrows) near the edges of the Cu substrate. (d) When the Cu and W foils are flattened and the cooling rate is slow through the melting point, only small cracks are observed. (e) On the flattened foils, crumples and large cracks are largely minimized. (f) Enlargement of the image in (e), with the EDX spectrum (inset) indicating the presence of SiOx which is highlighted by red circles in (c), (e), and (f). The presence of nanoscale cracks (indicated by green arrows in (c),

(e), and (f)) is possibly initiated by SiOx nanoparticle etching.

Lowering the cooling rate through the Cu melting point (parameters are in the experimental section, Figure 3.2) in an ellipse-shaped Cu surface, can suppress the 81

formation of large micron-sized cracks, but does not eliminate them completely (Figure 5.5).

Figure 5.5: When the Cu and W foils were not flattened prior to the growth, cracks generated at the centre of an elliptical shaped Cu surface cannot be avoided, even with a slow cooling through the Cu melting point.

Quartz is the most commonly used tube material for the CVD growth of BNNS.50,136,143,281 Carrying out the growth inside a quartz tube, however, often generates SiOx nanoparticles, which etch the nanosheet and ultimately create small cracks in the nanosheet (Figure 5.4c, e, f). Replacing the quartz tube with an alumina tube, or effectively insulating the BNNS from SiOx nanoparticles is expected to yield better quality BNNS. Several experiments were attempted still using quartz tubes, but with the Cu/W substrate placed in an alumina boat. A small alumina boat is then placed upside down in it but with a slim aperture allowing borazine gas to access the substrate

(Figure 5.6a). SiOx nanoparticles have been largely eliminated, and nearly crack free BNNS is obtained (Figure 5.6b).

Figure 5.6: (a) Schematic illustration of the BNNS grown in a confined space. Cu/W foils are kept inside two alumina boats to minimize direct exposure to SiOx nanoparticles. A slim opening was maintained to allow the precursor gas (mainly borazine gas from the decomposed ammonia borane) to make contact with the Cu surface. (b) SEM image of large and nearly crack free BNNS. Very few SiOx nanoparticles are observed. The striations are probably due to small crumples.

5.2.4 DF-TEM analysis of the crystallinity of BNNS Solid and melted Cu substrates can dramatically affect the thickness and crystallinity of BNNS, which is highly visible in the DF-TEM images (Figure 5.7). These images were acquired by imaging with a diffracting spot from one particular pattern. These images were then colored and merged.205,236 Figure 5.7a shows a BF-TEM image of two BNNS triangular islands that overlap. The two sets of hexagonal spots obtained from the dotted circular region in Figure 5.7a suggest that these two triangular grains have a relative rotation of about eight degrees, as determined from the vertical angle of the two lines in Figure 5.7b. A schematic diagram of the arrangement of the two BNNS triangular islands and the related SAED patterns from the circled regions are presented in Figure 5.7c. DF-TEM images (Figure 5.7d, e) formed by selecting the ringed spots 1-210 (green and red circles in Figure 5.7b) from the two different crystals highlight the regions comprising a seven-layered BNNS triangular pyramid (colored green) and the other BNNS triangle (colored red). Figure 5.7f shows a montage of the two BNNS triangular islands where the two differently oriented crystals are easily recognized by their colors.

Figure 5.7: Grain size analysis using the DF-TEM technique. (a) BF-TEM image of two merged BNNS triangles. (b) SAED pattern of the overlapping region (marked by a dotted circle in (a)) shows two sets of spots with a hexagonal symmetry. (c) Schematic diagram of the overlapping triangular pyramids and the corresponding SAED patterns from the marked circular areas. (d) and (e) Color-coded DF-TEM images, derived by selecting the ringed (red and green) 1-210 spots in (b). (f) Combined DF-TEM image, obtained by merging the images in (d) and (e). (g) DF- TEM images show that BNNS grown on solid Cu is polycrystalline, and the grains are mostly a few hundred nanometers in size. Some overlapping grains are identifiable from their mixed colors. (h) Large single-crystalline BNNS of several microns in size is obtained on melted Cu substrate. Inset is the red green blue (RGB) color model image. 84

This technique was applied to image the crystallinity of BNNS grown on both solid and melted Cu substrates (Figure 5.7g, h). Large numbers of differently oriented small crystals with lateral sizes of a few hundred nanometers are readily identifiable for the BNNS grown on the solid Cu sample, within the 2.4 μm × 2.4 μm area (Figure 5.7g). In some regions, there are overlappings of different crystals, as indicated by the mixed colors. In contrast, for the BNNS growth on melted Cu substrate, a single crystalline region of 4.25 μm × 2.4 μm is observed (Figure 5.7h). The larger crystalline BNNS on the melted Cu substrate compared to the solid Cu substrate can be attributed to the absence of grain boundaries, step edges, rolling-induced topography, and other defects in the liquid Cu surface.157

5.2.5 Stacking order of CVD grown BNNS The stacking order of bi-layers of BNNS grown on melted Cu substrate was analyzed first. Figure 5.8a is the low magnification BF-TEM image of BNNS that was grown on melted Cu substrate. The SAED pattern from the circular area marked in Figure 5.8a shows three sets of hexagonal spots. Using a small aperture filter along the each spot marked in Figure 5.8b, the corresponding DF-TEM images can be obtained (images are marked by 1, 2 and 3 in Figure 5.8). Figure 5.8c is the combined DF-TEM images, where two BNNS domains (colored in red and green respectively) are seen overlapped with each other. The appearance of yellow color ribbons (due to the mixing of red and green colors, according to RGB color model which is shown in Figure 5.7) across the domain boundary region confirms that the two domains are overlapped. The region was also analyzed under HAADF-STEM imaging mode (Figure 5.8d). An FFT from this region reveals 9.6° rotational orientation difference between the sheets, which matches the angular difference in the SAED pattern of the two related crystals (Figure 5.8b). An FFT filtered image of Figure 5.8d clearly demonstrates the formation of Moire pattern in that region, indicating the absence of any ordered stacking in these bi-layers (Figure 5.8f). Thus, the bi-layers of BNNS on the melted Cu substrate are mostly located along overlapping edges of two BNNS crystals.

Figure 5.8: (a) BF-TEM image of BNNS on a lacey carbon TEM grid. (b) SAED pattern taken from the red-circled spot in (a) reveals three differently oriented h-BN crystals in that region. 1, 2, and 3 are DF-TEM images obtained along the red, green, and blue aperture filters marked in (b). (c) Combined DF-TEM color image (from 1, 2, and 3) reveals the crystalline structure in each region. Overlapping is observed along grain boundaries. (d) Raw ADF-STEM image from the overlapped region shown in (c). (e) FFT pattern obtained from (d), showing two different sets of hexagonal spots, rotated about 9.6° with respect to each other. (f) FFT filtered image of (d) showing a clear Moiré pattern developed due to the absence of definite stacking orders.

The stacking sequences in the BNNS grown on solid Cu substrate, from bi-layer region to multilayer triangles, were then studied using aberration-corrected HAADF-STEM. HAADF-STEM images are very sensitive to the thickness and atomic number295, which allows the composition of the atom/column sites to be identified238, and consequently, the number of layers and the stacking order to be elucidated. Due to the multilayer nature of BNNS on solid Cu substrate, the numbers of layers present in a particular BNNS region was determined first, by the focused electron beam rastering over an area of ~ 2 nm by 2 nm. Electron beam-induced ablation caused layers of BNNS within this region to be stripped away until a hole formed. It was then possible to measure the changes in intensity in the HAADF-STEM images relative to that of the hole (vacuum). The intensity profile in Figure 5.9b shows the background signal of the hole and stepwise increase in intensity due to additional BN monolayer.

Figure 5.9: (a) Raw HAADF-STEM image where a hole (dark) is visible. (b) Intensity profile along the rectangle in (a) shows a stepwise increment in intensity corresponding to vacuum, 1 layer, and 2 layers of BN.

Figure 5.10a shows an HAADF-STEM image with a monolayer region in the center and a bi-layer around it. The intensity profile (the profile next to Figure 5.10a) from the monolayer clearly identifies the N and B atoms due to their different atomic number (the signal intensity is sensitive to approximately the atomic number squared), and it is clear that the centres of the hexagonal cell (marked with arrows) are empty. The corresponding profile taken from the bi-layer region shows that there is now an atom at the center of the hexagonal cell (marked with arrows). The intensity profile suggests that the N atoms of the second layer are superimposed on the B atoms of the 87

first layer, and the B atoms of the second layer sit at the centers of the hexagonal cells of the first layer. This corresponds to AB stacking, which is the energetically favored structure.46–48 This is confirmed by the broad agreement between the experimental results and an HAADF-STEM simulation of an AB stacking sequence (Figure 5.10a).

Figure 5.10b shows an HAADF-STEM image of three-layer and four-layer regions. The image has been deconvoluted using the maximum entropy method, as implemented by the DeConv HAADF plug-in from HREM Research Inc., to reduce noise and enhance contrast. This region is part of the same crystal shown in Figure 5a, in which the first two layers are confirmed to have an AB sequence. Therefore, in the third layer, there are six different stacking possibilities, i.e. ABA, ABA´, ABB, ABB´, ABC, and ABC´. By comparing the simulated images from these candidate structures (Figure 5.11) with the experimental HAADF-STEM image (Figure 5.10b), the possibilities can be narrowed down to ABA, ABB´ or ABC´. The intensity profiles arising from these possibilities were then simulated and compared with experiment. The intensity profile simulated on the basis of ABA stacking (the profile next to Figure 5.10b) provides the best match with experiment. Confirming the three layer sequence to be ABA, the four layer region in Figure 5.10b can have again another six types of stacking such as ABAA, ABAB, ABAC, ABAA´, ABAB´, ABAC´. However, this intensity matching analysis was not continued beyond three layers, since as the number of layers increases, the contrast variations become smaller, and due to the inherently noisy signal in the experimental images, reliable and unambiguous matching becomes impossible.

Figure 5.10: Identification of stacking orders in BNNS. (a) FFT filtered HAADF- STEM image of BNNS showing bi-layer surrounding a monolayer region. Simulated images together with normalized experimental (black) and calculated (red) intensity profiles along the arrows are shown. The center positions of the hexagonal cells are marked by black arrows. The experimental bi-layer structure matches that calculated for AB stacking. (b) Deconvoluted HAADF-STEM image of the 3/4-layer region, where the trilayer is ABA stacked. (c) FFT filtered HAADF-STEM image of bi-/tri- layer region. The bi-layer adopts AC´, and the tri-layer is AC´B stacked. 89

ABA stacking ABA´ stacking ABB stacking

ABB´ stacking ABC stacking ABC´ stacking Figure 5.11: Simulated annular ADF-STEM images of the six different types of h-BN stacking possible for the tri-layered BNNS in Figure 5.10b. The first two layers of this tri-layered structure have been found to be AB (Figure 5.10a). Therefore, six other stacking orders are possible in the third layer.

In the CVD grown BNNS on solid Cu substrate, the AC´ stacking order was also observed in the first two layers, as observed in the HAADF-STEM image in Figure 5.10c. This structure has been theoretically predicted to be highly energetically unfavorable. The HAADF-STEM simulated image and the intensity profile along the <0100> direction (white arrows) show a very good match with the experimental intensity profile for the bi-layer region (Figure 5.10c). As with the preceding case, probing the stacking order of the surrounding tri-layer region was carried out. The good match between the experimental and simulated intensity profiles strongly suggest that the tri-layer is AC´B stacked. Simulated HAADF-STEM images of all the bi- and tri-layer stacking are summarized in Figure 5.12.

AB stacking AB´ stacking AB´A stacking AC Stacking

ACA stacking ACC stacking ACC´ stacking AC´ stacking

ACÁ stacking ACÁ´ stacking ACĆ stacking ACĆ´ stacking

AA stacking AAB stacking AAC stacking AAC´ stacking

AA´ stacking AA´B stacking AA´C stacking AA´C´ stacking Figure 5.12: Simulated ADF-STEM images of bi- and tri-layers of BNNS, having various stacking configurations.

It is hard to estimate the relative proportions of various stacking orders reported here since ADF-STEM is a highly localized technique. A wider study would be needed in order to provide a good statistical analysis of the preferential growth sequence in CVD grown BNNS. Different physical properties may arise when stacking orders in h-BN change. It has been demonstrated that AA and AB stacking in bi-layer h-BN impart different optical and topological properties.46 When used as the substrates in nanoelectronics, the various stacking orders in BNNS may affect the electronic properties of the devices. To experimentally probe these relationships, it would be ideal to control the stacking order over a large BNNS.

5.2.6 EELS analysis to identify monolayer BNNS from the few-layer The intensity profile that is generated based upon the HAADF-STEM images (Figure 5.10) can be correlated to determine the thickness of the BNNS (Figure 5.9), but this requires the presence of a background (vacuum signal) as a reference. The thickness determination can also be effectively done without drilling a hole in the film (under highly localized beam scanning/electron beam ablation). EELS is found as an effective method to differentiate 1 layer from 2+ layers by means of ELNES of the N edge, particularly for monolayers, and by quantitative EELS analysis of extracted core loss edge intensities. Figure 5.13a shows an HAADF image of a tetrahedrally-shaped multilayer BNNS island, where the edges of the layers appear bright. This is most noticeable around the base of the tetrahedron, where some contamination (mainly residual poly(methylmethacrylate) (PMMA)) is present. The EELS spectrum (not shown) did indicate the presence of C as well as B and N. The island is believed to sit on a large BNNS monolayer. An EELS core-loss spectrum image was acquired from the rectangular region (Figure 5.13a). The spectrum image was processed to remove the background signal by fitting an exponential in front of each edge of interest. The edge of interest was then integrated over a 30 eV wide window, starting at approximately halfway up the leading edge of the core loss edge of interest. The N K- edge color map is shown in Figure 5.13a with the number of layers labelled. The thickness intensity profile (b) derived from this map indicates seven distinct steps, each of around 45 x 104 counts per step. The intensity of the first step was measured to be 44 x 104 counts – indicating that it is a monolayer rather than some multiple thereof. Mapping the B edge (not shown) produced very similar results. The spike that occurs 92

at the step between the monolayer and bi-layer is real and is evident in both the HAADF image and the elemental map (Figure 5.13a). This most likely arises from contamination that is built up around the base of the tetrahedral BNNS island. Using quantitative EELS to map the N K (or B K) edge intensity across steps, enables the average intensity per step to be determined. From this, it is possible to estimate whether the thinnest region of the specimen consists of a monolayer, or is a multilayer, and if so, how many layers are present.

Figure 5.13c shows an EELS spectrum of the N K-edge from a monolayer of h-BN (red). This was confirmed to be a monolayer from direct imaging by HAADF-STEM. Equivalent EELS spectra for bi-layer (green) and 8-layer (blue) BNNS are also shown. The fine structure differences between the 2 and 8 layers are minimal. Nevertheless, subtle differences exist between the spectra from multilayer and monolayer films. The peak at 418 eV, which is highly apparent for multilayer films, is noticeably absent from the monolayer spectrum. According to McDougall et al.284, this peak is the result of anti-bonding interactions involving N-2s orbitals and B-2pxy orbitals. Since this interaction can only occur between layers, its absence is indicative of a monolayer.

Figure 5.13: Using EELS to identify the number of layers of BNNS: (a) HAADF image of 8-layered BNNS in a tetrahedrally-shaped island. EELS spectral imaging was acquired from the marked rectangular region. The N K-edge color intensity map of the rectangular region in (a) with layer numbers is shown. (b) Intensity profile across the map showing 8 distinct layers. The spike at the first step is due to contamination along the base of the tetrahedrally-shaped BNNS island. (c) EELS spectra of the nitrogen K- edge for 1, 2, and 8 layer BNNS. The peak at 418 eV appears from regions 2 or more layers thick but is absent in the spectrum of the monolayer.

5.3 Conclusion Methods to improve the quality of BNNS grown on a Cu substrate using CVD have been demonstrated in this chapter. Washing the ammonia borane precursor with pentane, followed by vacuum drying effectively removed the organic impurities that are responsible for the formation of graphite-like structures on top of the BNNS. The quartz tube is conducive to the formation of SiOx nanoparticles that etch the film. The formation of SiOx nanoparticles can be largely eliminated by shielding the growth substrate inside a confined space between two alumina boats. By flattening the Cu/W 94

substrate prior to growth and by using slow cooling through the Cu melting point, large single crystalline monolayer h-BN with sizes larger than 4.25 μm × 2.4 μm can be obtained. It is demonstrated that CVD grown BNNS adopts AB, ABA, AC´, and AC´B stacking orders, which are known to have higher energies than the most stable AA´ configuration. EELS has proven to be effective in determining the number of layers by means of N K-edge ELNES. The N K-edge ELNES is sensitive to the number of layers, with an energy loss peak at 418 eV being absent for monolayer but present for multilayer films. N (or B) K-edge quantification enables relative film thickness changes at steps in the film to be measured and thus the absolute number of monolayers in a film to be obtained without reference to the intensity of the background (vacuum). These findings are instrumental for the fabrication of high-quality BNNS via CVD and would spur studies on stacking order dependent properties and performances of BNNS.

Chapter 6

Reactivity of Cu with the presence of BNNS: from the perspective of atomically thin passivation

6.1 Introduction Metallic components are susceptible to various degrees of corrosions in different kinds of environment, thus shortening its expected service life and increasing the operation cost. The common route of preventing the corrosion is to apply a coating over the metallic surface, and generally polymer303,304 and metallic/ceramic305 coating are employed to protect the base metal. The coatings, however, are thick in order to achieve effective protection, and this can change the dimension, property, and esthetic view of the underlying components. Coating materials such as polymer will not withstand high temperatures and metals becomes unstable under harsh acidic conditions.

For the next generation nanoelectronics, nanoscale coating is needed to accommodate the compact design. Two-dimensional (2D) materials that can be fabricated into atomic thin sheets as a coating over the substrate can be a great choice. Graphene has recently been considered for this purpose since it is robust and flexible, and the hexagonal honeycomb structure can effectively block any species, including helium.306 Mixed results, however, have been reported.307–312 Good short-term anti-corrosion was reported,307–309 but over time accelerated Cu oxidation and corrosion were observed in the presence of graphene compared to the bare Cu substrate.313,314 This acceleration is because highly conductive graphene assisting electrons transfer in the two-component galvanic cell between Cu and graphene, facilitating oxygen reduction and Cu oxidation around the defects in the long run.

h-BN could, therefore, be considered for protection due to its insulating nature258,315 impermeability to small molecules (pore dia. 1.2 Å in the hexagon97), robustness and transparency.50 Moreover, it has an excellent chemical stability in most aquatic environments.93 Few-layered BNNS (7~8 nm thick, ~20 layers) has been reported to slow down the corrosion of underlying Cu substrate in a very diluted NaCl solution (0.1M) for a few minutes.245 BNNS has shown good oxidation resistance on Cu substrate at 500 °C for 30 mins (5 nm, i.e.~15 layers, BNNS grown on Ni substrate first and then transferred to Cu).92 Learned from the case of graphene which shows mixing results under various conditions, careful studies of protection by BNNS under different environments are necessary.

This work reports systematic studies on the reactivity of Cu with the presence of BNNS in water, oxygen, and air. DFT calculations, XPS, EELS analysis, EIS, LPR, and AC- SECM analyses were employed to study the reaction mechanism and to evaluate the practicality of using BNNS as an ultrathin passivation. The efficacy of protection is found closely related to the quality of the BNNS, where domain boundaries and defects are detrimental. The oxidation of underlying Cu substrate towards water was accelerated with the presence of discontinuous BNNS while slow reaction was observed when Cu was nearly fully covered by BNNS. These findings indicate that the perfection of protection will be on how to remove the defects in BNNS by improving the synthesis or how to selectively patch up the defects if not avoidable in the synthesis.

6.2 Results At first, the stability of the underlying Cu substrate fully coated by BNNS was studied in air, oxygen, and water. The BNNS grown on solid Cu surface at 1000 °C (details in the experimental section) was selected for this purpose. Figure 6.1a is a SEM image of BNNS on Cu substrate. Large numbers of multilayer islands (seen in white contrast) are found on top of the mono- to bi-layers of BNNS. The AFM height profile analysis

(Figure 6.1b, c) from this sample transferred to the SiO2/Si substrate also confirms that the islands are sited on top of mono- to bi-layer BNNS.

Figure 6.1: (a) A SEM image of the completely covered solid Cu substrates by BNNS.

Monolayered BNNS on top of Cu substrate is hard to distinguish under the optical microscope due to its poor white light absorbance (< 1.5%).250 Keeping the samples under UHV condition does not change any optical contrast (Figure 6.2a). If kept in air, however, only after 3 days, random red stripes and spots appeared on the BNNS coated substrates (Figure 6.2b), which seem to be associated with oxidized Cu.313,314 To understand the reason for much higher Cu oxidation in air in presence of BNNS coating, the samples were then kept in pure deionized water (oxygen in water was reduced by blowing nitrogen for few hours) and oxygen, the two active constituents of air that would react with Cu. Optical images were taken after 3 days. Under oxygen atmosphere, change to the contrast was negligible and the sample appeared similar to that kept under UHV (Figure 6.2a). Under water, dark red stripes were observed (Figure 6.2c), indicative of more oxidation of Cu. These results prove that with the presence of BNNS, Cu tends to react with water more and not so much with oxygen.

Figure 6.2: Oxidation of Cu surface in the presence of BNNS. When fully covered, (a) Cu surface experiences a negligible change in contrast when kept in oxygen or UHV; (b) Red spots and stripes show up after being kept in air for 3 days; (c) Dark red stripes were observed after being kept in water for 3 days. The Reddish region in the optical image is associated with oxidized Cu.

To elucidate the formation mechanism of these red stripes, Cu substrate partially covered by BNNS (Figure 6.3a) was prepared and tested under similar conditions. For the air exposed sample (Figure 6.3b), Cu substrate underneath the triangular BNNS, now, clearly seems to oxidize more compared to the bare state. Red lines indicating relatively more oxidized positions are overlapping with the edges of BNNS triangles (indicated by the black arrows). Also, along the domain boundaries (indicated by blue arrow), more Cu oxidation have been observed.

Figure 6.3: (a) A SEM image of Cu substrate partially covered by BNNS. Black regions are BNNS. (b) Optical image of the corresponding sample shown in (a), after 3 days of air exposure. Black and blue arrows indicating more Cu oxidation along the edges and domain boundaries of BNNS, respectively.

Keeping the samples under oxygen for 3 days; again, no change in optical contrast was apparent when compared with the as-prepared (Figure 6.4a), further proving the relative stability of Cu under oxygen with the presence of BNNS. In contrast, the region below the BNNS in water appeared redder than the bare one (Figure 6.4b). Raman mapping analysis within 500 – 700 cm-1 wavelength range (Figure 6.4b) demonstrates the correlation between the oxides and the BNNS in terms of relative positions and appearance (triangle shapes) (Figure 6.4c). There are strong Raman bands from the coated regions (region 2) that are associated with Cu2O peaks, in contrast to featureless Raman spectrum collected from the bare (region 1). Additionally, more intense Cu oxide peaks (greenish and reddish regions in the Raman mapping) are observed directly below the BNNS multilayer islands, edges and domain boundaries (see also Figure 6.3b).289,316 Thus, the red stripes observed for the fully covered Cu (Figure 6.2b, c) are due to oxidation of Cu directly under the defects in the nanosheets.

Figure 6.4: When partially covered, (a) Cu shows negligible change in contrast after being kept in oxygen/UHV; (b) strong oxidation of underlying Cu was observed after 3 days in water exposure. (c) Raman mapping from the marked rectangular region in (b) reveals a close correlation between the oxidation and the BNNS, where more oxidation is seen underneath the BNNS. (d) XPS analysis reveals more oxidation of the BNNS coated Cu compared with bare Cu.

100

XPS spectra (Figure 6.4d) of a bare Cu substrate and fully coated one after 1-month air exposure reveal weak satellite peaks of Cu 2p3/2 and Cu 2p1/2, attributed to a thin 317 layer of Cu2O. From the peak deconvolution, small amounts of CuO/Cu(OH)2 were also identified in the BNNS coated sample.317 The Cu2+ compounds may form directly underneath the defects of BNNS where more intense Raman peaks were observed (Figure 6.4c). No Cu2+ peak was detected on the bare Cu sample under XPS analysis. These results corroborate that BNNS accelerates Cu oxidation.

The partially grown BNNS on melted Cu substrate (the SEM image is shown in Figure 6.5a) were also examined under the optical microscope after 3 days of air exposure. The sample shows similar phenomenon to that observed on solid Cu substrate: strong Cu oxidation underneath the BNNS are apparent (Figure 6.5b).

To understand why the oxidation of Cu directly underneath BNNS is accelerated in water, DFT calculations have been performed on adsorption and dissociation of water molecules on defect-free monolayer BN/Cu(111) and on OH-terminated zigzag edge BN to model reactions at the perimeter interface between a BNNS island and the Cu(111) surface. Higher density and faster growth of BNNS have been observed on Cu(111) than all other planes.302

101

Water dissociation on the metal surface can be considered as a stepwise process318–322: * H2O + *  H2O ……...………………………………………………… (6.1) * * * H2O  OH + H ……………………………………………………….. (6.2) OH*  O* + H* …………………………………………………………. (6.3)

* where the asterisk (*) denotes surface, and M denotes adsorbed species M (M = H2O,

OH, O, H). DFT calculations demonstrate that H2O* (Eq. 6.1) preferably adsorbs on top of Cu atoms on the Cu(111) surface and is oriented almost parallel to the surface, which are in a good agreement with the previous DFT results.320,321 The calculated adsorption energy of H2O on Cu(111), Ead(H2O) = E(H2O/Cu(111)) – E(Cu(111)) –

E(H2O), is -0.35 eV. Here, E(H2O/Cu(111)), E(Cu(111)), and E(H2O) represents the energy of the water/surface system, the energy of the Cu(111) surface, and the energy of a free water molecule, respectively.

Figure 6.6 demonstrates the overall change in energy upon water dissociation. The first * H abstraction from H2O on the Cu(111) surface (Eq. 6.2) is an exothermic process, as shown by the black line. OH* and H* intermediates prefer to be separated at large distances due to repulsion: when OH* and H* on the Cu(111) surface are growing closer, the configuration becomes less stable. The last step of water dissociation on Cu(111), H abstraction from OH* (Eq. 6.3), is not favorable energetically, in comparison with the OH* + H* configuration. Thus, the calculations demonstrate that on a Cu(111) surface, water dissociation into OH* and H* species is the most favorable channel.

102

For the BNNS/Cu(111) interface, the most stable configuration is that N atoms sit on top of Cu atoms with N–Cu distance of 2.94 Å, while B atoms are located on top of the face centered cubic (fcc) hollow sites, which agrees with the theoretical studies.174,269 Transition metal support can considerably modify the chemical properties of inert monolayer BN due to the mixing between metal-d and BN-p orbitals, and electron sharing at the BN-metal interface.174,256,269,323,324 DFT calculations show weaker interactions between a water molecule and the terrace of BN/Cu(111) compared to that with the pure Cu(111) surface. It has been found that * H2O is physisorbed on the terrace of BN/Cu(111), and H abstraction from H2O and OH* is strongly energetically unfavorable (Figure 6.6, red line), because the relative energies of OH* + H* and O* and 2H* configurations are significantly positive in comparison with the free H2O. Intercalation of H2O between BNNS and Cu(111) is essentially endothermic and requires 1.5 eV (Figure 6.6, blue line). Once sandwiched * * * between BNNS and Cu, H2O can dissociate into OH and H with the energy release of 1 eV. A pair of OH* and H* species underneath BNNS, however, is not energetically favorable in comparison to adsorption on the pure Cu(111) surface, mainly because of the large size of OH*. The instability of OH* underneath BNNS promotes further H

103

* * * abstraction from OH leading to the complete dissociation of H2O into O and 2H . Such dissociation is clearly energetically favorable, with the final configuration being more stable than H2O dissociation on the pure Cu(111) surface, as can be seen from * * * * Figure 6.6. Optimized configurations of H2O , OH , H , and O between BNNS and Cu(111) are shown in Figure 6.7. The stabilization of the final dissociation products underneath BNNS arises from the additional interaction of O* with the B atom on top of O*, together with Cu-O* interactions. Indeed, it is well known that atomic oxygen adsorbs preferably on fcc and hexagonal close packed (hcp) threefold hollow sites on Cu(111), with a small preference for the fcc site.325 In the case of the BN/Cu(111) system, B atoms are located on top of fcc hollow sites on Cu(111), and oxygen will interact simultaneously with Cu and B atoms of the BN layer. Such an effect can lead to an excess of highly reactive atomic oxygen on the Cu(111) surface underneath BNNS and therefore promotes oxidation of the Cu surface along with the mechanisms described for the pure Cu surface.326 (a) (b)

104

There still exists a question as to how water can intercalate between BNNS and Cu(111). Intercalation of atoms or small molecules in between h-BN or graphene layers and metal supports has recently attracted intense interest254,327,328 because it opens up new exciting ways to tune chemical reactions329 and to regulate the coupling between the two-dimensional layers and the supports330. Such intercalation usually starts from the edges or defects of the cover layer. It was, therefore, investigated, how the edge of BNNS affects H2O intercalation, choosing an OH-terminated zigzag edge of BNNS on a Cu(111) surface as the model. This choice is stipulated by recent findings showing that OH-B terminated BNNS is the most stable among possible H, O, and OH terminations.331

105

DFT calculations demonstrate that when H2O adsorbs on Cu atoms near the OH- terminated zigzag edge of BNNS, the H2O molecular plane is nearly parallel to the surface (Figure 6.8a), and the adsorption energy is similar to the case of the pure Cu(111) surface. Upon the first H abstraction from water, H* adsorbs on the Cu(111) surface close to the BNNS edge (Figure 6.8b), while OH* bridges a surface Cu atom and an OH-terminated B atom at the edge (Figure 6.8c). The edge is consequently strongly distorted due to interaction with the additional OH* group. Such a configuration is more stable when compared to OH* adsorption on the pure Cu(111) (Figure 6.6), indicating that interaction with the OH-terminated zigzag edge of BNNS stabilizes the OH* intermediate. The second H abstraction from OH* results in the adsorption of H* on Cu(111) near the edge of BNNS, while O* is stabilized at the fcc hollow site on Cu(111) underneath the edge B atom (Figure 6.8d). This simulation demonstrates that water molecules preferably dissociate along the edge of BNNS on Cu(111), with H* atoms remaining on the Cu(111) surface while O* atoms adsorb on the Cu(111) surface underneath BNNS. Such an effect would lead to an excess of highly reactive atomic O* underneath BNNS, which consequently promotes Cu oxidation. The simulation results are reflected by the more pronounced oxidation observed along the edges (Figure 6.3b and Figure 6.4b). Note that Cu oxide/hydroxide was found in the XPS and Raman spectra, meaning that B and O do not form stable chemical bonds, rather intermediate bonds during the oxidation of Cu underneath BNNS.

The simulated interaction between boron and oxygen makes it worthwhile to to see if h-BN itself is oxidized at the atomic scale. The XPS analysis of the B 1s and N 1s peaks of the BNNS on Cu after 6-month air exposure does not reveal any B-O or N-O peaks (Figure 6.9a, b). A 5-layer thick h-BN triangular pyramid is shown in an ADF- STEM image (Figure 6.9c), and the layer numbers were identified from the intensity profile.238 Using EELS analysis, the presence of –B (OH) along BNNS edges can be identified.332 EELS mapping was thus carried out to examine the possible presence of B-O bonds in the marked regions (green rectangle in Figure 6.9c). The extracted average EELS spectra from the blue and red rectangular areas (Figure 6.9d and e respectively) did not indicate any O K-edge peak at 532 eV, both on the edges and in

106

the plane. XPS and EELS results indicate that B and O interaction is more in a transition state and does not form stable chemical bond.

Figure 6.9: XPS analysis of (a) B 1s and (b) N 1s peaks of the BNNS on Cu after 6 months of air exposure. B-O or N-O peaks are not detected from the peak deconvolution analyzes. (c) ADF-STEM image of the 5-layer h-BN triangular pyramid. The intensity profile indicates that the layer numbers is correlated with the ADF-STEM intensity. (d, e) EELS analysis of the red rectangle (along the edge) and blue rectangle (along the plane) in (c), respectively. No oxygen was detected in both places.

107

Mixed reports are available as to the efficacy of graphene as an ultrathin anti-corrosion layer.307,308,313,314,333 Some claim that the high electronic conductivity of graphene accelerate the Cu oxidation around the defects in the long run313,314, while it is suggested recently that the strong bonding between graphene and the substrate such as Ni and Co prevents the migration of active reactants and consequently prevents oxidation of the metal substrates, even with defects in graphene.333 It is demonstrated above that the partially coated Cu by BNNS experienced enhanced reactivity along domain boundaries and defects while fully covered Cu shows only spotty oxidization. The kinetics of Cu oxidation slows down significantly after around one week in air (Figure 6.10). Under similar conditions, the degree of Cu oxidation with the presence of BNNS is much less severe compared with that of graphene,313,314 judged by the optical color contrast. Also, the insulating nature of BNNS is expected to protect the strongly bounded materials more efficiently than graphene. This means that protection of Cu could be achieved through developing high-quality BNNS that features large domains and limited defects. Acknowledging the importance of ultrathin coating, the performance of ultrathin BNNS coatings towards anticorrosion application has been investigated and the results are compared with the bare Cu substrates in 0.5 M NaCl solution.

Figure 6.10: Optical microscopy image of fully coated solid Cu substrate with BNNS, (a) after 1 day (b) after 1 week and (c) 1 year of air aging. The degree of Cu oxidation can be correlated from the optical contrast variation313,314 (dark red for heavily oxidized Cu surface). The optical contrast variations indicate that the Cu oxidation in the presence of BNNS slowed down significantly after 1 week. It is assumed that after the formation of initial thin layer of copper oxide (favoured by B-O bonds, as discussed in the DFT calculation section above), further oxidation is slowed down due to the retardation of electron transport by the insulating BNNS.

108

It is demonstrated that high quality of BNNS (several microns large domains, mostly 1-2 layers) can be grown on melted Cu, in contrast to small, polycrystalline, and multiple-layered BNNS grown on the solid Cu.157 For the corrosion test, EIS analysis was first carried out on bare and BNNS coated Cu samples to determine the impedance (Figure 6.11). Optical microscopy, SEM, XPS and Raman analysis were performed before and after the EIS tests to analyze the integrity of the BNNS and chemical states of the sample surface.

Figure 6.11: Comparison of EIS test results between the bare and BNNS coated Cu substrates: (a) Nyquist and (b) Bode plots, acquired by EIS in 0.5 M NaCl solution after 5 minutes of immersion. (c) Nyquist and (d) Bode plots after 1 hour immersion in 0.5 M NaCl solution. Impedance values are the highest for BNNS coated melted Cu substrate. Note that both the Nyquist and Bode plots are presented as acquired for a surface of area of 0.785 cm2 and repeated for at least 3 times on each sample to check reproducibility. 109

The impedance values are represented by the Nyquist and Bode plots (Figure 6.11a, b). The electrical equivalent circuit (EEC) presented in Figure 6.12 was found to provide an appropriate fit for the experimental data. The EIS tests demonstrate a considerable increase in impedance as the result of BNNS coating. Bare melted Cu shows relatively higher resistance to corrosion than the bare solid Cu, which can be attributed to the larger grain size and smoother surface (Figure 6.13), with lesser reactive Cu atoms exposed in comparison to the solid Cu.334 For the BNNS coated melted Cu substrate, the impedance value at low frequency (also known as DC limit impedance) is about one order of magnitude higher than that of the bare melted Cu substrate (Figure 6.11b). The impedance at lower frequencies (e.g. <0.1Hz) corresponds to the charge transfer in the defective domains of the surface layer (either through the native oxide film or the BNNS coating). This result is comparable to that of graphene coated Cu307,308, which means that hexagonal networks in both BNNS and graphene can effectively block the ion diffusion. The higher impedance in the BNNS coated melted Cu substrate in comparison to the solid Cu substrate is attributed to the high quality of BNNS.

Figure 6.12: Electrical equivalent circuit used to fit the EIS data.

After 1-hour immersion, the impedance of the bare (both solid and melted) Cu increased substantially compared with the BNNS coated Cu substrate (Figure 6.11c, d). This indicates the formation of thicker corrosion oxides on the bare Cu than on the coated, corroborating the protective BNNS over the Cu substrate. Moreover, the BNNS coated melted Cu experienced the least increase in impedance, indicating the best corrosion resistance associated with high quality of BNNS. Both the surface and charge transfer impedances at lower frequencies of BNNS treated/melted Cu specimen decreased after one-hour immersion in NaCl solution (compared to the result of 5 min immersion). This is likely due to adsorption of ions at the surface facilitating charge transfer process. Comparing Figure 6.11b and Figure 6.11d, the BNNS treated/melted 110

Cu specimen did not show a reduction in surface film impedance in the mid frequency domain (i.e. 0.1Hz

Figure 6.13: Electron backscattered diffraction (EBSD) analysis of the (a) solid and (b) melted Cu substrates. Very large grains are seen on the melted Cu substrate. Inset (top right in b) is the EBSD map of a large area including the grain shown in b. The scale bars in the images are 500 μm.

After the EIS tests (1 hour in 0.5 M NaCl solution), there is a dramatic change in the optical appearance compared with the samples before the EIS test (Figure 6.14a, b). Corrosion (reddish under optical microscope313,314) are seen all over the surface of the bare solid (Figure 6.14c) and melted Cu (Figure 6.14d), in contrast to isolated regions on BNNS coated solid Cu (Figure 6.14e) and only a few small spots on BNNS coated melted Cu surface (Figure 6.14f).

Figure 6.14: (a-f) Optical micrograph images of the bare and BNNS coated Cu substrates before and after the EIS tests. The BNNS coated melted Cu is found to be the least corroded after the EIS test. Scale bar is 100 μm.

111

SEM images, before and after one hour immersed EIS tested samples are shown in Figure 6.15. After the test, the bare Cu samples are found corroded all over the surface (Figure 6.15b). Numerous pits are found on the bare substrates, indicating the Cl- ions penetrating into the substrates.335,336 Cu (I and II) oxide308 deposits have been found mostly along the rolled directions in the bare and BNNS coated solid Cu substrates (Figure 6.15b, d). Although, the multilayer BNNS islands on the solid Cu substrate157 (Inset of Figure 6.15c), are expected to protect the underlying Cu substrate more effectively than the monolayer regions; interestingly, large numbers of pits have been found on top of those BNNS islands while isolated pits are detected on the monolayer regions (inset of Figure 6.15d). The appearance of large numbers of pits on top of the BNNS islands indicates the triangles as defective spots for corrosion to proceed. SAED pattern (shown in the inset of Figure 6.15d) obtained from a single BNNS triangular pyramid, however, shows only one set of distinctive six-fold hexagonal symmetrical pattern, the signature for single crystalline BNNS. Therefore, the only defect which can present in the BNNS islands is the vacancy defect. During the synthesis step, BNNS islands were found preferentially along the high energy ridges and rough areas on the Cu surface.157 Some imperfections, thus, might arise during the formation of the multilayer islands. Indeed, central hole on some of the as-synthesized BNNS islands is detected under high magnification SEM analysis (Figure 6.16). The exact mechanism of the pit formation in the multilayer islands, however, requires further analysis and is beyond the scope of this work.

On the BNNS coated melted Cu substrate, the morphology of the nanosheet is seen mostly unaltered after the EIS test (Figure 6.15e, f). Pits, whose are likely formed along the nanosheet defects, are detected only in isolated regions. Due to the largely unchanged surface morphology of the sample, the Cl- ion penetration is presumed to occur along the cracks and defects of the nanosheet307,311, which are less on the melted substrate compared to the BNNS on the solid Cu.

112

Figure 6.15: SEM images of (a) bare solid and (b) bare melted Cu substrates, and (c- f) BNNS coated Cu substrates before and after the EIS test. (c) and (e) are the SEM images of BNNS coated solid and melted Cu substrates respectively. Cu oxide layers and pits are visible after the test, on solid and melted Cu substrate (b). Large pits associated with the corrosion products are highly visible for BNNS coated solid Cu (d), while pitting is hardly visible for BNNS coated melted Cu (f). Inset of (c) is a high magnification image of a four-layered tetrahedral shaped BNNS island. SAED pattern obtained from this BNNS tetrahedron identifies it as single crystalline (inset of (c)). However, after corrosion, large numbers of pits are detected on these multilayer tetrahedrons (inset of (d)). Scale bars in blue and green correspond to 5 and 1 μm respectively. 113

Figure 6.16: (a) A five-layered triangular h-BN pyramid and (b) two inverted triangles, imaged under the SEM mode inside TEM at 80 kV. Small holes in the center of the pyramids are apparent in both images, which formed during the growth.

XPS analysis on the EIS tested samples was carried out to analyze the corrosion products. Strong Cu 2p3/2 and Cu 2p1/2 satellite peaks (Figure 6.17a) confirm the 2+ 317 presence of Cu compounds [CuO or Cu(OH)2] on the bare Cu. The ratio between integrated peak areas associated with CuO/Cu(OH)2 and Cu/Cu2O indicates the degree of oxidation. The BNNS coated solid and melted Cu have clearly suffered less corrosion, as evidenced by the weak satellite peak of Cu 2p1/2 and Cu 2p3/2 peak (Figure 6.17a). A similar trend from the Cl 2p peak analysis (Figure 6.17b) has been observed, where the integrated peak area of the Cl 2p peak for the BNNS coated melted Cu substrate is the smallest among all samples.

Figure 6.17: XPS analysis of (a) Cu 2p peaks and (b) Cl 2p peaks indicates that BNNS coated melted Cu suffers the least corrosion.

114

Further Raman analysis confirms less oxidation on the BNNS coated Cu samples (Figure 6.18).313 The intense peaks at 531 and 614 cm-1, small peaks at 338 and 414 cm-1 and other peaks appearing at higher wavenumbers suggest that bare Cu surface 289,313 underwent Cu(OH)2 or CuO formations, corroborating the XPS results. For the BNNS coated sample, weak Cu oxidation peaks were detected on the solid Cu while they are nearly invisible for the melted Cu.

Figure 6.18: Optical images of (a) bare solid, (b) bare melted Cu sample before the EIS test, (c) bare solid, (f) bare melted Cu, (e) BNNS coated solid, and (f) BNNS melted Cu after the EIS test. (g) Raman spectra from the circled regions in (a-f). Bare Cu samples are seen heavily oxidized while the BNNS coated melted Cu sample suffered the least oxidation. Scale bar is 10μm.

In addition to the anti-corrosion study in 0.5 M NaCl solution, the short and long-term stability of the BNNS coated Cu substrates at 200 °C in air was examined. For this purpose, partially BNNS coated solid and melted Cu substrates (details in the experimental section) were prepared to examine the anti-oxidation behavior of Cu in the presence of BNNS (Figure 6.19). Initially, oxidized the BNNS coated Cu substrates

115

were oxidized at 200 °C in air for 10 minutes. BNNS can indeed protect the underlying Cu substrates from air oxidation within this short time (Figure 6.19a, b), as was reported in the previous studies.91,92,245 The reddest spot in the optical image of Figure 6.19a is correlated with the strongest oxidation. A Raman mapping taken within the 500-700 cm-1 wavelength range from the marked dashed rectangular region (Figure 6.19a) shows the distribution of oxides (Figure 6.19b). The Raman spectra collected from the highlighted regions in Figure 6.19a indicate weak Cu2O formation (331, 525 and 620 cm-1) on top of the BNNS while the bare Cu substrate has suffered a severe oxidation (high Raman oxidation peak intensities). To examine the efficiency of BNNS protection on Cu for a long term, the BNNS coated samples was kept at 200 °C in air for 10 days and analyzed the substrates under the optical and Raman microscopy (Figure 6.19c-d). The BNNS coated regions turned grayish, which is associated with the presence of CuO (282, 334, 602 cm-1)316. The bare Cu regions turned greenish

(Figure 6.19c), which is due to the presence of Cu2(CO3)(OH)2, Cu(OH)2NO2, or other complex Cu oxides and hydroxide313,337.The course of reaction in the BNNS coated and uncoated Cu substrates over the long term oxidation, therefore, seems different. The molecules from the atmosphere, such as water, carbon dioxide, nitrogen dioxide, sulfur dioxide etc. cannot interact with the underlying Cu substrate in the presence of BNNS while those are readily available to react with the bare Cu substrate. Raman mapping (collected at 450-700 cm-1 range from the marked square region in Figure 6.19c) clearly shows that the heavily oxidized regions are the uncovered Cu (Figure 6.19d). BNNS thus can protect the underlying substrate for a short term. In the long run, oxygen ion migration occurs along the BNNS edges and defects and gradually the whole Cu surface underneath the BNNS is oxidized.

116

Figure 6.19: Oxidation of Cu substrates partially covered by BNNS at 200 °C in air for 10 minutes and 10 days. (a) After 10 min, BNNS coated and bare Cu regions are observed as whitish and red regions, respectively, in the optical image. (b) Raman mapping of the marked area in (a) taken in the range of 450-700 cm-1. Images marked with 1 and 2 represents the Raman spectra from the circled areas of (a). (c) Optical image of the sample after 10 days. Grayish areas are BNNS covered and the greenish regions are uncovered. (d) Raman mapping from the marked rectangular region in (c) taken in the range of 450-700 cm-1. (3) and (4) represents the Raman spectra from the highlighted regions in (c). BNNS can protect the Cu substrate for a short term. In the long run, oxygen ion migration occurs along the BNNS edges and defects and gradually the whole Cu surface underneath the BNNS is oxidized.

117

LPR is a common non-destructive electrochemical technique to determine the corrosion resistance. The resistance, derived from the inverse of the slope of the potential vs current curves determines the corrosion rate of the samples. The smaller slope values reflect higher resistance in the BNNS coated Cu (almost double that of the bare Cu substrate). Currents generated from the BNNS coated Cu are lower than the bare Cu at the same potential (Figure 6.20), which indicates a good protection by BNNS.

Figure 6.20: Linear polarization resistance (LPR) curves of the bare and BNNS coated Cu substrates (collected on an area of 0.785 cm2). Resistance to corrosion was determined from the inverse slope value of the trendline of each curve. The BNNS coating has doubled the charge transfer resistance.

Being insulated, BNNS could prevent the formation of galvanic cells and thus slow down the corrosion. To acquire local information regarding insulating and/or electro- activity of the 1~2 layered BNNS, AC-SECM technique was resorted (methods are in the experimental section). Normalized AC approach curves (Impedance vs. tip- substrate separation distance curves, with the tip approaching the surface) of the surface of the bare and BNNS coated Cu samples are presented in Figure 6.21a, b. A negative feedback, i.e., decreasing normalized impedance as the ultra-micro-electrode

118

(UME) tip (dia. 25 μm) approached the Cu surface, was observed for the bare solid and melted Cu at the entire frequency spectrum, i.e. 1 kHz to 100 kHz (Figure 6.21a). This indicates electrically conducting characteristic of the surface. In contrast, a positive feedback was found for the BNNS coated samples (Figure 6.21b) at low frequencies (i.e. f < 8 kHz), while negative feedback was observed at high-frequency domains (i.e. f > 15 kHz). The AC response at low frequencies is more sensitive toward surface properties (insulating or conducting), whereas at high frequency the local ionic strength of electrolyte dominates the response. Multiple AC-SECM tests at various locations yielded similar data, confirming that mono- to few-layers BNNS can effectively insulate the conducting Cu surface.

Figure 6.21: Normalized AC-SECM approach curves from 1 kHz to 100 kHz in 1 mM NaCl solution, obtained using a Pt microelectrode with 25 μm diameter. (a) bare solid and melted Cu (inset), and (b) BNNS coated solid and melted Cu substrates (inset), respectively. The positive feedback indicates the insulating nature of the BNNS coating.

These LPR and AC-SECM results indicate that BNNS could be a candidate for protection in seawater like environment. The intrinsic insulating nature of BNNS makes the formation of the galvanic cells impossible, and the hexagonal network is impermeable to small ions and thus difficult for charge transfer to occur, both of which slow down the corrosion reaction. Thickness in relation to corrosion is much less important compared with the quality of the BNNS, where defects and domain boundaries in the nanosheet appear to be detrimental. BNNS grown on melted Cu is of predominantly mono- or bi-layers with large domains (so fewer domain boundaries) 119

and very smooth surfaces (less in-plane defects)157, and it therefore effectively protect the underlying Cu substrate.

6.3 Conclusion The reactivity of Cu substrate coated by few-layered BNNS towards water, oxygen and air has been studied from the perspective of ultrathin passivation. Compared with the exposed section, Cu that is underneath BNNS shows an accelerated reaction towards water. DFT calculations indicate that dissociation of H2O into O* and 2H* is most energetically favorable along the BNNS boundaries, and the interactions between O* and B lead to accumulation of O atoms between BNNS and Cu, which accelerates the formation of Cu oxide/hydroxide. Being chemically inert and thermally stable, BNNS could be a great candidate for atomically thin protective coating. Experimentally, the quality of the nanosheet rather than thickness has proven to be the critical factor for protection. BNNS coating can dramatically change the oxidation pathways of Cu when heated at 200 °C for up to 10 days, proving strong resistance to oxidation. The defects in BNNS allow air to diffuse through the coating and then oxidize the underlying Cu gradually. Mono- and bi-layered BNNS grown on melted Cu with larger domains and lesser defects has shown outstanding anti-corrosion resistance in 0.5 M NaCl solution than the multilayer BNNS grown on solid Cu. The challenge is to how to eliminate the defects in BNNS so the contact between the corrosion reagent and the underlying substrate is removed. One immediate solution could be patching up the defects as using inert atomic layer coatings as in the case of graphene.311 For the long-term solution, it could be necessary to grow high-quality BNNS with large grain size and minimized defects. Achievements in this regard will also benefit the exploration of graphene-based electronics where BNNS has been found to be the best substrate.19

120

Chapter 7

Conclusion and future outlook

7.1 Summary The main focus of this dissertation is to grow high-quality BNNS that is free from impurities and defects, and features large domains and homogeneity. A detailed characterization study has been conducted to analyse the quality of the BNNS. The stability of BNNS in air and its potential as an anti-corrosion protective layer on Cu substrate have been also reported in this dissertation. The major findings from this work can be summarized in the following sections.

7.1.1 Growth of large and homogenous BNNS Cu substrates, both solid and melted, are used for the growth of BNNS. The samples grown on these two substrates were analysed by optical, AFM, XPS, Raman, SEM and TEM techniques to confirm the quality of the BNNS. The rolling-induced defects and grain boundaries on the solid Cu surface are found to lead to the formation of multilayer island-like growth of BNNS on those sites. Thus, inhomogenous growth with nanocrystalline domains of BNNS is observed on top of the solid Cu substrate. By carrying out the BNNS growth on melted Cu substrate, the imperfections that are observed on the solid Cu substrate can be eliminated, which consequently leads to the formation of homogeneous (1-2 layers) BNNS with large domains (more than 4.25 μm × 2.4 μm).

7.1.2 Crack- and carbon-free growth of BNNS Although higher quality BNNS can be grown on melted Cu substrate when compared with solid Cu substrate, micron sized cracks are detected in the nanosheet after the growth on melted Cu substrate. This is due to the elliptical shape of the Cu substrate, which generates a large temperature gradient from the edge towards the center during cooling, and consequently, creates high tensile stress for the as-grown BNNS. By

121

flattening Cu and W foils and slowly cooling near the Cu melting point, these large cracks can be avoided. In addition to the large cracks, large numbers of nanosized cracks are seen all over the nanosheet. SiOx nanoparticles (generated from the quartz tube) have been detected inside all of these nanosized cracks, which possibly etched the nanosheet. The formation of these cracks has been largely eliminated by carrying out the growth inside two alumina boats, but leaving a small aperture to give access to borazine gas.

In many of the experiments, micron-sized graphite-like particles are observed on top of BNNS. The residual solvents from ammonia borane are likely to be responsible for the generation of these graphite-like particles. Using pentane treated ammonia borane followed by ultra-high vacuuming for several days can produce carbon-free growth of BNNS.

7.1.3 Observation of unusual stacking in BNNS h-BN is generally considered to be AA´ stacked, where the boron atoms sit on top of nitrogen atoms and vice versa in the succeeding layers. Theoretical studies, however, confirm that AB stacking (similar to graphite) is also energetically favorable. Using the HAADF-STEM imaging technique, several types of stacking orders have been found in the CVD grown BNNS, such as AB, ABA, AC´, and AC´B. Among these, AC´ stacking was considered to be highly unstable according to the theoretical calculations. These findings could be useful for perfecting the CVD growth of large and uniform BNNS. The interesting stacking orders would spur research interest in how to control the stacking structure, its impact on the properties of BNNS, and on nanoelectronics where BNNS is used as a key component.

7.1.4 EELS assisted thickness analysis A new way to identify monolayer BNNS from 2+ layers by the EELS near edge fine structure analysis of the N K-edge, was also developed. This is useful, since identification of monolayer BNNS from its bulk counterpart is often a tedious job under conventional TEM analysis. For the monolayer BNNS, the third peak in the N K-edge is always found to be absent. Since the third peak (anti-bonding interactions 122

within N 2s to B 2pxy orbitals) appears from the interaction of subsequent layers, its absence is indicative of monolayer BNNS. Once the monolayer is identified, the successive layers can be determined from the quantitative EELS analysis of extracted core-loss edge intensities.

7.1.5 BNNS as an ultrathin passivation layer on Cu substrate BNNS is a large-band-gap insulator, stable at high temperature, and is generally considered inert in most aqueous environments, which makes it a promising candidate for a passivation layer. Contrary to intuition, however, BNNS was found to catalyze the oxidation of the underlying Cu substrate. Systematic studies that involved keeping the BNNS coated Cu substrate in air, oxygen, and water environments proved that water is responsible for the accelerated Cu oxidation in the presence of BNNS. DFT calculations suggest that the boron atoms give additional stability to the oxygen atoms underneath the nanosheet. Oxygen can diffuse along the defects and edges of the nanosheet, and consequently, oxidize the underlying Cu substrate. Therefore, the growth of large crystalline BNNS is essential for the effective protection of the underlying Cu substrate. The electrochemical tests in 0.5M NaCl (close to seawater concentration) of the BNNS grown on melted Cu show that it has outstanding anti- corrosion resistance, much higher than the multilayer BNNS grown on solid Cu. This has been attributed to the large domains and fewer defects in the BNNS grown on melted Cu substrate.

7.2 Suggestions for future work

7.2.1 BNNS growth

The CVD growth of BNNS should be carried out inside alumina tubes to avoid SiOx nanoparticles, which has been found to etch the nanosheet. The cooling rate through the Cu melting point temperature should be optimized to avoid cracks in the BNNS.

As ammonia borane gives off a large volume of borazine in a sudden decomposition, the control of precursor gas flow is often difficult. Controlled flow of borazine using

123

a mass flow controller is recommended. Use of an alternative source of precursor can be another option.

In this study, Cu substrate was used for the growth of BNNS, where melted Cu was shown to be beneficial for the high-quality growth. Other transition metal substrates, such as Ag, Au, and non-transition-metal substrates, such as Zn and Ge can be used as substrates in their melted condition. Although, the growth of BNNS is found better on melted Cu than on solid Cu substrate, the exact growth mechanism of BNNS on melted Cu substrate is still unclear. Future work thus needs to be focused on unravelling the mechanism of BNNS grown on liquid metal surface through experimental and detailed theoretical analysis, which should aid in growing large and homogeneous BNNS.

Transferring the BNNS onto SiO2/Si wafer, glass, or polymeric substrates often leaves polymeric residues and generates defects and cracks in the nanosheet, which lower the overall performance of the nanosheet. Further research is needed to directly grow BNNS on those substrates. Alternatively, polymer-free and less destructive transfer methods need to be established for the potential use of BNNS in graphene-based nanoelectronics.

7.2.2 Origin of h-BN stacking orders and their characterization More work needs to be carried out to understand the exact origin of various h-BN stacking orders. Growing BNNS on other substrates such as Ni and Pt, and comparing the stacking order with that grown on Cu may help to explain the impact of substrates on various stacking orientations. HAADF-STEM analysis, combined with SEM and EBSD results, may help to clarify the impact of substrate grain orientation on the stacking order.

In addition, EELS spectra of BNNS with different stacking orders can be collected to find out their correlation. A similar database can be prepared to establish the relationship between stacking orders and Raman, XPS, UV-vis spectrum, etc. The stacking-order-dependent optical and electronic properties can also be determined if the control and exact orientation of the BNNS are achieved.

124

REFERENCES

1. Morozov, S. V. et al. Giant Intrinsic Carrier Mobilities in Graphene and Its Bilayer. Phys. Rev. Lett. 100, 016602 (2008). 2. Mak, K. F., Lee, C., Hone, J., Shan, J. & Heinz, T. F. Atomically Thin MoS 2 : A New Direct-Gap Semiconductor. Phys. Rev. Lett. 136805, 2–5 (2010). 3. Terrones, M. et al. Experimental and theoretical studies suggesting the possibility of metallic boron nitride edges in porous nanourchins. Nano Lett. 8, 1026–1032 (2008). 4. Kroto, H. W., Heath, J. R., O’Brien, S. C., Curl, R. F. & Smalley, R. E. C60: Buckminsterfullerene. Nature 318, 162–163 (1985). 5. Iijima, S. Helical microtubules of graphitic carbon. Nature 354, 56–58 (1991). 6. Boehm, H. P., Clauss, A., Fischer, G. O. & Hofmann, U. Das Adsorptionsverhalten sehr dünner Kohlenstoff-Folien. Zeitschrift für Anorg. und Allg. Chemie 316, 119–127 (1962). 7. Nagashima, A., Tejima, N., Gamou, Y., Kawai, T. & Oshima, C. Electronic Structure of Monolayer Hexagonal Boron Nitride Physisorbed on Metal Surfaces. Phys. Rev. Lett. 75, 3918–3921 (1995). 8. Nagashima, A., Tejima, N., Gamou, Y., Kawai, T. & Oshima, C. Electronic dispersion relations of monolayer hexagonal boron nitride formed on the Ni(111) surface. Phys. Rev. B 51, 4606–4613 (1995). 9. Novoselov, K. S. Electric Field Effect in Atomically Thin Carbon Films. Science 306, 666–669 (2004). 10. Novoselov, K. S. et al. Two-dimensional atomic crystals. Proc. Natl. Acad. Sci. 102, 10451–10453 (2005). 11. Coleman, J. N. et al. Two-dimensional nanosheets produced by liquid exfoliation of layered materials. Science 331, 568–71 (2011). 12. Aufray, B. et al. Graphene-like silicon nanoribbons on Ag(110): A possible formation of silicene. Appl. Phys. Lett. 96, 183102 (2010). 13. Liu, H. et al. Phosphorene: An Unexplored 2D Semiconductor with a High Hole Mobility. ACS Nano 8, 4033–4041 (2014). 14. Li, L. et al. Black phosphorus field-effect transistors. Nat. Nanotechnol. 9, 372– 377 (2014). 15. Dávila, M. E., Xian, L., Cahangirov, S., Rubio, A. & Le Lay, G. Germanene: a novel two-dimensional germanium allotrope akin to graphene and silicene. New J. Phys. 16, 095002 (2014). 16. Gibney, E. The super materials that could trump graphene. Nature 522, 8–10 (2015). 17. Akinwande, D., Petrone, N. & Hone, J. Two-dimensional flexible nanoelectronics. Nat. Commun. 5, 5678 (2014). 18. Dean, C. R. et al. Boron nitride substrates for high-quality graphene electronics. Nat. Nanotechnol. 5, 722–726 (2010). 19. Weitz, R. T. & Yacoby, A. Graphene rests easy. Nat. Nanotechnol. 5, 699–700 125

(2010). 20. Bresnehan, M. S. et al. Integration of Hexagonal Boron Nitride with Quasi- freestanding Epitaxial Graphene: Toward Wafer-Scale, High-Performance Devices. ACS Nano 6, 5234–5241 (2012). 21. Kim, K. K. et al. Synthesis and Characterization of Hexagonal Boron Nitride Film as a Dielectric Layer for Graphene Devices. ACS Nano 6, 8583–8590 (2012). 22. Lee, K. H. et al. Large-Scale Synthesis of High-Quality Hexagonal Boron Nitride Nanosheets for Large-Area Graphene Electronics. Nano Lett. 12, 714– 718 (2012). 23. Wang, L. et al. Monolayer Hexagonal Boron Nitride Films with Large Domain Size and Clean Interface for Enhancing the Mobility of Graphene-Based Field- Effect Transistors. Adv. Mater. 26, 1559–1564 (2014). 24. Heterostructures, N. et al. Flexible and Transparent MoS 2 Field- Effect Transistors on Hexagonal Boron. ACS Nano 7, 7931–7936 (2013). 25. Wentorf, R. H. Cubic Form of Boron Nitride. J. Chem. Phys. 26, 956 (1957). 26. Monteiro, S. N., Skury, A. L. D., de Azevedo, M. G. & Bobrovnitchii, G. S. Cubic boron nitride competing with diamond as a superhard engineering material – an overview. J. Mater. Res. Technol. 2, 68–74 (2013). 27. Taniguchi, T., Teraji, T., Koizumi, S., Watanabe, K. & Yamaoka, S. Appearance of n-Type Semiconducting Properties of cBN Single Crystals Grown at High Pressure. Jpn. J. Appl. Phys. 41, L109–L111 (2002). 28. Pedersen, H., Chubarov, M., Högberg, H., Jensen, J. & Henry, A. On the effect of water and oxygen in chemical vapor deposition of boron nitride. Thin Solid Films 520, 5889–5893 (2012). 29. Zhao, Z., Yang, Z., Wen, Y. & Wang, Y. Facile Synthesis and Characterization of Hexagonal Boron Nitride Nanoplates by Two-Step Route. J. Am. Ceram. Soc. 94, 4496–4501 (2011). 30. Alkoy, S., Toy, C., Gönül, T. & Tekin, A. Crystallization behavior and characterization of turbostratic boron nitride. J. Eur. Ceram. Soc. 2219, 1415– 1422 (1997). 31. Chopra, N. G. et al. Boron Nitride Nanotubes. Science 269, 966–967 (1995). 32. Golberg, D., Bando, Y., Stéphan, O. & Kurashima, K. Octahedral boron nitride fullerenes formed by electron beam irradiation. Appl. Phys. Lett. 73, 2441 (1998). 33. Chen, Z.-G. et al. Novel Boron Nitride Hollow Nanoribbons. ACS Nano 2, 2183–2191 (2008). 34. Ingo, G. M. et al. Novel route to high-yield synthesis of sp2-hybridized boron nitride nanoplates on stainless steel. J. Mater. Chem. 21, 10268 (2011). 35. Deepak, F. ., Vinod, C. ., Mukhopadhyay, K., Govindaraj, A. & Rao, C. N. . Boron nitride nanotubes and nanowires. Chem. Phys. Lett. 353, 345–352 (2002). 36. Qiu, Y. et al. Synthesis of continuous boron nitride nanofibers by solution coating electrospun template fibers. Nanotechnology 20, 345603 (2009). 126

37. Zheng, M., Gu, Y., Xu, Z. & Liu, Y. Synthesis and characterization of boron nitride nanoropes. Mater. Lett. 61, 1943–1945 (2007). 38. Yu, Y., Chen, H., Liu, Y., White, T. & Chen, Y. Preparation and potential application of boron nitride nanocups. Mater. Lett. 80, 148–151 (2012). 39. Pakdel, A., Bando, Y. & Golberg, D. Morphology-Driven Nonwettability of Nanostructured BN Surfaces. Langmuir 29, 7529–7533 (2013). 40. Weng, Q., Wang, X., Zhi, C., Bando, Y. & Golberg, D. Boron Nitride Porous Microbelts for Hydrogen Storage. ACS Nano 7, 1558–1565 (2013). 41. Yin, J., Li, X., Zhou, J. & Guo, W. Ultralight Three-Dimensional Boron Nitride Foam with Ultralow Permittivity and Superelasticity. Nano Lett. 13, 3232–3236 (2013). 42. Huang, B., Lee, H., Gu, B.-L., Liu, F. & Duan, W. Edge stability of boron nitride nanoribbons and its application in designing hybrid BNC structures. Nano Res. 5, 62–72 (2011). 43. Huang, B., Lee, H., Gu, B., Liu, F. & Duan, W. Edge stability of boron nitride nanoribbons and its application in designing hybrid BNC structures. Nano Res. 5, 62–72 (2012). 44. Mukherjee, R. & Bhowmick, S. Edge stabilities of hexagonal boron nitride nanoribbons: a first-principles study. J. Chem. Theory Comput. 7, 720–724 (2011). 45. Lin, Y. & Connell, J. W. Advances in 2D boron nitride nanostructures: nanosheets, nanoribbons, nanomeshes, and hybrids with graphene. Nanoscale 4, 6908–6939 (2012). 46. Kim, C.-J. et al. Stacking Order Dependent Second Harmonic Generation and Topological Defects in h-BN Bilayers. Nano Lett. 13, 5660–5665 (2013). 47. Ribeiro, R. M. & Peres, N. M. R. Stability of boron nitride bilayers: Ground- state energies, interlayer distances, and tight-binding description. Phys. Rev. B 83, 235312 (2011). 48. Constantinescu, G., Kuc, A. & Heine, T. Stacking in Bulk and Bilayer Hexagonal Boron Nitride. Phys. Rev. Lett. 111, 036104 (2013). 49. Li, Q. et al. Grain Boundary Structures and Electronic Properties of Hexagonal Boron Nitride on Cu(111). Nano Lett. 15, 5804–5810 (2015). 50. Song, L. et al. Large Scale Growth and Characterization of Atomic Hexagonal Boron Nitride Layers. Nano Lett. 10, 3209–3215 (2010).

51. Nozaki, H. & Itoh, S. Structural stability of BC2N. J. Phys. Chem. Solids 57, 41–49 (1996). 52. Pokropivny, V. V et al. Boron Nitride Analogs of Fullerenes (the Fulborenes ), Nanotubes , and Fullerites (the Fulborenites ). J. soli 154, 214–222 (2000). 53. Liu, Y., Zou, X. & Yakobson, B. I. Dislocations and Grain Boundaries in Two- Dimensional Boron Nitride. ACS Nano 6, 7053–7058 (2012). 54. Gibb, A., Alem, N. & Chen, J. Atomic resolution imaging of grain boundary defects in monolayer chemical vapor deposition-grown hexagonal boron nitride. J. Am. Chem. Soc. 135, 6758–6761 (2013).

127

55. Cretu, O., Lin, Y. & Suenaga, K. Evidence for active atomic defects in monolayer hexagonal boron nitride: a new mechanism of plasticity in two- dimensional materials. Nano Lett. 14, 1064–1068 (2014). 56. Yazyev, O. V. & Chen, Y. P. Polycrystalline graphene and other two- dimensional materials. Nat. Nanotechnol. 9, 755–767 (2014). 57. Lyalin, A., Nakayama, A., Uosaki, K. & Taketsugu, T. Theoretical predictions for hexagonal BN based nanomaterials as electrocatalysts for the oxygen reduction reaction. Phys. Chem. Chem. Phys. 15, 2809–20 (2013). 58. Jin, C., Lin, F., Suenaga, K. & Iijima, S. Fabrication of a Freestanding Boron Nitride Single Layer and Its Defect Assignments. Phys. Rev. Lett. 102, 195505 (2009). 59. Lipp, A., Schwetz, K. A. & Hunold, K. Hexagonal boron nitride: Fabrication, properties and applications. J. Eur. Ceram. Soc. 5, 3–9 (1989). 60. Engler, M., Lesniak, C. & Damasch, R. Hexagonal Boron Nitride (hBN): Applications from Metallurgy to Cosmetics. Ceram. forum Int. 84, 49–53 (2007). 61. Li, J., Dahal, R., Majety, S., Lin, J. Y. & Jiang, H. X. Hexagonal boron nitride epitaxial layers as neutron detector materials. Nucl. Instruments Methods Phys. Res. Sect. A Accel. Spectrometers, Detect. Assoc. Equip. 654, 417–420 (2011). 62. Schieber, M. et al. Polycrystalline BN and LiF Based Semiconductor Alpha Particle and Neutron Detectors. in 2006 IEEE Nuclear Science Symposium Conference Record 3728–3731 (IEEE, 2006). doi:10.1109/NSSMIC.2006.353804 63. Watanabe, K., Taniguchi, T. & Kanda, H. Direct-bandgap properties and evidence for ultraviolet lasing of hexagonal boron nitride single crystal. Nat. Mater. 3, 404–409 (2004). 64. Solozhenko, V. L., Lazarenko, A. G., Petitet, J.-P. & Kanaev, A. V. Bandgap energy of graphite-like hexagonal boron nitride. J. Phys. Chem. Solids 62, 1331–1334 (2001). 65. Zupan, J. & Kolar, D. Optical properties of graphite and boron nitride. J. Phys. C Solid State Phys. 5, 3097–3100 (1972). 66. Uddin, M. R., Li, J., Lin, J. Y. & Jiang, H. X. Carbon-rich hexagonal (BN)C alloys. J. Appl. Phys. 117, 215703 (2015). 67. Zhang, A., Teoh, H. F., Dai, Z., Feng, Y. P. & Zhang, C. Band gap engineering in graphene and hexagonal BN antidot lattices: A first principles study. Appl. Phys. Lett. 98, 023105 (2011). 68. Watanabe, K., Taniguchi, T., Niiyama, T., Miya, K. & Taniguchi, M. Far- ultraviolet plane-emission handheld device based on hexagonal boron nitride. Nat. Photonics 3, 591–594 (2009). 69. Liu, Z. et al. In-plane heterostructures of graphene and hexagonal boron nitride with controlled domain sizes. Nat. Nanotechnol. 8, 119–24 (2013). 70. Britnell, L. et al. Electron tunneling through ultrathin boron nitride crystalline barriers. Nano Lett. 12, 1707–10 (2012). 71. Petrone, N. et al. Flexible Graphene Field-Effect Transistors Encapsulated in 128

Hexagonal Boron Nitride. ACS Nano 9, 8953–8959 (2015). 72. Kobayashi, Y., Kumakura, K., Akasaka, T. & Makimoto, T. Layered boron nitride as a release layer for mechanical transfer of GaN-based devices. Nature 484, 223–227 (2012). 73. Boldrin, L., Scarpa, F., Chowdhury, R. & Adhikari, S. Effective mechanical properties of hexagonal boron nitride nanosheets. Nanotechnology 22, 505702 (2011). 74. Lee, C., Wei, X., Kysar, J. & Hone, J. Measurement of the elastic properties and intrinsic strength of monolayer graphene. Science 321, 385–388 (2008). 75. Li, C., Bando, Y., Zhi, C., Huang, Y. & Golberg, D. Thickness-dependent bending modulus of hexagonal boron nitride nanosheets. Nanotechnology 20, 385707 (2009). 76. Duan, Z.-Q., Liu, Y.-T., Xie, X.-M. & Ye, X.-Y. A simple and green route to transparent boron nitride/PVA nanocomposites with significantly improved mechanical and thermal properties. Chinese Chem. Lett. 24, 17–19 (2013). 77. Wang, Y., Shi, Z. & Yin, J. Boron nitride nanosheets: large-scale exfoliation in methanesulfonic acid and their composites with polybenzimidazole. J. Mater. Chem. 21, 11371 (2011). 78. Zhi, C. Y., Bando, Y., Tang, C. C., Kuwahara, H. & Golberg, D. Large-Scale Fabrication of Boron Nitride Nanosheets and Their Utilization in Polymeric Composites with Improved Thermal and Mechanical Properties. Adv. Mater. 21, 2889–2890 (2009). 79. Khan, U. et al. Polymer reinforcement using liquid-exfoliated boron nitride nanosheets. Nanoscale 5, 581–587 (2013). 80. Song, W.-L. et al. Polymer/Boron Nitride Nanocomposite Materials for Superior Thermal Transport Performance. Angew. Chemie Int. Ed. 51, 6498– 6501 (2012). 81. Sevik, C., Kinaci, A., Haskins, J. B. & Çağın, T. Characterization of thermal transport in low-dimensional boron nitride nanostructures. Phys. Rev. B 84, 085409 (2011). 82. Ouyang, T. et al. Thermal transport in hexagonal boron nitride nanoribbons. Nanotechnology 21, 245701 (2010). 83. Kumar, A. & Pal, D. Lattice Thermal Conductivity of Boron Nitride Crystals at Temperatures 1.5 to 300 K. Phys. status solidi 129, K9–K12 (1985). 84. Jo, I. et al. Thermal Conductivity and Phonon Transport in Suspended Few- Layer Hexagonal Boron Nitride. Nano Lett. 13, 550–554 (2013). 85. Zhi, C., Xu, Y., Bando, Y. & Golberg, D. Highly Thermo-conductive Fluid with boron nitride nanofillers. ACS Nano 5, 6571–6577 (2011). 86. Taha-Tijerina, J. et al. Electrically Insulating Thermal Nano-Oils Using 2D Fillers. ACS Nano 6, 1214–1220 (2012). 87. Yu, J. et al. Interfacial modification of boron nitride nanoplatelets for epoxy composites with improved thermal properties. Polymer (Guildf). 53, 471–480 (2012). 88. Wang, X., Weng, Q., Wang, X., Li, X. & Zhang, J. Biomass-Directed Synthesis 129

of 20 g High-Quality Boron Nitride Nanosheets for Thermoconductive Polymeric Composites. ACS Nano 8, 9081–9088 (2014). 89. Lin, Y., Williams, T. V. & Connell, J. W. Soluble, Exfoliated Hexagonal Boron Nitride Nanosheets. J. Phys. Chem. Lett. 1, 277–283 (2010). 90. Kim, K. K. et al. Synthesis of Monolayer Hexagonal Boron Nitride on Cu Foil Using Chemical Vapor Deposition. Nano Lett. 12, 161–166 (2012). 91. Li, L. H., Cervenka, J., Watanabe, K., Taniguchi, T. & Chen, Y. Strong Oxidation Resistance of Atomically Thin Boron Nitride Nanosheets. ACS Nano 8, 1457–1462 (2014). 92. Liu, Z. et al. Ultrathin high-temperature oxidation-resistant coatings of hexagonal boron nitride. Nat. Commun. 4, 1–8 (2013). 93. Rand, M. J. & Roberts, J. F. Preparation and Properties of Thin Film Boron Nitride. J. Electrochem. Soc. 115, 423–429 (1968). 94. Sainsbury, T. et al. Oxygen Radical Functionalization of Boron Nitride Nanosheets. J. Am. Chem. Soc. 134, 18758–18771 (2012). 95. Hummers, W. S. & Offeman, R. E. Preparation of Graphitic Oxide. J. Am. Chem. Soc. 80, 1339–1339 (1958). 96. Marcano, D. C. et al. Improved Synthesis of Graphene Oxide. ACS Nano 4, 4806–4814 (2010). 97. Yan, K. et al. Ultrathin Two-Dimensional Atomic Crystals as Stable Interfacial Layer for Improvement of Lithium Metal Anode. Nano Lett. 14, 6016–6022 (2014). 98. Cun, H., Iannuzzi, M., Hemmi, A. & Roth, S. Immobilizing Individual Atoms beneath a Corrugated Single Layer of Boron Nitride. Nano Lett. 13, 2098–2103 (2013). 99. Li, X., Yin, J., Zhou, J. & Guo, W. Large area hexagonal boron nitride monolayer as efficient atomically thick insulating coating against friction and oxidation. Nanotechnology 25, 105701 (2014). 100. Cho, D.-H., Kim, J.-S., Kwon, S.-H., Lee, C. & Lee, Y.-Z. Evaluation of hexagonal boron nitride nano-sheets as a lubricant additive in water. Wear 302, 981–986 (2013). 101. Lee, C. et al. Frictional characteristics of atomically thin sheets. Science 328, 76–80 (2010). 102. Lei, W., Portehault, D., Liu, D., Qin, S. & Chen, Y. Porous boron nitride nanosheets for effective water cleaning. Nat. Commun. 4, 1777 (2013). 103. Balmain, W. H. Bemerkungen über die Bildung von Verbindungen des Bors und Siliciums mit Stickstoff und gewissen Metallen. J. Prakt. Chem. 27, 422– 430 (1842). 104. Rudolph, S. Boron Nitride (BN). Am. Ceram. Soc. Bull. 79, 50 (2000). 105. Wang, L., Meric, I., Huang, P., Gao, Q. & Gao, Y. One-dimensional electrical contact to a two-dimensional material. Science 342, 614–617 (2013). 106. Li, L. H. et al. Large-scale mechanical peeling of boron nitride nanosheets by low-energy ball milling. J. Mater. Chem. 21, 11862 (2011). 130

107. Yao, Y. et al. Large-scale production of two-dimensional nanosheets. J. Mater. Chem. 22, 13494 (2012). 108. Lin, Y., Williams, T. V., Cao, W., Elsayed-Ali, H. E. & Connell, J. W. Defect Functionalization of Hexagonal Boron Nitride Nanosheets. J. Phys. Chem. C 114, 17434–17439 (2010). 109. Huang, J. Y., Yasuda, H. & Mori, H. HRTEM and EELS Studies on the Amorphization of Hexagonal Boron Nitride Induced by Ball Milling. J. Am. Ceram. Soc. 83, 403–409 (2004). 110. Streletskii, a. N. et al. Destruction, amorphization and reactivity of nano-BN under ball milling. J. Alloys Compd. 483, 313–316 (2009). 111. Zhao, G. et al. One-Step Exfoliation and Hydroxylation of Boron Nitride Nanosheets with Enhanced Optical Limiting Performance. Adv. Opt. Mater. 4, 141–146 (2016). 112. Zeng, Z. et al. An Effective Method for the Fabrication of Few-Layer-Thick Inorganic Nanosheets. Angew. Chemie Int. Ed. 51, 9052–9056 (2012). 113. Kovtyukhova, N. I. et al. Reversible Intercalation of Hexagonal Boron Nitride with Brønsted Acids. J. Am. Chem. Soc. 135, 8372–8381 (2013). 114. Du, M., Wu, Y. & Hao, X. A facile chemical exfoliation method to obtain large size boron nitride nanosheets. CrystEngComm 15, 1782 (2013). 115. Li, X. et al. Exfoliation of Hexagonal Boron Nitride by Molten Hydroxides. Adv. Mater. 25, 2200–2204 (2013). 116. Lin, Y. et al. Aqueous Dispersions of Few-Layered and Monolayered Hexagonal Boron Nitride Nanosheets from Sonication-Assisted Hydrolysis: Critical Role of Water. J. Phys. Chem. C 115, 2679–2685 (2011). 117. Han, W.-Q., Wu, L., Zhu, Y., Watanabe, K. & Taniguchi, T. Structure of chemically derived mono- and few-atomic-layer boron nitride sheets. Appl. Phys. Lett. 93, 223103 (2008). 118. Liu, Y.-T., Xie, X.-M. & Ye, X.-Y. Tuning the solubility of boron nitridenanosheets in organic solvents by using block copolymer as a ‘Janus’ modifier. Chem. Commun. 49, 388–390 (2013). 119. Singhal, S. K. et al. Gold-Nanoparticle-Decorated Boron Nitride Nanosheets: Structure and Optical Properties. Part. Part. Syst. Charact. 30, 445–452 (2013). 120. Meyer, J., Chuvilin, A. & Algara-Siller, G. Selective sputtering and atomic resolution imaging of atomically thin boron nitride membranes. Nano Lett. 9, 2683–2689 (2009). 121. Spear, K. E. Thermochemical Modeling of Steady-State CVD. in Eletrochem. Soc. 81–97 (1984). 122. Pearson, H. Handbook of Chemical Vapor Deposition. Noyes, Park Ridge, NJ (Noyes Publications, 1992). 123. Spear, K. & Dirkx, R. Role of high-temperature chemistry in CVD processing. Pure Appl. Chem. 62, 89–101 (1990). 124. Takahashi, T., Itoh, H. & Takeuchi, A. Chemical vapor deposition of hexagonal boron nitride thick film on iron. J. Cryst. Growth 47, 245–250 (1979).

131

125. Sano, M. & Aoki, M. Chemical vapour deposition of thin films of BN onto fused silica and sapphire. Thin Solid Films 83, 247–251 (1981). 126. Arya, S. & D’amico, A. Preparation, properties and applications of boron nitride thin films. Thin Solid Films 157, 267–282 (1988). 127. Paine, R. & Narula, C. Synthetic routes to boron nitride. Chem. Rev. 90, 73–91 (1990). 128. Phani, A. R. Thin films of boron nitride grown by CVD. Bull. Mater. Sci. 17, 219–224 (1994). 129. Li, J. et al. Preparation of Polyborazylene-Derived Bulk Boron Nitride with Tunable Properties by Warm-Pressing and Pressureless Pyrolysis. Chem. Mater. 22, 2920–2929 (2010). 130. Auwa, W., Suter, H. U., Sachdev, H. & Greber, T. Synthesis of One Monolayer of Hexagonal Boron Nitride on Ni (111) from B-Trichloroborazine (ClBNH)3. Chem. Mater. 16, 343–345 (2004). 131. Mu, F. & Sto, K. Symmetry versus Commensurability : Epitaxial Growth of Hexagonal Boron Nitride on Pt (111) From B-Trichloroborazine. 3, 3464–3467 (2005). 132. Frueh, S. et al. Pyrolytic Decomposition of Ammonia Borane to Boron Nitride. Inorg. Chem. 50, 783–792 (2011). 133. Whittell, G. R. & Manners, I. Advances with Ammonia-Borane: Improved Recycling and Use as a Precursor to Atomically Thin BN Films. Angew. Chemie Int. Ed. 50, 10288–10289 (2011). 134. Son, M., Lim, H., Hong, M. & Choi, H. C. Direct growth of graphene pad on exfoliated hexagonal boron nitride surface. Nanoscale 3, 3089–3093 (2011). 135. Kim, G. et al. Growth of high-crystalline, single-layer hexagonal boron nitride on recyclable platinum foil. Nano Lett. 13, 1834–1839 (2013). 136. Gao, Y. et al. Repeated and Controlled Growth of Monolayer, Bilayer and Few- Layer Hexagonal Boron Nitride on Pt Foils. ACS Nano 7, 5199–5206 (2013). 137. Wang, X. et al. ‘Chemical Blowing’ of Thin-Walled Bubbles: High-Throughput Fabrication of Large-Area, Few-Layered BN and Cx-BN Nanosheets. Adv. Mater. 23, 4072–4076 (2011). 138. Wang, X. et al. High-yield boron nitride nanosheets from ‘chemical blowing’: towards practical applications in polymer composites. J. Phys. Condens. Matter 24, 314205 (2012). 139. Qin, L., Yu, J., Li, M., Liu, F. & Bai, X. Catalyst-free growth of mono- and few- atomic-layer boron nitride sheets by chemical vapor deposition. Nanotechnology 22, 215602 (2011). 140. Yu, J. et al. Vertically Aligned Boron Nitride Nanosheets: Chemical Vapor Synthesis, Ultraviolet Light Emission, and Superhydrophobicity. ACS Nano 4, 414–422 (2010). 141. Qin, L., Yu, J., Kuang, S., Xiao, C. & Bai, X. Few-atomic-layered boron carbonitride nanosheets prepared by chemical vapor deposition. Nanoscale 4, 120–123 (2012). 142. Zhang, C., Hao, X., Wu, Y. & Du, M. Synthesis of vertically aligned boron 132

nitride nanosheets using CVD method. Mater. Res. Bull. 47, 2277–2281 (2012). 143. Ismach, A. et al. Toward the Controlled Synthesis of Hexagonal Boron Nitride Films. ACS Nano 6, 6378–6385 (2012). 144. Ammonia, D. et al. Chemical Vapor Deposition of Boron Nitride Nanosheets on Metallic Substrates via Decaborane/Ammonia Reactions. Chem. Mater. 23, 4414–4416 (2011). 145. Müller, F., Hüfner, S., Sachdev, H., Gsell, S. & Schreck, M. Epitaxial growth of hexagonal boron nitride monolayers by a three-step boration-oxidation- nitration process. Phys. Rev. B 82, 075405 (2010). 146. Yin, J. et al. Aligned Growth of Hexagonal Boron Nitride Monolayer on Germanium. Small 11, 5375–5380 (2015). 147. Jang, A.-R. et al. Wafer-Scale and Wrinkle-Free Epitaxial Growth of Single- Orientated Multilayer Hexagonal Boron Nitride on Sapphire. Nano Lett. 16(5), 3360-3366 (2016). 148. Rohr, C., Boo, J. & Ho, W. The growth of hexagonal boron nitride thin films on silicon using single source precursor. Thin Solid Films 322, 9–13 (1998). 149. Behura, S., Nguyen, P., Che, S., Debbarma, R. & Berry, V. Large-Area, Transfer-Free, Oxide-Assisted Synthesis of Hexagonal Boron Nitride Films and Their Heterostructures with MoS2 and WS2. J. Am. Chem. Soc. 137, 13060– 13065 (2015). 150. Lee, Y.-H. et al. Growth selectivity of hexagonal-boron nitride layers on Ni with various crystal orientations. RSC Adv. 2, 111–115 (2012). 151. Guo, N. et al. Controllable growth of triangular hexagonal boron nitride domains on copper foils by an improved low-pressure chemical vapor deposition method. Nanotechnology 23, 415605 (2012). 152. Shi, Y. et al. Synthesis of few-layer hexagonal boron nitride thin film by chemical vapor deposition. Nano Lett. 10, 4134–4139 (2010). 153. Yang, P. C., Prater, J. T., Liu, W., Glass, J. T. & Davis, R. F. The formation of epitaxial hexagonal boron nitride on nickel substrates. J. Electron. Mater. 34, 1558–1564 (2005). 154. Müller, F. et al. Epitaxial growth of hexagonal boron nitride on Ag(111). Phys. Rev. B 82, 113406 (2010). 155. Simonson, R. J., Paffett, M. T., Jones, M. E. & Koel, B. E. A vibrational study of borazine adsorbed on Pt(111) and Au(111) surfaces. Surf. Sci. 254, 29–44 (1991). 156. Bresnehan, M. S. et al. Prospects of direct growth boron nitride films as substrates for graphene electronics. J. Mater. Res. 29, 459–471 (2013). 157. Khan, M. H. et al. Synthesis of Large and Few Atomic Layers of Hexagonal Boron Nitride on Melted Copper. Sci. Rep. 5, 7743 (2015). 158. Bresnehan, M. S., Bhimanapati, G. R., Wang, K., Snyder, D. W. & Robinson, J. a. Impact of Copper Overpressure on the Synthesis of Hexagonal Boron Nitride Atomic Layers. ACS Appl. Mater. Interfaces 6, 16755–16762 (2014). 159. Stehle, Y. et al. Synthesis of Hexagonal Boron Nitride Monolayer: Control of Nucleation and Crystal Morphology. Chem. Mater. 27, 8041–8047 (2015). 133

160. Orofeo, C. M., Suzuki, S., Kageshima, H. & Hibino, H. Growth and low-energy electron microscopy characterization of monolayer hexagonal boron nitride on epitaxial cobalt. Nano Res. 6, 335–347 (2013). 161. Sutter, P., Lahiri, J., Zahl, P., Wang, B. & Sutter, E. Scalable synthesis of uniform few-layer hexagonal boron nitride dielectric films. Nano Lett. 13, 276– 281 (2013). 162. Sutter, P., Lahiri, J., Albrecht, P. & Sutter, E. Chemical Vapor Deposition and Etching of High-Quality Monolayer Hexagonal Boron Nitride Films. ACS Nano 5, 7303–7309 (2011). 163. Lu, J. et al. Step flow versus mosaic film growth in hexagonal boron nitride. J. Am. Chem. Soc. 135, 2368–73 (2013). 164. Orlando, F. & Larciprete, R. Epitaxial growth of hexagonal boron nitride on Ir (111). J. Phys. Chem. C 116, 157–164 (2011). 165. Caneva, S. et al. Nucleation Control for Large, Single Crystalline Domains of Monolayer Hexagonal Boron Nitride via Si-Doped Fe Catalysts. Nano Lett. 15, 1867–1875 (2015). 166. Vinogradov, N. A. et al. One-Dimensional Corrugation of the h-BN Monolayer on Fe(110). Langmuir 28, 1775–1781 (2012). 167. Kim, S. M. et al. Synthesis of large-area multilayer hexagonal boron nitride for high material performance. Nat. Commun. 6, 8662 (2015). 168. Caneva, S. et al. Controlling Catalyst Bulk Reservoir Effects for Monolayer Hexagonal Boron Nitride CVD. Nano Lett. 16, 1250–1261 (2016). 169. Corso, M. Boron Nitride Nanomesh. Science 303, 217–220 (2004). 170. Müller, F., Hüfner, S. & Sachdev, H. One-dimensional structure of boron nitride on chromium (110)–a study of the growth of boron nitride by chemical vapour deposition of borazine. Surf. Sci. 602, 3467–3476 (2008). 171. Lu, G., Wu, T., Yuan, Q. & Wang, H. Synthesis of large single-crystal hexagonal boron nitride grains on Cu–Ni alloy. Nat. Commun. 6, 6160 (2015). 172. Morscher, M., Corso, M., Greber, T. & Osterwalder, J. Formation of single layer h-BN on Pd(111). Surf. Sci. 600, 3280–3284 (2006). 173. Preobrajenski, a. B., Vinogradov, a. S. & Mårtensson, N. Monolayer of h-BN chemisorbed on Cu(111) and Ni(111): The role of the transition metal 3d states. Surf. Sci. 582, 21–30 (2005). 174. Laskowski, R., Blaha, P. & Schwarz, K. Bonding of hexagonal BN to transition metal surfaces: An ab initio density-functional theory study. Phys. Rev. B 78, 045409 (2008). 175. Bokdam, M., Brocks, G., Katsnelson, M. I. & Kelly, P. J. Schottky barriers at hexagonal boron nitride/metal interfaces: A first-principles study. Phys. Rev. B 90, 085415 (2014). 176. Grad, G., Blaha, P., Schwarz, K., Auwärter, W. & Greber, T. Density functional theory investigation of the geometric and spintronic structure of h-BN/Ni(111) in view of photoemission and STM experiments. Phys. Rev. B 68, 1–7 (2003). 177. Liu, Y., Bhowmick, S. & Yakobson, B. I. BN White Graphene with ‘Colorful’ Edges: The Energies and Morphology. Nano Lett. 11, 3113–3116 (2011). 134

178. Zhang, Z., Liu, Y., Yang, Y. & Yakobson, B. I. Growth Mechanism and Morphology of Hexagonal Boron Nitride. Nano Lett. 16, 1398–1403 (2016). 179. Chang, C.-K. et al. Band Gap Engineering of Chemical Vapor Deposited Graphene by in Situ BN Doping. ACS Nano 7, 1333–1341 (2013). 180. Park, J.-H. et al. Large-area monolayer hexagonal boron nitride on pt foil. ACS Nano 8, 8520–8528 (2014). 181. Tay, R. Y. et al. Growth of Large Single-Crystalline Two-Dimensional Boron Nitride Hexagons on Electropolished Copper. Nano Lett. 14, 839–846 (2014). 182. Wu, W. et al. Growth of Single Crystal Graphene Arrays by Locally Controlling Nucleation on Polycrystalline Cu Using Chemical Vapor Deposition. Adv. Mater. 23, 4898–4903 (2011). 183. Liu, Z. et al. Direct Growth of Graphene/Hexagonal Boron Nitride Stacked Layers. Nano Lett. 11, 2032–2037 (2011). 184. Lewis, A. M., Derby, B. & Kinloch, I. A. Influence of Gas Phase Equilibria on the Chemical Vapor Deposition of Graphene. 3104–3117 (2013). 185. Yan, Z. et al. Toward the Synthesis of Wafer-Scale Single-Crystal Graphene on Copper Foils. ACS Nano 6, 9110–9117 (2012). 186. Edwards, R. S. & Coleman, K. S. Graphene film growth on polycrystalline metals. Acc. Chem. Res. 46, 23–30 (2013). 187. Paths, A. E. & Nucleation, G. Activation Energy Paths for Graphene Nucleation and Growth on Cu. ACS Nano 6, 3614–3623 (2012). 188. Yan, Z. et al. Toward the Synthesis of Wafer-Scale Single-Crystal Graphene on Copper Foils. ACS Nano 6, 9110–9117 (2012). 189. Paffett, M. T., Simonson, R. J., Papin, P. & Paine, R. T. Borazine adsorption and decomposition at Pt(111) and Ru(001) surfaces. Surf. Sci. 232, 286–296 (1990). 190. Kidambi, P., Blume, R. & Kling, J. In situ observations during chemical vapor deposition of hexagonal boron nitride on polycrystalline copper. Chem. Mater. 26, 6380–6392 (2014). 191. T.B. Massalski, H. O. Binary Alloy Phase Diagrams. (ASM International: Materials Park, 1990). 192. J.A. Venables. Introduction to surface and thin film processes. (Cambridge University Press, 2000).

193. Okada, M. et al. Direct Chemical Vapor Deposition Growth of WS2 Atomic Layers on Hexagonal Boron Nitride. ACS Nano 8, 8273–8277 (2014). 194. Suk, J. W. et al. Transfer of CVD-Grown Monolayer Graphene onto Arbitrary Substrates. ACS Nano 5, 6916–6924 (2011). 195. Kang, J., Shin, D., Bae, S. & Hong, B. H. Graphene transfer: key for applications. Nanoscale 4, 5527–5537 (2012). 196. Regan, W., Alem, N., Alemán, B. & Geng, B. A direct transfer of layer-area graphene. Appl. Phys. Lett. 96, 10–13 (2010). 197. Lee, Y. et al. Wafer-Scale Synthesis and Transfer of Graphene Films. Nano Lett. 10, 490–493 (2010). 135

198. Bae, S. et al. Roll-to-roll production of 30-inch graphene films for transparent electrodes. 5, 1–5 (2010). 199. Wang, D.-Y. et al. Clean-Lifting Transfer of Large-area Residual-Free Graphene Films. Adv. Mater. 25, 4521–4526 (2013). 200. Amet, F. et al. Tunneling spectroscopy of graphene-boron-nitride heterostructures. Phys. Rev. B 85, 073405 (2012). 201. Song, Y. et al. Triggering the atomic layers control of hexagonal boron nitride films. Appl. Surf. Sci. 313, 647–653 (2014). 202. Sevak Singh, R. et al. Band gap effects of hexagonal boron nitride using oxygen plasma. Appl. Phys. Lett. 104, 163101 (2014). 203. Wang, M. et al. A Platform for Large-Scale Graphene Electronics - CVD Growth of Single-Layer Graphene on CVD-Grown Hexagonal Boron Nitride. Adv. Mater. 25, 2746–2752 (2013). 204. Garcia, A. G. F. et al. Correction to Effective Cleaning of Hexagonal Boron Nitride for Graphene Devices. Nano Lett. 13, 2314–2314 (2013). 205. Huang, P. Y. et al. Grains and grain boundaries in single-layer graphene atomic patchwork quilts. Nature 469, 389–392 (2011). 206. Gibb, A., Alem, N. & Zettl, A. Low pressure chemical vapor deposition synthesis of hexagonal boron nitride on polycrystalline metal foils. Phys. Status Solidi 250, 2727–2731 (2013). 207. Mattox, D. M. in Handbook of Physical Vapor Deposition (PVD) Processing 1–24 (Elsevier, 2010). doi:10.1016/B978-0-8155-2037-5.00001-0 208. Sajjad, M., Ahmadi, M., Guinel, M. J.-F., Lin, Y. & Feng, P. Large scale synthesis of single-crystal and polycrystalline boron nitride nanosheets. J. Mater. Sci. 48, 2543–2549 (2012). 209. Nakhaie, S. et al. Synthesis of atomically thin hexagonal boron nitride films on nickel foils by molecular beam epitaxy. Appl. Phys. Lett. 106, 213108 (2015). 210. Han, W.-Q., Yu, H.-G. & Liu, Z. Convert graphene sheets to boron nitride and boron nitride–carbon sheets via a carbon-substitution reaction. Appl. Phys. Lett. 98, 203112 (2011). 211. Lin, T.-W. et al. Converting Graphene Oxide Monolayers into Boron Carbonitride Nanosheets by Substitutional Doping. Small 8, 1384–1391 (2012). 212. Gong, Y. et al. Direct chemical conversion of graphene to boron- and nitrogen- and carbon-containing atomic layers. Nat. Commun. 5, 3193 (2014). 213. Han, W., Bando, Y., Kurashima, K. & Sato, T. Synthesis of boron nitride nanotubes from carbon nanotubes by a substitution reaction. Appl. Phys. Lett. 73, 3085 (1998). 214. Kim, G. et al. Catalytic Conversion of Hexagonal Boron Nitride to Graphene for In-Plane Heterostructures. Nano Lett. 15, 4769–4775 (2015). 215. Zeng, H. et al. ‘White Graphenes’: Boron Nitride Nanoribbons via Boron Nitride Nanotube Unwrapping. Nano Lett. 10, 5049–5055 (2010). 216. Nag, A. et al. Graphene Analogues of BN: Novel Synthesis and Properties. ACS Nano 4, 1539–1544 (2010). 136

217. Wang, L., Sun, C., Xu, L. & Qian, Y. Convenient synthesis and applications of gram scale boron nitride nanosheets. Catal. Sci. Technol. 1, 1119 (2011). 218. Zhang, S. et al. Ultrathin BN nanosheets with zigzag edge: one-step chemical synthesis, applications in wastewater treatment and preparation of highly thermal-conductive BN–polymer composites. J. Mater. Chem. A 1, 5105 (2013). 219. Gorbachev, R. V et al. Hunting for Monolayer Boron Nitride: Optical and Raman Signatures. Small 7, 465–468 (2011). 220. Golla, D. et al. Optical thickness determination of hexagonal boron nitride flakes. Appl. Phys. Lett. 102, 2011–2014 (2013). 221. Wu, Q. et al. Single Crystalline Film of Hexagonal Boron Nitride Atomic Monolayer by Controlling Nucleation Seeds and Domains. Sci. Rep. 5, 16159 (2015). 222. Shehzad, M. A. et al. Nematic Liquid Crystal on a Two Dimensional Hexagonal Lattice and its Application. Sci Rep 5, 13331 (2015). 223. Kim, D. W., Kim, Y. H., Jeong, H. S. & Jung, H.-T. Direct visualization of large-area graphene domains and boundaries by optical birefringency. Nat. Nanotechnol. 7, 29–34 (2012). 224. Son, J.-H. et al. Detection of graphene domains and defects using liquid crystals. Nat. Commun. 5, 3484 (2014). 225. Jia, C., Jiang, J., Gan, L. & Guo, X. Direct optical characterization of graphene growth and domains on growth substrates. Sci. Rep. 2, 707 (2012). 226. Duong, D. L. et al. Probing graphene grain boundaries with optical microscopy. Nature 490, 235–239 (2012). 227. T. Kuzuba, K. Era, T. I. and T. S. A low frequency raman-active vibration of hexagonal boron nitride. Solid State Commun. 25, 863–865 (1978). 228. Hoffman, D. M., Doll, G. L. & Eklund, P. C. Optical properties of pyrolytic boron nitride in the energy range 0.05-10 eV. Phys. Rev. B 30, 6051–6056 (1984). 229. Nemanich, R. J., Solin, S. A. & Martin, R. M. Light scattering study of boron nitride microcrystals. Phys. Rev. B 23, 6348–6356 (1981). 230. Geick, R., Perry, C. & Rupprecht, G. Normal Modes in Hexagonal Boron Nitride. Phys. Rev. 146, 543–547 (1966). 231. Ferrari, A. C. & Basko, D. M. Raman spectroscopy as a versatile tool for studying the properties of graphene. Nat. Nanotechnol. 8, 235–246 (2013). 232. Gao, G. et al. Artificially Stacked Atomic Layers: Toward New van der Waals Solids. Nano Lett. 12, 3518–3525 (2012). 233. Hite, J. K., Robinson, Z. R., Eddy, C. R. & Feigelson, B. N. Electron Backscatter Diffraction Study of Hexagonal Boron Nitride Growth on Cu Single-Crystal Substrates. ACS Appl. Mater. Interfaces 7, 15200–15205 (2015). 234. Cho, H. et al. Growth kinetics of white graphene (h-BN) on a planarised Ni foil surface. Sci. Rep. 5, 11985 (2015). 235. Dias, J., Kidambi, P. R., Hofmann, S. & Ducati, C. EBSD study of substrate- 137

mediated growth of hexagonal boron nitride. J. Phys. Conf. Ser. 522, 012070 (2014). 236. Lee, G.-H. et al. High-Strength Chemical-Vapor-Deposited Graphene and Grain Boundaries. Science 340, 1073–1076 (2013). 237. Robertson, A. W. et al. Structural distortions in few-layer graphene creases. ACS Nano 5, 9984–9991 (2011). 238. Krivanek, O. L. et al. Atom-by-atom structural and chemical analysis by annular dark-field electron microscopy. Nature 464, 571–574 (2010). 239. Hartel, P., Rose, H. & Dinges, C. Conditions and reasons for incoherent imaging in STEM. Ultramicroscopy 63, 93–114 (1996). 240. Odlyzko, M. L. & Mkhoyan, K. A. Identifying Hexagonal Boron Nitride Monolayers by Transmission Electron Microscopy. Microsc. Microanal. 18, 558–567 (2012). 241. Ahn, C. C. Transmission Electron Energy Loss Spectrometry in Materials Science and The EELS Atlas. (Wiley-VCH Verlag GmbH & Co. KGaA, 2004). doi:10.1002/3527605495 242. Cretu, O. et al. Structure and Local Chemical Properties of Boron-Terminated Tetravacancies in Hexagonal Boron Nitride. Phys. Rev. Lett. 114, 075502 (2015). 243. Nemes-Incze, P., Osváth, Z., Kamarás, K. & Biró, L. P. Anomalies in thickness measurements of graphene and few layer graphite crystals by tapping mode atomic force microscopy. Carbon N. Y. 46, 1435–1442 (2008). 244. Zitzler, L., Herminghaus, S. & Mugele, F. Capillary forces in tapping mode atomic force microscopy. Phys. Rev. B 66, 155436 (2002). 245. Li, L. H., Xing, T., Chen, Y. & Jones, R. Boron Nitride Nanosheets for Metal Protection. Adv. Mater. Interfaces 1, 1300132 (2014). 246. Xue, Y. et al. Excellent electrical conductivity of the exfoliated and fluorinated hexagonal boron nitride nanosheets. Nanoscale Res. Lett. 8, 49 (2013). 247. Suzuki, S., Pallares, R. M. & Hibino, H. Growth of atomically thin hexagonal boron nitride films by diffusion through a metal film and precipitation. J. Phys. D. Appl. Phys. 45, 385304 (2012). 248. Lei, W., Portehault, D., Dimova, R. & Antonietti, M. Boron Carbon Nitride Nanostructures from Salt Melts: Tunable Water-Soluble Phosphors. J. Am. Chem. Soc. 133, 7121–7127 (2011). 249. Ingo, G. M. & Padeletti, G. X-ray photoelectron spectroscopy and secondary- ion mass spectrometry of boron nitride thin films on austenitic stainless steel. Thin Solid Films 228, 276–279 (1993). 250. Westheimer, G. Optimal Magnification in Visual Microscopy*†. J. Opt. Soc. Am. 62, 1502 (1972). 251. Tauc, J., Grigorovici, R. & Vancu, A. Optical Properties and Electronic Structure of Amorphous Germanium. Phys. status solidi 15, 627–637 (1966). 252. Stern, F. Elementary Theory of the Optical Properties of Solids. Solid State Phys. 15, 299–408 (1963).

138

253. Márcia, L., Ansaloni, S., Martins, E. & Sousa, B. De. Boron Nitride Nanostructured: Synthesis, Characterization and Potential Use in Cosmetics. Mater. Sci. Appl. 2013, 22–28 (2013). 254. Zhang, Y. et al. Hexagonal Boron Nitride Cover on Pt(111): A New Route to Tune Molecule–Metal Interaction and Metal-Catalyzed Reactions. Nano Lett. 15, 3616–3623 (2015). 255. Sutter, P. & Sutter, E. Thickness determination of few-layer hexagonal boron nitride films by scanning electron microscopy and Auger electron spectroscopy. APL Mater. 092502, 1–7 (2014). 256. Preobrajenski, A. B. et al. Adsorption-induced gap states of h-BN on metal surfaces. Phys. Rev. B 77, 085421 (2008). 257. Yang, Y., Fu, Q., Wei, M., Bluhm, H. & Bao, X. Stability of BN/metal interfaces in gaseous atmosphere. Nano Res. 8, 227–237 (2014). 258. Joshi, S. et al. Boron Nitride on Cu(111): An Electronically Corrugated Monolayer. Nano Lett. 12, 5821–5828 (2012). 259. Li, X. et al. Large-area synthesis of high-quality and uniform graphene films on copper foils. Science 324, 1312–1314 (2009). 260. Wolf, G., Baumann, J., Baitalow, F. & Hoffmann, F. P. Calorimetric process monitoring of thermal decomposition. Thermochim. Acta 343, 19–25 (2000). 261. Baitalow, F., Baumann, J., Wolf, G., Jaenicke-Rößler, K. & Leitner, G. Thermal decomposition of B–N–H compounds investigated by using combined thermoanalytical methods. Thermochim. Acta 391, 159–168 (2002). 262. Gomez-Aleixandre, C., Diaz, D., Orgaz, F. & Albella, J. M. Reaction of diborane and ammonia gas mixtures in a chemical vapor deposition hot-wall reactor. J. Phys. Chem. 97, 11043–11046 (1993). 263. Baril, G. et al. An Impedance Investigation of the Mechanism of Pure Magnesium Corrosion in Sodium Sulfate Solutions. J. Electrochem. Soc. 154, C108–C113 (2007). 264. Eckhard, K. & Schuhmann, W. Alternating current techniques in scanning electrochemical microscopy (AC-SECM). Analyst 133, 1486 (2008). 265. Izquierdo, J., González, S. & Souto, R. M. Application of AC-SECM in Corrosion Science : Local Visualization of Heterogeneous Chemical Activity in AA2024 Surfaces. Int. J. Electrochem. Sci. 7, 11377–11388 (2012). 266. C.T. Koch. Determination of core structure periodicity and point defect density along dislocations. (Arizona State University, 2002). 267. Wu, Z. & Cohen, R. E. More accurate generalized gradient approximation for solids. Phys. Rev. B 73, 235116 (2006). 268. Soler, M. et al. The SIESTA method for ab initio order- N materials. J. Phys. D. Appl. Phys. 14, 2745–2779 (2002). 269. Lyalin, A., Nakayama, A., Uosaki, K. & Taketsugu, T. Adsorption and Catalytic Activation of the Molecular Oxygen on the Metal Supported h-BN. Top. Catal. 57, 1032–1041 (2014). 270. Lyalin, A., Nakayama, A., Uosaki, K. & Taketsugu, T. Functionalization of Monolayer h-BN by a Metal Support for the Oxygen Reduction Reaction. J. 139

Phys. Chem. C 117, 21359–21370 (2013). 271. Uosaki, K. et al. Boron Nitride Nanosheet on Gold as an Electrocatalyst for Oxygen Reduction Reaction: Theoretical Suggestion and Experimental Proof. J. Am. Chem. Soc. 136, 6542–6545 (2014). 272. Wyckoff, R. W. G. Crystal Structures. (John Wiley & Sons, 1963). 273. Preobrajenski, A. B., Cunniffe, J. P., Taketsugu, T., Lyalin, A. & Ma, N. From Graphene Nanoribbons on Cu(111) to Nanographene on Cu(110): Critical Role of Substrate Structure in the Bottom-Up Fabrication Strategy. ACS Nano 9, 8997–9011 (2015). 274. Yuzuriha, T. H. & Hess, D. W. Structural and optical properties of plasma- deposited boron nitride films. Thin Solid Films 140, 199–207 (1986). 275. Chen, W., Li, Y., Yu, G. & Li, C. Hydrogenation: a simple approach to realize semiconductor− half-metal− metal transition in boron nitride nanoribbons. J. Am. Chem. Soc. 132, 1699–1705 (2010). 276. Tang, Q., Bao, J., Li, Y., Zhou, Z. & Chen, Z. Tuning band gaps of BN nanosheets and nanoribbons via interfacial dihalogen bonding and external electric field. Nanoscale 6, 8624–8634 (2014). 277. Liu, L. et al. Graphene Oxidation: Thickness-Dependent Etching and Strong Chemical Doping. Nano Lett. 8, 1965–1970 (2008).

278. Lee, G. et al. Flexible and Transparent WS2 Field- Effect Transistors on Hexagonal Boron-Graphene Heterostructures. ACS Nano 7, 7931–7936 (2013). 279. Geng, D. et al. Uniform hexagonal graphene flakes and films grown on liquid copper surface. Proc. Natl. Acad. Sci. 109, 7992–7996 (2012). 280. Wu, Y., Fan, Y., Speller, S. & Creeth, G. Large single crystals of graphene on melted copper using chemical vapor deposition. ACS Nano 6, 5010–5017 (2012). 281. Pakdel, A., Zhi, C., Bando, Y., Nakayama, T. & Golberg, D. Boron Nitride Nanosheet Coatings with Controllable Water Repellency. ACS Nano 5, 6507– 6515 (2011). 282. Kelly, B. T. The anisotropic thermal expansion of boron nitride. Philos. Mag. 32, 859–867 (1975). 283. Zobelli, a., Gloter, a., Ewels, C., Seifert, G. & Colliex, C. Electron knock-on cross section of carbon and boron nitride nanotubes. Phys. Rev. B 75, 245402 (2007). 284. McDougall, N. L., Nicholls, R. J., Partridge, J. G. & McCulloch, D. G. The Near Edge Structure of Hexagonal Boron Nitride. Microsc. Microanal. 20, 1053– 1059 (2014). 285. Hao, Y. et al. The Role of Surface Oxygen in the Growth of Large Single- Crystal Graphene on Copper. Science 342, 720–723 (2013). 286. Mohsin, A., Liu, L., Liu, P., Deng, W. & Ivanov, I. Synthesis of millimeter-size hexagon-shaped graphene single crystals on resolidified copper. ACS Nano 7, 8924 – 8931 (2013). 287. Li, Q., Chou, H., Zhong, J., Liu, J. & Dolocan, A. Growth of adlayer graphene on Cu studied by carbon isotope labeling. Nano Lett. 13, 486–490 (2013). 140

288. Wu, T., Ding, G., Shen, H. & Wang, H. Triggering the Continuous Growth of Graphene Toward Millimeter-Sized Grains. Adv. Funct. Mater. 23, 198–203 (2013). 289. Niaura, G. Surface-enhanced Raman spectroscopic observation of two kinds of adsorbed OH− ions at copper electrode. Electrochim. Acta 45, 3507–3519 (2000). 290. Chan, H. Y. H., Takoudis, C. G. & Weaver, M. J. Oxide Film Formation and Oxygen Adsorption on Copper in Aqueous Media As Probed by Surface- Enhanced Raman Spectroscopy. J. Phys. Chem. B 103, 357–365 (1999). 291. Hawaldar, R. et al. Large-area high-throughput synthesis of monolayer graphene sheet by Hot Filament Thermal Chemical Vapor Deposition. Sci. Rep. 2, 682 (2012). 292. Wang, L. et al. Growth and Etching of Monolayer Hexagonal Boron Nitride. Adv. Mater. 27, 4858–4864 (2015). 293. Shmeliov, A. et al. Impurity induced non-bulk stacking in chemically exfoliated h-BN nanosheets. Nanoscale 5, 2290–2294 (2013). 294. Marom, N. et al. Stacking and Registry Effects in Layered Materials: The Case of Hexagonal Boron Nitride. Phys. Rev. Lett. 105, 046801 (2010). 295. Pennycook, S. J. Z-contrast stem for materials science. Ultramicroscopy 30, 58– 69 (1989). 296. Hiura, H., Miyazaki, H. & Tsukagoshi, K. Determination of the Number of Graphene Layers: Discrete Distribution of the Secondary Electron Intensity Stemming from Individual Graphene Layers. Appl. Phys. Express 3, 095101 (2010). 297. Heldebrant, D. J., Karkamkar, A., Linehan, J. C. & Autrey, T. Synthesis of ammonia borane for hydrogen storage applications. Energy Environ. Sci. 1, 156–160 (2008). 298. Ramachandran, P. V. & Gagare, P. D. Preparation of Ammonia Borane in High Yield and Purity, Methanolysis, and Regeneration. Inorg. Chem. 46, 7810–7817 (2007). 299. Ramachandran, P. V., Mistry, H., Kulkarni, A. S. & Gagare, P. D. Ammonia- mediated, large-scale synthesis of ammonia borane. Dalt. Trans. 43, 16580– 16583 (2014). 300. Ci, L. et al. Atomic layers of hybridized boron nitride and graphene domains. Nat. Mater. 9, 430–435 (2010). 301. Sharma, S., Kalita, G. & Vishwakarma, R. Opening of triangular hole in triangular-shaped chemical vapor deposited hexagonal boron nitride crystal. Sci. Rep. 5, 10426 (2015). 302. Wood, G. E. et al. van der Waals epitaxy of monolayer hexagonal boron nitride on copper foil: growth, crystallography and electronic band structure. 2D Mater. 2, 025003 (2015). 303. Redondo, M. I. & Breslin, C. B. Polypyrrole electrodeposited on copper from an aqueous phosphate solution: Corrosion protection properties. Corros. Sci. 49, 1765–1776 (2007).

141

304. Stratmann, M., Feser, R. & Leng, A. Corrosion protection by organic films. Electrochim. Acta 39, 1207–1214 (1994). 305. Segarra, M. et al. Copper and CuNi Alloys Substrates for HTS Coated Conductor Applications Protected from Oxidation. Mater. Sci. Forum 426-432, 3511–3516 (2003). 306. Bunch, J. S. et al. Impermeable Atomic Membranes from Graphene Sheets. Nano Lett. 8, 2458–2462 (2008). 307. Prasai, D., Tuberquia, J. C., Harl, R. R., Jennings, G. K. & Bolotin, K. I. Graphene: Corrosion-Inhibiting Coating. ACS Nano 6, 1102–1108 (2012). 308. Huh, J.-H. et al. Enhancement of seawater corrosion resistance in copper using acetone-derived graphene coating. Nanoscale 6, 4379–4386 (2014). 309. Singh Raman, R. K. et al. Protecting copper from electrochemical degradation by graphene coating. Carbon N. Y. 50, 4040–4045 (2012). 310. Kang, D. et al. Oxidation Resistance of Iron and Copper Foils Coated with Reduced Graphene Oxide Multilayers. ACS Nano 6, 7763–7769 (2012). 311. Hsieh, Y.-P. et al. Complete Corrosion Inhibition through Graphene Defect Passivation. ACS Nano 8, 443–448 (2014). 312. Chen, S. et al. Oxidation Resistance of Graphene-Coated Cu and Cu/Ni Alloy. ACS Nano 5, 1321–1327 (2011). 313. Zhou, F., Li, Z., Shenoy, G. J., Li, L. & Liu, H. Enhanced Room-Temperature Corrosion of Copper in the Presence of Graphene. ACS Nano 7, 6939–6947 (2013). 314. Schriver, M. et al. Graphene as a Long-Term Metal Oxidation Barrier: Worse Than Nothing. ACS Nano 7, 5763–5768 (2013). 315. Kubota, Y., Watanabe, K., Tsuda, O. & Taniguchi, T. Deep Ultraviolet Light- Emitting Hexagonal Boron Nitride Synthesized at Atmospheric Pressure. Science 317, 932–934 (2007).

316. Rashad, M., Rüsing, M., Berth, G., Lischka, K. & Pawlis, A. CuO and Co3O4 Nanoparticles: Synthesis, Characterizations, and Raman Spectroscopy. J. Nanomater. 2013, 1–6 (2013). 317. Biesinger, M. C., Lau, L. W. M., Gerson, A. R. & Smart, R. S. C. Resolving surface chemical states in XPS analysis of first row transition metals, oxides and hydroxides: Sc, Ti, V, Cu and Zn. Appl. Surf. Sci. 257, 887–898 (2010). 318. Carrasco, J., Hodgson, A. & Michaelides, A. A molecular perspective of water at metal interfaces. Nat. Mater. 11, 667–674 (2012). 319. Gokhale, A. A., Dumesic, J. A. & Mavrikakis, M. On the Mechanism of Low- Temperature Water Gas Shift Reaction on Copper. J. Am. Chem. Soc. 130, 1402–1414 (2008). 320. Michaelides, A., Ranea, V. A., de Andres, P. L. & King, D. A. General Model for Water Monomer Adsorption on Close-Packed Transition and Noble Metal Surfaces. Phys. Rev. Lett. 90, 216102 (2003). 321. Phatak, A. A., Delgass, W. N., Ribeiro, F. H. & Schneider, W. F. Density Functional Theory Comparison of Water Dissociation Steps on Cu, Au, Ni, Pd, and Pt. J. Phys. Chem. C 113, 7269–7276 (2009). 142

322. Wang et al. Surface Structure Sensitivity of the Water−Gas Shift Reaction on Cu( h kl ) Surfaces: A Theoretical Study. J. Phys. Chem. B 107, 557–562 (2003). 323. Gao, M., Lyalin, A. & Taketsugu, T. Oxygen activation and dissociation on h- BN supported Au atoms. Int. J. Quantum Chem. 113, 443–452 (2013). 324. Gómez Díaz, J. et al. Hexagonal boron nitride on transition metal surfaces. Theor. Chem. Acc. 132, 1350 (2013). 325. López-Moreno, S. & Romero, A. H. Atomic and molecular oxygen adsorbed on (111) transition metal surfaces: Cu and Ni. J. Chem. Phys. 142, 154702 (2015). 326. Soon, A., Todorova, M., Delley, B. & Stampfl, C. Oxygen adsorption and stability of surface oxides on Cu(111) : A first-principles investigation. Phys. Rev. B 73, 165424 (2006). 327. Preobrajenski, A. B. et al. Impact of Oxygen Coadsorption on Intercalation of Cobalt under the h-BN Nanomesh. Nano Lett. 9, 2780–2787 (2009). 328. Grånäs, E. et al. Oxygen Intercalation under Graphene on Ir(111): Energetics, Kinetics, and the Role of Graphene Edges. ACS Nano 6, 9951–9963 (2012). 329. Sutter, P., Sadowski, J. T. & Sutter, E. A. Chemistry under Cover: Tuning Metal−Graphene Interaction by Reactive Intercalation. J. Am. Chem. Soc. 132, 8175–8179 (2010). 330. Wang, Y. & Xu, Z. Water Intercalation for Seamless, Electrically Insulating, and Thermally Transparent Interfaces. ACS Appl. Mater. Interfaces 8, 1970– 1976 (2016). 331. Lee, K. H. et al. Nanocrystalline-Graphene-Tailored Hexagonal Boron Nitride Thin Films. Angew. Chemie Int. Ed. 53, 11493–11497 (2014). 332. Xiao, F. et al. Edge-Hydroxylated Boron Nitride Nanosheets as an Effective Additive to Improve the Thermal Response of Hydrogels. Adv. Mater. 27, 7196–7203 (2015). 333. Weatherup, R. S. et al. Long-Term Passivation of Strongly Interacting Metals with Single-Layer Graphene. J. Am. Chem. Soc. 137, 14358–14366 (2015). 334. Miyamoto, H., Harada, K., Mimaki, T., Vinogradov, A. & Hashimoto, S. Corrosion of ultra-fine grained copper fabricated by equal-channel angular pressing. Corros. Sci. 50, 1215–1220 (2008). 335. El Warraky, a., El Shayeb, H. a. & Sherif, E. M. Pitting corrosion of copper in chloride solutions. Anti-Corrosion Methods Mater. 51, 52–61 (2004). 336. Husain, E. et al. Marine Corrosion Protective Coatings of Hexagonal Boron Nitride Thin Films on Stainless Steel. ACS Appl. Mater. Interfaces 5, 4129– 4135 (2013). 337. Bertolotti, G. et al. Micro-Raman study of copper hydroxychlorides and other corrosion products of bronze samples mimicking archaeological coins. Anal. Bioanal. Chem. 402, 1451–1457 (2012).

143