Energy Dispersion Model Using Tight Binding Theory

Rochester Institute of Technology RIT Scholar Works

Theses

6-2016

Energy Dispersion Model using Tight Binding Theory

Follow this and additional works at: https://scholarworks.rit.edu/theses

Recommended Citation Vajpey, Divya S., "Energy Dispersion Model using Tight Binding Theory" (2016). Thesis. Rochester Institute of Technology. Accessed from

This Thesis is brought to you for free and open access by RIT Scholar Works. It has been accepted for inclusion in Theses by an authorized administrator of RIT Scholar Works. For more information, please contact [email protected]. Energy Dispersion Model using Tight Binding Theory

Divya S. Vajpey Energy Dispersion Model using Tight Binding Theory Divya S. Vajpey

A Thesis Submitted in Partial Fulﬁllment of the Requirements for the Degree of Master of Science in Electrical Engineering

Approved by:

Dr. Sean Rommel Advisor Date Professor

Dr. Santosh Kurinec Date Professor

Dr. Jing Zhang Date Kate Gleason Assistant Professor

Dr. Sohail Dianat Date Department Head

Department of Electrical and Microelectronic Engineering

Rochester Institute of Technology Rochester, New York

June, 2016 Acknowledgments

I would like to thank my advisor, Dr. Rommel for introducing me to this topic and for his constant support and guidance. On the innumerable occasions that I’ve been lost during this research, Dr. Rommel has always been there to point me in the right direction. I thank him especially for always being there to provide good sound advice - personally and professionally. I also thank my thesis committee - Dr. Kurinec and Dr. Zhang for their invaluable inputs and support. I would also like to thank Dr. James Moon for introducing me to device physics and inspiring my interest in this ﬁeld. A special thank you to Ms. Patricia Vicari and Department of Electrical Engineering for helping me to pursue my interest. Lastly, I would like to thank my family and all my friends for being a constant source of support and encouragement.

i Dedication

For my mother, sister and father who always encouraged me to push beyond my limits.

"There is a freedom waiting for you, In the breezes of the sky, And you ask, ’What if I fall?’ But darling... ’What if you ﬂy?’"

-Erin Hanson

ii Abstract

2-D Semiconductors are novel materials in the field of nano-electronics. Their unusual transport properties have led to an extensive research attention towards similar materials. Graphene has carriers that exhibit an effective "speed of light" (106 m/s) in the low energy range of <0.5 eV. These massless fermions exhibit various quantum electrodynamics(QED) in the low energy range that can be explored. It is therefore of interest to understand and simulate the nature of the bands and understand their interactions along the edges and corners of the first brillouin zone. Various forms of monolayer graphene exist - A Graphene nano-ribbon is a sheet of graphene of narrow width and high aspect ratio existing in 2 different edge orientations. This variation in the band gap of different morphologies is known to affect the On current characteristics for corresponding GNRFETs and other applications.

The first brillouin zone of monolayer MoS2 is hexagonal in shape and similar to that of graphene. However, in the case of transition metal dichalcogenides (TMDs), the effect of different layers stacked is the important factor affecting the energy bands.

MoS2 in bulk form is an indirect band gap semiconductor and in monolayer form is a direct band gap semiconductor. The main diﬀerence lies in the structure of the unit cell in either cases. TMDs contain a metal atom d-orbital and a chalcogen p-orbital interaction. In the case of graphene, the highest complexity orbital interaction is

between 2 pz orbitals. This research proposes to achieve a common energy dispersion model for diﬀerent hybridized structures, using tight binding theory. The goal is to achieve a suitable starting point to obtain practical electronic transport calculations for complex atomic structures. We begin with analyzing the electronic properties by obtaining the an- alytic solution of the wave function, from the Schrödinger equation. We obtain the energy dispersion relation by solving the Hamiltonian. Construction of the Hamil- tonian matrix is the most crucial step in this process and the matrix elements are

iii derived using nearest neighbor (NN) interactions. From thereon, methods to solve the Hamiltonian matrices mathematically and MATLAB codes to achieve them are discussed. In this research, we explore the electronic and quantum transport properties with the help of a common model that works across sp2 and sp3d hybridized atomic orbitals. We speciﬁcally deal with 2-D materials like Graphene Nano-Ribbons (GNR) and

Molybdenum Disulﬁde (MoS2). The width/layer-tunable band-gap of these materials is favorable to a variety of modern day applications. A variety of external factors are also known to vary the electronic properties of these materials, like - electric and magnetic ﬁelds, external pressure, temperature, strain, etc. Considering spin-orbit coupling, leads to valley physics and coupled spin in monolayers, makes it possible to control the spin and valley correspondingly. These factors are not accounted for in the current model and can build build upon as extensions.

The eﬀect of adding layers in Graphene and MoS2 is a avenue for future scope and research.

iv Contents

Abstract iii

Table of Contents v

List of Figures vii

List of Tables viii

List of Symbols ix

1 Introduction 1 1.1 0D, 1D, 2D, 3D Semiconductors ...... 1 1.2 Factors aﬀecting Energy Bands ...... 2 1.3 Methods of calculating Band Structure ...... 4 1.4 Organization of Document ...... 6

2 Atomic Hybridization & Hamiltonian Matrix 8 2.1 Eﬀect of Diﬀerent Atomic Structures ...... 8 2.1.1 Graphene ...... 11

2.1.2 Molybdenum Disulﬁde (MoS2)...... 14 2.2 Linear Combination of Atomic Orbitals (LCAO) ...... 16 2.3 Real and Reciprocal Space ...... 16 2.3.1 Unit cells - Real space ...... 17 2.3.2 Reciprocal Spaces ...... 19 2.3.3 Graphene Reciprocal Space ...... 21 2.3.4 Molybdenum Disulﬁde Reciprocal Space ...... 23 2.4 Schrödinger Equation ...... 24 2.5 Hamiltonian Matrices ...... 24 2.5.1 Basis of Orbitals ...... 25 2.5.2 Hamiltonian for one-dimensional atomic structure ...... 25 2.5.3 General Hamiltonian ...... 26 2.5.4 Hamiltonian for sp2 Hybridiation (Special Case : Graphene) . 28 3 2.5.5 Hamiltonian for sp d Hybridisation (Special Case : MoS2).. 30 2.6 Conclusion ...... 33

v CONTENTS

3 Tight Binding Theory and Solving for Energy Dispersion Relation 34 3.1 Introduction ...... 34 3.2 Introduction ...... 34 3.3 Tight Binding Parameters ...... 36 3.3.1 Graphene ...... 36

3.3.2 Tight Binding Parameters for MoS2 ...... 37 3.4 Solving the Hamiltonian ...... 38 3.4.1 Graphene ...... 38

3.4.2 MoS2 ...... 39

3.5 Energy Dispersion - Graphene and MoS2 ...... 40 3.5.1 Graphene ...... 40 3.5.2 Comparison to existing work - Graphene ...... 42

3.5.3 MoS2 ...... 43 3.6 Graphical User Interface ...... 45

4 Edge Eﬀect of Graphene & Layers Eﬀect on MoS2 48 4.1 Graphene ...... 48 4.1.1 A-GNR ...... 49 4.1.2 Z-GNR ...... 54

4.2 Monolayer/Bilayer eﬀect in MoS2 ...... 57

5 Summary & Future Work 59 5.1 Summary ...... 59 5.2 Future Work ...... 60

A MATLAB Code for Graphene 61 A.1 Case 1 - π bands ...... 61 A.2 Case 2 - π & σ bands ...... 65

B MATLAB Code for Armchair GNR 69

C MATLAB Code for Zigzag GNR 71

D Tight Binding simpliﬁcations and independent parameters for MoS2 74

E MATLAB Code for monolayer MoS2 78

vi List of Figures

2.1 sp orbitals ...... 10 2.2 sigma and pi bonds ...... 11 2.3 Monolayer of 2D Graphene ...... 12 2.4 Layered 2D Graphene ...... 13 2.5 Interatomic orbital interaction of Graphene ...... 13 2.6 MoS2 atom ...... 15 2.7 Real Space ...... 17 2.8 Bravais ...... 18 2.9 Reciprocal ...... 21 2.10 Reciprocal Space of Graphene ...... 22 2.11 1st Brillouin Zone of Graphene ...... 22 2.12 σ and π bands of Graphene ...... 23 2.13 3D 1st BZ of Graphene ...... 23 2.14 Nearest neighbor vectors of Graphene ...... 28

3.1 Vppπ =1eV ...... 40

3.2 Vppπ = 2.7eV ...... 41 3.3 Energy Band Dispersion with all TB Parameters ...... 41 3.4 3D E-k for Graphene ...... 42

3.5 Energy Dispersion for Monolayer MoS2 [1] ...... 44 3.6 2D and 3D representation of Graphene Energy Band Diagram . . . . 45 3.7 Armchair GNR GUI ...... 46 3.8 Zigzag GNR GUI ...... 46 3.9 Monolayer Molybdenum Disulﬁde GUI ...... 47

4.1 Armchair and Zigzag edged Graphene Nano-Ribbon ...... 49 4.2 Energy Dispersion for A-GNR with width n=6 ...... 52 4.3 Energy Dispersion for A-GNR with width n=7 ...... 52 4.4 Energy Dispersion for A-GNR with width n=8 ...... 53 4.5 Energy Dispersion for A-GNR with width n=22 ...... 53 4.6 Energy Dispersion for Z-GNR with width n=5 ...... 56 4.7 Energy Dispersion for Z-GNR with width n=16 ...... 56 4.8 Energy Dispersion for Z-GNR with width n=30 ...... 57

4.9 Energy Dispersion for Bilayer ((a) & (b)) and Bulk MoS2 [1] . . . . . 58

vii List of Tables

2.1 Types of hybridization [2] ...... 9 2.2 Seven types of lattice crystal systems [3] ...... 19

3.1 Tight Binding Parameters for Graphene [4] ...... 37

3.2 Tight Binding Parameters for MoS2 [1] [5] ...... 37

D.1 Independent Tight Binding Parameters for MoS2 [1] ...... 77

viii List of Symbols

Term Description Units/Value

E Energy Dispersion eV Ψ Wave Function U Potential Energy eV ~k Wave vector R~ Lattice vector Hˆ Hamiltonian Operator BZ Brillouin Zone BCC Body Centered Cubic FCC Face Centered Cubic HCC Hexagonal Centered Cubic GNR Graphene Nano-Ribbon A-GNR Armchair Graphene Nano-Ribbon Z-GNR Zigzag Graphene Nano-Ribbon LCAO Linear Combination of Atomic Orbitals ~ ~ ~ δ1, δ2, δ3 Nearest Neighbour Vectors

Vintra intra-atomic hybridization between orbitals |2 > & eV |3 >

Vσ between two |2 > states, anion and cation of same eV orbital

Vppπ between two 2pz states eV

Vppσ between two 2pz states eV

Vspσ between s and p orbitals eV

Vssσ between two s orbitals eV

σ between two |2 > states, 2 anions/cations of same orbital x tm,n 86 Tight Binding Parameters for MoS2 eV

E0, Ess Energy between orbitals eV a,b unit cell lattice vectors DFT Density Functional Theory LSDA Local Spin Density Approximation

ix Chapter 1 Introduction

Over the years, semiconductors have provided various stages to explore funda- mental questions of physics. As technology increased, the semiconductors and the heterostructure devices also evolved - as the traditional concepts reached their per- formance limits. Since their discovery in the early 20th century, semiconductors such as Si, Ge, GaAs and InP have come a long way and, other materials such as GaN, InAs, InN etc., have been explored to make better devices. It is believed that control over conductivity in any device is crucial to the evolution of new ideas and materials. Modiﬁcations to the structures of current devices, application of strain to materials used, alternative materials etc., were some of the several successful attempts made to improve device operation [6] [7]. This research is aiming to aid in the same bid - to explore new alternative materials that can possibly open further avenues for future scientists.

1.1 0D, 1D, 2D, 3D Semiconductors

Crystals are solid substances that consist of highly ordered, periodic arrangement of atoms known as lattices. Crystal structures produce diﬀerent material properties, depending on the periodic potential created across the solid. Traditionally, most semiconductors used in electronic and optoelectronic devices were of 3-D crystal form. The most common example is the silicon complementary metal-oxide-semiconductor.

1 CHAPTER 1. INTRODUCTION

A 3-D crystal, when constrained from one of the directions, forms a 2-D quantum well. When a long wire is drawn from the 3-D crystal, in such a way that, the length of the wire is very large and the diameter of the wire is small enough to cause quantum conﬁnement, it is known as a 2-D quantum wire. Two-dimensional crystal structures gained spotlight when it was possible to obtain atomically thin sheets of graphene. Other two-dimensional crystal structures include

molybdenum disulﬁde (MoS2), tungsten disulﬁde (WS2), tungsten diselenide (WSe2), etc [8]. These 2-D crystals are also known as transition metal dichalcogenides (TMDs).

They have the chemical formula of MX2, where ’M’ is the transition metal chemically bonded to 2 chalcogenide atoms ’X’. A perfect 2-D crystal does not have any dangling bonds. However, when it is formed from 3-D crystals, they have dangling bonds at the surface, which are passivated by dielectrics like hydrogen or lattice matched or strained heterostructures.A 2-D crytal when folded at its longitudinal ends, forms a one-dimensional structure known as a nanotube. When folded from all directions, the structure is known as a zero-dimensional buckyball. The perfect example for a 1-D structure is an atomic chain. When the atoms in the chain, connect to form an atomic ring, the structure is considered to be zero- dimensional. A solitary atom is the perfect example for a zero-dimensional structure.

1.2 Factors aﬀecting Energy Bands

To understand the current in a semiconductor, it is essential to know the ﬁxed and mobile charges present and the transport of mobile carriers through the material. Energy bands in a crystal provides us with this framework. Energy bands are a energy levels occupied by diﬀerent electrons of a single isolated atom. As per Pauli’s exclusion principle, each orbital forms discretized energy levels. As the atoms combine to form molecules, the energy bands represent energies for the molecular orbitals. The collection of multiple such molecules forms a solid, and the energy levels are close

2 CHAPTER 1. INTRODUCTION

enough to become continuum. Band gaps are formed as they are ranges of energy that are not occupied by any orbital. There are many internal and external factors that cause the nature of a material and band gaps to vary:

• Doping : Introduction of a foreign material in a crystalline semiconductor like Silicon or Germanium can dramatically change the conductivity of the semiconductor. Generally, it has been observed that elements from Group III like Boron (B), Aluminum (Al) or Gallium (Ga) are added to make the semiconductor p-type conducting by producing a electron (or hole) deﬁciency. Each impurity atom creates a hole. To make an n-type conductive semiconductor, Antimony (Sb), Arsenic (As) or Phosphorous (P) are added as impurities (dopants). Each impurity atom donates an electron.

• Strain Strain can be introduced locally or globally for any device. Local strain is observed due to stressed overlayers, embedded source/drain stressors or stressed

contacts. For example - a Silicon Nitride (Si3N4) or Silicon Carbide (SiC) ﬁlm is applied on source and drain regions to induce uniaxial stress. Global strain is generally observed due to lattice mismatch in the materials used. Examples include strained SiGe, strained Si on relaxed SiGe on insulator (SGOI), etc. Particularly enhanced eﬀects of strain are observed in 2-D semiconductors.

• Electric and Magnetic fields When properly applied, both, electric and magnetic fields are known to modify energy bands of the material/device under consideration. These fields give us the ability to modulate the transport properties of the device to best suit our needs.

3 CHAPTER 1. INTRODUCTION

• Eﬀect of width and layers in 2-D semiconductors These eﬀects are observed primarily in 2-D semiconductors. Depending on the width of the nano-ribbon, for example - Graphene Nano-Ribbon (GNR), it is known to exhibit both metallic and insulating properties. This is due to the

eﬀect of the pz or π orbitals in the z-direction of the carbon atom. The eﬀect of 0p0, 0d0 and higher orbitals makes the band gap tunable when there is more than a single layer of the 2-D semiconductor. This is observed in Graphene and

other materials like MoS2, WS2, etc,. This eﬀect is explained in further details in chapter 4.

1.3 Methods of calculating Band Structure

There are multiple methods/theories of calculating the band structure of a material. This research simulates ab-initio calculations for band structures of 2-D semiconductors using a combination of a few of the theories described below.

• Density Functional Theory (DFT) : This is one of the latest theories to understand the electronic structure of molecules and atoms. This theory though founded in the 1970s, was never utlized till date because of the number of flaws in it. Recently, a lot of modifications have been made and this theory is one of the widely accepted theories for nuclei, atoms, molecules, solids, quantum and classical fluids. The theory can be generalized to incorporate situations like spin, free energy, time-dependent states, bosons,etc. It operates on the principles of Local Density Approximation (LDA) and Generalized Gradient Approximation (GGA).

• Pseudopotential Theory : The pseudopotential theory, replaces the coulombic potential term for the core electrons in the schrödinger equation, with an eﬀective potential, known as

4 CHAPTER 1. INTRODUCTION

the pseudopotential. This pseudopotential includes the core electrons and the

nucleus of the atom. A cut-oﬀ radius rc needs to be established to separate the valence orbitals from the core orbitals. Generally, the "closed" shells or electron- full orbitals are considered to be core orbitals and the remaining shells as the valence orbitals. It follows the small-core approximation which assumes no signif- icant overlap between core and valence electrons and nonlinear core corrections can be performed for the cases when the interactions are non-negligible.

• Bloch Theorem : This theorem works on the principle of periodic potentials of the lattice, which can be replicated throughout the solid. The schrödinger equation has to be solved to obtain a suitable periodic potential.

• Perturbation Theory : This theory is a set of approximations performed to a complex system in terms of a simpler one. We begin with a smaller simpler system, who’s mathematical solution is known. Additional "perturbations" are added to this system and since they are small, the energy levels and associated eigen values are expressed as corrections and can be calculated using approximate methods. There are many variations to this theory - Time dependent/independent, degenerate/non- degenerate perturbation theory. Relevant theories are covered further on, in this text.

• Tight Binding Theory : In the tight-binding model, it is assumed that the crystal/ionic potential of the atom is strong. For this reason, the electron in the atomic orbital is tightly bound to its own atom and uninﬂuenced by external factors. Tight-binding theory expresses the electron wave function in terms of atomic orbital and utilizes the Bloch wave function heavily. This research uses tight binding theory

5 CHAPTER 1. INTRODUCTION

primarily for its ab-initio band structure calculations.

Detailed explanations of the pseudopotential, bloch and tight-binding theories are covered further ahead.

1.4 Organization of Document

This document begins with a basic information about semiconductors and their evolution over the years. The foray into 2-D semiconductors and their applications is briefly discussed. This research deals with 2-D semiconductors primarily. All the subsequent explanations made, would be with 2-D semiconductors as the reference, unless specified otherwise. The need to understand energy bands in the current day semiconductor market is explained. We begin with a brief background about the most effective theories to calculate the band structures and also, the theories on which this research is based. Chapter two covers the basic concepts that are needed to obtain the energy band structure for any material. It begins with a brief overview of the different crystal structures that exist in nature. The different atomic hybridizations is explained in detail and the subsequent effect on their band structures. Atomic structures of 2-D semiconductors in particular is explained - with special references to Graphene and Molybdenum Disulfide. The theory of linear combination of atomic orbitals (LCAO) is one of the most important principles in molecular theory. The expression of electronic wave functions in terms of bloch wave function, and the application to molecular and solid physics is explained. Reciprocal space of unit cells and Brillouin zones form an integral part of the bloch theorem. Brillouin zones form the peridodic structure which can be replicated throughout the solid - once the energy band of the 1st Brillouin zone is obtained, the entire solid can be represented with the same energy band. Hamilto- nian matrix is the heart and soul of the energy band calculations. The latter part of

6 CHAPTER 1. INTRODUCTION

this chapter, starts to connect the dots to the Hamiltonian matrix - the construction, factors affecting and the solution. The tight binding theory is closely linked to the building of the Hamiltonian matrix. The tight binding parameters for general solids (1-D and 2-D) and in specific to Graphene and Molybdenum Disulfide is the highlight of Chapter 3. Chapter 4 covers special scenarios to be considered due to the different morphologies of Graphene Nano-ribbons and Molybdenum Disulfide available. The density of states for Graphene is also calculated. The primary goal of this research is to aid students understand band structure calculations in the easiest way possible. A model simulation GUI is created as an interface, to facilitate better. MATLAB is used as the primary tool for all band structure and GUI calculations. The codes for all calculations are presented in the appendices at the end.

7 Chapter 2 Atomic Hybridization & Hamiltonian Matrix

Atomic hybridization is a crucial step towards achieving the band structure of any material. Earlier theories like the Valence Band theory fell short in describing the shapes of the bonds formed between atoms and electrons, which was explained by hybridization. Most of the modern day chemistry uses the a combination of the two theories, along with molecular orbital theory. We obtain the hybridization by adding up the number of bonded atoms and lone pair electrons as sp, sp2 or sp3. This chapter covers the theory by which these complex structures are formed and how the orientation and symmetry aﬀects the Hamiltonian Matrix. In the latter part of this chapter, the concept of the Schrödinger equation is introduced. This leads to the derivation of the Hamiltonian Matrix. A general Hamiltonian matrix

is derived and the exercise is repeated for GNRs and MoS2. Relevant Hamiltonian matrix eigen values are related to the inter-atomic potentials and atomic symmetry.

2.1 Eﬀect of Diﬀerent Atomic Structures

As mentioned earlier, the type of hybridization is decided on the basis of the number of bonded atoms and lone pair electrons. This also sets the dimensional geometry of the material. Table 2.1 below, describes this relation eﬀectively.

8 CHAPTER 2. ATOMIC HYBRIDIZATION & HAMILTONIAN MATRIX

Table 2.1: Types of hybridization [2]

Bonded atoms Lone pairs Hybridization Geometry

4 0 sp3 Tetrahedral (3-D)

3 0 sp2 Trigonal (2-D)

2 0 sp Linear (1-D)

3 1 sp3 Trigonal Pyramidal

2 1 sp2 Bent at an angle of 120°

2 2 sp3 Bent at an angle of 109°

Orbitals with zero, one, two and three units of angular momentum are called s, p, d and f orbitals respectively [9]. 2-D semiconductors are sp2 hybridized and exhibit planar geometry. However, there are higher hybridizations that exist like sp3d and sp3d2, but are outside the scope of this research. Fig. 2.1 depicts typical s and p orbitals.

9 CHAPTER 2. ATOMIC HYBRIDIZATION & HAMILTONIAN MATRIX

(b) p orbital

(a) s orbital

From Antonio H Castro Neto.,Les houches notes on graphene, arXiv:1004.3682, 2010. Reprinted with permission of Cornell University Library.

Figure 2.1: Typical s and p orbital structures [4]

These orbitals interact with each other and create an interatomic potential. The lower value of the energy expectation state which is created due to in-phase interactions is known as the bonding state. The other state with out-of-phase interactions is known as the anti-bonding state. This wave function is orthogonal to that of the bonding state [9]. There are also non-bonding orbitals present that contain lone pair of electrons and are neither bonding or antibonding. When two hybrid orbitals from diﬀerent atoms interact with each other, sigma bonds (σ bonds) are created. The interacting orbitals can be 2 p orbitals or between an s and p orbital, as seen in Fig. 2.1 a & b. These bonds are characterized by their electron density being along the bond axis. The overlap between 2 p orbitals creates pi bonds (π bonds). These bonds have their electron density above and below their bond axis, as observed in Fig. 2.2 [10].

10 CHAPTER 2. ATOMIC HYBRIDIZATION & HAMILTONIAN MATRIX

From Eric V Anslyn and Dennis A Dougherty, Modern physical organic chemistry University Science Books, 2006. Reprinted with permission of University Science Books.

Figure 2.2: Typical σ and π bonds [10]

2.1.1 Graphene

Graphene is an ideal two-dimensional allotropic form of carbon. The carbons atoms interconnect with each other to form a periodic arrangement of hexagonal network as illustrated in Fig. 2.3.

11 CHAPTER 2. ATOMIC HYBRIDIZATION & HAMILTONIAN MATRIX

Resimulated from L. Brey and H. A. Fertig,Phys. Rev. B 73, 235411. ©American Physical Society. Resimulated and printed with permission of APS.

Figure 2.3: Monolayer of 2D Graphene showing Armchair and ZigZag edges [11]

Carbon has atomic number 6 and has an electronic conﬁguration of 1s2 2s2 2p2.

Carbon has 4 valence electrons and they are distributed into s, px, py and pz bond orbitals. Graphene is a planar sp2 hybridized structure and essentially contains 3 σ and one π bond. Due to its honeycomb lattice structure, graphene has 2 diﬀerent edges. The top and bottom edges are zigzag shaped and the left and right edges are armchair shaped. The energy dispersion relation for Graphene changes along each edge. Graphene has 2 atoms per unit cell as seen in Fig. 2.3. The blue colored dots represent an atom A and red colored dots represent the 2nd atom B. The highlighted red and blue rectangles represent Zigzag and Armchair GNRs of unit atom width respectively.

The graphene lattice structure can be deﬁned by primitive lattice vectors a=a0(1,0) √ and b=a0(1/2, 3/2), where, a0 is the graphene lattice constant (=0.246 nm). Bilayer graphite is shown in Fig. 2.4. The distance between adjacent graphene layers is c = 1.42 Å.

12 CHAPTER 2. ATOMIC HYBRIDIZATION & HAMILTONIAN MATRIX

Figure 2.4: Layered Structure of 2D Graphene with interplanar distance c = 1.42 Å

The interaction of the outermost orbitals containing valence electrons (2s2 2p2) creates an inter-atomic potential and aﬀects the E-k by forming π and σ bonds in graphene. As the π bonds are perpendicular to the plane, unlike the σ bonds, they aﬀect the E-k relation when graphene layers are stacked (bilayer or multiple layered graphene structure). Graphene’s interatomic interactions can be seen in Fig. 2.5 below. (a), (b) and (c) are the σ bonds and (d) is the π bond.

Figure 2.5: Interatomic orbital interaction of Graphene (a)Vssσ, (b)Vspσ, (c)Vppσ, (d)Vppσ [4]

From Antonio H Castro Neto.,Les houches notes on graphene, arXiv:1004.3682, 2010. Reprinted with permission of Cornell University Library.

13 CHAPTER 2. ATOMIC HYBRIDIZATION & HAMILTONIAN MATRIX

2.1.2 Molybdenum Disulﬁde (MoS2)

The general structure for all transition metal dichalcogenides generally remains same with diﬀerent lattice constant values. Monolayer Molybdenum disulﬁde (ML- MDS) is a transition metal dichalcogenide (TMD). TMDs generally have the chemical

formula of MX2, where M is the transition metal chemically bonded to 2 atoms of chalcogenide atoms (X). Other examples of TMDs include Tungsten disulﬁde

(WS2), Tungsten diseluride (W Se2) etc. TMDs generally are not perfectly planar like Graphene Nano-Ribbons. This is because the bonds involve s-, p- and d- orbitals. Due to this reason, the two M-X bonds stick out of the plane. However, these structures have no broken/dangling bonds at the surface like perfect 3-D crystals [8].

The bandgap is known to vary from monolayer to bulk structures. In MoS2, the band gap transitions for bulk materials from a indirect band gap of 1.3 eV to a direct band gap of 1.8 eV in monolayer structures [12]. Structure of a typical ML-MDS and a unit atom structure is shown in Fig. 2.6.

14 CHAPTER 2. ATOMIC HYBRIDIZATION & HAMILTONIAN MATRIX

From Tight-binding model and direct-gap/indirect-gap tranisition in single-layer and multi-layer

MoS2, Physical Review B, E Cappelluti et al. Reprinted with permission of APS.

Figure 2.6: (left) Model atomic structure of MoS2. The small green rectangle represents unit cell of monolayer MoS2, which is doubled in a bulk structure (red rectangle). (right) Detailed trigonal prismatic structure for 2 layers of MoS2 with lattice constants. [13]

In Fig. 2.6, (left) we observe the structure of MoS2 atoms in a bulk structure. The blue and yellow circles represent Mo and S atoms respectively. The arrangement of atoms in monolayer or bulk structure is shown in further detail in the Fig. alongside

(right). The MoS2 structure forms an almost perfect trigonal prismatic structure with Mo atoms sandwiched between 2 layers of S atoms. The lattice constants a, b are Mo- Mo or S-S distances and nearest neighbour Mo-S distances respectively. The values

p2 assigned are experimentally observed to be a = 3.16 Å and b = 7/12a. c0 = 6.14 Å

and d are the distances between Mo-Mo planes and 2 monolayers of MoS2 respectively. u = 1.586 Å and w are distances between Mo-S planes and S-S planes respectively.

Again, the band gap also changes with the number of layers of MoS2 involved. A transition from direct gap in monolayer to a smaller indirect gap is observed when

15 CHAPTER 2. ATOMIC HYBRIDIZATION & HAMILTONIAN MATRIX

more than a single layer is involved.

2.2 Linear Combination of Atomic Orbitals (LCAO)

Molecular orbitals consist of atomic orbitals - described by wave mechanics as a wave form - Ψ. The wave function of an electron/molecule gives all the information about it [9]. However, to obtain the exact wave function of an electron is diﬃcult, as it is obtained by solving the Schrödinger equation which has no closed form solution [14]. Hence, the molecular orbital wave function is obtained by summing up the individual wave functions of the atoms contained in the solid. This process is known as the Linear Combination of Atomic Orbitals (LCAO). This is one of the standard methods to solve the periodic potential problems in the theory of electron motions in solids [15]. An oversimpliﬁed implementation of LCAO is given below,

Ψ = Ψ1 ± Ψ2 ± · · · ± Ψn (2.1)

where, Ψ is the wave function of the solid and Ψ1, ..., Ψn are the wave functions of the ’n’ atoms/molecules within the solid. This theory was originally proposed by Bloch and is used as an intermediary method for interpolation. This theory is used along with the tight binding theory and Bloch’s theorem described in further detail ahead in this chapter. To understand the application of the LCAO, it is essential to understand real space and reciprocal space structures of diﬀerent unit cells. This is covered in the following sections.

2.3 Real and Reciprocal Space

In solids, atomic valence levels broaden into bands that are mobile and each

electronic state characterized by a momentum p or wave number k = p/~, that is #» restricted to a Brillouin zone [9]. The wave vector k , gives information about the

16 CHAPTER 2. ATOMIC HYBRIDIZATION & HAMILTONIAN MATRIX

electron motion and its direction. It tells us about the number of oscillations it can cover in a unit distance. To understand Brillouin zones, it is essential to ﬁrst know the unit cell structures of diﬀerent solids in real and reciprocal spaces.

2.3.1 Unit cells - Real space

Solids are made up of lattices which is a periodic arrangement of atoms. Real space is the arrangement of atoms in a unit cell in the solid. Any lattice in real space, #» deﬁned by R,

#» #» #» #» R = X a + Y b + Z c (2.2) where, a, b and c are vectors inclined at angles α, β and γ angles to the x, y and z axes respectively.

Figure 2.7: Real Space Primitive unit cell

Fig. 2.7 is a primitive unit cell which repeats throughout the solid. If this were a simple cubic unit cell, all the side would be equal and same length, say, a. The atoms are present at each of the 8 vertices of the cube and all the internal angles would be 90°. Cesium Chloride (CsCl) exhibits a unit cell of the simple cubic kind. In general, there exists two kinds of lattices - Bravais and non-Bravais lattice structures. In a Bravais lattice, all the lattice points are equivalent as compared to a non-Bravais

17 CHAPTER 2. ATOMIC HYBRIDIZATION & HAMILTONIAN MATRIX lattice. The simple cubic lattice described earlier is a Bravais lattice. Bravais lattices also include the hexagonal, face centered cubic (FCC) and body centered cubic (BCC) types. The cubic system, is one of the 14 crystal systems. The lattice structures for 5 Bravais cubic cyrstal structures is described in Fig. 2.8 below [3].

(a) Simple Cubic (b) Hexagonal Lattice

(d) FCC Lattice (c) BCC Lattice

Figure 2.8: Bravais Lattice Structures

The hexagonal lattice has an atom at each of its 12 vertices and one on each of its hexagonal faces. The Body Centered Cubic (BCC) Lattice, has an atom at each of its 8 vertices and one in the center of its rectangular body. The Face Centered Cubic Lattice (FCC) is similar in structure to the BCC structure. The atoms at the vertices are similarly placed and instead of the atom at the center of the body, an atom is placed centrally at each of the 6 rectangular faces.

18 CHAPTER 2. ATOMIC HYBRIDIZATION & HAMILTONIAN MATRIX

The non Bravais lattices have non-equivalent lattice points. Bravais or non-Bravais, there are seven crystal systems present in general. These are enlisted in Table 2.2.

Table 2.2: Seven types of lattice crystal systems [3]

From RHEED Transmission Mode and Pole Figures Thin Film and Nanostructure Texture Analysis, Gwo-Ching Wang, Toh-Ming Lu. Reprinted with permission. System Conventional cell & axes Number of Lattices

Cubic a = b = c 3 α = β = γ = 90°

Hexagonal a = b 6= c 1 α = β = 90°, γ = 120°

Triclinic a 6= b 6= c 1 α 6= β 6= γ

Trigonal a = b = c 1 α = β = γ < 120°, 6= 90°

Monoclinic a 6= b 6= c 2 α = γ = 90°, 6= β

Tetragonal a = b 6= c 2 α = β = γ = 90°

Orthorhombic a 6= b 6= c 4 α = β = γ = 90°

2.3.2 Reciprocal Spaces

The reciprocal space is also known as the kx-ky plane and is the Fourier transform #» of the spatial lattice in real space. This space gives us the limit for wave vector k in the ﬁrst Brillouin Zone of the lattice, to calculate the energy dispersion relation. Any point in this space, can be described by,

19 CHAPTER 2. ATOMIC HYBRIDIZATION & HAMILTONIAN MATRIX

#» # » # » # » K = MK1 + NK2 + RK3 (2.3) # » # » # » Where, X, Y, Z, M, N and R, are integers and vectors K1, K2 and K3 are given by,

#» #» # » 2π( b × c ) K1 = #» #» #» a · ( b × c )

#» #» # » 2π( c × a ) K2 = #» #» #» b · ( c × b )

#» #» # » 2π( a × b ) K3 = #» #» #» c · ( a × b )

#» #» #» and a , b and c are unit vectors in real space. For every lattice in the real space, there exists a lattice in the reciprocal space or k-space or momentum space. The reciprocal lattices for Fig. 2.8, as in Fig. 2.9 below respectively. BCC lattice structures are very common in metals like iron (Fe), lithium (Li), chromium (Cr), sodium (Na) and tungsten (W). Examples of FCC lattices include Copper (Cu), Gold (Au), Silver (Ag) and Nickel (Ni).

20 CHAPTER 2. ATOMIC HYBRIDIZATION & HAMILTONIAN MATRIX

(a) Simple Cubic Reciprocal Space (b) Hexagonal Lattice Reciprocal Space

Figure 2.9: Reciprocal Lattice Structures

It can be seen from Fig. 2.9, that the reciprocal space of a simple cubic lattice of side ’a’, is the same lattice, with sides of length 2π/a. The hexagonal lattice also has the same structure in reciprocal space, however, the structure is rotated by an angle of 30°, along the z-axis. For both BCC and FCC lattices of side a, the sides are of length 4π/a. The ﬁrst Brillouin zones for each of the lattices are shown in the highlighted blue polyhedral structure.

2.3.3 Graphene Reciprocal Space

For Fig.2.3 as the structure in real space, the reciprocal space of monolayer graphene is as seen in Fig.2.10 below.

21 CHAPTER 2. ATOMIC HYBRIDIZATION & HAMILTONIAN MATRIX

Figure 2.10: Reciprocal Space of Graphene

Fig.2.11 below depicts the 1st Brillouin zone (BZ) of graphene. The BZ of graphene is hexagonal in shape and the corners of the Brillouin Zone are known as Dirac points. Dirac Points are the points in the BZ where the energy is zero. The center of the BZ is known as the Γ (gamma) point. The Dirac points are named as K and K‘ points alternatively and M is the midpoint between K and K‘ points.

Γ = (0,0)

2π (1, √1 ) K = 3a 3

2π M= 3a (1, 0)

‘ 2π K = 3a (0, 1)

Figure 2.11: 1st BZ of Graphene with co-ordinates

The π and π∗ bands (bonding and anti-bonding respectively) lie at the corners of the hexagonal Brillouin zone as seen in Fig. 2.12 below.

22 CHAPTER 2. ATOMIC HYBRIDIZATION & HAMILTONIAN MATRIX

Figure 2.12: σ and π bands of Graphene in 1st BZ [4]

From Antonio H Castro Neto.,Les houches notes on graphene, arXiv:1004.3682, 2010. Reprinted with permission of Cornell University Library.

A 3D Brillouin zone for Graphene is seen in Fig. 2.13 below.

Figure 2.13: 3D 1st BZ of Graphene

2.3.4 Molybdenum Disulﬁde Reciprocal Space

The in-plane Brillouin zone for MoS2 is hexagonal in shape, similar to Fig. 2.11 for Graphene. The points of high symmetry vary in the case of MoS2 to Γ = (0, 0), √ K = 4π/3a(1, 0) and M = 4π/3a(0, 2 3/2).

23 CHAPTER 2. ATOMIC HYBRIDIZATION & HAMILTONIAN MATRIX

2.4 Schrödinger Equation

The Schrödinger equation calculates on a quantitative basis the energy levels for a conﬁning potential. The most general form of the Schrödinger equation is as follows:

Hˆ Ψ = EΨ (2.4) where, Ψ is the wave function of an electron in unit cell, E is the energy and Hˆ is the Hamiltonian operator. The Schrödinger equation implies that when a Hamiltonian operates on a wave function Ψ, the result would be proportional to the same wave function and the proportionality constant E is the energy of the wave function. Hence, by solving the Schrödinger equation, and diagonalizing the Hamiltonian, we can obtain the energy eigen values with respect to the wave vector (k). The Schrödinger equation can be solved analytically but to obtain numerical solutions, there are numerical methods, most of which, convert wavefunction Ψ into a column vector [Ψ(t)] and the diﬀerential operator Hˆ into a matrix [H]. This changes the above mentioned Schrödinger equation into a matrix equation,

d ih¯ [Ψ(t)] = [H][Ψ(t)] (2.5) dt

The solution to the above equation is to ﬁnd the eigen values Eα and eigenvectors α of the matrix [H] [16]:

[H][α] = Eαα (2.6)

2.5 Hamiltonian Matrices

From Eq. 2.6, it is observed that if the electron states are represented by linear combination of atomic orbitals, the electron energy bands are found to depend on a set

24 CHAPTER 2. ATOMIC HYBRIDIZATION & HAMILTONIAN MATRIX

of orbital energies and inter atomic Hamiltonian matrix elements [9]. Atomic orbitals are used as basis to obtain the matrix representation of the Hamiltonian operator. The derivation of Hamiltonian matrix elements is elaborated in sections 2.5.2 - 2.5.5.

2.5.1 Basis of Orbitals

The concept of basis functions provides us a computational tool to analyze the Schrödinger equation. The wave function of an atomic orbital, can be expressed in terms of a set of basis functions [16],

M #» X #» Φ( r ) = φmum( r ) (2.7) m=1 #» #» where, um r is the basis function for the wave function Φ( r ). The wave function, can again be expanded into a column vector as described earlier. The advantage of using basis functions is that one can choose the values of the basis orbitals, so that the size of the Hamiltonian matrix reduces to reduce complexity. We begin with building the Hamiltonian matrix for a simple one-dimensional atomic solid and the general Hamiltonian structure for any solid.

2.5.2 Hamiltonian for one-dimensional atomic structure

By method of ﬁnite diﬀerences, the Hamiltonian matrix is given by,

Hn,m = [Un + 2t0]δn,m − t0δn,m+1 − t0δn,m−1 (2.8)

where, δn,m is the Kronecker delta function. The value is one, if n is equal to m, else, is zero. To simplify or expand the value of the Hamiltonian, we consider a one-dimensional solid composed of N atoms, whose Hamiltonian matrix is described by [16],

25 CHAPTER 2. ATOMIC HYBRIDIZATION & HAMILTONIAN MATRIX

|1 > |2 > · · · |N − 1 > |N >

|1 > Eo Ess ··· 0 Ess ········· (2.9)

|N − 1 > 0 0 ··· Eo Ess

|N > Ess 0 ··· Ess Eo

The matrix elements signify the nearest neighbor interactions, where |1 >, |2 >, |3 >,... |N > are atomic orbitals of each of the N-atoms respectively. This implementation follows a condition known as the periodic boundary condition (PBC). This condition implies that the solid does not end abruptly and instead, the last atom wraps around and overlaps the ﬁrst atom forming a ring in a one-dimensional solid. In case of a non-periodic solid, where there is abrupt truncation of the solid, ele-

ments H1,N and HN,1 would be zero. Periodic boundary conditions reduces complexity mathematically however doesnt take into consideration the eﬀect of surfaces [9].

The matrix elements E0 and ESS are arbitrary energy eigen values. E0 is a symbol

for diagonal elements of the matrix and ESS symbolizes the oﬀ diagonal elements. In

simple terms, E0 and Ess signify intra- and inter-atomic potentials respectively. The subscript ‘ss’ does not stand for or signify the s orbital for the solid and is meant to be used as a generic symbol.

2.5.3 General Hamiltonian

In the earlier case, the unit cell of the one-dimensional solid consists of a single atom. Hence, the size of the Hamiltonian matrix is N × N. Similarly, in a solid containing a unit cell with 2 atoms, the size of the matrix would be 2N × 2N. In general, the size of the Hamiltonian depends on the arbitrary number of basis functions per unit cell, ‘b’. The size of the Hamiltonian would then be (b × b). Consider a unit cell

26 CHAPTER 2. ATOMIC HYBRIDIZATION & HAMILTONIAN MATRIX

‘n’, connected to ‘m’ neighboring unit cells by Hamiltonian Hnm. The Schrödinger equation from (1) changes to:

X [Hnm]{φm} = E{φn} (2.10) m

Where, {φm} is a (b x 1) column vector denoting the wave function for unit cell ‘m’ and is given by,

#» #» i k · d m {Φm} = {Φo} exp (2.11)

This is deﬁnition is similar to that of "Bloch Theorem". Bloch theorem is one of the standard methods for solving problems related to periodic potential. It elaborates on making a linear combination of atomic orbitals located on the various atoms of the crystal. The wave functions for electrons in a crystal, have eigen states and can be written in the form,

#» # » X i k ·dm Ψ(r) = e Φn(r − Ri) (2.12)

Ri where, Ψ is the complete wave function and Φn is the wave function of atomic orbital # » of atom ‘n’ at a position Ri and dm is the vector connecting nearest neighboring atoms. The sum is extended over all the atoms at respective positions in the unit cell of the crystal. On, the basis of the Bloch Theorem, the Schrödinger equation from (1) can be modi- ﬁed to,

#» E{Φo} = [h( k )]{Φo}

27 CHAPTER 2. ATOMIC HYBRIDIZATION & HAMILTONIAN MATRIX with,

#» #» # » # » X i k ·(dm−dn) [h( k )] = [Hnm]e (2.13) m

2.5.4 Hamiltonian for sp2 Hybridiation (Special Case : Graphene)

Applying, the above tight binding and Bloch theorem to Graphene, we begin with the equation for the wave function having 2 atoms per unit cell A and B.

X X Ψ(r) = ΨA(RA)Φ(r − RA) + ΨB(RB)Φ(r − RB) (2.14)

RA RB where, Φ(r − RA) and Φ(r − RB) are the wave functions of the pz orbitals of graphene,

located at origins RA and RB respectively.

If τ1, τ2 and τ3 are the vectors connecting the nearest carbon atoms and are deﬁned as,

√ τ 3 τ1 1 = a(0,1/ ) √ τ2 = a(-1/2,-1/2 3) √ τ3 τ3 = a(1/2,-1/2 3) τ2

Figure 2.14: Nearest neighbor vectors of Graphene

If only the pz orbitals of graphene are considered, the Hamiltonian changes from

28 CHAPTER 2. ATOMIC HYBRIDIZATION & HAMILTONIAN MATRIX the general expression to,   pc pa  z z   #» #» #»  H = pc V (exp(i k · τ ) + exp(i k · τ ) + exp(i k · τ ))  z π ppπ 1 2 3  #» #» #»  a  pz Vppπ(exp(−i k · τ1) + exp(−i k · τ2) + exp(−i k · τ3)) π (2.15)

The following points need to be kept in mind before forming the Hamiltonian matrix for all the π and σ bands of graphene [9]:

• The superscripts ‘a’ and ‘c’ stand for anion and cation respectively. Graphene may not have anions and cations per se, but they symbolize the the atoms A and B from Fig. 2.3.

• For the π bands, the pz orbitals in A interact only with the pz orbitals in atom

B and the interactions with other orbitals - s, px and py is zero.

• For the σ bands, s orbital of the anion will have zero interaction with other

orbitals -px and py of the same anion. Similar theory is applicable to the cation.

• Vintra is the intra-atomic hybridization potential between orbitals |2> and |3> or rather inter-atomic interaction between s-orbital of anion and p-orbitals of cation and vice-versa.

• Vσ is the tight binding parameter between two |2> states or in other words, anion and cation of same orbital.

• σ is the tight binding parameter between two |2> states or in other words, between 2 anions/cations of same orbital.

Considering all the interacting atomic orbitals of Graphene, leads to the Hamiltonian of Graphene as [?],

29 CHAPTER 2. ATOMIC HYBRIDIZATION & HAMILTONIAN MATRIX

  |0 > |0 > |1 > |1 > |2 > |2 > |3 > |3 >      pc pa sa sc pc pa pc pa   z z x x y y     c  |0 > pz π VppπhAB(k) 0 0 0 0 0 0     a ∗  |0 > pz VppπhAB(k) π 0 0 0 0 0 0      |1 > sa 0 0 V eik·τ1 V 0 V 0   σ σ intra intra  H =    c −ik·τ1  |1 > s 0 0 Vσe σ 0 Vintra 0 Vintra     c ik·τ2  |2 > px 0 0 Vintra 0 σ Vσe Vintra 0      |2 > pa 0 0 0 V V e−ik·τ2 0 V   x intra σ σ intra     c ik·τ3  |3 > py 0 0 Vintra 0 Vintra 0 σ Vσe    a −ik·τ3 |3 > py 0 0 0 Vintra 0 Vintra Vσe σ (2.16) The ﬁrst two columns stand for the π bands and remaining 6 columns are for the 3 σ bands. The above 8 × 8 matrix can be split into 2 × 2 and 6 × 6 matrices for ease of calculations. The above matrix can be individually or as a whole be diagonalized to get the Energy-dispersion relation.

3 2.5.5 Hamiltonian for sp d Hybridisation (Special Case : MoS2)

MoS2 has an eleven-band electronic band structure. The eleven bands are composed of ﬁve 4d orbitals of Mo and six 3p orbitals of S. The six orbitals of S, includes both the top and bottom layers which sandwiches the intermediate Mo layer. Out of the eleven bands, there are four conduction bands and seven valence bands [13]. We follow notation from E. Cappelluti et. al. [13], for the ﬁve 4d orbitals of Mo

2 2 2 2 4d(x − y ) and 4d(xy) as d2, 4d(xz) and 4d(yz) as d1, 4d(3z − r ) as d0; pxy is used

to denote the S 3px, 3py orbitals and pz for the 3pz orbital. The unit cell of MoS2 is deﬁned by a 11-fold Hilbert space vector, given by,

30 CHAPTER 2. ATOMIC HYBRIDIZATION & HAMILTONIAN MATRIX

† † † † 2 2 † 2 2 † † † † Φi = (p(i, x, t) , p(i, y, t) , p(i, z, t) , d(i, 3z − r ) , d(i, x − y ) , d(i, xy) , d(i, xz) , d(i, yz) ,

† † † p(i, x, b) , p(i, y, b) , p(i, z, b) )

(2.17)

where, all the p and d orbitals in the above equation create electrons in the 0α0 orbital of the Mo atom in the i-unit cell. In p(i, x, t) and p(i, y, t), the ’t’ stands for the top layer in the S atom and ’b’ in the other p-orbital notation stands for the bottom layer. The Hilbert space is an abstract vector space whose exact nature depends on the system. Each eigenstate of the observable in the Hilbert space, corresponds to the eigenvector of the respective operator, which gives us the eigenvalues associated to that state. Considering nearest neighbor interactions, there are 12- tight binding parameters for MoS2 - they are described in detail in the following chapter. In terms of orbital basis, the Hamiltonian for MoS2 is described as,

ˆ (1L) X ˆ † (1L) ˆ H = Φi (k)H(i, j) (k)Φj(k) (2.18) i,j,k

The above equation is a minimalistic tight binding Hamiltonian representation for

MoS2. This model considers only the diagonal matrix elements and the hopping terms between atoms. This model takes into account the mirror symmetry in the crystal - xy plane (µ1), mirror symmetry in yz plane (µ2) centered around each atomic position and three-ﬁeld rotation symmetry (R3) [17]. These relevant symmetry operations reduce the number of independent tight binding parameters involved in the Hamiltonian. The following equations describe the matrix elements:

• Diagonal elements:

(1L) (1) (2) Hi,i = i + 2ti,i cos(k · δ1) + 2ti,i [cos(k · δ2) + cos(k · δ3)] (2.19)

31 CHAPTER 2. ATOMIC HYBRIDIZATION & HAMILTONIAN MATRIX

• Positively symmetric under µ2 for (i, j) = (3, 5), (6, 8), (9, 11),

(1L) (1) (2) (3) Hi,j = 2ti,j cos(k·δ1)+ti,j [exp(−ik·δ2)+exp(−ik·δ3)]+ti,j [exp(ik·δ2)+exp(ik·δ3)] (2.20)

• Negatively symmetric under µ2 for (i, j) = (1, 2), (3, 4), (4, 5), (6, 7), (7, 8), (9, 10), (10, 11),

(1L) (1) (2) (3) Hi,j = −2iti,j sin(k·δ1)+ti,j [exp(−ik·δ2)−exp(−ik·δ3)]+ti,j [− exp(ik·δ2)+exp(ik·δ3)] (2.21)

• Diﬀerent hopping exists between Mo and S atoms. For positively symmetric atoms i and j, which includes pairs (3,1), (5,1), (4,2), (10,6), (9,7), (11,7), (10,8),

(1L) (4) Hi,j = ti,j [exp(ik · δ4) − exp(ik · δ6)] (2.22)

• While, for negatively symmetric (i,j) = (4,1), (3,2), (5,2), (9,6), (11,6), (10,7), (9,8), (11,8), the relation is deﬁned as,

(1L) (4) (5) Hi,j = ti,j [exp(ik · δ4) + exp(ik · δ6)] + ti,j δ5 (2.23)

(1L) (1L) ∗ • At all times, the Hamiltonian requires H(i,j) (k) = H(j,i) (k)

The resulting complete Hamiltonian matrix, with individual matrix elements is

described in detail in appendix . MoS2 has a 11 × 11 matrix which solves to obtain

the 11 bands. MoS2 has 12 tight binding parameters - ∆0, ∆1, ∆2, ∆p, ∆z, Vddδ,

Vddπ, Vddσ, Vppδ, Vppπ, Vppσ, Vpdπ. These parameters describe the atomic levels - l = 0, l = 1 and l = 2 for the Mo d orbitals and the in-plane and out-of-plane S orbitals. These tight-binding or Slater-Koster parameters is described in detail in the upcoming chapter.

32 CHAPTER 2. ATOMIC HYBRIDIZATION & HAMILTONIAN MATRIX

2.6 Conclusion

The energy dispersion models for Graphene, GNRs and MoS2 is described in this chapter. There are other models that exist - three band and ﬁve band models which consider fewer bands and hence, calculations are must simpler. There are models that

disregard the role of pz orbitals of S atoms. However, this is found to be unavoidable to get a complete understanding of the evolution of band structures [13]. The crucial aspect is to understand the various factors that play into aﬀecting the band gap that the material exhibits. Most important are the Γ and K points which are crucial to the band structures in monolayer and multi-layered compounds. The changes in topology of the atomic structure in the Brillouin zone is responsible for the variation in single layered to multi-layered compounds.

33 Chapter 3 Tight Binding Theory and Solving for Energy Dispersion Relation

3.1 Introduction

Tight binding model is found most accurately the simulation of the band structure and control the transition from direct band gap material in single-layered systems to indirect band gap material in multi-layered systems. It assumes that there exists strong crystal or ionic potentials. This implies that, when an electron is captured while traveling through the lattice, it remains bound to the ion for a long period of time before it moves ahead. Thus, during this period, the electron is tightly bound to its own atom and is unaﬀected by other atoms [18].

3.2 Introduction

The Hamiltonian H is a quantum operator which relates the total energy of a particle. Hence, it can be represented as the sum of the kinetic and potential energies [18], ~2 d2 H = − 2 + V (x) (3.1) 2m0 dx

34 CHAPTER 3. TIGHT BINDING THEORY AND SOLVING FOR ENERGY DISPERSION RELATION

V(x) is the crystal potential and can be deﬁned using the LCAO theory as,

X V (x) = v(x − Xj) (3.2) j

= v(x) + V 0(x) (3.3) where, v(x) and V 0(x) are the potentials due to the atom itself and other atoms respectively. The general expression for energy, as per Bloch’s theorem is given as,

1 X ik(Xj −X 0 ) E(k) = exp j < φv(x − Xj0 )|H|φv(x − Xj) > (3.4) N j,j0

Energy is expressed as a double summation over j and j’ and is a function of the

diﬀerence Xj − Xj0 . Hence, the above expression can be, by keeping Xj0 arbitrarily at zero, rewritten as,

0 X ikXj E(k) =< φv(x)|H|φv(x) > + exp < φv(x)|H|φv(x − Xj) > (3.5) j

V 0(x) is the potential observed due to other atoms. This term is negligible when the atom is considered to be at the origin. Hence, the ﬁrst term in the above equation 3.5, can be rewritten as,

− 2 d2 < φ (x)|H|φ (x) >=< φ (x)|[ ~ + v(x)]|φ (x) > + < φ (x)|V 0(x)|φ (x) > v v v 2m dx2 v v v (3.6) The second term in equation 3.5, involves nearest neighbor interactions. The second term can be expanded as,

2 2 h d 0 < φv(x)|H|φv(x−a) >=< φv(x)|− 2 +v(x−a)|φv(x−a) > + < φv(x)|V (x−a)|φv(x−a) > 2m0 dx (3.7)

35 CHAPTER 3. TIGHT BINDING THEORY AND SOLVING FOR ENERGY DISPERSION RELATION

Equation 3.7 has 2 terms - the ﬁrst of which, can be ignored, as it there is negligible

overlap between 2 different atoms - φv(x) and φv(x − a). The second term, proves to be a more influential parameter known as the overlap integral. It is also symbolized as the parameter 0Γ0 and is dependent on the interaction between orbitals centered at two different neighboring atoms. As the number of atoms and orbitals interacting increases, the complexity of the above equation increases. There are more interacting parameters or ’tight binding’ parameters involved increases and it is found to vary from material to material. Another important tight binding parameter, usually observed in band structure cal-

0 0 culations, is the hopping/transfer integral −‘t /‘Vppπ.If t → 0, the electrons do not hop into neighboring sites and the system is considered to be isolated. However, if t > 0, the electrons can stay in either/both electronic states at reduced kinetic energies. The overlapping integral, on the other hand, takes into consideration, the atomic interaction between the diﬀerent layers. The overlapping integral, in this research, is ignored for simplicity of calculations.

3.3 Tight Binding Parameters

In general, its been observed that, the tight binding parameters are functions of interatomic distances. Hence, the knowledge of the kind of bond - σ or π between two atoms is crucial to know, before we begin on band structure calculations. In this section, the tight binding parameters used for energy band calculations for each of

the three cases - Monolayer Graphene, GNRs and monolayer MoS2 is discussed.

3.3.1 Graphene

Typical value of transfer integral for the π orbitals of graphene lies in the range of 2.9-3.1 eV (used in Case 1 of this discussion). The tight binding parameters for graphene used in case 2 of this discussion can be found in table 3.1 below.

36 CHAPTER 3. TIGHT BINDING THEORY AND SOLVING FOR ENERGY DISPERSION RELATION

Table 3.1: Tight Binding Parameters for Graphene [4] Tight Binding Parameter Value

Vintra -2.77 eV

Vσ -12.3578 eV

Vppπ -2.4 eV

Vppσ 8.4 eV

Vspσ 5.4 eV

Vssσ -5 eV

3.3.2 Tight Binding Parameters for MoS2

Since, we do not consider the full-range eleven-band tight binding Hamiltonian (FTBH) model, the 12 original tight binding parameters reduce to the following six parameters,

Table 3.2: Tight Binding Parameters for MoS2 [1] [5] Tight Binding Parameter Expression Calculated Value (eV)

(1) t δ1 = a1 3.18

(2) t δ2 = a1 + a2 3.18

(3) t δ3 = a2 3.18

(4) t δ4 = −(2a1 + a2)/3 1.836

(5) t δ5 = (a1 + 2a2)/3

δ6 = (a1 − a2)/3 1.836

(6) t δ7 = −2(a1 + 2a2)/3

δ8 = 2(2a1 + a2)/3 3.6719

δ9 = 2(a2 − a1)/3

The detailed description of each of these Tight Binding and their independent

37 CHAPTER 3. TIGHT BINDING THEORY AND SOLVING FOR ENERGY DISPERSION RELATION

parameters is enlisted in Appendix D. The next step that follows tight binding calculations is the simulation of the respective band structures.

3.4 Solving the Hamiltonian

The Hamiltonian matrices are solved by diagonalization. This section will elaborate on the mathematical steps and the simulation ﬂow followed on a mathematical software like MATLAB. The increase in the level of complexity of solving the Hamiltonian matrix to obtain band structure calculations is clearly observed as we move from

Monolayer Graphene to GNRs to monolayer MoS2.

3.4.1 Graphene

In the case of Graphene, both hand-calculations and MATLAB simulations are

consistent with each other. The hand-calculations are simpler when only the pz orbitals are considered.

Case 1 : π bands, When π = 0 By following simple dot and cross product rules, we obtain,

E(k) = |hAB(k)| (3.8) q √ √ √ √ = ±Vppπ [exp(−ikya/ 3) + 2 exp(ikya/2 3)] · [exp(ikya/ 3) + 2 exp(−ikya/2 3)cos(kxa/2)] q √ √ 2 = ±Vppπ 1 + 2 exp(−ikya 3/2) cos(kxa/2) + 2 exp(ikya 3/2)cos(kxa/2) + 4 cos (kxa/2) q √ 2 E(k) = ±Vppπ 1 + 4 cos(kxa/2) cos(kya 3/2) + 4 cos (kxa/2) (3.9)

38 CHAPTER 3. TIGHT BINDING THEORY AND SOLVING FOR ENERGY DISPERSION RELATION

Case 2 : When π 6= 0, similar calculations lead to,

Eπ,±(k) = π ± |Vppπ||hAB(k)|, (3.10)

Eσ,1,±(k) = σ − Vintra ± Vsigma, (3.11)

r 2 Vintra 3Vintra 2 Eσ,2,±(k) = σ + + + V ± |VintraVsigma||γk|, (3.12) 2 2 σ r 2 Vintra 3Vintra 2 Eσ,3,±(k) = σ + − + V ± |VintraVsigma||γk|, (3.13) 2 2 σ

3.4.2 MoS2

The Hamiltonian for MoS2 is a 11 × 11 matrix. The size of this matrix hugely complicates the energy dispersion relation to follow simple dot and cross product or mathematical rules. Such simulations should be performed, preferably on softwares like MATLAB. There are 3 main commands in MATLAB that perform the main calculations,

[V,D]=eig(H); eigst = sum(D); E1(Nk,:)=sort(real(eigst));

The ﬁrst command, calculates for matrix (H), the Diagonal matrix (D) whose diagonal elements are eigen values of H. The matrix V correspondingly consists of the the eigen vectors from matrix H. The second command, calculates the sum of the diagonal eigen values and the real values are subsequently sorted in order to get the energy dispersion relation. This matrix is a (Nk×11) matrix, where Nk, decides the density of matrix calculations. The arbitrary value for this parameter is kept at 101 simulation points [16].

39 CHAPTER 3. TIGHT BINDING THEORY AND SOLVING FOR ENERGY DISPERSION RELATION

3.5 Energy Dispersion - Graphene and MoS2

3.5.1 Graphene

Limiting the values of k to that of the 1st BZ, we get the E-k relation for the π bands of graphene for case 1, as observed in Fig. 3.1 and Fig. 3.2.

Figure 3.1: Vppπ = 1 eV, π = 0

For case 2, when all the tight binding parameters are taken into consideration (as mentioned in table 3.1), the E-k changes to the following:

40 CHAPTER 3. TIGHT BINDING THEORY AND SOLVING FOR ENERGY DISPERSION RELATION

Figure 3.2: Vppπ = 2.7eV , π = 0

Figure 3.3: Energy Band Dispersion with all TB Parameters (1 Hartree = 27.21 eV)

From both the above cases, it can be seen that Graphene exhibits degeneracy at K point. The full model shows 4 bands in conduction and valence bands - 3 for the

41 CHAPTER 3. TIGHT BINDING THEORY AND SOLVING FOR ENERGY DISPERSION RELATION

σ bonds and one for the π bond. The conduction band minimum and valence band maximum is calculated from that of the π bond and hence, gives the characteristic to the material. Following is the 3D representation of the E-k for Graphene,

Figure 3.4: 3D Representation E-k for Graphene

The 6 Dirac points forming the hexagonal 1st BZ of graphene can be observed. It is noticed that at these points, the energy becomes zero.

3.5.2 Comparison to existing work - Graphene

The π and σ bands calculated by the simple tight binding method, discussed in this research, is consistent with the more sophisticated calculations and methodologies in [19], [?]. [19] calculates band structures from the 3-D Brillouin zone and relates the

42 CHAPTER 3. TIGHT BINDING THEORY AND SOLVING FOR ENERGY DISPERSION RELATION

results obtained from single layered structures. Hamiltonian elements are calculated without tight binding approximations or theory. The linear nature of the dispersion relation for Graphene near the Dirac points, is consistent throughout. Modifying Graphene structures to quasi - 1D structures known as Graphene nanoribbons increases the band gap with width reduction due to confinement effects [20]. The edge effects play a crucial role in bandgap engineering of graphene. [20] uses Den- sity Functional Theory (DFT) and Local Density Approximation (LDA) in addition to the basic tight binding theory to calculate the nature of Graphene and Graphene nano-ribbons. The findings of this research with respect to metallic/insulating nature of the GNR’s reestablishes earlier findings.

3.5.3 MoS2

The ﬁgure below shows the simulations obtained from the minimalistic model described earlier. It can be observed that along the boundaries of the ﬁrst Brillouin Zone, Γ-M-K-Γ, the model preserved the basic nature of the material and orbital

nature of the bands. Monolayer MoS2 exhibits a direct gap of energy.

In the case of monolayer MoS2, the conduction band minimum and valence band maximum is observed at K point. However, this shifts, when the number of layers is increased. This can be seen from the energy dispersion relation in Fig. 3.5.

43 CHAPTER 3. TIGHT BINDING THEORY AND SOLVING FOR ENERGY DISPERSION RELATION

Figure 3.5: Energy Dispersion for Monolayer MoS2 [1]

0 0 The valence band maximum is found to shift from ’K’ to Γ point in bilayer MoS2.

This research shows simulations only for monolayer MoS2. The application of tight binding calculations to multiple layers can be considered as future work. It is also observed that if the bands belonging to the same positive or negative symmetry group, the level anticrossing introduces orbital character mixing. Hence, to obtain accurate band structure simulations, it is necessary to include higher order coupling terms. This adds to the scope and extent of the research by which it can be taken forward.

This theory successfully simulates the basic nature of Monolayer MoS2. However, in comparison to earlier work by Ridolﬁ and others [21], the highest point of the conduction band does not match. This is because of the diﬀerence in the values of the tight binding parameters used. However, the lowest point of the valence band lies at ’K’ for both the models. In comparison with the DFT calculations used in [1], the simulations match perfectly in terms of the nature of the electronic bands.

44 CHAPTER 3. TIGHT BINDING THEORY AND SOLVING FOR ENERGY DISPERSION RELATION

3.6 Graphical User Interface

The graphical user interface has been setup for Graphene, GNRs and MoS2. Fig. 3.6 below, is a screenshot of the GUI for Graphene.

Figure 3.6: 2D and 3D representation of Graphene Energy Band Diagram

45 CHAPTER 3. TIGHT BINDING THEORY AND SOLVING FOR ENERGY DISPERSION RELATION

Figure 3.7: Armchair GNR GUI

Figure 3.8: Zigzag GNR GUI

46 CHAPTER 3. TIGHT BINDING THEORY AND SOLVING FOR ENERGY DISPERSION RELATION

Figure 3.9: Monolayer Molybdenum Disulﬁde GUI

The input parameters are the as per case 1 of this research - ’Graphene’ from the

pop-up menu, the tight binding parameter (Vppπ) and lattice constant value. The last parameter ’width’ is a term that would only aﬀect the band structure of A-GNR and/or Z-GNR. The GUI for A-GNR and Z-GNR cases is shown in Fig. 3.7 and Fig. 3.8 respectively.

There is a dropdown on the top-left corner that changes the energy dispersion according to material/ nano-structure chosen. The input parameters for the two special cases of Graphene would now in addition include the width of the GNR.

The GUI for MoS2 has all the tight binding parameters as input. MoS2 has 88 tight binding parameters and hence, are not available as modiﬁable inputs to this simulation.

A better understanding of the band structure in the case of MoS2 would be observed, if there were an extension to change these values and also the number of layers. This can be considered as an addition to the future work of this research.

47 Chapter 4

Edge Eﬀect of Graphene & Layers Eﬀect on MoS2

4.1 Graphene

As seen from the earlier chapters, monolayer graphene exhibits two kinds of struc- tural formations - Armchair and Zigzag GNRs. The electronic properties depends strongly on the size and geometry of the nanoscale carbon systems [11]. Boundary conditions considered for each of these cases, leads to diﬀerent varying results. The Armchair Graphene Nano-Ribbon (A-GNR) can be metallic or semi-conducting depending on the width of the GNR. This behavior is observed primarily due to the

topology of the π bands or the pz orbitals which is a signiﬁcant contributing factor in the ﬁrst Brillouin zone of the zigzag edge but has no role in the armchair edge. The zigzag GNR is always metallic. Fig. 4.1 below depicts a typical Armchair and Zigzag edged Graphene Nano-Ribbon with a width of 2 atoms.

48 CHAPTER 4. EDGE EFFECT OF GRAPHENE & LAYERS EFFECT ON MOS2

Figure 4.1: Armchair and Zigzag edged Graphene Nano-Ribbons with width = 2

The boundary condition for Z-GNRs is for the wave function to disappear on a single sublattice at each edge. This leads to conﬁned electronic states, with wave functions extended throughout the entire system and involve both the sublattices - unit atoms A and B (from ﬁgure 2.3). In the case of Z-GNRs there are surface states that are strongly localized on the nonvanishing edges. However, in A-GNRs, the wave function vanished on both sublattices and edges respectively. The general expression for any GNR to be metallic is [11],

L = (3M + 1)a0 (4.1)

where, ’M’ can be any integer and a0 is the graphene lattice constant.

4.1.1 A-GNR

Using the tight binding approximation and hard wall boundary condition, wave #» vector ( k ) is discretized in the conﬁned direction. A-GNR is found to be metallic when n = 3m + 2 (where, m is any integer)and is insulating otherwise [22]. From the tight binding approximation, we get,

49 CHAPTER 4. EDGE EFFECT OF GRAPHENE & LAYERS EFFECT ON MOS2

1 n X X ikxxA |ψ >A= e i φA(i)|Ai >, NA i=1 xAi 1 n X X ikxxB |ψ >B= e i φB(i)|Bi > . (4.2) NB i=1 xBi

φA and φB are the components of atoms A and B sublattices in the y direction, which is perpendicular to the edge. |Ai> and |Bi> are the wave functions of the pz

orbitals of the carbon atoms A and B respectively. To solve for φA and φB using hard-wall boundary conditions, we get,

φA(0) = φB(0) = 0,

φA(n + 1) = φB(n + 1) = 0. (4.3)

√ 3qya Choosing φA(i) = φB(i) = sin( 2 i) and substituting in (4.3), we get,

2 pπ qy = √ , p = 1, 2, ··· , n. (4.4) 3a n + 1

where, qy is the discretized wave vector in the y direction and a = 1.42 Å. A nor- q NX (n+1) malization condition is used to ﬁnd NA and NB as 2 , where Nx is the total number of atoms in x-direction which is not quantized. The wave function changes from (2.14) and (4.2) to,

50 CHAPTER 4. EDGE EFFECT OF GRAPHENE & LAYERS EFFECT ON MOS2

"s n √ # 2 3q a X X ikxxA y |ψ > = CA e i sin ( i)|Ai > (4.5) Nx(n + 1) 2 i=1 xAi "s n √ # 2 3q a X X ikxxB y +CB e i sin ( i)|Bi > (4.6) Nx(n + 1) 2 i=1 xBi

Using Tight Binding, the Hamiltonian is given by,

X X H = i|i >< i| − ti,j(|i >< j|), (4.7) i where, < i, j > denotes nearest neighbors.

Here, in a perfect A-GNR, ti,j = t= Vppπ and i = = π. Back substituting in the original Schrödinger equation, we get,

E = ± |µ|, (4.8) √

ikx 3a −ikxa µ = −t[2e a/2 cos ( qy) + e ] (4.9) 2

Here, ± is for the conduction/valence bands respectively. The limits for kx is same

π 3kxa π as that of Graphene from previous sections : − 2 ≤ 2 ≤ 2 . So, kx varies from −π/3a to π/3a.

51 CHAPTER 4. EDGE EFFECT OF GRAPHENE & LAYERS EFFECT ON MOS2

Figure 4.2: Energy Dispersion for A-GNR with width n=6

Figure 4.3: Energy Dispersion for A-GNR with width n=7

52 CHAPTER 4. EDGE EFFECT OF GRAPHENE & LAYERS EFFECT ON MOS2

Figure 4.4: Energy Dispersion for A-GNR with width n=8

Figure 4.5: Energy Dispersion for A-GNR with width n=22

Observing Fig. 4.2 - Fig. 4.5, it can be noted that A-GNR is metallic for a width of n=8 atoms and insulating for all other lengths. These simulated results match with

53 CHAPTER 4. EDGE EFFECT OF GRAPHENE & LAYERS EFFECT ON MOS2

existing energy band relations of A-GNR - [23] [4] [?] [24] [22].

4.1.2 Z-GNR

For a Z-GNR of length ’N’ atoms, it is observed that ’N/2’ atoms of type A, alternate in their placement with ’N/2’ atoms of type B. Periodic boundary conditions are imposed in a direction which is parallel to the edge of the nanostructure. Electronic confinements are created near the Dirac points. With similar tight binding calculations as the A-GNR, the linear dispersion relation can be predicted to be linear near |k| = 2π/3. This occurs due to the edge states which appear as special modes with partially flat bands. Under special conditions, they lead to separate modes in the two valleys. Considering nearest neighbour interactions and LCAO theory,we can attempt to understand the parrtial flat bands/ degeneracy with charge density distribution. The amplitude of the charge density is assumed to be proportional to radius. Wak- abayashi et al. [24], solve analytically the edge states for Z-GNR by first setting up the Bloch components of the linear combination of atomic orbitals at each edge state as exp(ik(n−1)), exp(ikn) and exp(ik(n+1)). For the wave function at each site to be exact at E = 0, the total sum of the components of each wave function should cancel each other out. The condition for this to occur is,

eik(n+1) + eikn + x = 0, eikn + eik(n−1) + y = 0, x + y + z = 0 (4.10)

This leads the wave components in the x, y and z directions as [−2 cos(k/2)]e(ik(n+1/2)), [−2 cos(k/2)]e(ik(n−1/2)) and [−2 cos(k/2)]2e(ikn) respectively [25]. The imbalance at the zigzag edges leads to the following boundary conditions [25],

ΦK±A(r[0]) = 0, ΦK±B(r[N+1]) = 0 (4.11)

54 CHAPTER 4. EDGE EFFECT OF GRAPHENE & LAYERS EFFECT ON MOS2 where, ’r’, stands for the co-ordinate of the atom on the chain. The eigen values for energy is obtained by solving the below equation,

= ±(η − k)eηW (4.12) √ where, η = 2 k2 − [25]. In terms of constructing the Hamiltonian, the nearest neighbor matrix elements have the eigen value, 2 cos(ka/2), the next nearest neighbor interaction is always one and the remaining have eigen value zero. The size of the Hamiltonian matrix is always double the width of the nanoribbon width - irrespective of the edge orientation (armchair or zigzag). Sample Hamiltonian matrix for a Z-GNR of width 2 atoms, is given by,   0 2 cos(ak/2) 0 0     2 cos(ak/2) 0 2 cos(ak/2) 0    H =    0 2 cos(ak/2) 0 2 cos(ak/2)     0 0 2 cos(ak/2) 0

Diagonalizing the above matrix, gives us the energy dispersion relation for Z-GNRs. Fig. 4.6 - Fig. 4.8 below, shows the same for Z-GNRs of varying widths.

55 CHAPTER 4. EDGE EFFECT OF GRAPHENE & LAYERS EFFECT ON MOS2

Figure 4.6: Energy Dispersion for Z-GNR with width n=5

Figure 4.7: Energy Dispersion for Z-GNR with width n=16

56 CHAPTER 4. EDGE EFFECT OF GRAPHENE & LAYERS EFFECT ON MOS2

Figure 4.8: Energy Dispersion for Z-GNR with width n=30

It can be observed that the highest valence band and lowest conduction bands are always degenerate at |k| = π, irrespective of the length. Also, the degeneracy, at these two points, gets flatter as the width increases. A consistent pair of flat bands is observed in the interval of 2π/3 ≤ |k| ≤ π [23]. The zigzag ribbon exhibits a special edge state. The energy states are flat at fermi energy and give sharp peaks in density of states as the charge density is strongly localized on the edge states. The same is not applicable for armchair GNRs. Varying the width of the GNR is observed to have an important effect on the edge state.

4.2 Monolayer/Bilayer eﬀect in MoS2

Similar to the edge eﬀect observed in Graphene, MoS2 also exhibits band gap variation depending on the number of layers. The highest point in conduction band

maximum shift from monolayer MoS2 ’K’ point to bilayer MoS2 ’Gamma’ point. When

bulk MoS2 is considered, the valence band minimum shifts from ’K’ point to a point

57 CHAPTER 4. EDGE EFFECT OF GRAPHENE & LAYERS EFFECT ON MOS2

between ’K’ and ’Gamma’ point. [1] simulates and compares tight binding calculations to DFT calculations and ﬁnds that the results are in good agreement. Fig, 4.9b, depicts the overlap matrix element strength between the layers. The gradient from red to white depicts the transition from the strongest to the weakest element.The overlapping matrix elements represent the splitting of single-layer states. The overlap matrix was till now considered zero for simplicity of calculations. However, as the size of the nano-structures increase in dimension, so does their level of complexity in terms of band gap engineering. A number of assumptions are made to cause least number of changes to the Hamilto-

nian of monolayer MoS2 and obtain bilayer and bulk MoS2 energy band relations. [1] decides on a gauge that makes it easier to form the Hamiltonian matrix for each of

the monolayers and then combines for a resulting bulk MoS2 structure, shown in Fig. 4.9c.

Figure 4.9: Energy Dispersion for Bilayer ((a) & (b)) and Bulk MoS2 [1]

58 Chapter 5 Summary & Future Work

5.1 Summary

In this research, we explained and implemented an ab-initio tight binding theory based process in an attempt to understand bandgap engineering in 2-D semiconductors. Ignoring the overlapping integrals in the case of Graphene and GNRs, simpliﬁed the Hamiltonian matrix and the calculations involved. As we moved from sp2 to sp3d

hybridized structures, (or moved to MoS2), the Hamiltonian matrix increased complex- ities in the calculations. Solving the Hamiltonian matrices to obtain energy dispersion relation gave us an understanding of the factors involving band-gap tunability. An Armchair Graphene Nano-Ribbon exhibits both metallic and insulating properties, depending on its width. Whereas, Zigzag Graphene Nano-Ribbon is always metallic irrespective of its width.

Monolayer MoS2 changes from a direct band gap material to an indirect band gap

material for Bilayer MoS2. As we transition towards bulk MoS2, the minimum and maximum points of the valence and conduction bands respectively vary - changing the electrical and quantum properties of the material. All results are found to be in good agreement with previous research. Diﬀerent methods and approximations have been discussed throughout this work. Another method of performing calculations is the Full Tight Binding Hamiltonian

59 CHAPTER 5. SUMMARY & FUTURE WORK

(FTBH) theory and expand on the k.p theory at the dirac points. Irrespective of the method used, the results are found to be fairly consistent across the diﬀerent theories. MATLAB Code for all the simulations and the Graphical User Interface (GUI) is provided in the appendices.

5.2 Future Work

The calculations shown in this research, can be expanded to accommodate new theories where d-orbital is believed to play a crucial role in the splitting of bands at the high-symmetry points [26]. Current ﬁrst-principle calculations reveal a gap as small as 1µeV . Whereas, when p-orbitals are considered, this gap is predicted to be

as high as 25µeV − 50µeV . This is observed because the pz orbitals do not hybridize with the px and py orbitals but, in fact hybridize with the dxz and dyz orbitals to form the π bands. These d-orbitals split due to spin-orbit coupling and the gap of the π band is found to be linearly dependent on the splitting of the d-orbitals. Other possible future extensions to the current work includes the effect of extrinsic spin-orbit coupling. Graphene loses its property to be doubly degenerate when placed on a substrate or under an electric field (for e.g., gate bias) and this increases the splitting of bands. This condition is known as extrinsic spin orbit coupling. In the case of extrinsic spin-orbit coupling, d-orbitals give negligible contribution to this effect. Studies [26] show that for an electric field of 1V/nm, the extrinsic gap created in 10µeV and the relation between the two is found to be linear. Splittings greater than 1meV occur generally due to charge transfer between the substrate and graphene. Energy dispersion relations including such effects can be solved using perturbation theory. The d-orbitals are found to make a bigger impact in the density of states calculations.

60 Appendix A: MATLAB Code for Graphene

A.1 Case 1 - π bands

clear;figure; a=1.42; otherA=0; V_ss_sigma=-5; V_sp_sigma=5.4; V_pp_sigma=8.4; V_pp_pi=-2.4; Es=-19.38; Ep=-11.07;

E_pi = (otherA^2 * Es) + ((1-otherA^2)*Ep); E_sigma=((1-otherA^2)*Es/3) + ((2+otherA^2)*Ep/3); V_intra=(1-otherA^2)*(Es-Ep)/3; V_sigma=(-2/3*V_pp_sigma) + ((1-otherA^2)*V_ss_sigma/3) - (2*sqrt... (2*(1-otherA^2))*V_sp_sigma/3); d1=[1 0]*a;d2 = [-1 sqrt(3)]*a/2; d3=[-1 -sqrt(3)]*a/2;

%%%%% Gamma-M Direction l=1;m=0;kmax=2*pi/3/a;Nt=101; for Nk=1:Nt

k=[l m]*kmax*(Nk-1)/(Nt-1); p1 = exp(1i*sum(k.*d1));p2 = exp(1i*sum(k.*d2));p3 = exp(1i*sum(k.*d3... )); gamma_k = (p1+p2+p3); h=[E_pi V_pp_pi*gamma_k 0 0 0 0 0 0; V_pp_pi*conj(gamma_k) E_pi 0 0 0 0 0 0; 0 0 E_sigma V_sigma*p1 V_intra 0 V_intra 0; 0 0 V_sigma*conj(p1) E_sigma 0 V_intra 0 V_intra; 0 0 V_intra 0 E_sigma V_sigma*p2 V_intra 0; 0 0 0 V_intra V_sigma*conj(p2) E_sigma 0 V_intra; 0 0 V_intra 0 V_intra 0 E_sigma V_sigma*p3; 0 0 0 V_intra 0 V_intra V_sigma*conj(p3) E_sigma];

61 APPENDIX A. MATLAB CODE FOR GRAPHENE

[V,D]=eig(h); eigst = sum(D); E(Nk,:)=sort(real(eigst)); X1(Nk) = (Nk-1); end hold on; plot(X1,E/27.21,'b');

%%%%%% Gamma-K(-ve) Direction l=1;m=1/sqrt(3);kmax=2*pi/3/a; for Nk=1:Nt

%%%%%E-k on basis of Les Houches and Zheng+ Brey& Fertig kx=2*pi/(3)/a; ky=linspace(0,2*pi/3/sqrt(3),101)/a;

62 APPENDIX A. MATLAB CODE FOR GRAPHENE

Z=abs(2*exp(1i*a.*kx/2).*cos(sqrt(3)*a.*ky/2) + exp(-1i*a.*kx)); E_pi_plus=E_pi+V_pp_pi*(Z); E_pi_minus=E_pi-V_pp_pi*(Z); E_sigma1_plus=E_sigma-V_intra+V_sigma; E_sigma1_minus=E_sigma-V_intra-V_sigma;

E_sigma2_plus=E_sigma+V_intra/2+sqrt((3*V_intra/2)^2+V_sigma^2+... V_intra*V_sigma*(Z)); E_sigma2_minus=E_sigma+V_intra/2+sqrt((3*V_intra/2)^2+V_sigma^2-... V_intra*V_sigma*(Z)); E_sigma3_plus=E_sigma+V_intra/2-sqrt((3*V_intra/2)^2+V_sigma^2+... V_intra*V_sigma*(Z)); E_sigma3_minus=E_sigma+V_intra/2-sqrt((3*V_intra/2)^2+V_sigma^2-... V_intra*V_sigma*(Z)); k=linspace(100,200,101); plot(k,E_pi_plus/27.21,'g'); plot(k,E_pi_minus/27.21,'g'); line([k(1) k(101)],[E_sigma1_plus/27.21 E_sigma1_plus/27.21],'color',... 'g'); line([k(1) k(101)],[E_sigma1_minus/27.21 E_sigma1_minus/27.21],'color... ','g'); plot(k,E_sigma2_plus/27.21,'g'); plot(k,E_sigma2_minus/27.21,'g'); plot(k,E_sigma3_plus/27.21,'g'); plot(k,E_sigma3_minus/27.21,'g');

% Line Demarcations M=X1(Nt); line([0 0],[X1(1)+0.2 E(1)/27.21-0.2],'Color','k'); line([M M],[X1(1)+0.2 E(1)/27.21-0.2],'Color','g');

% set xlabel ticks lowerlim=-X1(101);higherlim=k(101); set(gca,'XTick',[lowerlim,0,M,higherlim]);

63 APPENDIX A. MATLAB CODE FOR GRAPHENE

set(gca,'XTickLabel',{'K','\Gamma','M','K'},'Fontsize',12); grid minor; xlabel('Wave Vector','Fontsize',14); ylabel('Energy in units of Hartree','Fontsize',14); title('E-K','Fontsize',14);

64 APPENDIX A. MATLAB CODE FOR GRAPHENE

A.2 Case 2 - π & σ bands

clear;figure; a=1.42; otherA=0; V_ss_sigma=-5; V_sp_sigma=5.4; V_pp_sigma=8.4; V_pp_pi=-2.4; Es=-19.38; Ep=-11.07;

%%%%% Gamma-M Direction l=1;m=0;kmax=2*pi/3/a;Nt=101; for Nk=1:Nt

65 APPENDIX A. MATLAB CODE FOR GRAPHENE

[V,D]=eig(h); eigst = sum(D); E(Nk,:)=sort(real(eigst));

X1(Nk) = (Nk-1); end hold on; plot(X1,E/27.21,'b');

%%%%%% Gamma-K(-ve) Direction l=1;m=1/sqrt(3);kmax=2*pi/3/a; for Nk=1:Nt

X2(Nk) = -(Nk-1);

66 APPENDIX A. MATLAB CODE FOR GRAPHENE

end plot(X2,E/27.21,'r');

%M-K Direction %%%%%E-k on basis of Les Houches and Zheng+ Brey& Fertig kx=2*pi/(3)/a; ky=linspace(0,2*pi/3/sqrt(3),101)/a; Z=abs(2*exp(1i*a.*kx/2).*cos(sqrt(3)*a.*ky/2) + exp(-1i*a.*kx)); E_pi_plus=E_pi+V_pp_pi*(Z); E_pi_minus=E_pi-V_pp_pi*(Z); E_sigma1_plus=E_sigma-V_intra+V_sigma; E_sigma1_minus=E_sigma-V_intra-V_sigma;

67 APPENDIX A. MATLAB CODE FOR GRAPHENE

% Line Demarcations M=X1(Nt); line([0 0],[X1(1)+0.2 E(1)/27.21-0.2],'Color','k'); line([M M],[X1(1)+0.2 E(1)/27.21-0.2],'Color','g');

% set xlabel ticks lowerlim=-X1(101);higherlim=k(101); set(gca,'XTick',[lowerlim,0,M,higherlim]); set(gca,'XTickLabel',{'K','\Gamma','M','K'},'Fontsize',12); grid minor; xlabel('Wave Vector','Fontsize',14); ylabel('Energy in units of Hartree','Fontsize',14); title('E-K','Fontsize',14);

68 Appendix B: MATLAB Code for Armchair GNR

clear;figure; a=1.42; t=1; %t=2.7; eps = 0; n=5; plot_colour=['r','g','b','m','k','c','y','r-','g-','b-','m-','k-','y-... ','c-','r--','g--','b--','m--','k--','y--','c--','or','og','ob','...

om','ok','oy','oc','r*','g*','b*','m*','k*','y*','c*']; for p=1:n kx=linspace(-pi/3,pi/3,100)/a; q=2.*p*pi/sqrt(3)/a/(n+1);

Z=(2*exp(1i.*kx*a/2).*cos(sqrt(3)*a.*q/2))+(exp(-1i.*kx*a)); E=eps+abs(t*Z); k=kx/pi; hold on; plot(k,E,plot_colour(p),'LineWidth',2) plot(k,-E,plot_colour(p),'LineWidth',2); end set(gca,'XTick',[-k(100),0,k(100)]); set(gca,'XTickLabel',{'-\pi','0','\pi'});

69 APPENDIX B. MATLAB CODE FOR ARMCHAIR GNR

ylim([-3 3]); xlabel('Wave Vector k_x/3a'); ylabel('Energy/t'); str=sprintf('E-K Armchair NanoRibbons(n=%d)',n); title(str);

70 Appendix C: MATLAB Code for Zigzag GNR

clear; a=sqrt(3)*1.42; t=2.7;

% Length ofA-GNR ==>N N=30; k=linspace(-pi,pi,1000);

%%%% Hamiltonian generation

H=ones(2*N,2*N); x_atom=1; y_atom=1; l=1;m=0;kmax=pi/a; for Nk=1:N

k=[l m]*kmax*(Nk-1)/(N-1);

%%% code for generating hamiltonian starts here---> for i=1:2*N

for j=1:2*N

if(x_atom==y_atom) if(i==j) H(i,j)=0; elseif ((j-i)==1)

H(i,j)=2*cos(sum(k.*a/2)); elseif ((j-i)==-1)

H(i,j)=2*cos(sum(k.*a/2)); else

71 APPENDIX C. MATLAB CODE FOR ZIGZAG GNR

H(i,j)=0; end elseif(abs(x_atom-y_atom)==1) if(abs(i-j)==1) H(i,j)=1; else H(i,j)=0; end elseif(abs(x_atom-y_atom)>1) H(i,j)=0; end if(mod(j,2)==0) y_atom=y_atom+1; if(y_atom>N) y_atom=1; end end

end if(mod(i,2)==0) x_atom=x_atom+1; if(x_atom>N) x_atom=1; end end end

H=t*H; %%% <--- code for generating hamiltonian end here

[V,D]=eig(H); eigst = sum(D); E(Nk,:)=sort(real(eigst));

72 APPENDIX C. MATLAB CODE FOR ZIGZAG GNR

X1(Nk) = (Nk-1); end figure; hold on; plot(X1,E,'b'); plot(-X1,E,'b');

D=2*X1(Nk)/3; line([D D],[10 -10]); line([-D -D],[10 -10]);

%% set xlabel ticks set(gca,'XTick',[-X1(Nk),-D,0,D,X1(Nk)]); set(gca,'XTickLabel',{'-\pi','-2\pi/3','0','2\pi/3','\pi'},'Fontsize'... ,12);

xlabel('Wave Vector k_x'); ylabel('Energy/t'); str=sprintf('E-K ZigZag NanoRibbons(n=%d)',N); title(str);

73 Appendix D: Tight Binding simpliﬁcations and independent parameters for MoS2

By performing symmetry operations, the FTBH has 86 parameters that can be simpliﬁed and reduced to 36 parameters. The resulting relations are described below [1],

1. For indexes : (α = 1, β = 2); (α = 4, β = 5, γ = 3); (α = 7, β = 8, γ = 6); (α = 10, β = 11, γ = 9),

α = β (D.1)

1 3 (1) t(2) = t(1) + t (D.2) α,α 4 α,α 4 β,β (2) 3 1 (1) t = t(1) + t (D.3) β,β 4 α,α 4 β,β (2) (1) tγ,γ = tγ,γ (D.4) √ 2 (2,3) 3 (1) 1 (1) tγ,β = ± tγ,α + tγ,β (D.5) √2 2 2 (2,3) 3 (1) (1) (1) tα,β = ± (tα,α − tβ,β) − tα,β (D.6) 4 √ 2 1 3 (1) t(2,3) = t(1) ± t (D.7) γ,α 2 γ,α 2 γ,β

2. For indexes : (α = 1, β = 2, α0 = 4, β0 = 3, γ0 = 3); (α = 7, β = 88, α0 = 10, β0 = 11, γ0 = 9),

74 APPENDIX D. TIGHT BINDING SIMPLIFICATIONS AND INDEPENDENT PARAMETERS FOR MOS2

(4) 1 (5) 3 (5) t 0 = t 0 + t 0 (D.8) α ,α 4 α ,α 4 β ,β (4) 3 (5) 1 (5) tβ0,β = tα0,α + tβ0,β (D.9) 4 4√ √ 2 2 (4) (4) 3 (5) 3 (5) tβ0,α = tα0,β = − tα0,α + tβ0,β (D.10) √ 4 4 2 (4) 3 (5) t 0 = − t 0 (D.11) γ ,α 2 γ ,β (4) 1 (5) t 0 = − t 0 (D.12) γ ,β 2 γ ,β (4) (5) t9,6 = t9,6 (D.13) √ 2 (4) 3 (5) t = t (D.14) 10,6 2 11,6 (4) 1 (5) t = − t (D.15) 11,6 2 11,6

The independent parameter values calculated on the basis of DFT is given in table D.1 below [1].

Tight Binding Parameter Value (eV)

1 = 2 1.0688

3 -0.7755

4 = 5 -1.2902

6 -0.138

7 = 8 0.0874

9 -2.8949

10 = 11 -1.9065 (1) t1,1 -0.2069 (1) t2,2 0.0323 (1) t3,3 -0.1739

...Continued on next page

75 APPENDIX D. TIGHT BINDING SIMPLIFICATIONS AND INDEPENDENT PARAMETERS FOR MOS2

Table D.1 – continued from previous page

First column Second column

(1) t4,4 0.8651 (1) t5,5 -0.1872 (1) t6,6 -0.2979 (1) t7,7 0.2747 (1) t8,8 -0.5581 (1) t9,9 -0.1916 (1) t10,10 0.9122 (1) t11,11 0.0059 (1) t3,5 -0.0679 (1) t6,8 0.4096 (1) t9,11 0.0075 (1) t1,2 -0.2562 (1) t3,4 -0.0995 (1) t4,5 -0.0705 (1) t6,7 -0.1145 (1) t7,8 -0.2487 (1) t9,10 0.1063 (1) t10,11 -0.0385 (5) t4,1 -0.7883 (5) t3,2 -1.3790 (5) t5,2 2.1584 (5) t9,6 -0.8836 (5) t11,6 -0.9402

...Continued on next page

76 APPENDIX D. TIGHT BINDING SIMPLIFICATIONS AND INDEPENDENT PARAMETERS FOR MOS2

Table D.1 – continued from previous page

First column Second column

(5) t10,7 1.4114 (5) t9,8 -0.9535 (5) t11,8 0.6517 (6) t9,6 -0.0686 (6) t11,6 -0.1498 (6) t9,8 -0.2205 (6) t11,8 -0.2451

Table D.1: Independent Tight Binding Parameters for MoS2 [1]

77 Appendix E: MATLAB Code for monolayer MoS2

clear all;figure; %%% Values taken from Ab-initio Tight-Binding Hamiltonian for ... Transition Metal Dichalcogenides % Shiang Fang,1 Rodrick Kuate Defo,1 SharmilaN. Shirodkar,2 % Simon Lieu,1 GeorgiosA. Tritsaris,2 and Efthimios Kaxiras1,2 %1Department of Physics, Harvard University, Cambridge, ... Massachusetts 02138, USA. %2School of Engineering and Applied Sciences, Harvard University, ... Cambridge, Massachusetts 02138, USA. %(Dated: September 10, 2015) %%%%K, Gamma,M values taken from Three-band tight-binding model... for monolayers of group-VIB transition metal dichalcogenides

% Gui-Bin Liu,1,2 Wen-Yu Shan,3 Yugui Yao,1 Wang Yao,2,* and Di Xiao3... ,ï£¡ %1School of Physics, Beijing Institute of Technology, Beijing ... 100081, China %2Department of Physics and Center of Theoretical and Computational ... Physics, The University of Hong Kong, Hong Kong, China %3Department of Physics, Carnegie Mellon University, Pittsburgh, ... Pennsylvania 15213, USA %(Received 28 May 2013; revised manuscript received 20 July 2013; ... published 26 August 2013) %% values a = 3.18; a1 = a*[1 0]; a2 = a*[-1/2 sqrt(3)/2]; d1 = a1; d2 = a1+a2; d3 = a2; d4 = -((2*a1)+a2)/3; d5 = (a1+(2*a2))/3; d6 = (a1-a2)/3; d7 = -2*(a1+(2*a2))/3; d8 = 2*((2*a1)+a2)/3; d9 = 2*(a2-a1)/3; t1 = zeros; t2 = zeros; t5 = zeros; t6 = zeros; t1(1,1) = -0.2069;

78 APPENDIX E. MATLAB CODE FOR MONOLAYER MOS2

t1(2,2) = 0.0323; t1(3,3) = -0.1739; t1(4,4) = 0.8651; t1(5,5) = -0.1872; t1(6,6) = -0.2979; t1(7,7) = 0.2747; t1(8,8) = -0.5581; t1(9,9) = -0.1916; t1(10,10) = 0.9122; t1(11,11) = 0.0059; t1(3,5) = -0.0679; t1(6,8) = 0.4069; t1(9,11) = 0.0075; t1(1,2) = -0.2562; t1(3,4) = -0.0995; t1(4,5) = -0.0705; t1(6,7) = -0.1145; t1(7,8) = -0.2487; t1(9,10) = 0.1063; t1(10,11) = -0.0385; t5(4,1) = -0.7883; t5(3,2) = -1.3790; t5(5,2) = 2.1584; t5(9,6) = -0.8836; t5(11,6) = -0.9402; t5(10,7) = 1.4114; t5(9,8) = -0.9535; t5(11,8) = 0.6517; t6(9,6) = -0.0686; t6(11,6) = -0.1498; t6(9,8) = -0.2205; t6(11,8) = -0.2451; eps = zeros;

79 APPENDIX E. MATLAB CODE FOR MONOLAYER MOS2

eps(1) = 1.0688; eps(2) = 1.0688; eps(3) = -0.7755; eps(4) = -1.2902; eps(5) = -1.2902; eps(6) = -0.1380; eps(7) = 0.0874; eps(8) = 0.0874; eps(9) = -2.8949; eps(10) = -1.9065; eps(11) = -1.9065; %% for set alpha= 1, beta=2 t2(1,1) = t1(1,1)/4 + 3*t1(2,2)/4; t2(2,2) = 3*t1(1,1)/4 + t1(2,2)/4; t2(1,2) = sqrt(3)*(t1(1,1) - t1(2,2))/4 - t1(1,2); t3(1,2) = -sqrt(3)*(t1(1,1) - t1(2,2))/4 - t1(1,2); %% for set alpha= 4, beta= 5, gamma=3 t2(4,4) = t1(4,4)/4 + 3*t1(5,5)/4; t2(5,5) = 3*t1(4,4)/4 + t1(5,5)/4; t2(3,3) = t1(3,3); t2(3,5) = sqrt(3)*t1(3,4)/2 - t1(3,5)/2; t3(3,5) = -sqrt(3)*t1(3,4)/2 - t1(3,5)/2; t2(4,5) = sqrt(3)*(t1(4,4) - t1(5,5))/4 - t1(4,5); t3(4,5) = -sqrt(3)*(t1(4,4) - t1(5,5))/4 - t1(4,5); t2(3,4) = t1(3,4)/2 + sqrt(3)*t1(3,5)/2; t3(3,4) = t1(3,4)/2 - sqrt(3)*t1(3,5)/2; %% for set alpha= 7, beta= 8, gamma=6 t2(7,7) = t1(7,7)/4 + 3*t1(8,8)/4; t2(8,8) = 3*t1(7,7)/4 + t1(8,8)/4; t2(6,6) = t1(6,6); t2(6,8) = sqrt(3)*t1(6,7)/2 - t1(6,8)/2; t3(6,8) = -sqrt(3)*t1(6,7)/2 - t1(6,8)/2; t2(7,8) = sqrt(3)*(t1(7,7) - t1(8,8))/4 - t1(7,8); t3(7,8) = -sqrt(3)*(t1(7,7) - t1(8,8))/4 - t1(7,8); t2(6,7) = t1(6,7)/2 + sqrt(3)*t1(6,8)/2; t3(6,7) = t1(6,7)/2 - sqrt(3)*t1(6,8)/2;

80 APPENDIX E. MATLAB CODE FOR MONOLAYER MOS2

%% for set alpha= 10, beta= 11, gamma=9 t2(10,10) = t1(10,10)/4 + 3*t1(11,11)/4; t2(11,11) = 3*t1(10,10)/4 + t1(11,11)/4; t2(9,9) = t1(9,9); t2(9,11) = sqrt(3)*t1(9,10)/2 - t1(9,11)/2; t3(9,11) = -sqrt(3)*t1(9,10)/2 - t1(9,11)/2; t2(10,11) = sqrt(3)*(t1(10,10) - t1(11,11))/4 - t1(10,11); t3(10,11) = -sqrt(3)*(t1(10,10) - t1(11,11))/4 - t1(10,11); t2(9,10) = t1(9,10)/2 + sqrt(3)*t1(9,11)/2; t3(9,10) = t1(9,10)/2 - sqrt(3)*t1(9,11)/2; %% for set alpha= 1, beta= 2, alpha'= 4, beta'= 5, gamma'=3 t4(4,1) = t5(4,1)/4 + 3*t5(5,2)/4; t4(5,2) = 3*t5(4,1)/4 + t5(5,2)/4; t4(5,1) = -sqrt(3)*t5(4,1)/4 + sqrt(3)*t5(5,2)/4; t4(4,2) = t4(5,1); t4(3,1) = -sqrt(3)*t5(3,2)/2; t4(3,2) = -t5(3,2)/2; t4(9,6) = t5(9,6); t4(10,6) = -sqrt(3)*t5(11,6)/2; t4(11,6) = -t5(11,6)/2; %% for set alpha= 7, beta= 8, alpha'= 10, beta'= 11, gamma'=9 t4(10,7) = t5(10,7)/4 + 3*t5(11,8)/4; t4(11,8) = 3*t5(10,7)/4 + t5(11,8)/4; t4(11,7) = -sqrt(3)*t5(10,7)/4 + sqrt(3)*t5(11,8)/4; t4(10,8) = t4(11,7); t4(9,7) = -sqrt(3)*t5(9,8)/2; t4(9,8) = -t5(9,8)/2; %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% part ... done H1 = zeros(11,11);H2 = zeros(11,11); %%%%% gamma-M Direction %l=1;m=1/sqrt(3);kmax=pi/a; Nt=101; kx = linspace(0,pi/a,Nt); ky = linspace(0,pi/sqrt(3)/a,Nt);

81 APPENDIX E. MATLAB CODE FOR MONOLAYER MOS2

GM = [0,0;4*pi/3/a,0]; dist_GM = pdist(GM,'euclidean'); X1 = linspace(0,dist_GM,Nt); for Nk=1:Nt k=[kx(1,Nk) ky(1,Nk)]; %%% Forming the Hamiltonian for i = 1:11 for j = 1:11 if(i == j)

H1(i,j) = eps(i) + (2*t1(i,i)*cos(sum(k.*d1))) + (2*t2(i,... i)*(cos(sum(k.*d2)) + cos(sum(k.*d3))) ); end %%+ symmetry if(((i==3) && (j==5)) || ((i==6) && (j==8)) || ((i==9) && (j... ==11)))

H1(i,j) = 2*t1(i,j)*cos(sum(k.*d1)) + t2(i,j)*(exp(sum(-1... i*k.*d2)) + exp(sum(-1i*k.*d3))) + t3(i,j)*(exp(sum(1... i*k.*d2)) + exp(sum(1i*k.*d3))); H1(j,i) = conj(H1(i,j)); end %%- symmetry if(((i==1) && (j==2)) || ((i==3) && (j==4)) || ((i==4) && (j... ==5)) || ((i==6) && (j==7)) || ((i==7) && (j==8)) || ((i... ==9) && (j==10)) || ((i==10) && (j==11)))

H1(i,j) = -2i*t1(i,j)*sin(sum(k.*d1)) + t2(i,j)*(exp(sum... (-1i*k.*d2)) - exp(sum(-1i*k.*d3))) + t3(i,j)*(-exp(... sum(1i*k.*d2)) + exp(sum(1i*k.*d3))); H1(j,i) = conj(H1(i,j)); end %% dont know first neighbourX pairs+ symm if(((i==3) && (j==1)) || ((i==5) && (j==1)) || ((i==4) && (j... ==2)) || ((i==10) && (j==6)) || ((i==9) && (j==7)) || ((i... ==11) && (j==7)) || ((i==10) && (j==8)) )

82 APPENDIX E. MATLAB CODE FOR MONOLAYER MOS2

H1(i,j) = t4(i,j)*(exp(sum(1i*k.*d4)) - exp(sum(1i*k.*d6)... )) ; H1(j,i) = conj(H1(i,j)); end %% dont know first neighbourX pairs- symm if(((i==4) && (j==1)) || ((i==3) && (j==2)) || ((i==5) && (j... ==2)) || ((i==9) && (j==6)) || ((i==11) && (j==6)) || ((i... ==10) && (j==7)) || ((i==9) && (j==8)) || ((i==11) && (j... ==8)))

H1(i,j) = t4(i,j)*(exp(sum(1i*k.*d4)) + exp(sum(1i*k.*d6)... )) + t5(i,j)*exp(sum(1i*k.*d5)); H1(j,i) = conj(H1(i,j)); end

H2(9,6) = t6(9,6)*(exp(sum(1i*k.*d7)) + exp(sum(1i*k.*d8)) + ... exp(sum(1i*k.*d9))); H2(11,6) = t6(11,6)*(exp(sum(1i*k.*d7)) - exp(sum(1i*k.*d8))... /2 - exp(sum(1i*k.*d9))/2); H2(10,6) = sqrt(3)*t6(11,6)*(-exp(sum(1i*k.*d8)) + exp(sum(1i... *k.*d9)))/2; H2(9,8) = t6(9,8)*(exp(sum(1i*k.*d7)) - exp(sum(1i*k.*d8))/2 ... - exp(sum(1i*k.*d9))/2 ); H2(9,7) = sqrt(3)*t6(9,8)*(-exp(sum(1i*k.*d8)) + exp(sum(1i*... k.*d9)))/2; H2(10,7) = 3*t6(11,8)*(exp(sum(1i*k.*d8)) + exp(sum(1i*k.*d9)... ))/4;

H2(11,7) = sqrt(3)*t6(11,8)*(exp(sum(1i*k.*d8)) - exp(sum(1i*... k.*d9)))/4; H2(10,8) = H2(11,7);

83 APPENDIX E. MATLAB CODE FOR MONOLAYER MOS2

=H2(10,8); H2(8,11)=H2(11,8); end end [V,D]=eig(H1+H2); eigst = sum(D); E1(Nk,:)=sort(real(eigst)); end plot(X1,E1,'b'); hold on; %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%... %% % %%%%%%%M-K Direction Nt=101; kx = linspace(pi/a,4*pi/3/a,Nt); ky = linspace(0,pi/sqrt(3)/a,Nt); temp=Nt;

MK = [pi/a,0;4*pi/3/a,pi/sqrt(3)/a]; dist_MK = pdist(MK,'euclidean'); X2 = linspace(0,dist_MK,Nt); for Nk=1:Nt k=[kx(1,Nk) ky(1,temp)]; %%% Forming the Hamiltonian for i = 1:11 for j = 1:11 if(i == j)

H1(i,j) = eps(i) + (2*t1(i,i)*cos(sum(k.*d1))) + (2*t2(i,... i)*(cos(sum(k.*d2)) + cos(sum(k.*d3))) ); end %%+ symmetry if(((i==3) && (j==5)) || ((i==6) && (j==8)) || ((i==9) && (j... ==11)))

H1(i,j) = 2*t1(i,j)*cos(sum(k.*d1)) + t2(i,j)*(exp(sum(-1...

84 APPENDIX E. MATLAB CODE FOR MONOLAYER MOS2

i*k.*d2)) + exp(sum(-1i*k.*d3))) + t3(i,j)*(exp(sum(1... i*k.*d2)) + exp(sum(1i*k.*d3))); H1(j,i) = conj(H1(i,j)); end %%- symmetry if(((i==1) && (j==2)) || ((i==3) && (j==4)) || ((i==4) && (j... ==5)) || ((i==6) && (j==7)) || ((i==7) && (j==8)) || ((i... ==9) && (j==10)) || ((i==10) && (j==11)))

H1(i,j) = t4(i,j)*(exp(sum(1i*k.*d4)) + exp(sum(1i*k.*d6)... )) + t5(i,j)*exp(sum(1i*k.*d5)); H1(j,i) = conj(H1(i,j)); end

H2(9,6) = t6(9,6)*(exp(sum(1i*k.*d7)) + exp(sum(1i*k.*d8)) + ... exp(sum(1i*k.*d9))); H2(11,6) = t6(11,6)*(exp(sum(1i*k.*d7)) - exp(sum(1i*k.*d8))...

85 APPENDIX E. MATLAB CODE FOR MONOLAYER MOS2

/2 - exp(sum(1i*k.*d9))/2); H2(10,6) = sqrt(3)*t6(11,6)*(-exp(sum(1i*k.*d8)) + exp(sum(1i... *k.*d9)))/2; H2(9,8) = t6(9,8)*(exp(sum(1i*k.*d7)) - exp(sum(1i*k.*d8))/2 ... - exp(sum(1i*k.*d9))/2 ); H2(9,7) = sqrt(3)*t6(9,8)*(-exp(sum(1i*k.*d8)) + exp(sum(1i*... k.*d9)))/2; H2(10,7) = 3*t6(11,8)*(exp(sum(1i*k.*d8)) + exp(sum(1i*k.*d9)... ))/4;

H2(11,7) = sqrt(3)*t6(11,8)*(exp(sum(1i*k.*d8)) - exp(sum(1i*... k.*d9)))/4; H2(10,8) = H2(11,7);

H2(11,8) = t6(11,8)*(exp(sum(1i*k.*d7)) + exp(sum(1i*k.*d8))... /4 + exp(sum(1i*k.*d9))/4); H2(6,9)=H2(9,6);H2(6,11)=H2(11,6);H2(6,10)=H2(10,6); H2(8,9)=... H2(9,8); H2(7,9)=H2(9,7); H2(7,10)=H2(10,7);H2(7,11)=H2(11,7);H2(8,10)... =H2(10,8); H2(8,11)=H2(11,8); end end [V,D]=eig(H1+H2); eigst = sum(D); E2(Nk,:)=sort(real(eigst)); temp=temp-1; end plot(dist_GM+X2,E2,'b'); % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% ... part done %%%%%%%K-Gamma Direction % Nt=101;

86 APPENDIX E. MATLAB CODE FOR MONOLAYER MOS2

kx = linspace(0,4*pi/3/a,Nt); ky=0; KG = [0,0;4*pi/3/a,0]; dist_KG = pdist(KG,'euclidean'); X3 = linspace(dist_KG,0,Nt); for Nk=1:Nt k=[kx(1,Nk) ky]; %%% Forming the Hamiltonian for i = 1:11 for j = 1:11 if(i == j)

H1(i,j) = eps(i) + (2*t1(i,i)*cos(sum(k.*d1))) + (2*t2(i,... i)*(cos(sum(k.*d2)) + cos(sum(k.*d3))) ); end %%+ symmetry if(((i==3) && (j==5)) || ((i==6) && (j==8)) || ((i==9) && (j... ==11)))

87 APPENDIX E. MATLAB CODE FOR MONOLAYER MOS2

==11) && (j==7)) || ((i==10) && (j==8)) )

H1(i,j) = t4(i,j)*(exp(sum(1i*k.*d4)) + exp(sum(1i*k.*d6)... )) + t5(i,j)*exp(sum(1i*k.*d5)); H1(j,i) = conj(H1(i,j)); end

H2(9,6) = t6(9,6)*(exp(sum(1i*k.*d7)) + exp(sum(1i*k.*d8)) + ... exp(sum(1i*k.*d9))); H2(11,6) = t6(11,6)*(exp(sum(2i*k.*d7)) - exp(sum(1i*k.*d8))... /2 - exp(sum(1i*k.*d9))/2); H2(10,6) = sqrt(3)*t6(11,6)*(-exp(sum(1i*k.*d8)) + exp(sum(1i... *k.*d9)))/2; H2(9,8) = t6(9,8)*(exp(sum(1i*k.*d7)) - exp(sum(1i*k.*d8))/2 ... - exp(sum(1i*k.*d9))/2 ); H2(9,7) = sqrt(3)*t6(9,8)*(-exp(sum(1i*k.*d8)) + exp(sum(1i*... k.*d9)))/2; H2(10,7) = 3*t6(11,8)*(exp(sum(1i*k.*d8)) + exp(sum(1i*k.*d9)... ))/4;

H2(11,7) = sqrt(3)*t6(11,8)*(exp(sum(1i*k.*d8)) - exp(sum(1i*... k.*d9)))/4; H2(10,8) = H2(11,7);

H2(11,8) = t6(11,8)*(exp(sum(1i*k.*d7)) + exp(sum(1i*k.*d8))... /4 + exp(sum(1i*k.*d9))/4); H2(6,9)=H2(9,6);H2(6,11)=H2(11,6);H2(6,10)=H2(10,6); H2(8,9)=... H2(9,8);

88 APPENDIX E. MATLAB CODE FOR MONOLAYER MOS2

H2(7,9)=H2(9,7); H2(7,10)=H2(10,7);H2(7,11)=H2(11,7);H2(8,10)... =H2(10,8); H2(8,11)=H2(11,8); end end [V,D]=eig(H1+H2); eigst = sum(D); E3(Nk,:)=sort(real(eigst)); end plot(dist_GM+dist_MK+X3,E3,'b'); %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% part ... done % Line Demarcations line([dist_GM dist_GM],[5 -8],'Color','k'); line([dist_GM+dist_MK dist_GM+dist_MK],[5 -8],'Color','k'); line([0 dist_GM+dist_MK+dist_KG],[0 0],'Color','k','LineStyle','--'); % set xlabel ticks set(gca,'XTick',[0,dist_GM,dist_GM+dist_MK,dist_GM+dist_MK+dist_KG]); set(gca,'XTickLabel',{'\Gamma','M','K','\Gamma'},'Fontsize',12); grid minor; xlabel('Wave Vector','Fontsize',14); ylabel('Energy(eV)','Fontsize',14); title('E-K','Fontsize',14);

89 Bibliography

[1] Shiang Fang, Rodrick Kuate Defo, Sharmila N Shirodkar, Simon Lieu, Geor- gios A Tritsaris, and Efthimios Kaxiras. Ab initio tight-binding hamiltonian for transition metal dichalcogenides. Physical Review B, 92(20):205108, 2015.

[2] Chem faculty otterbein.

[3] Gwo-Ching Wang and Toh-Ming Lu. RHEED Transmission Mode and Pole Figures: Thin Film and Nanostructure Texture Analysis. Springer Science & Business Media, 2013.

[4] AH Castro Neto, F Guinea, NMR Peres, Kostya S Novoselov, and Andre K Geim. The electronic properties of graphene. Reviews of modern physics, 81(1):109, 2009.

[5] Gui-Bin Liu, Wen-Yu Shan, Yugui Yao, Wang Yao, and Di Xiao. Three-band tight-binding model for monolayers of group-vib transition metal dichalcogenides. Physical Review B, 88(8):085433, 2013.

[6] Kern Rim, Judy L Hoyt, and James F Gibbons. Fabrication and analysis of deep submicron strained-si n-mosfet’s. Electron Devices, IEEE Transactions on, 47(7):1406–1415, 2000.

[7] Toshifumi Irisawa, Toshinori Numata, Naoharu Sugiyama, and Shin-ichi Takagi. On the origin of increase in substrate current and impact ionization eﬃciency in strained-si n-and p-mosfets. Electron Devices, IEEE Transactions on, 52(5):993– 998, 2005.

90 BIBLIOGRAPHY

[8] Debdeep Jena. Tunneling transistors based on graphene and 2-d crystals. Pro- ceedings of the IEEE, 101(7):1585–1602, 2013.

[9] Walter A Harrison. Electronic structure and the properties of solids: the physics of the chemical bond. Courier Corporation, 2012.

[10] Eric V Anslyn and Dennis A Dougherty. Modern physical organic chemistry. University Science Books, 2006.

[11] Luis Brey and HA Fertig. Electronic states of graphene nanoribbons studied with the dirac equation. Physical Review B, 73(23):235411, 2006.

[12] Habib Rostami, Ali G Moghaddam, and Reza Asgari. Eﬀective lattice hamiltonian for monolayer mos 2: Tailoring electronic structure with perpendicular electric and magnetic ﬁelds. Physical Review B, 88(8):085440, 2013.

[13] E Cappelluti, Rafael Roldán, JA Silva-Guillén, Pablo Ordejón, and F Guinea. Tight-binding model and direct-gap/indirect-gap transition in single-layer and multilayer mos 2. Physical Review B, 88(7):075409, 2013.

[14] P.S.K.K.D.B.K. Dr. N . K. Verma. Comprehensive Chemistry XII. Laxmi Publi- cations.

[15] John C Slater and George F Koster. Simpliﬁed lcao method for the periodic potential problem. Physical Review, 94(6):1498, 1954.

[16] Supriyo Datta. Quantum transport: atom to transistor. Cambridge University Press, 2005.

[17] Jean-Louis Calais. Band structure of transition metal compounds. Advances in Physics, 26(6):847–885, 1977.

[18] M Ali Omar. Elementary solid state physics: principles and applications. Addison- Wesley, 1993.

91 BIBLIOGRAPHY

[19] GS Painter and DE Ellis. Electronic band structure and optical properties of graphite from a variational approach. Physical Review B, 1(12):4747, 1970.

[20] SM-M Dubois, Zeila Zanolli, Xavier Declerck, and J-C Charlier. Electronic properties and quantum transport in graphene-based nanostructures. The European Physical Journal B, 72(1):1–24, 2009.

[21] Emilia Ridolﬁ, Duy Le, TS Rahman, ER Mucciolo, and CH Lewenkopf. A tight- binding model for mos2 monolayers. Journal of Physics: Condensed Matter, 27(36):365501, 2015.

[22] Huaixiu Zheng, ZF Wang, Tao Luo, QW Shi, and Jie Chen. Analytical study of electronic structure in armchair graphene nanoribbons. Physical Review B, 75(16):165414, 2007.

[23] Kyoko Nakada, Mitsutaka Fujita, Gene Dresselhaus, and Mildred S Dresselhaus. Edge state in graphene ribbons: Nanometer size eﬀect and edge shape dependence. Physical Review B, 54(24):17954, 1996.

[24] Katsunori Wakabayashi, Mitsutaka Fujita, Hiroshi Ajiki, and Manfred Sigrist. Electronic and magnetic properties of nanographite ribbons. Physical Review B, 59(12):8271, 1999.

[25] Katsunori Wakabayashi, Yositake Takane, Masayuki Yamamoto, and Manfred Sigrist. Electronic transport properties of graphene nanoribbons. New Journal of Physics, 11(9):095016, 2009.

[26] Sergej Konschuh, Martin Gmitra, and Jaroslav Fabian. Tight-binding theory of the spin-orbit coupling in graphene. Physical Review B, 82(24):245412, 2010.

[27] A. H. Castro Neto. Selected Topics in Graphene physics. ArXiv e-prints, April 2010.

92 BIBLIOGRAPHY

[28] Rupali Kundu. Tight-binding parameters for graphene. Modern Physics Letters B, 25(03):163–173, 2011.

[29] J-C Charlier, PC Eklund, J Zhu, and AC Ferrari. Electron and phonon properties of graphene: their relationship with carbon nanotubes. In Carbon nanotubes, pages 673–709. Springer, 2008.

[30] Yugui Yao, Fei Ye, Xiao-Liang Qi, Shou-Cheng Zhang, and Zhong Fang. Spin-orbit gap of graphene: First-principles calculations. Physical Review B, 75(4):041401, 2007.

[31] Matlab documentation 2015.

[32] DJ Klein. Graphitic polymer strips with edge states. Chemical Physics Letters, 217(3):261–265, 1994.

[33] Mitsutaka Fujita, Katsunori Wakabayashi, Kyoko Nakada, and Koichi Kusakabe. Peculiar localized state at zigzag graphite edge. Journal of the Physical Society of Japan, 65(7):1920–1923, 1996.