First principles study on surface modification of antimony doped tin dioxide nanoparticles

1.1 Antimony doped tin dioxide ATO 2 1.2 Introduction to surface modification 4 1.3 Surface modification and crosslinking 6 1.4 Advantages to nanoparticle functionalization 8 2.1 Availab

FIRST PRINCIPLES STUDY ON SURFACE MODIFICATION OF ANTIMONY DOPED TIN

DIOXIDE NANOPARTICLES

MONG YU SIANG

NATIONAL UNIVERSITY OF SINGAPORE

2013

FIRST PRINCIPLES STUDY ON SURFACE MODIFICATION OF ANTIMONY DOPED TIN DIOXIDE

NANOPARTICLES

Mong Yu Siang (B.Sc (Hons), NUS)

A THESIS SUBMITTED FOR THE DEGREE OF MASTER OF SCIENCE

DEPARTMENT OF CHEMISTRY NATIONAL UNIVERSITY OF SINGAPORE

2013

Thesis Declaration

The work in this thesis is the original work of Yu Siang Mong, performed independently under the supervision of Prof Hanson Cheng, Chemistry Department, National University of Singapore, between Aug/2010 and Dec/2012

Acknowledgement

I would like to express my deepest gratitude to my supervisor, Professor Hansong Cheng for his invaluable advice and guidance The unrelenting encouragement, trust and support have been deeply helpful and dear to me I would also like to thank mentors and colleagues who have studied and worked together with me; Chenggang Zhou, Bo Han, Zhangxian Chen, Wanchao Li and Ran Li The stimulating discussions and continual assistance has expedited the research and improved the quality

of my work tremendously

Lastly, I would like to thank the Ministry of Education (MOE) for granting me deferment from my scholarship obligations to carry out this research work, and the National University of Singapore for providing the facilities to carry out the research work reported herein

1.1 Antimony doped tin dioxide (ATO) 2 1.2 Introduction to surface modification 4 1.3 Surface modification and crosslinking 6 1.4 Advantages to nanoparticle functionalization 8

2.1 Available methods and approximations 13 2.2 Density functional theory 16 2.3 Computational method 18

4 Probing electrophysical properties of ATO 38

4.1 O2 adsorption on pure tin dioxide 39 4.2 O2 adsorption on ATO 41 4.3 Water adsorption on pure tin dioxide 46 4.4 Water adsorption on ATO 52

5 Surface modification using dichloroacetylene 60

5.1 Establishing surface model 62 5.2 Adsorption of dichloroacetylene 63 5.3 Cross-linking nanoparticles 68 5.4 Alternatives to dichloroacetylene 72

6 Surface modification using butadiene 75

6.1 Establishing surface model 77 6.2 Isomers of butadiene 78 6.3 Adsorption of trans-butadiene 80 6.4 Adsorption of cis-butadiene 83 6.5 Feasibility of overall reaction 86 6.6 Energy barrier of cis-butadiene adsorption 89 6.7 Cross-linking of cis-butadiene adsorbed

nanoparticles

91 6.8 Improvements to film performance 93

Summary

Transparent metal oxide films are widely utilized as transparent electrodes in optoelectronic devices and considerable attention has been devoted by researchers to improve the material performance of these films while achieving a thinner and more flexible film at a lower cost Antimony doped tin dioxide films synthesized from nanoparticles has been deemed to possess the potential to satisfy these requirements but currently known variations in methods of synthesis and doping continues

to fall short of substituting for the industrial standard of tin doped indium oxide This work essentially proposes and proves that the novel approach

of using organic molecules to functionalize antimony doped tin dioxide nanoparticle surface will serve to lift the performance of transparent films Our simulations show that antimony is capable of acting as an effective electron donor which serves to facilitate the molecular adsorption of oxygen and the mixed adsorption of water Two separate modes of modification were studied and the first involved the interaction between dichloroacetylene and hydroxyl groups on a hydroxylated surface The triply bonded carbons of dichloroacetylene were found to act as an effective linker to strengthen the cohesiveness between nanoparticles while simultaneously acting as a pathway for electrons to flow between

neighboring particles In the absence of a hydroxylated surface, butadiene was found capable of interacting strongly with the surface and cross-linking neighboring nanoparticles via a photochemically allowed [2+2] cycloaddition reaction In addition, butadiene passivates the surface and reduces the chances of unwanted nanoparticle agglomeration and allow for films to be manufactured only when the appropriate radiation is introduced Such cross-linking serves to bring nanoparticles closer and enhances the electrical conductivity of films while simultaneously impart higher mechanical strength to allow fabrication of thinner films

List of Tables

Table 1: Energy of system corresponding to surface doping site

Table 2: Summary of calculation for O2 end-on adsorption mode on doped SnO2 (100) surface at all possible sites

Table 3: Summary of calculation for O2 side-on adsorption mode on doped SnO2 (100) surface at all possible sites

Sb-Table 4: Summary of calculated data for water adsorption on pure SnO2 at varying levels of dissociation

Table 5: Summary of adsorption energies for one water molecule at individual sites

Table 6: Summary of calculated data for water adsorption on Sb doped SnO2 at varying levels of dissociation

Table 7: Reaction energy of equation (1) at all possible surface sites

Table 8: Summary of calculated carbon-halogen and hydrogen-halogen bond strength, with their respective energy difference

Table 9: Summary of cis-butadiene adsorption strength corresponding to the separation distance of surface oxygen atoms at each adsorption site

List of Figures

Figure 1: Schematic representation of functionalization strategy via two proposed methods

Figure 2: Schematic representation of a roll to roll film production process

Figure 3: (a) Unit cell of pure SnO2 bulk phase with periodic boundary conditions imposed on the left and (b) multiple unit cells illustrating periodic boundary conditions on the right

Figure 4: (a) Unit cell of Sb doped SnO2 bulk phase with periodic boundary conditions imposed

Figure 5: (a) front and (b) side view of ATO bulk phase model with periodic boundary conditions imposed

Figure 6: Density of States of (a) pure SnO2 (b) Sb-doped SnO2 bulk phase

Figure 7: Density of States of Sb doped SnO2 with Projected Density of States highlighting contribution of Sb around the Fermi level

Figure 8: Side view of (a) Sb doped SnO2 (100) surface on the left and (b) pure SnO2 (100) surface on the right

Figure 9: Illustration of surface segregated Sb3+ on the left and Sb5+ on the right

Figure 10: Total surface electron density of (a) pure SnO2 (100) surface on the left and (b) Sb-doped SnO2 (100) surface on the right

Figure 11: Side view of (a) side-on and (b) end-on adsorbed O2 pure SnO2 (100) surface

Figure 12: Top view of the Sb doped SnO2 (100) surface to Dotted ovals indicate two unique O2 side-on adsorption sites

Figure 13: (a) Top view and (b) side view of O2 adsorbed on Sb doped SnO2 (100) surface

Figure 14: Side view of single water molecule associatively adsorbed on pure SnO2 (100) surface

Figure 15: Side view of monolayer coverage of associatively adsorbed water molecules on pure SnO2 (100) surface

Figure 16: Side view of single water molecule dissociatively adsorbed on

Sb doped SnO2 (100) surface

Figure 17: Schematic diagram of surface model from top view, with all water adsorption sites labeled

Figure 18: Schematic diagram of surface model corresponding to sites as indicated in figure 12

Figure 19: Side view of adsorbed water molecules at seventy-five percent dissociation on Sb doped SnO2 (100) surface

Figure 20: Hydroxylated Sb doped SnO2 nanoparticle (100) surface with number denoting unique surface sites suitable for reaction with dichloroacetylene

Figure 21: Dichloroacetylene physisorbed to the hydroxylated surface of

Sb doped SnO2 nanoparticles

Figure 22: Energy diagram showing the change in energy with respect to the expected reaction pathway

Figure 23: Surface of (a) two unlinked ATO nanoparticles and (b) two ATO nanoparticles cross-linked by two triply bonded carbon atoms

Figure 24: Density of states corresponding to diagrams as shown in figure 23(a) on the left and figure 22(b) on the right

Figure 25: (a) Top and (b) side view of Sb doped SnO2 (100) surface model used to study butadiene adsorption

Figure 26: Diagrammatic representation of (a) trans-butadiene on the left and (b) cis-butadiene on the right

Figure 27: Top view of Sb doped SnO2 (100) surface model with four unique adsorption sites highlighted

Figure 28: Side view of cis-butadiene adsorbed onto Sb doped SnO2 (100) surface

Figure 29: Energy diagram illustrating energies associated with adsorption

of cis-butadiene on Sb doped SnO2

Figure 30: Density of States (DOS) of (a) unmodified and (b) cis-butadiene cross-linked Sb doped SnO2 nanoparticles

Figure 31: Density of States (DOS) of (a) unlinked and (b) cis-butadiene cross-linked Sb doped SnO2 nanoparticles with projected density of states

of carbon

Figure 32: Surface of two ATO nanoparticles cross-linked by two butadiene molecules

cis-List of Abbreviations

ATO antimony doped tin dioxide

CVD chemical vapor deposition

DFT density functional theory

DNP double numeric polarized

DOS density of states

GGA generalized gradient approximation

HRTEM high resolution transmission electron

microscopy ITO tin doped indium oxide

LCD liquid crystal display

LDA local density approximation

LST linear synchronous transit

VASP Vienna Ab initio Simulation Package

XANES x-ray absorption near edge structure

XRD x-ray diffraction

Chapter 1: Introduction

Transparent conducting films are extensively utilized as transparent electrodes in optoelectronic devices such as touch panels, flat panel displays and solar panels [1] These films have to be transparent because passive displays like liquid crystal displays (LCD) do not emit light itself and requires a backlight module which projects visuals through these electrode films As such, they are usually incorporated with the touch panel module within an electronic device where a functional layer is sandwiched between two such transparent electrodes

These transparent conducting films can be made of metal oxides or conducting polymers Conducting polymers have undergone tremendous development over the years Typical examples would include polyaniline and polyethylene-di-oxythiophene (PEDT) [2, 3] They are relatively cheap, light weight and more flexible than metal oxide films However, their organic nature has generally led to concerns over their lifespan and material stability[4] Therefore, metal oxides such as tin doped indium oxide (ITO) are the current material of choice for the electronics industry because it possesses excellent electrical conductivity and optical properties while being significantly more durable

1.1: Antimony doped tin dioxide (ATO)

Despite having these advantages, research has been conducted intensively for many years to search for a substitute to ITO This is because indium is very rare [5] and this rarity has resulted in indium being very expensive Among the doped metal oxides considered, the cheaper antimony doped tin dioxide (ATO) has been long deemed as potentially capable of achieving comparable performance with the more expensive tin doped indium oxide Antimony doped tin dioxide offers comparable performance where the transmittance values are generally reported to be above eighty percent [6-8] and the resistivity around 7.9 x 10-4 Ωcm [9-11]

Antimony doped tin dioxide films are traditionally manufactured using methods such as magnetron sputtering, chemical vapor deposition (CVD) or spray pyrolysis The metal oxide material is deposited as a thin layer on a substrate and the layer is intentionally kept thin to maximize transmittance The substrate used may be either made of glass or an organic polymer Increasing demands for portable consumer electronics has continually pushed for the lighter and more flexible polymers to be adopted as the preferred substrate over glass However, the fact that traditional film processing methods require a post-treatment heating process has hindered their adoption The film is typically exposed to

elevated temperatures in the region of a few hundred degrees Celsius for

a couple of hours Under these conditions, polymer substrates with a much lower glass transition temperature of 75˚C [12] will degenerate rapidly

As a consequence, the prerequisite to the widespread adoption of polymers as the substrate is to keep heating to a minimum during the annealing process The challenge is to improve the quality of films manufactured such that the annealing process is no longer necessary Current methods to improve film performance generally involve varying methods of synthesis and the doping levels of antimony in tin dioxide However, the results have continued to fall short of expectations without the annealing process and novel strategies must be explored if a breakthrough is to be achieved

The challenge is to improve the quality of films made by antimony doped tin dioxide nanoparticles, so that the annealing process may be kept to a minimum The advantages of working with nanoparticles include greater control over film thickness and stoichiometry however the mere deposition of nanoparticles without any post-treatment would mean that gaps can be found between nanoparticles These inter-particle gaps

negatively impact both electrical conductivity and mechanical strength as inter-particle binding strength is reduced, and electrons have to travel across a significant distance to reach neighboring particles We believe these gaps can be bridged by chemical agents that can bind to and link

up neighboring nanoparticles The benefits would include increased cohesive strength and reduce inter-particle distances Such a method would involve the surface functionalization of nanoparticles

1.2: Introduction to surface modification

Surface functionalization of semiconductor nanoparticles is an increasingly important area in the development of new semiconductor based materials The direct attachment of molecules can tailor the electrophysical and surface chemical properties associated with the semiconductor As a result, these molecules can impart new functionalities

to the semiconductor such as molecular recognition or passivity Incorporation of these molecules into semiconductors essentially combines the desired properties of organic and inorganic materials The additional functionality brought about by functionalization can be very useful in expanding the range of applications for semiconductor materials and lead to further technological development in optoelectronics devices, microelectronic computing devices and the patterning of semiconductor

materials The benefits of creating hybrid organic/semiconductor materials offers increased flexibility in creating materials with tunable surface properties by changing the functional groups associated with the molecule

Considerable attention was given to both the selection of surface modification agents and the metal oxide system The modification agent has to be carefully selected based on the inherent properties of the molecule as well as the reactivity with the surface Fundamental understanding of the metal oxide system is thus necessary when selecting modification agents and designing appropriate modification strategies

For tin dioxide alone, the large number of possible dopants and multiple surface orientations has served to complicate the selection process In this instance, antimony doped tin dioxide was selected over fluorine doped tin dioxide despite fluorine being a slightly more effective dopant This is because fluorine was reported by Esteves and co-workers

to segregate on the surface of fluorine doped fin oxide powders [13] and the presence of negatively charged anions on nanoparticles will have significant implications on surface modifications as well as electrical conductivity A surface segregated with negatively charged ions can

hinder electron hopping from nanoparticle to nanoparticle, which may result in reduced electric conductivity The electronegativity of fluorine may also withdraw electrons from critical surface active sites and increase the difficulty of surface modifications

1.3: Surface Modification and crosslinking

Computational simulations using density functional theory (DFT) were used to study the viability of the surface reactions and the potential electrical and mechanical improvements Depending on whether the surface is hydroxylated or clean, different surface modification strategies have to be adopted The challenge here is to identify surface modification agents with compatible geometry and reactivity The surface oxygen atoms were found to be selective for certain unsaturated molecules The reaction is not indiscriminate as preliminary studies found reactions with closer triply bonded carbons from acetylene to be unfavorable Fortunately, strong binding with the surface was found for both dichloroacetylene and cis-butadiene

Surface modification using these molecules may serve to passivate the nanoparticle surface prior to film production, bring nanoparticles closer to reduce the inter-particle gaps and offer greater uniformity in the

arrangement of nanoparticles For dichloroacetylene, the chlorine functional groups is used to react with the hydroxyl groups on a prehydroxylated antimony doped tin oxide nanoparticle surface and the resultant triply bonded carbon linker between them acts a bridge for electrons to flow This linker offers increased electrical conductivity by improving the ease of electron transfer since electrons are now considered to move in an intra-particle mechanism as opposed to an inter-particle mechanism For the cis-butadiene molecule, the conjugated carbons adsorb onto the surface active sites made up of under-coordinated surface oxygen atoms The resultant surface can be further cured under UV radiation and take advantage of Woodward-Hoffman’s rules to link two nanoparticles via a [2+2] cycloaddition reaction Linked nanoparticles are expected to be more uniformly arranged when deposited, possess increased mechanical strength, flexibility and electrical conductivity while not significantly compromising optical transparency

Figure 1: Schematic representation of functionalization strategy via two proposed methods

1.4: Advantages of nanoparticle functionalization

With the surface modification strategies proposed, we believe that the cross-linking of nanoparticles to form a stable transparent conducting oxide film can be achieved with little heating once deposited onto a polymer substrate The additional mechanical strength provided by the modification to the film, combined with the flexibility of polymer substrates makes a roll to roll film making process While the nanoparticles and the appropriate functionalization have to be manufactured via a batch production process in a reactor, the film can be manufactured using a much more efficient roll to roll process We envision the film making process to start by producing nanoparticles in a reactor

before being transported and sprayed onto a polymer substrate on a belt system before undergoing some post-processing and rolled up

Figure 2: Schematic representation of a roll to roll film production process

This thesis focus on the organic functionalization of Sb doped SnO2

nanoparticle surfaces to improve the mechanical and electrical properties

of transparent conducting films made from these nanoparticles The organic modifications are envisioned to be performed by either wet chemistry methods or vacuum based methods In Chapter 1, we provide a brief introduction to transparent conducting oxides films and their alternatives while highlighting their associated challenges and advantages The materials making up inorganic transparent conducting oxides will be discussed and the advantages of Sb doped SnO2 will be highlighted

Chapter 2 briefly discusses the currently available computational methods while providing a qualitative comparison between them Among which, density functional theory (DFT) will be discussed and highlighted to be an excellent compromise between accuracy and costs The models that were used in this thesis as well as the individual computational details associated with each model will be discussed and validation performed The validation of models is of significant importance as this study intends for the models and methods to maintain a high degree of relevance and draw credibility conclusions from the results Chapter 3 highlights the improvement in electrical properties with the introduction of Sb dopant into antimony We discuss the importance of a critical doping level where electrical conductivity improves with increased doping up to a certain point As it is computationally intensive to study all the possible surface orientations of a Sb doped SnO2 nanoparticle, we leverage on available literature to select the orientation believed to be the most dominant and justify the preferred doping site of the Sb dopant on the nanoparticle surface Chapter 4 encompasses the use of both molecular oxygen and water to probe the Sb doped SnO2 nanoparticle surface in order to gain valuable insights that will facilitate the development of novel surface modification strategies And Chapter 5 and 6 will detail the interactions between dichloroacetylene and cis-butadiene with the nanoparticle

surface as well as cover the mechanical and electronic improvements that they offer to flexible transparent conducting films

Chapter 2: Theoretical methodology

Computational simulations are important techniques that are utilized in various areas of science, and are not limited to engineering science, biological science, physics and chemistry These techniques grow increasingly prominent because they are highly accurate and relative cheap when used to simulate and predict the outcomes of an experiment

In addition, the rapid improvements in the processing power of modern computers have enabled chemical systems to be simulated much more quickly than before Such simulations are used to calculate the structure and properties of various systems and are conducted in conjunction with experiments that involve characterization and synthesis The highly accurate results provide detailed and meaningful understand to the studies conducted up to the atomic level Examples of data obtainable would include the expected coordinates of constituent atoms in three dimensions, spectroscopic diagrams, dipoles, electronic structures and the associated energies

Many methods and models have been developed over the years to improve the efficiency and accuracy of simulations And they can be differentiated into two broad categories, namely the molecular mechanical model and the quantum mechanical model The molecular mechanical

model is fundamentally based upon the laws of Newtonian mechanics to facilitate the prediction of stable structures and their associated properties However, it is essential an empirical method that neglects the effects of electrons and is therefore not recommended for the simulation

of chemical reactions or treat complex systems where electronic effects are often critical

On the other hand, quantum mechanical methods take the electronic effects into account and treat a collection of atoms as nuclei and electrons Information about a system is obtained solving the Schrödinger equation,

where is the Hamiltonian operator for a particular system, is the

wavefunction and E is the energy associated with the system

2.1: Available methods and approximations

There are different quantum mechanical methods and each differs

in the exact nature of various approximations As a result, they offer various levels of accuracy and efficiency Ab-initio models, density functional theory models and semi-empirical models are subsets of the quantum mechanical method Semi-empirical methods incorporate parameters derived from experiments into the calculation and this reduces

the resource demands on computers This enables the semi-empirical method to be used for large systems without being overly costly However, its application is limited for systems where the necessary parameters have been well developed Therefore, complex systems that have not been thoroughly studied would not be applicable

Ab initio methods leverages on the fundamental laws of quantum

mechanics to solve the Schrödinger equation and unlike semi-empirical methods, it does not include experimental data It provides highly accurate predictions for a large variety of complex systems that have yet

to be thoroughly studied However, this wave function based method suffers the drawback of being computationally demanding and expensive Density functional theory (DFT) models involve calculations that rely on the Hohenberg-Kohn theorem which expresses the total energy of a system in total electron density instead of a single wavefunction This

difference as compared to the ab initio method, essentially leads to a

reduction in the computational resources required while remaining highly accurate

In solving the Schrödinger equation, the wavefunction provides

us with information on the energy and structural properties of the system

studied However, the systems commonly studied tend to involve more than one atom and solving the associated Schrödinger equation is not trivial It requires some assumptions to be made since it cannot be explicitly solved for multi-electron systems and this leads to approximations An example would be the well-known Born-Oppenheimer approximation where the nuclei are assumed to be static and the electrons moving This approximation is valid since electrons have a much smaller mass that that of nuclei and thus move much faster than the nuclei Thereafter, electron distribution within a system is based on the nuclei coordinates and allows for the kinetic energy term of the nuclei in

to be omitted while the term representing inter-nuclei repulsion becomes a constant This leads to significant simplification of the calculation and the resultant total energy calculated with the incorporation of the Born-Oppenheimer approximation is the sum of electronic energy and inter-nuclear repulsion energy

Even with the inclusion of the Born-Oppenheimer approximation, solving the Schrödinger equation a many body system remains very much

a challenge As such, the variation method and the perturbation theory are utilized to determine the wave function The variation method involves the selection of an initial wavefunction and the energy of this initial wave

function is expected to change with the variation of one or more parameters By varying these parameters, effort is made to lower the energy and this energy minimization process is expected to lead to a wave function that is close to the true wave function The perturbation theory is generally viewed as complementary to the variation principle It aims to simplify the calculations of a complex system by treating the system as a simple one, before introducing gradual changes or perturbations that will lead to a proper representation of a complex system

2.2: Density functional theory (DFT)

Density functional theory based upon the Hohenberg-Kohn theorem[16] and the Kohn-Sham scheme[17] essentially considers a many body system as single body system and this framework makes it possible

to avoid having to deal with wave function right at the start According to the Hohenberg-Kohn theorem, the external local potential of a system with many electrons can be represented as a functional of the ground state density ( unique to the system To illustrate the Hohenberg-Kohn theorem, we revisit quantum mechanical theory A general expression of

a N-electron Hamiltonian operator Ĥ is as follows:

= + + ext

where , and ext represents the electronic kinetic energy operator, the electron-electron interaction operator and the local static external potential And if the ground state wave function is represented by , the ground state energy can be expressed as the following:

E0 =

=

The kinetic energy T and electron-electron interaction energy U are

noticeably independent of potential Kohn and Sham scheme further simplified the problem by introducing the Kohn-Sham orbitals and wavefunction ks The expression of energy is thus changed to:

Eks = The simplification involves treating the system as a non-interacting system while preserving the electron density We note that the operators are now expressed differently:

where is the kinetic energy operator of the non-interacting system and

is the effective external potential or the Kohn-Sham potential

The effective external potential is defined as:

where is the classic coulomb energy between electrons and is the exchange-correlation energy And the electron density for a system with

N electrons is determined from the Kohn-Sham orbitals, as expressed as:

2

As a result, the many body system is simply treated as a single body Although calculations are made simpler and less expensive, Kohn- Sham equations does not guarantee a completely accurate result The lack

of understanding the exchange correlation functional means that this functional has to be guessed and this is usually achieved using DFT methods such as local density approximation (LDA) or generalized gradient approximation (GGA) The exchange-correlation functional is assumed to vary with spatial coordinates in the LDA method, where the each special coordinate is associated with electron density The GGA method tries to improve upon the LDA method by considering the gradients of electron density together with the electron density at each spatial coordinate when determining the exchange-correlation functional

2.3: Computational method

Density Functional Theory (DFT)[17] calculations under the generalized gradient approximation (GGA) were performed using the

Perdew-Burke-Ernzerhof (PBE) [18, 19] exchange relation-correlation

functional as implemented in the Vienna Ab initio Simulation Package

(VASP) [20, 21] The projector augmented wave method (PAW) [22, 23] was used to describe the core electrons of atoms and the plane wave basis set [23, 24] represented the valence orbitals Electronic energies were calculated with the self-consistent field (SCF) tolerance of 10-5 eV and the structural optimizations were performed until the total energy of the system was converged to less than 10-4 eV Spin polarization was included for all calculations performed unless specified otherwise The charge distribution was analyzed using the Bader scheme [25]

The modeling of adsorption was achieved by placing a optimized molecule of interest approximately 1.6 to 2.0 Å above the multi-layered (100) surface unit cell with periodic boundary conditions imposed The adsorbate molecule was pre-optimized in a 10Å x 10Å x 10Å supercell All atoms including the adsorbate molecules were allowed to relax in all three directions The molecular adsorption energy is calculated as:

pre-Eads = Esurf+molecule – Esurf - Emolecule , (1)

where Esurf+molecule is the energy of the surface with one adsorbed

molecular oxygen, Esurf is the energy of the surface system studied, and

Emolecule is the energy of the molecule of interest A negative adsorption energy indicates a favorable adsorption

2.4: Bulk phase model

In order to achieve complete understanding of the nanoparticles, two separate models consisting of the bulk phase model and the surface model were created to represent both the pure and Sb doped SnO2 The surface model provides detailed information regarding surface and molecule at the atomic level while the bulk phase model gives us insight

to the electronic properties of the material Pure SnO2 nanoparticles had been experimentally determined to be of the tetragonal rutile phase using XRD [26-28] The interior of the SnO2 nanoparticle was thereby modeled

as bulk phase of SnO2 and a computational model of tetragonal rutile crystal structure with each Sn4+ ion is surrounded by six O2- ions in a tetrahedron conformation was adopted to model for the interior of the nanoparticle The tetragonal supercell model consists of 16 Sn and 32 O atoms with periodic boundary conditions imposed All the atoms and cell parameters of the supercell are fully optimized The Brillouin zone integration was sampled using the Monkhorst and Pack scheme[29] with (4x4x4) k-point mesh

Figure 3: (a) Unit cell of pure SnO2 bulk phase with periodic boundary conditions imposed on the left and (b) multiple unit cells illustrating periodic boundary conditions on the right Sn atoms are denoted in grey, oxygen atoms in red

To obtain a model for the bulk phase of Sb doped SnO2, one tin atom was replaced by antimony to give us a model consisting of 15 Sn, 32

O and 1 Sb atom with periodic boundary conditions The substitution of one out of sixteen metal atoms gives us a bulk phase model with 6.25 mole percent doping The direct substitution of antimony for tin at 6.25 mole percent doping is justified because it has been experimentally determined using x-ray diffraction (XRD) to directly substitute for Sn in the lattice even with the doping level well over 10% [30, 31] And with the substitution of antimony for tin, the bulk phase model represents an electrically conductive Sb doped SnO2

Figure 4: (a) Unit cell of Sb doped SnO2 bulk phase with periodic boundary conditions imposed Sn atoms are denoted in grey, oxygen atoms in red and antimony atom in purple

2.5: Validation of method

The computational method was tested by calculating the parameters of the bulk SnO2 structure The optimized lattice constants of

a = 4.83 Å , c = 3.24 Å were found to be in excellent agreement with

experimentally and theoretically obtained values [32, 33] The calculated cohesive energy of was found to be -15.85 eV which is in reasonable agreement to the reported value of -16.17 eV [33]

2.6: Surface Model

In this thesis, we consider our nanoparticles to be spherical in nature and sufficiently large at approximately 20-30nm in diameter Spherical nanoparticles can be associated with many different orientations and it is too computationally expensive to study all the possible facets so the most commonly observed orientation was selected to represent the

nanoparticle surface Stroppa et al [34] used a combination of high

resolution transmission electron microscopy (HRTEM) and CRYSTAL06 program package to determine that the surface energy of the (100) orientation is the lowest among the low index Sb doped SnO2 surface orientations when the doping level is under fourteen percent As a result, ATO nanoparticles are expected to be predominantly associated with the (100) plane to minimize surface energy and our surface model adopts the (100) orientation to represent the ATO nanoparticle surface

The (100) surface was described using a flat slab model with periodic boundary conditions imposed on the unit cell The choice of a flat surface model is justified by the fact the molecular absorption is not expected to be significantly altered for large nanoparticles of 20-50nm [35] This is because the relatively large particle diameters of 20-50nm give rise to relatively small nanoparticle surface curvatures compared to the size of the molecular species that we intend to study

The (100) surface slab is obtained from the bulk crystal structure of SnO2 by cleaving atoms layer by layer to obtained the desired termination while maintaining the stoichiometry of SnO2 The (100) orientation is unique such that the stoichiometry can only be maintained when both

sides of the slab are the same Even with this constrain, it is possible to generate surface slabs with several different type of terminations However, the oxygen terminated (100) surface has been reported to be the most stable [36, 37] and therefore we model the (100) surface based

on these information using Accelrys' Material Studio software package

In our surface models, we maintain at least five stoichiometric layers while keeping the bottommost layer fixed The bottommost layer is fixed to simulate the bulk phase of a nanoparticle and the layers extending upwards represent surface layers Five stoichiometric layers is sufficiently thick to obtain reasonable results because no change in surface energy was reported when the number of layers of the slab is two

or higher [33]

Chapter 3: Introduction of antimony

The introduction of the dopant Antimony into SnO2 has been widely reported to result in improved electrical conductivity [38, 39] This has been widely attributed to Sb serving as a n-type dopant in SnO2 and increasing the charge carrier concentration [9] of the material Sb has been experimentally determined using x-ray diffraction (XRD) to directly substitute for Sn in the lattice even with the doping level well over 10% [30, 31] However, the doping of Sb into SnO2 is nontrivial as Sb has two stable oxidation states, namely Sb3+ and Sb5+ Experimental findings fromx-ray absorption near edge structure (XANES) [40] measurements has revealed a tendency for both to coexist at any one time Sb5+ acts as an electron donor relative to Sn4+ in SnO2, and Sb3+ acts as an electron trap

As a result, the challenge in achieving high conductivity is to achieve a significantly higher proportion of Sb5+ than Sb3+ in Sb-doped SnO2 (ATO) [41] This is complicated by the fact that Sb3+ tends to dominate with increasing doping level[42] From the literature review conducted, the doping percentage has been found [43] to influence the ratio of Sb5+ to

Sb3+ and the doping percentage between 5-10% [30] has been generally reported to be effective in achieving high electrical conductivity As a result, the bulk phase model representing ATO as shown in figure 5

adopts a doping level of 6.25% with the direct substitution of Sn for Sb, thereby representing an electrically conductive ATO material

Figure 5: (a) front and (b) side view of ATO bulk phase model with periodic boundary conditions imposed The intersection of two grey and red lines represents Zn and O atom respectively For clarity, the Sb atom and the surrounding O atoms are highlighted as purple and red balls respectively.

3.1: Properties of bulk phase ATO

Following the geometrical optimization of the ATO system, the

lattice constants a and c of the optimized unit cell were found to be 4.794

Å and 3.235 Å respectively They were appreciably smaller than that of the pure SnO2 as was shown earlier This phenomenon has been reported [44]

in literature and the decrease in unit cell volume has been attributed to the presence of Sb5+ The effective ionic radii of Sn4+, Sb3+ and Sb5+ has been found [45] to be 69pm, 76pm and 60pm Therefore, the substitution

of Sb5+ for Sn4+ leads to a reduction of the lattice constants and the substitution of Sb3+ for Sn4+ increases the lattice constants This indicates