Exfoliated Graphene for Photothermal Biomedical Applications Matthew David John Quinn January 2018 a Thesis Submitted for the De

Exfoliated Graphene for Photothermal Biomedical Applications

Matthew David John Quinn

January 2018

A thesis submitted for the degree of Doctor of Philosophy of The Australian National University.

This work was performed entirely by Matthew Quinn.

This research has been supported by an Australian Research Training Program (RTP) Scholarship.

Signed:______

Acknowledgements

I have met lots of amazing people during my PhD and have truly appreciated the friendships I have been lucky enough to have made both at Swinburne and at The ANU. Both the School of Chemistry and Biochemistry at Swinburne and the Department of Applied Mathematics at The ANU, are places with very interesting, caring and friendly people. Thanks for all the great times and the memories!

My supervisor Shannon Notley has created many amazing opportunities for me throughout my studies and has been a great support over the years we have been working together. My thanks also go to my other supervisors and those special lecturers throughout my studies who put in the extra effort, ultimately passing on some of their enthusiasm.

My biggest thanks of course go to my family who have supported me throughout my studies. I could not have done any of this without their support and guidance.

Thank you to everyone who was a part of this.

Abstract

When a material’s size is reduced to the nanoscale, it is well understood that in addition to the dramatically increased surface area to volume ratio, new properties emerge distinct from those present in the bulk material. When graphite is exfoliated down to graphene, several altered and highly attractive properties emerge which may enable new photothermal roles as well as allowing significant improvements to existing strategies.

One such nanosize tuning method that is highly suitable for layered materials is surfactant assisted liquid exfoliation of the bulk material to single and few layered sheets. Successfully isolated graphene presents a range of remarkable properties such as an unsurpassed thermal conductivity of ≈ 5000 W m-1 K-1,1-5 a very high mechanical strength with a Young’s modulus of ≈ 1 TPa,6-7 a broad optical absorptivity with a transmission of 97.7 % independent of wavelength in the visible spectrum (and a large portion of the infrared),8-9 as well as an ultrahigh electron mobility of ≈ 200,000 cm2 V-1 s-1.10-12

The development of photothermal agents for a range of biomedical applications is an area of huge interest and promise with high impact outcomes possible. In this study, single and few layer graphene has been explored for use as a photothermal agent for a range of biomedical roles such as thermal ablation of cancerous cells and photothermally controllable drug release.

With this focus, several biomedical photothermal applications were explored including the thermal ablation of cancerous glioma-neuroblastoma cells through photothermal conversion at the target site by the graphene microplates. By exploiting the significant absorption of graphene in the near-infrared, substantial amounts of energy can be delivered deep within biological tissue allowing a highly-localized region of dramatic heating to be achieved resulting in tissue ablation and cell death. This study also explores the use of graphene as a photothermal trigger to activate the controlled release of drug payloads in three different carrier systems. These systems include a graphene loaded (and stabilised) Pickering emulsion carrier with both oil in water and water in oil types possible. A photothermal coring of the emulsions was successfully achieved demonstrating the photothermally induced collapse of the emulsion.

Several graphene entrained lipid nanocarrier systems were explored with near-infrared activation inducing phase transitions. Small angle X-ray scattering was used to dynamically monitor photo-activated, reversible phase transitions.

An injectable hybrid graphene-surfactant-α-cyclodextrin thermoreversible gel system with graphene as an intrinsic component was also explored within this study. Photothermal drug release switching and controllable release rates were demonstrated with this biocompatible carrier system showing a highly versatile photothermally activatable drug depot.

Graphene is a material well suited for photothermal biomedical applications particularly when prepared via liquid exfoliation. This study explores the interactions of specific light energies with surfactant assisted liquid exfoliated graphene for photothermal applications and shows that graphene has a high transduction efficiency, is thermally stable and is intrinsically suitable towards stealth strategies, suggesting that graphene could be a significant addition to a range of photothermal biomedical applications.

Publications First author papers 1. Quinn, M. D. J.; Ho, N. H.; Notley, S. M., Aqueous Dispersions of Exfoliated Molybdenum Disulfide for Use in Visible-Light Photocatalysis. ACS Appl. Mater. Interfaces 2013, 5 (23), 12751-12756.

2. Quinn, M. D. J.; Du, J.; Boyd, B. J.; Hawley, A.; Notley, S. M., Lipid Liquid- Crystal Phase Change Induced through near-Infrared Irradiation of Entrained Graphene Particles. Langmuir 2015, 31 (24), 6605-6609.

3. Quinn, M. D. J.; Vu, K.; Madden, S.; Notley, S. M., Photothermal Breaking of Emulsions Stabilized with Graphene. ACS Applied Materials and Interfaces 2016, 8 (16), 10609-10616.

4. Quinn, M. D. J.; Wang, T.; Du, J. D.; Boyd, B. J.; Hawley, A.; Notley, S. M., Graphene as a Photothermal Actuator for Control of Lipid Mesophase Structure. Nanoscale 2017, 9 (1), 341-348.

5. Quinn, M. D. J.; Wang, T.; Al Kobaisi, M.; Craig, V. S. J.; Notley, S. M., PEO- PPO-PEO Surfactant Exfoliated Graphene Cyclodextrin Drug Carriers for Photoresponsive Release. Mater. Chem. Phys. 2018, 205, 154-163.

6. Quinn, M. D. J.; Tao, W.; Shannon, M. N., Surfactant Exfoliated Graphene as a Near-Infrared Photothermal Ablation Agent. Biomedical Physics & Engineering Express, (accepted for publication) 2018.

Supporting author papers 1. Vucaj, N.; Quinn, M. D. J.; Baechler, C.; Notley, S. M.; Cottis, P.; Hojati-Talemi, P.; Fabretto, M. V.; Wallace, G. G.; Murphy, P. J.; Evans, D. R., Vapor Phase Synthesis of Conducting Polymer Nanocomposites Incorporating 2d Nanoparticles. Chem. Mater. 2014, 26 (14), 4207-4213.

2. Pham, V. T. H.; Truong, V. K.; Quinn, M. D. J.; Notley, S. M.; Guo, Y.; Baulin, V. A.; Al Kobaisi, M.; Crawford, R. J.; Ivanova, E. P., Graphene Induces Formation of Pores That Kill Spherical and Rod-Shaped Bacteria. ACS Nano 2015, 9 (8), 8458- 8467.

3. Wang, T.; Quinn, M. D. J.; Nguyen, S. H. T.; Yu, A.; Notley, S. M., Graphene Films Using a Thermally Curable Surfactant. Adv. Mater. Interfaces 2016, 3 (15).

4. Wang, T.; Quinn, M. D. J.; Notley, S., A Benzoxazine Surfactant Exchange for Atomic Force Microscopy Characterization of Two Dimensional Materials Exfoliated in Aqueous Surfactant Solutions. RSC Advances 2017, 7 (6), 3222-3228. Table of Contents Acknowledgements ...... 5 Abstract ...... 6 Publications ...... 8 First author papers ...... 8 Supporting author papers ...... 8 Thesis introduction ...... 15 Chapter 1 ...... 18 1.1 Introduction to graphene ...... 18 1.1.1 Carbon allotropes ...... 18 1.1.2 A predicted unstable 2D material ...... 20 1.2 Graphene key properties arise from the electronic structure ...... 22 1.2.1 Graphene and the primitive cell ...... 23 1.2.2 The Brillouin zone of graphene ...... 24 1.2.3 Review of relevant solid state physics ...... 25 1.2.4 Graphene, a zero-gap semiconductor ...... 26 1.2.5 Electrons at the degeneracy points ...... 28 1.2.6 Bandgap in bi- and few layer graphene...... 30 1.2.7 Density of states ...... 32 1.3 Brief Introduction to the key properties of graphene ...... 33 1.3.1 Optical properties of graphene ...... 35 1.3.2 Thermal properties ...... 37 1.3.3 Electronic transport properties ...... 39 1.3.4 Mechanical strength ...... 45 1.4 Production methods for types of graphene ...... 47 1.4.1 Chemical vapour deposition ...... 48 1.4.2 Micromechanical cleavage (scotch tape method) ...... 50 1.4.3 Liquid exfoliation of graphite ...... 51 1.4.4 Graphene oxide (Hummers method) ...... 52 1.4.5 Reduced graphene oxide ...... 53 1.4.6 Liquid exfoliation – solvent approach...... 54 1.4.7 Surfactant assisted liquid exfoliation ...... 56 1.4.8 Relevant properties of SALE graphene...... 57 1.5 Photothermal principles ...... 59 1.5.1 Targeted photothermal applications ...... 60 9

1.5.2 Photothermal therapy principles ...... 60 1.5.3 Photothermal material and radiation selection ...... 60 1.6 Nanomaterials as photothermal agents ...... 61 1.6.1 Gold nanoparticles ...... 62 1.6.2 Gold nanoparticles as photothermal agents ...... 63 1.6.3 Graphene derivatives as photothermal agents ...... 65 1.6.4 Nanomaterials, cytotoxic or biocompatible? ...... 67 A2.0 Chapter 2A Graphene exfoliation theory ...... 70 A2.1 Introduction to surfactant assisted liquid exfoliation (SALE) ...... 70 A2.1.2 Bulk graphite ...... 71 A2.1.3 Minimizing the interfacial tension with surfactant ...... 72 A2.1.4 Sonication to effect exfoliation ...... 77 A2.2 Enthalpy of mixing ...... 79 A2.2.1 Key parameters of mixing enthalpy ...... 81 A2.3 Practical parameters of exfoliation ...... 83 A2.4 The (dual) role of surfactants ...... 86 A2.4.1 Surfactant types employed ...... 86 A2.4.2 Surfactant behavior ...... 87 A2.4.3 Suspension stability – Coulomb repulsion ...... 88 A2.4.4 Suspension stability - steric stability ...... 90 A2.4.5 Surfactant selection for targeted applications ...... 91 A2.4.6 Continuous addition of surfactant ...... 93 A2.5 Sonication - shear exfoliation ...... 94 A2.5.1 Ultrasound ...... 95 A2.5.2 Sonication to shear ...... 96 A2.6 Cavitation...... 99 A.2.6.1 Cavitation mechanism of exfoliation ...... 100 A2.7 Defect discussion ...... 101 B2.0 Chapter 2b - Graphene exfoliation and characterization...... 105 B2.1 Typical preparation procedure ...... 105 B2.2 Suspension characterization ...... 107 B2.2.1 Raman spectroscopy ...... 108 B2.2.2 UV-visible absorbance ...... 122 B2.2.3 Atomic force microscopy ...... 129 B2.2.4 Transmission electron microscopy...... 133 B2.2.5 Scanning electron microscopy ...... 138 B2.2.6 Dynamic light scattering – particle sizing ...... 140 B2.2.7 Zeta potential ...... 149 B2.2.8 X-ray diffraction ...... 154 B2.2.9 Suspension stability ...... 156 B2.2.10 XPS ...... 157 B2.2.11 SANS ...... 160 B2.3 General laser parameters ...... 162 B2.3.1 Lasers used in photothermal experiment ...... 162 A3.0 Chapter 3A – Photothermal properties of graphene ...... 165 A3.1 The photothermal process ...... 165 A3.2 Optical properties ...... 165 A3.2.1 Broad absorbance spectrum of graphene ...... 165 A3.2.2 Quantitative absorbance properties ...... 170 A3.2.3 Graphene absorbance - quantitative perspective ...... 171 A3.3 Thermal properties- general and applications focus ...... 173 A3.3.1 Phonon dispersion in graphene ...... 173 A3.3.2 Specific heat ...... 174 A3.3.3 Thermal conductivity ...... 175 A3.4 Interfacial thermal resistance ...... 180 A3.5 Thermal properties of graphene in suspension ...... 184 A3.5.1 Ballistic transport persistence ...... 185 A3.5.2 Clustering and percolation ...... 185 A3.5.2 Aspect ratio ...... 186 A3.5.3 Layer number influence on interfacial resistance ...... 187 A3.5.4 Polymer/surfactant coatings ...... 187 A3.5.5 Vapour formation around particles ...... 189 A3.5.6 Potential oxidation during irradiation ...... 190 A3.5.7 Summary of thermal properties ...... 192 B3.0 Chapter 3B – Assessing the photothermal properties of graphene ...... 194 B3.1 Materials and methods ...... 194 B3.1.1 Suspension selection...... 194 B3.1.2 Experimental design ...... 194 B3.1.3 Photothermal curve...... 195

B3.1.4 Laser power ...... 196 B3.1.5 Transducer concentration ...... 196 B3.1.6 Background heating ...... 196 B3.1.7 Photothermal cycling ...... 197 B3.1.8 Photothermal efficiency method ...... 197 B3.1.9 Laser flash analyser ...... 198 B3.2 Results and discussion ...... 199 B3.2.1 Photothermal curve ...... 199 B3.2.2 Background heating ...... 201 B3.2.3 System relationship to laser power ...... 202 B3.2.4 System relationship to concentration ...... 203 B3.2.5 Photothermal efficiency ...... 205 B3.2.6 Stability of graphene ...... 211 B3.2.7 Thermal property characterisation ...... 218 Chapter 3 Conclusions ...... 221 4.0 Chapter 4 Graphene for the photothermal ablation of cancerous cells ...... 223 4.1 Introduction ...... 223 4.1.1 Biological interactions of graphene ...... 223 4.1.2 Cytotoxicity studies ...... 226 4.1.3 Thermal ablation ...... 228 4.2 Materials and methods ...... 231 4.2.1 Tissue culturing ...... 231 4.2.2 Irradiation experiment ...... 232 4.3 Results and discussion ...... 233 4.3.1 Seeding density and incubation time ...... 233 4.3.2 Surfactant cytotoxicity ...... 234 4.3.3 Graphene cytotoxicity ...... 237 4.3.4 Thermal ablation ...... 245 Chapter 4 Conclusions ...... 250 5.0 Chapter 5 Photothermal breaking of emulsions stabilised with graphene ...... 252 5.1 Introduction ...... 252 5.1.1 Emulsions for drug delivery ...... 252 5.1.2 Polyethylene oxide solubility properties ...... 254 5.2 Materials and methods ...... 255 5.2.1 Preparation of emulsions ...... 255 5.2.2 Emulsion characterization – emulsion stability ...... 256 5.2.3 Breaking of emulsions ...... 256 5.3 Results and discussion ...... 257 5.3.1 Emulsion system control ...... 257 5.3.2 Emulsion stability ...... 260 5.3.3 Induced emulsion breakage ...... 267 Chapter 5 Conclusions ...... 271 6.0 Chapter 6 Graphene as a photothermal actuator for control of lipid mesophase structure ...... 274 6.1 Introduction ...... 274 6.1.1 Lipids in drug release ...... 274 6.2 Materials and methods ...... 275 6.2.1 Graphene preparation ...... 275 6.2.2 Lipids ...... 276 6.2.3 Formulation preparation ...... 276 6.2.4 Small angle X-ray scattering (SAXS) ...... 277 6.2.5 Crossed polarised light microscopy ...... 278 6.3 Results ...... 278 6.3.1 Equilibrium structural studies ...... 278 6.3.2 Photothermal activation of phase transitions using graphene ...... 283 6.3.3 Polarized microscopy for confirmation of phase transition boundaries ...... 286 6.3.4 Reversibility of the photothermal effect and dependence on laser power ...... 287 Chapter 6 Conclusions ...... 292 7.0 Chapter 7 Controllable release from graphene loaded tunable α-cyclodextrin gels ...... 294 7.1 Introduction ...... 294 7.2 Materials and methods ...... 296 7.2.1 Graphene preparation ...... 296 7.2.2 α-CD gel preparation and characterization ...... 296 7.2.3 α-CD gel drug release parameters ...... 297 7.3 Results and discussion ...... 298 7.3.1 α-CD-surfactant drug release studies ...... 298 7.3.2 Surfactant composition influences on gel properties ...... 303 Chapter 7 Conclusions ...... 316 Summary chapter ...... 318 Thermal ablation summary ...... 319

Pickering summary ...... 320 Lipids summary ...... 321 α-cyclodextrin summary ...... 323 Final comment ...... 324 References ...... 325

Thesis introduction

The primary focus of this study is to initially gain an insight into the photothermal ability of surfactant assisted liquid exfoliated (SALE) graphene and assess the suitability of this material in biomedical roles. Following the base assessments, the focus will transition to the biomedical photothermal applications of graphene, covering direct photothermal ablation and several photothermally activated drug release systems.

The production method will be explored in depth along with the characterization of graphene sheets produced in Chapter 2. The base photothermal assessments including direct transduction efficiency and photostability of the material is assessed in Chapter 3. Chapter 4 explores the interaction of graphene within biological environments from both a literature based review as well as an experimental exploration into the cytotoxic effects, and an in vitro demonstration of NG108-15 photothermal ablation.

Three photothermally activated drug release systems, Pickering stabilised emulsions, lipid nanostructures and an α-cyclodextrin (α-CD) -surfactant gel depots, are then explored through Chapters 5, 6 and 7 respectively. These chapters follow the structure of a brief introduction to the drug system, a detailed report of the methods of carrier system preparation and then exploring the photo-activation of said systems.

These chapters show SALE graphene to be a material capable of efficiently transducing the incident near-infrared (NIR) light to thermal energy on a highly-localized scale to allow controllable cell death or activation of drug release. Graphene is shown to be highly stable both from a colloidal and chemical perspective suggesting it is well suited for such roles. The biocompatibility is explored and assessed to be moderately cytotoxic and routes towards optimization of such properties are discussed.

Complete cell death is achieved upon irradiation of NG108-15 cells with graphene incorporated, with a discussion of routes towards optimization, such as via specific targeting. The Pickering stabilised emulsions allowed a demonstration of the localized photothermal activation of the emulsion carrier, but was not further explored due to a lack of reversibility. The lipid nanostructure systems were an excellent demonstration of the localized transduction and the reversibility of the lipids allowed in situ photothermal cycling and a detailed exploration into the phase transition profiles. The α-CD-surfactant-graphene drug depots allowed a full demonstration of photothermally activated release switching profiles providing a well-suited system to demonstrate the stability, the versatility and the broad photothermal abilities of the SALE graphene.

Chapter 1

1.1 Introduction to graphene

Graphene is a material of remarkable properties and has justifiably attracted huge attention since its successful isolation in 2004 by Andre Geim and Konstantin Novoselov and the subsequent awarding of the Nobel prize in physics in 2010.7 This material, with a thickness of a single atom, has a thermal conductivity approaching theoretical limits,1-5, 13 a transmission that is independent of wavelength in the visible region (and extending far into the infrared),8-9 a remarkable strength (elastic/Young’s modulus)6-7 along with a very high electrical conductivity10-12 and a range of other properties making it a highly promising material for a range of high impact applications. This two-dimensional (2D) material has provided the scientific community with an exciting, easily obtained, atomically thin material allowing bench-top experimental studies of a range of fundamental physics and quantum electrodynamics not previously possible.14-16 This material, which presents some of the most remarkable physical properties known to man, consists of a simple arrangement of carbon.

1.1.1 Carbon allotropes

Carbon is a group four element with four valence electrons available for covalent bonding. This makes it capable of forming single, double and triple bonds, with the most common allotropes being diamond and graphite. The presence of a single sheet of graphite is considered a new allotrope and is referred to as graphene.

These three allotropes of carbon, diamond, graphite and graphene, have vastly different properties which is due to the configuration of atoms and the bonding present. Diamond is a material connected in a three-dimensional lattice which results in a material of very high strength and hardness, (Young’s modulus E ≈ 910 GPa, Knoop hardness 5700 – 10,400 kg mm-2)17 with a high thermal conductivity (k) (k ≈ 2200 W m-1 K-1)17-19 The tetrahedral arrangement of atoms consists of sp3 hybridized bonding and involves all valence electrons subsequently resulting in a material of low electrical conductivity (resistivity 100 GΩ m) due to the absence of free electrons to act as charge carriers.17

Graphite is not connected in a three-dimensional network in the same fashion that diamond is, with the planar honeycomb lattice of carbon atoms being stacked vertically and being weakly held together with attractive van der Waals (vdW) forces. The layered aspect of this material and the attractive intermolecular forces holding said layers together results in a ‘soft’ material (Ea ≈ 1060 GPa, Ec ≈ 36.5 GPa, E(parallel to plane) ≈ 4.5 GPa) which is easily delaminated as in the case of graphite pencils.20

Within graphite, the carbon atoms are arranged in a trigonal planar, honeycomb configuration 2 - which consists of sp hybridization and leads to a high thermal conductivity (kab = 390 W m 1 -1 -1 -1 K , kc = 2 W m K ). This atomic configuration also leads to high electrical conductivities -6 -1 (resistivityab = 2.5 - 5.0 x10 Ω m ) due to the sharing of electrons throughout the lattice which is a result of the sp2 hybridization.

As graphene is simply a single layer of graphite, the atomic configuration is identical to that of graphite. However, the isolation of a single sheet and the subsequent absence of neighboring atoms, leads to some remarkable changes to the optical, electrical, thermal and mechanical properties.

Figure 1.1. Molecular structure of (a.) diamond, (b.) graphite (Bernal ABA stack) and (c.) graphene with the interlayer spacing of graphite labeled at 3.35 Å.3 21

All three of these material types consist solely of carbon atoms, with only the arrangements and bond types varying. However, these different allotropes result in vastly different material properties. From the common, cheap and unexciting graphite, to graphene, a material with some of the most remarkable properties that will likely allow significant technological advancements in the near future.

1.1.2 A predicted unstable 2D material

2D materials were predicted to be unstable due to the thermodynamic requirement for the existence of out-of-plane bending with interatomic interaction generating a mathematical paradox.22 Despite being only a single atom in thickness, graphene is not truly a 2D material but is a pseudo 2D material. This was initially revealed via transmission electron microscopy (TEM) imaging and scanning-probe microscopy on free standing and substrate supported graphene respectively. These imaging techniques showed pronounced out-of-plane deformations with heights of ≈ 1 nm.22-25

Graphene had been studied theoretically for approximately 60 years prior to its successful isolation, with the monolayer properties being explored to describe the properties of various carbon-based materials, namely, graphite.26-28 While many theoretical studies regarding graphene had been made, it was widely believed that 2D materials like graphene would be impossible due to thermodynamic stresses.

The stability of this pseudo 2D material, is due to the ripple formation resulting from the partially decoupled bending and stretching modes.22 The temporal and spatial modulation of the carbon-carbon (C-C) bond lengths due to the thermal vibrations and interatomic interactions force carbon atoms to occupy space in the third dimension.22 This forms dynamic ripples which lower the free energy, stabilizing the material. Such events can lead to the formation of electron-hole puddles which can induce an asymmetric distribution of bond lengths, which in turn forces the lattice to become non-planar. This is even more pronounced at graphene sheet edges or near defect sites where the asymmetry of the bond lengths is amplified, requiring a greater density of ripples is present to minimize the free energy.

The bond length asymmetry alone is quite dramatic even at room temperature i.e. 300 K. In the ground state of graphene all bonds are equivalent as conjugation leads to an averaging of the atomic distances. However, the asymmetry of first neighbor bond lengths even at room temperature, can vary from the average conjugated bond length of 1.42 Å out to 1.31 Å (double bonds) and 1.54 Å (single bonds). 23 These deviations result in significant deviations from planarity.

The pseudo 2D nature of graphene is an important factor when considering the fundamental properties of graphene and ultimately allow this single atomic layer of carbon atoms to be stable when removed from neighboring layers.

The first major graphene-related substance discovered was the buckminsterfullerene C60 or the buckyball.29 This was a soccer-ball-like configuration of carbon atoms which was found in common lamp soot and found to be very stable. Following the discovery of the buckyball was the discovery of carbon nanotubes (CNTs), rolled up sheets of graphene with a range of

21 attractive properties such as high thermal (k > 3000 W m-1 K-1)30 and electrical conductivity and again high tensile strength.31-32 CNTs quickly drew a large amount of attention due to these impressive properties and the area expanded rapidly with single walled CNTs (SWCNTs) and multi-walled CNTs (MWCNTs) being explored for a wide range of applications.33-34

In 2004, graphene was isolated experimentally, resulting in Professors Andre Geim and Kostya Novoselov receiving the 2010 Nobel prize in physics.7 This initial isolation of graphene monolayers was achieved via the scotch tape method, whereby graphene is removed from a graphite sample by adhesive tape (see Section 1.5 for more detail). The key to enabling this isolation was to deposit the micromechanically exfoliated graphene onto a 300 nm thick silicon wafer, which exploits a small but finite difference in the reflected light intensities of graphene making the contrast observable.35-36

1.2 Graphene key properties arise from the electronic structure

As with graphite, graphene sheets have three valence electrons (in the 2s and 2p states) combining to form strong bonds with neighboring atoms in plane.17 The hybridization of one s and two p orbitals forms a trigonal planar bond configuration with a sigma bond (a high strength bond) providing the in-plane strength of graphene.6-7 The fourth valence electron occupies the 2pz orbital which extends perpendicular from the atom (Figure 1.2 a). The p orbitals can bind covalently with neighboring carbon atoms to form a nodal plane pi bond. 37

Figure 1.2. (a.) Image of graphene hybridization constituents, and (b.) schematic describing the delocalization of the pi bonds and resulting conjugation.

As each p orbital has only one extra electron, the pi band is half filled. As these pi orbitals are half filled, they follow the tight-binding characteristic strongly (no/ or few neighboring influences) leading to large Coulomb energies and strong collective effects.37 Importantly, the pi electrons can be considered as delocalized, like the electrons within many metals. This quasi-free pi electron is shared among all atoms in the layer and is responsible for the electrical conductivity of graphene which will be discussed in greater detail in Section 1.4.

1.2.1 Graphene and the primitive cell

The thickness of an individual graphene layer is considered to 0.335 nm, which is determined from the interlayer distance of graphite and also correlates with an approximate summation of the sp2 hybridized carbon=carbon double bond (C=C) atomic and orbital heights.38-40 The unit cell of an individual graphene layer contains two inequivalent carbon atoms A and B, which despite each having the same number of surrounding atoms, have different relative positions. The inequivalence of these two atoms result in what is referred to as a complex lattice where the A and B atoms form two sublattices. The AB distance is of a ≈ 1.42 Å as shown in Figure 1.3.17, 41-42 23, 43 The fundamental lattice displacements are defined by AA’

= a1 and AA”=a2 with a distance of a1 = √3 * 1.42 Å = 2.46 Å.

Figure 1.3. Schematic of graphene showing distances between atoms, the distance for each vector of the unit cell, the angle of the sp2 hybridized bonds as well as the inequivalent carbon atoms represented with green and grey balls.

1.2.2 The Brillouin zone of graphene

To consider the electronic structure of graphene one must first shift from the real space direct lattice to the reciprocal lattice. This is achieved by identifying the primitive unit cell of graphene within the Bravais lattice as shown in Figure 1.4 a, and then converting from the real space vectors (ax) to the reciprocal vectors (bx) (Figure 1.4 b). By drawing the crystal plane lines of symmetry to the constructed reciprocal lattice, the first boundary can be identified indicating the first Brillouin zone (BZ).

The first boundary is that of a hexagon with sides of a distance 4pi/3a from its center. The density of electrons in reciprocal space is 2A where A is the area of the crystal and in this zone, each atom has exactly 1 electron. The BZ of graphene therefore said to have 2N electron states and the second BZ is empty.44

Figure 1.4. (a.) Simple unit cell of graphene where each of included portion of the atoms on the vertices sum to one atom, given two atoms per unit cell. (b.) Schematic showing the conversion from the real space vectors ax to the reciprocal space vectors bx and the vector equation. (c.) Building the reciprocal lattice from the bx vectors, overlaying the crystal planes, and highlighting the BZ around the origin where no crystal planes are crossed (green) as well as the second BZ (gold). Figure 1.4 d shows the first BZ of graphene with the middle of the BZ (Γ), the edge of a rectangular face (K, K’), and the middle of a rectangular face (M).

It is important to note that the reciprocal lattice can be constructed from either the A or B sublattices, as such within the reciprocal lattice all the unit cells are equivalent. This allows the physical properties to be studied within the first BZ, enabling the electronic structure of the entire lattice to be extrapolated.

1.2.3 Review of relevant solid state physics

We can see that to understand many of the key properties of graphene, it is best to consider the material from the reciprocal space perspective. In addition to understanding the space we are working from, we must review the laws governing the properties of graphene, in this case the most relevant perspective is from that of solid state physics.

Graphene is a very simple material from the perspective of components, simply comprising of carbon atoms covalently bonded together and arranged in a hexagonal lattice. It is primarily the arrangement of the atoms and the subsequent behavior of the electrons within this material which dictate the remarkable properties of graphene. Therefore, the focus must 25 be on the electronic structure and the relevant fundamental laws which dictate the resulting properties.

The energy dispersion (E-k – relation) is the most relevant characteristic which describes the behavior of electrons, connecting the electron’s energy to its quasi momentum. Equation 1 shows the link between the wave vector of the electrons to the energy of the electrons giving the energy dispersion. Following the De Broglie relations, the momentum p can be expressed as a wavenumber q using Planck’s reduced constant ћ (h bar). As this is the momentum of an electron within a solid, m* represents the effective mass of the electron with ν representing the velocity.

푝 = ℏ푞 = 푚∗휈 Equation 1.1

An understanding of the methods to calculate values such as the momentum of electrons in solids is critical in understanding the behavior of electrons within graphene and therefore allowing a comprehensive understanding of the unique properties of this remarkable material.

1.2.4 Graphene, a zero-gap semiconductor

The bandgap of graphene has an energy of zero at several positions within the BZ, classing this material as a zero-gap semiconductor.37, 45-46 Graphene is well known to have remarkable electron transport abilities with velocities of 200,000 cm2 V-1 s-1 reported in the early seminal reports (for suspended graphene).10-11 However, since the initial reports on the fundamental electronic transport properties, extensive efforts have been made to explore opening a bandgap within graphene to allow integration within current planar silicon based technology. Such gap inducing strategies have shown that strain and morphology confinement strategies, as with graphene nanoribbons, can achieve such semi-conducting properties.47-48 To understand why it is possible for graphene to have such remarkable properties and straddle several potential electrical roles, we must understand the physics of its properties when acting as a zero-gap semi-conductor or a semimetal.

Electrons follow the Pauli exclusion principle, where no two electrons of the same spin can occupy the same energy state, and as such electrons are classed as Fermions. Within atoms and solids, electrons will fill the available energy levels (electron shells) from the lowest possible energy states up until all electrons are occupying a state. At the highest energy state occupied by an electron, the Fermi level is defined.

When the valence and conduction bands of a material are overlapped, allowing partial occupation of the conduction band, the material will be classed as a conductor. When the bands are only slightly overlapping, it is classed as a semimetal, and when the bands are separated by an energy range which can allow an electron transition but only under sufficient excitation (i.e. not an excessively large gap), the material is classed as a semiconductor. Graphene, under certain conditions or with slight variations, can present properties of both a semimetal and a semiconductor.

For monolayer graphene the Fermi level is exactly at the point where the conduction and valence bands meet.16, 37, 49 These positions are referred to as Dirac points.16, 37, 49 As the bandgap is touching, graphene ought to be classed as a semimetal, however as this implies an overlap, the definition of zero-gap semiconductor gives a precise and descriptive label to the electrical properties of graphene.

The pi electrons of graphene are weakly associated with any one atom and are shared throughout the material making graphene electrically conductive. As the pi band of graphene occupies two levels, both the pi and the anti-pi (pi*), but with only one electron, the band can be regarded as only being half filled at any point. As conduction may only occur when there are available empty states within the energy band for electrons to be accelerated to, the small (touching) gap allows for easy conversion between electron and holes, leading to the very high transport mobilities (conductivity is proportional to the product of mobility and carrier concentration).

Doping in both the n and p directions results in a relative shift in the position of the Fermi level to that of the band positions in graphene. Doping graphene, most commonly with nitrogen or boron, can result in a modified electronic structure, which can allow tailoring

27 towards specific applications.50-51 For example with additional electrons added to the graphene (e.g. through the addition of a group 5 phosphorous, arsenic or nitrogen atom) then the Fermi level is higher than the Dirac points and is referred to as being n-doped (the opposite situation is for greater holes to be present, p-doping, Fermi level below Dirac points, e.g. with boron, gallium).52

1.2.5 Electrons at the degeneracy points

Once the first BZ has been defined along with the density of electrons, the electronic dispersion can be determined. Figure 1.5 a shows the electronic dispersion of graphene, showing the pi valence band below the plane and the pi* conductance band above the plane which are noninteracting.53 These two bands touch at six points, the degeneracy regions referred to as the Dirac points. The symmetry of these positions allows the six points to be reduced to a pair, K and K’ which are independent of one another. The zero-bandgap at the K points is a consequence of the identical environment of the two carbon atoms in the graphene unit cell.16, 45, 53

At low energies, which are the most relevant in electron transport, the bands have linear dispersions, and can be viewed as a cone touching at the EDirac which for monolayer graphene is the Fermi level. The linear energy-momentum relationship was shown by Novoselov et al. in 2005 showing that the conduction and valence bands intersect with no energy gap at q = 0, where q represents wave vector.16, 43 This linear dispersion (Figure 1.5) is valid within approximately ± 0.6 eV where the electrons show an effective rest mass of zero.54 As the band structure is symmetric about the Dirac point, electrons and holes should have the same properties.

Figure 1.5. (a.) The band structure for pi- and pi*-bands in the first BZ. The electron energy band extends to the second reciprocal space boundary showing the high symmetry of the contour plots for the neighboring zones. The electronic dispersions show the trigonal warping effects away from the K- and K’- points as well as the saddle point singularity at the M point. 1.5 b shows the enlargement of the Dirac cone highlighting the linear dispersion. Reproduced with permission from the American Physical Society.43

The energy of electrons (EG) in graphene has a linear dispersion:

퐸퐺 = 휈퐹푝 Equation 1.2

Where νF is the Fermi velocity and p is the momentum. This implies that the electrons (within this linear region) behave as a zero rest-mass, relativistic Dirac Fermions. The density of states (DOS) increases linearly with energy unlike the constant DOS of 2D systems with a parabolic dispersion. The Fermi velocity can be calculated from the nearest neighbor hopping t (t ≈ 2.5 eV43) and the lattice constant a (a = 0.14 nm see Figure 1.2) via the tight binding 6 -1 approximation to give νF ≈ 10 m s , approximately 1/300 the velocity of light. This indicates the velocity at which the carriers (electrons and holes) in graphene move.

Therefore, within the Dirac cones (degeneracies at the K and K’ positions) for monolayer graphene the electrons have an effective rest mass of zero. This dictates several of the unique

29 properties of graphene from the universal optical transmission of 97.7 % over a wide range of wavelengths and the remarkable electron mobilities of 200,000 cm2 V-1 s-1.8, 10-11

The energy dispersion of monolayer graphene (Figure 1.5) is plotted in reciprocal space and shows the energy (z axis) as a function of kx and ky (x and y axis’). The dispersion presented is for that of the first BZ, however it does extend past this point slightly showing the high symmetry of the Dirac cones. The two bands (the pi and pi*) within the BZ consist of a single pi electron of graphene, and the second BZ is empty.

The electronic dispersion within the first BZ is calculated from the tight binding approximation with either a nearest neighbor approximation (lower fidelity) or both the nearest and next nearest neighbor approximation (higher fidelity). The hopping parameter is a critical parameter here as it represents the overlap between electrons orbitals from neighboring atoms. The two equivalent lattice sites (A and B) results in interesting electron hopping properties within the 2D graphene lattice, resulting in the chirality in the carrier dynamics.43

The central high energy gap position Γ, is at the furthest position from the degeneracy points and has the highest energy gap of ≈ 20 eV.37, 55 The M position, the saddle points, are referred to as the van Hove singularities and are characterised by energy gaps of approximately ≈ 4.5 eV.55-56 The region extending from the Dirac point to the van Hove position are of particular relevance for optical absorption properties of interest to this study. The significance of these points will be discussed in greater detail in Chapter 3.

1.2.6 Bandgap in bi- and few layer graphene

It is important to note that several stacking configurations become (increasingly) possible with increasing layers and importantly the stacking configuration will influence the electronic structure. The most stable stacking in graphite is the Bernal stacked ABA configuration (Figure 1.1.). Other configurations such as the ABC rhombohedral stack are less stable but can provide interesting insights into the electronic structure changes.56

AB(A) stacked bi-layer graphene, consisting of 4 atoms per unit cell, has two sets of the conduction and valence bands form with a zero-bandgap,53 however, the electrons now have an associated mass which alters the linear E-k dispersion.9 This behavior results in the characterization of massive Dirac Fermions.53 The split bands result from the stacked atoms within bi-layer graphene, hybridizing to form a higher energy band, which is separated from the lowest band by t ~ 0.3 eV, the interlayer hopping parameter.46

ABA stacked tri-layer graphene presents a small overlap of the conduction and valence bands at low energy giving it semi-metallic properties.57 Tri-layer graphene also consists of two sets of conduction and valence bands. Both bi- and tri- layer graphene can have a bandgap opened via the application of a static electric field perpendicular to the layer.46, 57-58

Interlayer coupling in few layer graphene (tetra layer) leads to a significant change to the electronic structure with increasing splits to the conductance and valence bands. However despite the increased hybridization, hyperbolically dispersing bands with no bandgap are present.56 Novoselov and Geim reported in 2004 that the few layer graphene (which ranged from 1 to 3 layers) presented a small overlap in the bandgap.49

It has been reported that a bandgap can be opened within bi-layer graphene via the application of an external electric field to destroy the inversion symmetry,59-60 leading to the Mexican hat shaped E-k dispersion.59 Additionally, the presence of adsorbates on the surface of a bi- layer graphene sheet stack can also lead to an opened bandgap.61 This was explored for materials with an adsorption energy (with respect to fractional regions of a single lattice) ranging from approximately 0.3 to 1.8 eV.61 The influence of adsorbates on the electronic structure is of particular importance within this study as the graphene sheets employed here

31 in are present with a surfactant coating. However, the surfactants are weakly adsorbed through hydrophobic-hydrophobic and vdW forces and are likely to present only minor influences to the band structure.62-63

1.2.7 Density of states

In solid state physics a state is described by the wave vector q, such that one wave vector corresponds to one state. As electrons follow the Pauli repulsion rule they fill the available states as per the Fermi-Dirac distribution. In normal conditions one k vector corresponds to one state. But sometimes even within one k the electrons can have more degrees of freedom. If an electron has different spins or is situated within different energy valleys these degeneracy factors can limit the possible degrees of freedom and may alter the possible DOS.

Figure 1.6. (a.) DOS for graphene showing the Dirac point where the DOS is 0 and the van Hove singularity which corresponds to the M points of the BZ. The DOS is calculated as a function of energy (in units of t) calculated from the energy dispersion. (b.) Zoom in of a showing the linear relationship close to the Dirac point, reproduced with permission from the American Physical Society.37

Within graphene the DOS at the K and K’ positions can be observed to degenerate (completely) when the energies are closer to zero (at zero) showing semi-metallic behavior (Figure 1.6.).37 Unlike 2D electron layers in semiconductors, the carrier mobility in graphene can remain high even when the density is completely restricted at the Dirac point.11 It is the degeneracy of states near the Fermi level that enables a continuous tuning of the carrier concentration from electrons to holes.7

To briefly review to key characteristics of the graphene electronic structure. The zero- bandgap at the K points is due to the identical environment of the two carbon atoms in the graphene unit cell.16, 45, 53 These degeneracy regions at the K and K’ positions have a linear dispersion within a certain energy range, which is indicative of the massless characteristic of the electrons in this region and as such the electrons are characterised by the Fermi velocity. It is the behavior of the electrons acting as massless Dirac fermion which results in the high transport mobilities as well as the opacity independent of wavelength in the optical region. Additionally, the total energy range of the graphene bandgap is from 0 to ≈ 20 eV providing a significant range of absorbing energy states.

1.3 Brief Introduction to the key properties of graphene

Is important to define the nomenclature surrounding graphene when the different production routes and starting materials result in highly varied end products with ranging lateral dimensions, layer numbers and oxygen content.

Figure 1.7. Schematic defining the graphene family and corresponding appropriate nomenclature. Reproduced with permission from John Wiley and Sons.64

Whilst establishing a universal nomenclature for a material group is always important to avoid confusion within a field and ensure a more directed approach to research within a specific community, in the booming field of graphene research it all the more critical due to the many types of graphene. Depending on the route of production, the end graphene product is very different, and as such, differently prepared graphene materials will have completely different properties.

The most relevant properties to consider when classifying graphene materials is the layer number, the lateral dimensions as well as the oxygen content which provides a good indication of the crystalline quality of the material. Graphene suspensions prepared via the surfactant assisted liquid exfoliation (SALE) method results in predominantly < 5 layers with a typical C:O ratio of ≈ 60:1 and typical lateral dimensions of ≈ 250 - 300 nm (Chapter 2b), therefore the term graphene microplates is the most appropriate according to this classification system. As this method produces a sheet thickness distribution, the strictly technical terminology applicable is single and few layer graphene microplates, within this study the graphene material produced via this method will be referred to as single and few layered graphene, graphene microplates, or simply as graphene. In addition, due to the convolution in the field, the term pristine is often included to distinguish from graphene oxide (GO), as well as imply the original crystalline structure of the starting material is maintained.

1.3.1 Optical properties of graphene

The optical properties of graphene are those of the most relevance within this study. The ability for graphene to absorb strongly in the NIR region opens up the ability for this material to be employed for a number of photothermal biomedical applications.

1.3.1.1 Broad band absorbance – fine structure constant / Dirac fermions

One of the key optical properties of graphene is its broad and consistent absorbance band extending out into the infrared (IR) region. The ability of graphene to non-specifically absorb makes it highly versatile for a range of applications and most importantly allows the transduction of NIR light. Radiation in the NIR has particular biological relevance due to the low absorption coefficient of water and many other biological components, for radiation of this energy, as will be covered in more depth in Chapter 3.65

The broadband absorbance of graphene is a result of the broad ranging energies of the bandgap. As discussed, graphene is a zero-gap semiconductor, which is to say that its bandgap touches at specific locations (K and K’ points), whilst there are substantial energy gaps at other positions of the bandgap (Figure 1.5).

At very low incident energies, for example far IR radiation, absorbance within the pi bond results in intraband excitations, corresponding to positions close to the Dirac points.9 As the energy of the incident radiation increases, for example to the visible region, the electronic excitations transition to interband excitations. These events correspond to positions slightly further away from the K points.9

As the incident energy increases further, to the ultraviolet (UV), the transitions move toward the M saddle point (where the bandgap energy is approximately 4.5 eV, the van Hove singularity) and the massless Dirac fermion behavior of the electrons breaks down. This can be observed within the absorbance scan as the absorbance peak at approximately 270 nm. It is this broad energy range of the bandgap which results in the broad absorbance of graphene.

As the broadly absorbing spectra of graphene is dictated by the electronic structure, any disruptions to the hexagonal carbon lattice will result in changes to the absorbance profile. This is particularly evident when comparing the spectra of graphene materials produced via chemical oxidation routes to that of a pristine graphene which gives a direct insight into the role of the defect free, fully conjugated nature of graphene on the optical properties.

1.3.1.2 Quantitative absorbance properties

In addition to the broadband absorbance spectrum, graphene also transmits a highly specific 97.7 % of incident light per layer (2.3 % opacity) independent of the wavelength within the visible and (a portion of the) IR spectrum.8, 55 This arises as a result of the electrons near the Dirac points behaving as massless Dirac fermions. Electrons with such behaviors will have a universal dynamic conductivity which results in the transmission in these regions being solely defined by the fine structure constant.8

The fine structure constant is a fundamental physical constant which describes the coupling between light and relativistic electrons. As stated in Section 1.3, the zero-effective mass of the electrons near the Dirac point result in velocities of approximately 106 m s-1, a value approximately 1/300 the speed of light. These particles are therefore appropriately described by the Dirac equation rather than the Schrödinger equation.

It is due to the behavior of the electrons near these Dirac cones that dictates the specific opacity of graphene at 2.3 %, independent of the wavelength, as well as being almost linearly additive.9 It was recently demonstrated by Zhu et al. that the additional 2.3 % opacity of subsequent layers is not strictly linear showing a subtle decrease in absorbance with increasing layers. This has been attributed to the optical conductance of the graphene stack, specifically the interlayer interactions which result in the slight decrease in opacity per graphene layer.

From this information, we can see that suspensions consisting a sheet thickness distribution, will likely present attractive absorbance properties, a critical parameter for the role of a photothermal agent.

1.3.2 Thermal properties

Along with the versatile optical properties graphene also has a very high thermal conductivity reported as high as 5000 W m-1 K-1.1-5 This is greater than for that of both graphite 390 W m- 1 K-1 (graphite has been reported as high as 4180 W m-1 K-1 in the ab direction for highly crystalline stress annealed pyrolytic graphite) and diamond 2200 W m-1 K-1.17 The remarkably high thermal conductivity of graphene is attributed to a range of contributory factors such as the reduced incoherent scattering from neighboring sheets, the contributions of the flexural mode as well as the strong bonding within graphene coupled with the low mass of the carbon atoms.3 Additionally the reduction in phonon-rough boundary scattering and phonon spatial confinement effects have been suggested to contribute significantly.1

In crystal lattices the acoustic phonons are the primary heat carriers and are considered as quanta of crystal lattice vibrations.17, 67 The transverse (TA) and longitudinal (LA) acoustic 4 -1 4 -1 phonon modes have group velocities of vTA ≈ 1.36x10 m s and vLA ≈ 2.13x10 m s which are attributed to the strong in-plane sp2 bonds and the small mass of carbon atoms.68-71 This results in ballistic, scattering-free heat flow characteristics over short distances in graphene.3

Figure 1.8. Phonon dispersion in graphene, reproduced with permission from Cambridge University Press.3

The phonon dispersion showing the relationship between E (energy) and q (wavevector) presented above in Figure 1.8 maps out the Γ-M vectors of the BZ and plots the energy of each individual phonon mode across these vectors. The energy can be compared to the frequency (Ѡ) of the phonon mode via E=ћѠ.3 The longitudinal modes (L) represent compressive waves (atomic displacements along the wave propagation). While the transverse modes (T) correspond to the shear waves (in-plane displacements perpendicular to the propagation direction).72 The 2D nature of graphene also allows for the presence of out-of- plane atomic displacements known as flexural phonons.3

In three-dimensional materials, the thermal conductivity is commonly limited by phonon scattering by other phonons from any direction or from interaction with defects in the crystal lattice, conductive electrons and interfaces. If a crystal is perfect, then the main limit to phonon propagation are phonon–phonon scattering due to crystal lattice anharmonicity.

As the number of layers of graphene are decreased the thermal conductivity increases as there are fewer incoherent phonon modes present (Figure 1.9).73-74 The thermal conductivity properties will be expanded in greater detail in Chapter 3.

Figure 1.9. Measured and theoretical thermal conduction of graphene as a function of layer number. Reproduced with permission from Nature Publishing Group.74

The flexural mode (ZA) of graphene has also been established to play a critical role in the superior thermal conductivity of graphene.3, 73, 75 The ZA mode is the easiest mode to be excited due to its low frequency and carries a large proportion of the vibrational energy.73 The thermal conductivity of graphene has been shown to be greatly decreased when the flexural mode is supressed highlighting the important role of this flexural mode.76

It is important to consider the target application of graphene within this project to be employed as a photothermal transducing agent within several biomedical systems. As such, the thermal conductivity could play a significant role in heat dissipation, but any improvement it can impart to a system will be dictated by the thermal interface resistance of each system. This will be explored in greater detail in Chapter 3.

1.3.3 Electronic transport properties

As stated in Section 3, the pi electrons in graphene are quasi-free and are shared among all atoms in the layer. The quasi-free pi electron is primarily responsible for the electrical conductivity of graphene and the high mobilities and mobility densities arise as a result of the zero-bandgap energy regions around the inequivalent K and K’ positions of the BZ. 39

It is the properties of carriers (electrons/holes) near the Dirac point which governs the outstanding carrier transport properties of graphene as the charge carriers behave as relativistic particles with a zero-rest mass and therefore approximately 1/300th speed of light.16, 37, 43

The electron mobilities reported for freely suspended graphene are as high as ≈ 200,000 cm2 V-1 s-1. 10-12 In the ballistic transport regime (scattering free) where carriers move with the Fermi velocity of graphene.53 On longer scales, the carriers undergo elastic and inelastic collisions and transport becomes diffusive.37, 53, 77 These values are decreased substantially when the mobility is tested on a substrate supported graphene flake where values decrease to approximately 20,000 cm2 V-1 s-1.78 When supported on a substrate, inelastic scattering can occur involving the surface phonons of the substrate.53, 79-80 The surface phonons generate an electric field that extends away from the surface, coupling with the graphene carriers. By selecting substrates with higher surface phonon frequencies, such as boron nitride (BN), this can be avoided and higher mobilities have been reported.53, 81

Additionally, a common limitation to carrier mobilities in crystals is through phonon scattering, as such the mobilities of a material are expected to be greater at lower temperatures where the phonon action is minimized. The mobilities within graphene are practically independent of temperature however indicating the primary limitation is due to scattering from defects.49 It does appear that with increasing layers, a greater influence can be observed from temperature changes (≈ 15,000 cm2 V-1 s-1 at 300 K and ≈ 60,000 cm2 V-1 s-1 at 4 K).49

As the conductivity is proportional to the product of the mobility and the carrier concentration, in this case both electrons and holes, it is important to consider the carrier density within graphene. The density of electrons can be calculated from the DOS (in this scenario the number of states is the number of electrons below the Fermi level, thermal excitation complicates this though). The electron (and the hole (p)) density (n) for intrinsic graphene (in the conduction band) can be calculated at 9x1010 e- cm-2 for intrinsic graphene according to Equation 1.3.82 2 휋 푘푇 82 푛 = 푝 = 푛푖 = ( ) Equation 1.3 6 ℏ휈퐹

Where k is the Boltzmann constant and T is the absolute temperature. In direct comparison the previous highest reported electron mobility was for that of Indium antimonide (InSb, approximately 77,000 cm2 V-1 s-1)83 and semiconducting nanotubes (100,000 cm2 V-1 s-1).84 It important to note that the electron mobility in graphite crystals have been reported as high as 1,000,000 cm2 V-1 s-1 at low temperatures.85-86 As the carrier mobility drops off with increasing sheet number, it is expected that there must be an optimum layer number where the mobilities will proceed to increase. One further relevant comparison is against silicon which has an electron density of approximately 1400 cm2 V-1 s-1 and a hole density of approximately 450 cm2 V-1 s-1 at room temperature. Additionally, 2D electron gasses have presented some remarkably higher carrier mobilities exceeding 107 cm2 V-1 s-1.87 The fully reduced (GO) monolayers exhibited conductivities ranging between 0.05 and 2 S cm-1 and field effect mobilities of 2-200 cm2 V-1 s-1 at room temperature.88

Unsurprisingly, with the addition of layers there are significant changes to many of the fundamental properties such as the thermal conductivity and the electronic carrier mobility. For example, bi- and then tri-layer graphene exhibit a shift towards parabolic dispersions and a corresponding decrease in mobilities where the ratio of the average mobility is approximately consistent with the ratio of the effective mass of the charge carriers.57 The presence of carriers within additional layers may lead to a many-body renormalization of the Fermi velocity of graphene.43

The small overlap between the valence and conductance bands gives few layer graphene semi-metal properties resulting in electron and hole concentrations of ≈ 1013 cm-2 (at a gate voltage of ≈ 100 V), room-temperature mobilities of ≈ 10,000 cm2 V-1 s-1 as well as very large sustainable currents of > 108 A cm-2.49 In the ground breaking paper published by Novoselov et al. in 2004 they reported the typical dependence of sheet resistivity on the gate voltage for single and few layer graphene, showing a sharp peak of several kiloohms and the 41 subsequent drop off at high gate voltages.49 From this, the conductivity of graphene can be observed to increase linearly with the gate voltage on both sides of the resistivity peak resembling the ambipolar field effect characteristic of semiconductors. However, it does not show a zero-conductance region (i.e. Fermi level between each band).

1.3.3.1 Hall effect

In a conductor with the current moving from left to right (electrons moving to the left) and a magnetic field applied perpendicular to the current, a voltage will be created in the transverse (upwards) direction. This influences the electron trajectory to shift transversely (upwards) resulting in a charge forming at the top surface of the conductor as more electrons have accumulated there in relation to the bottom surface. The electric field which results between the two surfaces to push the electrons back into an even distribution, results in the Hall voltage (VH).89

The electron mobility of graphene was shown to be practically independent of temperature suggesting that they were still limited by scattering on defects. From the electron mobility and the surface charge density the mean free path can be determined at ≈ 0.4 μm indicating that the 2D gas is at the most several Å from the surface.49 The electrons in this system can be referred to as a 2D gas as the electronic transport is strictly 2D This was confirmed via Shubnikov–de Haas (ShdH) oscillation studies which showed the linear dependence of ShdH oscillations’ frequencies (BF) on the gate voltage (Vg) indicating the holes and electrons were proportional to the concentrations.

Figure 1.10. (a.) The ShdH oscillations of graphene as a function of charge carrier concentration. (b.) The Hall voltage of graphene (left y axis) showing the linear region at low applied magnetic field where a direct response between applied Bo and hall voltage can be observed (classical Hall regime), followed at higher applied Bo with the characteristic plateaus of the quantum hall regime. The ShdH voltage (longitudinal voltage) which can be seen to oscillate as the quantum hall effect occurs. Reproduced with permission from the Nature Publishing Group.16

Each of the ShdH oscillations in Figure 1.10 a corresponds to the population of one Landau level. When the Dirac point is passed at 0, the sequence is not interrupted between electrons and holes.37 When a low magnitude magnetic field is applied to graphene, the effect on the Hall voltage is linear as the charge separation is occurring at greater extents. The longitudinal voltage can be seen to be constant at these low magnetic fields indicating that the current is not being affected by the magnetic field. This follows the classical Hall regime.

At higher magnetic fields the Hall voltage can be seen to incrementally plateau and the longitudinal voltage shows corresponding oscillations. This is representative of the anomalous integer quantum hall effect.37 The quantum Hall effect can be observed for 2D conductors as the Landau level forms and the DOS is significantly altered.

The studies of the ShdH oscillations performed by Geim and Novoselov in 2004 confirmed that the electronic transport within graphene was strictly two-dimensional. The linear dependence of the oscillations’ frequencies indicated that the Fermi energies of the holes and electrons were proportional to the concentration, which is qualitatively different to the three- dimensional dependence scenario.

휀퐹 ∝ 푛 Equation 1.4 (2D)

2/3 휀퐹 ∝ 푛 Equation 1.5 (3D)

The electrical conductivity is also strongly influenced by layer number, whereby substrate 2 -1 -1 (SiO2) supported monolayer graphene has mobilities of ≈ 10,000 - 15,000 cm V s reported under ambient conditions.7, 49 Nagashio et al. reported a decrease in electron mobility to ≈ 2,000 cm2 V-1 s-1 for 2 layers (compared to a monolayer reported here as ≈5,000 cm2 V-1 s-1) followed by an approximately even electron mobility with increasing layer number.35 It has also been shown that multilayer graphene presents a parabolic E-k dispersion compared to that of monolayer which presents the linear dispersion. Additionally, sheet resistivity increases with a decrease in layer number. At the monolayer the resistivity curve changes shape but presents a resistivity close to that of, but lower than bi-layer graphene.35

1.3.4 Mechanical strength

The atomic configuration, specifically the sp2 covalent carbon to carbon bonding,7 also leads to a tensile strength (capacity of material to withstand pulling/stretching loads) of 130 GPa and a Young’s modulus (elastic stiffness) reported as high as ≈ 1 TPa.6, 90-92 In comparison, steel has a tensile strength of ≈ 400 MPa and a Young’s modulus of ≈ 200 GPa.93 We can also consider the Young’s modulus of graphite, ≈ 0.7 - 1 TPa for the in-plane, not hugely lower than that of monolayer graphene. The in-plane sp2 bonds between adjacent carbon atoms are among the strongest in nature with a bonding energy of approximately 4.9 eV.3, 94 This is in stark contrast to the vdW forces which are attributed to the inter sheet strength, estimated at approximately 50 meV.3, 95-96

Table 1. The bond energies for relevant graphene bonds and related bond types for comparison. All values collected from Pierson, 1995.17

Type of bond Approximate bond energy Note (kJ mole-1)

p 347 CH radical

sp3 434 Methane

sp2 442 Ethylene

sp 506 Acetylene

The Young’s modulus of graphene can be directly related to the flexural phonon mode such that it is possible to extract the Young’s modulus from the thermal vibrations.73 The frequency of the flexural mode is proportional to the square root of the Young’s modulus.73

The Young’s modulus of graphene is generally measured via AFM indentation experiments on graphene drumheads with two-dimensional elastic modulus (E2D) been reported at approximately 350 N m-1.6, 92 The experimentally obtained E2D value is then converted to the

E3D Young’s modulus by dividing the force by the thickness of the graphene sheet which is determined via the distance between the graphite interspace layers 3.35 Å (or the distance from the graphene to the underlying substrate).90, 97-98 99 The relationship between layer number and the bending rigidity has been shown with the rigidity increasing with layer number.98

The in-plane Young’s modulus of graphene can counter-intuitively be increased (from ≈ 0.9 TPa to 1.7 TPa) through the introduction of defects despite predictions suggesting otherwise.92, 100 This has been attributed to the suppression of the out-of-plane fluctuations by defects.92

However, this chemical doping must be of a specific concentration and type to achieve the counter-intuitive increase in Young’s modulus. Chemically derived monolayers of graphene have been explored and a Young’s Modulus of 0.25 TPa was reported indicating a substantial decrease in the stiffness properties of graphene when the electronic structure is interrupted.101 Interestingly, it was reported that the electrical conductivity of the sheets scaled inversely with the elastic modulus, suggesting that the oxygen bridges were in fact playing a role in bond reinforcement.101

These reported values indicate that graphene is incredibly strong in-plane with a substantial flexibility. The mechanical properties of graphene are critical to understand not only from a simplistic perspective of mechanically reinforcing applications etc., but because the deformation of graphene is governed by the Young’s modulus, interfacial energy and layer numbers, and as such the electronic structure can be strongly influenced by changes to these parameters.22 Therefore understanding the Young’s modulus of graphene correctly is paramount in a vast number of applications currently being explored and must be understood correctly.

1.4 Production methods for types of graphene

With the discovery and isolation of graphene via the scotch tape method came a range of widely varying alternative production routes. These methods of production lead to subtle differences in the macro configuration and in some cases the atomic structure of the final product, with significantly varying properties. Therefore, the production method must be carefully considered for any specific target application.

There are several well established graphene production methods; (i) physical exfoliation (scotch tape method), which allows high quality sheets to be prepared but at very low yield,49, 102 (ii) chemical vapour deposition (CVD) which produces large sheet sizes but at a high price and with a low yield103-104 and (iii) liquid exfoliation of graphite which can produce larger quantities but with a weaker control over the morphology, thickness and size. Liquid exfoliation can be performed via both sonication and shear approaches or via chemical routes which come at a cost to some of the most desirable properties.105-106 Alternatively, sonication and shear methods in appropriate solvents or in the presence of surfactant, such as the method employed herein, can produce defect free sheets at high concentrations.107-109 Limitations around the control of sheet morphology and thickness occur with the shear and sonication approaches, however the small sheet sizes open up the ability to incorporate these graphene materials within a range of nano- and micro-scale systems.

These production methods can be considered as either top down approaches (micromechanical cleavage and liquid exfoliation methods), or bottom up approaches (CVD). It is well understood the top down approaches lead to products with high polydispersity of size, shape and thickness (when relevant) and require cleaning/isolation (e.g. centrifugation), while bottom up approaches can be more difficult, they often result in a higher quality material with greater control over morphology during synthesis.

The most common and highly successful “bottom-up” route towards graphene production is that of CVD whereby graphene sheets are grown on a substrate from gaseous precursors in a highly-controlled environment.

1.4.1 Chemical vapour deposition

While the micromechanical cleavage method is rightly credited with the first isolation of monolayer graphene, CVD was being extensively explored with great promise prior to the success of the scotch tape method.

CVD is a bottom up method whereby graphene sheets are grown on a substrate from gaseous precursors in a highly-controlled environment. The sheet sizes produced via this production method are impressively large opening up the possibility for larger scale electronics where both transparency and low resistance are required.110 The sheet sizes achievable are now considered limited only by the size of the (copper) substrate, a limitation which has been largely addressed and optimized via roll-to-roll production methods.110-112 This is a particularly attractive route for graphene production which can be compatible with existing planar silicon technology.

The epitaxial preparation of a material is the deposition of one material on top of another in an ordered manner. When the material lattices are the same (and therefore the lattice constants) this is referred to as homoepitaxy, when there is a misfit, it is referred to as heteroepitaxy. Heteroepitaxy can lead to defects due to strain or relaxation. CVD is a form of epitaxial growth and as such the selection of appropriate substrates is crucial for the quality of the final graphene produced.

The general approach to CVD is the introduction of source precursor gasses, usually obtained via the pyrolytic decomposition of a hydrocarbon gas, into a reactor chamber containing the sample substrate. Under high energy (heat, plasma) conditions and with a typical pressure of a few millibar (e.g. ≈ 50 mbar112), the gasses react with each other or the substrate resulting in the growth of a thin film.

Metals are commonly used as the CVD substrate catalyst with great success.17, 113-117 The process of graphene CVD can be considered as the breakdown of precursors (usually hydrocarbon gases), followed by the graphitization of the precursors.17, 118 The presence of the metal catalyst allow the graphitization reaction to occur at approximately 1000 °C111, 119 (despite the sublimation point of graphite being approximately 3500 °C120). The self-limiting approach involves the growth of a graphene layer out from nucleation points, once the surface is covered the metal is no longer available to aid in the catalysis of the graphitization and as such the growth stops at a single layer.

As the direct lattice of graphene is a hexagonal structure, growth on hexagonal crystallographic surfaces is a desirable route to high quality graphene. When a lattice mismatch of less than 1% results as with Co (0001)116 and Ni (111)115 surfaces the graphene is considered commensurate and referred to as homoepitaxial growth. When a lattice mismatch is greater than 1 %117 than the growth is considered incommensurate and heteroepitaxial as occurs with iridium, Ir (111)113, palladium, Pd (111)116 and rutile, Ru (111).114 Recently the use of copper foils has allowed the production of uniform single layer deposition of graphene providing a significant improvement to the quality of produced graphene.111-112 The low solubility of carbon in copper appears to aid in making this growth process self-limiting.112

There are of course many challenges that must be overcome in CVD production of graphene such as the presence of carbon emissions from within the substrate during the cooling phase (commonly occurring at grain boundaries) which can lead to multilayer regions.121

Some limitations associated with CVD are the difficulty of control of sheet layer numbers, the inhomogeneity of the crystalline phases as the growth of the sheets starts from different seeding points. In addition, it is also considered to be a quite expensive process. One of the key original examples was that of the deposition of ethylene onto nickel surfaces in 1979.115

Currently CVD is understood to produce high quality, large sheet sizes but at a high price and with a low yield.103-104 CVD is a highly promising route towards scaling this technology and could potentially allow a planar production method (sheets which can be integrated with our current silicon technology) to become an affordable route towards large, high quality graphene.

1.4.2 Micromechanical cleavage (scotch tape method)

The key method of production which enabled the first isolation of graphene is the micromechanical exfoliation (scotch tape) method which is capable of producing the highest quality graphene sheets, which are of tens of microns in size.49, 102 Micromechanical exfoliation or micromechanical cleavage is a “top-down” approach allowing high quality sheets to be prepared but at very low yield.49, 102 This approach to exfoliation has been used for decades by researchers primarily in the field of crystal growth and crystallography.122-123 This method starts from a piece of bulk graphite (usually highly oriented pyrolytic graphite, HOPG) and through the use of pure mechanical forces, the delamination of individual graphene layers is achieved.

It was this method which resulted in the first isolation of graphene and the awarding of the Nobel Prize in Physics in 2010 to Geim and Novoselov. A key aspect of this initial isolation was the enabled visualization of the monolayer graphene by contrast against the 300 nm thick silicon wafer (SiO2/Si) that it was deposited on. The micromechanical exfoliation method has since been optimized to yield single layer graphene on the order of millimeters.124 This method is still credited as being the best route for producing the highest quality graphene, in terms of structural integrity, particularly when prepared from HOPG.118 By ensuring the parent material is of high quality, and by avoiding introducing any damage to the graphene crystal lattice, the highest quality monolayers can be achieved via this method. Therefore, sheets prepared via micromechanical exfoliation are the most appropriate choice for fundamental studies of the properties of monolayer graphene such as the remarkably high electrical and thermal conductivities and the tensile strength measurements. However, this production method is incredibly slow and therefore the amounts producible are very low, the parent material of raw graphite is however very cheap, except for when using HOPG.

As with any “top-down” approach there is little control over the shape or size of the graphene exfoliated. This method is highly suited for preparing sheets to explore the fundamentals of graphene but produces negligible yield, uncontrollable sizes, and must of course, be handled to some extent for use, not an insignificant challenge for a material consisting of only one layer of atoms.

1.4.3 Liquid exfoliation of graphite

Throughout this project, a liquid exfoliation method was employed to prepare the graphene sheets used, as such, it is on this method that a greater focus shall be applied.

Graphene exfoliation can be achieved in a liquid environment via a number of varying strategies, but primarily via the ultrasonication of starting bulk graphite materials, followed by a cleaning/isolation centrifugation stage. There are several alterations to this method which either target a pretreatment of the starting bulk graphitic material, such as via intercalation or oxidation, or to minimize the interfacial tension of the graphite/solvent interface as per the organic solvent and surfactant approaches.125-127

1.4.4 Graphene oxide (Hummers method)

Chemical exfoliation is well understood to be capable of producing large quantities but at some cost to the most desirable properties.105-106 One such graphite pretreatment approach prior to sonication is to oxidize the graphite. As graphite becomes oxidized the interlayer spacing is expanded approximately 2 or 3 times, which decreases the interlayer cohesive forces, allowing for easier separation of sheets. GO has been extensively explored and shown to have many interesting properties such as being luminescent under continuous wave irradiation,128 as well as presenting broad photoluminescence under visible excitation.129 GO is an insulting material with resistances of (RS) ≈ 1012 ohm square-1,130 and also has a high specific surface area attributed to the 2D morphology.

GO and rGO are two graphene materials which have been in use and explored long before the isolation of pristine graphene via the scotch tape method (more than a century).131 GO is most commonly prepared via the revised Hummers method which is an extensive process which oxidizes graphite and in turn separates out the individual graphene layers.132 This results in a graphitic structured material which is decorated with various functional groups (primarily comprising of hydrogen and oxygen). The Hummer’s method uses sulfuric acid

(H2SO4), sodium nitrate (NaNO3) and potassium permanganate (KMnO4) to oxidize the graphite.132

The clear disadvantage to this method is that the pristine continuous conjugation within the graphene sheets is lost upon the oxidation. This loss of conjugation leads to the loss of the remarkable electrical and thermal conductivities and the remarkable tensile strength.92, 130 Additionally, GO is a semiconductor with an energy bandgap. The greater the extent of oxidation, the larger this bandgap becomes, this can be useful for many applications but can also limit the suitability of GO for other roles.133

GO sheets are hydrophilic and as such stabilizing agents are not required which can make suspension handling far easier, with no requirement for additives to either sterically or electrostatically stabilise the suspensions. However, due to the disruption to the electronic structure the absorbance of GO out into the IR is significantly decreased compared to that of the starting material.

The production method of GO is well established and GO has many applications where it is highly suited such as for light emitting devices134 and bio-imaging.129 However, as the route of reducing GO returns many of the attractive properties of intrinsic graphene, it is certainly highly advantageous to perform this secondary chemical processing stage.

1.4.5 Reduced graphene oxide

Some of these properties lost through the oxidation process can largely be recovered by reducing the material after preparation but not to the extent of the pristine starting material. A common route to reducing GO is through treatment with hydrazine hydrate or through thermal reduction.88, 135-138

The reduction of GO does recover a substantial extent of the attractive properties, but not completely. The absorbance, for example, is largely recovered through reduction, but does not completely recover to the pristine state. Additionally, rGO is often (depending on extent of reduction) no longer hydrophilic and as such requires a stabilizing agent to be added. Sizes of GO or rGO flakes are usually approximately on the micron scale,101, 129 however, alterations to the production procedure can make the sheets far smaller.66

The conductivity of the GO sheets is also recovered through reduction and conductivity values of approximately 350 S cm-1 have been reported.139 Additionally when rGO is prepared into thin films the conductivity has been reported as high as approximately 1300 S cm-1.140

The oxidation and reduction methods should not be discounted as they can allow large quantities of highly exfoliated graphene sheets to be produced. GO is also very easily dispersed due to its inherent charge imparted by the extensive oxidation which can be a key benefit for many applications. GO reduction methods have improved dramatically on original GO properties and the method is can produce a high yield whilst keeping cost of production low. As this is a top-down approach the sheet sizes are of irregular morphology and require cleaning after preparation.

1.4.6 Liquid exfoliation – solvent approach

Liquid exfoliation is one graphene production method which can produce very high quantities of non-oxidized, small sized, single and few layered sheets of minimal defect content.107, 141 Liquid exfoliation relies on the selection of an appropriate liquid medium with an approximately matched surface energy to that of the solid phase. Such solvents will minimize the interfacial tension between the liquid and the graphene flakes, effectively lowering the energy required for the two materials to be in contact.142 If the interfacial tension is too high, the cohesive energy between graphite sheets will dominate and the graphite/graphene sheets will tend to adhere to each other. By matching the surface energies of both the graphite (interlayer) and the liquid phase the interfacial tension is minimized and the energetic barrier for separation in minimized.142

Graphite exists as layers of carbon atoms whereby the individual layers are primarily held together by vdW forces. These vdW forces are relatively weak compared to the intrasheet bond energies and can be overcome easily when the surface energies are matched to effect separation of these layers. The surface energy of graphite is considered as the cohesive energy between each sheet layer and has been reported with a significant range in literature, centering around ≈ 70 mJ m-2.138, 143-146 Hernandez et al. and a number of other groups have reported higher concentrations of graphene being produced when the solvent surface tension is approximately 41 mN m-1.107, 147 This surface tension corresponds to a surface energy of approximately 70 mJ m-2 (assuming the solvent surface entropy to be 0.1 mJ m-2 K-1), indicating that the surface energy of graphite (water/graphite) is approximately 70 mJ m-2.107

Figure 1.11. (a.) Schematic representing the cohesive energy of solvent and (b.) that of the vdW forces between the sheets.

Once the surface energies are approximately matched, the exfoliation is achieved using mechanical forces in suspension, such as via ultrasonication, and the shear forces produced via the mechanical displacements (pressure waves) of the sonication. This results in the ultimate separation of the graphene sheets from the original graphitic stack.

Common methods target intercalation (commonly with alkali metals) to expand the graphene

(and other 2D analogues such as molybdenum disulfide, MoS2) sheets prior to introduction of the mechanical forces.148-151 One of the first examples of liquid exfoliation of graphene like materials involved an intercalation process and was reported in 2003 (a year earlier than Geim and Novoselov’s seminal micromechanical cleavage reports).152 However, minimizing the interfacial tension via solvent selection or surfactant addition is another simpler, and in some ways, far more effective route to achieving liquid exfoliation.105, 107, 125, 153-154

This method does not produce a significant extent of defects in the sheets (discussed further in Section A.2.7) allowing the attractive properties of graphene to be maintained. However, the sheet sizes produced do limit some properties on a larger scale, such as, the electrical conductivity which is substantially limited due to needing to cross edge boundaries. The sheet thicknesses or the number of layers produced will cover a range from single out to approximately 10 layers, and whilst capable of producing significant concentrations, the 55 exfoliation rate is not as high as for surfactant exfoliation in an aqueous environment.109 Additionally, consideration must be made as to the handling of the prepared suspensions. Ensuring aggregation is avoided is a key consideration and stabilizing agents can play a critical role, allowing month scale stability.105, 109, 141

1.4.7 Surfactant assisted liquid exfoliation

Preparing graphene suspensions in organic solvents presents several significant drawbacks associated with the cost, toxicity and the difficulty of removing the (often) high-boiling point solvents. The liquid exfoliation method can be altered slightly to allow the exfoliation to occur in aqueous conditions. This can be achieved through the addition of surfactant to alter the surface energy of the solid phase following initial separation via sonication. The SALE method will be explored in greater depth in Chapter 2, however, in this section, a broad overview of the most relevant factors will be provided.

The surfactant role in assisting the exfoliation follows that of the role of organic solvents in that the interfacial tension is minimized allowing easier creation of new surfaces, in this case, the exfoliation of graphene sheets. Upon the initial small-scale exfoliation/separation of sheets, the surfactant adsorbs to the new surface with hydrophobic section of the surfactant is adsorbed, with the polar group extended into solution. Therefore, the interfacial tension between the liquid and the graphene/graphite surface (with adsorbed surfactant present) is now dictated by the surfactants interaction with the water. As this value is quite low, ≈ 5 mJ m-2, the work required to create new surfaces is reduced significantly from that of the graphite/water interfacial tension. As such, the energy required to effect the exfoliation of sheets and the creation of new surface is lowered, readily allowing delamination upon exposure to the intense, high energy ultrasonic waves.

Following this mechanism, two key aspects must be considered. The first is that the initial separation of the sheets must occur prior to any intercalation of the surfactant or organic solvent. The second consideration is that of the mechanism of the surfactants in resisting sheet restacking events.

Once a sheet has been separated, if only for a brief period, and even if the separation is incomplete, nearby surfactant molecules can adsorb to the surface of the sheets on a very fast time scale, preventing restacking. Graphene sheets produced via this method are subsequently coated in surfactant which allows facile dispersibility and strong stabilization in solution. The presence of the surfactant on the surface of the exfoliated graphene can be detrimental for several roles, such as electrical transport in films, photocatalytic roles etc., however, the surfactant also opens a wide range of opportunities explored within this project.

1.4.8 Relevant properties of SALE graphene

There are several key benefits to the SALE production method which make it well suited for photothermal applications and in general a promising route towards scalability of graphene production. The SALE method can produce very high concentrations of exfoliated graphene with quantities of up to 15 mg mL-1 having been reported.109 Additionally, this method can be applied to a range of layered materials including MoS2, BN, tungsten disulfide (WS2) and bismuth telluride (Bi2Te3) opening up a vast range of opportunities (important to note that liquid phase exfoliation in general, as well as several other methods, can prepare this range of 2D materials, not solely SALE).

By continuously adding surfactant during the sonication process the concentrations can be increased significantly. By selecting surfactants which do not contain an electrolyte co-ion, the surfactant concentrations can be increased significantly without resulting in a salting out effect. Further, many surfactant types are being broken continuously under ultrasonication, by adding fresh surfactant throughout, the desired liquid phase conditions can be maintained.

In addition to the high concentrations achievable via the SALE method, this exfoliation procedure is environmentally friendly in that it avoids the use of harsh organic solvents such as n-methyl-2-pyrrolidone (NMP), and allows the selection of safer (and cheaper) surfactants to be employed in water. The sediment graphite can also be recycled allowing a highly efficient use of the starting material.107

The previous comments are related more to the scalability of the SALE method; however, surfactant exfoliation of graphene also allows a high level of fine tune control of the produced graphene sheets.

The surface properties of surfactant exfoliated graphene can be controlled through the selection of different surfactants. The surfactant types can generally be considered from the following perspectives, non-ionic versus ionic, large or small as well as containing functional groups or functional group free. Selections can be made to alter the surface charge, the biocompatibility, the achievable concentration as well as to influence the particle size and the stability in suspension. From an applications perspective, the surfactants may be selected to influence the electrical and thermal properties, or the ability to solvent exchange or even surfactant exchange.

In addition to the controllable surface properties the absorbance profile for pristine graphene is quite strong out into the IR enabling potential in situ biomedical roles. The improved thermal properties achieved through exfoliation and the maintenance of the intact defect free sheets additionally allows any heat produced at the site of the sheet to be dissipated to shift from local to bulk heating of a system. Further simply the size of the sheet produced allow incorporation to small scale systems such as within porous gel networks or lipid nanostructures or on the surfaces of emulsion droplets.

The potential cytotoxicity of carbonaceous materials is well established as being highly biocompatible, with applications in thin carbon film coatings for implants having been extensively explored along with a range of other study focuses.155-156 Additionally extensive studies have reported the improvements to particle biocompatibility,157-158 stealth159 and bi specific targeting160-163 via adsorbing polyethylene oxide (PEO) groups onto the surface of (nano)materials. This is a strong advantage of this production method of graphene, allowing the tailoring of surface properties, for example by preparing the suspension with large, non- ionic surfactants.

With these considerations taken into account, the SALE method is the ideal route towards graphene preparation for in situ biomedical photothermal transduction to convert NIR into intense hot spots at a specified location to either induce drug release or for the thermal ablation of cancerous cells.

1.5 Photothermal principles

When considering the remarkable properties of graphene, several of the inherent properties can be seen to be well-suited towards being exploited in tandem. One such pairing is the optical and thermal properties of graphene. As graphene absorbs a wide range of wavelengths and the extent of this absorption can be manipulated with sheet thickness, the optical properties can be tuned to allow efficient absorption of a wide range of wavelengths, including the biologically relevant NIR. In addition, the sheet size can be altered to potentially allow greater heat transfer within a system. With these somewhat tunable optical and thermal properties, a biologically relevant photothermal role appears to be well suited to the material.

1.5.1 Targeted photothermal applications

The applications explored within this study aim to make use of the intrinsic properties of SALE graphene, as such a focus on two specific roles emerged. The roles were (i) to employ graphene to photothermally ablate cancerous cells, and (ii) to embed graphene within several drug delivery systems and photothermally activate drug release. These applications are explored throughout this manuscript along with the fundamental analyses of the photothermal ability of SALE graphene.

1.5.2 Photothermal therapy principles

Photothermal excitation refers to the irradiation of a photothermal agent with an appropriate wavelength of light such that the molecules are excited and the following relaxation leads to the release of thermal energy.164 Interest around photothermal materials has steadily increased over recent years, particularly for materials capable of absorbing NIR radiation for biomedical applications due to the deep penetration through tissue achievable in this wavelength region. The recent explosion of nanotechnology has allowed a number of challenges to be addressed, opening up potential applications including biomedical imaging,65 drug delivery,165-167 cancer cell diagnostics and cancer therapeutics,65, 166, 168-169 many of which may work in unison with photothermal triggering methods.169-170

1.5.3 Photothermal material and radiation selection

NIR is a particularly appropriate wavelength region for biomedical photothermal applications due to its ability for deep tissue penetration.65 Water and hemoglobin are major absorbers of the visible light, acting to limit its penetration into the biological tissue (Figure 1.12). Each of these materials have their lowest absorption coefficient in the NIR (650 – 900 nm) positioning NIR as an appropriate light source to reach the nanoparticulate transducers.

Figure 1.12. Biological absorbance of NIR radiation. Reproduced with permission from the Nature Publishing Group.65

When selecting a material to act as a potential photothermal agent, an important consideration is the biocompatibility as well as a strong NIR absorbance capability. Two materials that satisfy several of these requirements are graphene and the well-established benchmark photothermal agent group, gold nanoparticles (GNPs).

1.6 Nanomaterials as photothermal agents

In the last thirty years, hyperthermic therapy and the photothermal agents employed, have become an established discipline, with both a comprehensive range of applications and functional materials studied. This area has seen many promising results and in some cases been taken through to clinical trials.171 One of the key advantages associated with photothermal therapy (PTT) methods, is the highly specific nature of the treatment and therefore lower required dosage levels, reducing potential side effects. PTT is also considered to be a more affordable therapeutic method as it does not require high-powered expensive biomedical equipment.172

A significant range of photothermal agents have been explored in recent years, including the highly successful GNPs, CNTs, GO and rGO, and polymeric and metallic nanoparticles 61

NPs,173 gold containing compounds and gold colloids.169, 174-176

1.6.1 Gold nanoparticles

GNPs have properties that position these materials as being suitable agents for PTT. These include the high absorption efficiency without the effect of photobleaching and tunable localized surface plasmon resonance (SPR) via morphological changes.166, 168, 173, 177 SPR resonance is the event of charge density oscillations which are confined to metallic nanoparticles and metallic nanostructures.177 The biocompatibility of GNPs was an initial limitation, however coating strategies are being explored to reduce the significant cytotoxicity that has been reported by some groups.173, 178-179

Figure 1.13. GNPs absorbance with morphology reproduced with permission from Elsevier.180

As for all materials reduced in size into the nanoscale, the surface area to volume ratio is increased dramatically compared to the bulk counterpart, often having effects of increased reactivity along with quantum confinement effects. GNPs in particular have shown a significant increase in light scattering and absorption efficiency (up to 4-5 times) when brought down into this size scale.173 Due to the advantageous scattering properties, gold nanospheres have been employed for cancer cell imaging via confocal microscopy and dark field microscopy.169, 181

Manipulations to the morphology of the gold nanomaterials can directly alter the materials absorbance characteristics. Gold nanorods (GNRs) with suitable aspect ratios can absorb and scatter strongly in the NIR region.169 Gold nanomaterials with specific shapes e.g. –rods, - stars, -cages and shells, present altered localized SPR properties and subsequently are suited towards alternative potential applications. It is due to the tunable absorbance characteristics that GNPs have been considered the benchmark for photothermal applications.

The tunable absorbance properties of GNRs is well understood and does position these particles as being highly suited towards photothermal roles. The absorption coefficient (at 785 nm) of GNRs with an aspect ratio of ≈ 3.5 was reported as 4.6 x 109 M-1 cm-1 by Orendorff et al. in 2006 compared to that of exfoliated graphene reported at 5 x 106 M-1 cm- 1 by Paton et al. in 2016.182-183 The significantly greater (≈ 1000 x greater) absorbance of the optically optimized GNRs is limited to some extent by some toxicity factors,184 as well as the requirement of such high aspect ratio conformations.185

One of the problems which arises from the use of the anisotropic GNPs is the tendency for the morphologically altered materials to revert to spheres upon heating losing the old morphology specific properties.186-188 This change in morphology often occurs at high temperatures such as boiling (100 °C) and may still be functional as photothermal agents as the thermal ablation temperature is stated at 40 C.186

1.6.2 Gold nanoparticles as photothermal agents

In 2005 Huang et al., reported on successful studies into gold nanorods (GNRs) in both cancer cell imaging and PTT applications. In each circumstance NIR irradiation of GNRs was

63 carried out in vitro.169 The GNRs were conjugated with anti-epidermal growth factor receptor (anti-EGFR) monoclonal antibodies and demonstrated to selectively kill cancer cells leaving the healthy cells unaffected.169 By conjugating the GNRs with antibodies to selectively target the tumour cells, damage to non-malignant cells was reduced dramatically whilst maximizing the concentration of photothermally active agents at the cancer site.169 Targeting the EGFR molecule selective delivery to solid tumours such as brain, breast and lung (which are known to over-express this cell receptor) can be achieved.169

An example of a cancer treatment via GNR PTT has also been demonstrated by Chou, et al., who showed a novel specific delivery system that could provide stable storage of the GNRs. By establishing a functionalized chitosan-conjugated non-ionic surfactant-based nanocarrier system, GNR delivery showed improved tumour targeting and decreased liver uptake.165 This has considerable advantages as the in vitro cellular uptake and subsequently the photothermal effect was enhanced. This group went on to show that intravenous injection of this GNR carrier system followed by in vivo NIR laser irradiation resulted in efficient thermo-ablation in vivo.165

Among the wide variety of PTT agents studied, GNPs have emerged as one of the more promising agents. Gold compounds have had a long history in medicine, for example colloids of gold being employed for rheumatoid arthritis.189 GNPs have had a lot of success however there are several caveats associated with the use of these nanomaterials. The prominent concern is that GNPs have demonstrated severe toxicity in biological samples which has been addressed to some extent.178-179 Further concerns include build-up of nanoparticles in both the environment and within organs, the inability of the human immune system to remove or breakdown specific nanomaterials,190-191 potential membrane damage through free radical formation, as well as a general hesitation surrounding the booming, relatively unknown field of nanoparticle technology.

In response to these caveats, coatings are being developed to reduce the potential risks, however the protective capabilities of these coatings remain to be fully understood and while these coatings may reduce the danger of these materials, they can also act to reduce the efficacy of the therapeutic agent.190-191

The intrinsic optical properties of GNPs, along with the successes in areas such as coating and conjugation have all contributed to positioning GNPs as the current benchmark in the growing field of nano-oncology.

1.6.3 Graphene derivatives as photothermal agents

There are a range of materials other than GNPs which have been explored in recent years as photothermal agents including several graphene derivatives. One such graphitic photothermal agent explored in recent years is GO due to its significant optical absorption in the NIR region. In a study by Tian et al., in 2011 polyethylene glycolated (PEGylated, PEG = PEO) GO was shown to be an efficient photothermal agent demonstrating successful photothermal ablation of tumours in animal experiments.174 The motivation of this report was to exploit a carbonaceous material, with anticipated low toxicity, and employ the NIR absorbance for NIR transduction.

One further graphene derivative was explored as a photothermal agent by Robinson et al., who showed RGO to be a more efficient NIR photothermal agent than its oxidized counterpart.66 To reduce the radiation power required to achieve photothermal ablation with GO, a highly-reduced GO was prepared and studied. The PEGylation was derived by chemically reducing covalently PEGylated nanosize GO, partially restoring the polyaromatic structure seen in graphene sheets, in order to increase the NIR absorbance by more than six- fold.66

Figure 1.14. NIR absorbance increase with conjugation, reduced GO versus GO.66 Reproduced with permission from the American Chemical Society.

This report by Robinson et al. was a key step in establishing graphene materials as an effective NIR photothermal agent. As discussed in Section 1.4, by altering the production method of the graphene sheets to a method that avoids the loss of the aromatic structure initially, the NIR absorbance can be further improved. It is therefore critical to ensure that the production method will not impart defects into the graphene sheets during production, to allow lower concentrations to achieve sufficient heating and reduce potential cytotoxicity.

A group of additional graphene derivative materials explored in recent years as potential NIR photothermal agents are single and multi-walled CNTs.192-194 CNTs are an allotrope of carbon and take the form of a cylindrical, rolled up graphene sheet(s).33 In 2005, Kam et al., demonstrated that SWNTs could be used as both biological transporters and selective photothermal ablation agents capable of targeting cancer cells.195 CNTs have been shown to absorb strongly within the NIR range with an absorption coefficient at 808 nm of ε = 4.72 mL mg-1 m-1.195

As with GNPs, there are many graphene analogue (layered) nanomaterials which may be explored for these roles. One such is MoS2 which has a significant optical absorbance in the NIR region (absorption coefficient at 800 nm of ε = 3350 mL mg-1 m-1).196

Figure 1.15. Absorbance profile of exfoliated MoS2.

Several reports have shown MoS2 can be employed successfully within this role both in vitro and in vivo.197-198 However, some questions remain around the biocompatibility of the material consisting of a heavy metal, although its natural presence in the human body indicates that this is a material worth pursuing. The issue of potential cytotoxicity of these materials is of critical importance and must be explored thoroughly, and all ambiguities must be addressed before any further steps can be made.

1.6.4 Nanomaterials, cytotoxic or biocompatible?

The biocompatibility behavior of nanomaterial photothermal agents is a crucial factor for any material considered for biological applications. Carbonaceous materials are considered to be well tolerated by animal cells and as such graphene, comprising solely of carbon atoms, with mostly uncharged bulk properties and with no heavy metals present, can be assumed to be likely non-toxic.156 Further as biocompatibility is controlled predominantly by the interface of the biomaterial and mammalian cells, the behavior of an exfoliated material will likely 67 present similar properties to the bulk counterpart.156 However, as will be discussed further in Chapter 4, the morphology of materials exfoliated down to single and few layers plays a significant role in the interactions with mammalian cells.

These important factors, along with several other considerations are explored in greater depth in Chapter 4. Within this chapter, the extent of toxicity of SALE graphene is explored and compared against the not-insignificant amount of data available in literature to date. The mechanisms and trends associated with nanoparticle toxicity are explored along with alternative agents, potential routes towards optimization as well as an experimental in vitro assessment of the photothermal ability.

A2.0 Chapter 2A - Graphene exfoliation theory

All graphene suspensions used throughout this study were prepared by surfactant assisted liquid exfoliation (SALE), with subtle variations to target more favorable surface chemistries, sizes or concentrations. This chapter explores the theory and practice of this production method, followed by the graphene characterization theory and the most relevant experimental data.

As there are minor differences in the graphene preparation techniques employed for each of the biomedical applications explored within Chapters 3,4,5 and 6, data representative of all suspensions is presented within this chapter.

A2.1 Introduction to surfactant assisted liquid exfoliation (SALE)

Exfoliation of a bulk, layered material in a liquid phase can be readily achieved by selecting solvents which minimize the interfacial tension between the liquid medium and the graphene sheets, effectively lowering the energy required to create new surfaces.107 This can be achieved via a number of methods, primarily by exfoliating in the presence of a solvent with an appropriate surface energy, such as N-methylpyrrolidone (NMP).107 One such liquid exfoliation strategy is to adjust the surface energy of the graphite/graphene through the addition of surfactants. Within this study the surfactant role in altering the surface energy to enable efficient exfoliation was employed to isolate single and few layers of graphene from bulk graphite.

Once the interfacial tension has been minimized, the exfoliation is achieved using mechanical forces in suspension. The mechanical forces in suspension are commonly introduced through ultrasonication, and the shear forces produced via the mechanical displacements (pressure waves). SALE is capable of producing very high quantities of minimally defected, non-oxidized, small sized, single and few layered sheets.107, 141 This method avoids the use of expensive, high boiling point and often toxic organic solvents, as well as allowing the surface chemistry of the exfoliated particles to be controlled through surfactant selection.

A2.1.2 Bulk graphite

Graphite exists as layers of carbon atoms whereby the individual layers are primarily held together by van der Waals (vdW) forces. The SALE method exfoliates individual layers away from the bulk material to isolate single and few layer graphene.

In order to understand the mechanism of how these sheets are separated, we must first review the forces that are binding these layers together. VdW forces refer to a broad range of mechanisms which dictate attractive intermolecular forces. The three primary mechanisms are the weak London forces, dipole-dipole forces and the dipole induced dipole forces.

London dispersion forces, first described by Fritz London in 1930, are temporary attractive forces that results when the electrons in two adjacent neutral atoms (or molecules) occupy positions that make the atoms form temporary dipoles. At any given moment, a neutral molecule will have a dipole moment due to fluctuations in the electron distribution in the molecule. The electric field of this dipole configuration will then polarize neighboring molecules inducing an additional dipole moment. Despite this force being very weak, and only present over a very small range, it accounts for a very large proportion of intermolecular attractive forces. This force is sometimes called an induced dipole-induced dipole attraction.

The second of the two vdW forces briefly addressed in this manuscript is the dipole-dipole forces occur between permanently polarized molecules where the positive end of one molecule is attracting the negative end of a neighboring molecule.

The third of the vdW forces is the dipole induced dipole scenario. In this event the configuration of the electrons of a molecule at a particular moment in time can lead to the dipole influence on a nearby molecule. Is can in turn induce a dipole to be formed within that molecule which in turn influence a neighbor, and so on and so forth.

These three mechanisms are relatively weak interactions in respect to an individual event, however these events become of greater significance due to the large number of these interactions. It is these forces in addition to others including hydrophobic-hydrophobic and pi stacking, that lead to the interlayer forces of layered vdW solids such as graphite. It is important to note that this is not an all-encompassing review of the interlayer forces of graphite, but these are highly relevant mechanisms by which the layers are held together.

A2.1.3 Minimizing the interfacial tension with surfactant

As briefly mentioned earlier, the SALE method relies on minimizing the interfacial tension between the liquid phase and solid phases. When the surface energy of the graphene/graphite has been lowered sufficiently via the adsorption of surfactant, the exfoliation can occur readily allowing isolation of single and few layer graphene sheets. To understand this mechanism, we shall first review the key concepts of the surface chemistries most relevant to the separation event.

Cohesive energy, often referred to as binding energy, refers to the energy required to convert one constituent molecule into a completely independent molecule. For example, to separate out a water molecule from all surrounding water molecules. The energy required to break all bonds with surrounding water molecules corresponds to this value and we can therefore consider the cohesive energy as the inter-molecular energy.

The cohesive energy of a solid refers to the energy to completely separate out every atom from all surrounding atoms. However, throughout this study when the cohesive energy of layered materials such as graphite, and some other two-dimensional (2D) materials are referred to, the cohesive energy referred to is representative of the energy required to separate individual layers of the respective materials apart.

Surface energy is the work per unit area (J m-2) done by the force that creates the new surface and the work of cohesion is 2 x the surface energy. The surface energy is often best described through the example of the liquid air interfacial surface tension. Surface tension arises due to an imbalance of forces at the air liquid interface whereby the attractive forces of neighboring molecules influence the interfacial molecules from below. This leads to a net contractile force within the liquid causing as small a surface area as possible. Surface energy and surface tension can be considered as interchangeable when referring to a liquid air interface.

An interface is the boundary region between two adjacent bulk phases and if one of the phases is the vapour or a gas, it is referred to as a surface. Matter within the few molecular diameters that constitute the width of an interface has different physical properties and energy characteristics to that of the bulk and must therefore be considered carefully.

From this surface chemistry perspective, the exfoliation event can be described starting from the fundamental equations:

The first law of thermodynamics can be expressed as:

∆푈 = ∆푞 + ∆푤 = ∆푞 − 푃∆푉 Equation 2.1

Where U represents internal energy, q represents heat, w is work, P is pressure and V is volume.

And the Second law with entropy S and temperature T:

∆푞 ∆푆 = and is therefore ∆푞 = 푇∆푆 Equation 2.2 푇

The second law can be substituted into the first law:

∆푈 = 푇∆푆 − 푃∆푉 Equation 2.3

Then as:

퐺 = 푈 − 푇푆 + 푃푉 Equation 2.4

So:

∆퐺 = ∆푈 − 푆∆푇 − 푇∆푆 + 푃∆푉 + 푉∆푃 Equation 2.5

∆퐺 = 푇∆푆 − 푃∆푉 − 푆∆푇 − 푇∆푆 + P ∆푉 + 푉∆푃 Equation 2.6

∆퐺 = −푆∆푇 + 푉∆푃 Equation 2.7

To give the fundamental thermodynamic equation where G represents the change in Gibbs free energy.

For an open system where material can enter or leave the system the change in Gibbs Free Energy when molecules are added to the system is given by:

∆퐺 = ∑푖 휇푖∆푛푖 Equation 2.8

Where µ is the chemical potential of species i and n is the number of particles of species i added to the system.

Additionally, as we are considering changes in surface area, ∆A, the energy changes associated with creating or removing interfaces must also be considered. The energy per unit area of an interface is given by the surface energy γ. Therefore, the energy change associated with changes in interfacial areas for each interface, j is given by;

∆퐺 = ∑푗 훾푗 ∆퐴푗 Equation 2.9

This is the thermodynamic definition of surface energy. When this is included, the fundamental equation becomes:

∆퐺 = 푉∆푃 − 푆∆푇 + ∑푖 휇푖∆푛푖 + ∑푗 훾푗 ∆퐴푗 Equation 2.10

In a closed system, in the absence of chemical reactions, at constant pressure and constant temperature, only one term remains.

∆퐺 = ∑푗 훾푗 ∆퐴푗 Equation 2.11

If we now consider the physical process of exfoliation in a solvent. This involves the loss of graphene-graphene interface and the creation of graphene-solvent interface. For each unit area of graphene that is exfoliated, two units of graphene-solvent interface are formed. Thus, the free energy associated with exfoliation of graphite in solution over an area A is;

∆퐺 = 훾퐺푆 × 2퐴 − 훾퐺퐺 × 퐴 Equation 2.12

Where, 훾퐺푆, is the surface energy of the graphene-solvent interface and 훾퐺퐺 is the surface energy of the graphene-graphene interface.

Therefore, minimizing the interfacial tension between the graphite and the solvent, maximizes the thermodynamic favourability of exfoliation, noting that 훾퐺퐺, is determined by the material properties of graphite and cannot be experimentally manipulated.

Figure 2.1. (left) Schematic describing the energy required to create new surface. (right) Free energy associated with exfoliation of graphite in solution over an area A (fixed creation of 1 m2).

This approach provides a strong description of the system requirements and energetics from a basic thermodynamic perspective. However, the general consensus of the scientific community working in this field is generally to approach the event by describing the enthalpy of mixing which does appear to describe the events reasonably well.

It has been widely reported that the liquid exfoliation method is attributed to the small net energetic cost during the exfoliation process. The energy balance for the graphene and solvent system can be best described as the enthalpy of mixing per unit volume as initially described by Hernandez et al..107 This description (see Equation 2.13) shows that when the graphene and solvent surface energies are approximately matched, the mixing enthalpy will be at a minimum. This allows exfoliation to occur more easily.

A2.1.4 Sonication to effect exfoliation

Once the enthalpy of mixing has been minimized through an approximate match of the surface energies of each phase, separation can be readily achieved via sonication. The exact mechanism of how the sonication effects the exfoliation of a single graphene layer from a 77 graphite stack is unclear, however there are two primary theories. The first theory is that the separation results due to shear forces caused by sonication and the second is that cavitation events, on or near the particle surfaces, results in separation via a number of suggested mechanisms. Throughout this chapter the strong evidence supporting the shear exfoliation via sonication mechanism will be explored. Regardless of the exact mechanism, sonication has been shown to be highly effective at exfoliating graphene under these described conditions.

The first consideration must be to explore how this process is intialised. The initial separation must be occurring via a mechanism completely separate from the roles of the surfactant or the organic solvent, these agents make the process easier energetically, but do not initialize it. As the total area of the sheets in contact dictates the strength of the sheet attraction (areal interfacial binding), at sheet edges the energy of such a connection is lower, and the sonication events (be it cavitation or shear forces) will have a greater result here. The initial separation, where the overall driver is of course the sonication, should be considered at a small scale, such as in Figure 2.2, where the join of two sheets at the sheet edge is shown.

Figure 2.2. Initialization schematic of the exfoliation mechanism and progression of separation along join. Graphite / graphene sheets labelled with honeycomb structure for identification purposes.

The opening of an initial graphite sheet separation then allows for surfactant/organic solvent molecules to enter the gap. This leads to the lowering of the surface energy of the graphite sheet in the case of surfactant and in both scenarios, a low interfacial tension. The lowering of the interfacial tension, in addition with the now broken (or decreased) cohesive forces between the graphite layers (as the neighboring region of the sheets become separated, the total average area energetic attraction is lowered), leads the progression of the sheet separation along the join/faces of the graphite sheets.

The restacking scenario is quite simple, the surfactant molecules are adsorbed to the newly created surface, and therefore either electrostatically repel neighboring sheets or simply sterically prevent restacking (as discussed in Section A2.4). In the organic solvent scenario, the restacking events may occur more readily, possibly explaining the lower yields. The energetic motivation for restacking is lowered by the decreased interfacial tension, and therefore the sheets must orient closely enough, then have the solvent expelled from between the sheets to allow the restacking.

A2.2 Enthalpy of mixing

As mentioned briefly, the balance of these forces can be best represented by calculating the enthalpy of mixing as defined by Hernandez et al..107 This equation accounts for both the energy required to separate all molecules (solvent and graphene sheets) and the energy to bring them back together, as a graphitic dispersion accounting for the reduction of sheet thickness. See Equation 2.13.

∆퐻푚푖푥 2 2 107 ≈ (훿퐺 − 훿푠표푙) 휙 Equation 2.13 푉푚푖푥 푇푓푙푎푘푒

Where ∆Hmix is the enthalpy of mixing, Vmix is the volume of the mixture, the factor of two is required as two surfaces are being accounted for, the solid graphite flakes and the solvent.

Tflake refers to the thickness of the final product, for example if the final thickness is to be a

79 monolayer, the Tflake value is 0.335 nm. The volume ratio 휙 is that of which is contributed by the graphite sheets and the δG and δsol represent the surface energies for each respectively.

To see the relationship between the set surface tension and the optimal exfoliation conditions, a slightly expanded equation can be considered with the mass (MG), flake size (T) and density (ρ) of graphite.

푀퐺 퐺 푠표푙 퐺−푠표푙 107 ∆퐻푚푖푥 ≈ 2 (퐸푠푢푟 + 퐸푠푢푟 − 2퐸푖푛푡푒푟 ) Equation 2.14 푇2휌퐺

G Here we can see that the surface energy of the graphite (E sur) is added to the surface energy sol G-sol of the solvent (E sur) and the interfacial energy between the two is subtracted (2E inter). If we consider a neat water scenario, the surface tension will be 72.8 N m-1, corresponding to a surface energy of approximately 100 mJ m-2. The surface energy of the graphite sheets can be assumed to be approximately 70 mJ m-2 and so we see a positive enthalpy emerging. EG- sol inter represents the solvent-graphite binding energy per unit area. This endothermic parameter of the equation can be considered as the energy associated with the new surface area available and the increased interaction between the solvent and the solid. The interfacial energy can be accounted for by considering that the interactions are predominantly dispersive and therefore the geometric mean approximation can be made as follows199:

퐺−푠표푙 퐺 푠표푙 1/2 199 퐸푖푛푡푒푟 ≈ (퐸푠푢푟 ∗ 퐸푠푢푟) Equation 2.15

The heat of mixing can then be observed to increase as the solvent surface energy proceeds to go lower than the surface energy of the graphite (Figure 2.3). The further away from matched surface energies, the greater the solvent-graphite binding energy becomes. The solvent surface energy is determined from the surface tension via Equation 2.16:

푠표푙 푆표푙 107 훾 = 퐸푠푢푟 − 푇푆푆푢푟 Equation 2.16

-1 sur Where 훾 represents the surface tension in N m and E sol is the solvent surface energy. The sur -2 - S sol value represents the solvent surface entropy where an approximation of 0.1 mJ m K 1 is used.107, 200 The T value represents temperature in kelvin.

This has been demonstrated several times by showing the extent of exfoliation as a function of surface tension, whereby the values corresponding to the enthalpy of mixing minimum show the maximum concentrations produced.107, 147 However, it is important to note that it is the surface energy which is of importance, not the surface tension. The surface energy can however be determined from the surface tension for the organic solvent mechanism.

A2.2.1 Key parameters of mixing enthalpy

With an understanding of the enthalpy of mixing and its influence on the exfoliation process, the importance of controlling a range of parameters including liquid surface energy, temperature, extent of exfoliation and the starting concentration of bulk graphite can be readily demonstrated. The most relevant parameter is the surfactant concentration for SALE, which is easily monitored via surface tension.

The enthalpy of mixing equation can be plotted as a function of surface tension, temperature and the final flake thickness to give a clear indication of the trend and show the optimal conditions for exfoliation (Figure 2.3).

Figure 2.3. (a.) Enthalpy of mixing of graphite to monolayer graphene at 25 °C as a function of surface tension. (b.) Enthalpy of mixing minimum as a function of temperature at a fixed surface tension of 41 mJ m-2. (c.) Enthalpy of mixing as a function of final sheet layer mono- , bi-, and tri-layer, at a fixed surface tension of 41 mJ m-2 and at 25 °C.

Figure 2.3 a shows the trend of enthalpy as a function of surface tension, with a clear minimum at ≈ 41 mN m-1. The addition of surfactant molecules can be understood to adsorb onto the surface of newly created graphite/graphene surface, lowering the interfacial tension. As the surface energy of the water is dropped closer to that of the interlayer forces, the enthalpy of mixing decreases as the work required to create new surfaces is reduced significantly. At the point where the two phases have a surface energy of approximately 70 mJ m-2 (which corresponds to an organic solvent surface tension of 41 mN m-1), the minimum is reached. The continued deviation away from this point as the surface energy of the water is decreased past that of the solid, results in a greater interfacial tension and an increase in energy required to mix. With this information, the enthalpy of mixing as a function of temperature can be readily determined with a clear trend identifiable at the energetic minimum (Figure 2.3 b). This is simply representing the change in surface energy as a function of temperature and therefore a deviation away from the energetic minimum at room temperature. This is included to highlight the susceptibility of the method to temperature changes and the importance of maintaining a fixed temperature during exfoliation in order to achieve reproducible results.

The relationship between the extent of exfoliation can also be easily demonstrated by adjusting the final thickness of the exfoliated material as shown in the example in Figure 2.3 c as the enthalpy of mixing calculates the energy required to separate out each layer and then mix all layers within the liquid phase, have an increased thickness (for example a bi-layer), the energy of mixing will be lower.

It is important to point out that exfoliation can be achieved outside of the minimum region implying that the equation is a good guideline for exfoliation but not extremely robust. Further, a range of additional parameters will strongly influence the yield achieved such as flask shape, uniformity of sonication probe, temperature fluctuations both in the bulk solution and on a smaller scale in terms of cavitation along with several other parameters. The wide range of variables and the possible exfoliation with small deviations outside of the optimal conditions, could account for some of the (somewhat) contradictory reports showing successful exfoliation outside of conditions providing the energetic minimum.

A2.3 Practical parameters of exfoliation

Whilst working from a position that the enthalpy of mixing provides a robust starting point for further considerations, the exfoliation conditions do not need to be strictly confined to the minimum net energy to achieve substantial exfoliation. The temperature can be shifted considerably from optimal values with good results as can the liquid phase surface energy. As such practical considerations should therefore be made.

Surfactant concentration does not strictly need to be maintained, the temperature does not need to be strictly controlled, the sonication power can also be varied substantially whilst still achieving good exfoliation. However, to achieve reproducibility these parameters must be understood and controlled.

The sonication source power and the output power are two important factors of the exfoliation process and should therefore be monitored to understand subtle differences between batches. The source power can be measured at the generator and was automatically displayed by the Q700 Qsonica. The output power can be approximately determined by measuring the heating rate of a known volume. This is only a very approximate determination which does not consider equilibrium with the room temperature or any power changes within individual experiments due to changes to tip surface, temperature or pressure.

As can be seen in Figure 2.4, the bulk temperature of the sample increases dramatically over a short period of time if not controlled, showing the importance of controlling the temperature during sonication. This measurement was performed in 200 mL of MilliQ water with a 13 mm flat tip probe driven by the Qsonica 700 unit at an amplitude of 60 %. The source power was recorded throughout this sonication period to show the decrease in required source power to achieve the constant amplitude of 60 % as the temperature increases.

Figure 2.4. (a.) Water sample sonication with no temperature control showing the deviation of temperature (blue squares) and source power (red circles) over the sonication time with the room temperature plotted (green triangles) for comparison. (b.) The same experiment but with temperature control at the optimized sample preparation conditions.

As sonication introduces a large amount of mechanical energy into the sample, a large portion of this introduced energy is converted to heat. As the surface energies (liquid and solid) fluctuate with temperature, it is critical to control the temperature of the sample during sonication time. Additionally, an increase of bulk temperature will decrease the working energy of the sonicator to achieve the set amplitude as the water is more energetic at higher temperatures. By maintaining the bulk temperature, we can largely address this problem. However, localized heating is unavoidable due to effects such as cavitation. It is very difficult to address these events during the preparation process, and it is also important to take them into consideration as the occurrence of localized high temperature and pressure regions could easily lead to oxidation of the materials being prepared. However, the product shows no indication of oxidation and therefore the cavitation events are not leading to significant oxidation.

The enthalpy of mixing equation does appear to give a reasonably robust guideline to the mechanisms of the SALE method, with a reasonable correlation of the minimized interfacial tension with a spike in concentration. However, several factors complicate this including reports from several groups showing spikes at different concentrations, the increased surface area upon exfoliation which leads to surfactant adsorption and removal, the different 85 chemistries of surfactants which can limit the production (for example counter ions for some ionic surfactants). Many of these factors will be discussed throughout this chapter.

A2.4 The (dual) role of surfactants

The targeted role of the surfactant molecules to minimize the interfacial tension has been discussed previously, however in addition to this initial role of enabling the exfoliation, the surfactant molecules also stabilise the suspensions via electrostatic or steric repulsion mechanisms. Additionally, the varying behaviors and properties of the surfactants employed within this study are of high importance and must be covered to gain a complete understanding of the systems explored.

A2.4.1 Surfactant types employed

A range of different surfactant types were used within this study with the majority of experimental systems employing triblock copolymers (Pluronic®). Triblock copolymers consist of extended regions of polypropylene oxide (PPO) and polyethylene oxide (PEO) (Figure 2.3). The PEO segments are considered to be hydrophilic and the PPO segments are hydrophobic.

Figure 2.5. Schematic of non-ionic triblock copolymers predominantly employed within this study, ethylene oxide and propylene oxide components.

Monomeric ionic surfactants were also employed for specific roles within this study including the preparation of thin films and for atomic force microscopy surfactant exchange sample preparation methods.

The primary surfactants employed were the larger non-ionic surfactants F108 and F127 due to the excellent stability properties, the larger graphene concentration range achievable, as well as the lack of any counter-ions present with ionic surfactants. Table 2.1 presents the surfactant types employed within this study, listed with several key properties of each.

A2.4.2 Surfactant behavior

A surfactant is a surface active molecule that has both a polar and non-polar section and may be charged or uncharged. These molecules will preferentially concentrate at surfaces and interfaces, giving the definition of being surface active molecules. This is a dynamic process with a balance being reached between the molecules tendency to adsorb at interfaces or to mix in solution due to thermal motion.

At an air/liquid interface, a surfactant molecule will orient itself with its non-polar portion in the air phase as the surface energy of the polar region is much greater than the surface energy of the non-polar region. Surfactant molecules will always orient in such a way as to minimise the interaction energy.

Upon the initial preparation of the graphite water mix, the graphite does not remain suspended and quickly sediments to the bottom of the vessel. The addition of dissolved surfactant quickly aids in forming a suspension of the smallest graphite particles (upon mixing), indicating the surfactant is adsorbing to the graphite, and aiding in its interaction with the water.

The adsorption of surfactant to graphite occurs via the hydrophobic segment of the surfactant molecule to the basal plane of the graphite. The hydrophilic segment then extends into solution either providing stability to the sheets via electrostatic (ionic surfactants) or steric repulsion (non-ionic surfactants) (Figure 2.6).

Figure 2.6. Adsorption schematic of triblock copolymers to graphene.

A2.4.3 Suspension stability – Coulomb repulsion

The DLVO theory (named after Derjaguin, Landau, Verwey and Overbeek) describes the mechanism for stabilization of colloidal suspensions. It consists of two primary components, attractive forces which arise from vdW interactions, as well as repulsive forces which rise from electrostatic interactions and ultimately osmotic repulsion.

The greater the surface charge of a particle the larger to electric field effect extends into solution and the larger the sphere of influence is. Within this region of influence there will be several factors determining the distribution of the ions resulting in the formation of the electrical double layer. At a small distance from the particle, a high concentration of electrolytes of opposite charge will form. Slightly further away from the particle, a higher concentration of both positive and negatively charged electrolytes will be present (Figure 2.7). This is a dynamic distribution of salts in reality, however, we can consider the electrolytes to travel with the particles.

Figure 2.7. Schematic of the electrical double layer showing the charge density on the left, and the distribution of positive and negative ions around the colloid. Reproduced from www.zeta-meter.com.

As two particles approach, each surrounded with a high electrolyte concentration region, the two electrical double layers overlap, resulting in a further increased concentration of electrolytes. This causes water to be driven into the region between the particles via osmotic pressure which forces the particles apart and results in a stable suspension. If the salt concentration is raised significantly then the relative increase in salt concentration observed upon the overlap of the two approaching particles is not as great and therefore the osmotic pressure is not as strong. When the magnitude of the particle surface charge is lower and the Debye length (length of particles electric field) shorter, the net attraction of the vdW forces overcome the repulsive forces, leading to particle aggregation and suspension collapse.

In aqueous solution, many colloidal particles have the ability to develop a charge at the surface. This is often due to functional groups whether originally present such as negatively charged oxygenated sites at the edge of a graphene sheet, or can be induced via adsorbed charged species from solution. For SALE graphene, the surface charge of the sheets can be controlled through surfactant selection, to be positively, negatively and largely uncharged particles. The surface charge or surface potential will then influence the distribution of the charged groups, such as electrolytes in solution.

For a graphene suspension prepared with an ionic surfactant, such as cetyltrimethylammonium bromide (CTAB) or sodium dodecyl sulfate (SDS), the DLVO theory largely explains the stability properties. Further, the addition of many such surfactants, such as CTAB, results in the substantial increase of salt through the counter ion (bromine), which can lead to suspension destabilization.

The properties of the ionic surfactants will dictate the total charge of the intrinsic (slightly) negative charge of the graphene sheets. This must be carefully considered when dialyzing any suspensions as the removal of a positive ionic surfactant will bring the surface potential through from a net positive charge to a net negative charge. Therefore, the suspension will experience a minimum charge magnitude as it passes through the isoelectric point which will also lead to suspension collapse.

A2.4.4 Suspension stability - steric stability

By employing large triblock copolymers (Pluronics®) the surfactant molecules also act to prevent re-aggregation as the PPO segments are adsorbed onto the basal plane and the large PEO groups extend into solution, sterically preventing suspension collapse.201 The PPO segments of the surfactants adsorb onto the hydrophobic basal plane of the graphite/graphene sheets providing the anchor point. The PEO segments then extend out into solution, and it is these segments which sterically prevent nearby sheets from coming into close contact and prevent aggregation.

Similar to the electrical double layer (EDL) scenario of two approaching and overlapping particle, in which the increased electrolyte concentration leads to osmotic repulsion, in this scenario it is the increased polymer concentration / decreased water localized concentration which induces the osmotic repulsion.67

A2.4.5 Surfactant selection for targeted applications

When surfactants are considered for a specific role, a range of factors must be addressed. The long term stability of the suspension will be strongly influenced by the surfactant chemistry and several studies have shown that the stability of the large non-ionic surfactants is particularly good.

Simple considerations ensuring that there is sufficient concentration of surfactant must be made as well. There are several key factors which will depend strongly on the concentration of surfactant, ensuring a sufficient coverage of the particles and avoiding excessive counter ion concentrations.

As the available surface area of the graphite/graphene particles increase dramatically with exfoliation, the concentration of surfactant must be set such that there will be sufficient coverage of particles post exfoliation. Limitations around the final concentrations are observed if this concentration is not appropriately set.

The critical micelle concentration (CMC) is the concentration at which the surfactant molecules begin to self-assemble (into micelles), as it becomes more entropically favorable

91 for the surfactant molecules to orient such that the hydrophobic segments to interact with each other, rather than with the liquid. The surface tension can be adjusted readily below the CMC, but above this concentration changes to the concentration will not result in significant alterations in the surface tension. This is an important consideration as above the CMC, excess surfactant will not contribute to exfoliation, other than acting as a reservoir, and should therefore be avoided in most scenarios. Many of the non-ionic surfactants have broad concentration ranges (with appropriate surface tensions) away from the CMC points which enables a wider range of concentrations to be employed.

It has been well established that PEO, which is also named polyethylene glycol (PEG) is a highly versatile material with excellent biocompatibility. Many studies have shown that PEG can both aid in improving the biocompatibility of nanoparticles as well as aid in stealth strategies, avoiding physiological defense mechanisms. As the triblock copolymers employed within this study consist largely of PEO, it is anticipated that the attractive properties of PEO will be applied to the graphene sheets. A more extensive discussion will be made as to the biocompatibility of the materials explored in Chapter 4.

Ionic surfactants are well understood to present greater cytotoxicity to mammalian cells and as such are only explored as a comparison to the nonionics. However, the smaller surfactants do allow for far easier dialysis and filtration. As such they were explored for use in thin film preparation.

As discussed briefly in Section A2.4.1, the non-ionic surfactants do not contain any counter- ions and as such there is no risk to excessive salt concentrations when large quantities of surfactant are added. For this reason, as well as the high suspension stability, the anticipated improvements to biocompatibility and stealth strategies as well as the broad regions away from the CMC at appropriate concentrations, the non-ionic surfactants were also employed within this study in a slightly altered production technique involving the continuous addition of surfactant. This allows a simple alteration to the production method to be employed to achieve dramatically larger concentrations. A2.4.6 Continuous addition of surfactant

The continuous addition of surfactant is one approach whereby the depletion of surfactant from solution either from adsorption to ever increasing surface area of product, or the degradation of surfactant, can be addressed, allowing high quantities of graphene to be prepared.

This method relies on a surfactant being employed which does not have any significant influence on the solution conditions within a practical concentration range. The tri-block copolymers are quite appropriate for these roles as they provide excellent dispersibility properties and do not introduce counter-ions into the suspension. The tri-block copolymers can be dissolved at high concentrations allowing highly concentrated suspensions to be prepared. Therefore, the concentration can be set, and an appropriate addition rate (both rate and concentration must be considered) with little concern about over or under addition. In the scenario where the rate is too low, the production would simply be slowed, and at times where the addition is too fast the rate may not be optimized but the primary concern would simply be excess surfactant.

The presence of excess surfactant may not have any negative influence depending on the application. However, the uncertainty of the ratios of free surfactant to adsorbed surfactant increase the challenge of controlling the system tightly which is necessary for the experiments performed within this study. Therefore, for the majority of situations, a single addition method was used to allow a complete understanding of the suspensions prepared.

The continuous addition method does provide several considerable advantages for suspension preparation, primarily in the increased concentrations achievable. As briefly mentioned in Section A2.4.2, the adsorption of surfactant to graphite and graphene is an important consideration as during exfoliation there is a dramatic increase in surface area. The continuous addition method can be used to readily address the removal of available surfactant

93 molecules simply by adding more. The continuous addition method therefore operates with an assumption that the surfactant is always maintained at, or above the target concentration.

Further, the extreme conditions within the suspension during ultrasonication required to separate the graphitic layers, can result in surfactant degradation. The triblock copolymers will therefore no longer be contributing in the same manner to decrease the graphite/graphene surface energy and are in fact quite difficult to control and predict, as the ratios of sizes and possible chemical species are highly variable. Careful analysis of the final product concentrations and sheet characterization data (such as zeta potential) allows an estimate to be made as to the extent and significance of such events. Results largely indicated that surfactant degradation was not having a significant influence during the production and can largely be ignored. One situation where the degradation (of the triblock copolymers) was significant was in regards to the formation of toxic byproducts which can be addressed by simple dialysis (Chapter 4).

Once all key parameters such as temperature, concentration etc., have been controlled, the length of sonication time during continuous addition of surfactant can be as long as desired.

A2.5 Sonication - shear exfoliation

From the enthalpy of mixing equation, the energy required to separate the sheets from one another has been defined and the role of surfactant to achieve said matching has been discussed. The next consideration in the production method is the mechanical forces effecting the sheet separation in solution in relation to the sonication. The key parameters of sonication will be briefly discussed and then the introduction of energy into the system via sonication will be linked to the shear forces experienced by the graphene particles resulting in exfoliation.

A2.5.1 Ultrasound

It is important to understand the mechanism of the ultrasound in order to understand its role in the exfoliation of graphene.

Sound is transmitted through physical mediums such as water as longitudinal mechanical waves composed or regions of high and low pressure. The regions/periods of high pressure are referred to as compressions and periods of low pressure are called rarefactions. The distance between two successive compressions equals the wavelength. The pressure amplitude of the wave is the maximum change in pressure from the equilibrium value and can be referred as the intensity of a soundwave. The frequency of soundwaves refers to the number of complete waves per second with the SI unit of Hz or hertz.67, 202

Soundwaves can travel through any material medium, however, the speed depends on the properties of that medium. The speed of sound in water (at 25 °C) is 1493 m s-1.202

The sonication process starts with the conversion of mains voltage into high frequency electrical energy (generally 20 kHz) via the sonicator generator. The ultrasonic transformer or transducer then converts this electrical energy into mechanical vibrations of a fixed frequency which are directed down the probe or horn. The displacement of the tip, induced by the vibrational energy transferred down the probe, creates the sound waves within the liquid medium at the frequency of the vibration. Ultrasound is considered to be frequencies of 20,000 Hz or greater. The extent of the displacement of the probe tip will dictate the intensity of the sound wave produced. The amplitude refers to the longitudinal displacement of the probe and is presented as a percentage of the maximum. For most sonicator systems the amplitude value is the primary method of controlling the intensity of the produced ultrasound.

The high frequency and high pressure sound waves propagate out from the sonicator tip, passing over; and around; and buffeting the graphite particles. When the sheet orientation is appropriate the soundwaves pass across the surface, resulting in a high energy shear force parallel to the sheet face. The waves act to peel individual layers off from the neighboring layers, effecting the isolation of single and few layer graphene.

The shape of the probe is important in determining the output intensity. Both the dimensions of the entire probe, as well as the shape of the tip face, are important factors. The shape of the probe (in terms of length and taper) will result in an equal vibrational energy, in the top wider region of the probe, to be localized in a smaller region and increase the spatial intensity of the produced sound waves. Therefore, the tip face and probe morphology dictate the intensity profile of the ultrasound output and the required energy to achieve the set amplitude. For example, the larger the tip face of the sonicator the greater the working power required to achieve the set amplitude value.

The output power is difficult to accurately maintain as wear to the sonicator tips result in a different contact face and therefore a different energy being introduced into the system. For all suspensions prepared within this study a replaceable tip was fitted and exchanged when visible wear could be identified. As discussed previously, the output power also fluctuates with temperature and pressure however the vessels used for sonication were all open to atmospheric conditions.

A2.5.2 Sonication to shear

Sonication is well established as a successful method for exfoliating particles, but the mechanism by which this occurs is not completely clear. The primary mechanism suggested is that the sonication leads to shear forces parallel to the sheet. This event is aided by the platelet morphology of the graphite can either orient to allow the wave front to pass over (parallel) or across (perpendicular) the sheet, or for a specific sheet/stack to be buffeted along and eventually cycle back to the region of high intensity sonication.

Shear is a type of strain and occurs when two adjoining layers of a solid slide relative to each other and in a direction parallel to their surfaces of contact.67 To calculate the average shear σ experienced by two such faces:

퐹 휎 = Equation 2.17 퐴

Where F is the force applied and A is the cross-sectional area of material area perpendicular to the applied force. A Newtonian fluid moving along the solid boundary of a material will incur a shear stress on that boundary, in this scenario we are considering the shear stress experienced by the boundary of a graphitic surface as fluid moves past it. As the fluid moves past the stacked sheets the velocity differential between from the bulk fluid rate, the rate closer to the surface of the sheets and the rates above and below the sheets will result in an induced shear stress.203

The shear strain experienced by the material is proportional to the strain rate and viscosity of the fluid. The applied stress can be related to the shear rate γ given with the viscosity η by Newton’s law:

휎 = 휂훾 Equation 2.18

To give:

퐹 = 휂훾퐿2 Equation 2.19

We will assume for all shearing calculations that the surface energy of the liquid phase has been set to approximately match that of the solid phase, (i.e. surface tension 41 mJ m-2 and surface energy 71.8 mJ m-2). The sheets can then be considered at various states of stacking / overlapping and the L represents length, to provide the area of the sheets.

Paton et al. determined a calculation which allowed the minimum shear required to achieve separation which can be calculated via:

2 푠표푙 퐺 [√퐸푠푢푟 − √퐸푠푢푟] 훾 = Equation 2.20203 휂퐿

As with the enthalpy equation we can assume the surface energies of graphite and the adjusted solution (surfactant addition to a minimized interfacial tension) are 71 and 71.8 mJ m-2. For sheet exfoliation of length 250 nm the shear force required to achieve delamination is estimated at ≈ 1.01 x 104 s-1.

Now the enthalpy of mixing and the shear force required to exfoliate a single sheet have been defined along with the output power from the sonicators, we can attempt to join the two to understand the exfoliation achieved from sonication. We have determined that upon adsorption of surfactant, the interfacial energy in minimized, at which point the enthalpy of mixing indicates that several Joules would be capable of effecting exfoliation. The shearing equations indicate that the shear stress required to achieve this exfoliation is approximately 1.01 x 104 s-1.203

A very approximate connection can be drawn between the frequency of the ultrasonication, 20,000 kHz or 2 x 104 S-1, and the minimum shear stress required determined by Paton et al. of 1.01 x 104 s-1. The frequency of the sonication cannot be directly converted to a shear force, however, the considerably larger frequency can be assumed to achieve shear forces within solution exceeding that of the minimum required rate.

A2.6 Cavitation

As discussed briefly, soundwaves are transmitted through water as mechanical waves. The waves compress and stretch the molecular spacing of the medium through which it passes. When the negative pressure is large enough, the distance between molecules of the liquid exceeds the minimum molecular distance required to remain intact, resulting in the creation of voids. This event is referred to as cavitation and can lead to vapour cavities with internal pressures of several thousand atmospheres and temperatures of several thousand degrees kelvin.204-205 The cavitation process involves the rapid expansion and compression of the bubbles until a collapse occurs. It is at the point of collapse that the extreme conditions occur releasing of a high powered shockwave. These intense but small scale events must be taken into consideration as they can lead to rapid probe tip face wear and could be affecting the sample in several ways.

As the tip face is constantly changing (on a very small scale) due to cavitation at the surface and general wear from extended sonication periods, the reproducibility of batches, while not incomparable, is rarely exact and post processing of the exfoliated batch is always required in terms of achieving identical concentrations and comparable sheet sizes and sheet thickness distributions.

The high intensity sonication region is highly localized with the intensity dropping off rapidly both radially and axially. Therefore, sonication batches of large volumes such as 200 mL, the procedure relies on the bulk agitation of the medium. Whereby the intense sonication mixes

99 the liquid continuously bringing source material into the active region. The positioning of the sonicator tip is also an important factor. If the tip is positioned close enough to the base of the reaction flask, the active sonication region will be altered, partially reflected and potentially lead to regions of increased or decreased intensities or dead zones. As such, the sonication probe was positioned at the same height, and the same reaction flask was employed for all exfoliations.

Figure 2.8. Image showing the approximate high intensity region of the sonication in water.

A.2.6.1 Cavitation mechanism of exfoliation

In addition to the shear force exfoliation mechanism, cavitation events have been attributed to the exfoliation of such layered materials during sonication. It is important to note that cavitation events will likely be occurring during shearing at the rates described by Paton et al..

During the sonication, cavitation-induced bubbles in close proximity to the sheets, which upon collapsing release micro-jets and shockwaves which cause compressive stress waves within the graphite stacks. Once the compressive stress wave spreads to the free interface of graphite a tensile stress wave will be reflected back into the body leading to intensive tensile stress within the graphite stack. When intensive low pressure regions (sucking discs) are present, the combination of the tensile stress within the sheet and the “normal force” pulling the top layer away from the stack result in exfoliation.206 It has also been suggested that micro-jets may split the graphite flakes in a wedge-like fashion.206

A2.7 Defect discussion

The liquid exfoliation technique is said to produce defect free ‘pristine’ graphene sheets. This statement refers specifically to the absence of defects in the basal plane of the sheets, both as the presence of oxygen and as disruptions in the crystal lattice. This is particularly relevant when considering the suitability of this material (and production route) for optical applications.

Possible defects in graphene may be present such as disclinations, dislocations or as a result of grain boundaries.207 Dislocation and grain boundaries are of a more common consideration in this space, where dislocations can occur due to the absence of an atom resulting in non-six membered rings. Grain boundaries are the ultimate interface between single-crystalline grains in such polycrystalline materials.208 Additionally, the presence of oxygen or any functional group within the plane is considered to be a defect as these variations to the localized chemistry will alter the material properties. As liquid exfoliation is understood to not significantly alter the chemistry of the sheets, the presence of any such defects is attributed to the starting graphite material.

The sheet edges of graphene will not present the same atomic configuration as within the plane and commonly have functional groups present, these groups subsequently contribute a negative charge to the sheets.209 The liquid exfoliation process has been recently shown to introduce a greater extent (than initially suggested) of defects in the produced graphene sheets (as a collective), however, the location of such defects, edge or basal, remains to be convincingly determined.

101

Several key papers have emerged recently which are beginning to address the extent and location of such defects, a much needed resolution in this field.210 Such as the work performed by Bracamonte et al. in 2014.211 This group reported a convincing Raman study showing an increase of the ID/IG ratio, a Raman peak intensity analysis representing disorder (see Section B2.2.1.4), as a function of sonication time. This does not specifically point to edge or basal plane as they make clear. This effect could also simply be due to a reduction in size with sonication time, which then creates more edges, causing such an increase in the

ID/IG to occur.

An additional, and perhaps the strongest report towards basal plane defects was made by 212 Cancado et al. in 2011. This group focused more specifically on the ratio of the ID/IG with respect to the G peak FWHM. This strategy relies on the width of the G band increasing with bulk disorder, while the increase of edge quantity does not. The data reported by this group is perhaps the strongest argument for the presence of basal plane defects being introduced by sonication, however, not all together convincing due to the potential experimental error of the ion bombardment and the challenges of knowing the raster area and the ion beam current as pointed out by Pollard et al. in 2014.213

Another commonly reported study presenting defects specific to the basal plane is an STM analysis of graphene thin film containing liquid exfoliated graphene by Polyakova et al. in 2011.214 The key data presented within this study was to show defective / functionalized sites via STM topographic images within the graphene film. However, when careful attention is payed to the sample preparation, the nature of the samples, which were provided by an external group,154 can be understood as likely being overlapping sheets within the thin film, not individual graphene sheets. Therefore, the presence of defects assumed to be basal cannot be distinguished from graphene edges.

A study by Skaltsas et al. in 2013 approached the problem from an XPS perspective.215 The data presented therein showed a significant presence of a number of oxygenated groups, however, this study did not appear to distinguish between edge and basal plane locality. It has been demonstrated in several studies that under higher power and longer sonication conditions, that edge defect concentrations are greater.210

It is difficult to analyse with absolute confidence whether defects are being introduced within the plane for such small sheets as are produced via liquid exfoliation. However, the data presented to date indicates that there is likely some presence of basal plane defects induced via sonication (cavitation), but in populations sufficiently small so as to minimally influence the optical properties of the product.

103

B2.0 Chapter 2B - Graphene exfoliation and characterization

Chapter 2b transitions to a focus on the characterization of the graphene microplates. This chapter will briefly cover the practical procedure of exfoliation, followed by an explanation of the theory of each characterization method employed and the representative data obtained.

B2.1 Typical preparation procedure

The typical preparation method is outlined as follows. 200 ml of MilliQ filtered water is transferred to a glass beaker with an in-built cooling jacket. Graphite flakes obtained from Sigma Aldrich were then added at a concentration of 0.5 % w/w and the surfactant concentration was set at 0.1 % w/w, a value previously determined to be of sufficient concentration to sufficiently lower the graphite/solvent interfacial tension. The sample was allowed to sit until all surfactant was completely dissolved. A Julabo 900 W Frigomix U chiller was employed for suspension sonication, the chiller temperature was set to 7 °C and the sample temperature was allowed to equilibrate prior to sonication. This allowed for a constant temperature of 25 °C to be maintained during sonication periods.

The primary surfactants used within this study were SDS, CTAB, and the Pluronics F108, F127, L64 and P123 (see Table 2.1 for surfactant composition). All surfactants were purchased from Sigma Aldrich and used without further purification.

105

Table 2.1. Surfactant composition with most relevant specifications compiled from references 216 and 217-219 and experimental data.

Surfactant Composition Molecular Surface Tension PEO CMC at ID Weight at 0.1 % (25 °C, (wt. %) 25 °C (wt. (Da) N m-1) %)

CTAB C19H42BrN 364.45 ≈ 38 - 0.03

F108 PEO141:PPO44:PEO141 ≈ 14,600 41 80 4.5

F127 PEO101:PPO56:PEO101 ≈ 12,600 41 70 0.7

F68 PEO78:PPO30:PEO78 ≈ 8,400 50 80 0.3

L64 PEO13:PPO30:PEO13 ≈ 2,900 43 40 1.6

P123 PEO20:PPO70:PEO70 ≈ 5,750 34 30 0.03

SDS C12H25O4SNa 288.38 60 - 0.2

The sonication time of 4 hours was selected for most experiments as the observed production rate was linear within this region allowing consistency in terms of graphene concentration, remaining surfactant concentration and particle size of suspensions to be prepared.

Samples were then sonicated for 4 hours at an amplitude of 60 % with a 13 mm flat tip probe with replaceable tip. At the end of the sonication period the sample was collected and centrifuged at 3000 relative centrifugal force (rcf) for 10 minutes. After centrifugation, the supernatant was collected avoiding the sedimented, incompletely exfoliated materials. This was performed by pipetting the centrifuged suspension, avoiding the bottom ≈ 20 % of sample.

This procedure typically produced concentrations of 0.02 mg mL-1. However, concentrations of graphene suspensions can vary significantly with minor alterations to the production procedure depending on surfactant identity, surfactant concentrations during exfoliation as well as sonication power and period.109, 220

The selection of MilliQ water allows controllability of salt concentrations when required but it is not strictly necessary in all occasions. When a large triblock copolymer has been selected, the stability properties allow for small variations in salt concentrations with no significant effect. The salt concentration, and therefore the use of MilliQ filtered water was important when the more electrolytically sensitive ionic surfactants were employed.

The temperature is maintained for a number of reasons some of which were discussed in Section A2.3 The surface energy of the liquid medium fluctuates significantly with temperature and would dramatically influence the graphene yield. It is also important to avoid significant evaporation due to high liquid temperatures causing changes to the concentrations of the individual components. Further, it has also been briefly explored that oxidation will be encouraged at higher temperatures which would in turn result in a loss of the attractive properties targeted.

The initial concentration of graphite selected is far in excess of the minimum required quantity, however it has been demonstrated that having a larger initial concentration of graphite can lead to a greater yield of graphene.221 Ensuring the graphite concentration was not so high as to result in extensive adsorption of surfactant, altering the solution conditions.

B2.2 Suspension characterization

After the graphene suspensions have been exfoliated and centrifuged a range of characterization methods are employed to confirm the product properties. These techniques include determining the concentration of graphene, Raman spectroscopy, UV-visible absorbance spectroscopy (UV-vis), atomic force microscopy (AFM), Transmission and scanning electron microscopies (TEM/SEM), dynamic light scattering (DLS), zeta potential,

107

X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS) as well as small angle neutron scattering (SANS).

B2.2.1 Raman spectroscopy

The primary characterization method for the determination of graphene sheet thickness is Raman spectroscopy, due to the ease of analysis (when compared to AFM) and the unambiguous information which can be obtained from collected spectra.

Raman spectroscopy is the measurement of inelastic (Stokes and anti-Stokes radiation) excited from the materials being measured. The sample is irradiated with a source laser, typically 532 nm, which then produces inelastically scattered Stokes (as well as Rayleigh and anti-Stokes) radiation at specific wavenumbers (radiation energies, wavelengths) when phonons in the sample are either created or destroyed. The inelastic to elastic energy difference then corresponds to the bond energies within the samples atomic lattice. A cut-out filter removes the anti-Stokes and most of the Raleigh scattered radiation from reaching the charged coupled device (CCD) detector (at -70 °C) leaving (primarily) the Stokes peaks for analysis.

Figure 2.9. Schematic of Rayleigh, Stokes and anti-Stokes scattering showing the characteristic energy gap remaining after Stokes relaxation.

The energy (frequency) of the back scattered Raman signal is generally plotted as reciprocal centimeters (i.e. number of complete wave periods within 1 cm, cm-1) with the number of counts providing the intensity of the signal at the respective frequency.

It is important to understand the phonon dispersion within graphene in order to interpret the Raman spectra. There are 6 phonon dispersion bands in a single unit cell of graphene consisting of two carbon atoms. Of the 6 phonon bands three of the bands are acoustic (A) and three bands are optical (O) of which 2 modes are in plane (i) and 1 is an out of plane (o) mode. The direction of the vibrations is considered in respect to the nearest carbon-carbon atoms and stated as being longitudinal (L) or transverse (T) (Figure 2.10). The plot in Figure 2.10 b shows the frequency of each of the phonon modes at each respective location within the BZ.

Figure 2.10. (a.) First Brillouin zone of graphene with highlighted phonon paths represented in the dispersion image. (b.) The dispersion of acoustic and optical phonon modes in graphene. X-axis phonon wave vector which is represented by the Brillouin zone (BZ) position. Reproduced with permission from American Physical Society.69

A transverse wave consists of displacement occurring perpendicular to the direction of propagation of the wave and is often referred to as a shear wave, such waves are commonly produced via sonication. Longitudinal waves consist of displacement parallel to the propagation of the wave. Plate waves, such as the Lamb wave propagate parallel to the test surface throughout the thickness of the material and can only occur in materials a few

109 wavelengths thick. Lamb waves can be both symmetrical and asymmetrical, the latter is referred to as a flexural mode as the body of the plate bends as the two surfaces move in the same direction.

To address each relevant phonon mode, the graphene Raman spectrum key peaks G, D and 2D will be addressed and explained in detail, touching on some of the other peaks commonly present.

Figure 2.11. Raman spectrum of bi-layer graphene with most common modes present.

B2.2.1.1 G peak

The G peak appears at ≈ 1580 cm-1 and is a primary in-plane vibration mode,222-223 a vibrational mode caused directly via excitation from the external source (laser or heat etc.).

The G peak corresponds to the high-frequency in-plane E2g phonon at the Γ position (longitudinal iLO). This in-plane mode is representing the carbon to carbon in-plane stretching and will be present for single and few layered graphene as well as bulk graphite (Figure 2.12 a).

The G peak will show an approximately linear increase in intensity with sheet thickness however the relative intensity change of the G peak from graphene to graphite is not dramatic (Figure 2.12 b). The G peak also shows minor variations to its position, with an approximate 3-5 cm-1 increase of frequency for monolayer graphene compared to bulk graphite (Figure 2.12 c). The G peak will also show responses to several other parameters such as temperature. The stability and predictability of the G peak allows comparisons to be made against it to determine relative intensities and gain information regarding sheet thickness.

Figure 2.12 (a.) Image of G peak carbon in-plane stretching vibrational mode. (b.) G band intensity versus sheet numbers. (c.) G peak position as a function of sheet number. (d.) The graphical representations of examples of phonon scattering processes. Reproduced and adapted with permission from (a. and d.) IOP publishing (b.) the American chemical society and (c.) Elsevier. 224-226

B2.2.1.2 D peak

The D peak, which appears at 1350 cm-1, represents the breathing modes of six atom rings and requires a defect for activation. The D peak comes from the (i)LO phonons around the BZ corner K and is activated by double resonance otherwise referred to as in-plane intervalley K to K’ scattering (Figure 2.13).226 The D peak is referred to as the disorder peak and is representative of both the intervalley phonon scattering and defect scattering, present for sp3 hybridization. The first-order D peak is not visible in pristine graphene (with sizes larger than the laser diameter) as a charge carrier must be excited and inelastically scattered by a phonon, followed by a second elastic scattering by a defect or break in crystallinity, leading to recombination.

111

Figure 2.13 (a.) Schematic of the breathing mode of the six atom rings in graphene. (b.) Schematic of the intervalley double resonance scattering. Adapted with permission from IOP publishing.225

B2.2.1.3 2D peak

The third, and the most informative peak is the 2D peak, which appears at 2690 cm-1 and is the D peak second order over tone.223 As the 2D peak originates from a process where momentum conservation is satisfied by two phonons with opposite wave vector, no defects are required for their activation and are therefore always present. As the layer number increases, a significant decrease of the relative intensity of the lower frequency 2D1 peaks occurs (Figure 2.14 b).222 This results in a significant change of shape of the 2D peak with the appearance of a large shoulder for graphite.

Figure 2.14 (a.) Schematic of the 2D second order overtone intervalley phonon scattering. (b.) The 2D peak components for (top series) monolayer graphene and (bottom series) graphite. Adapted (a.) and reproduced (b.) with permission from IOP publishing.225 The 2D peak will become symmetrical and show an intensity approximately 4 times greater than the G peak for a monolayer.222 This dramatic increase in intensity for the 2D peak is due to this peak being produced by a resonant Raman scattering process in which an electron is promoted from the valence to the conduction band.223

The 2D mode is referred to as double-resonant Raman scattering as the electron exists in a real energy level both upon its initial excitation and after the creation of phonons.223 The double-resonant nature of the 2D peak results in a highly uncharacteristic energy-dispersive property of this peak. With a different excitation energy the electron is excited to a slightly different point in reciprocal space, therefore the wavevector and energy of the photon required to scatter the electron in the conduction band is slightly different, resulting in a different phonon being excited.223

There are two primary methods to determine the sheet thickness of graphene via Raman spectroscopy, the first of which is the intensity ratio between the 2D and G peaks and the second is the shape of the 2D peak. If the ratio of the intensities of the 2D to the G peak are approximately 2 to 1 and the 2D peak is symmetric the sample is monolayer. Another simple identifier is the presence of a higher frequency shoulder on the 2D peak, which is a clear indication of graphite (or many layers). Beyond approximately 5 layers the spectrum of the graphene stack becomes practically indistinguishable from graphite.222 A bi-layer graphene stack will show a broader up-shifted 2D band in comparison to that of graphene.222

113

Figure 2.15. (a.) the change in shape of the 2D peak as a function of sheet numbers. (b.) Intensity comparison of the 2D and G peaks to determine sheet thickness. Adapted with permission from Elsevier.226

The shape of the 2D peak changes as a function of sheet number as shown by Figure 2.15 a. To understand the change in shape and intensity of the 2D peak we can start by expanding the explanation of the peak in graphite. In graphite this peak is present with two components 222 the 2D1 and 2D2 (each with a and b constituents as well) at ¼ and ½ height of the G peak. With a single additional layer from mono- to bi-, the individual electronic bands are split. Two of the bands remain touching indicating that bi-layer graphene still consists of a zero- gap. The addition of neighboring sheets introduces the three-dimensional vibrational mode, which consists of a phonon dispersion along the c direction perpendicular to the plane.

This alteration to the bandgap results in the single 2D peak of monolayer graphene being split into four wave vectors corresponding to phonons of different frequencies.222 With the increased number of modes, the total maximum peak intensity is dropped significantly. Simply put, as the number of graphene layers increases an increasing number of modes combine to produce a wider shorter higher frequency peak.227

The full width half maximum (FWHM) of the 2D peak also provides insight as to the graphene thickness. A monolayer graphene sheet will have a 2D peak FWHM of approximately 20 cm-1.72, 222, 228 While a bi-layer will have a FWHM of ≈ 50 cm-1. B2.2.1.4 Defect analysis

There are several other peaks present on a graphene Raman spectrum which should be addressed. The D+D” peak which is at ≈ 2450 cm-1 and the 2D’ peak at ≈ 3250 cm-1. The D’+D” originates from the same phonon as causes the 2D peak and both peaks originate from second-order Raman scattering.

Figure 2.16. (a.) Raman spectra of mono-layer defected graphene showing the D peak with an intensity approximately 4 times greater than that of the G peak. (b.) Evolution of the D and G peaks as a function of defect quantity (induced by ion bombardment). Reproduced with permission from (a.) the Nature Publishing Group and (b.) Elsevier.72, 229

Detailed information can be determined through the comparison of the intensities of several disorder peaks. The D peak (at ≈ 1350 cm-1), which has already been discussed, as well as the D’ which appears as a shoulder of the G peak (at ≈ 1620 cm-1), and the D+D’ positioned at ≈ 2940 cm-1 are all disorder induced peaks (Figure 2.16 a). As the disorder in graphene increases the intensity of each of these three peaks increase.227, 230-231 Additionally, the FWHM for all the key Raman active modes of graphene increase with any disorder.226

From this information an approximation of the extent of disorder can be made primarily from 232 a comparison between the intensity of the D and the G peaks ID/IG. With a small increase to the extent of defects the ID/IG will increase as a higher defect intensity results in a greater 115 amount of scattering (Figure 2.16.b).227 As the extent of disorder is sufficient to transition the material into a more amorphous carbon structure, the ID/IG will begin to decrease, which can then be observed by far greater attenuation for all Raman peaks.227, 229

It is important to note that IG is proportional to the sample area, whereas ID is proportional to the length of the edge analyzed. In scenarios where the sheet size is very small, the edges will dominate and give a misleading estimate of defects. As such, little weighting was applied to the ID/IG throughout out this study, although the D peak was never observed to have a greater intensity than the G, and the largest ID/IG observed is presented in Figure 2.20. From this, we can confidently say that defects are not being introduced into the graphene sheets during the preparation method, particularly with the support of complementary techniques such as XPS.

B2.2.1.5 Laser spot size

As the laser wavelength is 532 nm and the magnification is 100 X for all collected spectra the spot size of the laser can be determined via the following equation:

퐷532 = 1.22휆/푁퐴 Equation 2.21

(NA = Numerical Aperture = 0.85 for the 100 X magnification lens; characterises the range of angles over which the system can accept or emit light)

Or for a 100 X lens, Laser spot diameter (100 X; 532.1 nm) = (1.22 x 532.1) / 0.85

= 763.72 nm

As the Raman spectra were collected using the 100 X lens and the 532 nm laser a spot size of ≈ 750 nm was present. The particle size of the sheets was determined to be ≈ 350 nm, with a high level of polydispersity. Therefore, the D peak is expected to be present for (almost) all Raman spectra collected for sheets prepared throughout this study as the sp3 hybridized carbon atoms of the edges will (always) be measured.

Figure 2.17. Raman spectrum graphite small and large sheet size with and without the D peak.

A clear example of the sheet size effect can be demonstrated with a large unexfoliated graphite flake (Figure 2.17). The larger size allows the edges to be deliberately avoided (or included) by the incident light showing the absence (presence) of the D peak.

B2.2.1.6 Improved sample preparation

Additionally, it is very difficult to prepare samples in such a way as to isolate individual sheets, as can be seen via the AFM scan images (Figure 2.24) where overlapping, sheet folding and aggregation are commonly observed. As such the Raman spectra collected can be understood to be always convoluted by neighboring/overlapping sheets, which is not representative of the conditions present of sheets in suspension (or in any of the systems explored within this project). This common convolution is shown in Figure 2.18 a., where the shape of the 2D peak can be observed with the higher intensity peak on the left and the right of the center of the peak. The transition from single to 5 layers only results in the emergence of a peak on the higher frequency side of the peak as shown in Figure 2.15.

117

Figure 2.18. (a.) 2D peak change in shape with sheet thickness with (b.) the sheet thickness distribution based on 2D peak shape.

This is therefore strongly influenced by the sample preparation procedure which was a simple dropwise addition to either an alumina filter or to a glass slide (see Section B2.2.1.8). By altering the sample preparation method to a combined surfactant exchange, Langmuir- Blodgett dip-coating method (method used for AFM sample preparation, developed by Wang et al.,233 (described in Section B2.2.3.2), the sheets could be better separated allowing more representative scans to be obtained.

Figure 2.19. Raman scans comparing sample preparation methods of (a.) simple dropwise addition to (b.) the Langmuir Blodgett (LB) surfactant exchange method.

An important factor to explore as sonication is well known to produce cavitation events which can comprise of remarkably high temperatures and pressures which could be reasonably expected to induce significant defects to sheets. The lack of defects suggests that if cavitation events are occurring in sufficiently close proximity to the graphene sheets, the result is not a minor defect site creation, but a cross-section sheet breakage (as opposed to sheet separation/exfoliation).

It can be difficult to confidently infer a large amount of definitive information from the

Raman spectra for exfoliated graphene. However, the improvement of the IG/I2D ratio upon the changing of sample preparation method, coupled with the anomalous shape of the 2D peak (Figure 2.18) strongly suggests that convolution of spectra is playing a significant role.

B2.2.1.7 Typical Raman sample preparation

Samples are typically prepared by adding a small volume to a glass slide or to a 0.22 μm pore size alumina filter (Whatman), dropwise to 500 μL and allowing to dry. Samples were also prepared via the AFM sample preparation method (Section B2.2.3) which involved a combined surfactant exchange and Langmuir-Blodgett dip-coating method developed by

119

Wang et al..233 The sample analysis was performed using the Renishaw inVia Reflex Raman Spectrometer using a 532 nm 50 mW laser with a 100 X lens (unless stated otherwise) with a 2400 L/mm grating. The 532 nm laser employed is a diode pumped solid-state laser. Samples were typically collected with an acquisition time of 2 seconds and acquisition number of 10 with the confocality set to normal and the laser power at 100% (corresponds to the laser power of 1.2 mW). Only baseline removal and curve fitting post processes were applied to raw data collected. All samples were collected at room temperature which was 25 ± 2 °C. Care was taken to avoid laser induced heating, primarily by working at a laser power range of approximately 1.2 mW.

B2.2.1.8 Representative experimental Raman data

The exfoliation down to single and few layers of graphene is confirmed from the Raman spectra shown in Figure 2.20. The positioning, shape and relative intensities of the three major peaks at 1350 cm-1, 1580 cm-1 and 2690 cm-1 corresponding to the D, G and 2D peaks can provide good information as to the thickness of the sheet(s) analyzed. Comparison against the spectrum of a bulk graphite flake provides an easy identification of the exfoliation effect. The 2D peak is shifted to lower wavenumber than bulk graphite and shows a single symmetric peak indicative of single- or ≈ bi-layer graphene.234-235

Figure 2.20 a. Raman spectra of (top series) graphite and (bottom series) exfoliated F108- graphene. (b.) F108-graphene Raman spectra with peak intensities and (c.) the 2D peak intensity and position. (d.) Graphite Raman spectra with peak intensities and (e.) the 2D peak intensity and position.

The Raman spectra collected provides thickness information through the ratio of the intensity of the G and 2D peaks, I2D/IG, which is 0.87:1 for the presented spectrum, indicating highly exfoliated graphene sheets.203, 228, 236 The shape and positioning of the 2D peak indicates a highly exfoliated sample approaching that of monolayer graphene. The D peak is expected 121 to be present for the Raman spectra collected for sheets of this size as the laser diameter is larger (0.76 µm) than the sheets and will therefore detect the sp3 hybridized carbon atoms of the sheet edges.107 The FWHM for the 2D peak is 70 cm-1 which indicates this is likely a bi- layer spectrum.72

B2.2.2 UV-visible absorbance

UV-visible absorbance measurements were employed throughout this study primarily to determine the suspension concentration, as well as to give some qualitative insight into the graphene electronic structure. UV-vis measures the extent of transmission of monochromatic light through the sample, indicating the wavelengths at which light is absorbed. In the visible region, the absorbance corresponds to the electronic excitations from the highest occupied molecular orbital to the lowest unoccupied molecular orbital.

Absorption of electromagnetic radiation (EMR) within a material occurs when the energy of the incident EMR is equal to that of the bond energy of the material. When the incident EMR is not of the same energy as that of the bonds, the light will transmit through the material (conduction of radiant energy through a medium), this is characterised by the transmittance of a material. Transmittance is defined as the ratio of radiant power transmitted (I) by a material to the incident radiant power (Io).

푇 = 퐼/퐼표 Equation 2.22

Which is usually expressed as a percentage. The transmittance can also be converted to absorbance to represent the amount of EMR absorbed as opposed to transmitted.

퐴 = − log10 푇 Equation 2.23 B2.2.2.1 Absorption coefficient and Beer’s law

Concentration determination was calculated using the absorption coefficient of 4237 mL mg- 1 m-1 at 750 nm determined by Paton et al.,183 was used throughout the study allowing concentrations to be readily determined using Beer’s law:

퐴 = 휀퐿퐶 Equation 2.24

Where A is absorbance with arbitrary units, ε represents the absorption coefficient, L is the path length and C the concentration.

A large range of values have been reported for the absorption coefficient of graphene despite the apparent simplicity of the procedure required to determine it. Values from as small as 1043 mL mg-1 m-1 to as large as 6600 mL mg-1 m-1 have been reported.107, 220 Paton et al. derived an equation to more robustly predict this value for graphene (and a number of other graphene analogous materials) by relating the absorption coefficient for dispersions of nanosheets to that of the optical absorption of a monolayer at normal incidence.183

3 log10 푒 183 훼(휆) = 퐴푀퐿(휆) Equation 2.25 8휌푁푆푑0

Where ρNS is the nanosheets density, d0 is the monolayer thickness and AML is the intrinsic nanosheet absorption. As discussed in Section 1.3, the optical opacity of a graphene monolayer is 2.3% and is independent of wavelength from approximately 400 to 900 nm.8 This value sits in the middle of the previously reported values, and accounts for the potential scattering events which may convolute the absorption coefficient value determined. As all suspensions employed were prepared via very similar production methods and the concentration ranges used were not dramatically different (and controlled were 123 concentrations differences were large) the single extinction factor of 4237 mL mg-1 m-1 was applied to all samples.

It is important to note that a range of other contributing factors including lateral size distribution, mean number of layers per flake and the presence of functional groups can influence the determined absorption coefficient.237 This highlights the importance of using similar preparation conditions when applying a single absorption coefficient value across multiple studies.

B2.2.2.2 Characteristic graphene spectrum

The characteristic UV-vis spectrum for graphene shows a peak at approximately 270 nm. As the wavelength is increased beyond 270 nm the absorbance decreases slightly and begins to flatten, extending almost flatly out into the IR region. This broad range of strong absorbance demonstrated by the typical UV-vis scan is attributed to the conjugation within the graphene sheet where the pi electrons are capable of absorbing a wide energy range of light. The optical properties of graphene will be discussed in greater detail in Chapter 3.

Figure 2.21. The absorbance scan above shows the characteristic absorption peak for graphene at 270 nm as well as the strong broad band absorption allowing wavelengths from the UV region through the infrared to be absorbed.

The absorbance @ 750 nm is used in conjunction with the extinction coefficient of 4237 mL mg-1 m-1 to determine the concentration of the graphene solution employed. It is important to note that the graphite/graphene absorbance scans will show little qualitative signal change. This method is primarily for the determination of concentration. However, as discussed further in Chapter 3, the extent of oxidation and some insight into the electronic structure can be determined from the absorbance spectra.

The small dips and rises in the spectra out in the near-infrared (NIR) are attributed to the changing of detectors at these wavelengths. This is different to the artifacts observed from the UV-1800 Shimadzu UV which arise as a result of lamp and detector changes.

B2.2.2.3 Monitoring concentration during production

The extent of exfoliation was monitored using UV-vis. The concentration of graphene in suspension at sequential stages of sonication was determined using a UV-1800 Shimadzu

125

UV Spectrophotometer via the absorbance at 750 nm using the extinction coefficient 4237 mL mg-1 m-1 determined by Paton et al. (Figure 2.22).183

Figure 2.22. (a.) Absorbance scans as a function of exfoliation time. (b.) concentration as a function of exfoliation time.

Measuring the absorbance of small extractions of suspension throughout the sonication period provides an excellent insight into the exfoliation process. The data presented in Figure 2.22 is for a 0.1 % initial F108 surfactant concentration of graphene. This linear region indicates that the suspension conditions have been set appropriately and the concentration of graphene is increasing linearly with sonication time.

It is important to note that samples are centrifuged after sonication to remove large, unexfoliated material from the suspension. The data presented above in Figure 2.22 is for sample post centrifugation and therefore the increase of concentration observed is for that of suspended particulates which are sufficiently light to avoid sedimentation during the centrifugation. This means that smaller particles, which are not necessarily exfoliated completely (to single and few layer) may be present if they are of sufficiently small size. This is important and flake breakage can occur readily during ultrasonication, as such the linear concentration presented above may not solely represent exfoliation, but simply the production of sheets of this size range. B2.2.2.4 Identifying oxidation

UV-visible absorbance can also indicate whether extensive oxidation has occurred, as the peak at ≈ 270 nm would shift to substantially lower wavelengths ≈ 230 nm.66 This is an important consideration when producing the material used for experiments and ensuring the samples have been prepared correctly without the loss of the target properties.

The absorbance spectra for a range of graphene materials was measured to compare the spectra peak positions as well as the relative decrease in absorbance out into the NIR. The absorbance spectra of GO and both partially and completely rGO samples, purchased from Graphenea show far weaker absorbance out in the NIR (Figure 2.23).

Figure 2.23. Absorbance spectra for GO, partially reduced rGO, completely reduced rGO and pristine F108 exfoliated graphene at a fixed concentration of 1 mg mL-1.

The characteristic absorbance scan for exfoliated pristine graphene shows the peak maximum at 270 nm indicative of non-oxidized graphene sheets.238 The wavelength scan also shows a strong and broad absorbance in the infrared region that is stronger than GO and rGO (Figure 2.23). The trend of the strengthening NIR absorbance with increasing conjugation is

127 particularly clear when an incompletely reduced GO sample is compared, showing a strong increase in absorbance between that of GO and fully reduced GO (Table 2.2).66

One of the key properties of surfactant exfoliated graphene microplates can be clearly demonstrated by directly comparing the absorbance profiles for pristine graphene with GO, rGO (with varying extents of reduction). As the absorbance spectra extend out towards the infrared (IR) region those graphene materials with higher conjugation networks can clearly be seen to absorb stronger. Therefore, a lower concentration is required for the target application.

Table 2.2. Commercially purchased graphene oxide (GO) and reduced graphene oxide (rGO) data, collated from Graphenea product datasheet. Graphene Absorption coefficient Absorbance at 808 C:O ≈ Particle type (mL mg-1 m-1) nm (at 1 mg mL-1) ratio Size (nm)

183 F108- 4237 @750 nm 47.73 ≈ 60:1 370 graphene 66 Reduced 7380 @ 265 nm 36.68 4.68 : 1 260 - 295 graphene

66 66 Partiallyoxide 7380 @ 265 nm 28.00 20 reduced

66 Graphenegraphene 6150 @ 230 nm 6.08 1.15 : 1 4300 - 16600 oxide

B2.2.2.5 Typical UV-visible sample preparations

The initial concentration of all suspensions prepared throughout this study were determined using a UV-1800 Shimadzu UV Spectrophotometer. This is a double beam photometric system with both deuterium and halogen lamps and uses a silicon photodiode detector. The only sample preparation required was the dilution of suspensions into a concentration range that was confirmed to be linear with absorbance. Dilution was performed with MilliQ water as it is critical to ensure the ionic strength of the suspension is not altered significantly, particularly at low surfactant concentrations. All measurements were performed in quartz cuvettes with a path length of 10 mm and a volume of 3.5 mL unless otherwise stated.

The absorbance spectra for the graphene materials was also measured using a UV-310PC Shimadzu UV-vis-NIR spectrometer. Shimadzu UV-vis-NIR spectrometers have three detectors, a photomultiplier tube (PMT) for the UV and visible regions, and InGaAs and cooled PbS detectors for the NIR.

B2.2.3 Atomic force microscopy

Scanning probe microscopy (SPM) is a class of analytical techniques for the measurement of solid materials which uses a small probe and recognizes the movement (deflection, attraction) at the end of a probe. AFM is one strategy of SPM which is derived from scanning tunneling microscopy (STM), which involves the rastering of a sharp tip, attached to a cantilever, over the surface of interest at a fixed distance (from the surface) and mapping the deviation of the tip cantilever position via a reflected laser. The forces inducing the tip movement arise from the repulsion or attraction of the tip to the surface. AFM is not limited to only conductive materials (and can therefore be employed in liquid) and a combination of forces dictate the tip repulsion and attraction. AFM is a highly suited analysis method for graphene sheets, allowing not only the determination of the lateral dimensions of samples, but due to the remarkable sub nanometer z resolution, the thickness of the sheets can be accurately measured.

129

B.2.2.3.1 Atomic force microscopy of graphene

The theoretical thickness of a monolayer graphene sheet is 0.335 nm (determined from the interlayer spacing of graphite), which is then additive with layers and can be directly measured via AFM (Figure 2.24).

Figure 2.24. (a. and c.) AFM image of graphene sheets with (b. and d.) the corresponding cross section to show thickness.

The AFM image shows multiple sheets in the scan area with the corresponding cross section positions displayed. The sheet morphology can be observed showing the irregular shapes, and the polydispersity of the sheet sizes is also illustrated. Each sheet can be observed to be approximately 2 nm thick. While the information able via this method can provide in-depth sheet thickness information, there are a range of limitations surrounding the sample preparation, the statistical robustness of the results, and the presence of surfactant on the surface of the material. These factors can combine to limit the effectiveness for SALE graphene analysis.

In order to accurately image a single layer of graphene that sheet must be completely flat on the substrate. This is a difficult task to achieve as the graphene sheets will tend to deposit on top of other sheets and not only appear thicker, but also have angles in respect to the surface, convoluting the images. This of course, is in addition to the presence of a surfactant layer of unknown thickness on both faces of the sheet.

Initial attempts of sample preparation involved adding small quantities (a drop) of graphene suspension to silicon wafers and exploring different heating methods to either attempt to calcine the surfactant in controlled atmospheres, or simply evaporate the solvent. These methods result in a combination of aggregation, oxidized samples, and excessive stacking and called for an improved approach.

A method was developed to employ a LB dip coating approach to allow the graphene sheets to be deposited flat on the substrate surface. The process required a solvent and surfactant exchange, which was achieved through the addition of a secondary benzoxazine surfactant which allowed the sample to be suspended and in chloroform. The chloroform suspension was then transferred dropwise to the LB trough dropwise and compressed. The silicon wafer could then be dipped through the surface layer, allowing the sheets to adsorb flat onto the substrate. This sample preparation method showed significant improvements over the previous methods, however, there are still a large number of sheets present which may partially aggregate and sometimes aren’t completely flat. This limits large scale analysis, leaving the statistical significance lacking robustness.

131

It is important to consider the surfactant layer on the surface of the graphene sheets, as well as the presence of water, which will result in larger thicknesses being measured.203, 233 It was shown by Wang et al., that a 2 nm step was present for individual sheet numbers, which was then altered to a ≈ 1 nm interval following a heat treatment stage.

In addition to the valuable thickness information the AFM scans also give lateral size information which is highly accurate, but again low in statistical significance. When combined with DLS and TEM, a greater representation of the sheet lateral sizes can be built.

B.2.2.3.2 Typical AFM sample preparation

Prepared graphene suspensions were extensively centrifuged at 3000 rcf for 30 minutes to remove any larger unexfoliated or aggregated particles. The supernatant was collected for use.

The surfactant exchange was performed by first performing high speed centrifuge of 25,000 rcf using a Thermal Scientific sorvall rc 6+ centrifuge for 2 hours to remove all liquid and collect the sample as a pellet. The precipitate was then re-dispersed in 5 mg mL-1 of a benzoxazine surfactant (BM1000) with stirring and mild bath sonication. This centrifugation and re-dispersing process was repeated three times. The samples were freeze dried following the centrifugation process and redispersed in chloroform with the aid of mild bath sonication and stirring.

The transfer of the silicon wafer substrate was performed using a NIMA LB trough following the surfactant exchange. The samples were diluted with fresh chloroform and then the chloroform suspension was dropped on the water surface using a glass syringe and a syringe pump (Harvard PHD2000) with an infuse speed of 6 mL min-1. The film of chloroform could be faintly seen on the surface after addition. The surface was compressed using the LB trough barriers and a small piece of silicon wafer was dipped slowly into the trough and then slowly withdrawn. The substrate was allowed to dry, then rinsed with water and ethanol and finally blown dry with pressurized nitrogen. When heat treated, the substrate was then calcined at 350 °C in an air muffle furnace for 24 hours in an attempt to remove the surfactant.

A Multimode 8 AFM (Bruker) was used to image graphene samples using ScanAsyst-Air mode. Bruker ScanAsyst-Air probes with nominal tip radii of 2–12 nm and a silicon nitride cantilever of spring constant, 0.4 N m-1 were employed. Analysis on flattened images was conducted using NanoScope Analysis software (V 1.5, Bruker).233 Samples were prepared on silicon wafer as described by Wang et al. in 2017.233

To review, graphene suspensions were characterised via AFM, whereby the sheet thickness and morphology can be determined and visualized. The thickness section shows a monolayer graphene sheet that likely has a water or surfactant coating as well as indicating that these are single and few layer suspensions.203, 233 The AFM image shows the polydispersity that is characteristic of these materials and gives an approximate comparison to the particle sizing analysis performed through DLS (see Section 2.2.6).

B2.2.4 Transmission electron microscopy

TEM is an electron microscopy technique which relies on the transmission of electrons through the sample to be imaged to build the final image. Electron microscopies use an electron beam to “illuminate” the sample, producing images with remarkable magnification. The electrons are directed and focused using electromagnetic lenses as opposed to glass lenses used in traditional light microscopy. By using electrons as the source, the maximum resolution is far greater than that of the theoretical limit of visible light.

The resolution (dmin) of any microscope is limited by the wavelength (λ) of the illumination source used:

133

휆 푑 ≅ Equation 2.26 푚푖푛 2

For light microscopy, the visible spectrum ranges from ≈ 400 – 700 nm resulting in a best resolution of 200 nm. The wavelength of electrons is adjustable by its energy, for example a 100 keV electron has a wavelength of ≈ 4 pm.

ℎ푐 퐸 = Equation 2.27 휆

Allowing the (fundamental/theoretically achievable) resolution to be as approximately 2 pm. Practical limitations see the maximum resolution of typically 0.1 – 0.2 nm.

TEM provides excellent images for lateral size information and some sheet thickness information can be obtained. In order to achieve high quality TEM images the sample must have a thickness comparable to the mean free path of the electron that passes through the sample, therefore the maximum sample thickness must be below several hundred nanometers. Usually to achieve any decent imaging the sample thickness must be a few tens of nanometers, depending on the electron gun power. As TEM relies in the transmission of electrons through the sample to build the image, the extent of electron attenuation provides good information as to the thickness of the sample being imaged. Further, the transmission of electrons means that the (spatial) intensity of electrons measured by the detector depends on the sample structure, again providing further insight into the sample. The exfoliated sheets position TEM as another highly suited characterization method.

Figure 2.25. Exfoliated F108-graphene laying across a TEM grid hole.

One limitation here is that it can be difficult to accurately and confidently identify the number of sheets. Simply counting the number of sheets at the edges is often possible as with Figure 2.26.

Figure 2.26. TEM image of a thicker, high sheet number graphene stack with visible sheet edges.

135

Additionally, high resolution TEM (HRTEM) can allow visualization of atomic structure of the sheets by showing the lines of atomic symmetry. This can then be supplemented with the electron diffraction pattern which can allow identification of aligned graphene layers, and provide information as to the thickness of the sheets.

Electron diffraction has been demonstrated as a graphene thickness analysis technique by several groups. Whereby the intensity of the inner set of diffraction points is more intense for monolayers than for multilayers.141, 239-240 Electron diffraction was not employed as a thickness analysis during this study as the AFM and Raman were considered sufficient for our purposes.

Additionally, identifying folded graphene sheets indicates that the layer number is sufficiently thin so as to allow low sheet rigidity for folding. Despite the high in-plane Young’s modulus of graphene, it is easily warped in the out-of-plane direction which induces strain energy altering the electronic structure along the fold.241 The folding present in Figure 2.27 indicates that the sheet number is sufficiently low as to allow such folding events.

Figure 2.27. Exfoliated F108-graphene with identifiable folding indicating low sheet number.

B2.2.4.1 Typical TEM sample preparation

All TEM suspension samples were all prepared by dropping 100 μL of 0.1 mg mL-1 of each respective sheet suspension onto Holey Carbon TEM grids supported on filter paper and vacuum suctioning briefly. C-flattm holey carbon-coated TEM support grid is an ultra-flat holey carbon-coated TEM support grid. The C-flattm is manufactured without plastics and therefore does not require any pretreatment prior to image analysis. The holey carbon grids provide a sample support platform which consists of 2 m pores that act to retain the sample whilst allowing the electrons to transmit readily through to the detector.

A JEOL 2100F TEM was used to image the exfoliated graphene sheets. Samples were imaged at 75 kV. The vacuum suctioning aided in drying the TEM grid and also by drawing the sheets predominantly over the holes within the grids, which is not critical, but does allow more images of freely suspended sheets to be collected.

137

Figure 2.28. The TEM image collected show several exfoliated graphene sheets, highlighting both the extent of exfoliation as well as the quantity of sheets that can be easily produced via the liquid exfoliation method.

B2.2.5 Scanning electron microscopy

As with TEM, SEM illuminates the sample with electrons rather than light. SEM measures the backscattered electrons which have bounced back from the sample, as well as secondary electrons which have been knocked out from the specimen by the incident electron. The intensity therefore depends on the atomic number and the sample morphology. SEM images are likened to that of a birds eye view, where only the surface of the sample is observed, and therefore do not provide sheet thickness information.

By observing sheet samples under SEM, the manner in which sheets deposit can be identified which is particularly helpful when attempting to understand the height profiles obtained via AFM. It can be seen in Figure 2.24 that sheets often sit vertically, on angles and are often folded-over creating misleading high points on the AFM profile.

SEM was also employed to measure the thickness of thin films prepared within the study (primarily for XPS sample preparation). The sample films were simply cut and placed vertically against a small support on the electrically conductive adhesive tape. The samples were coated with either gold or platinum to increase conductivity and addition conductive tape was placed over the top of the film connecting to the base on either side of the sample. The images could then be taken and the thickness of the films prepared could be easily determined (with care to ensure the working distance and scale were correct).

B2.2.5.1 Typical SEM sample preparation

The typical sample preparation procedure for the SALE nanosheets involved depositing 100 µL of 0.1 mg mL-1 onto 0.2 µm alumina filters (Whatman), allowing to dry and then cutting the filter into small fragments. The small filter fragments were then mounted via conductive adhesive tape onto an SEM stage and imaged at 1 kV.

Figure 2.29. SEM image at 50,000 X magnification for F108-graphene.

If required samples were coated with platinum at 10 mA for 2 minutes to avoid charge build up and to allow greater resolution. SEM images of the exfoliated graphene sheets were collected to contribute to general morphology and exfoliation elucidation using a Zeiss UltraPlus FESEM without any coating at a voltage of 1 kV. Samples were added dropwise to an Alumina filter.

139

Film thicknesses (prepared for XPS analysis were determined via cross section SEM. Samples were mounted vertically against supports with conductive tape connecting the films to the base plate minimizing the charge build up.

Figure 2.30. Cross sectional SEM images of CTAB-graphene filtered films for thickness determination. SEM image at (a.) 3,000 X and (b.) 12,500 X magnification and (c.) image of the mounted samples for thickness determination.

B2.2.6 Dynamic light scattering – particle sizing

DLS measurements were performed to determine the hydrodynamic particle size of a range of exfoliated graphene samples. Particle size is a highly important factor to consider for prepared suspensions providing key information regarding the nanosheets behavior and possible location within a drug delivery system as well as cell-particle interaction.

DLS relies on an incident light source to be directed at and scattered by the suspended particles. The detector, which is generally offset by 90 or close to 180°, will measure the scattered light signal and determine the size from the correlation function of the signal versus time (Figure 2.31).

Figure 2.31. (a.) Example DLS raw data showing the correlation function over time. (b.) The corresponding number particle size distribution (PSD).

The data presented in Figure 2.31 is for a 0.025 mg mL-1 F108-G sample. The number PSD diameter for this sample was determined to be 338.2 nm, with the average particle size determined to be ≈ 370.0 nm.

For smaller particles, the signal will have a shorter correlation coefficient as the speed/Brownian motion of the particle will be greater, therefore the scattered light signal will have a greater correlation for a shorter period of time. Larger particles will move slower and as such the scattered light from these particles will be more constant over a longer period. It is important to note that larger particles will dominate the scattering signal as intensity distributions tend to emphasize the contributions from species with the largest scattering contribution. This is usually the larger particles as the scattering intensity is to the 6th power of the size.242 This is particularly important when considering a highly polydisperse sample, such as SALE graphene.

All DLS data quoted throughout this study is the intensity and the number PSD. This was selected as the intensity value is the raw data representing the scattering intensity which is

141 what is directly measured by DLS. The number PSD accounts for the presence of larger particles and corrects the sizing data accordingly.

In order to determine the appropriate concentration range to perform DLS measurements, a series of graphene suspensions was prepared covering a range of concentrations. The concentration range in which the particles were measured at approximately the same size was identified and selected for future experiments.

Figure 2.32. (a.) Particle size as a function of concentration. (b.) Corresponding standard error of the PSD values.

Within the size range presented, the concentrations were approximately even (with perhaps a small sign of deviation away from the mean at 0.2 mg mL-1), and with increasing concentrations, the data became of poor quality. By looking at the standard error of the PSD values, a clear trend could be identified, showing concentration of the highest precision at 0.025 mg mL-1. Future PSD measurements were performed at this concentration.

When handling particle sizing data of highly polydisperse samples, the selected resolution of the data post-processing is important and can provide a valuable insight to the materials properties. Low resolution analysis can give a representative mean size value for the suspension, and was employed predominantly throughout this study. High resolution can provide information of the breakdown of separate size groups of particles, such as the removal of the largest particles in suspension over sonication time.

B2.2.6.1 Polydispersity of SALE graphene

Top down approaches to colloidal suspension preparation are well understood to produce polydisperse dispersions. The liquid exfoliation method is a top down approach, namely from graphite to graphene, and as such the suspensions produced result in characteristic highly polydisperse size ranges.

The typical size profile of a graphene suspension ranges from approximately 20 nm out to 3 μm diameter. These values will alter significantly depending on the size and identity of the starting material and the specific production method parameters such as centrifugation rates and times as well as the sonication power.

Figure 2.33. Example DLS raw data showing the polydispersity and the impact this has on the sizing data, with (red series) low resolution handling and (green series) high resolution data handling.

143

For the 10 samples of 0.025 mg mL-1 F108-graphene measured had a mean polydispersity index of 0.367 indicating a highly disperse suspension. This is in strong agreeance with data obtained via the imaging techniques, AFM, TEM and SEM.

Figure 2.34. Particle size of exfoliated nanosheets as a function of time with the three series showing tri-modal distributions (green - highest size, red – middle size distribution and blue – the smallest size distribution) and the respective trends of each.

Figure 2.34 shows the decrease in size only for the largest size mode. The two lower distribution modes stay approximately constant over the extended sonication period. As Figure 2.34 indicates, the resolution of the sizing distribution mode can significantly influence the determined size. Low resolution modes can be both a simpler way to view such trends, as well as more accurately reflect the general sizing trend of the polydisperse particles in suspension.

B2.2.6.2 Sonication effect of particle size

It is important to note that when high sonication powers are employed for exfoliation, the mean particle size will decrease over sonication time. This was demonstrated and accounted for by Khan et al. in 2010,210 by exfoliating graphene under low power conditions for extended times and monitoring the size trends.

Figure 2.35. (a.) Particle size as a function of sonication time at 2 kW probe system sonication power. (b.) Particle size as a function of time with 100 W probe system, with additional applied 2D morphology correction equation data.

For example, as shown in Figure 2.35 the particle size can be seen to decrease as a function of sonication time. It is important to note that this sample was prepared using ~ 2 kW sonication which would induce far greater and more rapid particle break down (the suspensions used throughout this study were prepared using a 100 W sonication probe system.

The amplitude of 80% was selected for most suspensions prepared via the Qsonica 100 W sonication system, as this intensity of the sonication was appropriate for producing highly exfoliated with no observable defects or size changes.

145

The Malvern Zetasizer ZS sizing equation assumes a spherical particle. It is clear from the TEM, AFM and SEM imaging that the particles have a very high aspect ratio and hence the confidence that may be applied to the light scattering approach must be considered carefully.

The mean particle hydrodynamic size was adjusted using a correction equation specific for 2D sheet morphology samples, developed by Lotya et al. in 2013.243 This group showed that the equivalent sphere radius, scales with the particle size of a thin planar exfoliated sheet using Equation 2.28 where a is the peak in the PSD in nanometers.

퐿 = (0.07 ± 0.03)푎1.5±0.15 Equation 2.28243

Whilst this correlation value does not provide absolute certainty to the DLS data, it does add a robustness, addressing the sphere-2D sheet assumption, increasing the confidence which may be applied.

B2.2.6.3 Typical particle size for surfactant exfoliated graphene

Interestingly the size differences were observed for suspensions prepared with different surfactants. The average hydrodynamic diameters of 370, 195 and 310 nm for materials exfoliated using F108, F68 and L64 respectively (Figure 2.36).

Figure 2.36. Hydrodynamic particle size (diameter) distribution (number PSD) for (a.) F108- G, (b.) F68-G and (c.) L64-G.

It is not completely clear (and beyond the scope of this study) whether the observed size difference is due to an alteration in the exfoliation mechanism, or possibly surfactant degradation events. A likely influence, not in isolation, is stability differences provided by the surfactants along with signal due specifically to the surfactant molecules.

The stability mechanisms of the surfactant exfoliated graphene suspensions were discussed in Section A2.4. To further elucidate these mechanisms and gain an understanding of the practical stability provided by the predominantly employed non-ionic surfactant F108, a range of experimental conditions were explored including ionic strength and pH (Figure 2.37) (for zeta potentials see Section B2.2.7, Table 2.4).

147

Figure 2.37. The particle size of the graphene suspension was measured as a function of ionic strength showing little change until 1 M of NaCl. This highlights the stability imparted to these suspensions by the large surfactant molecules and strongly indicates that the system is sterically stabilised.

As discussed in Section A2.4, the “concentration” of the polymeric surfactant chains increases significantly upon particle-particle approach. The effectively lowered water concentration then drives osmotic repulsion as the water molecules move in to the overlap region. By increasing the salt concentration, the relative decrease in water concentration during overlap is minimized and as such the repulsive mechanism is not as influential (Figure 2.37). Therefore, the polymeric steric stabilization can be overcome more easily, resulting in an increased amount of aggregation/ flocculation events, which is demonstrated by the mild increase in measured particle size. It is important to note the high salt concentration for the increased size, far above any working ionic strength conditions.

B2.2.6.4 Typical DLS particle size sample preparation

All DLS data reported within this study were collected using a Malvern Zetasizer Nano ZS which utilizes non-invasive backscattering signal. A 633 nm, 4 mW laser was employed, and the detector of this instrument is an avalanche photodiode detector, positioned at 13° and 173°. The intensity changes are analyzed with a digital autocorrelator which generates the correlation function. The maximum size range measurable on the system is from 0.3 nm to 10 microns, and the maximum concentration is stated at 40 % w/v, however concentrations measured were far lower.242

All samples were inverted 10 times and sonicated at low power (bath sonication) to reverse any minor scale aggregation/flocculation events. The Zetasizer was on for a minimum of 15 minutes prior to any measurements and all measurements were performed at 25 °C. All samples were measured for 10 runs with 10 cycles per run.

B2.2.7 Zeta potential

Zeta potential measurements were made for a range of exfoliated graphene samples to provide a greater insight into the stabilization mechanism of the suspensions and to detect significant defects if present. As touched on previously in Section A2.4, the DLVO and steric stabilization mechanisms apply strongly to the suspensions prepared within this study with both the smaller ionic as well as the larger non-ionic surfactants.

As discussed in Section A2.4, in aqueous solution, many colloidal particles have the ability to develop a charge at the surface which can have a significant influence on the system stability. The pH can therefore play a significant role in the charge presented under given conditions.

It is important to know the position at which the zeta potential is measured in relation to the particle to gain a clear understanding of the relevance of the zeta potential. The layer of electrolytes in immediate contact with the colloid itself will consist predominantly of the counterion of the colloids charge, this is referred to as the Stern plane. The diffuse layer consists of a high concentration of electrolytes which are influenced by both the surface potential and the stern layer. The point at which the electrical potential of the colloid is no longer influencing the electrolytes in solution is referred to as the slipping plane or the plane

149 of shear. It is at this point at which the zeta potential is measured and marks the end of the electrical double layer (EDL).

The EDL effectively compresses upon the increase of electrolyte concentration and as such the zeta potential can be considered as a function of ionic strength. Zeta potential is typically determined by particle microeletrophoresis whereby a dilute colloidal suspension is subjected to an electric field. This causes the colloids to migrate towards an electrode depending on the respective charges of particle and electrode. Therefore, the direction of migration is dictated by the type of charge while the speed of migration is primarily dictated by the magnitude of charge.

The equation to determine the electrophoretic mobility is as follows:

푣 휀푧 휇 = = 푓 ( ) ∗ 1000 푚2 ∙ 푉−1 ∙ 푠푒푐−1 Equation 2.29 퐸 푛

Whereby E is the electric field strength (V), ν is the velocity, ε is the permittivity of solution

(given by ε= εoD). Z represents zeta potential; n is the viscosity (cp) and the factor of 1000 is a conversion value. The f value is a scaling factor which is input as 1 when the EDL is small and 2/3 when the EDL is large in comparison to the particle size.

The Smoluchowski equation was applied throughout this study as it is appropriate for this particle size range and predominant observed charge. The Smoluchowski equation is valid for samples with a sufficiently thin electrical double layer in comparison to the particle radius.

The surface charge of a colloidal particle is an important consideration when dialyzing suspensions to remove surfactant and can lead to the complete collapse of a suspension quite easily. Zeta potential provides a method to monitor the charge of the particles and allow suspension collapses to be avoided and give a greater understanding of the stabilization mechanisms.

Zeta potential measurements can also aid by indicating whether significant changes in particle charge occurs over extended periods of sonication. A change in particle charge may indicate that chemical defects have occurred during the exfoliation process. One such defect which could occur is oxidation caused by bond cleavage associated with cavitation during sonication. The lack of any deviation away from the initial zeta potential indicates the absence of significant quantities of defects forming throughout the exfoliation process.

Figure 2.38. Zeta potential of an L64-graphene suspension as a function of sonication time.

Figure 2.38 shows little deviation of the electrophoretic mobility over the course of the extended sonication period. This suggests that if defects are being induced via this process, they are not sufficient to influence the surface charge of the particles, implying minimal defect presence.

151

Figure 2.39. Zeta potential as a function of pH for an F108-graphene suspension. The low magnitude and the absence of any deviation of the zeta potential over the pH of 3 through to 11 suggests that there were very few edge defects present in the prepared suspension.

From Figure 2.40 it can be immediately identified that upon the initial addition of NaCl, the zeta potential magnitude is decreased significantly. This shows the influence of the salt concentration on the stability mechanism, particularly any electrostatic repulsion stability, whilst also suggesting the importance of controlling salt concentrations of suspensions.

Figure 2.40. Zeta potential as a function of ionic strength for an F108-graphene suspension.

The zeta potential of the exfoliated graphene particles was determined through electrophoretic mobility measurements (Table 2.4). The graphene samples have a relatively low potential of –5 mV at pH 6 (Figure 2.39). This suggests that the graphene sheets have few or no defects in the basal plane and the charge arises from the edges. Importantly, the charge is low indicating that steric effects due to the adsorbed polymeric surfactants dominate the stabilization mechanism.

Table 2.4. Zeta potentials for a representative range of relevant surfactant stabilised suspensions.

Surfactant Electrophoretic Zeta potential - ID mobility (m2 V-1 s-1) Smoluchowski CTAB 1.887 +24.167(mV) F108 -2.385 -26.4 F127 -0.462 -6.286 L64 -1.308 -16.769 P123 -1.385 -18.147

Typical zeta potential sample preparation

The zeta potential measurements were performed using the Malvern Zetasizer ZS and were prepared identically to the particle size samples. Disposable clear zeta cuvettes with inbuilt electrode faces were employed throughout the study. Samples were measured at 25 °C with a pre-measurement agitation via a sonication bath and inversion of sample.

The measurements were performed in MilliQ water (pH 6.5 unless otherwise stated). Typically, 5 measurements of zeta potential using 4 separate samples were undertaken. The values reported throughout this study are an average of these 20 measurements.

153

B2.2.8 X-ray diffraction

XRD was employed for basic characterization of the graphene sheets, for a brief comparison against intercalated graphite and for characterization of the α-CD gels (see Chapter 7).

XRD is an X-ray scattering technique whereby the scattered intensity of the incident X-ray beam in measured as a function of the incident angle. For crystalline materials, there is periodic variation of electron density at atomic dimensions which leads to interference between the scattered X-rays (incoherent scattering) resulting in the formation of Bragg peaks (coherent scattering). For parallel planes of atoms, with a space of d between the planes, constructive interference only occurs when Bragg’s law is satisfied.

The XRD pattern of graphite is presented in Figure 2.41 and provides an appropriate starting point to describe the key information provided by XRD.

Figure 2.41. XRD pattern for graphite showing a clear peak at the 2θ angle of ≈ 27° and ≈ 54° which corresponds to a d spacing of ≈ 0.33 nm. This is the interlayer distance, calculated from the Bragg equation:

휆 = 2푑ℎ푘푙 ∙ 푆𝑖푛휃 Equation 2.30

0.15406 푛푚 푑 = 2 × 푆𝑖푛휃

푑 = 0.33 푛푚

The XRD pattern for monolayer graphene should ideally not show any c direction crystallinity (d spacing). However, as the production method produces a range of sheet thicknesses this can be clearly identified.

Figure 2.42. (a.) XRD of exfoliated graphene. (b.) Intercalated graphite versus non- intercalated.

The single and few layer exfoliated graphene sample presented almost identical XRD spectra to that of graphite. This is likely due to the presence of few layer graphene sheet stacks. The intercalated sample (purchased from Sigma-Aldrich, Expandable graphite flakes) shows the broader reflection peak as well as a small decrease in reflection angle indicating an expansion of the interlayer spacing (d ≈ 0.35 nm).

The a direction crystallinity can also be measured, and has been determined both theoretically and experimentally.244-245 The a spacing is of 2.46 Å (Figure 1.3). The analysis of the a

155 spacing allows the possibility of identifying the thermal expansion of graphene by observing a shift of the a spacing in response to changes in temperature. This is tightly related to this study as the thermal properties of graphene may aid in the photostability of the sheets through greater heat transfer.

Graphite is well understood to have a negative thermal expansion coefﬁcient (TEC) below 500 K.246-247 Graphene, a single layer of graphite, has been predicted to exhibit negative TEC as well, below 300 K, with a unique dependence of its lattice parameter with temperature due to the out-of-plane vibration modes that its membrane like topography allows. This has been demonstrated via synchrotron powder XRD and the TEC of graphene was confirmed to remain positive from 10 to 300 K.244

Typical XRD sample preparation

Graphite flakes were simply deposited on the titanium sample holder with double sided adhesive tape. Exfoliated graphene samples were filtered with 0.2 μm pore size alumina (Whatman) filters with sufficient concentration to either be free standing, or sufficiently thick as to ensure no background signal was detected. All XRD data was obtained using the Bruker DiffracPLUS X-ray diffractometer (XRD) employing Cu Kα radiation (λ = 0.15406 nm). The samples were scanned over a 2θ range of 5° to 90° at a scanning rate of 1 degree per minute.

B2.2.9 Suspension stability

The suspensions produced via the SALE method are highly stable and will remain in suspension for months at a time. It is important however to understand the stability on a finer scale in order to ensure that experiments reflect the intended conditions of the suspension.

Figure 2.43. Absorbance scans for an L64-graphene suspension (red series) one day after preparation and (blue series) six months after preparation.

As can be seen in Figure 2.43, the suspensions show consistent stability over extended periods of time (with brief and low power re-agitation). This allowed suspension samples to be bath sonicated prior to use for all experiments, however no suspension older than 4 weeks were employed within any studies.

B2.2.10 XPS

XPS is a surface sensitive technique that allows determination of the chemical environment of atoms. The surface of the sample is irradiated with X-rays causing the ejection of electrons. XPS is able to qualitatively determine the kinetic energy and quantitatively determine the number of electrons that have escaped from the material. XPS is commonly used for measuring the elemental composition, empirical formula, chemical state and the electronic state of a material.

The kinetic energy is generally determined using a cylindrical mirror analyzer (CMA). A CMA consists of two concentric metal cylinders with voltages applied at each end to create

157 an electric field. With a change of voltage, the acceptable kinetic energy will change, allowing the determination of the binding energy and number of the electrons.

XPS is highly informative for confirming the absence of oxidation for the prepared graphene microplates. A typical carbon to oxygen (C:O) ratio for GO is approximately 1:1.248 The C:O ratios determined for samples prepared via SALE were determined to be approximately 60:1. This is determined from the C1s peak which indicates that the bonding of the carbon atoms is predominantly sp2 hybridized.

Figure 2.44. XPS scans of CTAB-exfoliated graphene. (a.) Survey scan (b.) carbon 1s scan, (c.) oxygen 1s scan.

Therefore, from the XPS data the information of the chemical state from the topmost 1 – 12 nm can be collected. The carbon 1s scan shows the predominant chemical bonding state of the carbon to be C-C, determined from the binding energy. The smaller band beneath the graphitic C-C band likely corresponds to the C=C state. Importantly the scans do not represent that of GO, which will present a large C-O peak at 286.2 eV.106, 130, 249

Figure 2.45. XPS C1s spectra for dialyzed and non-dialyzed CTAB-graphene

It is important to consider the influence the surfactant may have on the measurement of a surface specific analysis tool such as XPS. To better understand the spectra collected, dialyzed and un-dialyzed graphene suspension (prepared as filtered films) were analyzed showing the removal of a small peak at approximately 284 eV likely corresponding to the C- N state.

Typical XPS sample preparation

Exfoliated graphene suspension was filtered to form a thin film using a 0.02 μm pore Anopore inorganic alumina filter (Whatman) purchased from Interpath Services Pty Ltd.. Smaller surfactant samples (CTAB MW = 364.45 Da) were selected to allow efficient filtration. Dialysis was performed prior to filtration to remove the vast majority of excess surfactant using high retention seamless cellulose dialysis tubing purchased from Sigma Aldrich with a molecular weight cut-off of 12294 Da.

159

A 20 mL aliquot of the 1 mg mL-1 SALE CTAB-graphene was filtered using a vacuum filter. The graphene sample filtrate was then washed with 100 ml of MilliQ filtered water and 25 mL of propanol without allowing the film to dry (Leaving 10 mL of liquid above the filter surface until all liquid had been filtered). After 5 minutes of drying under the vacuum, the filtrate was then placed in a drying oven at approximately 60 C for 30 minutes, in order to remove excess moisture.

B2.2.11 SANS

SANS is a well-established scattering technique for microstructure investigations which uses elastic neutron scattering. This may occur either via nuclear interaction with nuclei or by the magnetic moments associated with unpaired electrons (dipoles) in magnetic samples.250

SANS differs significantly from small angle X-ray scattering (SAXS) (which will be focused on in greater detail in Chapter 6) in that neutron scattering results from interactions with nuclei, while X-ray scattering arises from interactions with electrons. Since the interaction probability of a neutron is small, the neutron is generally able to penetrate well through condensed matter. Neutrons can also be reflected by some surfaces at incident glancing angles allowing them to be used for both bulk condensed matter and surface probes.

Neutron scattering is highly suited for condensed matter investigations as thermal/cold neutrons are a non-invasive probe, the do not alter the sample as they do not deposit energy into it. Neutron scattering lengths range significantly with the atomic number of the sample and are independent of momentum transfer Q. This allows deuterium to be used as a label in -13 SANS experiments giving significantly different scattering lengths (bH=-3.3739*10 cm -13 250 and bD = 6.671 * 10 cm) respectively.

A range of deuterated CTAB graphene suspensions were produced in H2O and D2O to allow for the matrix matching required for the SANS analysis. The Guinier plot obtained aided in approximately characterizing the thickness of the SALE graphene sheets as well as characterizing the sample as being platelet-like (Figure 2.46).

When D (the fractal exponent of the scattering objects) is representing a sample of fully dispersed platelet-like objects its value is ≈ 2.251 When larger agglomerates are present the D value will move more towards a range of 3 to 4.252 The gradient obtained for the low q range was -3.01 which suggests a slightly agglomerated dispersion of platelets. Furthermore, the presence of a uniform line in the low q range indicates the thin, platelet-like nature of the sample (Figure 2.46).

Figure 2.46. SANS Guinier plot of CTAB-graphene.

The SANS data was particularly useful in identifying the stability of the exfoliated graphene sheets as SANS data can readily identify whether graphene is present as single-layer or as a more agglomerated species.252 This is particularly relevant for all samples produced via the SALE method as the tendency of the materials to aggregate is an important consideration.

161

Typical SANS sample preparation

In order to prepare the exfoliated graphene sheets for the SANS, several graphene suspensions were prepared in both H2O and D2O mediums as well as with varying deuterated CTAB, allowing for matrix matching. The CTAB-graphene batches were produced using fully hydrogenated CTAB, tail deuterated CTAB, head hydrogenated CTAB, as well as fully deuterated CTAB, all in both H2O and D2O. All samples were produced via the single addition SALE method and treated as outlined in Section B2.1. All SANS analysis was performed at the National Institute of Standards and Technology (NIST). The SANS data were collected over a q range of 0.0036 to 0.5936 Å-1.

B2.3 General laser parameters

To demonstrate the versatility of graphene as a broadly absorbing NIR agent, several different lasers were used throughout this study, with a range of varying parameters such as beam collimation, distance from source, power and wavelength. It is important to be able to clarify a number of these parameters briefly.

B2.3.1 Lasers used in photothermal experiment

There are many types of lasers in use both in commercial industry as well as in laboratory development, which are generally classified by the laser medium. The three most widely used laser classes are gas, solid-state and semiconductor lasers.

Gas lasers are the most commonly used industrial lasers and can emit a wider range of wavelengths than the other classes. Some common gas laser mediums are helium-neon (He-

Ne) and carbon dioxide (CO2) lasers. CO2 lasers have a wavelength of λ = 10.64 μm and are often employed for cutting and joining of metallic materials.253 Solid state lasers consist of a laser medium of a nonconductive solid doped with laser-active ions. These ions are excited by optical pumping (by laser diodes) requiring the host material to be transparent to the absorption and emission wavelengths of the laser-active components. The Nd:YAG (neodymium:yttrium-aluminium-garnet) laser is a commonly used solid-state laser which can achieve continuous output powers in the kilowatt range. The Nd is the laser- active component.254-255

Semiconductor lasers consist of a semiconducting laser medium such as GaAs (wavelength 808 nm) and InGaAs (wavelength 910 – 980 nm) and are generally lower power lasers with a high electrical to optical efficiency. As with any laser, a diode laser requires a gain medium within an optical cavity and the semiconductor bandgap determines the emission. Semiconductor laser diodes are current pumped lasers and can operate at room temperature. These lasers can, however, be combined to produce lower beam quality kilowatt output powers when in an array.253, 256-258 Efficiencies have increased dramatically in recent years to examples of 100% output of photons in respect to injected electrons.

Two lasers were used for the photothermal experiments, an 808 nm and a 980 nm laser. The first laser employed was an 808 nm continuous wave diode laser with tunable power and was fitted with an optical fibre connection with FC-connector. The 808 nm diode laser was powered by a 6645A DC power supply with the output power controlled by the selected current. The 808 nm laser diode has a maximum output of 35 W and is classed as an IV laser product. The power output was measured using a PowerMax500D, Molectron Detector Incorporated.

The second employed was a 980 nm continuous wave diode with an output power of 185 mW fitted with an optical fibre connection. This was a customized laser diode built with a butterfly diode and mount and controlled using a modular laser diode controller LDC-3900.

163

A3.0 Chapter 3A - Photothermal properties of graphene

Prior to exploring direct applications of graphene in photothermal roles, it is important to understand the photothermal ability of the material and clearly understand the parameters defining the action. Initially, we can separate this into two categories of the optical properties of graphene, and the thermal properties.

Once the base optical and thermal properties have been reviewed, the practical parameters of the photothermal systems can be explored in greater detail.

A3.1 The photothermal process

When the energy of a photon is equal to the difference between the ground state and an excited state, absorption occurs. As the photon strikes the atom or molecule, an electron is promoted to a higher level.259 When graphene particles are irradiated with near-infrared (NIR) light, free electrons within graphene absorb the energy of the photons, and are excited, resulting in an increased kinetic energy. This then leads to electron-phonon scattering, as the electrons relax through electron-electron scattering. The electron-phonon scattering results in an increased particle temperature. The increased particle temperature then leads to energy exchange with the surrounding medium through phonon-phonon coupling.260-261

A3.2 Optical properties

A3.2.1 Broad absorbance spectrum of graphene

The broad absorbance profile of graphene indicates that there are electrons present at a wide range of energy states, enabling the absorption of the light for a wide range of incident energies. In comparison, diamond, which consists of the same atoms but in a tetrahedral sp3

165 configuration, has a wide transmittance range (from ultraviolet, UV, extending to lower energy electromagnetic radiation, EMR, at far-infrared, FIR, wavelengths). The wide transmittance range of diamond is due to the nature and strength of the diamond bonds and the corresponding bandgap (5.48 eV at 25 °C).262

Graphite, however, is optically opaque even when at a very limited thickness.17 Optical transitions within graphite from approximately 0 to 9 eV (137.76 nm), correspond to intra- and interband optical transitions involving primarily the π bands.263 These excitations have been shown to involve one electron per atom, the 2pz orbital electrons (extending perpendicular to the plane, which form the π bond. A broad optical absorption occurs near approximately 15 eV (82.66 nm) which is associated with the interband transitions involving three sigma electrons per atom which form the planar bonding network within graphite.263

Graphene is a black body absorber with an emissivity of 0.99 over a wide range of wavelengths.264 A graphene suspension (of sufficient concentration) appears completely black, indicating that all visible light has been absorbed.259 To understand the absorbance profile of graphene we must first transition to the relevant properties of the electronic structure.

As shown in Figure 3.1, the absorbance of graphene is strong and broad covering a wide range of wavelengths out into the NIR with an almost flat profile. This allows the biologically relevant NIR region to be employed. To define the photothermal ability of single and few layered graphene as a photothermal agent, the base parameters of material in this role must be thoroughly explored.

Figure 3.1. Absorbance spectra for surfactant (F108) exfoliated graphene, fully reduced graphene oxide, partially reduced graphene oxide (data from 66) and graphene oxide extending out past the NIR region. Conjugation versus absorbance schematic. All samples are set to 1 mg mL-1.

When comparing the optical properties of graphene oxide (GO), reduced graphene oxide (rGO) and pristine surfactant exfoliated graphene the role of the conjugation can be demonstrated clearly. The absorbance of the graphene oxide materials drops off sharply at higher wavelengths than that of the other graphene materials. When the conjugation is recovered partially through reduction this can be seen to increase and extend further out towards the infrared (IR). Pristine, fully conjugated graphene presents an absorbance profile which extends almost as a flat line out into the IR region.

It has been well established that for graphene oxide the absorbance spectrum has two characteristic features. A peak is present at approximately 233 nm, corresponding to the π- π* transition of the C=C bond. The shoulder at approximately 290 – 300 nm corresponds that of the π-π* transition of the C=O transition.265 For pristine graphene there is a single peak at ≈ 270 nm which corresponds to the π to π* excitation at the van Hove point.266

167

The broadband absorbance of graphene is a result of the broad ranging energies of the bandgap. The broad absorbing capability is due to graphene being a zero-gap semiconductor, which is to say that its bandgap touches at specific locations (K and K’ points), whilst there are substantial energy gaps at other positions of the bandgap (Figure 3.2).

The tight-binding model allows the calculation of electronic band structure, it relies on the electrons considered, to be tightly bound to the respective atom, and the electrons to have a limited interaction with surrounding atoms. The tight-binding approximation calculates the band structure from the wave function of the atom and assumes that the electron will have a similar wave function to the atomic orbital of the atom. When these conditions are met (i.e. negligible overlap or influence from neighboring atoms/electrons) the ionization energy will be a close approximation of the electron energy.267 The most influential parameter on the tight bonding model is the interatomic matrix elements, or simply the bond energies.

Figure 3.2. Bandgap of monolayer graphene with the Brillouin zone (BZ) positions labelled. Adapted with permission from Elsevier.55

The degeneracies at the K and K’ positions are referred to as Dirac cones. It is the presence of the K point degeneracies of the graphene bandgap (shown in Figure 3.2) which result in the electrons near these points behaving as massless Dirac fermions characterised by 6 -1 37, 55 velocities of vF ≈ 10 m s . At the Γ point of the BZ the band gap is at its widest with an energy of ≈ 20 eV.9 At low energies of incident radiation the optical response is dominated by the free carrier response (intraband). In the mid- to the NIR regions, the optical absorbance is dictated by the interband transitions and is nearly frequency independent.9

At very low incident energies, for example far infrared radiation, absorbance within the π bond results in intraband excitations, corresponding to positions close to the Dirac points.9 As the energy of the incident radiation increases, for example to the visible region, the electronic excitations transition to interband excitations. These events correspond to positions slightly further away from the K points.9 As the degeneracy points bring the band gap to the Fermi level, and the band gap extends right out to 20 eV at the Γ position, a broad range of light energies can be absorbed.

As the incident energy increases further, to the UV, the transitions move toward the M saddle point (where the bandgap energy is approximately ≈ 4.5 eV, the van Hove singularity)55-56 and the massless Dirac fermion behavior of the electrons breaks down. This can be observed within the absorbance scan shown in Figure 3.1 as the absorbance peak at approximately 270 nm. It is this broad energy range of the bandgap which results in the broad absorbance of graphene.

As the broadly absorbing spectra of graphene is dictated by the electronic structure, any disruptions to the hexagonal carbon lattice will result in changes to the absorbance profile. This is particularly evident when comparing the spectra of graphene materials produced via chemical oxidation routes to that of pristine graphene which gives a direct insight into the role of the defect free, fully conjugated nature of graphene on the optical properties.

169

A3.2.2 Quantitative absorbance properties

In addition to the broadband absorbance spectrum, monolayer graphene also transmits a highly specific 97.7 % of incident light per layer (2.3 % opacity) independent of the wavelength within the visible and (a portion of the) IR spectrum.8, 55 This arises because of the electrons near the Dirac points behaving as massless Dirac fermions. Electrons with such behaviors will have a universal dynamic conductivity which results in the transmission in these regions being solely defined by the fine structure constant.8

The fine-structure constant is a fundamental physical constant characterizing the interactions between elementary charged particles. This describes the coupling between light and relativistic electrons where charge of an electron (e) represents the strength of coupling of an elementary particle with the electromagnetic field via the following equation:

2 4휋휀표ℏ푐훼 = 푒 Equation 3.1

Where α denotes the fine structure constant, εo is the permittivity of free space, ℏ is the reduced Planck constant (4ћ=h/2π). The unitless value of the fine structure constant is given as 7.297 x 10-3 (≈ 1/137) as measured by the National Institute of Standards and Technology (NIST).

It is the nearest (and next nearest) neighbor electron hopping from the A and B inequivalent atomic sub-lattices which leads to the conductance of graphene sheets to be independent of the photon energy.26, 37 Further, electrons close to the Dirac points will exhibit the high frequency (dynamic) conductivity equal to e2/4ћ allowing the behavior of the Dirac fermions to dictate the this universal dynamic conductivity to give8:

휎(휔) = (1.01 ± 0.04)푒2/4ℏ Equation 3.28 This corresponds to an absorbance of:

4휋 퐴(휔) = ( ) 휎(휔) = 휋훼 ≈ 2.29% Equation 3.355 푐

The zero-effective mass of the electrons near the Dirac point result in velocities of approximately 106 m s-1, ≈ 1/300 the speed of light. These particles are therefore appropriately described by the Dirac equation rather than the Schrödinger equation.

The region of interest for this study is the NIR region, where the 2.3% opacity, is independent of frequency over a wide energy range. The Dirac cones of graphene have a linear dispersion within approximately ± 0.6 eV.54 Within this energy range the electrons have an effective rest mass of zero and as such they follow the Dirac equation and present the approximately linear 2.3 % opacity per layer.

A3.2.3 Graphene absorbance - quantitative perspective

A single layer of graphene absorbs ≈ 2.3 % of visible light, as such, graphene has largely been explored as a transparent material for a range of applications.103, 106, 111, 236 Even though graphene has been explored considered for transparent roles, the opacity of 2.3 % for a single layer of atoms is certainly not negligible and does open up a number of valuable opportunities. Further, with additional layers the absorbance grows approximately linearly with substantial absorbing capabilities being achieved with only a small addition to sheet numbers as shown in Figure 3.4.9

171

Figure 3.3. Absorbance of graphene as a function of sheet number. Reproduced with permission from IOP Publishing.9

It is due to the behavior of the electrons near these Dirac cones that dictates the specific opacity of graphene at 2.3 %, independent of the wavelength, as well as being almost linearly additive.9 It was recently demonstrated by Zhu et al. that the additional 2.3 % opacity of subsequent layers is not strictly linear, showing a subtle decrease in absorbance with increasing layers. This has been attributed to the optical conductance of the graphene stack, specifically the interlayer interactions which result in the slight decrease in opacity per graphene layer.

It is also important to note that the reflectivity of graphene cannot be neglected with increasing layers. While the reflectivity at 1 layer is < 0.1 %,8 when 10 layers are present, this increases to ≈ 2 %.268-269 From this information, we can see that suspensions consisting a sheet thickness distribution, will likely present attractive absorbance properties, a critical parameter for the role of a photothermal agent. A3.3 Thermal properties- general and applications focus

The thermal conductivity of graphene has been reported as high as ≈ 5000 W m-1 K-1.1-5 As such, the presence of graphene within a drug delivery matrix or aqueous suspension may impart some improvement to the respective system thermal properties. There is however, a large difference between the thermal conductivity within a single, pristine graphene sheet and the enhancement this material may have on a larger scale. As such, several considerations need to be addressed, such as the interfacial resistances to heat transfer (Kapitza resistances), the thermally insulating polymer surfactant coatings, the sheet thickness distribution which is inversely proportional to the thermal conductivity, and the size of the sheets in suspension.

Some key focusses of the thermal properties are the specific heat, the thermal conductivity, the thermal dissipation from the material and the stability of the materials at higher temperatures.

A3.3.1 Phonon dispersion in graphene

To understand the thermal properties of graphene the vibrational modes (phonons) must be understood as they contribute to ≈ 99 % of the thermal conductivity within graphene.261 From the interatomic interactions and space group symmetry the dynamic matrix can be determined.73 As discussed briefly in Chapter 1, the unit cell of graphene consists of 2 carbon atoms which results in 3 acoustic (A) and 3 optical (O) phonon modes. It is generally accepted that due to the low velocities, the optical phonons contribute little to the thermal transport of graphene.261, 270 As the heat transfer within crystals is dominated by the role of the acoustic phonons, it is there that we will focus our attentions.

The high thermal conductance in graphene is due to the ballistic, scattering-free heat flow characteristics of the material.3 Near the centre BZ, the transverse (TA) and longitudinal (LA) 4 -1 acoustic modes have linear dispersions. The high group velocities vTA ≈ 1.36x10 m s and 173

4 -1 2 vLA ≈ 2.13x10 m s are attributed to the strong in-plane sp bonds and the small mass of carbon atoms.68-71 The flexural mode of graphene has also been established to play a critical role in the superior thermal conductivity of graphene.3, 73, 75

The flexural mode (ZA) is a characteristic mode for solid plates or rods and the free surface boundary condition results in two sets of Lamb waves, one symmetrical and one asymmetrical.73 The lowest frequency asymmetrical Lamb wave is the flexural mode which has a defining parabolic dispersion curve. In contrast to the high velocity TA and LA modes, the ZA mode is quadratic and is the easiest mode to be excited due to its low frequency and therefore carries most of the vibrational energy.73 This is of particular importance in graphene due to the very large specific surface area, where essentially all atoms are exposed and as such, all phonon modes are surface modes.73

A3.3.2 Specific heat

When considering the thermal properties of graphene, with a focus on the transfer of light to thermal energy, the specific heat properties, particularly in terms of heat dissipation are important factors. Specific heat is defined as the amount of energy necessary to raise the temperature of a specific quantity of a particular material by 1 °K (J g-1 K-1).202 As such the specific heat determines the thermal energy stored and the rate of cooling, represented by the thermal time constant τ:

휏 = 푅퐶푉 Equation 3.4

Where R is the thermal resistance for heat dissipation, V is the volume of the body and C is the specific heat. The thermal time constants for nanoscale objects are often very short, graphene has been shown to have a thermal time constant of 0.1 ns,271 and 1 ps for the relaxation of individual phonon modes.272-273 The specific heat is stored by both lattice vibrations as well as by free conduction electrons.3 However, phonons dominate the specific heat of graphene at all practical temperatures.274-275

-1 -1 At room temperature, the specific heat of graphite is CP ≈ 0.7 J g K , which is 30 % higher than that of diamond due to the higher density of states (DOS) at low phonon frequencies arising from the weak interlayer forces.276-277 This is expected to be similar for graphene at room temperature as the ZA modes would likely be fully excited. In the event of suppression of the ZA modes (say through growth on a substrate), the specific heat can be decreased.3 As such we can expect freely dispersed graphene sheets in suspension to present no significant ZA suppression.

From this we can understand the specific heat of graphene will likely be quite like that of graphite and that the dissipation of heat would likely be a faster process. This is an important consideration when the desired role for this material is to be repeatedly heated to high temperatures. The fast dissipation of thermal energy may allow thermal damage to be minimised and therefore the protection of attractive material properties, along with a more efficient heat delivery mechanism.

A3.3.3 Thermal conductivity

Thermal transport occurs in the presence of a temperature gradient where both electrons and phonons play an important role as thermal energy carriers.73 Diffusive thermal transport occurs in large systems at or at high temperatures where the thermal transport ability is limited by phonon scattering mechanisms.73 Therefore in the diffusive regime the thermal conductivity is constant with respect to the length of the structure. The electronic contribution of thermal transport is ≈ 10 W m-1 K-1 which is a negligible contribution to the overall thermal conductivity.278 The phonon contribution to the thermal conductivity is therefore the dominant component and is referred to as the lattice thermal conductivity.

175

Quantum physics dictates that a discrete amount of energy is required to trigger vibrations within a given system. The quantity of this required energy is equal to the frequency of the vibration multiplied by Planck’s constant. As carbon atoms are small in both mass and size and tightly bound within diamond, graphite and graphene, the quantum energies required to achieve the vibration is large. Therefore, the vibrations occurring predominantly at high frequencies. This is important as a common limitation to phonon transfer within a crystal is the presence of anharmonic phonon scattering. At ordinary temperatures there a few atomic vibrations present that can impede the passage of thermal waves and therefore the thermal conductivity is observed to be quite high for these three carbon systems systems.17

The thermal conductivity of a material is the speed of thermal transfer over a specific distance with units of W m-1 K-1 (J m-1 K-1 s-1).67 The thermal conductivity κ of a material can be related to the specific heat by:

휅 ≈ ∑ 퐶휈휆 Equation 3.5

Where ν and λ represent the appropriately weighted phonon group velocity and mean free path respectively.3, 279

The thermal conductivity of graphene microplates may provide a key advantage for this material in its use as a photothermal agent through high thermal dissipation. An increased heat distribution from the localised heating site could allow the graphene sheets to remain below the thermal oxidation point, allowing repetitive activation in the system. Ultimately providing enhanced heat dissipation from individual particles and bulk phase heat transfer improvements.280-281

As stated previously, the thermal conductivity has been reported as high as ≈ 5000 W m-1 K- 1 for isotopically purified samples with large grain sizes.1, 13 To give a comparison, the thermal conductivity of natural diamond is ≈ 2200 W m-1 K-1 and isotopically pure diamond is ≈ 3200 W m-1 K-1.75, 282 Additionally, graphite in plane thermal conductivity at room temperature is ≈ 2000 W m-1 K-1.283

However, in the c axis graphite presents a thermal conductivity that is far lower, 6 W m-1 K- 1 at room temperature. This significant decrease is attributed to the inter-plane van der Waals (vdW) interactions (analogous to graphene attached to a substrate). The interlayer thermal conductivity of graphite is far greater (18 GW m-1 K-1) which indicates that the thermal resistance across the interface dominates that within the material.3

The flexural modes have been shown to play a critical role in enabling the thermal conductivity of graphene to be so large.76 This has been demonstrated by looking at the suppression of the ZA modes when attached to a substrate (compared to suspended graphene) highlighting the important role of this flexural mode.

The thermal conductivity can also be observed to decrease when the width of the material (graphene nanoribbon) becomes narrower than the phonon mean free path, which is reported 284 as λ0 ≈ 600 nm for suspended graphene. The confinement leads to greater phonon scattering and a subsequent loss of thermal conductivity.

When the characteristic dimension of a material becomes comparable to the wavelength of the mean free path of electrons and phonons, quantum confinement and classical interface scattering effects can emerge and strongly influence to material properties.75

177

The de Broglie wavelength of an electron can be defined by the following equation:

2휋ℏ 휆 = Equation 3.6 푒 √2푚∗퐸

Where ћ is the reduced Planck constant, m* is the electron effective mass and E is the electron energy.

Due to such quantization restrictions, the thermal transport properties can present a reversed thickness dependence, i.e. with thinner materials the thermal conductivity of nanomaterials increases.75 In two-dimensional systems, the allowable wave vector states lie on well separated planes in the reciprocal space, (importantly only one plane for single layer graphene). As such, materials with one or more dimensions being comparable or smaller than the wavelength of the electrons or phonons result in a strongly restricted phase space that can satisfy the energy and momentum conservation requirements for phonon-phonon scattering.

Further, atomically smooth surfaces, such as suspended flat single layer graphene cannot scatter phonons diffusely due to the absence of a velocity component perpendicular to the surface, i.e. a flexural ZA mode.

Thermal conductance (σ) and thermal conductivity (κ) can be related from the defining equations:

휅 휎 = Equation 3.7 퐿 푠

Where s and L are the cross-sectional area and the length respectively. The thermal conductance is relevant for the ballistic transport regime, a common transport mechanism for nanomaterials. Ballistic transport is the transport of quanta of thermal energy ћѠ via phonons (or electrons) through a conductor without scattering, the electrons do not collide with anything that changes the energy or momentum. The term arises as the electron behaves like a projectile traveling through the conductor.285

Ballistic transfer usually only occurs at very low temperatures where the phonon density is too weak for phonon-phonon scattering. The ballistic thermal conductance therefore is not dependent of the length of the system because of the infinite phonon mean path.68, 286 Graphene has recently been shown to have crystal quality high enough to allow ballistic transport at room temperature.286

The contribution from the ZA mode in the ballistic regime at low temperatures has been shown both theoretically and experimentally to follow a temperature dependence scaling at T1.5.286-288 There is no universally accepted fact on the relative contribution of the ZA mode to the thermal conductivity in graphene with contributions estimations ranging from 70 % to 20 %.289-290 However, the consistency of the temperature dependence scaling of the ZA mode lends strong support to the quasi-ballistic regime of the thermal transport in graphene.3, 291

To review the thermal properties of graphene briefly and its relevance within this project, we can surmise that the specific heat value (slightly larger than that of graphite) will likely influence the cooling rate of the particles. As the phonons dominate in the contribution to specific heat, in particular the flexural mode phonon increases the specific heat capacity, the material will release heat at a slower rate however the heat transfer coefficient is of greater importance.

179

The interfacial thermal conductance, however, is considerably lower than that of the in-plane conductance with reports ranging from ≈ 20 to 200 MW m-2 K-1.3, 75, 292 Due to this significant decrease, the benefit of the remarkable thermal conductivity will likely not significantly improve the thermal dissipation and therefore the thermally induced oxidation of photothermally heated graphene should be carefully studied.

To explore why the interfacial thermal transport is so much lower than that of the in-plane thermal transport properties, thermal boundary resistance mechanisms must be explored.

A3.4 Interfacial thermal resistance

Heat transfer across the interface between two materials is usually accompanied by a finite temperature drop.293-296 This phenomena was first reported by P. L. Kapitza in the study of heat transfer across helium/solid boundary.296-297 This decrease in thermal transport across interfaces between two media is usually limited by the mismatch of phonon spectra in the two media.261, 297-298

Figure 3.4. Steady-state temperature field across the liquid-nanoparticle interface where the temperature can be seen the decrease dramatically with the distance from the particle showing the characteristic drop of the Kapitza length (x-axis unit radius of particle). Reproduced with permission from the American Physical Society.299

To estimate the rate of heat loss from the particle to its surroundings QW, the following equation can be used:

푄푊 = 퐴푠푢푟푓푎푐푒 ∙ 퐺 ∙ (푇1 − 푇푤,푠 ) Equation 3.7

Where the surface area of the particle Asurface and the water temperature at the particle surface

(T1) are combined with the thermal conductance at the water-particle interface (Tw,s). The thermal conductance (G) relates the temperature drop at the interface to the heat flux crossing the interface. This constitutes the coupling parameter between the particle and surrounding medium.260

The interfacial thermal conductance for graphene has been shown (coupled with a SiO2 substrate) at ≈ 50 MW m-2 K-1 a considerably lower value than that of the in-plane thermal conductivity of graphene, which will decrease the heat dissipation from the sheets 181 significantly.291 Further, when a larger body of work is considered with different media, the interfacial thermal conductance can be seen to be considerably lower than that of the in-plane conductance with reports ranging from ≈ 20 to 200 MW m-2 K-1. 3, 75, 292

As the interfacial thermal conductance decrease is largely attributed to the phonon frequency mismatch, a number of studies have been exploring these interactions to elucidate the underlying mechanism of the dramatic decrease.291, 293, 300-301 It was proposed by Hu et al. that as the heat transfer phonons and the heat storage phonon modes are of substantially different frequencies the thermal energy must be transferred from one internal mode to the other.300 That is to say, as heat is added to graphene for example via NIR laser absorption, the heat is added to the higher frequency phonons predominantly, in order for this heat to be passed out it needs to transfer to the lower frequency phonons. This results in what has been referred to as a large “internal” resistance as the coupling between the high and low frequency modes is weak.261

The magnitude of the interfacial resistance R can be approximated with:

∆푇 푅 = Equation 3.8 푗

Where the heat flux across the (liquid-solid) interface is represented with j. The Kapitza resistance effects are particularly influential when the dimensions of the system (particle size) are comparable to the Kapitza length.261

푙퐾 = 퐾푅 Equation 3.9

Where K represents the thermal conductivity of the liquid. It is apparent that the large majority of the attractive thermal properties are lost through the interfacial thermal resistance from phonon scattering at the interface between the two phases. This is of particular relevance as throughout this study the photothermally employed graphene sheets are either in aqueous suspension, in an emulsion or within a polymer matrix. As such the mismatch of phonon frequencies will be significant and as such we can expect little improvement to the heat dissipation, despite the dramatic thermal conductivity of graphene.

This effect was convincingly demonstrated by Shima et al. who showed that with materials - with a dramatic thermal conductivity contrast (k of bulk Ag and Fe3O4 are 429 and 9.7 W m 1 K-1 respectively) the same bulk k enhancement was observed (in the same base fluid.302

183

Figure 3.5. Thermal enhancements for materials of significantly different thermal conductivities. Ag and Fe3O4. Reprinted with permission from Shima et al., Copyright 2014, American Chemical Society.302

A3.5 Thermal properties of graphene in suspension

The most appropriate way to assess the potential thermal enhancements in an aqueous photothermal role is by considering the system as a nanofluid and drawing from the information collected in this newly emerging field. By doing so, we can understand a range of key components to this system including the interfacial resistance, stability of the particles, volume fraction, aspect ratio, distance between particles or the potential percolating clustered pathways. The field of thermal transfer in nanofluids is currently rife with controversy with a convincingly large number of experimental groups showing anomalous results of far greater enhancements than theoretically predicted.303-305

The thermal conductivity variation in stable nanofluids can be accounted for via the Maxwell equation derived from the effective mean field theory.306-307 Through this approach the thermal conductivity of a nanofluid is determined by the thermal conductivity of the base fluid, the nanoparticles, and the volume fraction of the nanoparticles. A3.5.1 Ballistic transport persistence

The possibility of ballistic transport extending out from one particle and reaching a nearby neighbouring particle is one potentially significant mechanism by which the bulk thermal properties of a nanofluid may be increased. The phonon mean free path in liquid is severely limited and can be estimated at approximately 1-2 nm. However, ballistic transport could persist in small interparticle distances to improve bulk k.308

A key consideration in this regard is the distance between particles in suspension whereby distances of < 5 nm may allow ballistic phonons to reach a neighbouring particle. Brownian motion of the particles would further reduce such distances allowing practical volume fractions (< 5 wt. %) to allow connections.

Liquid layering around the graphene particles may also contribute to a greater than anticipated bulk thermal conductivity. The simple effect of increased ordering of water around the particle will lead to greater heat transfer through this region of the medium compared to that of the bulk.309 This idea builds from the fairly straightforward concept that with increasing pressure the matrix material density and interfacial bonding properties are influenced in such a way as to enhance the propagation of phonons.293, 310

A3.5.2 Clustering and percolation

The creation of connected paths throughout the liquid medium via transiently contacting particles can lead to significant increases to the thermal conductivity of the bulk material. Interestingly, despite the dynamic and transient nature of such connections the thermal contributions can be substantial. The connections can however, be due to and potentially lead to the instability of the suspension ultimately leading to collapse. Clustering events have been experimentally observed, and these events are expected to influence the bulk thermal

185 conductivity significantly as the heat can move much faster through these solid high- conductivity phases.303, 309

A3.5.2 Aspect ratio

As the clustering events in suspension lead to such a dramatic enhancement of the thermal conductivity, exploring materials which could potentially create such transient networks more efficiently is a logical next step.308 It has been demonstrated in a number of studies that the increase in the aspect ratio shows an increase in the k enhancement.311-312 This lends additional substantial weighting to the clustering transport improvement mechanism. By targeting materials with greater aspect ratios these properties can be harnessed and used to increase the likelihood of such events.

By combining these potentially attractive properties with a stable suspension, perhaps these connections can be employed to see stable, thermally superior suspensions. Such approaches have been explored through linear aggregates where networks can be triggered externally, to switch between high and low thermal conductivities.313

This therefore lends the suspensions to be tailored toward higher concentrations and larger particle size and high aspect ratios in order to achieve percolation connection events.

A3.5.3 Layer number influence on interfacial resistance

The sheet edges of graphene promote strong boundary scattering of the phonons, resulting in a decreased thermal conductivity. This trend has been demonstrated to be strongly layer dependent and that the resistance will decrease with increasing layer number.314 It has been shown that when the layer number is below 28 the thermal conductivity will increase with temperature.314 Additionally, regarding nanoparticle suspensions, it was recently demonstrated that the Kapitza length is dependent of the number of layers of graphene in few layer graphene.295

A3.5.4 Polymer/surfactant coatings

As it has been established that to increase the thermal transfer from within the particle to the liquid medium the phonon spectra overlap must be approximately overlapped, the functionalisation of particles will likely play a critical role. From a microscopic point of view the heat transfer across the interface is via the interaction of the atoms in each phase, increasing the interactions between them will enhance the heat transfer.261 One approach to achieve this is to attempt to bridge the phonon spectra by adsorbing an intermediate group to the surface which can act to transfer the phonon energies between the two faces. In this scenario ensuring a strong (covalent bonding) connection to the particle is important along with ensuring the intermediate possesses similar phonon spectra to each phase.315-316 It has been demonstrated that a weaker connection between the adsorbed material and the nanomaterial surface, will result in an increased interfacial resistance.317 The reverse has also been demonstrated showing that with compressive stress the interfacial resistance is enhanced.

The method of functionalisation is a key consideration as the introduction of chemical species to the graphene microplates will likely affect and ultimately interrupt the electronic structure, disrupting the targeted attractive properties. It has been demonstrated that a grafting density of as little as 0.0032 Å-2 leads to a decrease in k of ≈ 69%.318 As the introduction of functional 187 groups occurs, the sp2 hybridisation is broken, (commonly to sp3) resulting in a softening of the high-frequency phonon modes.318-319 This highlights the importance of the route of introduction of an intermediate, as well as shows the dramatic decrease in favourable properties which can result from a small extent of surface functionalisation.

It is important to note that the adsorption of surfactant on the surface of the graphene could potentially lead to a minor disruption to the electronic structure. However, the adsorption energy is far lower than that of a covalent bond, as such, it unlikely that any significant disruption will occur.

The graphene sheets prepared via surfactant assisted liquid exfoliation are subsequently coated in adsorbed surfactant. The selection of this surfactant allows a large degree of control of the surface properties of the sheets which can be incredibly useful in many scenarios. The presence of surfactant on the surface of the graphene microplates is however, quite likely to contribute to the interfacial resistance across the graphene liquid boundary as most polymers have a thermal conductivity in the range of approximately 0.5 W m-1 K-1, lower than that of water ≈ 0.6 W m-1 K-1.318, 320

As the heat conduction within polymers occurs via diffusive vibrational modes, the propagation of such modes is very short (on the order of a few bond lengths).293, 318 Long chain polymers accommodate bends and curves with very little resistance or energy penalty. As such, the configurational entropy drives polymer chains to assume curvilinear shapes, forming highly disordered and entangled amorphous structures. These random orientations lead to a tortuous path for the propagation of phonons.320 Therefore the distance phonons can travel within amorphous polymers is considered to be less than 10 nm.320-321

A balance must be made between a strong interaction energy to allow good phonon coupling from the nanoparticle to the polymer layer, while ensuring that the interaction of the bridging material is not so strong as to interrupt the electronic structure significantly. A solution to this challenge is to target covalent linkages to the edges of graphene sheets to both achieve stronger interaction energies (≈ 5 eV293) whilst not disrupting the conjugate sp2 structure.

The size of the graphene sheets is an important factor both on a macroscopic and microscopic scale. As mentioned previously, the clustering or percolation events can allow transient networks to be formed which can dramatically enhance the thermal conductivity. However, on the microscale the length of the sheets will also dictate the presence of longer wavelength phonons which have been shown to play a critical role in heat transfer from graphene to a polymer surrounding.293 This is most relevant on a very small size scale (sheet lengths less than ≈ 79 Å), however it does provide some insight as to the transfer mechanisms and routes to enhance such systems.

A3.5.5 Vapour formation around particles

As demonstrated by Roper in 2007322 and Kotaidis in 2005323 the heating of a localized nanoparticle leads to a thin layer of adjacent fluid being superheated. At temperatures of approximately 85 % of the Tcritical, the pressure is overcome due to the fluid surface tension and the liquid vapourises.322 The presence of this thin layer vapour cavity will therefore act as an insulating layer leading to non-equilibrium heating. This indicates that the temperatures observed in the bulk will underestimate the localised temperature at the photothermally activated nanoparticle site. Dramatic temperature gradients will be present at the site of the particle which result in ambient conditions being experienced as close as several particle radii’s away.314

These events have been thoroughly explored in relation to the particle morphology reversion of gold nanorods as the gold particles experience temperatures sufficiently high to melt the particle while the bulk temperature did not achieve the required temperatures.188 For graphene materials, a similar concern revolves around oxidation which would lead to a decrease in photothermal efficiency with the loss of aromatic structure. However, in the event

189 of oxidation events being observed (through Raman spectroscopy after the experimental heating) this could be used as an indicator of the localised temperatures occurring.

It has also been demonstrated that the increased pressure around nanoparticles, largely due to the curvature or the particle, (a factor which may not be so influential for the two- dimensional graphene) can act to prevent the formation of vapour. The presence of vapour formation around the localised heating sites, will completely block heat transfer due to the low conductivity of this phase. As such, the fluid around nanoparticles (assuming a sufficient radius) can sustain large heat fluxes well above the critical heat flux of the neat fluid.314

The sheet thickness distribution analysis in performed by Wang,233 Coleman,141 Lotya220 and Hernandez107 indicate the vast majority (≈ 67%107) of sheets are less than 5 layers thick. indicates that this material will predominantly present a thermal conductivity of ≈ 2250 W m-1 K-1.74 which is expected to impart a significant improvement to the bulk thermal transfer within suspensions. It has been demonstrated by Yu et al. in 2010 that the addition of GO and graphene microplates to aqueous solutions will provide a substantial increase to the bulk suspensions thermal conductivity. This group showed an 86 % increase in thermal conductivity for 5.0 wt. %.

A3.5.6 Potential oxidation during irradiation

A key property of the SALE graphene is the strong and broadband absorption allowing it to act as an efficient photothermal agent in the NIR. As discussed in Section A3.2, the attractive absorbance characteristics are due to the fully conjugated planes and minimal defect sites. It is therefore critical to ensure that at the high temperature being reached during irradiation, that the graphene sheets are not oxidising or losing this favourable structure in any other manner.

Oxidation events would also give an insight into the highly localized temperatures occurring at the nanoparticle as oxidation of graphene has been reported at temperatures of ≈ 300 °C in uncontrolled atmosphere.323-324 It is important to note that the oxidation of single layer graphene will occur more readily than that of a bi- or tri-layer graphene which have been reported to occur at ≈ 600 °C.324 This is particular pertinent for the single and few layer graphene, indicating a more robust photothermal agent.

The reaction of oxygen with a carbon surface such as graphite or graphene has been shown to have two primary steps, the activation of molecular oxygen on the surface and the stabilization of this species through the formation of covalent bonds.325 On a defect free region of graphite or single wall carbon nanotubes (SWCNT), oxygen has been demonstrated to adsorb and be bound with a binding energy of 0.12 eV, with no charge transfer between 326 the O2 and graphite indicating a vdW type interaction. The non-reactivity is attributed to the energy mismatch between the unoccupied states of O2 and the valence band of graphite which can be kinetically overcome at the site of a defect or edge.325

For materials with a higher availability of edges such as graphene, as opposed to (colloidal) graphite, the surface properties are dominated by the edge properties and as such the oxidation process predominantly follows the ‘one-step’ Eley-Rideal mechanism.327 The oxidation of a carbon surface happens by the direct collision of an oxygen molecule with a - -2 327 defect site, reducing the oxygen to species’ such as O , or O2 . The following intramolecular Coulomb repulsion breaks the O-O bond in an exothermic reaction with no energetic barrier, forming carbon oxygen functional groups.

- -2 O2(ads) + C + 2e  O2 (ads)

-2 - -2 O2 (ads) + 2e  2O (ads)

O-2(ads) + C  CO(ads) + 2e-

191

As the ratio between edges and defect free basal plane shifts from high edge quantity to low the mechanism shifts toward a two-step (adsorption then reaction). The most common oxidation products have been shown to be carboxylic acids, acid anhydrides and lactones.327

A3.5.7 Summary of thermal properties

Briefly, the consensus appears to be that the dramatic increases in nanofluids k enhancements is likely simply due to the instability of the suspensions, leading to aggregation. This is assumed to provide a short term (and less practical) increase in k through clustering of particles.

From this information, we can draw the conclusions that the thermal conductivity of surfactant coated graphene microplates will likely be imparting highly localised pockets of high thermal conductivity (primarily in-plane). The enhancement to the bulk medium properties will however, likely be limited due to interfacial resistance and the polymer layer.315-317 Through optimisation of characteristics such as particle size, aspect ratio, minimizing any defects which may limit the ballistic transport, the thermal conductivity could be enhanced. The graphene thermal activation of a drug release system or anti-cancer thermal ablation roles will likely be highly effective, but the thermal properties of graphene will likely contribute little to the broader system properties.

The extent of the contribution of internal thermal properties on the external heat dissipation within the system remains unclear with significant barriers requiring consideration such as the Kapitza resitance.294-296 However, percolation events313, 328 and Ballistic transfer linking309, 329 will likely result to allow substantial improvements on the bulk thermal transfer.308

193

B3.0 Chapter 3B - Assessing the photothermal properties of graphene

To practically assess the photothermal ability of graphene as a photothermal agent, a range of parameters must be explored. These include the achievable temperature ranges, the influence of concentration and laser power, as well as a direct measurement of the photothermal efficiency and the suspension thermal conductivity.

B3.1 Materials and methods

B3.1.1 Suspension selection

The graphene suspension for the photothermal assessments, was prepared and stabilised in F108 triblock copolymer surfactant, unless otherwise stated, which was selected due to its high relevance with the target applications (biocompatibility, stability, and yield). The suspensions were prepared via a single addition surfactant method, allowing tight control of the graphene to surfactant ratio and providing a substantial concentration range to work within. All suspensions were centrifuged at 3000 relative centrifugal force (rcf).

B3.1.2 Experimental design

The bulk suspension was bath sonicated for 5 minutes with minimum 10 inversions before and after the sonication bath prior to measuring out the required volume. A 0.5 mL aliquot was transferred into a 4 mL glass cuvette (cut down to hold a maximum of 2 mL) unless otherwise stated, and fitted into an insulated cuvette holder inside a light box. An 808 nm laser diode was positioned outside of the light box (to avoid an additional heat source) with an optic fibre positioned within the light box directed at the cuvette. A convex lens was used to focus the beam and was positioned such that the beam focussed in the middle (both face and depth) of the cuvette. This resulted in a spot size of approximately 1.7 mm and for all experiments referred to in this chapter, an 808 nm laser was employed at 500 mW, unless otherwise stated. Samples were also mixed via a magnetic stirring bar and plate for all experiments (Note: for photothermal efficiency experiments no insulation was used and no magnetic stirrer was present.).

A thermocouple was fitted to two automated arms allowing either y, z movement or x, y depending on configuration. The y, z configuration was used for most experiments and allowed precise positioning of the thermocouple tip at a fixed position from the incident beam and most importantly, reproducible positioning.

For all experiments referred to in this chapter the light box was fitted with an insulation layer to minimise the room temperature fluctuation effects on experiments. The temperature inside and outside of the box was also monitored to allow sufficient time for samples to re- equilibrate with the room temperature after heating experiments were required.

B3.1.3 Photothermal curve

Photothermal curves were produced to demonstrate the increase of bulk temperature as a function of irradiation time. These graphs show a rapid increase in bulk temperature which slows to a plateau when equilibrium is reached with the room temperature. When bulk temperatures are being reported, it is at a fixed irradiation time, representative of the maximum temperature, unless otherwise stated.

195

B3.1.4 Laser power

The second key parameter to explore was to determine an appropriate laser power to work with in addition to determining the relationship between laser power and observed bulk temperature. This was determined by ranging the 808 nm laser power from 100 mW to 500 mW at a fixed graphene concentration and recording the photothermal curves produced.

B3.1.5 Transducer concentration

The relationship between graphene concentration and the heating ability was also explored. This was achieved by preparing a graphene suspension at a substantial concentration of 0.65 mg mL-1 and then preparing a dilution set ensuring the only variable was the graphene concentration.

B3.1.6 Background heating

As NIR radiation can be minimally absorbed by water and surfactants, it is important to measure any background heating by water for the aqueous suspensions explored. These experiments were performed by transferring the 0.5 mL volume of MilliQ and surfactant solution at 0.1 wt. % filtered water into a cuvette and carrying out the appropriate 808 nm irradiation for 20 minutes in order to capture a relevant heating profile. This was then repeated for the relevant concentration neat surfactant solutions to confirm that no heat is being contributed from the surfactant.

B3.1.7 Photothermal cycling

F108-graphene samples were prepared as described in Chapter 2 with a concentration of 0.65 mg mL-1 and a looped photothermal curve experiment was performed with 10 loops. Samples were irradiated for 15 minutes and allowed to cool for 40 minutes between irradiation periods with the insulation removed from the light box to allow speedy cooling.

In order to determine any oxidation events during irradiation the absorbance and Raman spectra were measured as a function of complete cycle number. Raman and absorbance sample handling was performed as described in Chapter 2.

B3.1.8 Photothermal efficiency method

A cuvette was suspended in the air via a side arm clamp with no magnetic stirrer to avoid an additional heat source / sink (Figure 3.6 shows image of experimental setup). The sample was irradiated with ≈ 400 mW 808 nm for 15 minutes (with 40 minutes equilibration post and prior) and the concentration was set such that the absorbance was 0.1 a.u. at 808 nm to allow accurate determination of the transduction efficiency. The thermocouple was submerged 1 mm below the surface at a distance of 7.9 mm from the laser spot. An XPS-Q8 Universal high-performance motion controller/driver and two automated arms were used to provide reproducible positioning of the thermocouple, a key consideration when not mixing the samples during irradiation.

197

Figure 3.6. Image of the photothermal efficiency experimental configuration.

B3.1.9 Laser flash analyser

All thermal conductivity measurements were performed using a Linseiss Laser Flash Analyser (LFA) 1000. This apparatus measures the thermal diffusivity of the sample, from which the thermal conductivity is calculated. The LFA 1000 uses a YAG laser (25 J/pulse) and an InSb IR detector which requires liquid nitrogen cooling.

Liquid samples were added to a graphite sample holder with a quartz liquid specific cell with a sample z cross section of 1 mm. A laser pulse of 0.5 ms at 150 V was directed at the sample from below, and the rise in temperature on the top surface of the sample is measured and reported as a rise in voltage. The time taken for the top surface temperature to rise to the half maximum voltage indicates the thermal diffusivity (when the thickness, density and specific heat are also considered). Samples were measured at 25 °C in an uncontrolled atmosphere.

A three-layer fitting parameter (1 liquid sample layer, 2 quartz container layers) was employed to calculate the thermal conductivity for liquid samples. The quartz cell was spray coated with an aerosolised graphite (isopropyl alcohol mix) to ensure the emissivity of the surface was uniform.

Solid F108 pellets with ranging concentrations of freeze dried F108-graphene were prepared as a direct demonstration of the graphene contribution to the thermal conductivity of the matrix. The pellets were suspended in graphite sample holders with thicknesses between a range of 1.0 to 1.5 mm. The samples were also spray coated with graphite to ensure uniform emissivity.

The sample analysis was the same as described for the liquid cell, but a 3-layer fit was not selected.

B3.2 Results and discussion

B3.2.1 Photothermal curve

A demonstration of the base photothermal ability is shown in Figure 3.7, which shows the increase of bulk suspension temperature as a function of 500 mW 808 nm irradiation time with a spot size diameter of 1.7 mm, corresponding to a power density of 44.1 W cm-2. The base ability of graphene as a photothermal agent is critical in assessing and demonstrating the practicality of this material for this role.

199

Figure 3.7. Photothermal curve of 0.65 mg mL-1 F108-G (F108-graphene) irradiated with 808 nm 500 mW NIR.

The pi conjugation network of graphene allows strong absorbance of NIR radiation and transduction of the light to heat. The NIR is negligibly absorbed by both water and the surfactant (Figure 1.12) resulting in a system where any heating effect observed can be almost entirely attributed to the graphene transduction of NIR. As the graphene is irradiated and heated, the bulk suspension temperature slowly increases. The temperature at the localised site of heating (the particle) would be much greater than that observed in the bulk suspension.

The precision of the photothermal curve experiment was determined by performing three repeat irradiation experiments (0.65 mg mL-1 F108-G 500 mW) the maximum temperature after 15 minutes of exposure was 69 ± 0.8 °C (Figure 3.8).

Figure 3.8. The 0.65 mg mL-1 F108-G photothermal curve in triplicate to determine the precision and error of the experiment.

Temperature as a function of laser power was determined by irradiating a series of F108-G suspensions at a fixed concentration, at increasing power.

B3.2.2 Background heating

The background heating of water at 500 mW for 20 minutes showed a maximum rise of ≈ 2 °C which is practically negligible but critical to explore regardless (Figure 3.9). As the maximum rise observed was so small the values presented for experimental photothermal curves was not corrected for this value. Further it is important to not only show that the heating was due solely to the graphene photothermal agent, but further to show that 500 mW NIR irradiation could well be a practical non-harmful biological approach.

201

Figure 3.9. Photothermal curves for 0.017 mg mL-1 L64-G (red line), 0.1 % mg mL-1 neat L64 solution (green long dashes) and MilliQ filtered water (blue dots).

B3.2.3 System relationship to laser power

The relationship between laser power and photothermal heating via nanomaterial transducers is well established and has been demonstrated for a range of nanomaterial types.330-332 With greater laser power, there are a larger number of photons exciting the graphene electrons, leading to a greater magnitude of heating. As the laser power increases, assuming there are excess particles present, the observed heating effect will increase (Figure 3.10). Some practical limitations are present, for such demonstrations, such as ensuring the maximum laser power is not excessive, such that non-specific heating from surrounding tissue becomes significant.

Figure 3.10. (a.) F108-G photothermal curves as a function of laser power. (b.) shows the maximum temperature at 15 minutes as a function of laser power.

There was a linear increase of the equilibrium bulk temperature with laser power at the fixed graphene concentration of 0.65 mg mL-1. The gradient also changed with laser power showing a faster rise with laser power. The power of 500 mW was selected to provide a rapid rise in temperature whilst remaining low enough power to be biologically relevant. With excessive laser powers the background heating of water (which is minimal) could become a factor and as such laser powers are best to be minimised when appropriate.

B3.2.4 System relationship to concentration

Similarly, the relationship between concentration of the photothermal agent and the observed heating effect has been well demonstrated for a wide range of nanomaterials and follows a similar trend to that of laser power. With increasing concentration, there are more transduction sites allowing greater heating until the optical density has been saturated. At a fixed laser power and increasing concentration of nanomaterials will result in the increase of the heating effect again confined by the concentration range being within a range allowing excessive transmission or absorbance to be avoided.

In selecting an appropriate ratio between the concentration of particles and the laser power, the considerations are simply avoiding excessive transmission or absorption when the 203 particle concentration is too low or high respectively. When these parameters are met, the linear relationship between the concentration of photothermal agents and laser power can be readily demonstrated.

The NIR photothermal properties of the neat graphene suspensions was investigated by performing a concentration series of photothermal heating curves. It is important to note that the laser geometry, suspension volume and concentration will influence the temperature changes observed. Some slight fluctuations in the observed photothermal curve maximums are attributed to room temperature variations.

Figure 3.11. (a.) Photothermal curves as function of concentration showing in ascending order; 0, 0.075, 0.15, 0.225 and 0.30 mg mL-1 F108-graphene over 15 minutes of irradiation. (b.) The respective temperatures achieved after 15 minutes as a function of graphene concentration.

The linearity presented in Figure 3.11 b is specific to the range of data presented, at lower concentrations the linearity is lost as the increase in transmission leaves too few graphene transduction sites and the effect of the liquid medium will begin to influence the trend. At higher concentrations, the linearity is lost as the optical density becomes saturated.

B3.2.5 Photothermal efficiency

A key consideration is the efficiency at which the graphene microplates are able to convert the incident NIR light into heat. In order to determine the efficiency of the prepared graphene suspension, samples were photothermally heated under monitored conditions, followed by a cooling period when the laser was turned off (Figure 3.12). The photothermal efficiency could then be calculated via the method used by Roper et al. in 2007.322

(ℎ퐴(푇 −푇 )−푄 ) 휂 = 푚푎푥 푎푚푏 푂 Equation 3.10 퐼(1−10−퐴808)

Where η is the transduction efficiency, h is the heat transfer coefficient, A is the surface area of the container. Tmax is the maximum temperature at equilibrium, Tamb is the ambient temperature surrounding the experiment cell and Qo is the heat dissipated from light absorbed by the cell. I represents the incident laser power and A808 is the absorbance of the sample at the wavelength of the laser.

To determine the sample specific photothermal efficiency, the route of energy into the system must first be defined by Equation 3.11:

−퐴휆 푄퐼 = 퐼(1 − 10 )휂 Equation 3.11

Where QI represents the laser induced heat dissipated into the system, I is the laser power,

Aλ is the absorbance of the graphene at the wavelength of the laser (at the specific sample concentration). η is the transducing efficiency from the incident absorbance to thermal energy.

205

Next we consider the loss of heat from the system with the following Equation 3.12:

푄푒푥푡 = ℎ퐴(푇 − 푇푎푚푏) Equation 3.12

This describes the external heat flux or heat loss from the system (Qext) by accounting for the surface area of the cell A and the heat transfer constant h (determined via the sample system time constant introduced in Equation 3.15 The heat loss is also dependent on the difference between the sample temperature and the room temperature (T – Tamb), with a greater ∆T resulting in a greater heat loss.

We can then establish an energy balance equation as follows (Equation 3.13):

푑푇 ∑ 푚 퐶 = 푄 + 푄 − 푄 Equation 3.13 푖 푃푖 푑푡 퐼 푂 푒푥푡

Where on the left side of the equation the mass (m), the specific heat of the total sample, that is an estimated value for the combined glass walls and sample (CP) and the change of temperature over time will equal the energy input on the right side of the equation. On the right side of the equation QO is introduced which represents the energy lost through reflection or absorbance of the glass cuvette itself.

Roper then describes the introduction of a dimensionless driving force temperature θ, which is simply a ratio of the ∆T compared to that of the maximum as follows in Equation 3.14:

푇 −푇 휃 ≡ 푎푚푏 Equation 3.14 푇푎푚푏−푇푚푎푥

207

Along with a sample system time constant τS from Equation 3.15

Equation 3.15

Which provides the system equilibration time and is determined via the inverse of the Lnθ vs time plot gradient for the cooling data (or simply the gradient of the time versus -Lnθ as presented). Equation 3.15 also allows the determination of hA and therefore h as once the τs is known, a key step to determine η.

By then substituting Equations 3.14 and S3.15 into 3.13, Equation 3.16 can be determined:

푑휃 1 푄 +푄 = ( 퐼 푂 − 휃) Equation 3.16 푑푡 휏푆 ℎ퐴(푇푚푎푥−푇푎푚푏)

Which describes the combined events where the inverse of the system time constant, the laser induced heating and the heat loss will sum to the rate of the normalised ∆T (θ) as a function of time.

Then by shifting from the scenario of Equation 3.16, and turning off the incident laser, QI and QO = 0, simplifying Equation 3.16 to Equation 3.17:

푑휃 휃 = − Equation 3.17 푑푡 휏푆

When the system is at Tmax, this can be solved by including θ=1 at t=0 (100 % at 0 seconds) for Equation 3.18:

푡 푡 휃 = exp (− ) 퐿푛휃 = (− ) Equation 3.18 휏푆 휏푆

Therefore, by plotting the Ln of θ determined via Equation 3.14 against the time in seconds, the sample specific time constant can be determined, allowing the calculation of the heat loss components hA, the heat transfer constant with surface area via Equation 3.15.

If we then consider the system when at Tmax equilibrium, we know that external heat flux from Equation 3.11 must equal Equation 3.12 (Equation S19):

푄퐼 + 푄표 = ℎ퐴(푇푚푎푥 − 푇푎푚푏) Equation 3.19

Then finally by substituting Equation 3.11 into 3.19 we get Equation 3.10.

209

Figure 3.12. Photothermal efficiency analysis plots with the (a.) photothermal heating and cooling of the graphene suspension (red line) and the room temperature (blue series). (b.) The linear time data of the cooling period versus the natural logarithm of the driving force temperature (θ) from the cooling data marked in Figure 3.12 a.

-1 The ∑ 푚퐶푃 was calculated at 3.046 J K , I was measured at 0.393 W, Tamb = 21.44, Tmax =

27.14 and the Qo was measured with water present in the cuvette at 0.010 W. From the cooling data, the time constant for the heat transfer from the system can be determined from the gradient of the time vs. Lnθ plot, which was found to be -0.0043 (seconds). From the sum of the sample mass and specific heat (suspension 0.5 g and ≈ 4.184 J g-1 K-1, and cuvette walls 1.144 g and 0.837 J g-1 K-1),333 hA can be found to be 13.10 mW K-1 which corresponds to an efficiency of 79.83 %.

It is important to note that the sample was not sealed and not within a vacuum during the photothermal efficiency experiment, both of which will result in a slightly increased cooling rate which would exaggerate the efficiency value obtained. However, the sample temperature was measured by a thermocouple positioned approximately 2.3 mm away from the laser spot which will be considerably lower than the Tmax at the site of the irradiation, this will result in an underestimated Tmax and therefore an underestimated photothermal efficiency. It is a fair assumption that the underestimation of the Tmax will be of greater influence than the increased cooling rate from open air convection, as such we are confident that the efficiency quoted is likely a slight underestimation.

The average heat transfer constant can also be calculated via Equation 3.15 once the time constant has been determined. For the sample presented (with a surface area of 499.5 mm2) this was calculated as 7.96 W m-2 K-1.

The heating rate (°C/second) can be converted back to power (mW) to gain an alternative perspective of the photothermal efficiency of the graphene sheets. By taking the gradient from 15 – 45 seconds after the laser is turned on (allowing a brief equilibration period) the power in versus temperature rise data suggests an efficiency of ≈ 95 % when the absorbance and Qo is accounted for. The cooling rate data is far more reliable, as the heating sample is constantly being dampened by cooling with greater temperatures, as well as the cooling data being driven solely by the cooling effect.

As we can see from Figure 3.12, the approximate efficiency of transduction is ≈ 80 % for a concentration of 0.65 mg mL-1 F108-graphene suspension.

An initial estimation of energy absorbed, to gain the photothermal efficiency can be simply performed by taking the water temperature increase and dividing it by the product of the density and the specific heat.

B3.2.6 Stability of graphene

As several studies have explored the stability of nanoparticles in suspension sterically and ionically stabilised, a focus on the thermal influence to stability was taken. By assessing the stability from the photothermal maximums achievable as a function of time and heating cycles, not only is an insight into the stability of the suspensions made, but the photothermal stability of the material is simultaneously explored.

211

A key parameter to explore is to determine whether any damage is occurring during the irradiation and if so to what extent. As the photothermal agents will be at a higher temperature than the bulk observed temperature, it is possible that extreme temperature may be occurring locally and damaging the sheets. Further, one of the key advantages of graphene over gold nano rods for photothermal roles hypothesised is that the graphene sheets can achieve the same maximums numerous times while gold nano rods are significantly limited due to morphology reversion.

Further, one of the key advantages hypothesised for graphene in photothermal roles is that the graphene sheets should be thermally stable and able to achieve the same maximums repeatedly, a limitation of other photothermal materials such as gold nanorods (GNRs).187

The photothermal cycles are a key demonstration of the excellent stability of these suspensions, exposing the graphene sheets to vigorous temperature fluctuations and directly measuring the photothermally induced temperature maximums. This gives a result oriented view of the suspension stability leaving no room for doubt or ambiguity.

Figure 3.13 shows the photothermal cycling for 0.65 mg mL-1 F108-graphene immediately indicating that the maximum is not significantly decreasing with the number of cycles. The maximum bulk temperature for the 0.65 mg mL-1 sample was measured at 66.5 °C indicating that even at these substantially high temperatures the suspension was remaining stable, and not being damaged.

Figure 3.13. Suspension stability analysis via photothermal cycles showing the (a.) photothermal cycling of a 0.65 mg mL-1 F108 graphene suspension with series ascending from the lowest at 100 mW, through 200, 300, 400 and the highest series at 500 mW of 808 nm irradiation. (b.) The Tmax of each peak showing a minor decrease in maximums as a function of cycle number with the (c.) rate of decrease plotted against laser power. (d.) The -1 Tmax for an un-agitated and agitated 0.65 mg mL graphene suspension after 4 and 5 cycles respectively at 400 mW.

The photostability of the graphene sheets was confirmed by re-agitating a sample following several photothermal loops and recording a return to the previously observed Tmax, indicating the decrease was due to suspension stability. The return to an increased Tmax indicates that the sheets are not experiencing any thermally induced damage leading to the loss of attractive properties, as shown in Figure 3.13 d, It is important to note that while the suspension stability may lead to a minor decrease in Tmax over successive cycles, the stability of the suspension would not influence in situ thermal ablation. As the graphene sheets would not be present in a suspension during any application the loss of stability is not a key factor.

213

Further, stability of the suspension can be easily improved through surfactant selection (long chains improve suspension steric stability) as well as the concentration of surfactant along with a number of additional strategies.334

A slight decrease in maximums can be observed for all samples which is attributed to the stability of the suspensions, not to any loss of transduction properties of the nanosheets through oxidation. This was confirmed by re-agitating a sample after several photothermal cycles and observing the recovery of the Tmax. One possible explanation for these events is that the heating of the particles is taking the temperature of the adsorbed stabilising polymer surfactants to near or above the respective cloud points (> 100 °C for F108, 58 °C for L64).216 As such, the steric stabilisation decreases as it becomes more energetically favourable for the surfactant to interact with itself than with the suspension medium.142, 335 Additionally, the high temperatures experienced will act to increase the rate of adsorption/desorption of surfactant molecules through the increase of kT,336-338 which may result in a decreased concentration of surfactant at the graphene surface, reducing the repulsive mechanism. This will lead to minor flocculation events, removing suspended particles from the path of the laser as well as reducing the availability of atoms to absorb the source NIR, ultimately resulting in a decreased observed Tmax.

Figure. 3.14. The extended view of the suspension stability data showing the significant effect brought about by the evaporation and subsequent increase in concentration of the F108- graphene suspensions.

By extending the cycles, the significance of the evaporation can be observed. As the total volume of the suspension decreases, the concentration of graphene is effectively increasing while simultaneously the bulk temperature increasing dramatically with a decrease of the volume.

To determine any oxidation events during irradiation, absorbance and Raman spectra were collected after each experiment. The absorbance scan showed no change with the characteristic peak remaining at 268 nm, the only change being an increase to the concentration as a result of minor evaporation during cycling. The Raman similarly showed no change to the spectra, confirming that the samples are not being oxidised during the photothermal process. The Raman peak showed no identifiable features of oxidation such as 324 a change to the ID:IG towards that of GO or a red shift the G peak position. A Raman spectra measured for graphene oxide has been included to allow easy visual comparison, confirming the absence of oxidation events. This indicates that the graphene sheets are capable of being used for multiple activations as the photothermal ability is not lost.

215

Figure 3.15. (a.) UV-visible absorbance spectra and (b.) Raman spectra of the D and G peaks for graphene oxide (bottom series), F108-graphene pre (middle series) and post (top series) 10 x 500 mW photothermal cycles (series are offset).

The lack of oxidation after the extensive irradiation at 500 mW also provides an approximate indication that the localized temperature is likely not exceeding ≈ 250 °C, the point at which oxidation would be thermally induced.324

These results are in strong agreement with previously performed work whereby the photostability of gold nanorods was compared to that of rGO.339 The rGO absorbance spectrum showed negligible change following heating while the GNRs showed a significant change to the absorbance spectrum attributed to the melting of particles and loss of the rod morphology key to the absorbance properties.187, 322 We can understand from this data that the graphene based material can achieve substantial temperatures while remaining stable after multiple exposures.

B3.2.6.1 Stability of suspension

The stability of the surfactant stabilised suspension depends on a range of factors as discussed in Chapter 2. Key relevant parameters affecting the stability can be primarily split into two groups, the surfactant identity (size) and the temperature experienced. As demonstrated by Rahme et al. in 2007 the stability of surfactant stabilised nanomaterial colloidal suspensions will increase with surfactant concentration, PEO and PPO block lengths, or the overall length of the polymer340 the size of the tri-block copolymer. This allows suspensions exfoliated with the large F108 surfactant to remain stable for many months with minimal flocculation, aggregation and settling.

The steric stability imparted via these surfactants is dictated by the extent to which the polyethylene oxide (PEO) segments are extended out into solution away from the particle surface. Control over the position of the PEO chains can be achieved through manipulations to the ionic strength341-343 of the solution as discussed further in Chapter 5. However, similar disruptions/influences of the suspension stability can be achieved through the change of temperature particularly around the cloud point of the polymer surfactant as non-ionic polymer surfactants exhibit reverse solubility versus temperature. This behaviour is attributed to the dehydration of the hydrophilic polyethylene oxide segments.341, 344

These are particularly relevant factors for the use of this material as a photothermal agent, however these factors will strongly influence the testing of the materials however in situ, whether that be in a drug release carrier matrix or in tissue, the stability of the suspension is not a factor.

B3.2.6.2 Extended irradiation time

The L64 Pluronic surfactant has a low cloud point temperature of approximately 62 C at which point the surfactant undergoes a phase change demonstrated in Figure 3.16. This phase change would likely result in the aggregation and sedimentation of the nanosheets. In the event that the suspension appreciably aggregates then the sample material is no longer going to be present in the original large surface area condition and also likely no longer in the path of the laser.

217

Figure 3.16. Induced aggregation L64-stabilised suspension over extended exposure period.

Figure 3.16 shows the extended photothermal run for L64-graphene sample. This highlights the maximum temperature achieved and the subsequent decrease in temperature likely due to aggregation of nanosheets in suspension.

A photothermal experiment was performed over an extended period of time in order to determine if this event was occurring. The L64-graphene sample was selected as it was sufficiently concentrated and photothermally active to achieve temperatures greater than the cloud point of L64. The data was showing clear signs of a slow decrease of temperature from approximately 55 minutes (temperature = 60.57 °C) onwards.

B3.2.7 Thermal property characterisation

In order to determine the influence, the presence of the graphene microplates were having on the thermal transfer properties of the suspensions, the thermal conductivity of graphene suspensions and graphene loaded F108 surfactants was explored via laser flash analysis.

Figure 3.17. Raw thermal diffusivity signal for a graphite reference standard showing the increase in voltage (temperature) shortly following the laser pulse.

A graphite reference standard, with a thermal diffusivity of 1.1 cm2 s-1 was measured prior to sample experiments for system calibration. The shape of the graphite standard heating pulse curve shows the increase of temperature on the surface of the sample following the laser pulse (Figure 3.17). The rise in voltage is normalised, such that the minimum value becomes zero, and the half maximum is determined. The time at this point is interpolated allowing the sample diffusivity to be determined.

푙 훼 = 0.1388 ∙ Equation 3.20345 푡

Where α is the thermal diffusivity, l is the sample length (thickness, cm2) and t is the time at half maximum (seconds).

From the thermal diffusivity, the thermal conductivity (k) can be determined when the specific heat and the density are known.

219

푘 = 훼 ∙ 퐶푃 ∙ 휌 Equation 3.21

Where CP is the specific heat (J g-1 K-1) and ρ is the density (g cm-3). The specific heat of all samples was determined experimentally via differential calorimetry scanning (DSC) measurements and the density was measured via helium pycnometry using a micrometrics AccuPyc 1330 gas pycnometer.

Figure 3.18. (a.) The determined thermal conductivity of the aqueous samples as a function of F108-G concentration. (b.) The determined thermal conductivity of the solid F108 pellet samples as a function of F108-G concentration.

Clear enhancements to the bulk thermal conductivity can be observed for both the aqueous and pellet samples (Figure 3.18). The addition of the highly conductive graphene microplates is providing improved pathways for heat transfer within each matrix.

Within the pellets this is likely via an overlapping sheet mechanism, whereby the solid sheets are in direct contact and can propagate the relevant phonon modes without dramatic loss. The liquid sample improvement is attributed to percolation events as discussed in Section A3.5.

This suggests that with sufficient concentrations the thermal dissipation within a graphene suspension will be improved, likely to a minor extent. However, as no oxidation events were identified following the extended irradiation and photothermal cycling, it is possible that the small improvement to heat dissipation is aiding in minimising the thermal oxidation.

Chapter 3 Conclusions

These base experiments have demonstrated the ability of graphene in the role of a photothermal agent including showing the heating action is due solely to the graphene present, the relationships between concentration and heating as well as laser power and heating. Further the approximate photothermal efficiency of the transduction was also explored showing the efficiency was approximately 80 %. With these factors considered and explored, the base ability of graphene for the role of a photothermal agent has been well established.

221

4.0 Chapter 4 Graphene for the photothermal ablation of cancerous cells

4.1 Introduction

As the photothermal ability of graphene has been investigated and results demonstrate that this is a material which can achieve substantial in situ NIR induced heating, the next consideration is how can these photoresponsive materials be employed and how appropriate is the material for biological applications. As such, the ability of liquid exfoliated pristine graphene prepared using biocompatible surfactants to stimulate cell death on irradiation with NIR light in situ is presented within this chapter.

The near-infrared (NIR) region is a particularly useful energy range of light for biomedical applications as it is not strongly absorbed by water and biological materials as discussed in Chapter 1.65, 346 This allows greater penetration into biological tissue, allowing an effective external activation route for materials capable of strongly absorbing the appropriate wavelengths and transducing it to heat. A range of photothermal agents including graphene oxide (GO), reduced graphene oxide (rGO), carbon nanotubes (CNTs), molybdenum disulfide (MoS2), and gold nanoparticles/gold nanorods (GNPs/GNRs) have been explored, with varying degrees of success, in roles such as thermal ablation of tissue cells, with a focus on materials with a strong NIR absorbance profile.66, 165, 174, 192, 197, 339 Graphene has a strong absorption coefficient in the NIR and will likely be a more efficient transducer than many of the materials previously explored.183

4.1.1 Biological interactions of graphene

Any material posited as a biomedical agent must be an efficient photothermal transducer, while not being acutely toxic to allow in situ functionality. The biocompatibility of graphene materials depends on their intrinsic physical–chemical properties which in turn are dependent

223 on the raw materials and production methods used.347 As such, liquid exfoliated graphene, a material consisting almost entirely of carbon, is expected to present predominantly non-toxic properties,156 particularly when prepared with larger triblock copolymer surfactants.348-350

Initial biocompatibility studies are commonly performed via in vitro experiments (cell culturing), whereby assays are employed to measure the metabolic activity of the cells present following cell treatment and incubation. As such, in vitro tissue culture experiments can give an indication of the cell viability under the tested conditions. These assays are an excellent route to quickly determining the dose-dependent toxicity of nanoparticles, which amongst other key factors, is critical to understand for biological-nanoparticle interactions.

Depending on the route of production, the chemical nature and therefore the properties of the sheets obtained will be altered significantly. Specifically, the Hummers method and subsequent alterations to this original production route,132, 351 produces oxidized sheets with significantly different properties to sheets produced via exfoliation methods, such as absorbance characteristics and dispersion mechanisms. It is important to note that just as the properties mentioned will differ significantly between production methods, the interaction with biological materials will also change and therefore it should not be assumed that trends will apply across even only subtly different nanomaterials. However, we can gain some information in this little understood area, through looking at previous studies of the effects of graphitic nanomaterials on biological systems.

4.1.1.1 Biocompatibility review of similar materials

Graphene and graphene derivatives have predominantly been shown to be of low cytotoxicity and have been explored extensively in recent years in biomedical roles.352-354 In 2010, Hu et al. showed that GO and rGO exhibited minimal cytotoxicity to the mammalian cells which were taking up the nanosheets.355 The biocompatibility of graphene paper (prepared from GO suspensions) was also explored in 2008 by Chen et al., using L-929 (mouse fibroblast) cells with results indicating that (at least for a flat graphene surface) that graphene is a biocompatible material for this cell line.356 L-929 cells adhered and proliferated on these surfaces and a sub-confluent layer was reached after 48 hours of culture time.356

A study by Agarwal et al., in 2010, showed that rGO was biocompatible with a range of mammalian cells including neuroendocrine PC12, oligodendroglia cells, and osteoblasts.357 CNTs were also shown to inhibit the proliferation, viability and the neuritegenesis of PC12 in this study,357 indicating that the distinct nano-topographic features are likely dictating whether these chemically similar materials present as cytotoxic.

In 2012, Gollavelli et al. explored the cytotoxicity of GO to HeLa human cervical cancer cells and showed a graphene toxicity with a half-maximal inhibitory concentration (IC50) value of approximately 100 μg mL-1. This study also explored the lactate dehydrogenase (LDH) release in the presence of GO, showing a significant increase with GO concentrations out to 200 μg mL-1 indicating a decrease in cell membrane integrity. The reactive oxygen species (ROS) production was monitored via flow cytometry within this study, indicating significant presence of ROS out to 100 μg mL-1, with a marked increase observable at 200 -1 354 - μg mL . In 2014, Zhou et al. reported IC50 values of GO at values higher than 80 μg mL 1. Which was the highest concentration tested within this study against 3 cell lines, MDA- MB-231 human breast cancer cells, PC3 human prostate cancer cells and B16F10 mouse melanoma cells.

From these studies, we can see that the IC50 value for graphene derivative materials appears to be centered at ≈ 100 μg mL-1, a value classed as being of low cytotoxicity.358-359

As surfactant assisted liquid exfoliated (SALE) graphene has significantly different surface properties to those of other graphene based materials, it is necessary to define any cytotoxic effect. In addition to providing some generalised insight as to the biocompatibility of

225 graphene, this is also a critical factor in establishing a suitable concentration range for the role as a thermal ablation agent.

4.1.1.2 Nanoparticle stealth strategies

A critical consideration is the uptake of particles into the reticuloendothelial system (RES) and other physiological defence mechanisms which reduce the number of particles that reach the target site. It is a well-established strategy in the development of nanomaterials for biomedical applications to coat the nanoparticles in ‘inert’ polyethylene oxide (PEO), often referred to as PEGylation (PEG=PEO).157-159, 198 One such example of successful stealth coating via PEGylation was reported in 2006 by Niidome et al., in which they PEGylated GNRs prepared with cetyltrimethylammonium bromide (CTAB) to improve the viability. The GNRs coated with the PEG presented a viability of over 90 % (at the highest concentration tested 0.5 mM) compared those without which showed a 50 % viability at 0.05 mM.159

The surfactants predominantly employed within this study are triblock copolymers (Pluronics®) which consist of large PEO and polypropylene oxide (PPO) segments.216 Therefore it is expected that graphene with an adsorbed triblock copolymer coating will present greater biocompatibility. The triblock copolymers are adsorbed on to the hydrophobic basal plane via the PPO segments, leaving the hydrophilic PEO chains extending into aqueous solution.109, 334 It was expected that with more PEG groups present as is the case with the larger polymer chains, the cytotoxicity would be further dampened and as such the primary experiments were performed with the non-ionic surfactants L64 and F127.

4.1.2 Cytotoxicity studies

To explore the suitability of SALE graphene as a photothermal ablation agent, experiments were performed against mammalian NG108-15 cells in an attempt to closely represent the final application conditions. NG108-15 cells are a somatic cell hybrid of mouse neuroblastomas crossed with rat glioma. The neuronal cell line of NG108-15 was selected both to allow a well characterised starting point for future culturing and thermal ablation work and due to this cell lines experimental reproducibility.

Figure 4.1. Light microscopy images of NG108-15 cells at (a.) 20 X and (b.) 5 X magnification.

NG108-15 cells are a cross of mouse neuroblastomas and rat glioma hybrid cells and is an immortal mammalian cell line (Figure 4.1). As such, these cells should give a highly relevant platform for the assessment of NIR induced photothermal ablation of cancerous tissue with graphene in vitro.

4.1.2.1 Mechanisms of nanoparticle induced cell death

Cytotoxicity induced by nanoparticles can be caused via a wide range on mechanisms such as membrane deformation,235, 354, 360-361 reactive oxidative stress,362-364 UV-activation of electron hole pairs leading to bond splitting and the formation of radicals,365-367 protein denaturation,368-369 DNA damage and many more.370 Two of the more common and more widely applicable mechanisms are those of membrane deformation as well as generation of ROS, resulting in oxidative stress.

227

When oxidative species are at a normal concentration, the coupling conditions (e.g. oxidative phosphorylation371) within the mitochondrion are stable with ROS being generated at a low frequency.372 Under these conditions the ROS present are easily neutralised by antioxidant defences, including a range of antioxidant enzymes and glutathione. These natural defence mechanisms are overwhelmed with excessive ROS generation, leading to oxidative stress, commonly indicated by an accumulation of oxidised glutathione.360, 372 Excessive ROS concentrations have been shown to lead to a range of detrimental events such as DNA damage, chromatic condensation and lipid peroxidation, which eventually leads to apoptosis.354, 360, 372 Apoptosis is the mechanism by which cells can control the time of their own death through programmed cell death. Apoptosis can be triggered by a range of events such as irreparable DNA damage, or can be triggered in response to a biological circumstance such as a virus infection.371

Direct physical damage or membrane deformation is another toxicity mechanism attributed to nanoparticle cytotoxicity. For graphene microplates, this is of particular importance due to the two-dimensional (2D) morphology of the sheets providing a blade like edge, which can penetrate cell membranes. The rupturing of cell membranes is therefore dictated by the extent interaction of the nanoparticle to the cell where some interesting trends have emerged in the literature. One such influence is that with an increased hydrophilicity, materials tend to penetrate the cell in greater quantities, whilst presenting a decreased toxicity than for that of hydrophobic materials.373-374 Additionally, the lower the density of functional groups on the graphene derivative material, a lower extent of interaction with cells is observed, resulting in an increased viability. Again, indicating a hydrophobic basal plane may be inhibiting the entrance of the sheets to into the cells.

4.1.3 Thermal ablation

Thermal ablation is the local application of extreme temperatures to induce irreversible cell injury through tumour apoptosis and coagulative necrosis.375-376 In addition to NIR stimulation a range of approaches to achieve the in situ localized heating/cooling have been explored such as radiofrequency ablation, microwave ablation and cryo-ablation.375-379 As graphene is considered a reasonably biocompatible material and is strongly absorbing in the NIR it is an suitable candidate for NIR activated photothermal ablative therapies.

Thermal ablation is applied for the treatment of small un-resectable tumours for patients who are poor surgical candidates. Compared to surgery, thermal ablation has lower morbidity rates, increased preservation of surrounding tissues along with reduced hospitalization times and costs.380-381 However, thermal ablation has some limitations such as incomplete ablation and disease recurrences.382 It is important to note that at this stage several clinical trials have been performed for a range of thermal ablation approaches, such as radiofrequency-,383-384 microwave-385-386 and cryo-ablations.387-388 These clinical trials have had varying extents of success and can be seen as an encouragement to explore NIR activated systems.380-381

A sizeable number of NIR thermal ablation in vivo and in vitro studies have been performed for a range of nanoparticles, which have also provided an insight into the actions these particles have in physiological conditions.165, 198, 339, 352

In a study by Tian et al., in 2011 PEO coated GO (GO-PEG) was shown to be an efficient photothermal agent, demonstrating successful photothermal ablation of tumours in animal experiments.174 In this scenario the PEO is present primarily to help stabilise the GO, however it would also provide additional stealth advantages.174 This GO-PEG system was shown to be an adequate photodynamic agent when coupled with the photosensitizer molecule Chlorin e6 (Ce6), demonstrating a significant improvement to free Ce6.174 Ce6 is a photosensitive molecule with antitumour activity when irradiated.

In 2012, Yang Wan et al. demonstrated full survival and no observable damage within mice using ultra small rGO with a non-covalent PEO coating.339 The dosage applied was at ≈ 2.7 mg mL-1 (20 mg per kg of mouse, assuming mouse weight 250 g and blood volume 1.5 mL) 354, which is far higher than the reported IC50 values for rGO as determined via in vitro studies. 229

356-357 This indicates that these graphene derivative materials will likely present lower toxicities in vivo vs. in vitro. Yang et al. demonstrated successful reduction of tumour size (reported 100% removal) in mice via NIR activation of rGO and nano-rGO using a power density of 0.15 W cm-2 for 5 mins.339 They reported temperatures of 48 °C during activation and full survival rate. Graphene derivative materials have been reported in several studies to be generally hemocompatible, as well as to present a faster elimination with smaller particles.347

One example of cancer therapy via photothermal therapy has been demonstrated by Chou, et al., who showed a novel, specific delivery system, which could provide stable storage and delivery of gold nanorods. By establishing this functionalized chitosan-conjugated Pluronic- based nanocarrier system, GNR delivery showed improved tumour targeting and decreased liver uptake.165 This has considerable advantages as the in vitro cellular uptake and subsequently the photothermal effect was enhanced. This group went on to show that intravenous injection of this gold nanorods carrier system followed by in vivo NIR laser irradiation resulted in efficient thermo-ablation in vivo.165 The tumours showed complete resorption with no damage to surrounding tissue. Importantly, no observable cytotoxic effect was observed during the in vivo experiments with a loading concentration of 50 μg mL-1.165

Other nanomaterials with sufficient absorbance within the NIR have also been explored including MoS2. In 2013, Chou et al. employed chemically exfoliated MoS2 for the NIR thermal ablation of HeLa cells at a dosage of 20 and 40 ppm.389 Temperatures within the sample well were reported to reach 57 and 64 °C respectively after 20 minutes of irradiation resulting in zero viability.

4.2 Materials and methods

Graphene suspensions were prepared and characterised as described previously in Chapter 3.

4.2.1 Tissue culturing

The NG108-15 neuroblastoma cross glioma cells were obtained from the European Collection of Cell Cultures. Culturing was performed in a humidified atmosphere with 5%

CO2 at 37°C and in Dulbecco’s Modified Eagle Medium (DMEM). The DMEM was prepared with an additional 10% (v/v) fetal bovine serum (FBS), 1% (v/v) glutamine and 1% penicillin/streptomycin. Cells used for experiments were always at 70 – 80% confluence and no older than passage 20. All experiments were performed in independent triplicates and Triton X was used as the control at a concentration of 1 mg mL-1. Matching concentrations of graphene were included to the control samples where appropriate to ensure the background absorbance was equal.

The analysis assay employed for all tissue culturing experiments was the cell viability colorimetric assay kit, CellTiter 96 Aqueous One Solution Cell Proliferation Assay. This solution contains a tetrazolium compound (3-[4,5-dimethylthiazol-2-yl]-5-[3- carboxymethoxyphenyl]-2-[4-sulfophenyl]-2H-tetrazolium, inner salt) (MTS) and an electron coupling reagent (phenazine ethosulfate). The tetrazolium compound is reduced into a coloured formazan product by metabolically active cells, the absorbance at 490 nm of the resulting solution is measured after the appropriate incubation period indicating the extent of metabolically active cells.

For all cytotoxicity testing 48 hours of total incubation time was selected and the cells were exposed to the respective test material (surfactant or nanoparticle suspension) for 24 hours. The wells were seeded at 20,000 cells cm-2 in DMEM and incubated for 24 hours. At this

231 point 100 µL of the cell culture media was extracted and 100 µL of prepared test solution/suspension diluted in MilliQ and cell culture medium accordingly (i.e. 50 µL media, and appropriate MilliQ to test solution dilution). The cells were then cultured for a further 24 hours prior to MTS analysis.

At the end of the relevant exposure and incubation periods the cell culture media was extracted and the cells were washed with a warm (37 °C) phosphate buffer solution (PBS). After which 100 µL of the MTS solution (prepared at a 1:5 ratio with PBS) was added to the wells and incubated for 4 hours prior to measuring the absorbance of each well at 490 nm on a plate reader. The IC50 and the statistical analysis of the dose response curves were determined using GraphPad PRISM version 7.02.

4.2.2 Irradiation experiment

The NG108-15 cells were incubated in the presence of L64-graphene at a concentration of 17 µg mL-1 for 4 hours prior to irradiation. This concentration was selected as the cell viability after the analysis period would be sufficient to establish a cell death versus cell viability signal. Samples were categorized into the following groups, cells irradiated with and without graphene, as well as non-irradiated cells with and without graphene. Triton X was used as the positive control to show full cell death.

Irradiation experiments were performed inside an incubator to maintain the ambient conditions at a constant 37 °C and 5 % CO2. A 400 mW 808 CW laser was directed at an individual well via an optic fibre fed into the incubator through a small port in the rear wall of the incubator. The laser tip was positioned above a 96 well plate at a distance of 18 mm from the well-base, such that the spot size was 2.2 mm in radius resulting in a power density of 21.06 W cm-2. A range of irradiation times were explored and 60 minutes was selected to allow a clear demonstration of complete cell death.

4.3 Results and discussion

The cytotoxicity of the both the graphene and the surfactants used for exfoliation and stabilization of the suspensions, is a critical parameter for the use of graphene as photothermal ablation agent. As such, the cytotoxicity of the surfactants and surfactant exfoliated graphene used within this study were assessed against the NG108-15 cells (Figures 4.3 and 4.4).

The preliminary experimental parameters needed to be determined such as incubation times and seeding densities (See Figure 4.2). Once these initial parameters were established, the cytotoxicity for a range of relevant surfactants as well as the three key functional nanomaterials needed to be experimentally determined, at which point the irradiation experiments could be performed.

4.3.1 Seeding density and incubation time

The population densities and incubation times were then explored to determine the optimum parameters for the cytotoxicity tests and irradiation experiments. As can be seen in Figure 4.2, the initial seeding density resulted in different absorbance signals in after the MTS assay as with more cells alive, the greater the conversion of the tetrazolium to the formazan.

233

Figure 4.2. Absorbance at 490 nm for (a.) NG108-15 seeding densities at 48 hours incubation time and (b.) for 24, 48, 72 and 96 hours incubation time.

The seeding density of 20,000 cells cm-2 was selected as the signal was sufficiently high for all incubation times, but specifically for the 48 and 72-hour incubation times the cells could be seen to be in a healthy growth phase. The drop off in cell viability for the higher populations is attributed to the depletion of nutrients in the culture media and limited space for cell growth.

4.3.2 Surfactant cytotoxicity

With the seeding densities and incubation times determined the focus of the study turned to the cytotoxicity of the neat surfactants on the NG108-15 cells. It was important to determine whether the cytotoxicity of the surfactants would be acutely toxic against the NG108-15 cells to halt any further experiments (Figure 4.3). Additionally, this information provides an understanding of the extent of toxicity attributed to each individual component of the surfactant-graphene suspensions.

A concentration series for a representative group of surfactants was prepared and added to the relevant wells and the samples were incubated for 24 hours.

Figure 4.3. Dose response curves for the surfactants (a.) L64, (b.) F127 and (c.) CTAB against NG108-15 cells. These show the normalized cytotoxic response of the NG108-15 cells as a function of surfactant concentration Panel d. shows the IC50 of each surfactant.

It is important to note the dramatic concentration differences in the cytotoxicity curves between the Pluronic surfactants F127 and L64, to the cationic CTAB. The positively charged CTAB showed a toxicity level ≈ 1000 fold greater than the Pluronics, following the well- established trend in literature.390-391 This highlights the biocompatibility of the non-ionic surfactants and also has dictated the decision to continue employing them for suspension stabilization and preparation for this work. 235

Each of the dose dependency plots show a decreasing cell viability with increasing concentration of surfactant. The IC50 can then be determined from the concentration at half the maximum cytotoxic response to give a clear comparison between cytotoxicity of the different surfactants. The results follow previously observed IC50 values for non-ionic surfactants on mammalian cells,348, 392-393 as well as following described trends with larger surfactants (such as F108 and F127) showing a mildly decreased cytotoxicity when compared to that of the smaller non-ionic surfactants (L64).348

It is well understood that ionic surfactants will present far greater toxicity to cells as clearly demonstrated by Arechabala et al. in 1999.394 This group compared the ionic sodium dodecyl sulfate (SDS) and cationic (benzethonium chloride) surfactants against the non-ionic (Tween 80) via an MTT viability assay (dimethylthiazol diphenyl tetrazolium bromide) assay using human fibroblast cells.394 The MTT assays showed a marked increase in the cytotoxicity of -1 the cationic surfactant whereby the IC50 was determined at 8 μg mL . The anionic surfactant -1 -1 presented an IC50 at 62.0 μg mL and the Tween 80 that of 850 μg mL .

Table 4.1. IC50 comparison of ionic and non-ionic surfactants experimentally and in literature.

Surfactant Experimental Results Arechabala IC50

-1 -1 Identity IC50 (mg mL ) (mg mL )

Cationic 0.0001 (cationic) 0.008 (cationic)

0.062 (anionic) Non-ionic 0.015 (L64) 0.210 (Tween 80)

0.1 (F127)

We can see via the comparison in Table 4.1 that the same trends are observed between the two data sets showing that the cytotoxicity of the cationic surfactant is dramatically larger than that of the other classes. Further, the non-ionic surfactants demonstrate excellent biocompatibility. The size of the triblock copolymer is an additional factor as shown by Kier et al. in 1995. This group showed that with increasing size the cytotoxicity was significantly decreased.348 This is in addition to the suspension stability and high graphene production yield advantages, along with creating a site for bio-conjugation allowing specific targeting.

It is important to consider that the Arechabala group carried out their assays on human fibroblasts, a strongly attaching cell, particularly when compared to NG108-15 cells which can be very easily detached from the culturing surface. This is important not only to account for some level of difference between the two datasets, but critically when considering the ‘balling up’ mechanism likely induced by the surfactant molecules on the attached cells. The balling up mechanism of surfactants refers to the edges of the dirt/cell being wetted, lowering the adhesion energy to the substrate, resulting in the edges lifting which leads to a rolling or balling up of the dirt/cell.

In summary, from this information we can understand that the nonionic, larger tri-block copolymers with high PEO percentages will provide a suitable starting point in the attempt of incorporating graphene into the NG108-15 culture.

4.3.3 Graphene cytotoxicity

In addition to the neat surfactant analyses, the cytotoxicity of the graphene microplates was assessed to determine an appropriate working concentration for the thermal ablation experiments (Figure 4.4). It has been well demonstrated by a number of studies that the presence of a PEO surface coating onto a nanomaterial can significantly reduce the observed cytotoxicity on surrounding cells.395 As such the non-ionic surfactants were selected for the preparation of graphene used in the thermal ablation and cytotoxicity experiments. Further, it was expected that with more PEG groups present, as is the case with the large triblock copolymers (F108 and F127), the cytotoxicity would be further dampened and as such the thermal ablation experiments were performed with the non-ionic surfactants L64 and F127. A higher rate of PPO to PEO has been shown to increase the cytotoxicity and should be considered as a contributor to any observed toxicity.395 A study published in 2012 by

237

Redhead et al., demonstrated a relationship between both the ratio of PEO to PPO as well as the affinity of the copolymers for the biological membranes. This group carried out these studies against intestinal epithelial Caco-2 cells and the human microvascular endothelial cells HMEC-1 assessing both the LDH production and the MTS conversion.395 Despite the presence of PPO, the biocompatibility is expected to be increased substantially over that of an un-coated graphene sample.

The relationship between the increased concentration of graphene based materials and mammalian cell death has been shown previously to be dose dependent.353, 396-398 The concentration range presented in Figure 4.4 shows no cell death at 0.65 µg mL-1 out to full -1 -1 cell death at 34 µg mL and beyond. IC50 was determined to be 12.1 µg mL , a value which is classified as moderately cytotoxic.358-359

Figure 4.4. Dose response curve for L64-graphene. This shows the normalized cytotoxic response of the NG108-15 cells as a function of L64-graphene.

The cytotoxicity curve for the selected graphene suspension was then determined and an appropriate concentration selected for the irradiation experiments (Figure 4.4). To gain some perspective of what these values mean we can compare against mammalian in vitro studies with similar nanomaterials.

The cytotoxicity of the surfactant coated graphene sample on the NG108-15 cells after 24 hours of incubation, was slightly lower than for a number of similar experimental IC50 values reported in several studies (SALE-graphene ≈ 10 μg mL-1; graphene derivatives ≈ 100 μg mL-1).353, 396-398 This increased cytotoxicity is likely due to the sonication process breaking down the Pluronics into toxic components which sees the non-ionic surfactants presenting -1 396 IC50 values of approximately 1 µg mL , a dramatic decrease down from values of 1 mg mL-1.348, 392-393 It is very likely that the degraded surfactant is the dominant contributor to the observed cytotoxicity.396 We see simple strategies to minimize the cytotoxicity caused by degraded surfactant through dialysis and perhaps surfactant exchange. However, as the concentration range determined was sufficient to carry out the experiments required to demonstrate graphene as a photothermal agent, these strategies were not explored.

4.3.3.1 Role of morphology

An important consideration around the cytotoxicity of graphene sheets is the mechanism of cell entry or membrane rupture. Li et al. showed in 2013, with combined experimental and computational studies, that a few layer graphene microsheet class of graphene were able to enter a range of mammalian cells including human lung epithelial cells via spontaneous membrane penetration.361 The graphene type employed for this study was quoted as having a carbon to oxygen ratio within the range of 1:10 – 1:25, indicating that this is a considerably oxygenated graphene class. This group also suggest that the localized penetration is due to the sharp corners or at protrusions along the edges of the graphene sheets. This is then followed by a recognition of the hydrophobic basal plane as a damage associated molecular pattern and is treated as a hydrophobic microbe or particle leading to macrophage engulfing the foreign body likely via phagocytosis.

239

Additionally, a strong relationship between the density of edges and observed toxicity of graphene sheets has been demonstrated for bacterial cells, further supporting this dominant mechanism.235 The mild hydrophilic nature of the PEO segments is likely discouraging agglomeration at the cell membrane. As such, the blade-like action of the graphene sheets is inhibited and the membranes are not being punctured as readily. On the other hand, the negative charge of the graphene sheets would likely be encouraging the opsonisation of particles by the NG108-15 cells. Future studies exploring the uptake of particles into cells as a function of the size of PEO chains attached would be informative.

An additional study published by Pham et al. in 2015, reported a similar experimental and computational joint study, whereby a surfactant exfoliated graphene was used to prepare films with different extents of roughness. This group demonstrated a strong relationship between the density of edges and the toxicity of the sheets towards the bacterial cells studied, staphylococcus aureus and pseudomonas aeruginosa.235 The CTAB-graphene films were prepared via filtration and extensive washing was performed to remove the toxic CTAB allowing growth on the smooth samples whilst the rough samples inhibited the pathogenic bacterial growth.

This study also attributed the mechanism of cell entry to the formation of pores within the lipid bi-layer, an event which then causes an osmotic imbalance resulting in cell death.235 This was explored both experimentally and via computational studies, the latter indicating that by enforcing an increase in the effective lipophilicity of the graphene sheets, the characteristics of the interaction with the bilayer are altered. These results indicated that the full insertion of the graphene sheets into the membrane leads to the formation of pores, where the more lipophilic the graphene, the greater the extent of interaction with the bilayer tails and ultimately, the greater the extent of pore formation. This indicates that the sheets are not acting predominantly via a blade-like mechanism, as has been suggested previously.361, 399

The cytotoxicity of graphene and single walled CNTs was explored using neural phaeochromocytoma-derived PC12 cells and shown to induce cytotoxic effects which are both concentration and size dependent.353 An oxidative stress mechanism for the cytotoxicity was suggested as a ROS were generated by the cells after exposure to the graphene sheets.353 Additionally, Yi et al. reported that insertion of graphene through a lipid bilayer was dictated by the size of the graphene sheets. In this study it was shown via computational studies that the larger microsheets would adopt a near-perpendicular configuration while the smaller nanosheets would position parallel to the bilayer.138

When considering the results from each of these studies, the mechanical rupturing of both mammalian and bacterial cells can be seen as an important mechanism in the cytotoxic behavior of graphene microplates. Which in turn, indicates that it is not a chemical mechanism that is dictating any cytotoxic behavior.

4.3.3.2 Role of surface chemistry

A trend has formed in literature indicating that more hydrophilic graphitic nanomaterials lead to a minimized toxicity.398 The presence of a surfactant coating is all the more critical enabling the reduction of particle toxicity. The extent of oxidation has been demonstrated to influence GO cell cytotoxicity, whereby the increasing oxidation quantity was leading to greater levels of phagocytosis. As the sheets were being engulfed more readily, the rupturing of the cell membrane was less common a resulting in a decreased toxicity.373-374

Both the presence of defects in graphene sheets and the sheet thickness can influence the extent of oxidation events occurring with thinner, high defect materials presenting oxidation at lower temperatures.324 As such, the production method employed herein is particularly appropriate in minimising the likelihood of such events occurring, both by minimising any initial defects and producing a sheet thickness distribution. These attributes combined lead to the protection of the attractive properties of graphene for photothermal applications.

The cytotoxicity reported for GNPs has been attributed to the surface charge. In a study in 2004 by Goodman et al., the positively charged GNP (positive charge imparted by the CTAB 241 employed) presented an IC50 of 1 μM against the cell line of Cos-1 cells. When the GNPs were treated such that the surface charge was altered to be anionic, the IC50 increased to 70 μM.400 This was attributed to the ability of the cationic nanoparticles to interact with the negatively charged cellular membrane and the resultant membrane disruption.184, 400 A similar strategy was employed by Takahashi in 2006, extracting the CTAB from an aqueous solution of gold nanorods using a chloroform phase that contained phosphatidylcholine.401 This treatment, of simply removing the positively charged component, enhanced the biocompatibility of the gold nanorods compared to the CTAB coated nanoparticles.184, 401

As discussed in Chapter 2, the surface charge of graphene is of a low magnitude and is dictated by the surfactant coating. The surface charge of graphene is intrinsically negative, arising from the sheet edges, which contain oxygenated sites.209 This indicates that such materials should be intrinsically more likely to be opsonized by the cell and be largely biocompatible.

4.3.3.2.1 Pristine graphene

A clear comparison can be made, when trying to relate the surfactant exfoliated graphene to a non-coated counterpart, by comparing against a pristine, non-oxygenated graphene type of similar morphology and thickness distribution. A study performed by Zhang et al. in 2010 showed that few layered graphene, prepared via a radio frequency chemical vapor deposition 353 method, induced cytotoxic effects and mitochondrial injury. The IC50 value was approximately 10 μg mL-1 against neuronal PC12 cells (derived from pheochromocytoma of the rat adrenal gland). This study drew links between the graphene sheets and CNT’s also tested and concluded that the morphology differences were leading to different toxic mechanism again highlighting the importance of nanomaterial morphology in cytotoxicity.

This study also demonstrated a more in depth analysis by monitoring the release of LDH and caspase activation, which indicates necrosis (via membrane damage) and apoptosis respectively. In this study LDH study indicated that the graphene was predominantly inducing apoptosis in the PC12 cells out to the highest concentration tested 100 μg mL-1. This study also briefly looked at the production of ROS induced by the presence of graphene and determined both a time and concentration dependence with dramatic increases in ROS at 100 μg mL-1 after 4 and 24 hours.

We can therefore summarise that this pristine graphene material without any protective coating has demonstrated a morphology role, is triggering an apoptosis event and is also producing considerable concentration of ROS.

4.3.3.2.2 GO rGO

If we shift away from the mechanical influences on the cytotoxicity of graphene materials, and focus on the chemistry of the materials we can identify the significance of particle surface chemistry on cytotoxicity towards mammalian cells. A helpful comparison is to compare the properties of the highly charged and subsequently hydrophilic graphene oxide with its reduced counterpart with predominantly hydrophobic properties.

A study was published in 2015, showing a convincing trend between the extent of oxidation for prepared graphene oxide and the observed cytotoxicity against mouse embryo 398 fibroblasts. This group tuned the extent of oxidation by changing the amount of KMnO4 used during the Hummers preparation method and confirmed via X-ray photoelectron spectroscopy (XPS) carbon to oxygen ratios of approximately 2.2, 2.4 and 2.9. Interestingly, the trend observed showed that with decreasing levels of oxidized sites, the cytotoxicity increased. The mechanism of toxicity was attributed to high levels of ROS being induced, whereby the lower oxidized sample resulted in stronger indirect oxidative damages. This group concluded that the decreased oxidation degree results in a greater number of free electrons and therefore a stronger oxidative ability allowing high levels of conversions such as OH being produced from H2O2.

243

This is an important insight to the mechanisms of graphene material cytotoxic behavior showing both that the high levels of ROS produced are a key factor in cytotoxicity as well as elucidating the relationship between oxidation degree, free electrons and the extent of ROS production. This group showed that the IC50 was observed at a concentration of approximately 100 μg mL-1.

4.3.3.2.3 RGO

The reduced counterpart for GO, rGO is far more hydrophobic and as such shows very different cell interaction mechanisms, as demonstrated in a study against human liver carcinoma (HepG2). A key difference is that the hydrophilic GO is readily internalized by mammalian cells while rGO will predominantly be found adsorbed the membrane surface.373 This results in distinctly varied events for example internalized GO will result in NADPH oxidase dependent ROS formation and deregulation of the antioxidant/DNA repair/apoptosis related genes. The membrane adsorbed rGO, however, leads to the generation of ROS through physical interaction. Both demonstrated similar toxic responses and were shown to -1 follow a differential dose dependency with an approximate IC50 at 50 μg mL for rGO and 75 μg mL-1 for GO.373

From these results, we can assume that the trend would continue, and that pristine graphene would likely be hydrophobic and therefore not internalized. However, pristine graphene would likely act in a similar manner to rGO, physically inducing the production of high levels of ROS. As mentioned previously in Section 4.1, the close comparison pristine graphene -1 example showed an IC50 of 10 μg mL agreeing with the trend. This information can then be compared to that of the surfactant exfoliated graphene, where the experimental results -1 obtained within this study showed an IC50 of ≈ 10 μg mL .

Accepting that the separate studies were performed on several cell lines we can still see that the cytotoxicity values are decreasing with increased oxidation extents while there is a marked improvement with the tri-block copolymer coating. It is important to highlight that NG108-15 cells are weakly attaching cells and that the results obtained through testing against them will likely overestimate the cytotoxicity compared to the more robust cells, such as HepG2 and fibroblast cells.

The higher than anticipated cytotoxicity of SALE graphene is most likely due to surfactant degradation during ultrasonication. This can be easily addressed via a surfactant exchange or simply via dialysis. However, for the purposes of this study, the concentration was sufficient to demonstrate the thermal ablation of NG108-15 cells and therefore the optimization was not further explored.

4.3.4 Thermal ablation

The first photothermal biomedical role explored for this material was for direct thermal ablation of cancerous cells. This requires sufficiently low toxicity to allow the cells to be incubated with the graphene particles for an appropriate period. The IC50 for the experimental L64-graphene provided a substantial concentration range to advance to the thermal ablation studies.

This can be considered as a simplistic, and somewhat of a brute force attack, which does have some limitations in targeting and collateral damage. However, several targeting strategies have been explored for nanoparticles through attaching bio-conjugates allowing specificity to the target cancerous cells. The efficiency of the targeting methods though is still quite low and if the target is widely distributed then limitations around the activation would also apply.160, 165 However, despite these limitations a substantial role still remains for large scale ablation of bulk tumours which can be significantly beneficial in reducing the size of tumours.

245

From the cytotoxicity curve established for L64-graphene a working concentration 17 µg mL-1 was selected for the substantial viability observed after a significant period of time of exposure (24 hours versus the thermal ablation incubation time of 4 hours). The concentration was set as high as possible to maximize the temperature experienced by the NG108-15 cells, whilst ensuring the integrity of the cells was not extensively compromised.

The irradiation experiments were performed for a range of concentrations and irradiation times to determine appropriate parameters to induce complete cell death. The data presented in Figure 4.5 is for NG108-15 cells incubated in the presence of L64-graphene at a concentration of 17 µg mL-1 for 4 hours prior to irradiation. At this concentration in the neat photothermal curves indicate that the temperature should reach bulk temperatures of ≈ 55 °C.

Samples were split into several groups including ‘irradiated with and without graphene’ and ‘non-irradiated with and without graphene. Triton X was used as the control to show full cell death. As can be seen in Figure 4.5, the irradiated samples containing graphene showed successful thermal ablation after 60 minutes of irradiation.

Figure 4.5. Irradiation of NG108-15 cells with (red columns) and without (blue columns) the presence of graphene after (a.) 60 minutes and (b.) 30 minutes of irradiation.

The irradiated samples containing graphene were almost entirely killed after 60 minutes of irradiation at 400 mW. The incomplete cell death is likely simply due to the laser spot size not fully covering the cells. Therefore, it is the bulk temperature which must effect the cell death of cells not directly irradiated, a value which drops quickly over small distances. It can also be clearly seen that when no graphene is present, there is no significant temperature rise upon irradiation and no cell death results. With all considerations accounted for, these plots show a clear demonstration of the photoresponsive graphene inducing cell death upon activation via NIR radiation. Figure 4.5 b shows the same experiment but for only 30 minutes of exposure, showing the thermal ablation is likely on active after this point. Additionally, within this same group the viability is approximately even between the irradiated and non- irradiated cells.

Hyperthermic injury is the primary mode of cell damage for thermal ablation. Referring to the localized thermal increase injury effecting cell death through changes to the tumour microenvironment and the damages to the cell membrane as well as sub cellular levels as 247 discussed in Section 4.1.375 It is also important to note that cell death induced via thermal ablative methods occurs by both direct and indirect mechanisms.

Heat ablated regions are often described as having three zones, the central zone where temperatures are at the maximum, the peripheral region where mid-range temperatures are experienced and the surrounding tissue unaffected by the nearby temperature increase. The cells in the central region undergo ablation-induced coagulative necrosis, a form of tissue necrosis in which the injury denatures structural proteins and enzymes prohibiting the proteolysis (breakdown of proteins) of dead cells. Tissue architecture can be preserved for days (in vivo) until being removed by infiltrating leukocytes. The peripheral region receives sub lethal hyperthermia largely through thermal conduction.

Direct cellular damage such as that which occurs in the primary heating zone, occurs from the subcellular level to the tissue level. The damage achieved is dependant primarily on the thermal energy and the sensitivity of the target.

At temperatures around 40 – 45 °C, irreversible cell damage occurs only after prolonged exposure such as 30 or 40 minutes of irradiation time.375 In this temperature range the primary feature of cell damage is the inactivation of vital enzymes.375 At temperatures above 60 °C, the time required to irreversibly damage cells decrease exponentially. The primary mode of damage in this higher temperature range is protein denaturation which is immediately cytotoxic and leads to coagulative apoptosis.402 It is likely that the cells are experiencing both a highly localised high temperature damage mechanism, where sheets are in close proximity, as well as a predominantly lower temperature mode of cell damage, from the increased bulk temperature of the media.

From the results in Figure 4.5 we can see that the thermal ablation of cancerous cells can be achieved through the in situ activation of graphene microplates via NIR irradiation. In order to improve the efficacy of this system further, several simple strategies can be employed, such as optimising the laser power and increasing the biocompatibility of the particles through alternative coatings.

Additionally, routes to specifically target cells through bio-conjugation have been expanding recently and are undoubtedly the way forward to ensure the delivery of photothermal agents.160-163 A common method towards achieving the attachment of the bio-specific conjugates is via PEGylation, positioning SALE graphene as well prepared for both stealth and specific targeting roles. As surfactant exfoliated graphene has a coating of triblock copolymers adsorbed to the surface, this material is initially prepared suitable for such stealth applications as well as for specific targeting bio-conjugation.

As surfactant exfoliated graphene is intrinsically capable of addressing excessive uptake and allow sites for conjugation of bio-specific macromolecules, it can be considered highly appropriate to act as a photothermal agent. The facile and high yield production method, the strong absorbance in the NIR allowing efficient conversion of incident NIR radiation in situ, through to its stealth and biocompatible properties and thermal stability all lend themselves to address the requirements of a photothermal agent. The next step is to shift to in vivo studies and explore optimal specific targeting strategies and hopefully add a more stable, less toxic photothermal agent to the anti-cancer armoury.

249

Chapter 4 Conclusions

Surfactant assisted liquid exfoliated graphene was demonstrated to be a strongly photothermally responsive material in the NIR. The biocompatibility of SALE graphene was explored against NG108-15 mouse neuroblastoma cross rat glioma cells showing moderate -1 biocompatibility with an IC50 of 12.1 µg mL allowing substantial working concentrations for thermal ablation experiments. Finally, the graphene sheets were successfully employed to photothermally ablate the cancerous NG108-15 neuronal cells with NIR radiation.

251

5.0 Chapter 5 Photothermal breaking of emulsions stabilised with graphene

5.1 Introduction

In this study, pristine graphene particles prepared by the surfactant assisted liquid exfoliation (SALE) technique are explored to promote the stabilization of emulsions through adsorption at the oil-water interface. Highly localized phase separation of these highly stable emulsions could however be induced through photothermal heating of the graphene particles at the interface exposed to near-infrared (NIR) light. The graphene wettability, which is a key determinant in preventing droplet coalescence, can be altered through the adsorption of non- ionic block copolymer surfactants. Varying the aqueous solution conditions influenced the hydration of the surfactant providing a further opportunity to alter the overall particle wettability and hence stability of the emulsion.

In this way, highly stable oil in water (o/w) emulsions were produced with decane; and water in oil (w/o) emulsions were formed with toluene as the oil phase.These emulsions can be employed as versatile drug carriers, and with this as the focus, the NIR photoactivation of the emulsion breaking was explored.

5.1.1 Emulsions for drug delivery

Dispersions of an immiscible liquid in a continuous phase of another liquid, or emulsions, are common in many cosmetics,403 food , medical404-406 and industrial applications. For macroemulsions where the droplet size is typically greater than 1 μm, both o/w and w/o types are possible. Surfactants are usually added in order to reduce the interfacial tension (the energetic cost of creating oil-water surface area) between the oil and water phases that results in significantly improved stability of the dispersed liquid droplets against phase separation. The type of emulsion depends largely on the properties of the surfactant or emulsifier used as well as the relative proportion of oil and water in addition to the preparation method. Macroemulsions are not thermodynamically stable, due in large part to the high interfacial area. Adsorption and desorption of amphiphilic emulsifiers, such as surfactants at the oil- water interface, is dynamic, which eventually results in the formation of necks between closely associated droplets. The energy of removing a surfactant monomer from the interface is approximately the same as the thermal energy of the system. Increased stability of the emulsion can be achieved by using particles to form so called Pickering emulsions. These particles typically have intermediate wettability (contact angle with water ~ 90°) and their size means that the energy required to remove them from the interface is orders of magnitude greater than the thermal energy.407-408 Hence Pickering emulsions are stable over greater time periods than emulsions formed by using simple surfactants.

Particles of different morphologies, sizes and chemistries have been employed in the stabilization of emulsions409-410 with an increased focus on particulate carbon materials such as carbon nanotubes (CNTs)411-413 and graphene oxide (GO) in recent years.414-418 A typical feature is to tune the relative wettability using adjustments to pH which influences surface potential of particles with weakly ionizable groups.419 A recent study used GO in order to improve the stability of o/w emulsions through alterations of the pH of solution.418 The use of GO or similar atomically thin crystalline materials is attractive for a number of reasons, but perhaps the most important is the vast surface area to volume ratio.104, 420-421 Furthermore, graphene and analogous single layered two-dimensional (2D) particles such as MoS2 have interesting electronic, optical, thermal and mechanical properties.196, 422-423 Aside from the large interfacial area, 2D particles also strongly interact with incident light as the molecularly thin sheets allow all atoms to participate in absorption, which is distinct from the corresponding three-dimensional (3D) materials.424 Graphene also absorbs light across a broad spectrum. This includes in the NIR region for pristine graphene or highly reduced graphene oxide (rGO) due to the highly conjugated nature of extended π bonded network.66, 105 Hence, graphene shows great promise in biomedical applications where exposure to NIR radiation induces photothermal heating.

253

Whilst inhibiting phase separation is important, many applications such as in drug delivery may benefit from controlled breaking of emulsions on action of any external stimulus.425-426 Here, graphene stabilised emulsions are demonstrated along with localized phase separation induced via photothermal heating on absorption of NIR light by the interfacial graphene particles.

5.1.2 Polyethylene oxide solubility properties

Polyethylene oxide (PEO) is a highly versatile polymer which has excellent biocompatibility properties and as such has been explored extensively for polymeric drug release applications and many other biomedical roles.427-430 Polyethylene oxides come in a wide range of different forms, from simple changes to size, to the presence of additional functional groups, as well as the block copolymer modifications, such as the triblock copolymers employed within this study.

It has been well established that aqueous solutions of PEO’s present a unique phase behavior, where the solubility of the polymer is decreased upon the increase of temperature past a certain (specific to each polymer variation) critical temperature known as the cloud point.431- 432 This has been attributed to the loss of hydration number per ethylene oxide unit, which has been demonstrated to decrease from ≈ 4 water molecules per unit at 25 °C, to ≈ 2 water molecules per unit at 70 °C.431, 433-434 The decreasing of hydration subsequently results in the phase transition of the PEO.

This is an important consideration for the graphene suspensions prepared with the triblock copolymers from several perspectives including a sample handling perspective, a suspension stability perspective as well as for possible routes of sample control. As the triblock copolymers, consisting of polypropylene oxide PPO segments and PEO segments, are adsorbed to the basal plane, the PEO segments extend into the solution and the solubility of these chains is critical for total suspension characteristics. By altering the extent of hydration, and therefore the extent by which these chains extend into solution, the surface chemistry of the particles can effectively be controlled.

5.2 Materials and methods

SALE was performed as described in Chapter 2, with a key consideration into the identity of the surfactant. By choosing surfactants with varying hydrophilic-lipophilic balance (HLB), o/w or w/o emulsions could be prepared. Three different triblock copolymers were used in this study: F127, P123 and L64 (see Table 2.1 for surfactant composition data). The emulsification was performed (predominantly) using two different oils: decane and toluene that were of analytical grade. Chloroform and tetradecane were also explored to demonstrate altered environments.

5.2.1 Preparation of emulsions

Graphene suspension (2 mL) was added to the oil phase (2 mL) followed by agitation by hand for approximately 30 seconds. To test the influence of salt concentration, the aqueous phase was prepared at ranging concentrations of NaCl prior to emulsification. The samples were left to stand in a glass vial for 3 days prior to image capture.

The electrical conductivity of the emulsion was subsequently measured in order to determine whether o/w or w/o emulsions were formed.

255

5.2.2 Emulsion characterization – emulsion stability

Emulsion sets, and their respective blanks, were prepared and left to stand with images taken at regular intervals to observe the rate of emulsion destabilization.

Optical microscopy was performed in order to determine the size distribution of the emulsions prepared over time. A small drop of the emulsion to be measured was placed onto a microscope slide with a coverslip placed on top. After a 60 second waiting period, 3 images were taken at 3 magnifications and the drop size was measured using ImageJ software. The average of 3 drops per emulsion sample was used.

5.2.3 Breaking of emulsions

Photothermal heating of water and the aqueous graphene suspension with a concentration of 0.03 mg mL-1 was first investigated prior to experiments with emulsions. 2 mL samples were placed in cuvettes and irradiated with two different NIR lasers (808 nm @ 500 mW, and 980 nm @ 185 mW) for a period of 20 (and 15) minutes. A thermocouple measured the temperature of the sample at a height of approximately 12 mm above the incident laser spot position. Photothermal heating of the emulsions in the absence and presence of graphene was subsequently undertaken using the NIR lasers.

The laser was directed horizontally through the emulsion and the propagation position was identified via the light scattering center observed via a dyno-lite digital video feed.

5.3 Results and discussion

5.3.1 Emulsion system control

The two oil phases chosen for use in the preparation of particle stabilised emulsions were toluene and decane. Toluene is poorly soluble in the aqueous phase (less than 0.1% w/w at 293 K),435 whereas decane is insoluble. The interfacial energies of toluene-water and decane- water interfaces are 36.6 mNm-1 and 51.2 mNm-1 respectively.436-437 These interfacial tensions are of the order of the surface energy of the unmodified graphene (≈ 41 mNm-1), however surfactant adsorption can change this depending on the relative balance of hydrophilic and lipophilic components (or HLB value). Thus, the exfoliated graphene particles stabilised with various non-ionic surfactants, having differing affinities for the interface depending on the oil phase. However, the wettability of the particles can be further tuned using salt addition, which was the primary mechanism employed within this study.

Table 5.1. Relevant properties of solvents employed for the Pickering stabilised emulsions.

Solvent ID Density (g cm-3) Interfacial tension with water mNm-1 @ 25 °C

Chloroform 1.49 32.8

Decane 0.73 51.2

Tetradecane 0.764 43.4

Toluene 0.867 36.6

As discussed in Chapter 2, the surfactants act not only to aid in the exfoliation, the PPO segments also adsorb to the sheets and stabilise the graphene suspensions.

257

Under low electrolyte conditions, the hydrophilic nature of the PEO chains results in the chains being extended away from the interface and into solution. Increasing the concentration of the electrolyte in the aqueous phase results in the collapse of the PEO chain segments toward the surface of the graphene sheet. This is due in large part to the dehydration of the ethylene oxide groups that effectively reduces the hydrophilicity of the graphene-polymer particles.

It has been well demonstrated that the addition of salts can effect the dehydration of PEO chains.438-440 The addition of salt dehydrates the PEO via a salting out action similar to that of temperature. Salting out refers to the scenario whereby the addition of electrolytes results in a greater interaction between the water molecules and the salt molecules, which in turn means a decrease in interaction between water and the PEO groups.371 This therefore leads to a greater PEO-PEO interaction and the decreased solubility. It has also been suggested that the PEO moieties closer to the PPO segments are dehydrated first compared to those further away.441-442

Hence the addition of salt has two effects, the first relating to the surface forces between particles (i.e. DLVO and steric stability, as discussed in Chapter 2) and the second relating to the wettability of the particles. This latter property can be highlighted when observing the contrasting behavior of these suspensions in emulsions formed with different oil phases.

Figure 5.1 shows the emulsions formed when an aqueous graphene suspension is mixed with decane as a function of increasing salt concentration. Two different surfactants were used in this instance to stabilise the graphene sheets, L64 and P123 that have HLB values of 12-18 and 7-9 respectively. In Figure 5.1 it can be seen that the particles with the more hydrophobic surfactant P123, adsorbed spontaneously to form an emulsion in decane through efficient immobilization at the interface. This suggests in the absence of salt, the graphene-P123 particles have the intermediate wettability required to effectively stabilise the decane-water interface.

Figure 5.1. Decane in water particle stabilised emulsions formed from exfoliated graphene stabilised with the non-ionic surfactant L64 (top) and P123 (bottom). The salt concentration increases from left to right in the order 0.01, 0.05, 0.2, 0.5 and 1.2 M.

The graphene-P123 particles become increasingly hydrophobic upon increasing the electrolyte concentration resulting in an emulsion with lower stability. Furthermore, at salt concentrations greater than 0.2 M, significant aggregation of the graphene sheets was apparent through the presence of sediment material at the bottom of the vials. Conversely, the graphene-L64 particles are relatively hydrophilic in the absence of any added salt and favorably remain in the aqueous phase. Increasing the electrolyte again reduces the hydrophilicity of the graphene-L64 particles resulting in the formation of increasing amounts of stable emulsions. Particles that remained in the aqueous phase do however, show obvious signs of increased sedimentation at high salt concentrations after a few weeks, reflecting the reduced stability of these particles against aggregation.

259

O/w type emulsions were observed for all of the graphene-surfactant particles when the aqueous suspension was used in conjunction with decane. This was confirmed through electrical conductivity measurements of the emulsions that showed conductivities of the order of the aqueous phase.

Table 5.2. Emulsion phase conductivities for phase identity confirmation.

Phase identity Conductance (S)

Decane 0

Emulsion 475

Aqueous 525

5.3.2 Emulsion stability

The emulsions typically prepared using the surfactant modified graphene sheets were stable for many weeks with no obvious signs of droplet coalescence. This was initially determined from a macro perspective, simply observing the volume of emulsion over time with a series of images. The stability of the emulsion was markedly greater than those formed using just the non-ionic block co-polymer surfactants however the macro method resulted in messy data. An alternative method of optical microscopy and droplet size analysis was employed to provide more a more robust analysis method. (Figures 5.2, 5.3 and 5.4).

Figure 5.2. Optical microscopy images of emulsions formed with decane and (a.) aqueous L64-graphene suspension or (b.) aqueous surfactant solution. Droplet size distributions with and without graphene at time = 60 minutes.

Figure 5.3. Histogram showing the frequency of drop sizes for (red) L64-graphene stabilised emulsions compared to (blue) the emulsions without the particles present after 60 minutes of emulsion preparation. Cumulative distribution is included for both (red) the graphene containing and (blue) free emulsions.

261

Figure 5.4. Frequency of drop sizes for (red) L64-graphene stabilised emulsion and (blue) the graphene free emulsion after 24 hours. The increasing mean size of the droplets over time indicates faster destabilization of the non-particle stabilised emulsion drops. Cumulative distribution is included for both (red) the graphene containing and (blue) free emulsions.

The surfactants used herein are able to stabilise the oil-water interface without graphene. Figure 5.2 shows example optical microscopy images of the droplets formed in the o/w emulsions using the neat L64 surfactant and L64-graphene particles 1 hour after preparation. The mean size over 24 hours of the droplets is smaller for the sample with graphene indicating the improved stability afforded by the presence of particles at the decane-water interface (Figures 5.3 and 5.4). Indeed, the emulsions with graphene present were stable for a period of more than 6 months in comparison to those with simply the neat surfactant that showed earlier signs of degradation. Short term drop size distribution analysis was the most appropriate method to demonstrate the improved stability in the presence of the particles (Table 5.3). This can be looked at from a time perspective with respect to one surfactant type, as per Figures 5.5 and 5.6, to see the change over time with and without graphene present.

Figure 5.5. L64-graphene stabilised emulsion ageing of drop size after a 24 hour period. Presenting a median size shift from 4.41 µm to 7.19 µm. Cumulative distribution is included for both (red) day 1 and (blue) day 2.

Figure 5.6. Graphene free emulsion ageing of drop size after a 24 hour period. Presenting a median shift from 9.27 µm to 12.80 µm. Cumulative distribution is included for both (red) day 1 and (blue) day 2.

263

Graphene particles stabilised with the very hydrophilic (HLB > 24) surfactant F127 were also used in the preparation of emulsions with different oil phases as shown in Figure 5.7. The graphene-F127 particles showed similar behavior to the graphene-L64 particles when forming an emulsion with decane. That is, with increasing salt concentration, a greater volume of o/w emulsion was observed. In contrast to the graphene-L64 particles, the emulsion was stable for longer with the graphene-F127 particles and no significant sedimentation was apparent over a period of weeks. This reflects the greater ability of the F127 surfactant to sterically stabilise the particles against aggregation due to the longer ethylene oxide chains extending into the aqueous solution, even in the semi-dehydrated state due to the presence of high concentrations of electrolyte.

Figure 5.7. Emulsions formed using an aqueous suspension of F127-graphene with (top) decane and (bottom) toluene. The salt concentration increases from left to right in the order 0.01, 0.05, 0.2, 0.5 and 1.2 M. Figure 5.7 also shows the emulsions formed with the aqueous suspension of F127-graphene and toluene as the oil phase. Clearly there is a significant difference in the emulsions with toluene in comparison to decane. Conductivity measurements show that a water in oil (w/o) type emulsion is formed with a decreasing volume observed with increasing salt concentration. The interfacial tension between toluene and water (36.6 mNm-1) is significantly lower than that of decane and water (51.2 mNm-1) and indeed toluene has some affinity for water due to its (very) weakly polar nature. The formation of w/o emulsions suggests that the oil phase preferentially wets the F127-graphene particles however increasing the salt concentration further dehydrates the stabilizing PEO chains leading to increased hydrophobicity and diminished emulsion volume.

Table 5.3. Median size of L64-graphene stabilised and blank emulsions over time showing the rate of the increase in emulsion drop size over time is larger for the blank emulsion samples, indicative of a faster destabilization of the emulsion.

Day L64-Graphene Emulsion L64 Emulsion Median Median Size (µm) Size (µm)

1 4.413 9.271

2 7.197 12.802

The free energy required to remove the solid graphene particles from the interface can be estimated using Equation 5.1 where r is the graphene particle radius, γow is the decane-water interfacial tension and θ is the contact angle (Figure 5.8). From Equation 5.1, the maximum energy occurs when the contact angle of the particle at the three phase line is 90°.443-444 For graphene particles with an effective radius of 75 nm, the free energy is ≈ 1.5 x 105 kT. Even for a significantly lower contact angle of 20°, the free energy cost is still > 500 kT. This compares to the neat surfactant where the dynamic adsorption to; and desorption from the interface is of the order of the thermal energy of the system. The relative diffusion rate for surfactant monomers is also significantly higher than that of large particles making the

265 creation of the necessary fluctuation of the adsorbed layer density more likely to induce coalescence.

2 2 퐸 = 휋푟 훾표/푤(1 ± 푐표푠휃) Equation 5.1

Figure 5.8. Sum of free energy (divided by kT) as a function of (a.) contact angle for a 75 nm radius particle and (b.) radius of particle size at 90° contact angle.

The presence of graphene at the interface clearly inhibits phase separation. At the concentrations used here of 0.1 % w/w, little change to the viscosity or density of the aqueous phase is observed so gravitational effects are minimal. The exfoliated 2D materials have a vast surface area in comparison to many other nano and micrometer sized particles that have been previously used. Indeed, the fully exfoliated surface area of graphene is 2630 m2 g-1.445 The surface area of the graphene sheets generated using the surfactant assisted ultrasonic exfoliation technique however is somewhat lower due to the presence of few layer material and of the order of 500-1000 m2 g-1 depending on the process parameters (for an average 300 x 300 nm sheet stack of 5 layers ≈ 550 m2 g-1).

5.3.3 Induced emulsion breakage

The emulsions are temperature sensitive and the bulk emulsion breakage can be achieved through a simple hot plate demonstration (Figure 5.9). A combination of altered interfacial tensions (oil/particle, water/particle), surfactant solubility and desorption energy being surpassed lead to droplet coalescence.

Figure 5.9. The emulsions were heated and observed to determine the stability and breaking point. Images of L64-graphene emulsion (1.25 M NaCl), at 25, 42 46, 49, 53 and 55 °C showing the collapse of the emulsion above 42 °C.

L64-graphene particle stabilised o/w (decane/water) emulsions, as well as neat surfactant stabilised o/w emulsions were then irradiated using the 808 nm laser. Figure 5.10 shows a time series of optical images of the emulsions with the laser impinging from the left. In the neat surfactant sample, the laser is scattered by the oil droplets nearest to the cuvette wall. No changes in the images are observed with time indicating that the heating effect in the absence of graphene at the oil-water interface is minimal. However, the images of the graphene stabilised emulsions show that the laser light is progressively propagated through the emulsion. The laser transmission through the emulsion was detected as a function of time and shown in Figure 5.11. It can be seen that upon initial irradiation the transmitted light increases quickly indicating a reduction in scattering due to localized breaking of the emulsion. As droplets coalesce, the emulsion becomes more transparent due to a reduction in scattering allowing the laser light to penetrate further into the cell and eventually out the other side. In the absence of graphene at the interface, no significant laser light intensity was detected. Figure 5.11 b also shows an image of the graphene stabilised emulsion 267 demonstrating the “tunnel” created by the laser light due to the localized phase separation. The spot size of the laser is approximately 0.5 mm and the phase separated region remains for more than 2 weeks due to the relatively high viscosity of the o/w emulsion. No visible signs of phase separation were observed for the neat surfactant stabilised emulsion sample.

Figure 5.10. Emulsion stability at t = 1, 2, 3, 16 and 60 seconds of exposure to 500 mW CW laser with wavelength of 808 nm (top) in the presence of graphene and (bottom) in the absence of graphene. The laser light is initially only scattered at the cuvette wall however as the emulsion breaks, light scattering is progressively observed throughout the emulsion.

Figure 5.11. The transmitted light was measured showing the delayed transmission for (red) the graphene emulsion, compared to the completely scattered light occurring upon (blue) the blank emulsion irradiation. The delayed transmission indicates the time required for the photothermal “coring” to take place and allow the light to pass through to the detector.

It is clear that graphene present at the oil water interface has a significant effect on the stability of these emulsions. Equally, localized heating of the emulsion resulting in phase separation can be induced photothermally. The oil-water interfacial tension is highly sensitive to increases in temperature with the thermal conductivity of the continuous phase (water in this case) orders of magnitude smaller compared to that of graphene. This means that the heat generated due to light absorption by the graphene particles is not efficiently dissipated through the medium leading to a rapid, highly localized temperature increase and reduction in interfacial tension. The reduced oil-water interfacial tension changes the affinity of the graphene particles for the interface resulting in de-stabilization and droplet coalescence.

Other light sources could also be used due to the strong broad-spectrum absorbance of graphene (Figures 5.12 and 5.13 show the same experiment with a 980 nm wavelength laser). This opens up the potential for controlling the stability for a range of industrially relevant emulsion systems.

Figure 5.12. 980 nm laser at 185 mW irradiation of emulsions prepared (top) in the presence of graphene and (bottom) in the absense of graphene.

The images from left to right show the progression of the light scattering centre of the laser through the graphene containing emulsion and the lack of any progression for the blank emulsion. In addition to the primary experiment performed at 808 nm this experiment demonstrates the versatility of a material having such a broad absorbance spectrum.

269

Figure 5.13. The transmitted light was measured showing the delayed transmission for the graphene emulsion indicating the time required for the photothermal “coring” to take place and allow the light to pass through to the detector.

Figure 5.14. Photothermal heating curves for (red) the L64-graphene suspension, (blue) the emulsion phase during breakdown and (green) the neat water samples. The thermal diffusion within the emulsion phase is significantly decreased even at a distance of 12 mm from the irradiation site indicating the localized region of heating.

The surfactants used to stabilise the graphene against aggregation in the aqueous phase and tune the overall wettability of the particles themselves, become less soluble at elevated temperature due to the further dehydration of the polyethylene oxide chains. Indeed the cloud point for L64 is 58 °C in water but is further depressed to 40 °C in the presence of 1.0 M salt.446 It is likely then that the graphene particles become more hydrophobic. Together with the change in interfacial tension, the altered wettability of the particles provide the necessary driving forces to overcome the high energy cost associated with removing a particle from the oil-water interface.

Light with a wavelength in the NIR was used in this study to demonstrate the utility of photothermally induced phase separation of an emulsion with a view toward potential biomedical applications and in particular non-water soluble drug delivery. The loading capabilities of emulsions are well understood and already in use for drug delivery purposes.447-448

The laser spot size used here was relatively small giving rise to phase separation in a region of a similar scale. A greater ability to tightly target delivery of the payload can be achieved with such control over emulsion breaking.

Chapter 5 Conclusions

Tight control over the highly localized breaking of emulsions was achieved through the use of photothermal heating of the oil-water interface. The heating was confined to small volumes through the absorption of NIR radiation by graphene used to stabilise the emulsion droplets against coalescence. The wettability of the graphene particles was tuned through the adsorption of nonionic surfactants and the use of varying aqueous solution conditions in order to extend the lifetime of the emulsions by immobilizing the graphene to the oil-water interface. These Pickering emulsions using graphene showed high stability under ambient conditions until the action of the external heat trigger induced through exposure to NIR light resulted in phase separation. This study hence demonstrates a potential new option for targeted breaking of emulsions in biomedical applications such as drug delivery where the delivery of the payload to confined areas is a necessity to avoid potential side effects.

271

By demonstrating the localized NIR-graphene induced drug carrier breakdown, the in situ activation of graphene in an appropriate environment is shown. Additionally, the strategy within this system is not to simply to add the graphene in, but to find a beneficial role for the triblock copolymer coated nanosheets in addition to the photothermal activation. By both stabilizing the emulsion and providing the mechanism for release, the versatility of graphene for such roles is thoroughly demonstrated.

273

6.0 Chapter 6 Graphene as a photothermal actuator for control of lipid mesophase structure

6.1 Introduction

This study demonstrates the potential for use of graphene as a photothermal actuator across a range of lipid based systems of interest in controlled drug delivery. Surfactant assisted liquid exfoliated (SALE) graphene was incorporated within bulk lipid samples of varying lipid types: glyceryl monooleyl ether (GME), glyceryl monooleate (GMO) and phytantriol (PYT)(3,7,11,15-tetramethylhexadecane-1,2,3-triol). The three lipids were selected to demonstrate the wide range of systems SALE graphene can be incorporated within, in addition to demonstrating the versatility of the lipid systems for such drug release roles.

6.1.1 Lipids in drug release

Lipids have been studied extensively for applications including food science,449-450 nanostructure templating, biosensing,451-452 protein crystallization453-454 and drug delivery.425, 455-456 Lipidic materials are particularly attractive due to the intrinsic biocompatibility of these building blocks of the biological world. It is critical to understand the structure being formed within the lipid matrices for drug delivery applications, in order for the release rates and transition points to be determined and the intended functions to be achievable.

Amphiphilic lipid molecules pack into highly ordered liquid crystalline (LC) phases in the presence of water,457-458 other surfactants456, 459 and a range of nanoparticles,460-461 in a highly predictable and well-studied manner. Lipid molecules such as GME, GMO and PYT are synthetic or semi-synthetic lipidic materials commonly employed in cosmetic and pharmaceutical formulation (Figure 6.1). These can be loaded with high quantities of drugs with varied physiochemistry462 and have been recently explored as potential drug delivery vehicles. Phase transitions of these materials can be readily induced through temperature changes including the use of a near-infrared (NIR) photothermal agent.425, 460 The direct structural determination of the bulk lipid samples is achievable through synchrotron small angle X-ray scattering (SAXS) which allows the structural evolution of the phases to be observed when induced through heating / cooling with excellent time resolution. This shows a potential method in controllable localized heating within the body as a promising step towards producing highly efficient drug delivery systems.

Figure 6.1. Chemical structure of (a.) GME, (b.) GMO and (c.) PYT with carbons blue and oxygens red. Drawn using ACD/ChemSketch (Freeware), version 14.1, Advanced Chemistry Development, Inc., Toronto, ON, Canada, www.acdlabs.com, 2015.

Several nanomaterials have been explored in recent years for the role of lipid drug delivery NIR photothermal agents including gold nanoparticles,169, 175 however, the chemically and photothermally stable graphene derivative materials are yet to be explored. This work builds on these recent studies and demonstrates the use of graphene as a versatile photothermal agent in lipid-based liquid crystalline materials prepared from GME, GMO and PYT.

6.2 Materials and methods

6.2.1 Graphene preparation

SALE of graphene suspensions was performed as described in Chapter 2, with a key consideration into the identity of the surfactant. The triblock copolymer F108 was employed as the primary surfactant for the graphene suspensions mixed with the lipid dispersions, as

275 previous studies have showed suitable integration of large, high percentage polyethylene oxide (PEO) triblock copolymers with lipid dispersions.463

6.2.2 Lipids

The PYT was received as a gift from DSM Nutritional Products (Singapore) with a nominal purity of 98.7%. Myverol 18-99K, commonly used in place of high purity GMO due to almost identical phase behavior, was donated by Kerry Bio-Science (Norwich, NY). The analytical data indicates that Myverol 18-99K (certificate no. 31500455) contains 58.3% glyceryl monooleate (C18:1), 12.2% glyceryl monolinoleate (C18:2), 5.1% glyceryl monolinolenate (C18:3), 3.9% glyceryl monopalmitate (C16:0), 1.7% glyceryl monostearate (C18:0), 0.96% glyceryl monogadoleate (C20:1), 0.2% glyceryl arachidonate (C20:4), 0.1% free fatty acids, and 0.4% glycerol. Trace amounts of unquantified triglycerides are also believed to be present. The GME was a gift from Cee Chemicals (Blacktown, NSW, Australia).

6.2.3 Formulation preparation

The lipid matrices were prepared by heating to 75 °C, in order to reduce the viscosity of each lipid, then adding in the aqueous F108-graphene suspension (at ranging concentrations as appropriate) at a 1:1 ratio. Each sample was then vortexed to mix thoroughly. The heating/vortexing cycle was repeated three times to achieve homogenous mixtures, samples were placed on a “roller” overnight and then were stored at 4 °C for two weeks prior to any experiments.

6.2.4 Small angle X-ray scattering (SAXS)

SAXS data was collected on the SAXS/WAXS beamline at the Australian Synchrotron.464 An X-ray beam with a wavelength of 1.033 Å (12 keV) was selected. A silver behenate standard (d-spacing = 58.38 Å) was used for the q range calibration. The scattering patterns were acquired using a Pilatus 1M detector with active area 169 x 179 mm2 and with a pixel size of 172 μm. The total q range for the instrument configuration was 0.02 < q < 1.00 Å-1.

For temperature-dependent structural data, samples were transferred to a vertical sample holder and sandwiched between kapton polyimide tape. The sample holder was then mounted in front of the SAXS beam on a temperature controlled plate and an empty sample position was fitted with a thermocouple. Snapshots were collected at 5 °C intervals from 25 °C through to 70 °C and each temperature was held for 5 minutes to allow equilibration prior to the snapshots being taken. The calibration curves of lattice parameter were determined via the q value of the primary peak versus temperature, using the approach described by Fong et 460 al.. This allowed the apparent temperatures (Tapp) of the samples to be determined during the irradiation experiments. This is achieved by finding the q value of the sample irradiated and applying the appropriate phase calibration equation to determine the corresponding temperature. This method allows the apparent temperature experienced by the lipid matrix during the experiment to be determined. This was performed at a range of graphene concentrations within the bulk lipid samples.

For dynamic NIR irradiation studies, the samples were loaded onto a vertical sample holder and mounted in front of the SAXS beam with the laser optical outlet positioned at a distance of 12 mm from the sample in order to achieve full coverage of the sample, 5 mm diameter, to ensure homogenous heating. The irradiation experiments were performed using a 400 mW 808 nm continuous wave (CW) laser. Each sample was irradiated for 60 seconds with 100 ms SAXS images collected every 5 seconds. This allowed the mapping of the phase transitions as the samples were heated up via the photothermal transduction of the NIR

277 radiation. Additionally, the samples were allowed to cool with continued SAXS snapshots to capture the reversibility of the graphene loaded lipids.

Avoiding radiation damage from the synchrotron beamline during the data accumulation was important in order to gather clear scattering patterns. To ensure radiation damage was avoided, a test was performed in which the exposure time was incrementally increased until radiation damage occurred. This allowed a safe exposure time range to be established well below the point of radiation damage.

6.2.5 Crossed polarised light microscopy

The secondary phase boundary analysis method employed for this study was cross polarized light microscopy. Polarized lenses on an Olympus BX53 visible light microscope were used to image the samples mounted on a temperature controlled stage (Linkam). A small portion of each sample was deposited onto a microscope slide, and a coverslip was placed on top of the sample. Each sample was then observed with no polarized lens to ensure that excess water was present. The polarized lens was then replaced and the samples were heated at 5 °C min- 1 until the sample was within 5 °C of the expected transition temperature, then the heating rate was adjusted to 1 °C and held for 1 minute at step to allow equilibration.

6.3 Results

6.3.1 Equilibrium structural studies

Three lipids were selected and were all studied in excess water, at a prepared ratio of 1:1 lipid to water (50 wt. %). Under these circumstances the phases observed in excess water with increasing temperature included the inverse bicontinuous cubic phase QII, the reversed hexagonal phase HII and the isotropic inverse micellar phase LII. The inverse bicontinuous cubic phase present for both the GMO and the PYT systems was the double diamond Pn3̅m Q224 type. This inverse bicontinuous phase is connected in 3D space with cubic symmetry and consists of two interwoven networks of water channels, separated by the bilayers.465-466 224 The QII phase will henceforth be referred to as Q for clarity. The reversed hexagonal phase consists of a dense packing of water filled rods arranged on a two-dimensional (2D) hexagonal lattice separated by lipid bilayers and the LII phase consists of the reversed micellar phase with no long range ordered packing and is therefore characterised by the singular broad peak. The Q224 phase can be identified by the Bragg peak positions, which present reflections at spacing ratios of √2: √3: √4: √6: √8: √9 as shown in Figure 6.2 a. The

HII phase is identified by the reflection ratios of 1: √3: √4 as shown in Figure 6.2 b.

Figure 6.2. Representative scattering patterns for the three phases seen in the current study, (a.) the inverse bicontinuous cubic phase (Pn3̅m -type, Q224), (b.) the inverse hexagonal phase and (c.) the isotropic inverse micellar phase, identified with the corresponding reflection ratios (where appropriate). Phases presented are for PYT sample containing graphene (0.3 mg mL-1) at 25 °C, 50 °C and 65 °C respectively.

Initial scattering snapshots were collected for blank and graphene-loaded samples in order to determine whether the F108-graphene sheets would disrupt the packing of the lipids at room temperature. There were no significant differences in the scattering patterns between the graphene loaded samples and the corresponding blanks (Figures 6.3 – 6.5).

279

Figure 6.3. GME samples with (a.) no graphene present and (b.) 0.3 mg mL-1 present.

Figure 6.4. GMO samples with (a.) no graphene present and (b.) 0.3 mg mL-1 present.

Figure 6.5. PYT samples with (a.) no graphene present and (b.) 0.3 mg mL-1 present.

The only identifiable differences between the loaded and unloaded sample scattering plots was the influence of the F108 on the LC packing. As previously reported by Dong et al. in 2006, the presence of the large, hydrophilic triblock copolymer aids in the packing of the lipid molecules and results in cleaner scattering data.463 Importantly, despite some minor differences between the SAXS patterns, the primary Bragg peak position remained constant. It is important to note that with the addition of nanoparticles, it is expected that some small influences to the packing parameter may occur. As shown by Fong et al. in 2012, a concentration of 3 nM gold nanorods, a clear deviation of the lattice parameter can be observed. No lattice parameter deviations were observed within the graphene concentrations of 0.3 mg mL-1 being equivalent to 25 µM, dramatically greater than that of the GNR concentration employed by Fong et al.. This suggests that the graphene sheets may be capable of being in tighter contact, at higher concentrations, while being less obstructive o the lipid packing structure, perhaps through a scrolling mechanism (particularly relevant for hexagonal phases).

The PYT system is ideal for phase analysis of the lipid systems as full transitions from Q224 to HII to LII can be easily induced via a controlled temperature ramp as shown in Figure 6.6 c. This allows a more comprehensive view of the phase transitions occurring within the NIR irradiation experiments, with an additional in depth look at the lattice structure at any one temperature or phase. As the temperature was increased the primary Q224 peaks shifted to larger q values indicating that the spacing of the LC lattice decreased (Figures 6.6 a, b and c). This trend can be observed until a marked shift between phases occurred at approximately

50 °C where the HII phase emerged.

281

Figure 6.6. Changes in structure upon heating the equilibrium lipid + water systems. The temperature dependence of the lattice parameter, derived from the position of the primary reflection in the SAXS profiles, is presented for peak scattering position for (a.) GME + water system, (b.) GMO + water system and (c.) PYT + water system. The plots show the phase identity at each point presented as blue diamonds representing the Q224 phase, red squares show the HII phase and the green triangles represent the LII phase. These profiles are subsequently used as calibration temperature ramps for determining temperature changes during dynamic irradiation experiments. Figure 6.6 d shows the SAXS profiles obtained during temperature ramp for the PYT + water system from (top profile) 25 °C through to (bottom profile) 70 °C at 5 °C intervals.

The transition to the LII phase, within the PYT dispersion, occurred at approximately 65 °C which enables some indication of temperature, but is not a crystalline structure making it less useful in ‘calibrating’ the apparent temperature experienced by the matrix during the subsequent NIR irradiation experiments.

The lattice parameters of the structures formed by GME (Figure 6.6 a) and GMO (Figure 6.6 b) also followed the expected trend of decreasing with temperature, with the GMO system showing the characteristic marked decrease in lattice spacing associated with the transition 224 from Q to HII. The lattice parameter to temperature calibration was performed for each lipid dispersion to allow the NIR induced temperatures to be determined for each system.

-1 Figure 6.6. a shows the changes (in Tapp) for the 0.3 mg mL of graphene in GME bulk lipid sample upon NIR irradiation, showing the (Tapp) increase from 20.0 to 64.9 °C. No phase transition was recorded for this bulk lipid sample which is in agreement with previously reported values as the transition from hexagonal phase to LII phase is expected to occur at approximately 80 °C.467

Figure 6.6 b shows the converted SAXS plot for the laser irradiated F108-graphene loaded GMO sample. It can be seen that the temperature quickly increases up to 56.6 °C where the 224 phase transitions from Q to HII. The apparent temperature reaches a maximum of approximately 69.8 °C. From this data it can be observed that the GMO phase transitions are occurring at temperatures slightly different to pure GMO468 due to the purity of the Myverol sample which has been shown previously to significantly alter the transition temperatures.469 Figure 6.6 c shows the PYT converted irradiation plot demonstrating the full phase 224 progression through from Q to HII occurring at 52.3 °C, through to the isotropic inverse micellar LII phase at 72.1 °C.

6.3.2 Photothermal activation of phase transitions using graphene

Photothermally induced phase transition plots of the graphene loaded lipids show the phases present throughout the irradiation time and the corresponding temperatures. Primarily these diagrams are produced to accurately show the photothermal transducing capabilities of the graphene sheets. They demonstrate a potential application in inducing phase transitions within a potential drug delivery vehicle. All blank lipid samples in the absence of graphene

283 showed no temperature increase (determined via the position of the primary peak) upon irradiation with the laser (Figures 6.7 – 6.9). Some small differences can be observed between scattering patterns, but these are attributed to in homogenous mixing.

Figure 6.7. Unloaded GME sample (a.) prior to NIR exposure and (b.) after 60 seconds of exposure.

Figure 6.8. Unloaded GMO sample (a.) prior to NIR exposure and (b.) after 60 seconds of exposure.

Figure 6.9. Unloaded PYT sample (a.) prior to NIR exposure and (b.) after 60 seconds of exposure. The transitions for each of the lipid systems can be easily observed and compared in Figure 6.10 d, which shows the phase transition as a function of irradiation time and includes the presence of significant mixed phases during a transition point. The transition points, or lack thereof in the case of GME, correlate strongly with previously published phase diagrams (collated in Figure 6.11).458, 467, 470

Figure 6.10. The Tapp as a function of NIR irradiation time for (a.) GME + water, (b.) GMO + water and (c.) PYT + water all loaded at 0.3 mg mL-1 F108-graphene. The plots show the phase identity at each point presented as blue diamonds representing the Q224 phase, red squares show the HII phase and the green triangles represent the LII phase. The grey crosses for each data set represents the graphene free lipid samples. Figure 6.10 d shows the phase of each lipid versus irradiation time (for the first 60 seconds) with any mixed phase is represented by the respective color gradient.

285

Figure 6.11. Phase diagram at 50 wt. % for each of the three lipid systems explored within this study.

6.3.3 Polarized microscopy for confirmation of phase transition boundaries

The graphene loaded lipids were also studied via polarized light microscopy with a temperature controlled stage, which allowed a secondary method of identifying the phase boundaries of the lipids (Figure 6.12). The birefringent phases of each of the lipids allowed only polarized light to pass through the lipid samples, permitting the boundaries between anisotropic phases and isotropic phases to be identified. In this study, each of the lipid systems self-assembled to form the HII phase at some point in the phase progression at 50 wt. %, allowing this anisotropic phase to act as the transition boundary marker readily identified by the well-established characterization of its fan like texture.471

224 Figure 6.12. The Q to HII phase boundaries determined via the crossed polarized light microscopy method are presented in Table 6.12 a and Graph 6.12 b and compared against the values determined through the SAXS experiments. Figure 6.12 c shows the fan-like texture characteristic of the anisotropic inverse hexagonal phase and Figure 6.12 d shows the polarized light microscopy image series corresponding to the phase transition from Q224 to HII phase for the PYT systems.

6.3.4 Reversibility of the photothermal effect and dependence on laser power

The samples were irradiated with NIR in a single position multiple times in order to determine the reproducibility and stability of the systems. As can be seen in Figure 6.13 a, 224 the full reversibility is demonstrated as the phases shift from the initial Q to HII to LII, followed by a return to Q224 with cooling to room temperature. However, the maximum temperature reached was observed to decrease slightly over the course of the cycles. We attributed this to inhomogeneity caused by the positioning of the vertical sample holder and

287 the reduced viscosity of the lipids at the high temperatures. In addition, video monitoring of the samples showed them to be vigorously mixing at the maximum temperatures during the irradiation periods. It is likely that during this event the vigorous mixing leads to inhomogeneous localized graphene concentrations resulting in a small decrease in maximum

Tapp over time.

Figure 6.13. (a.) Reversibility of photothermal heating studied using three cycles of irradiation for the 0.3 mg mL-1 graphene-loaded PYT system with 60 seconds irradiation time followed by 180 seconds of recovery. The plot also shows the phases Q224 (blue diamonds),

HII (red circles) and LII (green triangles) at the respective temperature. Figure 6.13 b shows the relationship between the 808 nm CW laser power and the temperatures achieved after 60 seconds of exposure determined via the primary scattering peak position and the calibration equations determined via the temperature ramp experiment. This experiment was performed on a 0.06 mg mL-1 graphene loaded bulk PYT sample.

The relationship between laser power and maximum temperatures (Tapp) achieved within the bulk lipid samples was also determined, as shown in Figure 6.13 b. This demonstrated the linear relationship between concentration of nanoparticles and laser power and heating effect explored previously in similar systems.332, 389

These bulk lipid samples were quite stable despite the high temperatures and vigorous mixing occurring. It is important to clarify that the cause of the slightly decreased maximum temperature observed for the repeated irradiation experiment is likely not due to any decreased ability of the graphene material to transduce the NIR as explored in Chapter 3. This is critical when comparing graphene against the current NIR photothermal agent gold nanorods which show morphology reversion under certain conditions therefore losing the advantageous absorbance magnitude and peak positioning.186

An interesting phenomenon was observed when monitoring the cooling segments of irradiation experiments, whereby a mixed phase consisting of the two bicontinuous cubic phases with Ia3̅d and Pn3̅m spacegroup was identifiable for the PYT lipid systems, even after cooling back to ambient temperature (Figure 6.14). This Ia3̅d phase structure, which was confirmed via the peak positions in Table 6.14 b, was not present upon heating and is not expected to appear unless the water concentration decreases to approximately 25 wt. %.458 This non-equilibrium phase structure has been observed previously in photothermal experiments within a similar system using gold nanorods, and was attributed to the ingress of the water being suppressed during cooling.469 These non-equilibrium systems indicate that the internal structure of dispersed LC particles is not independent of thermal history and are not in thermodynamic equilibrium. It is important to determine the cause of any unexpected phase structures to avoid difficulties in translation of such systems for controlled release applications.

289

Figure 6.14. (a.) Overlay of the PYT scattering pattern at 25 °C (blue) prior to and (red) after heating via NIR irradiation showing the presence of the non-equilibrium Ia3̅d Q230 phase. The insert shows the two SAXS patterns overlapped to highlight the Ia3̅d peaks. Table b lists the expected and observed q values for the identifiable non-Q224 peaks.

Three different lipids were selected to demonstrate the wide range of drug carrier vehicles that can be loaded with F108-graphene without disrupting the packing and act as the controllable release agent. PYT is a cosmetic ingredient which exhibits similar lyotropic phase behavior to GMO463 and importantly, does not contain an ester group imparting greater chemical stability. PYT consists of a highly branched phytantyl tail with a tri-hydroxy head group (Figure 6.1).

When considering such systems for drug delivery roles, selecting a drug carrier with a favorable release profile is a critical consideration. The three lipids selected for this study were chosen as they are well studied materials with thoroughly understood and demonstrated loading and release capabilities.455, 472 However, each of the lipids show a release profile that is not ideal for stimulated drug release applications as they transition from fast to slow release profiles with temperature.455 These systems are still highly relevant as they demonstrate both the ability to be loaded with a wide range of physiochemically varied materials and provide highly relevant platforms to demonstrate the photothermal abilities of surfactant exfoliated graphene. This challenge can be readily addressed by selecting a lipid with a more favorable release profile, such as from the lamellar phase Lα to the highly porous Q224 phase demonstrated for the lipid monoelaidin.473

Opportunities to optimize these systems are also available through subtle alterations to the composition. One method of optimizing the lipid LC drug carrier system is to manipulate the required phase transition temperature as was shown by Fong et al. in 2012. This group showed that the addition of GMO to a PYT system suppressed the transition from the Q224 to 460 HII phases. A lower phase transition temperature will subsequently reduce the required heating and therefore lower the concentration of the photothermal agent charged with the role of heating the system (or alternatively allow a reduced light flux). These optimization strategies are particularly important as they will likely hold the solution for establishing a system with an appropriate mesophase progression. Additionally, this lowers the amount of photothermal agent required, therefore minimizing toxicity and increasing the overall efficiency of target applications.

Individual phases of an LC system can present significantly different viscosities.474 The range and manipulations to the viscosity of LC systems opens up additional routes of delivery, with low viscosity materials allowing for injectable pathways,475 while the higher viscosity materials may provide topical application routes. By inducing in situ phase transformations from high viscous states to low viscous states, these limitations can be circumvented, and with a more delicate approach, specific phases can be targeted for control of release rates.

Previous studies have shown that the nanoparticles incorporated within bulk lipids are located in the aqueous phase.470 In addition to the previously reported precise location of these similarly sized materials, the length of the PEO chains of the surfactant impart a strong hydrophilic nature to the graphene sheets. It is therefore with reasonable confidence that we can assume that the sheets are most likely located primarily within the bulk aqueous phase within the lipid matrix. While the strong localization and coupling between the sheets and the lipid nanostructures is clear due to the observed thermal influence.

291

Chapter 6 Conclusions

This study has demonstrated that SALE graphene is an efficient and versatile NIR photothermal agent capable of being incorporated within bulk lipid samples of varying lipid types without disrupting packing while being in sufficiently intimate contact to provide localized heating and induce phase transitions. The reversibility of the lipid transitions allowed demonstration of the ability of the photothermal agent to withstand the high temperatures reported with no loss of efficiency. The temperatures achieved through the localized heating of graphene indicates that the photoresponsive lipid systems could be phase switched to release payloads in vivo via NIR stimulation.

293

7.0 Chapter 7 Controllable release from graphene loaded tunable α-cyclodextrin gels

7.1 Introduction

In this study, liquid exfoliated graphene sheets were incorporated within α-cyclodextrin- triblock copolymer supramolecular hydrogels prepared with a range of polyethylene oxide (PEO) and polypropylene oxide (PPO) block sizes and ratios in an attempt to control the release properties. The strong photothermal activity of graphene is expected to allow externally activated drug release from within the gels via near-infrared (NIR) irradiation. Such supramolecular hybrid hydrogels are expected to show thermoreversible changes in viscosity, a key property required for the role of both an injectable and multiple release point drug delivery depot. NIR activation of the photoresponsive pristine graphene sheets will be attempted through external irradiation in the presence of a drug model, fluorescein. In addition, a range of characterisation techniques including X-ray diffractometry (XRD) and a study into the viscoelastic properties of the gels are employed with the aim of elucidating the gelation mechanisms and providing a greater understanding of the gel properties dictating the release.

The addition of the cyclic oligosaccharide molecule α-cyclodextrin (α-CD), into a surfactant exfoliated graphene suspension, allows a supramolecular network to be formed with the resulting gel incorporating a homogenous dispersion of graphene.476-477 It has been reported by several groups that the α-CD molecule acts as a host molecule and threads onto available PEO groups to form the network, whereby the host-guest interactions occur via non-covalent interactions, such as hydrogen bonding and hydrophobic to hydrophobic interactions.478-484

The preparation of graphene suspensions in the presence of the triblock copolymer surfactants allows the graphene sheets to be an intrinsic structural component of the gels, as opposed to an added agent post carrier preparation. The adsorption of the surfactant, and the resulting extension of the PEO chain into solution, results in a graphene sheet covered in gel complexation sites. In the presence of α-CD, the inclusion complexes are formed between the α-CD and the PEO which is attached to the graphene sheets, resulting in the α-CD- surfactant-graphene hybrid gel.

As the inclusion complexes are formed, the network extends throughout the medium and large-scale gelation results.484 The resulting complexes formed are necklace-like supramolecular structures referred to as polypseudorotaxanes (Figure 7.1).483

Figure 7.1. Schematic of α-CD-Pluronic-graphene threading mechanism.

Supramolecular hydrogels are highly attractive for drug delivery applications due to the thermoreversible nature,481, 485 the ease of preparation as well as the inherent biocompatibility and biodegradability.486-488 As such, these hydrogels have been extensively studied for a range of different protein and peptide deliveries.489-492 Systems have been designed to be pH or temperature sensitive in order to allow a triggered release, whilst providing protection to the payload prior to reaching the targeted site.493 With such temperature activated systems, the challenges then shift to activating localized heating in situ. The use of NIR light and suitable transducing materials may answer these challenges.

295

As discussed in Chapter 1, the NIR region extends from ≈ 650 to 900 nm and is known to be a suitable range for biological applications as absorption by water and many other biological tissue components in this region is weak, therefore non-specific photothermal heating is minimized.494 By incorporating a NIR responsive photothermal agent within the gel, the challenges of externally activating localized heating deep within organic tissue can be addressed.

Here, the effectiveness of graphene microplates as NIR thermoresponsive agents for the promotion of triggered release from an α-CD carrier is investigated.

7.2 Materials and methods

7.2.1 Graphene preparation

Surfactant assisted liquid exfoliation (SALE) of graphene suspensions was performed as described in Chapter 2, with a key consideration into the identity of the surfactant. The triblock copolymers F108, F68, L64 and P123 were employed for the graphene suspension preparation providing a range of sizes and PEO:PPO ratios (Table 7.1).

7.2.2 α-CD gel preparation and characterization

Solutions of α-CD and surfactant were prepared and mixed to give a final α-CD concentration of 100 mg mL-1 and 2 wt. % of surfactant unless otherwise stated.

To study the gelation process a Rheometrics Dynamic Stress Rheometer (DSR) with 25 mm diameter parallel plates was used. Dynamic time sweep measurements at 10 rad s-1 frequency and 10 Pa stress at 25 °C were conducted. These conditions were selected to ensure that the yield point of any of the gels was not exceeded.

All gels were freeze-dried prior to imaging, the dry gel was then transferred onto the scanning electron microscope (SEM) stage and mounted with electrically conductive tape. SEM images were performed using a Zeiss UltraPlus FESEM with gel samples coated with platinum at 10 mA for 2 minutes prior to imaging at a voltage of 1 kV. The samples were simply freeze dried prior to imaging and a small quantity of the dried sample was transferred to the electrically conductive tape covered SEM stage. The sample was pressed into the tape and then the stage was flipped and tapped whilst upside down to remove excess sample.

7.2.3 α-CD gel drug release parameters

A typical gel sample was composed of 2 wt. % surfactant, 0.03 mg mL-1 graphene suspension and 1 mg mL-1 fluorescein as the drug model and is referred to here as α-CD-surfactant- graphene. Gel components were mixed in 5 mL glass rounded 10 mm diameter cuvette tubes 75 mm in height to a total volume of 1 mL of gel with α-CD at a concentration of 100 mg mL-1. The fluorescein was loaded into the α-CD solution and the pH was adjusted to 9.5 with NaOH, to allow the fluorescein to fully dissolve. All components (excepting the nanosheets) were dissolved prior to mixing. Immediately following mixing, the mixture was sonicated for 10 seconds, then covered with parafilm to avoid moisture loss and allowed to stand for approximately 6 hours. The control samples were prepared both with and without graphene. To remove any fluorescein from the walls of the vial, 6 hours after preparation, 4 mL of phosphate buffer solution (PBS) was added on top of the gel and left to stand for 16 hours prior to the release test. Immediately prior to the release test the PBS was discarded, the vial rinsed 3 times with PBS and a fresh 4 mL aliquot of PBS was added to the vial. The sample to be irradiated was then placed in the path of the 808 nm, 500 mW beam and the control samples were placed inside the light box approximately 30 cm away from the primary sample. The temperature within the light box was measured at both of the sample positions, to ensure all samples experienced an even ambient temperature throughout the experiment. 297

The laser was then turned on for the required period and then switched off before sample extraction. Each sample had the aqueous phase mixed prior to a 750 µL aliquot being taken, after which 750 µL of fresh PBS was added to the vial and the samples returned to the light box. Fixed extraction times (laser off times) were maintained at 4 minutes throughout all experiments.

The aliquot was then centrifuged for 2 minutes at 4000 relative centrifugal force (rcf) to remove any graphene that had migrated into the PBS layer and then 500 µL was taken and the absorbance was measured at 490 nm. All release experiments were performed at room temperature which was monitored during all experiments and shown to be 22.0 °C ± 1.5 °C. An additional confirmatory experiment was performed using a lower power (75 mW) 785 nm laser diode at 37 °C.

7.3 Results and discussion

7.3.1 α-CD-surfactant drug release studies

Fluorescein was selected as a drug model due to its size, strong absorbance (and fluorescence) signal, and well-studied and predictable properties. In order to be able to monitor the concentration released from the drug carrier gels a standard curve was established, for future conversions (Figure 7.2).

Figure 7.2. Fluorescein calibration curve (with full wavelength scans present) with the insert showing the peak absorbance at 490 nm (in arbitrary absorbance units) as a function of fluorescein concentration (mg mL-1).

The unstimulated release profiles for each of the three surfactant systems F108, F68 and L64 demonstrates that the different compositions produce subtlety different gels which influence the release rates of the drug model significantly (Figure 7.3 a). The F108 and F68 surfactant based gels, both with an 80 wt. % PEO composition but with substantially different sizes (14,600 and 8,400 Da respectively), showed very similar release profiles, indicating that the size difference between these surfactants has a minimal effect on the final product. The release rates of the two high PEO composition surfactants showed a significant decrease when compared to that of the smaller surfactant with the lower PEO composition, L64 (2,900 Da).

299

Figure 7.3. (a.) Baseline release for each α-CD-surfactant showing the different release rates for F108 (blue circles), F68 (green triangles) and L64 (red squares). Error bars show the maximum and minimums and the data presented is the average of three samples. (b.) Concentration of fluorescein released from irradiated α-CD-F108-graphene gel (red series) and irradiated α-CD-F108 gel (blue triangles) demonstrating the photothermally induced drug release. To show the photothermally induced release the blank release values are subtracted from both series.

The larger release rate of the L64 based gel reflects a more open internal network that allows the fluorescein molecules (1.09 nm maximum length495) to discharge. The irradiated α-CD- F108-graphene gel showed a dramatic increase in the concentration of fluorescein released into the PBS compared to that of the stimulated graphene free gel (Figure 7.3 b). As such the blank sample data (sample with no graphene present) was subtracted from that of the graphene loaded sample to easily observe the stimulated release values. This demonstrated that no component of the α-CD-surfactant system (other than the photothermal agent) was absorbing sufficient amounts of the NIR light to induce any additional drug release. There was no identifiable difference in the unstimulated release rates of the graphene loaded gel or that of the irradiated un-loaded graphene gel (Figure 7.4). It is a fair assumption that the drug model fluorescein could interact with graphene in the gel through π-π stacking, however as no changes were observed between graphene loaded and graphene free unstimulated gel samples, if this event is occurring it is of a negligible extent and effect. Further, it is likely that the surfactant previously adsorbed during the exfoliation process will leave little room for significant graphene to fluorescein interaction.

Figure 7.4. Release profile for (red circles) irradiated α-CD-F108 gel, (blue triangles) non- irradiated α-CD-F108 and (green squares) non-irradiated α-CD-F108-graphene.

The irradiation profile of the α-CD-F108-graphene gel versus that of the unloaded counterpart indicates that the increase in fluorescein release was entirely due to the photothermal action of the graphene sheets, resulting in localized heating. This localized heating was likely expanding the gel network or simply inducing enough of a phase change to activate the increased diffusion of fluorescein. Further, it has been demonstrated previously that hydrogen bonding plays a significant role in the interaction between PEO and α-CD moieties.480, 496 Therefore, it is fair to assume that as the hydrogen bonding is disrupted by the localized high temperature conditions upon irradiation and that the inclusion complexes are weakened inducing gel expansion and breakdown with a corresponding spike in fluorescein release.480, 497

The thermoreversible nature of the prepared gels allowed an activation switch cycle to be demonstrated by simply alternating periods of irradiation and non-irradiation sequentially, and measuring the fluorescein release (Figure 7.5 a). The release of fluorescein during the photothermally activated periods presented a marked increase, while the non-irradiated sample showed a consistent low rate of release.

301

Figure 7.5. (a.) α-CD-F108-graphene gel photothermally induced drug release with alternating laser on off periods to demonstrate the reversible nature of the gel. (b.) The relative degree of release of the irradiated graphene loaded gel and the release rate of the control.

The release rates between the activated and non-activated periods could then be directly compared by looking at the ratio of the fluorescein concentration released during each segment for the sample and blank (Figure 7.5 b). This showed an approximate 50 % increase for the release rate of the irradiated sample compared to when the sample was not irradiated. This data also indicates a slow upwards trend, likely indicating a decrease in the integrity of the gel structure with multiple heating and cooling cycles, or a slight residual increase in baseline temperature.

In order to demonstrate the versatility of graphene in the role of photothermal drug release activator a confirmatory experiment was performed using a laser with a different wavelength of 785 nm (Figure 7.6). The same trend was observed showing the flexibility provided by the strong, broadband absorbance of the graphene sheets. This is of particular value as one of the limitations observed for gold nanoparticles is the specificity of wavelengths required for activation, a challenge circumvented by a broadly absorbing material. This again reinforces the suitability of graphene in this role.

Figure 7.6. Confirmatory irradiation experiment with 2 wt. % F108 at 37 °C using a 50 mW 785 nm laser. (red circles) F108-graphene based gel show a greater concentration of fluorescein released upon stimulation via the 785 nm laser compared to that of (blue triangles) the graphene free F108 based gel.

This experiment was performed using a lower power 785 nm laser diode (75 mW), which could be positioned within an incubator at 37 °C, allowing more physiologically relevant conditions to be demonstrated. The samples all showed the same trend as those performed at ≈ 22.0 °C with the 500 mW 808 nm.

From the fluorescein drug release experiments a clear correlation between the surfactant molecular architecture (likely due to the PEO compositions) and the observed release rates is demonstrated (Figure 7.6 a). In an attempt to understand the observed differences in release profiles the gel viscoelastic properties were explored.

7.3.2 Surfactant composition influences on gel properties

The gelation kinetics of α-CD with F108, F68, L64, and P123 surfactants were explored in this study (Table 7.1). The evolution of storage (G’) and the loss moduli (G’’) were measured

303 in time sweep experiments immediately after mixing the gel components (Figure 7.7). The G’ and G” increase as the α-CD molecules thread onto the PEO chains. The point at which the elasticity modulus crosses over the loss modulus is the gel point, a property characteristic of viscoelastic gel materials.

Table 7.1. Triblock copolymers surfactant composition and properties compiled from references 216 and 217.

Surfactant Mw (da) PEO Critical micelle concentration identity (wt. %) at 25 °C (wt. %) F108 14600 80 4.5 F68 8400 80 0.3 L64 2900 40 1.6 P123 5750 30 0.03

Figure 7.7. Gelation kinetics of the α-CD gels formed with 2% of (a.) F108, (b.) L64, (c.) F68 and (d.) P123, measured by the evolution of the storage (G’) (red circles) and loss (G’’) (blue triangles) moduli in time sweep experiments at 25 °C. Note the x-axis of panels b. and d. extends further than for that of panels a. and c.

The gel point of the α-CD based gels shows a clear dependency on the surfactant chemical structure, specifically the PEO ratio in their molecular structure (Figure 7.7). For the F108 system, a large triblock copolymer (14,600 Da) with a PEO composition of 80 wt. %, the PEO groups are readily available for α-CD threading, resulting in a short gel point time. Compared to that of the L64 system, a 40 wt. % PEO composition surfactant (2,900 Da), the gel time is considerably larger than that of the F108 system. This emerging trend is clearly demonstrated when comparing both the F108 and F68 (80 wt. % PEO, 8,400 Da) based gels to the L64 and then to the P123 (30 wt. % PEO, 5750 Da) based gels showing that with decreasing PEO composition, gel formation is slowed. It is also likely that the critical micelle concentration (CMC) values, which at 2 wt. % are exceeded for the L64 and P123 surfactants,

305 may be slowing the gel formation time as the micelles must be broken before the PEO chains are available to partake in the gel formation (Table 7.1 for CMC values).

The co-assembly of the -CD and PEO moiety of the surfactants in the gelation process is dependent on the geometry of these two components. The cross-section diameter of a PEO unit is 3.1 Å,483 and the diameter of the -CD cavity is 4.7 Å,483 providing matching geometries to allow the PEO chain to thread into the -CD rings. As the height of the -CD cavity (7.9 Å) is approximately twice the contour length of the PEO repeat unit, multiple - CD molecules can thread onto the polymeric surfactant chains.483 As the surfactant with the shorter PEO chains (F68) based gels showed no significant difference in the gel point time to that of the surfactant with the longer PEO chains (F108), it appears that even this significant difference in surfactant size does not influence the efficiency of the PEO to α-CD interaction.

The gel point for each surfactant system was also explored using a simple “inversion test” where the gels were considered formed when no visible movement could be observed for 30 seconds when placed upside down (Figure 7.8).498 The results correlated well with the rheologically determined gel points, but the inversion test experiment does have a bias, inferring a delayed end point as it relies on the viscosity being sufficient to stop the movement of the entire gel matrix, which may occur well after the cross over point of the storage and loss moduli.

Figure 7.8. Gel point for each α-CD surfactant system as measured via the inversion test 25 °C.

Additionally, the inversion test highlighted that the α-CD-P123 gels did not form a gel of sufficient stability to be used for the release assays and was therefore not further explored. This is supported by the maximum measured storage modulus (Table 7.2) for the P123 system which was considerably lower than the other gels exhibited. The F68 system showed the greatest overall increase in maximum storage modulus which is likely due to the high PEO content, in addition to the smaller size of the polymers, forming a strongly connected but brittle gel.499

307

Figure 7.9. Average gelling points (G’/G’’ Crossover point) as a function of (red bars) different surfactant and (green asteriks) the PEO wt. % for each surfactant.

Table 7.2. The gel point times and the maximum storage modulus (G)’ for respective surfactant based gels.

Surfactant Gel point (min) Maximum G’ (kPa) F108 32 206 F68 28 828 L64 57 268 P123 95 16

A concentration series of F108 based α-CD gels were prepared to further explore the gelling mechanism of these systems. This again demonstrated the relationship between increased PEO concentration and that of the faster gelling time (Figure 7.9). The full viscoelastic plots can be seen in Figure 7.10. These results follow similar trends previously reported in literature whereby gelling time and release rates are faster with the increased concentration of cross-linker, or in this case the inclusion complex components (α-CD).483, 486

Figure 7.10. Full α-CD-F108 viscoelastic gel profiles as a function of F108 concentration at (a.) 0, (b.) 1, (c.) 2 and (d.) 3 wt. % 25 °C.

In addition to studying the influence of surfactant concentration, a graphene concentration series was established which showed a distinct delay in the gel formation upon the addition of graphene (Figure 7.11). This was evident with even the lowest graphene concentration explored. Both the graphene present and graphene free samples showed a strong linear dependence on the surfactant concentration. No further change in the gelling time with increasing graphene concentration was observed (Figure 7.12 b).

309

Figure 7.11. Full α-CD-F108 viscoelastic gel profiles as a function of graphene concentration at (a.) 0, (b.) 0.01, (c.) 0.02 and (d.) 0.03 mg mL-1 25 °C.

As the α-CD molecules are much smaller than all other components within this system, they would likely be the most mobile group and therefore the gelation mechanism should be viewed as the α-CD molecules threading onto the PEO chains, as opposed to the PEO chains entering the cavity of the α-CD molecules. This graphene concentration series indicates that the initial concentration of nanosheets was sufficient to sterically inhibit the mobility of the α-CD molecules within the solution whilst the large-scale network is forming. This is reasonable as the graphene sheets are by far the least mobile component within the gel matrix due to size and geometry. The full viscoelastic plots of the graphene loaded gels as a function of surfactant concentration can be seen in Figure 7.12 c.

Figure 7.12. (a.) Average α-CD-F108 gel point as a function of F108 concentration with (blue triangles) and without (red circles) graphene (0.03 mg mL-1). (b.) The gel point as a function of graphene concentration at a fixed F108 concentration of 2 wt. %, showing a distinct delay in gel formation time and then negligible further change with increasing graphene concentration. (c.) Time sweep data for 0.03 mg mL-1 graphene loaded gel with F108 concentrations of 3 (series a.), 2 (series b.) and 1 (series c.) wt. %.

SEM images were used to study the morphology and micro-structure of the gels explored within this study. The SEM images show the porous three-dimensional network of the dried α-CD gels, giving an approximate indication of pore sizes within the network, however possible artifacts associated with the freeze-drying process must not be ignored (Figure 7.13).

311

Figure 7.13. SEM images of freeze-dried α-CD gels with (a.) F108-graphene, (b.) F108, (c.) 4 wt. % F108, (d.) F68, (e.) L64 and (f.) P123 at 5,000 X magnification. All gels were prepared at 2 wt. % respective surfactant concentration unless otherwise stated.

The crystals of the α-CD gels appear as lamellar flakes with well-defined edges in the majority of the freeze-dried samples as described previously.483 Several morphological variations can be observed when the SEM images of the different surfactant based α-CD systems are compared, the presence of graphene, however, does not appear to present an identifiable change similar to previous reports (Figures 7.13 a and b).476 The lamellar flakes forming for the neat 2% F108 gel (Figure 7.13 b) appear slightly smaller and less connected than those of the 4% F108 gel (Figure 7.13 c.). Interestingly the L64 gel (Figure 7.13 e) shows a larger and less dense network than the F108 and F68 based gels, which could potentially be linked to the increased release rate (Figure 7.13 a). The P123 gel appears to be a highly dense structure with much smaller pore sizes (Figure 7.13 f). The morphologies of both the 2% F108 (Figure 7.13 b) and F68 gels (Figure 7.13 d), both with a PEO wt. % of 80, do not appear significantly different, which correlates strongly with the almost identical release rates observed for each.

In a further attempt to gain insight into the microstructure of the α-CD gels, XRD scans were performed for each of the surfactant type gels as well as the neat α-CD powder (Figure 7.14). The XRD spectra of the freeze-dried gels show the characteristic strong peak at approximately 2θ = 20.0° (d = 4.44 Å) which is assigned as the 210 reflection.481, 496 The channel-type crystalline structure resulting from the long-chain nature of the guest-molecules (PEO segments) has been well studied and the corresponding peaks have been identified in both hydrated and freeze-dried gels.481 The 210 reflection can be observed as a sharper, more well defined peak for the 4% F108 α-CD gel present at 2θ = 19.905 while the F108 and L64 surfactant gel shows a broader peak at a slightly lower diffraction angle of 2θ = 19.795. The broader peak indicated a less crystalline structure with lower surfactant compositions.

313

Figure 7.14. XRD patterns for the F108 2%, F108 4% and L64 2% α-CD gels with the 2Theta position labelled for the maximum of each series respectively.

Table 7.3. α-CD-surfactant gels XRD peak positions.

Sample ID 2θ F108 2% 19.795 F108 2% graphene 19.795 F108 4% 19.905 L64 2% 19.605 F68 2% 19.895 P123 2% 19.955

The thermoreversible nature of these -CD gels determined previously,478 was explored to demonstrate the injectability of this drug delivery depot preloaded with both the photothermal and medically active agent (Figure 7.15). A substantial decrease in viscosity was observed to occur below or near the cloud point of the employed surfactants respectively suggesting that a sufficient phase change had occurred and the α-CD molecules were disassociating from the PEO segments. The gel system reformed after a short period of time post excitation, demonstrating the recovery of the complex. The ability of the gels to reform the stable complex after heating is particularly important for a potential drug depot application allowing spikes of drug release to be controlled.

Figure 7.15. Images of an α-CD-L64 based gel demonstrating the thermoreversible nature of these systems with the gel (a.) prior to heating, (b.) immediately after heating and (c.) 5 minutes after heating.

The gel studies showed a strong correlation of gel formation time with that of the PEO composition of the surfactant and with the observed release of fluorescein from the photothermally stimulated gels. As the gel formation time and the PEO composition will influence the packing structure of the final gel product, it is a fair correlation to draw on while not entirely elucidating the mechanism behind the release profile changes.

315

Chapter 7 Conclusions

A range of triblock copolymer surfactants with varying PEO compositions were explored for the formation of a pristine graphene hybrid α-CD gel. The graphene was then externally activated through NIR irradiation resulting in highly localized heating sufficient to controllably activate the release of a drug model from within the gel network. The photothermal activation control was then demonstrated through a switching experiment showing distinct changes to the relative release rates of the drug model. The thermoreversible nature of the gels and the ability to externally activate drug release photothermally demonstrates that pristine α-CD-graphene hybrid gels could be a highly versatile injectable drug delivery depot where the drug release is controlled externally.

317

Summary chapter

The goal of this study was to explore the suitability and ability of surfactant exfoliated graphene in a range of photothermal biomedical applications. The key applications to be explored were that of graphene in the role of a photothermal agent for both cancerous cell thermal ablation and externally activated drug release.

In order to achieve the external photothermal activation, light within the near-infrared (NIR) region (650 - 900 nm) was selected due to its low absorption within biological materials which allows greater penetration. Graphene has strong absorbing potential throughout this region which allows such systems to be designed with graphene incorporated and the NIR employed to deliver energy to the graphene sheets in situ.

The production method of exfoliating graphene is highly suited for such roles as it does not introduce excessive basal plane defects and the intrinsic optical properties of graphene are maintained. By performing the exfoliation in the presence of surfactant, the sheets are well stabilised for use, have controllable surface chemistry, an improved biocompatibility and are also prepared with sites for bio-conjugation, allowing future specific targeting strategies. Additionally, the production method is low cost, can produce very high yields and is environmentally friendly in that it avoids the use of toxic solvents. This method also provides an additional level of versatility for such applications in that the aqueous environment provides a low boiling point medium, a challenge associated with solvent strategies.

The photothermal efficiency of the surfactant assisted liquid exfoliated (SALE) graphene was explored and determined to be approximately 80 %, an important factor when considering non-specific heating and designing an optimised photothermal system. Additionally, the thermal stability was explored (specific to the conditions of applications) indicating a highly suitable material for roles in the temperature range. It is the thermal stability and the versatility of the production method which positions this material as an adequate photothermal agent for such applications.

The thermal stability provides a significant advantage over gold nanorods (GNRs), a clear benchmark in the field with an absorption coefficient almost 1000 times greater than that of graphene in the NIR. The chemical and thermal stability of graphene provides a key advantage over GNRs in this regard. There is no dramatic difference in the biocompatibility properties of the two materials (when appropriately coated), which essentially brings the most relevant consideration down to a balance between flux required (primarily considering non- specific heating) and the stability of the materials. Therefore, the graphene route is more appropriate for longer term multiple release strategies, while GNRs likely still present as a more appropriate strategy for single dose applications.

Thermal ablation summary

A key focus of this study was to demonstrate directly the photothermal conversion ability that exfoliated graphene has by thermal ablation of mammalian cells. In addition to this initial goal, it was considered an appropriate central focus to explore the biological interactions this material presents, particularly any toxic properties. As all of the applications explored within the project are biomedical, it is critical to explore the effect this material will have on mammalian cells.

The initial approach therefore was to explore the cytotoxicity of the components of the SALE-graphene (the surfactant and the graphene) on a single cell line with. The goal was to not only identify the effect on biological materials, but to also establish any trends and identify the responsible agent. As such cytotoxicity curves were established for a range of surfactants employed and for L64 exfoliated graphene samples. The cytotoxicity curves for the neat surfactants followed the observed trend in literature, presenting high IC50 values indicating strong biocompatibility properties. The exception being the positively charged

319 cetyltrimethylammonium bromide (CTAB) which showed highly toxic properties in comparison to the Pluronics.

The SALE-graphene showed a lower than expected half maximal inhibitory concentration

(IC50) value indicating a moderate extent of toxicity. This was attributed to the degradation of surfactant during ultrasonication which has been showed previously to produce toxic degradation products. The L64-graphene cytotoxic values (measured with a 24 hour incubation period) however did allow a substantial working concentration for the NIR photothermal ablation experiments and improvements to the toxicity properties of the suspensions were therefore not explored. The photothermal ablation of NG108-15 mammalian cells was effectively demonstrated at this limited concentration and at an appropriate laser power providing a direct demonstration of graphene in this role.

Pickering summary

A range of emulsions were prepared with several surfactants and oils providing a solid basis to begin studies introducing exfoliated graphene into each system. The non-ionic surfactant adsorbed to the basal plane of the graphene sheets allows strong control over the surface chemistry for suspensions prepared via the SALE method and two key systems were selected to demonstrate this. A predominantly hydrophilic surfactant, F127, and a predominantly hydrophobic surfactant, P123, were used for the exfoliation and therefore the tailoring of the graphene sheet properties.

With oil in water, and water in oil emulsions established, demonstrating the versatility of this drug carrier, the attention shifted to address the effect of the presence of the graphene particles within the emulsion. The assumption was that the stability would be significantly improved by the presence of the particles, from a Pickering stability perspective. Initial experiments were simply long time scale imaging of the full emulsion to quantify the volume, however, both the stability of graphene and non-graphene containing emulsions were both extensive (months), making the external environmental factors significant, convoluting the data. An alternative, shorter time scale analysis was selected, imaging small volumes of the emulsion droplets under an optical microscope. The droplet sizes were then measured via ImageJ, and it was determined that the data was highly robust, with large numbers of images and samples providing consistent data. The larger droplet sizes observed for the graphene free samples indicated a greater extent of coalescence, indicating the improved stability provided by the graphene sheets.

With the influence of the graphene on emulsion stability assessed, the focus shifted to NIR photothermal activation of the graphene, in order to controllably break (release) the emulsion (the drug). Sufficient volumes of emulsion were prepared (both graphene-loaded and graphene-free) and both an 808 nm and a 980 nm laser were directed at the emulsion for short periods. The irradiation event was monitored via transmission measurements (power meter behind the cuvette) as well as with video footage. Both the transmission data and the video footage indicated that the graphene-loaded emulsion showed localized breakage upon NIR irradiation, resulting in a coring effect through the emulsion.

To summarise, these graphene-loaded emulsions were shown to be highly stable, as well as being capable of loading a wide physiochemical range of drugs in an oil or a water internal phase. This system was also demonstrated to be suitable for NIR activation via the graphene sheets showing completely controllable, externally activated breakage of the drug carrier.

Lipids summary

Three lipids were selected for this project to demonstrate the wide range of systems the SALE graphene can be incorporated within, in addition to demonstrating the versatility of the lipid systems for such drug release roles. Glyceryl monooleyl ether (GME), glyceryl monooleate (GMO) and phytantriol (PYT) were selected as their respective phase transitions within the

321 temperature range relevant for this study are well known and the release rates have also been explored.

Synchrotron small angle X-ray scattering (SAXS) was employed to identify the phases present for each of the lipids with and without graphene, as well as over a temperature range of 20 °C to 70 °C. The presence of graphene, at all concentrations explored, did not disrupt the packing configuration allowing incorporation of the nanosheets within the lipid dispersions. Indeed, the presence of the F127, specifically selected for the graphene preparation for this reason, was shown to aid in a more uniform phase of the lipid dispersion.

A calibration curve of the lattice parameter, as determined by the primary Bragg peak q position, was established as a function of temperature. This allowed graphene loaded samples to have the apparent temperature (Tapp) determined during the NIR irradiation via the SAXS patterns. Therefore, for an irradiation experiment, not only the bulk temperature was determined, but also the lattice spacing and the presence of mixed phases, two factors critical to understand for any system posited as a drug carrier.

Finally, the reversibility of the lipids allowed for an in situ reversibility experiment demonstrating the controllable, multi-stage release ability of both the carrier, and the photothermal activator. Drug release experiments were not performed however as the release profiles for these lipids is unfavourable for photothermal activation. They do however demonstrate clearly that an alternative lipid, such as monoelaidin, could be employed in precisely this role.

The pristine graphene sheets did not disrupt the packing of the liquid crystals, while being in sufficiently intimate contact to provide localized heating and induce phase transitions. The phase progressions induced through heating using NIR irradiation of the entrained graphene particles within the bulk liquid crystal were studied using SAXS and confirmed using polarized optical microscopy. Increases in apparent temperature experienced by the matrix of up to 50 °C were observed by establishing a SAXS versus bulk temperature calibration curve, allowing in situ measurements. The studies demonstrate the potential for use of graphene as a photothermal actuator across a range of lipid based systems of interest in controlled drug delivery.

α-cyclodextrin summary

A range of the non-ionic triblock copolymers were employed to prepare varying types of α- cyclodextrin (α-CD) gels with a strategy of exploiting the established polyethylene oxide (PEO) α-CD inclusion complex to form the gel. Within this project, there was a strong focus to incorporate the exfoliated graphene sheets as an intrinsic structural component of the drug carrier, not to simply add the photoresponsive agent post carrier preparation. With this in mind, the surfactants composition was carefully selected for exfoliation, providing a range of graphene based suspensions with varying PEO components adsorbed (as part of the PEO:PPO:PEO triblock).

It was found that sufficient (and significant) quantities of exfoliated graphene could be easily incorporated within the gel network with no detriment and so the attention turned to attempting to demonstrate drug release from these gels. A drug model was selected (fluorescein) and loaded into the gel during preparation resulting in a gel composed of α-CD- surfactant-graphene as well as fluorescein. Baseline release and NIR activated release experiments were performed exploring the basic release profiles and demonstrating the NIR activation possible along with a controllable step-release possible due to the thermoreversible nature of the α-CD-surfactant gels.

A relationship between PEO composition and drug release rate was identified and as such efforts turned to identify the cause of these variations. Scanning electron microscopy (SEM), X-ray diffraction (XRD) and primarily the viscoelastic studies were employed to study the gels. The XRD and SEM supported the higher release rate of the lower PEO composition gel via a decreased crystallinity and increased pore size respectively. The viscoelastic studies

323 added a further correlation of gelling time to that of the drug release rates and PEO composition, indicating that with faster gelling (with higher PEO composition) a tighter gel was formed resulting in a slower rate.

A highly stable and biocompatible drug carrier was demonstrated with controllable step release activatable via external NIR activation in tandem with the presence of the intrinsic graphene component.

Final comment

SALE-graphene was demonstrated as suitable material for the thermal ablation of the NG108-15 cells via NIR irradiation. Indicating that not only does the broad absorbance of graphene provide versatility in wavelength selection, but that the extent of the absorbance is sufficient to carry out such roles.

The morphology of the SALE graphene along with the easily controllable surface chemistry, through surfactant selection, allowed the graphene microsheets to be incorporated within three drug delivery systems. The presence of the graphene was either added with no negative effects (lipids nanostructures), included as an additional stabilizing agent (Pickering emulsions), or present as an intrinsic component of the system (α-CD-surfactant-graphene gels). In all scenarios external NIR irradiation of the graphene microsheets was successfully employed to trigger a drug release, or drug release analogous event

References

1. Balandin, A. A.; Ghosh, S.; Bao, W.; Calizo, I.; Teweldebrhan, D.; Miao, F.; Lau, C. N., Superior thermal conductivity of single-layer graphene. Nano Lett. 2008, 8 (3), 902-907. 2. Ghosh, S.; Calizo, I.; Teweldebrhan, D.; Pokatilov, E. P.; Nika, D. L.; Balandin, A. A.; Bao, W.; Miao, F.; Lau, C. N., Extremely high thermal conductivity of graphene: Prospects for thermal management applications in nanoelectronic circuits. Appl. Phys. Lett. 2008, 92 (15). 3. Pop, E.; Varshney, V.; Roy, A. K., Thermal properties of graphene: Fundamentals and applications. MRS Bull. 2012, 37 (12), 1273-1281. 4. Yan, Z.; Nika, D. L.; Balandin, A. A., Thermal properties of graphene and few layer graphene: Applications in electronics. IET Circuits, Devices and Systems 2015, 9 (1), 4-12. 5. Shahil, K. M. F.; Balandin, A. A., Thermal properties of graphene and multilayer graphene: Applications in thermal interface materials. Solid State Commun. 2012, 152 (15), 1331-1340. 6. Lee, C.; Wei, X.; Kysar, J. W.; Hone, J., Measurement of the Elastic Properties and Intrinsic Strength of Monolayer Graphene. Science 2008, 321 (5887), 385-388. 7. Geim, A. K.; Novoselov, K. S., The rise of graphene. Nat. Mater. 2007, 6 (3), 183- 191. 8. Nair, R. R.; Blake, P.; Grigorenko, A. N.; Novoselov, K. S.; Booth, T. J.; Stauber, T.; Peres, N. M. R.; Geim, A. K., Fine Structure Constant Defines Visual Transparency of Graphene. Science 2008, 320 (5881), 1308-1308. 9. Zhu, S. E.; Yuan, S.; Janssen, G. C. A. M., Optical transmittance of multilayer graphene. EPL 2014, 108 (1). 10. Bolotin, K. I.; Sikes, K. J.; Jiang, Z.; Klima, M.; Fudenberg, G.; Hone, J.; Kim, P.; Stormer, H. L., Ultrahigh electron mobility in suspended graphene. Solid State Commun. 2008, 146 (9–10), 351-355. 11. Du, X.; Skachko, I.; Barker, A.; Andrei, E. Y., Approaching ballistic transport in suspended graphene. Nat. Nanotechnol. 2008, 3 (8), 491-495. 12. Chen, J. H.; Jang, C.; Xiao, S.; Ishigami, M.; Fuhrer, M. S., Intrinsic and extrinsic performance limits of graphene devices on SiO 2. Nat. Nanotechnol. 2008, 3 (4), 206-209. 13. Chen, S.; Wu, Q.; Mishra, C.; Kang, J.; Zhang, H.; Cho, K.; Cai, W.; Balandin, A. A.; Ruoff, R. S., Thermal conductivity of isotopically modified graphene. Nat. Mater. 2012, 11 (3), 203-207. 14. Semenoff, G. W., Condensed-Matter simulation of a three-Dimensional anomaly. Phys. Rev. Lett. 1984, 53 (26), 2449-2452. 15. Fradkin, E., Critical behavior of disordered degenerate semiconductors. II. Spectrum and transport properties in mean-field theory. Physical Review B 1986, 33 (5), 3263-3268. 16. Novoselov, K. S.; Geim, A. K.; Morozov, S. V.; Jiang, D.; Katsnelson, M. I.; Grigorieva, I. V.; Dubonos, S. V.; Firsov, A. A., Two-dimensional gas of massless Dirac fermions in graphene. Nature 2005, 438 (7065), 197-200. 17. H.O., P., Handbook of carbon, graphite, diamond and fullerenes. Properties, processing and applications. Elsevier: 1995; p 417. 18. Anthony, T. R.; Banholzer, W. F.; Fleischer, J. F.; Wei, L.; Kuo, P. K.; Thomas, R. L.; Pryor, R. W., Thermal diffusivity of isotopically enriched $^{12}\mathrm{C}$ diamond. Physical Review B 1990, 42 (2), 1104-1111.

325

19. Klemens, P. G.; Pedraza, D. F., Thermal conductivity of graphite in the basal plane. Carbon 1994, 32 (4), 735-741. 20. Petroski, H., The Pencil. Alfred A. Knoff: United States, 1989. 21. Al-Jishi, R.; Dresselhaus, G., Lattice-dynamical model for graphite. Physical Review B 1982, 26 (8), 4514-4522. 22. Deng, S.; Berry, V., Wrinkled, rippled and crumpled graphene: an overview of formation mechanism, electronic properties, and applications. Mater. Today 2016, 19 (4), 197-212. 23. Fasolino, A.; Los, J. H.; Katsnelson, M. I., Intrinsic ripples in graphene. Nat Mater 2007, 6 (11), 858-861. 24. Xu, K.; Cao, P.; Heath, J. R., Scanning Tunneling Microscopy Characterization of the Electrical Properties of Wrinkles in Exfoliated Graphene Monolayers. Nano Lett. 2009, 9 (12), 4446-4451. 25. Ishigami, M.; Chen, J. H.; Cullen, W. G.; Fuhrer, M. S.; Williams, E. D., Atomic structure of graphene on SiO 2. Nano Lett. 2007, 7 (6), 1643-1648. 26. Wallace, P. R., The band theory of graphite. Physical Review 1947, 71 (9), 622-634. 27. McClure, J. W., Diamagnetism of graphite. Physical Review 1956, 104 (3), 666-671. 28. Slonczewski, J. C.; Weiss, P. R., Band structure of graphite. Physical Review 1958, 109 (2), 272-279. 29. Kroto, H. W.; Heath, J. R.; O'Brien, S. C.; Curl, R. F.; Smalley, R. E., C60: Buckminsterfullerene. Nature 1985, 318 (6042), 162-163. 30. Kim, P.; Shi, L.; Majumdar, A.; McEuen, P. L., Thermal transport measurements of individual multiwalled nanotubes. Phys. Rev. Lett. 2001, 87 (21), 2155021-2155024. 31. Gao, G.; Çaǧin, T.; Goddard Iii, W. A., Energetics, structure, mechanical and vibrational properties of single-walled carbon nanotubes. Nanotechnology 1998, 9 (3), 184- 191. 32. Bandaru, P. R., Electrical properties and applications of carbon nanotube structures. Journal of Nanoscience and Nanotechnology 2007, 7 (4-5), 1239-1267. 33. Baughman, R. H.; Zakhidov, A. A.; De Heer, W. A., Carbon nanotubes - The route toward applications. Science 2002, 297 (5582), 787-792. 34. Kociak, M.; Guéron, S.; Reulet, B.; Khodos, I. I.; Gorbatov, Y. B.; Volkov, V. T.; Vaccarini, L.; Bouchiat, H., Superconductivity in ropes of single-walled carbon nanotubes. Phys. Rev. Lett. 2001, 86 (11), 2416-2419. 35. Nagashio, K.; Nishimura, T.; Kita, K.; Toriumi, A., Mobility variations in mono- and multi-layer graphene films. Applied Physics Express 2009, 2 (2). 36. Blake, P.; Hill, E. W.; Castro Neto, A. H.; Novoselov, K. S.; Jiang, D.; Yang, R.; Booth, T. J.; Geim, A. K., Making graphene visible. Appl. Phys. Lett. 2007, 91 (6). 37. Castro Neto, A. H.; Guinea, F.; Peres, N. M. R.; Novoselov, K. S.; Geim, A. K., The electronic properties of graphene. Rev. Mod. Phys. 2009, 81 (1), 109-162. 38. Shearer, C. J.; Slattery, A. D.; Stapleton, A. J.; Shapter, J. G.; Gibson, C. T., Accurate thickness measurement of graphene. Nanotechnology 2016, 27 (12). 39. Shioyama, H., The interactions of two chemical species in the interlayer spacing of graphite. Synth. Met. 2000, 114 (1), 1-15. 40. Bullett, D. W., A chemical pseudopotential approach to covalent bonding. I. Journal of Physics C: Solid State Physics 1975, 8 (17), 2695-2706. 41. Kurasch, S.; Meyer, J. C.; Künzel, D.; Groß, A.; Kaiser, U., Simulation of bonding effects in HRTEM images of light element materials. Beilstein Journal of Nanotechnology 2011, 2, 394-404. 42. Meyer, J. C.; Kurasch, S.; Park, H. J.; Skakalova, V.; Künzel, D.; Groß, A.; Chuvilin, A.; Algara-Siller, G.; Roth, S.; Iwasaki, T.; Starke, U.; Smet, J. H.; Kaiser, U., Experimental analysis of charge redistribution due to chemical bonding by high-resolution transmission electron microscopy. Nat Mater 2011, 10 (3), 209-215. 43. Das Sarma, S.; Adam, S.; Hwang, E. H.; Rossi, E., Electronic transport in two- dimensional graphene. Rev. Mod. Phys. 2011, 83 (2), 407-470. 44. Misra, P. K., Physics of Condensed Matter. Elsevier: United States, 2012. 45. Lu, G.; Yu, K.; Wen, Z.; Chen, J., Semiconducting graphene: Converting graphene from semimetal to semiconductor. Nanoscale 2013, 5 (4), 1353-1368. 46. Lee, Y.; Myhro, K.; Tran, D.; Gilgren, N.; Velasco Jr, J.; Bao, W.; Deo, M.; Lau, C. N. In Band gap and correlated phenomena in bilayer and trilayer graphene, Proceedings of SPIE - The International Society for Optical Engineering, 2013. 47. Jiao, L.; Zhang, L.; Wang, X.; Diankov, G.; Dai, H., Narrow graphene nanoribbons from carbon nanotubes. Nature 2009, 458 (7240), 877-880. 48. Ni, Z. H.; Yu, T.; Lu, Y. H.; Wang, Y. Y.; Feng, Y. P.; Shen, Z. X., Uniaxial Strain on Graphene: Raman Spectroscopy Study and Band-Gap Opening. ACS Nano 2008, 2 (11), 2301-2305. 49. Novoselov, K. S.; Geim, A. K.; Morozov, S. V.; Jiang, D.; Zhang, Y.; Dubonos, S. V.; Grigorieva, I. V.; Firsov, A. A., Electric field in atomically thin carbon films. Science 2004, 306 (5696), 666-669. 50. Agnoli, S.; Favaro, M., Doping graphene with boron: a review of synthesis methods, physicochemical characterization, and emerging applications. Journal of Materials Chemistry A 2016, 4 (14), 5002-5025. 51. Huang, X.; Yin, Z.; Wu, S.; Qi, X.; He, Q.; Zhang, Q.; Yan, Q.; Boey, F.; Zhang, H., Graphene-based materials: Synthesis, characterization, properties, and applications. Small 2011, 7 (14), 1876-1902. 52. Wang, H.; Maiyalagan, T.; Wang, X., Review on recent progress in nitrogen-doped graphene: Synthesis, characterization, and its potential applications. ACS Catalysis 2012, 2 (5), 781-794. 53. Avouris, P., Graphene: Electronic and photonic properties and devices. Nano Lett. 2010, 10 (11), 4285-4294. 54. Gierz, I.; Riedl, C.; Starke, U.; Ast, C. R.; Kern, K., Atomic hole doping of graphene. Nano Lett. 2008, 8 (12), 4603-4607. 55. Mak, K. F.; Ju, L.; Wang, F.; Heinz, T. F., Optical spectroscopy of graphene: From the far infrared to the ultraviolet. Solid State Commun. 2012, 152 (15), 1341-1349. 56. Mak, K. F.; Shan, J.; Heinz, T. F., Electronic structure of few layer graphene: Experimental demonstration of strong dependence on stacking sequence. Phys. Rev. Lett. 2010, 104 (17). 57. Craciun, M. F.; Russo, S.; Yamamoto, M.; Oostinga, J. B.; Morpurgo, A. F.; Tarucha, S., Trilayer graphene is a semimetal with a gate-tunable band overlap. Nat. Nanotechnol. 2009, 4 (6), 383-388. 58. Lau, C. N.; Bao, W.; Velasco Jr, J., Properties of suspended graphene membranes. Mater. Today 2012, 15 (6), 238-245. 59. McCann, E., Asymmetry gap in the electronic band structure of bilayer graphene. Physical Review B - Condensed Matter and Materials Physics 2006, 74 (16). 60. McCann, E.; Fal'ko, V. I., Landau-level degeneracy and quantum hall effect in a graphite bilayer. Phys. Rev. Lett. 2006, 96 (8).

327

61. Quhe, R.; Ma, J.; Zeng, Z.; Tang, K.; Zheng, J.; Wang, Y.; Ni, Z.; Wang, L.; Gao, Z.; Shi, J.; Lu, J., Tunable band gap in few layer graphene by surface adsorption. Sci. Rep. 2013, 3. 62. Lu, Y. H.; Chen, W.; Feng, Y. P.; He, P. M., Tuning the Electronic Structure of Graphene by an Organic Molecule. The Journal of Physical Chemistry B 2009, 113 (1), 2-5. 63. Yong-Hui, Z.; Kai-Ge, Z.; Ke-Feng, X.; Jing, Z.; Hao-Li, Z.; Yong, P., Tuning the electronic structure and transport properties of graphene by noncovalent functionalization: effects of organic donor, acceptor and metal atoms. Nanotechnology 2010, 21 (6), 065201. 64. Wick, P.; Louw-Gaume, A. E.; Kucki, M.; Krug, H. F.; Kostarelos, K.; Fadeel, B.; Dawson, K. A.; Salvati, A.; Vázquez, E.; Ballerini, L.; Tretiach, M.; Benfenati, F.; Flahaut, E.; Gauthier, L.; Prato, M.; Bianco, A., Classification Framework for Graphene-Based Materials. Angew. Chem. Int. Ed. 2014, 53 (30), 7714-7718. 65. Weissleder, R., A clearer vision for in vivo imaging. Nat. Biotechnol. 2001, 19 (4), 316-317. 66. Robinson, J. T.; Tabakman, S. M.; Liang, Y.; Wang, H.; Sanchez Casalongue, H.; Vinh, D.; Dai, H., Ultrasmall reduced graphene oxide with high near-infrared absorbance for photothermal therapy. J. Am. Chem. Soc. 2011, 133 (17), 6825-6831. 67. Michael Ference, H. B. L., Reginald J. Stephenson,, Analytical Experimental Physics, Third Revised Edition. The University of Chicago Press: 1943. 68. Mingo, N.; Broido, D. A., Carbon nanotube ballistic thermal conductance and its limits. Phys. Rev. Lett. 2005, 95 (9). 69. Nika, D. L.; Pokatilov, E. P.; Askerov, A. S.; Balandin, A. A., Phonon thermal conduction in graphene: Role of Umklapp and edge roughness scattering. Physical Review B - Condensed Matter and Materials Physics 2009, 79 (15). 70. Popov, V. N., Low-temperature specific heat of nanotube systems. Physical Review B - Condensed Matter and Materials Physics 2002, 66 (15), 1534081-1534084. 71. Muñoz, E.; Lu, J.; Yakobson, B. I., Ballistic thermal conductance of graphene ribbons. Nano Lett. 2010, 10 (5), 1652-1656. 72. Ferrari, A. C.; Basko, D. M., Raman spectroscopy as a versatile tool for studying the properties of graphene. Nat. Nanotechnol. 2013, 8 (4), 235-246. 73. Jiang, J. W.; Wang, B. S.; Wang, J. S.; Park, H. S., A review on the flexural mode of graphene: Lattice dynamics, thermal conduction, thermal expansion, elasticity and nanomechanical resonance. Journal of Physics Condensed Matter 2015, 27 (8). 74. Ghosh, S.; Bao, W.; Nika, D. L.; Subrina, S.; Pokatilov, E. P.; Lau, C. N.; Balandin, A. A., Dimensional crossover of thermal transport in few layer graphene. Nat. Mater. 2010, 9 (7), 555-558. 75. Shi, L., Thermal and thermoelectric transport in nanostructures and low-dimensional systems. Nanoscale and Microscale Thermophysical Engineering 2012, 16 (2), 79-116. 76. Seol, J. H.; Jo, I.; Moore, A. L.; Lindsay, L.; Aitken, Z. H.; Pettes, M. T.; Li, X.; Yao, Z.; Huang, R.; Broido, D.; Mingo, N.; Ruoff, R. S.; Shi, L., Two-dimensional phonon transport in supported graphene. Science 2010, 328 (5975), 213-216. 77. Chen, J. H.; Jang, C.; Ishigami, M.; Xiao, S.; Cullen, W. G.; Williams, E. D.; Fuhrer, M. S., Diffusive charge transport in graphene on SiO2. Solid State Commun. 2009, 149 (27- 28), 1080-1086. 78. Ni, Z. H.; Ponomarenko, L. A.; Nair, R. R.; Yang, R.; Anissimova, S.; Grigorieva, I. V.; Schedin, F.; Blake, P.; Shen, Z. X.; Hill, E. H.; Novoselov, K. S.; Geim, A. K., On resonant scatterers as a factor limiting carrier mobility in graphene. Nano Lett. 2010, 10 (10), 3868-3872. 79. Fratini, S.; Guinea, F., Substrate-limited electron dynamics in graphene. Physical Review B - Condensed Matter and Materials Physics 2008, 77 (19). 80. Perebeinos, V.; Avouris, P., Inelastic scattering and current saturation in graphene. Physical Review B - Condensed Matter and Materials Physics 2010, 81 (19). 81. Dean, C. R.; Young, A. F.; Meric, I.; Lee, C.; Wang, L.; Sorgenfrei, S.; Watanabe, K.; Taniguchi, T.; Kim, P.; Shepard, K. L.; Hone, J., Boron nitride substrates for high-quality graphene electronics. Nat. Nanotechnol. 2010, 5 (10), 722-726. 82. Fang, T.; Konar, A.; Xing, H.; Jena, D., Carrier statistics and quantum capacitance of graphene sheets and ribbons. Appl. Phys. Lett. 2007, 91 (9). 83. Hrostowski, H. J.; Morin, F. J.; Geballe, T. H.; Wheatley, G. H., Hall effect and conductivity of InSb. Physical Review 1955, 100 (6), 1672-1676. 84. Dürkop, T.; Getty, S. A.; Cobas, E.; Fuhrer, M. S., Extraordinary Mobility in Semiconducting Carbon Nanotubes. Nano Lett. 2004, 4 (1), 35-39. 85. Sugihara, K.; Kawamura, K.; Tsuzuku, T., Temperature dependence of the average mobility in graphite. J. Phys. Soc. Jpn. 1979, 47 (4), 1210-1215. 86. Dresselhaus, M. S.; Dresselhaus, G., Intercalation compounds of graphite. Advances in Physics 1981, 30 (2), 139-326. 87. Pfeiffer, L.; West, K. W.; Stormer, H. L.; Baldwin, K. W., Electron mobilities exceeding 107 cm2/V s in modulation-doped GaAs. Appl. Phys. Lett. 1989, 55 (18), 1888- 1890. 88. Gómez-Navarro, C.; Weitz, R. T.; Bittner, A. M.; Scolari, M.; Mews, A.; Burghard, M.; Kern, K., Electronic transport properties of individual chemically reduced graphene oxide sheets. Nano Lett. 2007, 7 (11), 3499-3503. 89. Madhuri Sharon, M. S., Graphene. United States, 2015. 90. Frank, I. W.; Tanenbaum, D. M.; Van Der Zande, A. M.; McEuen, P. L., Mechanical properties of suspended graphene sheets. Journal of Vacuum Science and Technology B: Microelectronics and Nanometer Structures 2007, 25 (6), 2558-2561. 91. Lee, G. H.; Cooper, R. C.; An, S. J.; Lee, S.; Van Der Zande, A.; Petrone, N.; Hammerberg, A. G.; Lee, C.; Crawford, B.; Oliver, W.; Kysar, J. W.; Hone, J., High-strength chemical-vapor-deposited graphene and grain boundaries. Science 2013, 340 (6136), 1074- 1076. 92. Lopez-Polin, G.; Gomez-Navarro, C.; Parente, V.; Guinea, F.; Katsnelson, M. I.; Perez-Murano, F.; Gomez-Herrero, J., Increasing the elastic modulus of graphene by controlled defect creation. Nat Phys 2015, 11 (1), 26-31. 93. Mahendran, M., The modulus of elasticity of steel. In Thirteenth International Specialty Conference on Cold-Formed Steel Structures, St. Louis, Missouri U.S.A., 1996. 94. Schabel, M. C.; Martins, J. L., Energetics of interplanar binding in graphite. Physical Review B 1992, 46 (11), 7185-7188. 95. Bullett, D. W., Chemical pseudopotential approach to covalent bonding. II. Bond lengths and bond energies in diamond, silicon and graphite. Journal of Physics C: Solid State Physics 1975, 8 (17), 2707. 96. Bullett, D. W., A chemical pseudopotential approach to covalent bonding. I. Journal of Physics C: Solid State Physics 1975, 8 (17), 2695. 97. Blakslee, O. L.; Proctor, D. G.; Seldin, E. J.; Spence, G. B.; Weng, T., Elastic constants of compression-annealed pyrolytic graphite. J. Appl. Phys. 1970, 41 (8), 3373- 3382. 98. Poot, M.; Van Der Zant, H. S. J., Nanomechanical properties of few layer graphene membranes. Appl. Phys. Lett. 2008, 92 (6).

329

99. Politano, A.; Chiarello, G., Probing the Young’s modulus and Poisson’s ratio in graphene/metal interfaces and graphite: a comparative study. Nano Research 2015, 8 (6), 1847-1856. 100. Jing, N.; Xue, Q.; Ling, C.; Shan, M.; Zhang, T.; Zhou, X.; Jiao, Z., Effect of defects on Young's modulus of graphene sheets: A molecular dynamics simulation. RSC Advances 2012, 2 (24), 9124-9129. 101. Gómez-Navarro, C.; Burghard, M.; Kern, K., Elastic Properties of Chemically Derived Single Graphene Sheets. Nano Lett. 2008, 8 (7), 2045-2049. 102. Novoselov, K. S.; Jiang, D.; Schedin, F.; Booth, T. J.; Khotkevich, V. V.; Morozov, S. V.; Geim, A. K., Two-dimensional atomic crystals. Proc. Natl. Acad. Sci. U. S. A. 2005, 102 (30), 10451-10453. 103. Kim, K. S.; Zhao, Y.; Jang, H.; Lee, S. Y.; Kim, J. M.; Kim, K. S.; Ahn, J. H.; Kim, P.; Choi, J. Y.; Hong, B. H., Large-scale pattern growth of graphene films for stretchable transparent electrodes. Nature 2009, 457 (7230), 706-710. 104. Novoselov, K. S.; Fal'Ko, V. I.; Colombo, L.; Gellert, P. R.; Schwab, M. G.; Kim, K., A roadmap for graphene. Nature 2012, 490 (7419), 192-200. 105. Li, D.; Müller, M. B.; Gilje, S.; Kaner, R. B.; Wallace, G. G., Processable aqueous dispersions of graphene nanosheets. Nat. Nanotechnol. 2008, 3 (2), 101-105. 106. Eda, G.; Fanchini, G.; Chhowalla, M., Large-area ultrathin films of reduced graphene oxide as a transparent and flexible electronic material. Nat. Nanotechnol. 2008, 3 (5), 270- 274. 107. Hernandez, Y.; Nicolosi, V.; Lotya, M.; Blighe, F. M.; Sun, Z.; De, S.; McGovern, I. T.; Holland, B.; Byrne, M.; Gun'ko, Y. K.; Boland, J. J.; Niraj, P.; Duesberg, G.; Krishnamurthy, S.; Goodhue, R.; Hutchison, J.; Scardaci, V.; Ferrari, A. C.; Coleman, J. N., High-yield production of graphene by liquid-phase exfoliation of graphite. Nat. Nanotechnol. 2008, 3 (9), 563-568. 108. Lotya, M.; Hernandez, Y.; King, P. J.; Smith, R. J.; Nicolosi, V.; Karlsson, L. S.; Blighe, F. M.; De, S.; Wang, Z.; McGovern, I. T.; Duesberg, G. S.; Coleman, J. N., Liquid Phase Production of Graphene by Exfoliation of Graphite in Surfactant/Water Solutions. J. Am. Chem. Soc. 2009, 131 (10), 3611-3620. 109. Notley, S. M., Highly concentrated aqueous suspensions of graphene through ultrasonic exfoliation with continuous surfactant addition. Langmuir 2012, 28 (40), 14110- 14113. 110. Wassei, J. K.; Kaner, R. B., Graphene, a promising transparent conductor. Mater. Today 2010, 13 (3), 52-59. 111. Bae, S.; Kim, H.; Lee, Y.; Xu, X.; Park, J. S.; Zheng, Y.; Balakrishnan, J.; Lei, T.; Ri Kim, H.; Song, Y. I.; Kim, Y. J.; Kim, K. S.; Özyilmaz, B.; Ahn, J. H.; Hong, B. H.; Iijima, S., Roll-to-roll production of 30-inch graphene films for transparent electrodes. Nat. Nanotechnol. 2010, 5 (8), 574-578. 112. Li, X.; Cai, W.; An, J.; Kim, S.; Nah, J.; Yang, D.; Piner, R.; Velamakanni, A.; Jung, I.; Tutuc, E.; Banerjee, S. K.; Colombo, L.; Ruoff, R. S., Large-area synthesis of high-quality and uniform graphene films on copper foils. Science 2009, 324 (5932), 1312-1314. 113. Kholin, N. A.; Rut'kov, E. V.; Tontegode, A. Y., The nature of the adsorption bond between graphite islands and iridium surface. Surf. Sci. 1984, 139 (1), 155-172. 114. Himpsel, F. J.; Christmann, K.; Heimann, P.; Eastman, D. E.; Feibelman, P. J., Adsorbate band dispersions for C on Ru(0001). Surf. Sci. 1982, 115 (3), L159-L164. 115. Eizenberg, M.; Blakely, J. M., Carbon monolayer phase condensation on Ni(111). Surf. Sci. 1979, 82 (1), 228-236. 116. Hamilton, J. C.; Blakely, J. M., Carbon segregation to single crystal surfaces of Pt, Pd and Co. Surf. Sci. 1980, 91 (1), 199-217. 117. Coraux, J.; N'Diaye, A. T.; Busse, C.; Michely, T., Structural coherency of graphene on Ir(111). Nano Lett. 2008, 8 (2), 565-570. 118. Mattevi, C.; Kim, H.; Chhowalla, M., A review of chemical vapour deposition of graphene on copper. J. Mater. Chem. 2011, 21 (10), 3324-3334. 119. Rashid, H. U.; Yu, K.; Umar, M. N.; Anjum, M. N.; Khan, K.; Ahmad, N.; Jan, M. T., Catalyst role in Chemical Vapor Deposition (CVD) process: A review. Reviews on Advanced Materials Science 2015, 40 (3), 235-248. 120. Abrahamson, J., Graphite sublimation temperatures, carbon arcs and crystallite erosion. Carbon 1974, 12 (2), 111-141. 121. László, P. B.; Philippe, L., Grain boundaries in graphene grown by chemical vapor deposition. New Journal of Physics 2013, 15 (3), 035024. 122. Bonaccorso, F.; Lombardo, A.; Hasan, T.; Sun, Z.; Colombo, L.; Ferrari, A. C., Production and processing of graphene and 2d crystals. Mater. Today 2012, 15 (12), 564- 589. 123. Schultz, R. A.; Jensen, M. C.; Bradt, R. C., Single crystal cleavage of brittle materials. International Journal of Fracture 1994, 65 (4), 291-312. 124. Geim, A. K., Graphene: Status and prospects. Science 2009, 324 (5934), 1530-1534. 125. Hasan, T.; Torrisi, F.; Sun, Z.; Popa, D.; Nicolosi, V.; Privitera, G.; Bonaccorso, F.; Ferrari, A. C., Solution-phase exfoliation of graphite for ultrafast photonics. Physica Status Solidi (B) Basic Research 2010, 247 (11-12), 2953-2957. 126. Green, A. A.; Hersam, M. C., Solution phase production of graphene with controlled thickness via density differentiation. Nano Lett. 2009, 9 (12), 4031-4036. 127. Maragó, O. M.; Bonaccorso, F.; Saija, R.; Privitera, G.; Gucciardi, P. G.; Iatì, M. A.; Calogero, G.; Jones, P. H.; Borghese, F.; Denti, P.; Nicolosi, V.; Ferrari, A. C., Brownian motion of graphene. ACS Nano 2010, 4 (12), 7515-7523. 128. Gokus, T.; Nair, R. R.; Bonetti, A.; Böhmler, M.; Lombardo, A.; Novoselov, K. S.; Geim, A. K.; Ferrari, A. C.; Hartschuh, A., Making graphene luminescent by oxygen plasma treatment. ACS Nano 2009, 3 (12), 3963-3968. 129. Sun, X.; Liu, Z.; Welsher, K.; Robinson, J. T.; Goodwin, A.; Zaric, S.; Dai, H., Nano- Graphene Oxide for Cellular Imaging and Drug Delivery. Nano research 2008, 1 (3), 203- 212. 130. Becerril, H. A.; Mao, J.; Liu, Z.; Stoltenberg, R. M.; Bao, Z.; Chen, Y., Evaluation of solution-processed reduced graphene oxide films as transparent conductors. ACS Nano 2008, 2 (3), 463-470. 131. Brodie, B. C., XXIII. - Researches on the atomic weight of graphite. Quarterly Journal of the Chemical Society of London 1860, 12 (1), 261-268. 132. Hummers Jr, W. S.; Offeman, R. E., Preparation of graphitic oxide. J. Am. Chem. Soc. 1958, 80 (6), 1339. 133. Shen, Y.; Yang, S.; Zhou, P.; Sun, Q.; Wang, P.; Wan, L.; Li, J.; Chen, L.; Wang, X.; Ding, S.; Zhang, D. W., Evolution of the band-gap and optical properties of graphene oxide with controllable reduction level. Carbon 2013, 62, 157-164. 134. Matyba, P.; Yamaguchi, H.; Chhowalla, M.; Robinson, N. D.; Edman, L., Flexible and metal-free light-emitting electrochemical cells based on graphene and PEDOT-PSS as the electrode materials. ACS Nano 2011, 5 (1), 574-580. 135. Li, X.; Wang, X.; Zhang, L.; Lee, S.; Dai, H., Chemically Derived, Ultrasmooth Graphene Nanoribbon Semiconductors. Science 2008, 319 (5867), 1229-1232.

331

136. Mattevi, C.; Eda, G.; Agnoli, S.; Miller, S.; Mkhoyan, K. A.; Celik, O.; Mastrogiovanni, D.; Granozzi, G.; Carfunkel, E.; Chhowalla, M., Evolution of electrical, chemical, and structural properties of transparent and conducting chemically derived graphene thin films. Adv. Funct. Mater. 2009, 19 (16), 2577-2583. 137. Schniepp, H. C.; Li, J. L.; McAllister, M. J.; Sai, H.; Herrera-Alonson, M.; Adamson, D. H.; Prud'homme, R. K.; Car, R.; Seville, D. A.; Aksay, I. A., Functionalized single graphene sheets derived from splitting graphite oxide. J. Phys. Chem. B 2006, 110 (17), 8535- 8539. 138. Wang, S.; Zhang, Y.; Abidi, N.; Cabrales, L., Wettability and surface free energy of graphene films. Langmuir 2009, 25 (18), 11078-11081. 139. López, V.; Sundaram, R. S.; Gómez-Navarro, C.; Olea, D.; Burghard, M.; Gómez- Herrero, J.; Zamora, F.; Kern, K., Chemical vapor deposition repair of graphene oxide: A route to highly conductive graphene monolayers. Adv. Mater. 2009, 21 (46), 4683-4686. 140. Su, Q.; Pang, S.; Alijani, V.; Li, C.; Feng, X.; Müllen, K., Composites of graphene with large aromatic molecules. Adv. Mater. 2009, 21 (31), 3191-3195. 141. Coleman, J. N., Liquid exfoliation of defect-free graphene. Acc. Chem. Res. 2013, 46 (1), 14-22. 142. Israelachvili, J. N., Intermolecular and Surface Forces (Second Edition). Academic Press: Suffolk, 1992. 143. Benedict, L. X.; Chopra, N. G.; Cohen, M. L.; Zettl, A.; Louie, S. G.; Crespi, V. H., Microscopic determination of the interlayer binding energy in graphite. Chem. Phys. Lett. 1998, 286 (5-6), 490-496. 144. Girifalco, L. A.; Good, R. J., A theory for the estimation of surface and interfacial energies. I. Derivation and application to interfacial tension. J. Phys. Chem. 1957, 61 (7), 904-909. 145. Hodak, M.; Girifalco, L. A., Fullerenes inside carbon nanotubes and multi-walled carbon nanotubes: Optimum and maximum sizes. Chem. Phys. Lett. 2001, 350 (5-6), 405- 411. 146. Zacharia, R.; Ulbricht, H.; Hertel, T., Interlayer cohesive energy of graphite from thermal desorption of polyaromatic hydrocarbons. Physical Review B - Condensed Matter and Materials Physics 2004, 69 (15), 155406-1-155406-7. 147. Halim, U.; Zheng, C. R.; Chen, Y.; Lin, Z.; Jiang, S.; Cheng, R.; Huang, Y.; Duan, X., A rational design of cosolvent exfoliation of layered materials by directly probing liquid- solid interaction. Nature Communications 2013, 4. 148. Whittingham, M. S., Intercalation Chemistry. Academic Press: New York, 1982. 149. Eda, G.; Yamaguchi, H.; Voiry, D.; Fujita, T.; Chen, M.; Chhowalla, M., Photoluminescence from chemically exfoliated MoS 2. Nano Lett. 2011, 11 (12), 5111-5116. 150. Kaskhedikar, N. A.; Maier, J., Lithium storage in carbon nanostructures. Adv. Mater. 2009, 21 (25-26), 2664-2680. 151. Lukowski, M. A.; Daniel, A. S.; Meng, F.; Forticaux, A.; Li, L.; Jin, S., Enhanced hydrogen evolution catalysis from chemically exfoliated metallic MoS2 nanosheets. J. Am. Chem. Soc. 2013, 135 (28), 10274-10277. 152. Viculis, L. H.; Mack, J. J.; Kaner, R. B., A chemical route to carbon nanoscrolls. Science 2003, 299 (5611), 1361. 153. Torrisi, F.; Hasan, T.; Wu, W.; Sun, Z.; Lombardo, A.; Kulmala, T. S.; Hsieh, G. W.; Jung, S.; Bonaccorso, F.; Paul, P. J.; Chu, D.; Ferrari, A. C., Inkjet-printed graphene electronics. ACS Nano 2012, 6 (4), 2992-3006. 154. Blake, P.; Brimicombe, P. D.; Nair, R. R.; Booth, T. J.; Jiang, D.; Schedin, F.; Ponomarenko, L. A.; Morozov, S. V.; Gleeson, H. F.; Hill, E. W.; Geim, A. K.; Novoselov, K. S., Graphene-based liquid crystal device. Nano Lett. 2008, 8 (6), 1704-1708. 155. Bosdorf, K.; Mollenhauer, I.; Boenick, U.; Willmann, G., Materials for Hip Joint Prostheses - Alternatives to Standard Materials. Biomedizinische Technik 1995, 40 (12), 356- 362. 156. Starý, V. r.; Bačáková, L.; Hornı́k, J.; Chmelı́k, V., Bio-compatibility of the surface layer of pyrolytic graphite. Thin Solid Films 2003, 433 (1–2), 191-198. 157. Jokerst, J. V.; Lobovkina, T.; Zare, R. N.; Gambhir, S. S., Nanoparticle PEGylation for imaging and therapy. Nanomedicine 2011, 6 (4), 715-728. 158. Van Vlerken, L. E.; Vyas, T. K.; Amiji, M. M., Poly(ethylene glycol)-modified nanocarriers for tumor-targeted and intracellular delivery. Pharm. Res. 2007, 24 (8), 1405- 1414. 159. Niidome, T.; Yamagata, M.; Okamoto, Y.; Akiyama, Y.; Takahashi, H.; Kawano, T.; Katayama, Y.; Niidome, Y., PEG-modified gold nanorods with a stealth character for in vivo applications. J. Controlled Release 2006, 114 (3), 343-347. 160. Green, H. N.; Martyshkin, D. V.; Rodenburg, C. M.; Rosenthal, E. L.; Mirov, S. B., Gold nanorod bioconjugates for active tumor targeting and photothermal therapy. Journal of Nanotechnology 2011. 161. Liao, H.; Hafner, J. H., Gold nanorod bioconjugates. Chem. Mater. 2005, 17 (18), 4636-4641. 162. Dougherty, T. J.; Gomer, C. J.; Henderson, B. W.; Jori, G.; Kessel, D.; Korbelik, M.; Moan, J.; Peng, Q., Photodynamic therapy. Journal of the National Cancer Institute 1998, 90 (12), 889-905. 163. Ganta, S.; Devalapally, H.; Shahiwala, A.; Amiji, M., A review of stimuli-responsive nanocarriers for drug and gene delivery. J. Controlled Release 2008, 126 (3), 187-204. 164. Menon, J. U.; Jadeja, P.; Tambe, P.; Vu, K.; Yuan, B.; Nguyen, K. T., Nanomaterials for photo-based diagnostic and therapeutic applications. Theranostics 2013, 3 (3), 152-166. 165. Choi, W. I.; Kim, J. Y.; Kang, C.; Byeon, C. C.; Kim, Y. H.; Tae, G., Tumor regression in vivo by photothermal therapy based on gold-nanorod-loaded, functional nanocarriers. ACS Nano 2011, 5 (3), 1995-2003. 166. Ahmad, M. Z.; Akhter, S.; Jain, G. K.; Rahman, M.; Pathan, S. A.; Ahmad, F. J.; Khar, R. K., Metallic nanoparticles: Technology overview and drug delivery applications in oncology. Expert Opinion on Drug Delivery 2010, 7 (8), 927-942. 167. Arias, J. L., Advanced methodologies to formulate nanotheragnostic agents for combined drug delivery and imaging. Expert Opinion on Drug Delivery 2011, 8 (12), 1589- 1608. 168. Akhter, S.; Ahmad, M. Z.; Singh, A.; Ahmad, I.; Rahman, M.; Anwar, M.; Jain, G. K.; Ahmad, F. J.; Khar, R. K., Cancer targeted metallic nanoparticle: Targeting overview, recent advancement and toxicity concern. Current Pharmaceutical Design 2011, 17 (18), 1834-1850. 169. Huang, X.; El-Sayed, I. H.; Qian, W.; El-Sayed, M. A., Cancer cell imaging and photothermal therapy in the near-infrared region by using gold nanorods. J. Am. Chem. Soc. 2006, 128 (6), 2115-2120. 170. Kuo, W. S.; Chang, C. N.; Chang, Y. T.; Yang, M. H.; Chien, Y. H.; Chen, S. J.; Yeh, C. S., Gold nanorods in photodynamic therapy, as hyperthermia agents, and in near-infrared optical imaging. Angew. Chem., Int. Ed. 2010, 49 (15), 2711-2715.

333

171. Matteini, P.; Ratto, F.; Rossi, F.; Pini, R., Emerging concepts of laser-activated nanoparticles for tissue bonding. Journal of Biomedical Optics 2012, 17 (1). 172. Patra, C. R.; Bhattacharya, R.; Mukhopadhyay, D.; Mukherjee, P., Fabrication of gold nanoparticles for targeted therapy in pancreatic cancer. Adv. Drug Delivery Rev. 2010, 62 (3), 346-361. 173. Ahmad, M. Z.; Akhter, S.; Rahman, Z.; Akhter, S.; Anwar, M.; Mallik, N.; Ahmad, F. J., Nanometric gold in cancer nanotechnology: Current status and future prospect. J. Pharm. Pharmacol. 2013, 65 (5), 634-651. 174. Tian, B.; Wang, C.; Zhang, S.; Feng, L.; Liu, Z., Photothermally enhanced photodynamic therapy delivered by nano-graphene oxide. ACS Nano 2011, 5 (9), 7000-7009. 175. Jain, P. K.; Lee, K. S.; El-Sayed, I. H.; El-Sayed, M. A., Calculated absorption and scattering properties of gold nanoparticles of different size, shape, and composition: Applications in biological imaging and biomedicine. J. Phys. Chem. B 2006, 110 (14), 7238- 7248. 176. Yang, K.; Feng, L.; Shi, X.; Liu, Z., Nano-graphene in biomedicine: Theranostic applications. Chem. Soc. Rev. 2013, 42 (2), 530-547. 177. Hutter, E.; Fendler, J. H., Exploitation of localized surface plasmon resonance. Adv. Mater. 2004, 16 (19), 1685-1706. 178. Connor, E. E.; Mwamuka, J.; Gole, A.; Murphy, C. J.; Wyatt, M. D., Gold nanoparticles are taken up by human cells but do not cause acute cytotoxicity. Small 2005, 1 (3), 325-327. 179. Grabinski, C.; Schaeublin, N.; Wijaya, A.; D'Couto, H.; Baxamusa, S. H.; Hamad- Schifferli, K.; Hussain, S. M., Effect of gold nanorod surface chemistry on cellular response. ACS Nano 2011, 5 (4), 2870-2879. 180. Pérez-Juste, J.; Pastoriza-Santos, I.; Liz-Marzán, L. M.; Mulvaney, P., Gold nanorods: Synthesis, characterization and applications. Coord. Chem. Rev. 2005, 249 (17-18 SPEC. ISS.), 1870-1901. 181. Huang, X.; Qian, W.; El-Sayed, I. H.; El-Sayed, M. A., The potential use of the enhanced nonlinear properties of gold nanospheres in photothermal cancer therapy. Lasers Surg. Med. 2007, 39 (9), 747-753. 182. Orendorff, C. J.; Murphy, C. J., Quantitation of Metal Content in the Silver-Assisted Growth of Gold Nanorods. The Journal of Physical Chemistry B 2006, 110 (9), 3990-3994. 183. Paton, K. R.; Coleman, J. N., Relating the optical absorption coefficient of nanosheet dispersions to the intrinsic monolayer absorption. Carbon 2016, 107, 733-738. 184. Alkilany, A. M.; Murphy, C. J., Toxicity and cellular uptake of gold nanoparticles: What we have learned so far? J. Nanopart. Res. 2010, 12 (7), 2313-2333. 185. Huang, X.; El-Sayed, M. A., Gold nanoparticles: Optical properties and implementations in cancer diagnosis and photothermal therapy. Journal of Advanced Research 2010, 1 (1), 13-28. 186. Xie, J.; Lee, J. Y.; Wang, D. I. C., Seedless, surfactantless, high-yield synthesis of branched gold nanocrystals in HEPES buffer solution. Chem. Mater. 2007, 19 (11), 2823- 2830. 187. Plech, A.; Kotaidis, V.; Grésillon, S.; Dahmen, C.; Von Plessen, G., Laser-induced heating and melting of gold nanoparticles studied by time-resolved x-ray scattering. Physical Review B - Condensed Matter and Materials Physics 2004, 70 (19), 1-7. 188. Inasawa, S.; Sugiyama, M.; Yamaguchi, Y., Laser-Induced Shape Transformation of Gold Nanoparticles below the Melting Point: The Effect of Surface Melting. The Journal of Physical Chemistry B 2005, 109 (8), 3104-3111. 189. Mahdihassan, S., Cinnabar-gold as the best alchemical drug of longevity, called Makaradhwaja in India. American Journal of Chinese Medicine 1985, 13 (1-4), 93-108. 190. Unrine, J. M.; Tsyusko, O. V.; Hunyadi, S. E.; Judy, J. D.; Bertsch, P. M., Effects of particle size on chemical speciation and bioavailability of copper to earthworms (Eisenia fetida) exposed to copper nanoparticles. Journal of Environmental Quality 2010, 39 (6), 1942-1953. 191. Priester, J. H.; Stoimenov, P. K.; Mielke, R. E.; Webb, S. M.; Ehrhardt, C.; Zhang, J. P.; Stucky, G. D.; Holden, P. A., Effects of soluble cadmium salts versus cdse quantum dots on the growth of planktonic pseudomonas aeruginosa. Environ. Sci. Technol. 2009, 43 (7), 2589-2594. 192. Zhou, F.; Xing, D.; Song, S.; Ou, Z.; Chen, W., Single-walled carbon nanotubes enhance near-infrared region photothermal therapy. Zhongguo Jiguang/Chinese Journal of Lasers 2009, 36 (10), 2676-2679. 193. Zhou, F.; Xing, D.; Ou, Z.; Wu, B.; Resasco, D. E.; Chen, W. R., Cancer photothermal therapy in the near-infrared region by using single-walled carbon nanotubes. Journal of Biomedical Optics 2009, 14 (2). 194. Singh, R.; Torti, S. V., Carbon nanotubes in hyperthermia therapy. Adv. Drug Delivery Rev. 2013. 195. Kam, N. W. S.; O'Connell, M.; Wisdom, J. A.; Dai, H., Carbon nanotubes as multifunctional biological transporters and near-infrared agents for selective cancer cell destruction. Proc. Natl. Acad. Sci. U. S. A. 2005, 102 (33), 11600-11605. 196. Quinn, M. D. J.; Ho, N. H.; Notley, S. M., Aqueous dispersions of exfoliated molybdenum disulfide for use in visible-light photocatalysis. ACS Appl. Mater. Interfaces 2013, 5 (23), 12751-12756. 197. Chou, S. S.; Kaehr, B.; Kim, J.; Foley, B. M.; De, M.; Hopkins, P. E.; Huang, J.; Brinker, C. J.; Dravid, V. P., Chemically exfoliated MoS2 as near-infrared photothermal agents. Angew. Chem., Int. Ed. 2013, 52 (15), 4160-4164. 198. Wang, S.; Li, K.; Chen, Y.; Chen, H.; Ma, M.; Feng, J.; Zhao, Q.; Shi, J., Biocompatible PEGylated MoS2 nanosheets: Controllable bottom-up synthesis and highly efficient photothermal regression of tumor. Biomaterials 2015, 39, 206-217. 199. Barton, A., Handbook of Solubility Parameters and Other Cohesion Parameters, Second Edition. CRC Press: 1991. 200. Dikin, D. A.; Stankovich, S.; Zimney, E. J.; Piner, R. D.; Dommett, G. H. B.; Evmenenko, G.; Nguyen, S. T.; Ruoff, R. S., Preparation and characterization of graphene oxide paper. Nature 2007, 448 (7152), 457-460. 201. Guardia, L.; Fernández-Merino, M. J.; Paredes, J. I.; Solís-Fernández, P.; Villar- Rodil, S.; Martínez-Alonso, A.; Tascón, J. M. D., High-throughput production of pristine graphene in an aqueous dispersion assisted by non-ionic surfactants. Carbon 2011, 49 (5), 1653-1662. 202. Jewett, S. a., Principles of Physics. Thomson Brooks/Cole: Canada, 2006. 203. Paton, K. R.; Varrla, E.; Backes, C.; Smith, R. J.; Khan, U.; O'Neill, A.; Boland, C.; Lotya, M.; Istrate, O. M.; King, P.; Higgins, T.; Barwich, S.; May, P.; Puczkarski, P.; Ahmed, I.; Moebius, M.; Pettersson, H.; Long, E.; Coelho, J.; O'Brien, S. E.; McGuire, E. K.; Sanchez, B. M.; Duesberg, G. S.; McEvoy, N.; Pennycook, T. J.; Downing, C.; Crossley, A.; Nicolosi, V.; Coleman, J. N., Scalable production of large quantities of defect-free few layer graphene by shear exfoliation in liquids. Nat. Mater. 2014, 13 (6), 624-630. 204. Flint, E. B.; Suslick, K. S., The temperature of cavitation. Science 1991, 253 (5026), 1397-1399.

335

205. Suslick, K. S.; Flannigan, D. J., Inside a collapsing bubble: Sonoluminescence and the conditions during cavitation. In Annu. Rev. Phys. Chem., 2008; Vol. 59, pp 659-683. 206. Yi, M.; Shen, Z., A review on mechanical exfoliation for the scalable production of graphene. Journal of Materials Chemistry A 2015, 3 (22), 11700-11715. 207. Yazyev, O. V.; Chen, Y. P., Polycrystalline graphene and other two-dimensional materials. Nat Nano 2014, 9 (10), 755-767. 208. Yazyev, O. V.; Louie, S. G., Topological defects in graphene: Dislocations and grain boundaries. Physical Review B - Condensed Matter and Materials Physics 2010, 81 (19). 209. Yi, M.; Shen, Z.; Liang, S.; Liu, L.; Zhang, X.; Ma, S., Water can stably disperse liquid-exfoliated graphene. Chem. Commun. 2013, 49 (94), 11059-11061. 210. Khan, U.; O'Neill, A.; Lotya, M.; De, S.; Coleman, J. N., High-concentration solvent exfoliation of graphene. Small 2010, 6 (7), 864-871. 211. Bracamonte, M. V.; Lacconi, G. I.; Urreta, S. E.; Foa Torres, L. E. F., On the nature of defects in liquid-phase exfoliated graphene. J. Phys. Chem. C 2014, 118 (28), 15455- 15459. 212. Cançado, L. G.; Jorio, A.; Ferreira, E. H. M.; Stavale, F.; Achete, C. A.; Capaz, R. B.; Moutinho, M. V. O.; Lombardo, A.; Kulmala, T. S.; Ferrari, A. C., Quantifying defects in graphene via Raman spectroscopy at different excitation energies. Nano Lett. 2011, 11 (8), 3190-3196. 213. Pollard, A. J.; Brennan, B.; Stec, H.; Tyler, B. J.; Seah, M. P.; Gilmore, I. S.; Roy, D., Quantitative characterization of defect size in graphene using Raman spectroscopy. Appl. Phys. Lett. 2014, 105 (25), 253107. 214. Polyakova, E. Y.; Rim, K. T.; Eom, D.; Douglass, K.; Opila, R. L.; Heinz, T. F.; Teplyakov, A. V.; Flynn, G. W., Scanning tunneling microscopy and X-ray photoelectron spectroscopy studies of graphene films prepared by sonication-assisted dispersion. ACS Nano 2011, 5 (8), 6102-6108. 215. Skaltsas, T.; Ke, X.; Bittencourt, C.; Tagmatarchis, N., Ultrasonication induces oxygenated species and defects onto exfoliated graphene. J. Phys. Chem. C 2013, 117 (44), 23272-23278. 216. Alexandridis, P.; Alan Hatton, T., Poly(ethylene oxide)poly(propylene oxide)poly(ethylene oxide) block copolymer surfactants in aqueous solutions and at interfaces: thermodynamics, structure, dynamics, and modeling. Colloids Surf., A 1995, 96 (1-2), 1-46. 217. Alexandridis, P.; Holzwarth, J. F.; Hatton, T. A., Micellization of poly(ethylene oxide)-poly(propylene oxide)-poly(ethylene oxide) triblock copolymers in aqueous solutions: thermodynamics of copolymer association. Macromolecules 1994, 27 (9), 2414- 2425. 218. Mata, J. P.; Majhi, P. R.; Guo, C.; Liu, H. Z.; Bahadur, P., Concentration, temperature, and salt-induced micellization of a triblock copolymer Pluronic L64 in aqueous media. J. Colloid Interface Sci. 2005, 292 (2), 548-556. 219. Mysels, K. J., Surface tension of solutions of pure sodium dodecyl sulfate. Langmuir 1986, 2 (4), 423-428. 220. Lotya, M.; King, P. J.; Khan, U.; De, S.; Coleman, J. N., High-concentration, surfactant-stabilized graphene dispersions. ACS Nano 2010, 4 (6), 3155-3162. 221. Lotya, M.; Hernandez, Y.; King, P. J.; Smith, R. J.; Nicolosi, V.; Karlsson, L. S.; Blighe, F. M.; De, S.; Zhiming, W.; McGovern, I. T.; Duesberg, G. S.; Coleman, J. N., Liquid phase production of graphene by exfoliation of graphite in surfactant/water solutions. J. Am. Chem. Soc. 2009, 131 (10), 3611-3620. 222. Ferrari, A. C.; Meyer, J. C.; Scardaci, V.; Casiraghi, C.; Lazzeri, M.; Mauri, F.; Piscanec, S.; Jiang, D.; Novoselov, K. S.; Roth, S.; Geim, A. K., Raman spectrum of graphene and graphene layers. Phys. Rev. Lett. 2006, 97 (18). 223. John Edward Proctor, D. A. M. A., Aravind Vijayarghavan, An Introduction to Graphene and Carbon Nanotubes. CRC Press, Taylor and Francis Group: USA, 2017. 224. Cong, C.; Yu, T.; Saito, R.; Dresselhaus, G. F.; Dresselhaus, M. S., Second-Order Overtone and Combination Raman Modes of Graphene Layers in the Range of 1690−2150 cm−1. ACS Nano 2011, 5 (3), 1600-1605. 225. Beams, R.; Gustavo Cançado, L.; Novotny, L., Raman characterization of defects and dopants in graphene. Journal of Physics Condensed Matter 2015, 27 (8). 226. Ferrari, A. C., Raman spectroscopy of graphene and graphite: Disorder, electron- phonon coupling, doping and nonadiabatic effects. Solid State Commun. 2007, 143 (1-2), 47- 57. 227. Isaac Childres, L. A. J., Wonjun Park, Helin Cao, Yong P. Chen, RAMAN SPECTROSCOPY OF GRAPHENE AND RELATED MATERIALS. Purdue University, 2013. 228. Yuan, W.; Li, C.; Li, D.; Yang, J.; Zeng, X., Preparation of single- and few layer graphene sheets using Co deposition on sic substrate. Journal of Nanomaterials 2011, 2011. 229. Lucchese, M. M.; Stavale, F.; Ferreira, E. H. M.; Vilani, C.; Moutinho, M. V. O.; Capaz, R. B.; Achete, C. A.; Jorio, A., Quantifying ion-induced defects and Raman relaxation length in graphene. Carbon 2010, 48 (5), 1592-1597. 230. Saito, R.; Hofmann, M.; Dresselhaus, G.; Jorio, A.; Dresselhaus, M. S., Raman spectroscopy of graphene and carbon nanotubes. Advances in Physics 2011, 60 (3), 413-550. 231. Saito, R.; Jorio, A.; Souza Filho, A. G.; Dresselhaus, G.; Dresselhaus, M. S.; Pimenta, M. A., Probing phonon dispersion relations of graphite by double resonance Raman scattering. Phys. Rev. Lett. 2002, 88 (2), 274011-274014. 232. Tuinstra, F.; Koenig, J. L., Raman Spectrum of Graphite. The Journal of Chemical Physics 1970, 53 (3), 1126-1130. 233. Wang, T.; Quinn, M. D. J.; Notley, S., A benzoxazine surfactant exchange for atomic force microscopy characterization of two dimensional materials exfoliated in aqueous surfactant solutions. RSC Advances 2017, 7 (6), 3222-3228. 234. Wall, M., Raman spectroscopy optimizes graphene characterization. Adv. Mater. Processes 2012, 170 (4), 35-38. 235. Pham, V. T. H.; Truong, V. K.; Quinn, M. D. J.; Notley, S. M.; Guo, Y.; Baulin, V. A.; Al Kobaisi, M.; Crawford, R. J.; Ivanova, E. P., Graphene Induces Formation of Pores That Kill Spherical and Rod-Shaped Bacteria. ACS Nano 2015, 9 (8), 8458-8467. 236. Kang, M. S.; Kim, K. T.; Lee, J. U.; Jo, W. H., Direct exfoliation of graphite using a non-ionic polymer surfactant for fabrication of transparent and conductive graphene films. J. Mater. Chem. C 2013, 1 (9), 1870-1875. 237. Su, R.; Lin, S. F.; Chen, D. Q.; Chen, G. H., Study on the Absorption Coefficient of Reduced Graphene Oxide Dispersion. The Journal of Physical Chemistry C 2014, 118 (23), 12520-12525. 238. Dan, L.; Müller, M. B.; Gilje, S.; Kaner, R. B.; Wallace, G. G., Processable aqueous dispersions of graphene nanosheets. Nat. Nanotechnol. 2008, 3 (2), 101-105. 239. Meyer, J. C.; Geim, A. K.; Katsnelson, M. I.; Novoselov, K. S.; Obergfell, D.; Roth, S.; Girit, C.; Zettl, A., On the roughness of single- and bi-layer graphene membranes. Solid State Commun. 2007, 143 (1-2), 101-109.

337

240. Shigeo, H.; Takuya, G.; Masahiro, F.; Ryuji, S.; Masukazu, H.; Koji, K.; Toru, A.; Tadahiro, Y.; Yoshio, M.; Kenji, W.; Masami, S., Carbon Nanofilm with a New Structure and Property. Japanese Journal of Applied Physics 2003, 42 (9A), L1073. 241. Kim, K.; Lee, Z.; Malone, B. D.; Chan, K. T.; Alemán, B.; Regan, W.; Gannett, W.; Crommie, M. F.; Cohen, M. L.; Zettl, A., Multiply folded graphene. Physical Review B - Condensed Matter and Materials Physics 2011, 83 (24). 242. Malvern, Zetasizer ZS nano specifications. In Zetasizer Nano series, Limited, M. I., Ed. 2017. 243. Lotya, M.; Rakovich, A.; Donegan, J. F.; Coleman, J. N., Measuring the lateral size of liquid-exfoliated nanosheets with dynamic light scattering. Nanotechnology 2013, 24 (26). 244. Jean, F.; Zhou, T.; Blanc, N.; Felici, R.; Coraux, J.; Renaud, G., Effect of preparation on the commensurabilities and thermal expansion of graphene on Ir(111) between 10 and 1300 K. Physical Review B - Condensed Matter and Materials Physics 2013, 88 (16). 245. Zakharchenko, K. V.; Katsnelson, M. I.; Fasolino, A., Finite temperature lattice properties of graphene beyond the quasiharmonic approximation. Phys. Rev. Lett. 2009, 102 (4). 246. Nelson, J. B.; Riley, D. P., The thermal expansion of graphite from 15°c. to 800°c.: part I. Experimental. Proceedings of the Physical Society 1945, 57 (6), 477. 247. Mounet, N.; Marzari, N., First-principles determination of the structural, vibrational and thermodynamic properties of diamond, graphite, and derivatives. Physical Review B 2005, 71 (20), 205214. 248. Sobon, G.; Sotor, J.; Jagiello, J.; Kozinski, R.; Zdrojek, M.; Holdynski, M.; Paletko, P.; Boguslawski, J.; Lipinska, L.; Abramski, K. M., Graphene Oxide vs. Reduced Graphene Oxide as saturable absorbers for Er-doped passively mode-locked fiber laser. Opt. Express 2012, 20 (17), 19463-19473. 249. Stankovich, S.; Dikin, D. A.; Piner, R. D.; Kohlhaas, K. A.; Kleinhammes, A.; Jia, Y.; Wu, Y.; Nguyen, S. T.; Ruoff, R. S., Synthesis of graphene-based nanosheets via chemical reduction of exfoliated graphite oxide. Carbon 2007, 45 (7), 1558-1565. 250. technology, N. i. o. s. a., Probing nanoscale structures - the SANS toolbox. Boualem Hammouda: Center for neutron research, Gaithersburg, MD 201899-6102, 2017. 251. Teixeira, J., Small angle scattering by fractal systems. J. Appl. Crystallogr. 1988. 252. Milner, E. M.; Skipper, N. T.; Howard, C. A.; Shaffer, M. S. P.; Buckley, D. J.; Rahnejat, K. A.; Cullen, P. L.; Heenan, R. K.; Lindner, P.; Schweins, R., Structure and morphology of charged graphene platelets in solution by small-angle neutron scattering. J. Am. Chem. Soc. 2012, 134 (20), 8302-8305. 253. Dirk Basting, G. M., Excimer Laser Technology. Springer science and business media: 2005. 254. Willett, C. S., Introduction to gas lasers: population inversion mechanisms. Pergamon Press: Washington, DC, 1974. 255. Scheps, R., Introduction to laser diode-pumped solid state lasers. SPIE Press: United states of america, 2002. 256. Shiuji Nakamura, S. F. C., Introduction to nitride semiconductor blue lasers and light emitting diodes. Taylor and Francis: 2000. 257. Klotzkin, D. J., Introduction to semiconductor lasers for optical communications: an applied approach. Springer-Verlag: New York, 2014. 258. Renk, K. F., An Introduction to Semiconductor Lasers. In Basics of Laser Physics: For Students of Science and Engineering, Springer Berlin Heidelberg: Berlin, Heidelberg, 2012; pp 371-382. 259. Chang, R., Chemistry 9th Edition. McGraw Hill: New York, 2007; Vol. 9. 260. Ekici, O.; Harrison, R. K.; Durr, N. J.; Eversole, D. S.; Lee, M.; Ben-Yakar, A., Thermal analysis of gold nanorods heated with femtosecond laser pulses. J. Phys. D: Appl. Phys. 2008, 41 (18). 261. Liu, Y. Nanoscale Thermal Transport at Graphene-Soft Material Interfaces. Virginia Polytechnic Institute and State University, Blacksberg, VA, 2016. 262. Kittel, C., Introduction to Solid State Physics, 8th Edition. John Wiley and Sons, Inc.: New Caledonia, 2004. 263. Taft, E. A.; Philipp, H. R., Optical Properties of Graphite. Physical Review 1965, 138 (1A), A197-A202. 264. Matsumoto, T.; Koizumi, T.; Kawakami, Y.; Okamoto, K.; Tomita, M., Perfect blackbody radiation from a graphene nanostructure with application to hightemperature spectral emissivity measurements. Opt. Express 2013, 21 (25), 30964-30974. 265. Zhang, M.; Parajuli, R. R.; Mastrogiovanni, D.; Dai, B.; Lo, P.; Cheung, W.; Brukh, R.; Chiu, P. L.; Zhou, T.; Liu, Z.; Garfunkel, E.; He, H., Production of graphene sheets by direct dispersion with aromatic healing agents. Small 2010, 6 (10), 1100-1107. 266. Backes, C.; Paton, K. R.; Hanlon, D.; Yuan, S.; Katsnelson, M. I.; Houston, J.; Smith, R. J.; McCloskey, D.; Donegan, J. F.; Coleman, J. N., Spectroscopic metrics allow in situ measurement of mean size and thickness of liquid-exfoliated few layer graphene nanosheets. Nanoscale 2016, 8 (7), 4311-4323. 267. Slater, J. C.; Koster, G. F., Simplified LCAO Method for the Periodic Potential Problem. Physical Review 1954, 94 (6), 1498-1524. 268. Casiraghi, C.; Hartschuh, A.; Lidorikis, E.; Qian, H.; Harutyunyan, H.; Gokus, T.; Novoselov, K. S.; Ferrari, A. C., Rayleigh imaging of graphene and graphene layers. Nano Lett. 2007, 7 (9), 2711-2717. 269. Bonaccorso, F.; Sun, Z.; Hasan, T.; Ferrari, A. C., Graphene photonics and optoelectronics. Nature Photonics 2010, 4 (9), 611-622. 270. Chen, G.; Zeng, T.; Borca-Tasciuc, T.; Song, D., Phonon engineering in nanostructures for solid-state energy conversion. Mater. Sci. Eng., A 2000, 292 (2), 155-161. 271. Ong, Z. Y.; Pop, E., Frequency and polarization dependence of thermal coupling between carbon nanotubes and SiO2. J. Appl. Phys. 2010, 108 (10). 272. Ong, Z. Y.; Pop, E.; Shiomi, J., Reduction of phonon lifetimes and thermal conductivity of a carbon nanotube on amorphous silica. Physical Review B - Condensed Matter and Materials Physics 2011, 84 (16). 273. Kang, K.; Abdula, D.; Cahill, D. G.; Shim, M., Lifetimes of optical phonons in graphene and graphite by time-resolved incoherent anti-Stokes Raman scattering. Physical Review B - Condensed Matter and Materials Physics 2010, 81 (16). 274. Nihira, T.; Iwata, T., Temperature dependence of lattice vibrations and analysis of the specific heat of graphite. Physical Review B - Condensed Matter and Materials Physics 2003, 68 (13), 1343051-13430516. 275. Benedict, L. X.; Louie, S. G.; Cohen, M. L., Heat capacity of carbon nanotubes. Solid State Commun. 1996, 100 (3), 177-180. 276. Tohei, T.; Kuwabara, A.; Oba, F.; Tanaka, I., Debye temperature and stiffness of carbon and boron nitride polymorphs from first principles calculations. Physical Review B - Condensed Matter and Materials Physics 2006, 73 (6). 277. Nicklow, R.; Wakabayashi, N.; Smith, H. G., Lattice dynamics of pyrolytic graphite. Physical Review B 1972, 5 (12), 4951-4962.

339

278. Yiǧen, S.; Tayari, V.; Island, J. O.; Porter, J. M.; Champagne, A. R., Electronic thermal conductivity measurements in intrinsic graphene. Physical Review B - Condensed Matter and Materials Physics 2013, 87 (24). 279. Jeong, C.; Datta, S.; Lundstrom, M., Thermal conductivity of bulk and thin-film silicon: A Landauer approach. J. Appl. Phys. 2012, 111 (9). 280. Abadeer, N. S.; Murphy, C. J., Recent Progress in Cancer Thermal Therapy Using Gold Nanoparticles. J. Phys. Chem. C 2016, 120 (9), 4691-4716. 281. Huang, J.; Park, J.; Wang, W.; Murphy, C. J.; Cahill, D. G., Ultrafast Thermal Analysis of Surface Functionalized Gold Nanorods in Aqueous Solution. ACS Nano 2013, 7 (1), 589-597. 282. Berman, R., Thermal conductivity of isotopically enriched diamonds. Physical Review B 1992, 45 (10), 5726-5728. 283. Balandin, A. A., Thermal properties of graphene and nanostructured carbon materials. Nat. Mater. 2011, 10 (8), 569-581. 284. Haskins, J.; Kinaci, A.; Sevik, C.; Sevinçli, H.; Cuniberti, G. t. p. T.; Çaǧin, T., Control of thermal and electronic transport in defect-engineered graphene nanoribbons. ACS Nano 2011, 5 (5), 3779-3787. 285. Mark Lundstrom, J. G., Nanoscale Transistors, Device Physics Modeling and Simulation. Folaida, USA, 2006. 286. Saito, K.; Nakamura, J.; Natori, A., Ballistic thermal conductance of a graphene sheet. Physical Review B - Condensed Matter and Materials Physics 2007, 76 (11). 287. Xiangfan Xu, Y. W., Kaiwen Zhang, Xiangming Zhao, Sukang Bae, Martin Heinrich, Cong Tinh Bui, Rongguo Xie, John T. L. Thong, Byung Hee Hong, Kian Ping Loh, Baowen Li, Barbaros Ozyilmaz, Phonon Transport in Suspended Single Layer Graphene. Mesoscale and Nanoscale Physics 2010. 288. Jiang, J. W.; Wang, J. S.; Li, B., Thermal conductance of graphene and dimerite. Physical Review B - Condensed Matter and Materials Physics 2009, 79 (20). 289. Lindsay, L.; Broido, D. A.; Mingo, N., Flexural phonons and thermal transport in multilayer graphene and graphite. Physical Review B - Condensed Matter and Materials Physics 2011, 83 (23). 290. Aksamija, Z.; Knezevic, I., Lattice thermal conductivity of graphene nanoribbons: Anisotropy and edge roughness scattering. Appl. Phys. Lett. 2011, 98 (14). 291. Mak, K. F.; Lui, C. H.; Heinz, T. F., Measurement of the thermal conductance of the graphene/ SiO2 interface. Appl. Phys. Lett. 2010, 97 (22). 292. Merabia, S.; Termentzidis, K., Thermal conductance at the interface between crystals using equilibrium and nonequilibrium molecular dynamics. Physical Review B - Condensed Matter and Materials Physics 2012, 86 (9). 293. Luo, T.; Lloyd, J. R., Enhancement of thermal energy transport across graphene/graphite and polymer interfaces: A molecular dynamics study. Adv. Funct. Mater. 2012, 22 (12), 2495-2502. 294. Swartz, E. T.; Pohl, R. O., Thermal boundary resistance. Rev. Mod. Phys. 1989, 61 (3), 605-668. 295. Alexeev, D.; Chen, J.; Walther, J. H.; Giapis, K. P.; Angelikopoulos, P.; Koumoutsakos, P., Kapitza Resistance between Few layer Graphene and Water: Liquid Layering Effects. Nano Lett. 2015, 15 (9), 5744-5749. 296. Kapitza, P. L., Heat transfer and superfluidity of helium II. Physical Review 1941, 60 (4), 354-355. 297. Loh, G. C.; Teo, E. H. T.; Tay, B. K., Compounded effect of vacancy on interfacial thermal transport in diamond-graphene nanostructures. Diamond Relat. Mater. 2011, 20 (8), 1137-1142. 298. Cahill, D. G.; Braun, P. V.; Chen, G.; Clarke, D. R.; Fan, S.; Goodson, K. E.; Keblinski, P.; King, W. P.; Mahan, G. D.; Majumdar, A.; Maris, H. J.; Phillpot, S. R.; Pop, E.; Shi, L., Nanoscale thermal transport. II. 2003-2012. Applied Physics Reviews 2014, 1 (1). 299. Merabia, S.; Keblinski, P.; Joly, L.; Lewis, L. J.; Barrat, J. L., Critical heat flux around strongly heated nanoparticles. Physical Review E - Statistical, Nonlinear, and Soft Matter Physics 2009, 79 (2). 300. Hu, L.; Desai, T.; Keblinski, P., Determination of interfacial thermal resistance at the nanoscale. Physical Review B - Condensed Matter and Materials Physics 2011, 83 (19). 301. Chen, Z.; Jang, W.; Bao, W.; Lau, C. N.; Dames, C., Thermal contact resistance between graphene and silicon dioxide. Appl. Phys. Lett. 2009, 95 (16). 302. Shima, P. D.; Philip, J., Role of Thermal Conductivity of Dispersed Nanoparticles on Heat Transfer Properties of Nanofluid. Industrial & Engineering Chemistry Research 2014, 53 (2), 980-988. 303. Eastman, J. A.; Choi, S. U. S.; Li, S.; Yu, W.; Thompson, L. J., Anomalously increased effective thermal conductivities of ethylene glycol-based nanofluids containing copper nanoparticles. Appl. Phys. Lett. 2001, 78 (6), 718-720. 304. Xie, H.; Lee, H.; Youn, W.; Choi, M., Nanofluids containing multiwalled carbon nanotubes and their enhanced thermal conductivities. J. Appl. Phys. 2003, 94 (8), 4967-4971. 305. Das, S. K.; Putra, N.; Thiesen, P.; Roetzel, W., Temperature dependence of thermal conductivity enhancement for nanofluids. J. Heat Transfer 2003, 125 (4), 567-574. 306. Maxwell, J. C.; Garnett, W., An elementary treatise on electricity. 2011; p 1-215. 307. James Clerk, M., A treatise on electricity and magnetism. 2010; p 1-442. 308. Angayarkanni, S. A.; Philip, J., Review on thermal properties of nanofluids: Recent developments. Adv. Colloid Interface Sci. 2015, 225, 146-176. 309. Keblinski, P.; Phillpot, S. R.; Choi, S. U. S.; Eastman, J. A., Mechanisms of heat flow in suspensions of nano-sized particles (nanofluids). Int. J. Heat Mass Transfer 2002, 45 (4), 855-863. 310. Shen, M.; Evans, W. J.; Cahill, D.; Keblinski, P., Bonding and pressure-tunable interfacial thermal conductance. Physical Review B - Condensed Matter and Materials Physics 2011, 84 (19). 311. Timofeeva, E. V.; Routbort, J. L.; Singh, D., Particle shape effects on thermophysical properties of alumina nanofluids. J. Appl. Phys. 2009, 106 (1), 014304. 312. Lee, K. J.; Yoon, S.-H.; Jang, J., Carbon Nanofibers: A Novel Nanofiller for Nanofluid Applications. Small 2007, 3 (7), 1209-1213. 313. Philip, J.; Shima, P. D.; Raj, B., Enhancement of thermal conductivity in magnetite based nanofluid due to chainlike structures. Appl. Phys. Lett. 2007, 91 (20), 203108. 314. Wei, Z.; Ni, Z.; Bi, K.; Chen, M.; Chen, Y., Interfacial thermal resistance in multilayer graphene structures. Phys. Lett. A 2011, 375 (8), 1195-1199. 315. Wang, Y.; Yang, C.; Mai, Y. W.; Zhang, Y., Effect of non-covalent functionalisation on thermal and mechanical properties of graphene-polymer nanocomposites. Carbon 2016, 102, 311-318. 316. Lin, S.; Buehler, M. J., The effect of non-covalent functionalization on the thermal conductance of graphene/organic interfaces. Nanotechnology 2013, 24 (16).

341

317. Wang, Y.; Zhan, H. F.; Xiang, Y.; Yang, C.; Wang, C. M.; Zhang, Y. Y., Effect of Covalent Functionalization on Thermal Transport across Graphene-Polymer Interfaces. J. Phys. Chem. C 2015, 119 (22), 12731-12738. 318. Wang, M.; Galpaya, D.; Lai, Z. B.; Xu, Y.; Yan, C., Surface functionalization on the thermal conductivity of graphene-polymer nanocomposites. International Journal of Smart and Nano Materials 2014, 5 (2), 123-132. 319. Chien, S. K.; Yang, Y. T.; Chen, C. K., Influence of chemisorption on the thermal conductivity of graphene nanoribbons. Carbon 2012, 50 (2), 421-428. 320. Henry, A., Thermal transport in polymers. In Annual review of heat transfer, Begell house: 2013; Vol. Volume 17. 321. Choy, C. L., Thermal conductivity of polymers. Polymer 1977, 18 (10), 984-1004. 322. Roper, D. K.; Ahn, W.; Hoepfner, M., Microscale heat transfer transduced by surface plasmon resonant gold nanoparticles. J. Phys. Chem. C 2007, 111 (9), 3636-3641. 323. Kotaidis, V.; Plech, A., Cavitation dynamics on the nanoscale. Appl. Phys. Lett. 2005, 87 (21), 1-3. 324. Liu, L.; Ryu, S.; Tomasik, M. R.; Stolyarova, E.; Jung, N.; Hybertsen, M. S.; Steigerwald, M. L.; Brus, L. E.; Flynn, G. W., Graphene oxidation: Thickness-dependent etching and strong chemical doping. Nano Lett. 2008, 8 (7), 1965-1970. 325. Ulbricht, H.; Moos, G.; Hertel, T., Physisorption of molecular oxygen on single-wall carbon nanotube bundles and graphite. Physical Review B - Condensed Matter and Materials Physics 2002, 66 (7), 754041-754047. 326. Giannozzi, P.; Car, R.; Scoles, G., Oxygen adsorption on graphite and nanotubes. J. Chem. Phys. 2003, 118 (3), 1003-1006. 327. Yang, R. T.; Wong, C., Kinetics and mechanism of oxidation of basal plane on graphite. The Journal of Chemical Physics 1981, 75 (9), 4471-4476. 328. Prasher, R.; Phelan, P. E.; Bhattacharya, P., Effect of aggregation kinetics on the thermal conductivity of nanoscale colloidal solutions (nanofluid). Nano Lett. 2006, 6 (7), 1529-1534. 329. Ye, L.; Liu, J.; Sheng, P.; Huang, J. S.; Weitz, D. A., Sound propagation in colloidal systems. Journal De Physique. IV : JP 1993, 3 (1), 183-196. 330. Quinn, M. D. J.; Vu, K.; Madden, S.; Notley, S. M., Photothermal Breaking of Emulsions Stabilized with Graphene. ACS Appl. Mater. Interfaces 2016, 8 (16), 10609- 10616. 331. Quinn, M. D. J.; Wang, T.; Du, J. D.; Boyd, B. J.; Hawley, A.; Notley, S. M., Graphene as a photothermal actuator for control of lipid mesophase structure. Nanoscale 2017. 332. Pattani, V. P.; Tunnell, J. W., Nanoparticle-mediated photothermal therapy: A comparative study of heating for different particle types. Lasers Surg. Med. 2012, 44 (8), 675-684. 333. Ltd., C. Crystran UV Visible IR Specialist Optics - Silica Glass. https://www.crystran.co.uk/optical-materials/silica-glass-sio2. 334. Smith, R. J.; Lotya, M.; Coleman, J. N., The importance of repulsive potential barriers for the dispersion of graphene using surfactants. New Journal of Physics 2010, 12. 335. Campese, G. M.; Rodrigues, E. M. G.; Tambourgi, E. B.; Pessoa Jr, A., Determination of cloud-point temperatures for different copolymers. Brazilian Journal of Chemical Engineering 2003, 20 (3), 335-337. 336. Adam, U. S.; Robb, I. D., Adsorption and exchange of polyelectrolytes on crystal surfaces. Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases 1983, 79 (11), 2745-2753. 337. Lal, M.; Turpin, M. A.; Richardson, K. A.; Spencer, D., CONFIGURATIONAL BEHAVIOR OF ISOLATED CHAIN MOLECULES ADSORBED FROM ATHERMAL SOLUTIONS. ACS Symp. Ser. 1974, (8), 16-27. 338. Silberberg, A., Adsorption of flexible macromolecules. IV. Effect of solvent-solute interactions, solute concentration, and molecular weight. The Journal of Chemical Physics 1968, 48 (7), 2835-2851. 339. Yang, K.; Wan, J.; Zhang, S.; Tian, B.; Zhang, Y.; Liu, Z., The influence of surface chemistry and size of nanoscale graphene oxide on photothermal therapy of cancer using ultra-low laser power. Biomaterials 2012, 33 (7), 2206-2214. 340. Rahme, K.; Gauffre, F.; Marty, J. D.; Payré, B.; Mingotaud, C., A systematic study of the stabilization in water of gold nanoparticles by poly(ethylene oxide)-poly(propylene oxide)-poly(ethylene oxide) triblock copolymers. J. Phys. Chem. C 2007, 111 (20), 7273- 7279. 341. Prasad, M.; Moulik, S. P.; Wardian, A.; Moore, S.; van Bommel, A.; Palepu, R., Alkyl (C10, C12, C14 and C16) triphenyl phosphonium bromide influenced cloud points of nonionic surfactants (Triton X 100, Brij 56 and Brij 97) and the polymer (Polyvinyl methyl ether). Colloid. Polym. Sci. 2005, 283 (8), 887-897. 342. Sardar, N.; Ali, M. S.; Kamil, M.; Kabir Ud, D., Phase behavior of nonionic polymer hydroxypropylmethyl cellulose: Effect of gemini and single-chain surfactants on the energetics at the cloud point. J. Chem. Eng. Data 2010, 55 (11), 4990-4994. 343. Alexandridis, P.; Holzwarth, J. F., Differential scanning calorimetry investigation of the effect of salts on aqueous solution properties of an amphiphilic block copolymer (poloxamer). Langmuir 1997, 13 (23), 6074-6081. 344. Naskar, B.; Ghosh, S.; Moulik, S. P., Solution behavior of normal and reverse triblock copolymers (pluronic L44 and 10R5) individually and in binary mixture. Langmuir 2012, 28 (18), 7134-7146. 345. Linseiss, LFA 1000/1000 HT Thermal diffusivity, thermal conductivity. Germany, 2016. 346. Smith, A. M.; Mancini, M. C.; Nie, S., Bioimaging: Second window for in vivo imaging. Nat. Nanotechnol. 2009, 4 (11), 710-711. 347. Pinto, A. M.; Gonçalves, I. C.; Magalhães, F. D., Graphene-based materials biocompatibility: A review. Colloids Surf., B 2013, 111, 188-202. 348. Kier, L. D.; Wagner, L. M.; Wilson, T. V.; Li, A. P.; Short, R. D.; Kennedy, G. L., Cytotoxicity of ethylene oxide/propylene oxide copolymers in cultured mammalian cells. Drug Chem. Toxicol. 1995, 18 (1), 29-41. 349. Ricci, E. J.; Lunardi, L. O.; Nanclares, D. M. A.; Marchetti, J. M., Sustained release of lidocaine from Poloxamer 407 gels. Int. J. Pharm. 2005, 288 (2), 235-244. 350. Park, H.; Park, K., Biocompatibility issues of implantable drug delivery sysems. Pharm. Res. 1996, 13 (12), 1770-1776. 351. Marcano, D. C.; Kosynkin, D. V.; Berlin, J. M.; Sinitskii, A.; Sun, Z.; Slesarev, A.; Alemany, L. B.; Lu, W.; Tour, J. M., Improved synthesis of graphene oxide. ACS Nano 2010, 4 (8), 4806-4814. 352. Yang, K.; Zhang, S.; Zhang, G.; Sun, X.; Lee, S. T.; Liu, Z., Graphene in mice: Ultrahigh in vivo tumor uptake and efficient photothermal therapy. Nano Lett. 2010, 10 (9), 3318-3323. 353. Zhang, Y.; Ali, S. F.; Dervishi, E.; Xu, Y.; Li, Z.; Casciano, D.; Biris, A. S., Cytotoxicity effects of graphene and single-wall carbon nanotubes in neural phaeochromocytoma-derived pc12 cells. ACS Nano 2010, 4 (6), 3181-3186.

343

354. Gollavelli, G.; Ling, Y. C., Multi-functional graphene as an invitro and invivo imaging probe. Biomaterials 2012, 33 (8), 2532-2545. 355. Hu, W.; Peng, C.; Luo, W.; Lv, M.; Li, X.; Li, D.; Huang, Q.; Fan, C., Graphene- based antibacterial paper. ACS Nano 2010, 4 (7), 4317-4323. 356. Chen, H.; Müller, M. B.; Gilmore, K. J.; Wallace, G. G.; Li, D., Mechanically strong, electrically conductive, and biocompatible graphene paper. Advanced Materials 2008, 20 (18), 3557-3561. 357. Agarwal, S.; Zhou, X.; Ye, F.; He, Q.; Chen, G. C. K.; Soo, J.; Boey, F.; Zhang, H.; Chen, P., Interfacing live cells with nanocarbon substrates. Langmuir 2010, 26 (4), 2244- 2247. 358. Malek, S. N. A.; Phang, C. W.; Ibrahim, H.; Abdul Wahab, N.; Sim, K. S., Phytochemical and Cytotoxic Investigations of Alpinia mutica Rhizomes. Molecules 2011, 16 (1), 583. 359. Souza, L. F.; Caputo, L.; De Barros, I. B. I.; Fratianni, F.; Nazzaro, F.; De Feo, V., Pereskia aculeata muller (Cactaceae) leaves: Chemical composition and biological activities. International Journal of Molecular Sciences 2016, 17 (9). 360. Nel, A.; Xia, T.; Mädler, L.; Li, N., Toxic potential of materials at the nanolevel. Science 2006, 311 (5761), 622-627. 361. Li, Y.; Yuan, H.; Von Dem Bussche, A.; Creighton, M.; Hurt, R. H.; Kane, A. B.; Gao, H., Graphene microsheets enter cells through spontaneous membrane penetration at edge asperities and corner sites. Proc. Natl. Acad. Sci. U. S. A. 2013, 110 (30), 12295-12300. 362. Oberdörster, G.; Oberdörster, E.; Oberdörster, J., Nanotoxicology: An emerging discipline evolving from studies of ultrafine particles. Environ. Health Perspect. 2005, 113 (7), 823-839. 363. Donaldson, K.; Tran, C. L., Inflammation caused by particles and fibers. Inhalation Toxicology 2002, 14 (1), 5-27. 364. Donaldson, K.; Stone, V.; Tran, C. L.; Kreyling, W.; Borm, P. J. A., Nanotoxicology. Occupational and Environmental Medicine 2004, 61 (9), 727-728. 365. Colvin, V. L., The potential environmental impact of engineered nanomaterials. Nat. Biotechnol. 2003, 21 (10), 1166-1170. 366. Hidaka, H.; Horikoshi, S.; Serpone, N.; Knowland, J., In vitro photochemical damage to DNA, RNA and their bases by an inorganic sunscreen agent on exposure to UVA and UVB radiation. Journal of Photochemistry and Photobiology A: Chemistry 1997, 111 (1-3), 205-213. 367. Moriguchi, T.; Yano, K.; Hokari, S.; Sonoda, M., Effect of Repeated Application of C60 Combined with UVA Radiation Onto Hairless Mouse Back Skin. Fullerene Sci. Technol. 1999, 7 (2), 195-209. 368. Vertegel, A. A.; Siegel, R. W.; Dordick, J. S., Silica nanoparticle size influences the structure and enzymatic activity of adsorbed lysozyme. Langmuir 2004, 20 (16), 6800-6807. 369. Karajanagi, S. S.; Vertegel, A. A.; Kane, R. S.; Dordick, J. S., Structure and function of enzymes adsorbed onto single-walled carbon nanotubes. Langmuir 2004, 20 (26), 11594- 11599. 370. Bhabra, G.; Sood, A.; Fisher, B.; Cartwright, L.; Saunders, M.; Evans, W. H.; Surprenant, A.; Lopez-Castejon, G.; Mann, S.; Davis, S. A.; Hails, L. A.; Ingham, E.; Verkade, P.; Lane, J.; Heesom, K.; Newson, R.; Case, C. P., Nanoparticles can cause DNA damage across a cellular barrier. Nat Nano 2009, 4 (12), 876-883. 371. David L. Nelson, M. M. C., Lehninger, Principles of biochemistry, sixth edition. W. H. Freeman and company: New York, 2013. 372. Barry Halliwell, J. M. C. G., Free radicals in biology and medicine, 5th edition. Oxford university press: United Kingdom, 2015. 373. Chatterjee, N.; Eom, H. J.; Choi, J., A systems toxicology approach to the surface functionality control of graphene-cell interactions. Biomaterials 2014, 35 (4), 1109-1127. 374. Dervishi, E.; Li, Z.; Watanabe, F.; Biswas, A.; Xu, Y.; Biris, A. R.; Saini, V.; Biris, A. S., Large-scale graphene production by RF-cCVD method. Chem. Commun. 2009, (27), 4061-4063. 375. Chu, K. F.; Dupuy, D. E., Thermal ablation of tumours: biological mechanisms and advances in therapy. Nat Rev Cancer 2014, 14 (3), 199-208. 376. Manthe, R. L.; Foy, S. P.; Krishnamurthy, N.; Sharma, B.; Labhasetwar, V., Tumor Ablation and Nanotechnology. Molecular pharmaceutics 2010, 7 (6), 1880-1898. 377. Dupuy, D. E.; DiPetrillo, T.; Gandhi, S.; Ready, N.; Ng, T.; Donat, W.; Mayo-Smith, W. W., Radiofrequency ablation followed by conventional radiotherapy for medically inoperable stage I non-small cell lung cancer. Chest 2006, 129 (3), 738-745. 378. Sabel, M. S., Cryo-immunology: A review of the literature and proposed mechanisms for stimulatory versus suppressive immune responses. Cryobiology 2009, 58 (1), 1-11. 379. Lubner, M. G.; Brace, C. L.; Hinshaw, J. L.; Lee Jr, F. T., Microwave tumor ablation: Mechanism of action, clinical results, and devices. Journal of Vascular and Interventional Radiology 2010, 21 (SUPPL. 8), S192-S203. 380. Haen, S. P.; Pereira, P. L.; Salih, H. R.; Rammensee, H. G.; Gouttefangeas, C., More than just tumor destruction: Immunomodulation by thermal ablation of cancer. Clinical and Developmental Immunology 2011, 2011. 381. Sonntag, P. D.; Hinshaw, J. L.; Lubner, M. G.; Brace, C. L.; Lee, F. T., Thermal Ablation of Lung Tumors. Surgical oncology clinics of North America 2011, 20 (2), 369-ix. 382. Paulet, E.; Aubé, C.; Pessaux, P.; Lebigot, J.; Lhermitte, E.; Oberti, F.; Ponthieux, A.; Calès, P.; Ridereau-Zins, C.; Pereira, P. L., Factors limiting complete tumor ablation by radiofrequency ablation. CardioVascular and Interventional Radiology 2008, 31 (1), 107- 115. 383. Ambrogi, M. C.; Fontanini, G.; Cioni, R.; Faviana, P.; Fanucchi, O.; Mussi, A., Biologic effects of radiofrequency thermal ablation on non–small cell lung cancer: Results of a pilot study. The Journal of Thoracic and Cardiovascular Surgery 131 (5), 1002- 1006. 384. Nguyen, C. L.; Scott, W. J.; Young, N. A.; Rader, T.; Giles, L. R.; Goldberg, M., Radiofrequency ablation of primary lung cancer: Results from an ablate and resect pilot study. Chest 2005, 128 (5), 3507-3511. 385. Feng, W.; Liu, W.; Li, C.; Li, Z.; Li, R.; Liu, F.; Zhai, B.; Shi, J.; Shi, G., Percutaneous microwave coagulation therapy for lung cancer. Chinese Journal of Oncology 2002, 24 (4), 388-390. 386. Wolf, F. J.; Grand, D. J.; Machan, J. T.; DiPetrillo, T. A.; Mayo-Smith, W. W.; Dupuy, D. E., Microwave Ablation of Lung Malignancies: Effectiveness, CT Findings, and Safety in 50 Patients. Radiology 2008, 247 (3), 871-879. 387. Wang, H.; Littrup, P. J.; Duan, Y.; Zhang, Y.; Feng, H.; Nie, Z., Thoracic masses treated with percutaneous cryotherapy: Initial experience with more than 200 procedures. Radiology 2005, 235 (1), 289-298. 388. Kawamura, M.; Izumi, Y.; Tsukada, N.; Asakura, K.; Sugiura, H.; Yashiro, H.; Nakano, K.; Nakatsuka, S.; Kuribayashi, S.; Kobayashi, K., Percutaneous cryoablation of small pulmonary malignant tumors under computed tomographic guidance with local

345 anesthesia for nonsurgical candidates. Journal of Thoracic and Cardiovascular Surgery 2006, 131 (5), 1007-1013. 389. Chou, S. S., Chemically Exfoliated MoS as Near-Infrared Photothermal Agents. Angew. Chem., Int. Ed. 2013, 52 (15), 4160-4164. 390. Shah, R., Eldridge, D., Palombo, E., Harding, I.,, Lipid Nanoparticles: Production, Characterization and Stability. Springer: 2015. 391. Doktorovova, S.; Souto, E. B.; Silva, A. M., Nanotoxicology applied to solid lipid nanoparticles and nanostructured lipid carriers - A systematic review of in vitro data. Eur. J. Pharm. Biopharm. 2014, 87 (1), 1-18. 392. Lane, T. A., Reduction in the toxicity of a component of an artificial blood substitute by supercritical fluid fractionation. Transfusion 1988, 28 (4), 375-378. 393. Clarke, M. S. F.; McNeil, P. L., Syringe loading introduces macro molecules into living mammalian cell cytosol. J. Cell Sci. 1992, 102 (3), 533-541. 394. Arechabala, B.; Coiffard, C.; Rivalland, P.; Coiffard, L. J. M.; Roeck-Holtzhauer, Y. D., Comparison of cytotoxicity of various surfactants tested on normal human fibroblast cultures using the neutral red test, MTT assay and LDH release. J. Appl. Toxicol. 1999, 19 (3), 163-165. 395. Redhead, M.; Mantovani, G.; Nawaz, S.; Carbone, P.; Gorecki, D. C.; Alexander, C.; Bosquillon, C., Relationship between the Affinity of PEO-PPO-PEO Block Copolymers for Biological Membranes and Their Cellular Effects. Pharm. Res. 2012, 29 (7), 1908-1918. 396. Wang, R.; Hughes, T.; Beck, S.; Vakil, S.; Li, S.; Pantano, P.; Draper, R. K., Generation of toxic degradation products by sonication of Pluronic(®) dispersants: implications for nanotoxicity testing. Nanotoxicology 2013, 7, 1272-1281. 397. Chang, Y.; Yang, S. T.; Liu, J. H.; Dong, E.; Wang, Y.; Cao, A.; Liu, Y.; Wang, H., In vitro toxicity evaluation of graphene oxide on A549 cells. Toxicol. Lett. 2011, 200 (3), 201-210. 398. Zhang, W.; Yan, L.; Li, M.; Zhao, R.; Yang, X.; Ji, T.; Gu, Z.; Yin, J. J.; Gao, X.; Nie, G., Deciphering the underlying mechanisms of oxidation-state dependent cytotoxicity of graphene oxide on mammalian cells. Toxicol. Lett. 2015, 237 (2), 61-71. 399. Gurunathan, S.; Han, J. W.; Abdal Dayem, A.; Eppakayala, V.; Kim, J. H., Oxidative stress-mediated antibacterial activity of graphene oxide and reduced graphene oxide in Pseudomonas aeruginosa. International Journal of Nanomedicine 2012, 7, 5901-5914. 400. Goodman, C. M.; McCusker, C. D.; Yilmaz, T.; Rotello, V. M., Toxicity of gold nanoparticles functionalized with cationic and anionic side chains. Bioconjugate Chem. 2004, 15 (4), 897-900. 401. Takahashi, H.; Niidome, Y.; Niidome, T.; Kaneko, K.; Kawasaki, H.; Yamada, S., Modification of gold nanorods using phosphatidylcholine to reduce cytotoxicity. Langmuir 2006, 22 (1), 2-5. 402. Nikfarjam, M.; Muralidharan, V.; Christophi, C., Mechanisms of focal heat destruction of liver tumors. Journal of Surgical Research 2005, 127 (2), 208-223. 403. Marku, D.; Wahlgren, M.; Rayner, M.; Sjoo, M.; Timgren, A., Characterization of starch Pickering emulsions for potential applications in topical formulations. Int. J. Pharm. 2012, 428 (1-2), 1-7. 404. Marto, J.; Gouveia, L.; Jorge, I. M.; Duarte, A.; Gonçalves, L. M.; Silva, S. M. C.; Antunes, F.; Pais, A. A. C. C.; Oliveira, E.; Almeida, A. J.; Ribeiro, H. M., Starch-based Pickering emulsions for topical drug delivery: A QbD approach. Colloids Surf., B 2015, 135, 183-192. 405. Chevalier, Y.; Bolzinger, M. A., Emulsions stabilized with solid nanoparticles: Pickering emulsions. Colloids Surf., A 2013, 439, 23-34. 406. Iwasaki, Y.; Takahata, Y.; Fujii, S., Self-setting particle-stabilized emulsion for hard- tissue engineering. Colloids Surf., B 2015, 126, 394-400. 407. Pickering, S. U., CXCVI. Emulsions. J. Chem. Soc., Trans. 1907, 91 (0), 2001-2021. 408. Ramsden, W., Separation of Solids in the Surface-Layers of Solutions and 'Suspensions' (Observations on Surface-Membranes, Bubbles, Emulsions, and Mechanical Coagulation). -- Preliminary Account. Proc. R. Soc. London 1903, 72 (477-486), 156-164. 409. Binks, B. P.; Lumsdon, S. O., Pickering emulsions stabilized by monodisperse latex particles: Effects of particle size. Langmuir 2001, 17 (15), 4540-4547. 410. Binks, B. P.; Clint, J. H., Solid wettability from surface energy components: Relevance to pickering emulsions. Langmuir 2002, 18 (4), 1270-1273. 411. Shan, Y.; Yu, C.; Yang, J.; Dong, Q.; Fan, X.; Qiu, J., Thermodynamically Stable Pickering Emulsion Configured with Carbon-Nanotube-Bridged Nanosheet-Shaped Layered Double Hydroxide for Selective Oxidation of Benzyl Alcohol. ACS Appl. Mater. Interfaces 2015, 7 (22), 12203-12209. 412. Wang, H.; Hobbie, E. K., Amphiphobic carbon nanotubes as macroemulsion surfactants. Langmuir 2003, 19 (8), 3091-3093. 413. Zhai, W.; Li, G.; Yu, P.; Yang, L.; Mao, L., Silver Phosphate/Carbon Nanotube- Stabilized Pickering Emulsion for Highly Efficient Photocatalysis. J. Phys. Chem. C 2013, 117 (29), 15183-15191. 414. Lin, K. Y. A.; Yang, H.; Petit, C.; Lee, W. D., Magnetically controllable Pickering emulsion prepared by a reduced graphene oxide-iron oxide composite. J. Colloid Interface Sci. 2015, 438, 296-305. 415. Guo, P.; Song, H.; Chen, X., Hollow graphene oxide spheres self-assembled by W/O emulsion. J. Mater. Chem. 2010, 20 (23), 4867-4874. 416. Zheng, Z.; Zheng, X.; Wang, H.; Du, Q., Macroporous graphene oxide-polymer composite prepared through pickering high internal phase emulsions. ACS Appl. Mater. Interfaces 2013, 5 (16), 7974-7982. 417. Woltornist, S. J.; Oyer, A. J.; Carrillo, J. M. Y.; Dobrynin, A. V.; Adamson, D. H., Conductive thin films of pristine graphene by solvent interface trapping. ACS Nano 2013, 7 (8), 7062-7066. 418. He, Y.; Wu, F.; Sun, X.; Li, R.; Guo, Y.; Li, C.; Zhang, L.; Xing, F.; Wang, W.; Gao, J., Factors that Affect Pickering Emulsions Stabilized by Graphene Oxide. ACS Appl. Mater. Interfaces 2013, 5 (11), 4843-4855. 419. P. Binks, B.; O. Lumsdon, S., Stability of oil-in-water emulsions stabilised by silica particles. Phys. Chem. Chem. Phys. 1999, 1 (12), 3007-3016. 420. Geim, A. K.; Grigorieva, I. V., Van der Waals heterostructures. Nature 2013, 499 (7459), 419-425. 421. Sham, A. Y. W.; Notley, S. M., A review of fundamental properties and applications of polymer-graphene hybrid materials. Soft Matter 2013, 9 (29), 6645-6653. 422. Huang, X.; Zeng, Z.; Zhang, H., Metal dichalcogenide nanosheets: preparation, properties and applications. Chem. Soc. Rev. 2013, 42 (5), 1934-1946. 423. Vucaj, N.; Quinn, M. D. J.; Baechler, C.; Notley, S. M.; Cottis, P.; Hojati-Talemi, P.; Fabretto, M. V.; Wallace, G. G.; Murphy, P. J.; Evans, D. R., Vapor phase synthesis of conducting polymer nanocomposites incorporating 2D nanoparticles. Chem. Mater. 2014, 26 (14), 4207-4213. 424. Britnell, L.; Ribeiro, R. M.; Eckmann, A.; Jalil, R.; Belle, B. D.; Mishchenko, A.; Kim, Y. J.; Gorbachev, R. V.; Georgiou, T.; Morozov, S. V.; Grigorenko, A. N.; Geim, A.

347

K.; Casiraghi, C.; Castro Neto, A. H.; Novoselov, K. S., Strong light-matter interactions in heterostructures of atomically thin films. Science 2013, 340 (6138), 1311-1314. 425. Quinn, M. D. J.; Du, J.; Boyd, B. J.; Hawley, A.; Notley, S. M., Lipid Liquid-Crystal Phase Change Induced through near-Infrared Irradiation of Entrained Graphene Particles. Langmuir 2015, 31 (24), 6605-6609. 426. Bardhan, R.; Lal, S.; Joshi, A.; Halas, N. J., Theranostic nanoshells: From probe design to imaging and treatment of cancer. Acc. Chem. Res. 2011, 44 (10), 936-946. 427. Kissel, T.; Li, Y.; Unger, F., ABA-triblock copolymers from biodegradable polyester A-blocks and hydrophilic poly(ethylene oxide) B-blocks as a candidate for in situ forming hydrogel delivery systems for proteins. Adv. Drug Delivery Rev. 2002, 54 (1), 99-134. 428. Ricardo, N. M. P. S.; Pinho, M. E. N.; Yang, Z.; Attwood, D.; Booth, C., Controlling the gelation of aqueous micellar solutions of ethylene-oxide-based block copoly(oxyalkylene)s. Int. J. Pharm. 2005, 300 (1-2), 22-31. 429. Yu, L.; Chang, G.; Zhang, H.; Ding, J., Temperature-induced spontaneous sol-gel transitions of poly(D,L-lactic acid-co-glycolic acid)-b-poly(ethylene glycol)-b-poly(D,L- lactic acid-co-glycolic acid) triblock copolymers and their end-capped derivatives in water. Journal of Polymer Science, Part A: Polymer Chemistry 2007, 45 (6), 1122-1133. 430. Yu, L.; Chang, G. T.; Zhang, H.; Ding, J. D., Injectable block copolymer hydrogels for sustained release of a PEGylated drug. Int. J. Pharm. 2008, 348 (1-2), 95-106. 431. Sagawa, N.; Shikata, T., Hydration Behavior of Poly(ethylene oxide)s in Aqueous Solution As Studied by Near-Infrared Spectroscopic Techniques. The Journal of Physical Chemistry B 2013, 117 (37), 10883-10888. 432. Saeki, S.; Kuwahara, N.; Nakata, M.; Kaneko, M., Upper and lower critical solution temperatures in poly (ethylene glycol) solutions. Polymer 1976, 17 (8), 685-689. 433. Matsuyama, A.; Tanaka, F., Theory of solvation-induced reentrant phase separation in polymer solutions. Phys. Rev. Lett. 1990, 65 (3), 341-344. 434. Shikata, T.; Okuzono, M.; Sugimoto, N., Temperature-dependent hydration/dehydration behavior of poly(ethylene oxide)s in aqueous solution. Macromolecules 2013, 46 (5), 1956-1961. 435. Yang, Y.; Miller, D. J.; Hawthorne, S. B., Toluene Solubility in Water and Organic Partitioning from Gasoline and Diesel Fuel into Water at Elevated Temperatures and Pressures. Journal of Chemical & Engineering Data 1997, 42 (5), 908-913. 436. Saien, J.; Akbari, S., Interfacial Tension of Toluene + Water + Sodium Dodecyl Sulfate from (20 to 50) °C and pH between 4 and 9. Journal of Chemical & Engineering Data 2006, 51 (5), 1832-1835. 437. Zeppieri, S.; Rodríguez, J.; López de Ramos, A. L., Interfacial Tension of Alkane + Water Systems. Journal of Chemical & Engineering Data 2001, 46 (5), 1086-1088. 438. Almgren, M.; Alsins, J.; Bahadur, P., Fluorescence quenching and excimer formation to probe the micellization of a poly(ethylene oxide)-poly(propylene oxide)-Poly(ethylene oxide) block copolymer, as modulated by potassium fluoride in aqueous solution. Langmuir 1991, 7 (3), 446-450. 439. Bahadur, P.; Li, P.; Almgren, M.; Brown, W., Effect of potassium fluoride on the micellar behavior of pluronic F-68 in aqueous solution. Langmuir 1992, 8 (8), 1903-1907. 440. Attwood, D.; Collett, J. H.; Tait, C. J., The micellar properties of the poly(oxyethylene) - poly(oxypropylene) copolymer Pluronic F127 in water and electrolyte solution. Int. J. Pharm. 1985, 26 (1-2), 25-33. 441. Jain, N. J.; Aswal, V. K.; Goyal, P. S.; Bahadur, P., Micellar structure of an ethylene oxide-propylene oxide block copolymer: A small-angle neutron scattering study. J. Phys. Chem. B 1998, 102 (43), 8452-8458. 442. Jain, N. J.; Aswal, V. K.; Goyal, P. S.; Bahadur, P., Salt induced micellization and micelle structures of' PEO/PPO/PEO block copolymers in aqueous solution. Colloids Surf., A 2000, 173 (1-3), 85-94. 443. Binks, B. P.; Lumsdon, S. O., Influence of particle wettability on the type and stability of surfactant-free emulsions. Langmuir 2000, 16 (23), 8622-8631. 444. Whitby, C. P.; Wanless, E. J., Controlling pickering emulsion destabilisation: A route to fabricating new materials by phase inversion. Materials 2016, 9 (8). 445. McAllister, M. J.; Li, J. L.; Adamson, D. H.; Schniepp, H. C.; Abdala, A. A.; Liu, J.; Herrera-Alonso, M.; Milius, D. L.; Car, R.; Prud'homme, R. K.; Aksay, I. A., Single sheet functionalized graphene by oxidation and thermal expansion of graphite. Chem. Mater. 2007, 19 (18), 4396-4404. 446. Sharma, R.; Bahadur, P., Effect of different additives on the cloud point of a polyethylene oxide-polypropylene oxide-polyethylene oxide block copolymer in aqueous solution. J. Surfactants Deterg. 2002, 5 (3), 263-268. 447. Bagwe, R. P.; Kanicky, J. R.; Palla, B. J.; Patanjali, P. K.; Shah, D. O., Improved drug delivery using microemulsions: Rationale, recent progress, and new horizons. Critical Reviews in Therapeutic Drug Carrier Systems 2001, 18 (1), 77-140. 448. Spernath, A.; Aserin, A., Microemulsions as carriers for drugs and nutraceuticals. Adv. Colloid Interface Sci. 2006, 128-130, 47-64. 449. Amar-Yuli, I.; Libster, D.; Aserin, A.; Garti, N., Solubilization of food bioactives within lyotropic liquid crystalline mesophases. Curr. Opin. Colloid Interfacial Sci. 2009, 14 (1), 21-32. 450. Larsson, K., Lyotropic liquid crystals and their dispersions relevant in foods. Curr. Opin. Colloid Interfacial Sci. 2009, 14 (1), 16-20. 451. Nazaruk, E.; Bilewicz, R.; Lindblom, G.; Lindholm-Sethson, B., Cubic phases in biosensing systems. Anal. Bioanal. Chem. 2008, 391 (5), 1569-1578. 452. Liu, Q.; Boyd, B. J., Liposomes in biosensors. Analyst 2013, 138 (2), 391-409. 453. Cherezov, V.; Clogston, J.; Misquitta, Y.; Abdel-Gawad, W.; Caffrey, M., Membrane protein crystallization in meso: Lipid type-tailoring of the cubic phase. Biophys. J. 2002, 83 (6), 3393-3407. 454. Kulkarni, C. V., In cubo crystallization of membrane proteins. In Advances in Planar Lipid Bilayers and Liposomes, 2010; Vol. 12, pp 237-272. 455. Phan, S.; Fong, W. K.; Kirby, N.; Hanley, T.; Boyd, B. J., Evaluating the link between self-assembled mesophase structure and drug release. Int. J. Pharm. 2011, 421 (1), 176-182. 456. Dong, Y. D.; Larson, I.; Barnes, T. J.; Prestidge, C. A.; Allen, S.; Chen, X.; Roberts, C. J.; Boyd, B. J., Understanding the interfacial properties of nanostructured liquid crystalline materials for surface-specific delivery applications. Langmuir 2012, 28 (37), 13485-13495. 457. Barauskas, J.; Johnsson, M.; Tiberg, F., Self-assembled lipid superstructures: Beyond vesicles and liposomes. Nano Lett. 2005, 5 (8), 1615-1619. 458. Barauskas, J.; Landh, T., Phase behavior of the phytantriol/water system. Langmuir 2003, 19 (23), 9562-9565. 459. Boyd, B. J.; Dong, Y. D.; Rades, T., Nonlamellar liquid crystalline nanostructured particles: Advances in materials and structure determination. J. Liposome. Res. 2009, 19 (1), 12-28. 460. Fong, W. K.; Hanley, T. L.; Thierry, B.; Kirby, N.; Waddington, L. J.; Boyd, B. J., Controlling the nanostructure of gold nanorod-lyotropic liquid-crystalline hybrid materials using near-infrared laser irradiation. Langmuir 2012, 28 (40), 14450-14460.

349

461. Fong, W. K.; Hanley, T. L.; Thierry, B.; Kirby, N.; Boyd, B. J., Plasmonic nanorods provide reversible control over nanostructure of self-assembled drug delivery materials. Langmuir 2010, 26 (9), 6136-6139. 462. Shah, J. C.; Sadhale, Y.; Chilukuri, D. M., Cubic phase gels as drug delivery systems. Adv. Drug Delivery Rev. 2001, 47 (2-3), 229-250. 463. Dong, Y. D.; Larson, I.; Hanley, T.; Boyd, B. J., Bulk and dispersed aqueous phase behavior of phytantriol: Effect of vitamin E acetate and F127 polymer on liquid crystal nanostructure. Langmuir 2006, 22 (23), 9512-9518. 464. Kirby, N. M.; Mudie, S. T.; Hawley, A. M.; Cookson, D. J.; Mertens, H. D. T.; Cowieson, N.; Samardzic-Boban, V., A low-background-intensity focusing small-angle X- ray scattering undulator beamline. J. Appl. Crystallogr. 2013, 46 (6), 1670-1680. 465. Oka, T.; Saiki, T.; Alam, J. M.; Yamazaki, M., Activation Energy of the Low-pH- Induced Lamellar to Bicontinuous Cubic Phase Transition in Dioleoylphosphatidylserine/Monoolein. Langmuir 2016, 32 (5), 1327-1337. 466. Seddon, J. M.; Templer, R. H., Chapter 3 Polymorphism of lipid-water systems. In Handbook of Biological Physics, 1995; Vol. 1, pp 97-160. 467. Popescu, G.; Barauskas, J.; Nylander, T.; Tiberg, F., Liquid crystalline phases and their dispersions in aqueous mixtures of glycerol monooleate and glyceryl monooleyl ether. Langmuir 2007, 23 (2), 496-503. 468. Hyde, S. T.; Andersson, S.; Ericsson, B.; Larsson, K., A cubic structure consisting of a lipid bilayer forming an infinite periodic minimum surface of the gyroid type in the glycerolmonooleat-water system. Z. Kristallogr. - New Cryst. Struct. 1984, 168 (1-4), 213- 219. 469. Dong, Y. D.; Tilley, A. J.; Larson, I.; Jayne Lawrence, M.; Amenitsch, H.; Rappolt, M.; Hanley, T.; Boyd, B. J., Nonequilibrium effects in self-assembled mesophase materials: Unexpected supercooling effects for cubosomes and hexosomes. Langmuir 2010, 26 (11), 9000-9010. 470. Fong, W. K.; Hanley, T. L.; Thierry, B.; Tilley, A.; Kirby, N.; Waddington, L. J.; Boyd, B. J., Understanding the photothermal heating effect in non-lamellar liquid crystalline systems, and the design of new mixed lipid systems for photothermal on-demand drug delivery. Phys. Chem. Chem. Phys. 2014, 16 (45), 24936-24953. 471. Corkery, R. W., A variation on Luzzati's soap phases. Room temperature thermotropic liquid crystals. Phys. Chem. Chem. Phys. 2004, 6 (7), 1534-1546. 472. Fong, W. K.; Hanley, T.; Boyd, B. J., Stimuli responsive liquid crystals provide 'on- demand' drug delivery in vitro and in vivo. J. Controlled Release 2009, 135 (3), 218-226. 473. Kulkarni, C. V., Nanostructural studies on monoelaidin-water systems at low temperatures. Langmuir 2011, 27 (19), 11790-11800. 474. Wu, Y.; Stefl, M.; Olzynska, A.; Hof, M.; Yahioglu, G.; Yip, P.; Casey, D. R.; Ces, O.; Humpolickova, J.; Kuimova, M. K., Molecular rheometry: direct determination of viscosity in Lo and Ld lipid phases via fluorescence lifetime imaging. Phys. Chem. Chem. Phys. 2013, 15 (36), 14986-14993. 475. Malmsten, M., Soft drug delivery systems. Soft Matter 2006, 2 (9), 760-769. 476. Zu, S. Z.; Han, B. H., Aqueous dispersion of graphene sheets stabilized by pluronic copolymers:Formation of supramolecular hydrogel. J. Phys. Chem. C 2009, 113 (31), 13651- 13657. 477. Xu, Y.; Sheng, K.; Li, C.; Shi, G., Self-assembled graphene hydrogel via a one-step hydrothermal process. ACS Nano 2010, 4 (7), 4324-4330. 478. Simões, S. M. N.; Veiga, F.; Torres-Labandeira, J. J.; Ribeiro, A. C. F.; Sandez- Macho, M. I.; Concheiro, A.; Alvarez-Lorenzo, C., Syringeable Pluronic-α-cyclodextrin supramolecular gels for sustained delivery of vancomycin. Eur. J. Pharm. Biopharm. 2012, 80 (1), 103-112. 479. Harada, A.; Li, J.; Kamachi, M., Double-stranded inclusion complexes of cyclodextrin threaded on poly(ethylene glycol). Nature 1994, 370 (6485), 126-128. 480. Liao, X.; Chen, G.; Liu, X.; Chen, W.; Chen, F.; Jiang, M., Photoresponsive pseudopolyrotaxane hydrogels based on competition of host-guest interactions. Angew. Chem., Int. Ed. 2010, 49 (26), 4409-4413. 481. Huh, K. M.; Ooya, T.; Lee, W. K.; Sasaki, S.; Kwon, I. C.; Jeong, S. Y.; Yui, N., Supramolecular-Structured Hydrogels Showing a Reversible Phase Transition by Inclusion Complexation between Poly(ethylene glycol) Grafted Dextran and α-Cyclodextrin. Macromolecules 2001, 34 (25), 8657-8662. 482. Li, J.; Harada, A.; Kamachi, M., Sol-gel transition during inclusion complex formation between α-cyclodextrin and high molecular weight poly(ethylene glycol)s in aqueous solution. Polym. J. 1994, 26 (9), 1019-1026. 483. Guo, C. G.; Wang, L.; Li, Y. K.; Wang, C. Q., Suprarmolecular hydrogels based on low-molecular-weight poly(ethylene glycol) and α-cyclodextrin. J. Appl. Polym. Sci. 2013, 129 (2), 901-907. 484. Ni, X.; Cheng, A.; Li, J., Supramolecular hydrogels based on self-assembly between PEO-PPO-PEO triblock copolymers and α-cyclodextrin. J. Biomed. Mater. Res., Part A 2009, 88 (4), 1031-1036. 485. Kataoka, T.; Kidowaki, M.; Zhao, C.; Minamikawa, H.; Shimizu, T.; Ito, K., Local and network structure of thermoreversible polyrotaxane hydrogels based on poly(ethylene glycol) and methylated α-cyclodextrins. J. Phys. Chem. B 2006, 110 (48), 24377-24383. 486. Hoare, T. R.; Kohane, D. S., Hydrogels in drug delivery: Progress and challenges. Polymer 2008, 49 (8), 1993-2007. 487. Fenyvesi, E.; Gruiz, K.; Verstichel, S.; De Wilde, B.; Leitgib, L.; Csabai, K.; Szaniszlo, N., Biodegradation of cyclodextrins in soil. Chemosphere 2005, 60 (8), 1001- 1008. 488. Nava-Ortíz, C. A. B.; Alvarez-Lorenzo, C.; Bucio, E.; Concheiro, A.; Burillo, G., Cyclodextrin-functionalized polyethylene and polypropylene as biocompatible materials for diclofenac delivery. Int. J. Pharm. 2009, 382 (1–2), 183-191. 489. Lakkakula, J. R.; Maçedo Krause, R. W., A vision for cyclodextrin nanoparticles in drug delivery systems and pharmaceutical applications. Nanomedicine 2014, 9 (6), 877-894. 490. Li, J.; Loh, X. J., Cyclodextrin-based supramolecular architectures: Syntheses, structures, and applications for drug and gene delivery. Adv. Drug Delivery Rev. 2008, 60 (9), 1000-1017. 491. Bromberg, L. E.; Ron, E. S., Temperature-responsive gels and thermogelling polymer matrices for protein and peptide delivery. Adv. Drug Delivery Rev. 1998, 31 (3), 197-221. 492. Osman, S. K.; Soliman, G. M.; Amin, M.; Zaky, A., Self-assembling hydrogels based on B-Cyclodextrin polymer and poly (ethylene glycol) bearing hydrophobic moieties for protein delivery. Int. J. Pharm. Pharm. Sci. 2014, 6 (7), 591-597. 493. Ma, X.; Zhou, N.; Zhang, T.; Guo, Z.; Hu, W.; Zhu, C.; Ma, D.; Gu, N., In situ formation of multiple stimuli-responsive poly[(methyl vinyl ether)-alt-(maleic acid)]-based supramolecular hydrogels by inclusion complexation between cyclodextrin and azobenzene. RSC Advances 2016, 6 (16), 13129-13136. 494. Weissleder, R., A clearer vision for in vivo imaging: Progress continues in the development of smaller, more penetrable probes for biological imaging. Nat. Biotechnol. 2001, 19 (4), 316-317.

351

495. Cvetkovic, A.; Picioreanu, C.; Straathof, A. J. J.; Krishna, R.; Van Der Wielen, L. A. M., Relation between pore sizes of protein crystals and anisotropic solute diffusivities. J. Am. Chem. Soc. 2005, 127 (3), 875-879. 496. Zhu, W.; Li, Y.; Liu, L.; Chen, Y.; Wang, C.; Xi, F., Supramolecular hydrogels from cisplatin-loaded block copolymer nanoparticles and α-cyclodextrins with a stepwise delivery property. Biomacromolecules 2010, 11 (11), 3086-3092. 497. Arunan, E.; Desiraju, G. R.; Klein, R. A.; Sadlej, J.; Scheiner, S.; Alkorta, I.; Clary, D. C.; Crabtree, R. H.; Dannenberg, J. J.; Hobza, P.; Kjaergaard, H. G.; Legon, A. C.; Mennucci, B.; Nesbitt, D. J., Defining the hydrogen bond: An account (IUPAC Technical Report). Pure Appl. Chem. 2011, 83 (8), 1619-1636. 498. Yu, L.; Zhang, H.; Ding, J., A subtle end-group effect on macroscopic physical gelation of triblock copolymer aqueous solutions. Angew. Chem., Int. Ed. 2006, 45 (14), 2232-2235. 499. Xue, R.; Xin, X.; Wang, L.; Shen, J.; Ji, F.; Li, W.; Jia, C.; Xu, G., A systematic study of the effect of molecular weights of polyvinyl alcohol on polyvinyl alcohol-graphene oxide composite hydrogels. Phys. Chem. Chem. Phys. 2015, 17 (7), 5431-5440.