Clemson University TigerPrints

All Dissertations Dissertations

8-2013 FACTORS CONTROLLING SYNTHESIS OF IRON OXIDE NANOPARTICLES AND THE EFFECT OF SURFACE CHARGE ON MAGNETIC HYPERTHERMIA Bin Qi Clemson University, [email protected]

Follow this and additional works at: https://tigerprints.clemson.edu/all_dissertations Part of the Materials Science and Engineering Commons

Recommended Citation Qi, Bin, "FACTORS CONTROLLING SYNTHESIS OF IRON OXIDE NANOPARTICLES AND THE EFFECT OF SURFACE CHARGE ON MAGNETIC HYPERTHERMIA" (2013). All Dissertations. 1173. https://tigerprints.clemson.edu/all_dissertations/1173

This Dissertation is brought to you for free and open access by the Dissertations at TigerPrints. It has been accepted for inclusion in All Dissertations by an authorized administrator of TigerPrints. For more information, please contact [email protected]. FACTORS CONTROLLING SYNTHESIS OF IRON OXIDE NANOPARTICLES AND THE EFFECT OF SURFACE CHARGE ON MAGNETIC HYPERTHERMIA ______

A Dissertation Presented to the Graduate School of Clemson University ______

In Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy Materials Science & Engineering ______

by Bin Qi August 2013 ______

Accepted by: Dr. Olin Thompson Mefford, Committee Chair Dr. John Ballato Dr. Christopher Kitchens Dr. Fei Peng ABSTRACT

Iron oxide nanoparticles (IONPs) have been widely studied in the theranostics application due to their promising magnetic properties, low cytotoxicity and attractive biocompatibility. Despite the numerous studies on the kinetic mechanisms of IONPs synthesis and thus the resulting size, shape and crystallinity; there are still considerable unsolved issues in quantitatively depicting the dependence between particle morphology and the reaction conditions.

To begin to explain some of these phenomena, the kinetic mechanism for the morphology and crystalline changes of IONPs with the ligand/precursor ratio in nanoparticle synthesis was investigated. During the synthesis of nanoparticles via thermal decomposition of iron precursors, the capping ligand-precursor ratio influences the resulting size of the iron oxide nanoparticles. As the molar ratio of aliphatic amines to iron precursor is increased, the average diameter of the synthesized iron oxide nanoparticles decreases. This trend is opposite to previously reported results. We investigated this phenomenon by independently varying the ligand chain length, the ligand-precursor molar ratio, and the degree of saturation of the aliphatic chain. Nuclear magnetic resonance spectra of precursor illustrated the presence of a primary amine peak before heating and the peak absence after heating, potentially indicating that the primary amine acts as reducing agent to promote the decomposition of the iron precursor. We hypothesize that the amine groups play a dominant role in the nucleation of the particles, while the chain length and degree of aliphatic saturation have only a minor effect on

ii particle size. The nanoparticles’ size and crystallinity were characterized with high resolution transmission electron microscopy, dynamic light scattering, and X-ray diffraction, and the magnetic properties were characterized by magnetometry.

Known ligand/precursor ratio effects on the IONPs size distribution, here in, we report that the resulting size of iron oxide nanoparticles synthesized by thermodecomposition could be modified by changing the ligand molecule length in the reaction solution. Transmission electron microscopy (TEM) and X-ray diffraction (XRD) measurements show that IONPs diameter increased from 16 nm to 25 nm with ligand length at a low molar ratio of ligand to iron precursor (1:2). However, there was no observable dependence of particle size on the ligand chain length at higher molar ratios

(30:1). In addition, particle size evolution differences with the reaction time between different ligand lengths in the solution were verified by dynamic light scattering (DLS) and AC susceptibility. To understand the mechanism of these phenomena and the factors contributing to the reaction, the kinetic process of the particle formation was simulated by the Monte Carlo algorithm. The goal was to investigate the effect of the length of the ligand molecule on the nucleation stage and growth of the particles at different ligand/precursor ratios and reaction time regimes. The subsequent results agreed well with the experimental findings suggesting our hypothesized mechanism of particle growth is correct.

To transfer the synthesized IONPs from a hydrophobic environment into an aqueous system for biomedical applications, we transferred particles from organic solvent

iii to aqueous solution by a one-step approach and investigated the heating efficiency changes of IONPs between in toluene and water in a bio-friendly alternating field (147 kHz and 41 kA/m) using AC calorimetery. Using specific absorption rate (SAR) to evaluating heating efficiency, a maximum SAR value was obtained with particles diameter of 22nm in both toluene and aqueous solution. Those particles with sizes greater or lower than the 22nm particles exhibited lower SAR values which suggests that the

22nm particles are at the optimal size at which the total contribution of the Brownian and the Néel relaxation mechanisms were maximized. It was observed that the SAR value is significantly affected by the concentration of iron in toluene, which is opposite to the published report. This could be due to the interparticle colloidal interactions in the AC field and form the localized ordering structure which could restrain the relaxation of

IONPs. A pH dependency of SAR was observed in aqueous solution, which confirms that the pH will tune the surface charge of the nanoparticles and further affect the colloidal stability and SAR value. The results above have the implications for IONPs size control and prediction in synthesis and optimization of IONPs colloidal performance in biomedical applications.

iv DEDICATION

First and foremost I would dedicate this dissertation to Dr. Thompson Mefford. It is my fortune along with my life that I could study and work in his group. Five-year research life gone fast, but it’s long enough for me to learn the way of living and the way of working. Thanks for bring me to the new world, and my memories shall never fade away.

I lovingly thank my wife Jia Emily Yu, who supported me each step of this long way. I know all your sacrifice for us, and I cannot accomplish this without your accompany. We have a long long way to go together in the future, and this is the best pass for our coming new journey.

I would also like to thank Steven Saville, Roland Stone, Dan D’Unger, Katie

Davis, Akeem Cruickshank and all other my colleagues in the lab. I’ll remember the laugh and tear we shared in the life and helps and efforts together we made in the work.

You all are my best friends and teammates.

Last but not least, I sincerely thank all my committee, Dr. Ballato, Dr. Kitchens and Dr. Peng. Your recommendations and guidance help me in the achieving of my progress. I could not even approach to my goal today without your commitment.

v ACKNOWLEDGEMENTS

Mr. Roland Stone at Clemson University performed NMR characterization in

Chapter 2.

Dr. Thomas Crawford and Mr. Longfei Ye at University of South Carolina performed VSM characterization in Chapter 2 and 4.

Dr. Cindi Dennis, Thomas Crawford, and Thompson Mefford, are responsible for much of the theory development discussed in Chapter 2.

Kim Ivey at Clemson University performed TGA and FTIR characterization in

Chapter 2.

Monte Carlo modeling calculations were performed using the Palmetto cluster high-performance computing (HPC) resource at Clemson University in Chapter 3.

Dr. Haijun Qian at Clemson University performed microtoming on the polystyrene encapsulated particles described in Chapter 4.

Dr. Thompson Mefford as my advisor in full of my graduate project assessed in design of project and revision of all publications.

vi TABLE OF CONTENTS

Page

TITLE PAGE… ...... i

ABSTRACT...... ii

DEDICATION ...... v

ACKNOWLEDGEMENTS ...... vi

LIST OF FIGURES ...... x

LIST OF TABLES ...... xix

CHAPTER 1 IRON OXIDE NANOPARTICLE FABRICATION AND MAGNETIC HYPERTHERMIA APPLICATIONS ...... 1

1.1 Introduction ...... 1 1.2 Magnetism overview ...... 2 1.2.1 Diamagnetism ...... 6 1.2.2 Paramagnetism ...... 7 1.2.3 Ferromagnetism ...... 8 1.2.4 Ferrimagnetism ...... 9 1.2.5 Antiferromagnetism ...... 9 1.2.6 Superparamagnetism ...... 10 1.3 Iron oxide nanomaterial synthetic approaches ...... 12 1.3.1 Top down method ...... 12 1.3.2 Bottom-up methods ...... 13 1.3.3 Morphology control ...... 25 1.4 Kinetics and thermodynamic process of nanocrystals formation and modeling method...... 27

vii Table of contents (Continued)

Page

1.4.1 Kinetics and thermodynamics process of nanocrystals formation ...... 27 1.4.2 Simulation of kinetic process in nanocrystals synthesis ...... 30 1.5 Magnetic hyperthermia application in tumor treatment ...... 39 1.5.1 Hyperthermia treatment ...... 39 1.5.2 Magnetic hyperthermia mechanism ...... 40

CHAPTER 2 INFLUENCE OF LIGAND-PRECURSOR MOLAR RATIO ON THE SIZE EVOLUTION OF MODIFIABLE IRON OXIDE NANOPARTICLES ...... 43

2.1 Introduction ...... 43 2.2 Experimental ...... 46 2.2.1 Materials ...... 46 2.2.2 Nanoparticles synthesis ...... 46 2.2.3 Characterization ...... 47 2.3 Results and Discussion ...... 49 2.4 Conclusion ...... 68

CHAPTER 3 EXPERIMENTAL OBSERVATION AND MONTE CARLO SIMULATION OF LIGAND MOLECULE LENGTH EFFECT ON THE IRON OXIDE NANOPARTICLES SIZE DISTRIBUTION ...... 69

3.1 Introduction ...... 69 3.2 Experimental ...... 71 3.2.1 Materials ...... 71 3.2.2 Synthesis of Iron Oxide Nanoparticles ...... 72 3.2.3 Characterization ...... 73 3.3 Modeling ...... 73 3.4 Results and discussion ...... 76

viii Table of contents (Continued)

Page

3.5 Conclusions ...... 91

CHAPTER 4 COLLOIDAL ENVIRONMENT DEPENDENCY OF SPECIFIC ABSORPTION RATE ON IRON OXIDE NANOPARTICLES ...... 92

4.1 Introduction ...... 92 4.2 Experimental ...... 94 4.2.1 Materials ...... 94 4.2.2 Nanoparticles synthesis ...... 95 4.2.3 Encapsulation of IONPs into polymer matrix ...... 96 4.2.4 Hydrophilic conversion of IONPs ...... 96 4.2.5 Characterization ...... 97 4.3 Results and discussion ...... 98 4.4 Conclusions ...... 109

CHAPTER 5 CONCLUSIONS ...... 111

CHAPTER 6 FUTURE WORK...... 113

6.1 IONPs complex synthesis ...... 114 6.2 Kinetics modeling including precursor chemical reactivity ...... 116 6.3 Anisotropic IONPs synthesis ...... 117 6.4 One pot synthesis of hydrophilic IONPs by thermodecompostiton procedure with tunable size...... 118 6.5 SAR change with field and frequency ...... 118

REFERENCES ...... 119

ix LIST OF FIGURES

Figure Page

1-1 Typical magnetization-field and susceptibility-temperature curve of diamagnetic material ...... 6

1-2 Typical magnetization-field and susceptibility-temperature curve of paramagnetic material ...... 7

1-3 Typical magnetization-field and susceptibility-temperature curve of ferromagnetic material ...... 8

1-4 Typical susceptibility-temperature curve of antiferromagnetic material ...... 10

1-5 Langevin function relation ...... 12

1-6 (a)TEM and (b) photomicrographs of the ground product of iron oxide (mill rotational speed = 600 rpm, milling time = 10 h). Reprinted with permission from [23], R. Arbain, et al.. Preparation of Iron Oxide Nanoparticles by Mechanical Milling, Minerals Engineering, 2011, 24 (1), 1-9. Copyright 2011 Elesvier...... 13

1-7 Schematic diagram of spray pyrolysis system. Reprinted with permission from[41] F. Iskandar, Nanoparticle Processing for Optical Applications – a Review Advanced Powder Technology 2009, 20 (4), 283-292. Copyright 2009 Elesvier...... 14

1-8 Schematic diagram of chemical vapor deposition reactor...... 15

x List of figures (Continued)

Figure Page

1-9 SEM image of iron oxides nanoparticles deposited on silica surface after exposure to the gas mixture Reprinted with permission from [28] Zhang et. al, Self-Assembled Patterns of Iron Oxide Nanoparticles by Hydrothermal Chemical-Vapor Deposition Applied Physics Letters 2001, 79 (25), 4207- 4209. Copyright 2009 American institute of physics...... 15

1-10 Schematic diagram of an autoclave reactor ...... 18

1-11 Schematic of a thermodecomposition reaction apparatus ...... 19

1-12 Schematic phase diagram, plotted as temperature versus oil fraction. Reprinted with permission from C. Okoli, et al., Comparison and Functionalization Study of Microemulsion-Prepared Magnetic Iron Oxide Nanoparticles Langmuir 2012, 28 (22), 8479-8485. Copyright. 2012 American Chemical Society...... 23

1-13 Schematic diagram of sol-gel processing steps...... 25

1-14 Schematic diagram of nucleation and growth stage in thermodecomposition process based on LaMer model Reprinted with permission from[77] Murray, et al., Synthesis and Characterizetion of Monodisperse Nanocrystals and Close-Packed Nanocrystal Assemblies Annu. Rev. Mater. Sci. 2000, 30, 68. Copyright 2000, Annual Reviews...... 28

xi List of figures (Continued)

Figure Page

1-15 Schematic diagram of magnetic nanoparticles size evolution process and magnetic moment changes with time. Reprinted with permission from [82]Kwon et al., Kinetics of Monodisperse Iron Oxide Nanocrystal Formation by "Heating-up" Process Journal of the American Chemical Society 2007, 129 (41), 12571-12584Simulation of kinetic process in nanocrystals synthesis. Copyright 2007, American Chemical Society...... 29

1-16 Algorithm diagram for a combined Monte Carlo-Brownian dynamics method. Reprinted with permission from Guarnieri F. Theory and Algorithms for Mixed Monte Carlo-Stochastic Dynamics Simulations J Math Chem 1995, 18 (1), 25-35. Copyright 1995, Springer...... 38

1-17 Schematic diagram of possible mechanisms for conversion of magnetic energy into heat Reprinted with the permission of [107]Kumar, et al., Magnetic Nanomaterials for Hyperthermia-Based Therapy and Controlled Drug Delivery Advanced Drug Delivery Reviews 2011, 63 (9), 789-808. Copyright 2011, Elsevier...... 41

2-1 TEM image of IONPs synthesized by decomposing 0.5 mmol Fe(acac)3 with

octadecylamine- Fe(acac)3 molar ratio of a) 1:1, b) 3:1, c) 10:1, and d) 60:1 in solution...... 50

2-2 TEM image of IONPs synthesized by decomposing 0.5 mmol Fe(acac)3 with

hexylamine- Fe(acac)3 molar ratio of a) 1:1, b) 3:1, c) 10:1, and d) 60:1 in solution ...... 50

xii List of figures (Continued)

Figure Page

2-3 TEM image of IONPs synthesized by decomposing 0.5 mmol Fe(acac)3 with

oleylamine- Fe(acac)3 molar ratio of a) 1:1, b) 3:1, c) 10:1, and d) 60:1 in solution ...... 51

2-4 HR-TEM image and diffraction patterns of IONPs synthesized by decomposing

0.5 mmol Fe(acac)3 with oleylamine- Fe(acac)3 molar ratio of 60:1 in solution ...... 51

2-5 Mean size of IONPs as a function of molar amount of (circle) oleylamine, (square) hexylamine and (triangle) octadecylamine used during the synthesis...... 52

2-6 TEM image of comparative sample and two reference samples synthesized by

decomposing 0.5 mmol Fe(acac)3 with (a)1.5 mmol of hexylamine and 13.8 ml of benzyl ether, (b)1.5 mmol of hexylamine and 10 ml of benzyl ether and (c) 30 mmol of hexylamine and 10 ml of benzyl ether in solution...... 54

2-7 DLS spectra of comparative sample and two reference samples synthesized by

decomposing 0.5 mmol Fe(acac)3 with 1.5 mmol of hexylamine and 13.8 ml of benzyl ether, 1.5 mmol of hexylamine and 10 ml of benzyl ether and 30 mmol of hexylamine and 10 ml of benzyl ether in solution...... 54

2-8 Schematic of the nucleation process at (a) low ligand-precursor ratio and (b) high ligand-precursor ratio ...... 56

xiii List of figures (Continued)

Figure Page

2-9 a) TGA and b) TGA derivative at the heat ramping rate of 3 oC/min from room temperature to 290 oC and isothermal for 10 min for: (black) the pure

Fe(acac)3 with no ligand, (red) pure octadecylamine only, and (blue) 1:1

molar ratio of Fe(acac)3 and octadecylamine...... 57

2-10 a) TGA and b) TGA derivative for the pure Fe(acac)3, hexylamine, oleylamine and octadecylamine only, and c) TGA and d) TGA derivative for 1:1 molar

ratio of Fe(acac)3 and hexylamine, oleylamine and octadecylamine, and e) DSC of all seven compounds above...... 58

2-11 a) TGA, b) TGA derivative and c) DSC with 1:1, 3:1, 10:1 and 60:1 molar ratio

of octadecylamine to Fe(acac)3 ...... 59

2-12 FTIR spectra of (a) pure octadecylamine and Fe(acac)3 at room temperature,

and (b)1:1 molar ratio of Fe(acac)3 and octadecylamine in benzyl ether before reaction and after heated up to 300oC and kept for 1h...... 60

2-13 NMR spectra of the solution containing 0.5 mmol Fe(acac)3 and 0.5 mmol octadecylamine in benzyl ether at room temperature: (a) before heating, (b) after heating ...... 62

2-14 XRD patterns for six different iron oxide nanoparticles samples, as a function of type and molar amount of ligand used during the synthesis. Peak locations correspond with that of a ferrite inverse spinel. Data are offset for clarity...... 65

xiv List of figures (Continued)

Figure Page

2-15 Hysteresis loop on the synthesized IONPs measured at 300 K in powder form with Kapton tape background removed and normalized to concentration of iron. (a) IONPs synthesized with octadecylamine, (b) IONPs synthesized with oleylamine...... 67

3-1 TEM images of IONPs obtained with ligand-precursor molar ratio of 1:2 (a) hexylamine (b) dodecylamine (c) octadecylamine, and 30:1 of (d) hexylamine (e) dodecylamine (f) octadecylamine...... 77

3-2 Mean size of IONPs changes with ligand molecule size in different ligand- precursor ratio...... 77

3-3 DLS data of IONPs synthesized by hexylamine and octadecylamine with ligand- precursor ratio of 1:2 and 30:1 ...... 78

3-4 A/C susceptibility plot of IONPs at different reaction time with ligand- precuorsor in 1:2 and 30:1 molar ratio of hexylamine and octadecylamine. 79

3-5 DLS curves of reagent solution containing 1:2 ligand-precursor ratio of a) hexylamine, b) dodecylamine and c) octadecylamine sampling at 125 °C, 175 °C, 200 °C, 250 °C, 300 °C, and 7.5 min, 15 min at 300 °C...... 79

3-6 TEM images of IONPs synthesized by hexylamine and octadecylamine with 1:2 and 30:1 ligand-precursor molar ratio sampling at different temperature, scale bars represent 50 nm...... 80

xv List of figures (Continued)

Figure Page

3-7 Schematic of effect of ligand length with different concentration on nanoparticle size. (a) long chain, low concentration, (b) short chain, low concentration, (c) long chain, high concentration, (d) short chain, high concentration...... 81

3-8 Images represent simulation results of nanocrystal size and number after 1.5 min, 4.5 min, 9 min and 15 min growth since the beginning of nucleation of solutions with ligand-precursor 1:2 and 30:1 of hexylamine and octadecylamine by Monte Carlo modeling ...... 83

3-9 Mean average diameter of nanoparticles acquired from simulation results...... 84

3-10 Histogram of particles diameter acquired from simulation results ...... 84

3-11 Histogram of particles velocity acquired from simulation results ...... 85

3-12 Images represent simulation results of nanocrystal size and number after 1.5 min, 4.5 min, and 9 min growth since the beginning of nucleation of solutions with ligand-precursor 1:2 and 30:1 of oleic acid by Monte Carlo modeling ...... 86

3-13 Histogram of particles diameter synthesized by oleic acid acquired from simulation results ...... 88

3-14 Histogram of particles velocity synthesized by oleic acid acquired from simulation results ...... 88

xvi List of figures (Continued)

Figure Page

3-15 XRD patterns for IONPs samples as a function of molecule size and molar ratio of ligand used during the synthesis. Peak locations correspond with that of a ferrite inverse spinel. Data are offset for clarity...... 89

3-16 Hysteresis loop on the synthesized IONPs measured at 300 K in powder form with Kapton tape background removed and normalized to concentration of iron. IONPs synthesized with (a) ligand-precursor 1:2 molar ratio, (b) ligand-precursor 10:1 molar ratio...... 90

4-1 TEM images of IONPs of a) 9 nm, b) 14 nm, c) 17 nm, d) 22 nm and e) 26nm in diameter ...... 98

4-2 Histograms of IONPs diameter in TEM images a) 9 nm, b) 14 nm, c) 17 nm, d) 22 nm and e) 26nm ...... 99

4-3. Schematic of AC calorimeter apparatus ...... 99

4-4 Specific absorption rate of IONPs in toluene with different size and concentration in the applied field of 41 kA/m and 147 kHz...... 101

4-5 Schematic diagrams of nanoparticles arrangement in different concentration with and without applied field a) low concentration, field off; b) low concentration, field on; c) high concentration, field off; d) high concentration, field on...... 103

xvii List of figures (Continued)

Figure Page

4-6 TEM images of IONPs encapsulated in polymer matrix in different concentration with and without applied field a) low concentration, field off; b) low concentration, field on; c) high concentration, field off; d) high concentration, field on...... 104

4-7 Hydrodynamic size of different size IONPs change with pH in aqueous solution ...... 105

4-8 Zeta potential of different size IONPs change with pH in aqueous solution .... 106

4-9 Specific absorption rate changes of 5 mg/ml IONPs in aqueous solution with pH from 4 to 12 in the applied field of 41 kA/m and 147 kHz...... 107

4-10 Hydrodynamic volume of different size IONPs in toluene...... 107

4-11 Specific absorption rate changes with hydrodynamic diameter of 5 mg/ml IONPs in aqueous solution with pH from 4 to 12 in the applied field of 41 kA/m and 147 kHz...... 108

4-12 X-ray diffraction patterns of oxidized IONPs with different sizes...... 109

xviii LIST OF TABLES

Table Page

1-1 Term of magnetism and conversions between SI and CGS unit ...... 5

2-1 Mean diameter and standard deviation (in units of nanometer) of synthesized IONPs as a function of type and ligand-precursor molar ratio...... 52

2-2 Amounts for each component used in preparing the comparative sample (varying solvent volume) and the two reference samples (varying molar amount of ligand)...... 53

2-3 DLS peak data of comparative sample and two reference samples synthesized by

decomposing 0.5 mmol Fe(acac)3 with (comparative)1.5 mmol of hexylamine and 13.8 ml of benzyl ether, (hexylamine-3:1)1.5 mmol of hexylamine and 10 ml of benzyl ether and (hexylamine-60:1) 30 mmol of hexylamine and 10 ml of benzyl ether in solution...... 55

2-4 Coherence length (CL) and ratio of coherence length to particle diameter of the IONPs synthesized with different types and ligand-precursor molar ratio. .. 66

3-1 IONPs particles size acquired from XRD patterns and calculated by Scherrer’s equation...... 90

4-1 Reaction and purification conditions of obtained nanoparticles ...... 100

xix CHAPTER 1 IRON OXIDE NANOPARTICLE FABRICATION AND

MAGNETIC HYPERTHERMIA APPLICATIONS

1.1 Introduction

Over the past decade, magnetic nanoparticles have received considerable attention because of their potential to revolutionize biomedical areas via their application.

Magnetic nanoparticles show unique properties such as superparamagnetism, high field irreversibility, and high saturation field due to their finite size and high specific surface area[1-3]. Thanks to their remarkable physical and chemical properties compared with their bulk counterparts, magnetic nanoparticles could be used as magnetic resonance imaging (MRI) contrast enhancement agents, treatment media in magnetic hyperthermia, drug delivery vehicles, and bioseparation carriers[1, 3].

Iron oxide nanoparticles are a promising material in biomedical application due to the material’s low cytotoxicity, high biocompatibility, and outstanding magnetic properties among the numerous magnetic material candidates. As a result, these particles have been widely investigated from the synthesis, to colloidal surface modification, and further to in vitro and in vivo biomedical tests[4-6]. From this work, the most pivotal factor that affects iron oxide nanoparticles biomedical performance is the magnetic properties of the nanoparticles, which is significantly dependent upon the particles size, shape, oxidation state, and crystallinity. Therefore, exploration of the particle formation

1 mechanisms is critical for the successful synthesis of finely controlled nanoparticles to be used in theranostic applications.

Despite the intensive motivation for the investigation of the mechanisms and the kinetics in the synthesis of these materials, there is still a considerable knowledge gap in relationship between the obtained particles size, shape, and crystallinity, and the reagent amount and reaction conditions, not to mention, the ability to precisely predict the particles morphology before reaction.

This dissertation is focused on the morphology control in the synthesis of iron oxide nanoparticles, the kinetic mechanisms of the nucleation and crystal growth process during the fabrication process, in order to control and predict particles size and shape.

These materials were further applied to a study on the magnetic hyperthermia properties to investigation how various morphology properties affect the iron oxide nanoparticles performance biomedical applications.

1.2 Magnetism Overview

The origin of magnetism mainly arises from the orbital and spin motions of electrons and the behavior of the interactions of the electrons. The nuclear magnetic moments of the nuclei in the material, a vital source in the application of magnetic resonance imaging (MRI) and nuclear magnetic resonance (NMR), contribution to net magnetization is negligible[7-8].

2 Main magnetic vectors include magnetic field, H, magnetization, M, and the magnetic induction, B. All three vectors could be represented in centimeter, gram, second units (CGS) or international system units (SI), which is abbreviated from French: Le

Système international d'unités. Because magnetostatics could be presented in two different ways: 1. fictitious magnetic poles (CGS), 2. current sources (SI)[9].

Consider a loop of radius r and current i, roughly equivalent to an atom with orbiting electrons, a magnetic field H at the center of the loop is given by

(1-1)

The magnetic moment m of the loop will be

(1-2)

Therefore the intensity of magnetization M, defined as magnetic moment per unit volume, could be written as

(1-3)

which has the same units as H. Mass magnetization, σ, is the magnetic moment per unit mass

(1-4)

3 Susceptibility is a common approach of determining the category of different magnetic materials. The dimensionless unit (volume) susceptibility is defined as the ratio of magnetization to magnetic field.

(1-5)

whereas the mass susceptibility is

⁄ (1-6)

There is another way of defining magnetic field using B, magnetic induction by magnetic field generated from an electric current loops with area A (Amperian loop):

. B could be determined by force F applied onto a particle with charge, q, ∮ and velocity, v, in an electric field using Lorentz force law

(1-7)

The relationship between B, H and M is given by

(1-8)

−7 -2 -2 in SI, where μ0 is vacuum permeability and equal to 4π×10 m·kg·s ·A . In CGS, B, H and M are numerically equal to one another, and B and H are used interchangeably due to

μ0 is unity in CGS units:

4 (1-9) but each have their own arbitrary unit names (Gauss, Oersted, and emu/cm3). B is proportional to H in a vacuum (outside of a material) where M equals zero (B = 0H in

SI, or B = H in CGS). The common magnetism terms in two unit systems and conversion factors have been summarized in Table 1-1.

conversion term of magnetism symbol SI unit CGS unit (SI/CGS)

magnetic field H A/m Oersted 4π/103

magnetic induction B Tesla (kgA-1s-2) Gauss 104

(volume) magnetization M A/m emu/cm3 10-3

mass magnetization σ Am2/kg emu/g 1

-1 (volume) susceptibility χv dimensionless dimensionless (4π)

mass susceptibility χ m3/kg emu/Oe·g 103/4π

magnetic moment m Am2 emu 103

-2 -2 −7 -1 vacuum permeability μ0 H/m (mkgs A ) dimensionless (4π×10 )

Table 1-1: Term of magnetism and conversions between SI and CGS unit

The type of magnetism can be classified into five major groups: 1. Diamagnetism

2. Paramagnetism 3. Ferromagnetism 4. Ferrimagnetism 5. Antiferromagnetism.

5 Diamagnetism and paramagnetism contain no collective magnetic interactions and therefore no magnetic domains in the materials. The other three groups have long-range magnetic order (magnetic domains) when the temperature is below a certain level. In general, a material exhibiting ferromagnetism or ferrimagnetism is classified as magnetic material and materials in the other groups are considered as nonmagnetic[10].

1.2.1 Diamagnetism

Diamagnetism is the intrinsic property of all matter. It is due to the magnetic moments generated by procession of the electron orbital that were aligned opposite to the applied field. Therefore, a diamagnetic material has no net magnetic moments but can slightly repel the applied field. The typical characteristic of diamagnetic material is temperature independent susceptibility. The M-H curve and χ-T curve is like:

Figure 1-1 Typical magnetization-field and susceptibility-temperature curve of diamagnetic material

6 Typical diamagnetic materials are bismuth, silver, copper, lead, quartz and water.

Superconductors may be considered as perfect diamagnets, since all fields are expeled from them[11].

1.2.2 Paramagnetism

The unpaired electrons existing in the paramagnetic materials would align their magnetic moment freely and reinforce the applied field, while magnetization is zero if the field is removed. Paramagnetism susceptibility follows Curies’ Law, which is inversely proportional to the temperature, which is due to the moments aligning efficiency is opposed by thermo fluctuation. Typical diamagnetic materials are metal element, including tungsten, aluminium and sodium. Ferromagnetic and antiferromagnetic materials can have paramagnetic behavior once temperature is over their Curie or Néel temperature[10, 12].

Figure 1-2 Typical magnetization-field and susceptibility-temperature curve of paramagnetic material

7 1.2.3 Ferromagnetism

Ferromagnetic materials, like iron, nickel and cobalt, exhibit strong magnetic moments even without an applied field. This is due to all the localized electrons having the tendency to maintain a parallel arrangement (magnetic domain). The intrinsic characteristics of ferromagnetic materials are a) spontaneous magnetization b) magnetic hysteresis and c) Curie temperature. Magnetization of ferromagnetic materials is temperature dependent and will decrease to zero when temperature passes the Curie point.

Figure 1-3 Typical magnetization-field and susceptibility-temperature curve of ferromagnetic material

In an applied field, the magnetization has an upper limit called saturation magnetization, beyond which, a magnetic field cannot raise the magnetization further, because the magnetic domains of substance have been sufficiently oriented. Once ferromagnetic materials have been magnetized, there will be a magnetization remanence left even if

8 field decreases to zero. To remove the magnetization residue to zero, a reverse magnetic field (coercive field) needs to be applied, the amplitude of this field is described as coercivity[13].

1.2.4 Ferrimagnetism

Ferrimagnetism exist in ionic compounds like oxides, which have two sublattices.

The electron spin and the magnetic moment of two sublattices are antiparallel aligned but unequal. This inequality is usually due to the different ions in sublattices (such as Fe2+,

Fe3+). Ferrimagnetism has almost the same magnetic behavior as ferromagnetism, but possess their own magnetization compensation zone/point, at this point ferrimagnetic substances reveal zero magnetic moment even below the Curie temperature due to the two sublattices having equal moments. Typical ferrimagnetic materials include magnetite, maghemite, and other metal oxides like cobalt, nickel and manganese oxides[14].

1.2.5 Antiferromagnetism

Antiferromagnets have a zero net magnetic moment, since magnetic moments of the two sublattices is antiparallel and equal to each other. The susceptibility follows

Curie-Weiss law for paramagnets above the Néel temperature. Typical materials are hematite, chromium and nickel oxide[10, 15].

9

Figure 1-4 Typical susceptibility-temperature curve of antiferromagnetic material

1.2.6 Superparamagnetism

Ferromagnets and ferrimagnets may response to the applied field similar to paramagnetic materials with much greater susceptibility. This magnetic behavior is named as superparamagnetism. Superparamagnetism occurs when a substance’s scale is reduced to a single domain, which is usually below 50 nm[16-18]. In this size scale, the magnetic moment of the particles will invert with time due to the thermal fluctuation. The mean time between two invertion states is called the Néel relaxation time N and it follows Néel-Arrhenius equation[19]

(1-10)

-9 -10 where 0 is attemping time which is in the scale of 10 -10 second, K is magnetic anisotropy energy density of materials, V is volume of materials, kB is Boltzmann constant and T is temperature.

10 The magnetization of superparamagnetic materials is zero with no applied field when measurement time  is larger than N. If  is smaller than N, the original ferro- or ferri- magnetic behavior of material would be observed. The blocking temperature could be derived from Néel-Arrhenius equation[20]:

(1-11)

which is temperature to drive N =  when keeping  as constant. For a typical laboratory measurement, (ln /0) is in the order of 20-25. Magnetization of superparamagnetic materials will follow S-shape function in the applied field which could be simplified as

Langevin function relation if particles are monodisperse and temperature T≥KV/kB[21]:

(1-12)

where n is the particles density, 0 is vacuum permeability, m is the magnetic moment, H is the applied field and L is Langevin function[22]

(1-13)

And suscptibility is given by

(1-14)

11

Figure 1-5 Langevin function relation

1.3 Iron oxide nanomaterial synthetic approaches

1.3.1 Top down method

The top down process is a type of method to produced nanoparticles from attrition in a mechanical device, generally described as “milling” solid substrates. The approach includes shaker mills, planetary ball mills, and attritor mills, based on the difference of milling device and media. The products usually have broad size distribution which can range from 10nm to 1000nm with varied morphology. Top down method could be used in preparation of nanocomposites and nano-grained bulk materials due to its convenient and promising process as well as massive yield rate[23-24].

12 a b

Figure 1-6 (a)TEM and (b) photomicrographs of the ground product of iron oxide (mill rotational speed = 600 rpm, milling time = 10 h). Reprinted with permission from [23], R. Arbain, et al..

Preparation of Iron Oxide Nanoparticles by Mechanical Milling, Minerals Engineering, 2011, 24 (1),

1-9. Copyright 2011 Elesvier.

1.3.2 Bottom-up methods

The bottom-up methods can be classified into (1)gas phase synthesis and (2) liquid phase synthesis. Gas phase synthesis approaches mainly include pyrolysis (spray pyrolysis) and chemical vapor deposition (CVD), whereas liquid phase synthesis are including coprecipitation, micellar template fabracation, thermodecomposition, hydrothermal, and sol-gel processes[14, 25-41].

1.3.2.1 Pyrolysis

Pyrolysis is a process that atomizes a solution to areosol and heats the droplets in the space where the constituents react to produce solid particles. A typical schematic diagram is as shown below. The atomizer disperses droplets into a gas. The furnace can work as both a dryer and a reactor. The produced solid nuclei will be tranfered and grown

13 in a precipitator. Park, et al.,[42] reported a procedure to fabricate iron-carbon, cobalt- carbon magnetic particles utilizing a pyrolysis method. The obtained magnetic particles were effectively prevented from oxidation by coating with a graphite shell. The pyrolysis process was employed to synthesize nanoparticles with a narrow size distribution but requires extreme synthesis condiditons (i.e. particles are produced near 800 ºC).

Nonetheless, this is the only viable method to prouduce certain nanomaterials. For example, Ogi, et al.,[43] obtained GaN nanoparticles of 23.41.7nm diameter by spray pyrolysis.

Figure 1-7 Schematic diagram of spray pyrolysis system. Reprinted with permission from[41] F.

Iskandar, Nanoparticle Processing for Optical Applications – a Review Advanced Powder Technology

2009, 20 (4), 283-292. Copyright 2009 Elesvier.

14 1.3.2.2 Chemical vapor deposition

Chemical vapor deposition is a type of inert gas condensation method, in a typical synthesis process, a carrier gas stream with precursors are delivered continuously to a vacuumed reaction chamber at high temperature (usually above 900 °C) In the reaction, precursor is decomposed and deposited on the substrate to form uniform high purity nanocrystal clusters. Zhang, et al.,[28] synthesized iron oxide nanoparticle clusters by

Figure 1-8 Schematic diagram of chemical vapor deposition reactor.

Figure 1-9 SEM image of iron oxides nanoparticles deposited on silica surface after exposure to the gas mixture Reprinted with permission from [28] Zhang et. al, Self-Assembled Patterns of Iron Oxide Nanoparticles by Hydrothermal Chemical-Vapor Deposition Applied Physics Letters 2001, 79 (25), 4207-4209. Copyright 2009 American institute of physics. chemical vapor deposition method using ferrocene (Fe(C5H5)2), water, and xylene at

15 1000 °C. Low fractions of precursor in the delivery gas and prompt diffusion rate of the nuclei in the reactor are pivotal factors for reaction success[44].

1.3.2.3 Coprecipitation

Coprecipitation is a convient approach to fabricate nanoparticles from metal salts, dual salts usually, in aqueous solutions by the addition of basic solutions at ambient or raised temperature. In iron oxide nanoparticles synthesis, Fe3+ and Fe2+ are employed with a typical stoichiometric ratio of 2:1, and follows the reaction:

(1-15)

The morphology of nanoparticles could be affected by the cation of base, anion of iron salt, temperature, pH, and addition rate. Coprecipitation is a robust method carried out in an eco-friendly reaction condition and easy to scale up for industrial production.

The drawback, however, is the products are polydisperse in size[45].

Babes and co-workers[46] reported using iron chloride as the metal salt, tetramethylammonium hydroxide (TMAOH) and ammonia as the base, to synthesize iron oxide nanoparticles in aqueous solution. The mean average size of the nanoparticles in this procedure could be tuned from 3.5 nm to 6.5 nm by adjusting the injecting rate of iron salt solution and type of base. Tronc, et al.,[47], and Iida[48] found that mean average size of iron oxide nanoparticles will increase from 9 nm to 37 nm by raising the

16 ferrous fraction in total in salt from 33% to 100% which further affected their satuation magnetization of the particles. Gupta[49] and Kim[50] found nitrogen bubbling in the coprecipitation process can not only prevent iron oxide nanoparticles from oxidation but decrease particles size.

1.3.2.4 Autoclave methods

Autoclave, also called the hydrothermal method, is a fabrication approach to obtain hydrophilic nanoparticles. These reactions are performed using alcoholic or aqueous media in a high pressure vessel, where the reaction temperature could reach above the ambient pressure boiling point of the solvent due to the high pressure environment. The autoclave method is a promissing ways of synthesizing nanocrystals of multiple materials or single crystal with better crystallinity than those from other processes. Xu and co-workers[51] investigated factors that would tune the morphology of autoclave synthesized hematite nanoparticles. Based on their research, particles size distribution and aggregation could be reduced by adding polyvinyl alcohol (PVA) in the solution. Liang, et al.,[52] reported the shape and phase of the iron oxide nanoparticles could be tuned by varing the solvent volume ratio between the ethanol and water while using FeSO4·(NH4)2SO4·6H2O and oleic acid as reagents. Jia and co-workers[53] fabricated single crystal iron oxide nanotubes using a FeCl3 solution in the presence of

NH4H2PO4 by the autoclave method at 220 °C for 48 h.

17

precursor and solvent

Figure 1-10 Schematic diagram of an autoclave reactor

1.3.2.5 Thermodecomposition

One of the most effective ways for synthesizing monodisperse nanoparticles with tailorable size is decomposing iron organic precursors at high temperature in nonaqueous media.[1]

Alivisatos, et al.,[54] reported that maghemite particles were obtained by decomposing FeCup3 (Cup: N-nitrosophenylhydroxylamine) in long chain amines at 250-

300 °C through precursor injection method. This injection method involves the iron precursors being injected into hot solvent, thus triggering the nucleation of particles at a precise moment. The resulting synthesized particles ranged from 4 to 10 nm in diameter and were suspended in an organic solvent. Hyeon, et al.,[55] also synthesized maghemite

18 using Fe(CO)5 as precursor and trimethylamine oxide as mild oxidant in organic solvent though the injection procedure. Sun and co-workers[56] prepared iron oxide nanoparticles with size from 4-20 nm using iron(III) acetylacetonate (Fe(acac)3) as precursor in dibenzyl ether in the presence of 1,2-hexadecandiol as reducing agent as well as oleic acid and oleylamine as capping ligands at high temperature(265-300 °C).

Figure 1-11 Schematic of a thermodecomposition reaction apparatus

Park, et al.,[57] successfully synthesized the monodisperse iron oxide nanoparticles in large scale by two-step procedure. In their reaction, iron-oleate was prepared using FeCl3 and sodium oleate in the mixture solution of ethanol, water and hexane at 70 °C, then iron-oleate complex was then decomposed at 320 °C in octadecene in the presence of oleic acid. The use of iron chloride salts as precursors were proposed by several groups

19 due to their nontoxicity. However, all these particles can only be suspended in organic solvent because of their hydrophobic surface coatings.

Sun and co-workers[56] described the procedure to remove ligands from magnetic nanoparticles in suspension of tetramethylammonium 11-aminoundecanoate in dichloromethane and dispersed the particles in an aqueous solution with phosphate buffer at neutral pH. Li, et al.,[58] reported a method to produce water-dispersible iron oxide nanoparticles in acidic or basic environment by decomposing Fe(acac)3 or FeCl3 in refluxing 2-pyrrolidone. 2-pyrrolidone in this case is both a high boiling point solvent as well as a capping ligand. Similarly, Li[59] and Hu[60] synthesized water-suspendable polyethylene glycol (PEG) coated nanoparticles by using monocarboxyl-terminated PEG or ,-dicarboxyl-terminated PEG as a capping ligands. Jun, et al.,[2] demonstrated formation of monodisperse iron oxide nanoparticles with tunable size of 4,6,9, and 12 nm using Fe(acac)3 as precursor and coating with 2,3-dimercaptosuccinc acid (DMSA) by ligands exchange method to stabilize in phosphate buffered aqueous solution. Cai and co- workers[61] directly synthesized water dispersible iron oxide nanoparticles by decomposing Fe(acac)3 in the presence of triethylene glycol which served as the reducing agent, high boiling point solvent, and stabilizer. The prepared particles were nearly monodisperse and could be suspended in aqueous solution without extra surfactant.

However, the drawback of this synthetic route is that only one particle size can be produced. Recently, Young, et al.,[62] reported that highly monodisperse iron oxide nanoparticles coating with oleylamine only were obtain by using Fe(acac)3 as precursor, benzyl ether as solvent, oleylamine as capping ligands. They conjugated anticancer drug

20 Methotrexate onto the particles by an available linker trichloro-s-triazine due to the capping ligands oleylamine are much easier to be replaced by the linker than other ligands such as oleic acid.

Besides iron oxide, many other metallic and metallic oxides nanoparticles were synthesized by thermo-decomposition procedure as well. Park, et al.,[63] reported synthesis of monodisperse MnO and CoO particles using metallic-oleate complex as precursor. Pure iron nanoparticles with narrow size distribution were obtained through the same method while heating above 380 °C. Zeng and co-workers[64] demonstrated a method to prepare monodisperse 3 nm FePt bimetallic core/shell nanoparticles coated with 3 nm iron oxide shell by decomposing Fe(acac)3. Sun, et al.,[56] synthesized

MFe2O4 (M=Co, Mn) Magnetic nanoparticles using Co(acac)3 or Mn(acac)3 partially substituted for Fe(acac)3 in 1:2 ratio as precursor and benzyl ether as solvent in the presence of 1,2-hexadecandiol, oleic acid and oleylamine. Kim, et al.,[37] modified the procedure from Sun's work and synthesized monodisperse MFe2O4 (M=Mn, Co, Ni) magnetic nanoparticles by using lauric acid and lauric amine as alternative ligands instead of oleic acid and oleylamine. Gyergyek, et al.,[65] synthesized 8 nm CoFe2O4 through various methods and compared their different magnetic properties due to magnetocrystalline anisotropy and large magneto-optical coefficients.

1.3.2.6 Micellar template

Micellar template, or microemulsion, is a universal and prompt method of synthesizing nanoparticles. Microemulsion is a liquid system containing oil and water

21 two immiscible component and surfactant molecules to keep the whole system as one phase. This system is transparent, isotropic and thermodynamically stable. Micellar template can be classified into oil-in-water micelles, water-in-oil inverse micelles and bicontinuous structures depending on the oil/water ratio as well as hydrophobic- hydrophilic property of the surfactant[66].

In nanoparticles synthesis, water-in-oil inverse micellar template method has been widely applied. Santra and co-workers [67] reported a robust procedure utilizing inverse micellar template to synthesis monodisperse bare and silica coated iron oxide nanoparticle with particle diameters smaller than 5 nm. Okoli, et al.,[68] found the iron oxide nanoparticles synthesized by water-in-oil inverse micelles template had over 3 times higher saturation magnetization than the particles obtain by oil-in-water template due to the larger particles size, higher crystallinity and anisotropic property.

22

Figure 1-12 Schematic phase diagram, plotted as temperature versus oil fraction. Reprinted with permission from C. Okoli, et al., Comparison and Functionalization Study of Microemulsion- Prepared Magnetic Iron Oxide Nanoparticles Langmuir 2012, 28 (22), 8479-8485. Copyright. 2012 American Chemical Society.

23 1.3.2.7 Sol-gel

Sol-gel approach is a wet chemical method, which synthesis a sol first and followed by conversion of the sol into a gel. The applied chemicals containing an inorganic precursors, typically a metal alkoxide M(OR), and an organic solvent. A typical sol-gel reaction includes two steps: hydrolysis and condensation. Hydrolysis is the process where the metal alkoxide reacts with a water molecule to obtain a metal hydroxide. The condensation process is where the obtained metal hydroxide condenses with remaining metal alkoxides or other metal hydroxide molecules to build a metal- oxide-metal 3-D network (gel). Then, gel will is processed by aging and drying stages to remove the liquid phase. Dehydration comes after the drying stage to remove the remaining hydroxyl groups. To prevent rehydration, the dehydration reaction temperatures can reach up to 800 °C. The final stage is to decompose the gels at high temperatures to collapse porous structure and organic residues[38].

Lu and co-workers[69] modified iron oxide surface properties via the sol-gel method to coat silica shells on the particles by using tetraethyl orthosilicate as precursor.

The thickness of the shell could be tuned by the precursor amount. Bersani and co- workers[38] synthesized iron oxide thin films and powders via sol-gel route using

Fe(NO3)3·9H2O as the precursor by dissolving in 2-methoxyethanol and acetylacetone and annealing at 500 °C in the air for 1h. A homogeneous hematite phase was obtained by this procedure, without the presence of a maghemite phase.

24

Figure 1-13 Schematic diagram of sol-gel processing steps.

Cui, et al.,[70] fabricated magnetite, maghemite, and hematite nanoparticles with diameters between 5 nm and 10 nm in a large scale at low temperature using iron(II) chloride tetrahydrate and propylene oxide in ethanol to synthesize the sol and produced three different phases by different aging and heat treatment processes. This conventional procedure could effectively narrow the size distribution of particles obtained by sol-gel method and easy to scale up for industrial application.

1.3.3 Morphology control

Size and shape tunable magnetic nanoparticles were synthesized by changing different reaction conditions in various reaction systems. For investigating size and shape changes of magnetic nanoparticles vs. reaction condition, some previous work has been

25 reported. Sun, et al.,[71] obtained 8 nm monodisperse iron oxide nanoparticles via 6 nm iron oxide nanoparticles as seeds using Fe(acac)3 as precursor in benzyl ether with presence of oleic acid and oleylamine as ligands as well as 1,2-hexadecandiol as reducing agent. Similarly, 10 nm iron oxide nanoparticles were obtained by using 8 nm magnetic nanoparticles as seeds. Using this seed-mediated growth, bigger magnetic nanoparticles up to 20 nm were synthesized. Jana, et al.,[72] reported synthesis of iron oxide nanoparticles using iron-oleate as precursor. The size of nanoparticles tended to increase with increasing of concentration of ligands in the solution and could be different with various chain lengths of ligands. The shape of magnetic nanoparticles can also be various while concentration of ligands was different. Park, et al.,[57] demonstrated the temperature effects on decomposition of iron-oleate complex. They found one oleate ligand dissociated from precursor at 200-240 °C, and the other two dissociated at 300 °C.

Moreover, the size of the nanoparticles tended to be larger while the solution was heated up to a higher temperature. Also diameter particles would grow from 11 nm to 14 nm while concentration of capping ligands increased from 1.5 mM to 4.5 mM. Hyeon and co-workers[31] synthesized one-nanometer-scale size controlled monodisperse magnetic nanoparticles from 6 nm to 15 nm by combining seeds media procedure and series of amount of iron oleate precursor. Bronstein, et al.,[73] reported that the structure of the iron oleate complex, which was modified by retreatment, could affect the size and shape of the iron oxide nanoparticles.

It has also been observed that nanoparticle size and shape are determined by the reaction time, temperature, and concentration of the reagents. Recently, Kim, et al.[74]

26 obtained monodisperse iron oxide nanocubes with 80 nm sides using Fe(acac)3 as precursor decomposing in benzyl ether with the presence of oleic acid a heating rate up to

20 °C/min. Wang and co-workers[75] obtained monodisperse iron oxide nanorods by decomposing Fe(CO)5 in the ionic liquid using oleic acid and oleylamine as capping ligand.

1.4 Kinetics and thermodynamic process of nanocrystals formation and modeling method

1.4.1 Kinetics and thermodynamics process of nanocrystals formation

For investigating the kinetic mechanisms of particles synthesis, a classical model founded by LaMer and Dinegar[76] to describe the synthesis of monodisperse sulfur hydrosols could potentially be applied to the synthesis iron oxide nanoparticles. In this model, monodisperse crystals could be obtained when (i) Initial concentration of precursor is dilute, and (ii) The nucleation stage ends in a short time. Murray and co- workers[77] induced Ostwald ripening in the LaMer's model and explained the kinetic process for the preparation of monodisperse quantum dots by the thermodecomposition procedure.

27

Figure 1-14 Schematic diagram of nucleation and growth stage in thermodecomposition process based on LaMer model Reprinted with permission from[77] Murray, et al., Synthesis and Characterizetion of Monodisperse Nanocrystals and Close-Packed Nanocrystal Assemblies Annu. Rev. Mater. Sci. 2000, 30, 68. Copyright 2000, Annual Reviews.

Pileni[78] demonstrated a chemical mechanism for the preparation of nanoparticles by a micelle procedure. It showed that the concentration and the type of the surfactant strongly affect the shape and size of nanoparticles. The size of the nanoparticles will increase with increasing of water/surfactant molar ratio. Shevchenko, et al.,[35] reported the role of nucleation rate in size control of CoPt3 nanoparticles. It was illustrated that nanoparticles tend to be smaller with higher nucleation. In preparation

28 of colloidal ternary ZnCdSe nanoparticles, Lee, et al.,[79] demonstrated that particles size tends to be larger at higher temperatures due to high diffusivity. Avrami[80] illustrated a theory in the calculation of the crystallization process of condensed matter which could match the experimental results of the formation of nanoparticles in solution, but lacked a physical meaning. Tsukimura and co-workers[81] developed a new model on crystallization rate of nanoparticles in solution and successfully quantified the relationship between time and crystallization rate in the growth from nanoparticles to crystals. Kwon, et al.,[82] demonstrated the kinetics process of iron oxide nanoparticles synthesized by the “heating up” (dissolve precursor in the solution before heating process) thermodecomposition route is essentially the same as “hot-injection” (inject precursor in the solution during the heating process at the nucleation temperature) process.

Figure 1-15 Schematic diagram of magnetic nanoparticles size evolution process and magnetic moment changes with time. Reprinted with permission from [82]Kwon et al., Kinetics of Monodisperse Iron Oxide Nanocrystal Formation by "Heating-up" Process Journal of the American Chemical Society 2007, 129 (41), 12571-12584Simulation of kinetic process in nanocrystals synthesis. Copyright 2007, American Chemical Society.

29 1.4.2 Simulation of kinetic process in nanocrystals synthesis

In scientific studies and biomedical applications, monodisperse nanoparticles with tunable size and shape are preferred[83]. Synthesis of controllable nanoparticles could address a perspective direction of acquiring homogeneous crystallinity, chemical composition, phase structure and surface activities.

A theoretical simulation algorithm could build a promising model including molecular interactions, particles motion behavior, nucleation and crystal growth factors with restrained boundary conditions to calculate and further predict the nucleation process, the particle growth route, resulting size distribution, shape, and even colloidal stabilities.[84]

In modeling of colloidal transport behavior, Brownian Dynamics, Molecular

Dynamics, and Monte Carlo are three algorithms having advantages by embodying individual particle motion tracks, interactions, collisions, and allocations in the computational hierarchy to precisely reproduce and predict the entire process in micron and even nano scale on space aspect[85-87]. Moreover, Brownian Dynamics, Molecular

Dynamics, and Monte Carlo simulations divide the systematic properties into discrete molecule and/or particles contribution by utilizing fundamental physics and chemical interactions[85-87], which gives researchers a better understanding of the research object in finite time and space dimensions, and therefore reveals the mechanism of the kinetics and thermodynamics in the system. Known the motion behavior of the nuclei embryo in the nucleation stage by applying Brownian Dynamics, Molecular Dynamics, or Monte

30 Carlo algorithm in the simulation, it is possible to calculate and track the size and position of the particles combining collision conditions or other size changes components in.

1.4.2.1 Brownian dynamics

Brownian dynamics is a certain algorithm case of the stochastic dynamic method, which was used to compute a system containing multi particles and some degrees of freedom. In Brownian dynamics, there is a basic assumption that random force contains no correlations in time and space[88-89], and therefore could be simplified from stochastic dynamic. Utilizing this time scale simplification or called separation, Brownian dynamics could be applied in solving the problem, which would be complicated via other methods such as molecular dynamics. In the hierarchy of Brownian dynamics, the motion, which is not the investigation candidate, should occur in markedly rapid time than the rest, and Brownian dynamics offers a shortcut to simulate and solve the problem by considering the rapid process as a random interaction. This simplification will lose some motion details microscopically but preserve the major motion of which is the focus: discrete particles. Van Gunsteren[90] and Ermak[91] built the fundamental theory of

Brownian dynamics by employing a systematic force, Fi(x(t)), into the ordinary Langevin equation[92]:

( ) (1-16)

31 where m is the mass of particle,  is friction coefficient, v is the velocity of particle and R is the random force of the system. On the other hand, 2-D Brownian motion follows a

Gaussian (normal) distribution as[92-93]:

(1-17) √

where y1 and y2 are the space position of the particle at time tn and tn+1, D is ¼ of the random force spectral density and  is the correlation coefficient related to particles space position y1 and y2. The particle motion distance r and velocity v in next time slot in

Brownian dynamics can be derived as:

{ }

(1-18)

[

] (1-19) where O[(Δt)4] is the big O notation which means the function of distance r changes in the rate of (Δt)4

32

∫ { } , (1-20)

(1-21)

(1-22)

Scala[92] developed Brownian dynamics in the simulation of polydisperse hard sphere in the absence of hydrodynamic interactions. Northrup and co-workers[85] reported using Brownian dynamics simulation to represent diffusion controlled bimolecular reactions and further explore the interaction between enzymes and substrates. Turq, et al.,[94] applied Brownian dynamics in the simulation of interacting cations and anions motion in saline to compute mass transfer coefficients.

1.4.2.2 Molecular dynamics

Molecular dynamics is the simulation approach of investigating problems on molecular level by numerically building and solving the Newtonian mechanics of individual molecules of a system. The initiation of the method could be dated back to

Alder and Wainwright in 1959[95] and Rahman in 1964[96]. However, the early studies were based upon the over simplified model, like a very limited number of molecules moving at the same velocity, and not until more recent time, Molecular dynamics was employed in the massive molecule/particles motion with complicated applied force and motion behaviors. The classic Molecular dynamics algorithm was developed by

33 Verlet[97]. Verlet algorithm takes into account of all of the interactions between each of the particles in the target system and expressed by the Newton force equation:

⃗⃗⃗ ∑ (1-23)

where r is the distance between two particles and F is the force between particles which is based on the interactions between the particles in the investigated system. Verlet also illustrated a difference equation including time step h in:

∑ (1-24)

This simple algorithm was further developed into velocity by the Verlet algorithm and has become the current choice of molecular dynamics simulations[98]. Due to its straightforward algorithm, Molecular dynamics has been widely applied in the simulation of the system with decided interactions. There are, however two major restrictions that limit the further development of Molecular dynamics to a universal algorithm. First is all the interactions between the particles have to be correctly accounted. Molecular dynamics is an “exact” algorithm without considering random forces. Therefore, it’s not applicable to the system with rapidly changing random intermolecular interactions. Second, one need to consider a vast number of particles when simulating a macroscopic model to bridge the gap between the molecular and the continuum levels[88]. Starr and co- workers[99] investigated the case of bead-spring polymer melt surrounding a nanoparticle by molecular dynamics modeling and illustrated the polymer-particles

34 interaction can effectively change the glass transition temperature of polymer. Naicker, et al.,[100] employed molecular dynamics simulations in surface energy and crystal structure calculation of three phases (anatase, brookite, and rutile) of titania. In their research, the calculated Ti-O bond length matched well within the reference experimental data. Surface energy difference in three phases with different crystal size agreed with their thermostability.

1.4.2.3 Monte Carlo simulation

Monte Carlo modeling was first introduced by Metropolis, et al.,[101] in 1953.

Nowadays, it has been developed in extensive diverse application due to its capability to simulate the scenario that occurring at rapid time scales by including random process in the algorithm which is similar to the Brownian dynamics[100]. The different part is that the classic Monte Carlo simulation is based on Markovian, a more mathematic process comparing to the Langevin equation. It can be used not only in the simulation of a system in the equilibrium state but also at quasi-states or even “chaos” conditions[100, 102].

Metropolis considered a particle i in the position of m first and then moves to the next position n randomly. Then the algorithm could be written as:

(1-25)

where 0 is a uniform random number between 0 and 1, rmax is largest displacement the particles can move. Then the acceptance or rejection will be determined by the potential

35 energy difference dV from position m to n. If dV<0, the move is accepted. If dV > 0, the

Boltzmann factor exp(-dV/kBT) will be employed as second determination condition.

Similarly, a uniform random number b0 between 0 and 1 is generated. The move is still accepted if b0 < exp(-dV/kBT), otherwise, the move will be rejected. Once the determination process is finish, the system will be taken as a new state whether particle moves or not. By repeating this process and inducing more particles of interest, Monte

Carlo simulation is therefore processed. It is necessary to combine Brownian dynamics and Monte Carlo simulation together when modeling the physical system including both multiple particle interactions with random force induced and case by case determination of particles motion behavior. Guarnieri[103-104] developed Monte Carlo simulation with

Brownian dynamics algorithm (MC-BD) which has been developed and utilized by Sanz and co-workers[102] in simulation of self-diffusion process in colloidal fluid and Hwang, et al.,[105] in determination of ion current in ion channel system. In the MC-BD algorithm, the individual particles position and velocity could be expressed as

( )

(1-26)

{ }( ) {

} (1-27)

36 where

∫ ( ) (1-28)

∫ ( ) (1-29)

37

Figure 1-16 Algorithm diagram for a combined Monte Carlo-Brownian dynamics method. Reprinted with permission from Guarnieri F. Theory and Algorithms for Mixed Monte Carlo-Stochastic Dynamics Simulations J Math Chem 1995, 18 (1), 25-35. Copyright 1995, Springer.

38 1.5 Magnetic hyperthermia application in tumor treatment

1.5.1 Hyperthermia treatment

Hyperthermia treatment is a type of approach to kill cancer cell by exposing body tissue to high temperature (upto 45 °C) with minimal damage to healthy tissues[106], to kill the cancer cells, and to activate the immune system[107]. In order to effectively eradicate the tumor cell, radiation therapy and/or chemotherapy are usually combined with the hyperthermia treatment. Clinical trails have shown that tumor size is significantly reduced when radiation therapy and/or chemotherapy is applied with hyperthermia[106, 108].

The hyperthermia method could be classified as local, regional, and whole-body hyperthermia. Local hyperthermia is applied in small area, usually in tissue scale, such as a tumor. Regional hyperthermia is used in larger area in organ scale. Whole body hyperthermia, just as its name implies, is used to treat metastatic cancer by elevating the body temperature to 42 °C. There are several ways of providing hyperthermia treatment which could as simple as using a hot room, wrapping the patient in hot blankets, infusing a warm fluid, or to utilize instruments such as ultrasound, infrared, microwave, induction heating, and magnetic hyperthermia[106, 109].

In general, healthy tissues are not likely to be damaged in hyperthermia treatment if the temperature has been controlled. Some tissues may be overheated, however, because of the regional inconsistencies in thermal diffusion in the body. People may feel

39 burns, discomfort or pain after treatment, but these side effects are usually temporary.

Infusion and whole body hyperthermia can cause serious adverse effects such as blood clots, bleeding, diarrhea, and nausea[110]. To minimumize the size effects, hyperthermia treatment need to be localized to a small area.

1.5.2 Magnetic hyperthermia mechanism

Magnetic hyperthermia is the promising therapeutic method of cancer treatment comparing to other hyperthermia process as it could localize the heat to the tumor region and therefore keep the heating efficiency without damaging normal cells[111]. Magnetic particles have been proposed in hyperthermia treatment in 1957 by Gilchrist and co- workers[112]. Heat of magnetic nanoparticles is majorly generated from either hysteresis loss as well as Brownian and Néel relaxation[113-114].

A typical way of evaluating magnetic nanoparticle heating efficiency under alternating magnetic field is using the specific absorption rate (SAR)[115] which has the units of W/g, which could be expressed from both field aspect as well as materials aspect:

(1-20)

where A is the area of hysterisis loop in the unit of J/g, f is the frequency of the alternating magnetic field, Cv is the volume heat capacity of the material in field. t is the probing time, T is the probing temperature and m is the mass of magnetic nanoparticle in the investigated cadidate.

40

Figure 1-17 Schematic diagram of possible mechanisms for conversion of magnetic energy into heat Reprinted with the permission of [107]Kumar, et al., Magnetic Nanomaterials for Hyperthermia- Based Therapy and Controlled Drug Delivery Advanced Drug Delivery Reviews 2011, 63 (9), 789-808. Copyright 2011, Elsevier.

The drawback of using SAR as evaluation standard is that SAR is highly field and frequency dependent. To minimize this effect, Kallumadil and co-workers[116] proposed a parameter intrinsic loss power (ILP) which is SAR/(fH2). However, the fundamental assumption of this theory that imaginary magnetic susceptibility is identical in different frequency has been proven has very limited applicable conditions[113-114]. On the other hand, Carrey, et al.,[117]developed the linear response theory (LRT) to depict SAR in field aspect by assuming magnetic materials response linearly to the field under the condition of 0MsVH0

41 magnetization, V the particle volume, H0 the magnetic field amplitude, kB the Boltzmann constant, and T is temperature. Based on the LRT, the SAR could be represent as

(1-21)

where  is the relaxation time, f is the field frequency and  is the density of magnetic materials. LRT can explain some hyperthermia research phenomena and matched well with experimental data in certain cases[118-119]. To enhance the SAR value, particle size and shape as well as colloidal stability play a signification role, which is also the primary object of magnetic nanoparticle synthesis for the application of hyperthermia.

42 CHAPTER 2 INFLUENCE OF LIGAND-PRECURSOR MOLAR

RATIO ON THE SIZE EVOLUTION OF MODIFIABLE IRON

OXIDE NANOPARTICLES

2.1 Introduction

Magnetic nanoparticles have been the subject of intense research over the past decade due to their unique magnetic properties and broad potential applications, including magnetic resonance imaging (MRI) contrast enhancement[120-121], drug delivery[122], magnetic hyperthermia[122], bioseparation[83], and high density data storage[123]. Typically, magnetite (Fe3O4) and/or maghemite (γ-Fe2O3) iron oxide nanoparticles (IONPs) are favored for biomedical applications due to their magnetism combined with their biocompatibility and environmental stability[3]. One of the most popular methods for synthesizing IONPs with narrow polydispersity is the thermodecomposition of an iron-organic precursor in a non-aqueous solution in the presence of capping ligands at high temperature[78, 124-126]. Park, et al.,[63] reported that narrow polydispersity IONPs ranging in size from 5 nm to 22 nm were synthesized using iron-oleate as the precursor and oleic acid as the ligand. Particle size was controlled by altering the solvent and thus the reflux temperature of the reaction. Sun and colleagues[56] demonstrated the formation of IONPs with narrow size distribution via thermodecomposition of iron(III) acetylacetonate (Fe(acac)3) in a solution of oleic acid,

43 oleylamine, and benzyl ether at 297 ºC. Using seeds from this initial reaction this group was able to grow larger particles by the introduction of additional iron precursor7.

Despite these advancements, the role of the ligand-precursor ratio in determining the size of the nanoparticles has been only reported by a few authors.[40, 72, 124, 127-

129] The general trend has been that as the amount of stabilizing ligand increases, the mean particle size also increases. For example, in their synthesis of IONPs prepared by decomposing iron-oleate in a solution of oleic acid and benzyl ether, Jana, et al.,[72] observed an increase in the size of the IONPs with increasing ligand to precursor molar ratio. Similarly, Park, et al.,[124] tuned the IONPs size by increasing the molar ratio of oleic acid to iron oleate (from 2:1 to 4.5:1) in the reaction. Meledandri and co- workers[127] found that the particle size increases at higher ligand/precursor ratios using both oleic acid and oleylamine in the reaction. Yin, et al.,[128] obtained similar results by decomposing iron pentacarbonyl in the presence of oleic acid as the capping ligand.

However, Guardia, et al.,[40] observed the reverse trend when decanoic acid was used as the ligand. They hypothesized this reverse relationship could be attributed to changes in the rate of growth of the particles due the low concentration of the decanoic acid. In addition, Alvarez and colleagues[129] tuned the particles’ shape from spherical to cubic

(while maintaining a similar size) by decreasing the ligand/precursor ratio while keeping the ligand/solvent ratio constant.

Despite the results above, the mechanism and influence of the capping ligand- precursor ratio on the synthesis of iron oxide nanoparticles has not been studied in detail.

44 The studies have focused mainly on the relative chemical ratio without considering the change of the total volume caused by including excess ligand. This change may “dilute” or “concentrate” the precursor molecules in the unit reaction space and further change the diffusion rate in the nucleation stage as well as the final particle size. Thus, the previous studies have not completely investigated the role of ligand on the nucleation growth of nanoparticles.

We have investigated the effect of the ligand-precursor ratio with a primary amine group on the size and size distribution of IONPs synthesized using thermodecomposition.

In order to minimize any concentration changes of the reagents that may affect the diffusion rate of nuclei in the reaction, we set the quantity of the precursor and the volume of solvent as constant for all subsequent reactions. This allows us to measure the effect of the capping ligands to iron precursor molar ratio (from 1:1 to 60:1) on the nucleation and growth of the particles. We used IONPs synthesized via thermodecomposition using Fe(acac)3 as a precursor; benzyl ether as the solvent; and 1- hexylamine, 1-octadecylamine, and oleylamine as the three stabilizing ligands. The main advantage of Fe(acac)3 methods is that NPs with narrow size distribution can be synthesized directly from commercially available high purity Fe(acac)3; no additional preparation of the iron precursor is required[56, 63]. The IONPs resulting from thermodecomposition with Fe(acac)3 have a hydrophobic surface, so they require an additional step to link hydrophilic molecules onto the IONPs to make them stable in water[36]. Aliphatic amines have been shown to be easily displaced by hydrophilic stabilizing molecules like PEG[130], and therefore they are the focus of this study.

45 In order to investigate the effects of the aliphatic amine on the nucleation growth stages of particle synthesis, samples under varying reaction conditions were characterized by transmission electron microscopy, X-ray diffraction, nuclear magnetic resonance, magnetometery, and dynamic light scattering. Through the characterization of the particles, we can attempt to explain the role of the capping ligand in dictating the size of nanoparticles produced via thermodecomposition, as well as possible mechanisms behind the change in produced particle size with increasing capping ligands-precursor ratio.

2.2 Experimental

2.2.1 Materials

The materials used were: iron(III) acetylacetonate (Fe(acac)3) (99%) and 1- octadecylamine (98%) was purchased from Alfa Aesar. Benzyl ether (99%) was purchased from Acros Organics. Hexylamine (99%) was purchased from TCI America.

Oleylamine (70%) was purchased from Aldrich. Ethanol (anhydrous, 95%) was purchased from Fisher Scientific. Hexane (HPLC grade) was purchased from EMD. All chemicals were used directly without further purification.

2.2.2 Nanoparticles synthesis

To investigate particle size changes with different molar amount of all three ligand we modified a procedure previously reported by Kang, et al.,[45] For this study,

Fe(acac)3 was used as the iron precursor, a primary aliphatic amine was employed as the ligand, and benzyl ether was used as the solvent. The only variables of this reaction were

46 the ligand type and its molar ratio to the iron precursor. For all studies, 0.5 mmol of

Fe(acac)3 were dissolved in 10 ml of benzyl ether. 1-hexylamine, 1-octadecylamine, and oleylamine were the three primary aliphatic amine stabilizing ligands used in this study.

Four different ligand-precursor ratio (1:1, 3:1, 10:1, and 60:1) were studied for each of the ligands, yielding a total of twelve reactions. As an example reaction, 0.1766g (0.5 mmol) of Fe(acac)3 were added into a three-neck flask along with 10 ml of benzyl ether.

The corresponding amount of aliphatic amine was then added. The reaction was stirred via a magnet stir bar under a nitrogen blanket. The solution was initially heated to 110 ºC and held for 20 min with a nitrogen purge. It was further heated to 300 ºC at a rate of 3 ºC per min, and kept at the reflux temperature for 1 hour, still under a nitrogen blanket. After cooling to room temperature, the resulting nanoparticles were precipitated by adding 30 ml of ethanol into the black prepared solution, isolated by centrifugation at 10000 RPM for 15 min, and washed using a mixture of ethanol and hexane (volume ratio 1:1) three times. After washing, the product was redispersed in hexane for long term storage.

2.2.3 Characterization

Samples for transmission electron microscopy (TEM) were obtained by dropping a diluted (~0.1mg of FeO per mL) hexane solution onto a carbon coated copper grid.

High-resolution TEM images were acquired at an accelerating voltage of 300 kV on a

Hitachi H-9500, and the mean average size of nanoparticles were acquired from TEM using FovaPro image analysis software. Images with magnification rate between 70000x and 150000x were selected as analysis candidates, in which one nanometer in length

47 contains approximately 10 to 15 pixels across. Approximately 400 nanoparticles were measured for each sample and each reaction conducted. From this dataset, the resulting mean and standard deviation of the size of the nanoparticles was determined to describe the polydispersity of the sample. The validity of the size distribution data were processed by t-test, and results prove that the size distribution of each sample was significantly different to others.

Thermogravimetric analysis (TGA) and differential scanning calorimetry (DSC) was carried out by SDT Q600 from TA Instruments. All samples were heated from room temperature to 290 °C at the ramping rate of 3 °C / min. Hydrodynamic diameter of

IONPs was obtained by a Zetasizer NanoZS dynamic light scattering (DLS) instrument from Malvern Instruments. FTIR was taken by Thermo-Nicolet Magna 550 FTIR instrument. Nuclear magnetic resonance (NMR) spectra were characterized in CDCl3 solution using a JEOL ECX-300 spectrometer. Each sample was run with 32 scans at

20oC. The instrument control and spectra analysis were carried out by Delta NMR

Processing and Control Software. X-ray powder diffraction (XRD) was performed at 40 kV and 40 mA using a Rigaku Ultima IV X-ray diffractometer with CuKα radiation (λ

=0.154056 nm) at a scanning rate of 0.5°/min. Hysteresis loops were measured at 300K from ±2387 kA/m (±30,000 Oe) using a vibrating sample magnetometer (VSM) on the physical property measurement system (PPMS) from Quantum Design, Inc. IONPs samples in hexane were loaded onto Kapton tape, allowed to dry fully, and sealed during measurement under vacuum. Following VSM analysis, the loaded samples were digested via concentrated nitric acid over two days and then the sample was diluted to 2% /wt

48 nitric acid. Inductively coupled plasma atomic emission spectroscopy (ICP-AES) (Perkin

Elmer Optima 3100 RL) was used to determine the amount of iron in the sample. The magnetization data was normalized by mass of iron in the sample. Blank Kapton tape was tested as background and was removed by subtracting from the data prior to plotting

(Figure 2-7).

2.3 Results and Discussion

Twelve different reactions were conducted to test the role of molar ratio and type of aliphatic amine on the formation of iron oxide nanoparticles. Again, the ligands tested were hexylamine, octadecylamine, and oleylamine at four different concentrations (0.5,

1.5, 5, and 30 mmol) each. To rephrase, this is molar ratio of ligand to precursor of 1:1,

3:1, 10:1, and 60:1. The resulting particles were characterized by TEM, XRD, DLS,

NMR, and VSM to gain insight on the effect of capping ligand on the nucleation and growth of the particles.

Changes in the mean size of the synthesized particle were measured with TEM.

For example, increasing the ratio of octadecylamine to precursor, decreased the size of the resulting particles (Figure 2-1). The same trend was observed in hexylamine and oleylamine (Figures 2-2 and 2-3).

49

Figure 2-1 TEM image of IONPs synthesized by decomposing 0.5 mmol Fe(acac)3 with octadecylamine- Fe(acac)3 molar ratio of a) 1:1, b) 3:1, c) 10:1, and d) 60:1 in solution.

Figure 2-2 TEM image of IONPs synthesized by decomposing 0.5 mmol Fe(acac)3 with hexylamine-

Fe(acac)3 molar ratio of a) 1:1, b) 3:1, c) 10:1, and d) 60:1 in solution

50

Figure 2-3 TEM image of IONPs synthesized by decomposing 0.5 mmol Fe(acac)3 with oleylamine-

Fe(acac)3 molar ratio of a) 1:1, b) 3:1, c) 10:1, and d) 60:1 in solution

a b

1 nm

Figure 2-4 HR-TEM image and diffraction patterns of IONPs synthesized by decomposing 0.5 mmol

Fe(acac)3 with oleylamine- Fe(acac)3 molar ratio of 60:1 in solution

The synthesized IONPs are single crystal which is confirmed by the high resolution image of a typical nanoparticle and its diffraction patterns. Image analysis of the samples confirmed that as the molar ratio of amine increases, the resulting particle size decreases (Table 2-1). It should be noted that differences between sizes as a function molar ratio are significantly different as confirmed by a two-tail t-test with a 95%

51 capping ligand type

ligand- precursor hexylamine octadecylamine oleylamine molar ratio

1:1 9 ± 3 11 ± 3 12 ± 3 3:1 8 ± 1 8 ± 1 8 ± 1 10:1 7 ± 1 7 ± 1 8 ± 2 60:1 5 ± 1 5 ± 1 7 ± 1

Table 2-1 Mean diameter and standard deviation (in units of nanometer) of synthesized IONPs as a function of type and ligand-precursor molar ratio.

Figure 2-5 Mean size of IONPs as a function of molar amount of (circle) oleylamine, (square) hexylamine and (triangle) octadecylamine used during the synthesis.

52 confidence interval. This trend is further illustrated in Figure 2-4 where the resulting particle size is shown to be highly molar ratio dependent, but the type of aliphatic amine appears to be less significant. It should be noted that this trend with molar ratio is the opposite of the previously reported trends as discussed above.

In order to verify that the change in nanoparticle size was a result of ligand- precursor ratio and not a result of the iron precursor being diluted by the capping ligand

(by increasing of total volume of the reaction solution), we also prepared a comparative sample. In this reaction, 1.5 mmol of 1-hexylamine, 0.5 mmol of Fe(acac)3, and 13.8 ml of benzyl ether (rather than the 10 ml used previously) were reacted using the same heating method as previously described. This sample was then compared to the sample containing the same amount of ligand as well as the sample with the same total volume(ligand-precursor of 3:1 and 60:1) (Table 2-2).

comparative reagent hexylamine-3:1 hexylamine-60:1 sample

Fe(acac)3 (mmol) 0.5 0.5 0.5 hexylamine (mmol) 1.5 1.5 30 hexylamine (ml) 0.2 0.2 4 benzyl ether (ml) 13.8 10 10 total volume (ml) 14 10.2 14

Table 2-2 Amounts for each component used in preparing the comparative sample (varying solvent volume) and the two reference samples (varying molar amount of ligand).

53

Figure 2-6 TEM image of comparative sample and two reference samples synthesized by decomposing 0.5 mmol Fe(acac)3 with (a)1.5 mmol of hexylamine and 13.8 ml of benzyl ether, (b)1.5 mmol of hexylamine and 10 ml of benzyl ether and (c) 30 mmol of hexylamine and 10 ml of benzyl ether in solution.

Figure 2-7 DLS spectra of comparative sample and two reference samples synthesized by decomposing 0.5 mmol Fe(acac)3 with 1.5 mmol of hexylamine and 13.8 ml of benzyl ether, 1.5 mmol of hexylamine and 10 ml of benzyl ether and 30 mmol of hexylamine and 10 ml of benzyl ether in solution.

54 TEM (Figure 2-5) of this comparative sample had a mean size (8 ± 2) nm, the same as the result obtained by the “3:1” experiment, which was (8 ± 1) nm.

These three samples were also compared using DLS (Figure 2-6 and Table 2-3), which measures the hydrodynamic size. Again the comparative sample most closely matched the “3:1” reaction in size, suggesting that the ligand to precursor molar ratio plays a key role in size determination, while the volume changes’ (i.e. concentration) impact on particle size are minor in comparison.

comparative hexylamine- hexylamine-

study 3:1 60:1 Z-average /nm 38.66 35.33 19.55 polydispersity index 0.168 0.244 0.048 Number diameter / nm 12.45 15.63 6.105 Standard deviation / nm 2.85 4.95 1.79

Table 2-3 DLS peak data of comparative sample and two reference samples synthesized by decomposing 0.5 mmol Fe(acac)3 with (comparative)1.5 mmol of hexylamine and 13.8 ml of benzyl ether, (hexylamine-3:1)1.5 mmol of hexylamine and 10 ml of benzyl ether and (hexylamine-60:1) 30 mmol of hexylamine and 10 ml of benzyl ether in solution.

Using a derivative of the LaMer model[76], we hypothesize that the nanoparticles are formed in four distinct stages (Figure 2-7). The first is thermal decomposition of the chelated iron precursor to produce free elemental components, which are described as

“embryos”. The second is nucleation of the particles, followed by the third stage: growth.

Fourth, the particles reach equilibrium in their final form. This hypothesis assumes that the capping ligand-precursor ratio affects the rate of nucleation, and therefore the final

55 size of the particles. For the experiments described above, it is assumed that aliphatic primary amine works as a mild reducing agent in addition to being a stabilizing ligand.

This may promote the decomposition rate of the iron precursor and, therefore, the nucleation rate.

Figure 2-8 Schematic of the nucleation process at (a) low ligand-precursor ratio and (b) high ligand- precursor ratio

To determine what role the primary amine ligand plays during the synthesis, the thermal decomposition of the iron precursor both in the presence and absence of the octadecylamine was characterized using TGA and NMR. TGA experiments (Figure 2-8) were conducted for pure Fe(acac)3, pure octadecylamine, and a well mixed paste of

Fe(acac)3 and octadecylamine in a 1:1 molar ratio. From the data we can clearly see that the Fe(acac)3 decomposition temperature, indicated by the decrease in weight, has shifted to higher temperatures when mixed with octadecylamine. Fe(acac)3 with octadecylamine

o started decomposing at ~150 C, and the maximum decomposition rate of Fe(acac)3 with octadecylamine (determined by the peak in the derivative data) was reached at 210 oC. In

56 o comparison, the pure Fe(acac)3 started decomposing at ~75 C, and the maximum

o decomposition rate of pure Fe(acac)3 was reached at 160 C. The second decomposition stage of the pure octadecylamine, indicated by the inflection point in the TGA and the second peak in the derivative, was at 240 oC, and demonstrates the evaporation of the ligand. These data indicate that the intermediate product formed by the mixture of

Fe(acac)3 and octadecylamine helps increase the thermal stability of both components.

Furthermore, the higher decomposition temperature of Fe(acac)3 in the presence of octadecylamine could prompt an initial burst of nucleation, improving the creation of nanoparticles with uniform size.

Figure 2-9 a) TGA and b) TGA derivative at the heat ramping rate of 3 oC/min from room o temperature to 290 C and isothermal for 10 min for: (black) the pure Fe(acac)3 with no ligand, (red) pure octadecylamine only, and (blue) 1:1 molar ratio of Fe(acac)3 and octadecylamine.

In order to investigate the thermal stability behavior changes of Fe(acac)3 in different reaction media, TGA and DSC analysis were employed on three amine ligands with 1 to 1 ligand to precursor molar ratio, octadecylamine to Fe(acac)3 molar ratio series from 1:1 to 60:1, and pure compound of Fe(acac)3 and all three ligands. The weight loss of Fe(acac)3 in this analysis process postponed to 150 C could be due to the purge

57 nitrogen flow rate decreased from 30 to 3 ml/min, which may minimize the possible volatilize and shift chemical equilibrium of Fe(acac)3 decomposition to a certain extent.

Figure 2-10 a) TGA and b) TGA derivative for the pure Fe(acac)3, hexylamine, oleylamine and octadecylamine only, and c) TGA and d) TGA derivative for 1:1 molar ratio of Fe(acac)3 and hexylamine, oleylamine and octadecylamine, and e) DSC of all seven compounds above.

58

Figure 2-11 a) TGA, b) TGA derivative and c) DSC with 1:1, 3:1, 10:1 and 60:1 molar ratio of octadecylamine to Fe(acac)3

It could be observed that the weight loss of Fe(acac)3 between 150 C and 250 C is due to the decomposition of Fe(acac)3, not evaporation from TGA and DSC curve. The decomposition peak of Fe(acac)3 turned to be less significant with increasing of ligand/precursor ratio which is because the decrease of the relative weight percentage decrease of Fe(acac)3. The weight loss of octadecylamine at 300 C in the samples with lower ligand/precursor molar ratio (1:1, 3:1) proved the formation of intermediate product between precursor and ligand. The weight loss of octadecylamine at 300 C in the samples with higher ligand/precursor molar ratio (10:1, 60:1) is because the evaporation of excess ligand in the solution. The endotherm peaks at 50 C represented the melting process of octadecylamine.

59 In order to determine the role of the functional group in the reaction, the sample of 1:1 molar ratio of Fe(acac)3 and octadecylamine in benzyl ether was characterized individually (a) and before and after reaction (b). Before the reaction by FT-IR (Figure 2-

9), we can see that the specific peaks of amine group are around 3300 cm-1 and in the fingerprint region (1650-1580 cm-1). But the peaks in the fingerprint region (1650-1580-1) are also overlapped with C=O and C=C peaks of Fe(acac)3. After the reaction, no N-H at

3300 cm-1 could be detected, which could be due to the relative intensity of the specific peaks (benzyl ether peaks are at 3033, 2858, 1549 cm–1)[131]. The loss of the peak during the reaction may be primarily due to the decomposition of the Fe(acac)3.

Therefore, FTIR is not suitable for tracking the presence/absence of the amine group in this system.

Figure 2-12 FTIR spectra of (a) pure octadecylamine and Fe(acac)3 at room temperature, and (b)1:1 molar ratio of Fe(acac)3 and octadecylamine in benzyl ether before reaction and after heated up to 300oC and kept for 1h.

60 In order to verify the role of the primary amine on the ligand in nucleation, NMR spectra were measured at room temperature on the solution containing 0.5 mmol octadecylamine, 10 ml of benzyl ether, and 0.5 mmol of Fe(acac)3 before and after heating (Figure 10). (The solution after heating was measured on the supernatant, to remove artifacts in the NMR spectra originating from the nanoparticles.) The primary amine peak (2.7 ppm) can be observed for the unreacted sample along with the aliphatic chain (1.4, 1.3, and 0.9 ppm). However, this primary amine peak is absent following the reaction, yet the aliphatic chain peaks remain, shifted slightly due (presumably) to binding to the newly formed IONPs. This observation strongly supports our hypothesis that the amine group works as reducing agent and was oxidized during the reaction.

Therefore, the amount of aliphatic primary amine present during the synthesis is what changes the rate of nucleation.

61

Figure 2-13 NMR spectra of the solution containing 0.5 mmol Fe(acac)3 and 0.5 mmol octadecylamine in benzyl ether at room temperature: (a) before heating, (b) after heating

62 Now that the role of the aliphatic primary amine has been established, the synthesis model can be revisited (Figure 2-7). When the iron precursor decomposes at elevated temperatures, the “embryo” iron will combine to form particle nuclei. The rate of this formation will be largely dictated by the reactivity of the precursor as well as the diffusion rate of the “embryo” iron. Since aliphatic primary amine works as reducing agent in this reaction, the presence of the additional ligand in the solution will further promote the nucleation rate. Alternatively, the rate of nucleation is lower in a solution with a low ligand-precursor ratio. When the particles enter the growth stage, the nuclei larger than a critical size, Dc, can continue to grow into particles. Alternatively, nanoparticles below Dc will dissolve into the solution since their high interfacial energy is thermodynamically unfavorable.

Therefore, during the growth stage, a lower nucleation rate (again dictated by ligand-precursor ratio) will allow more time for dissolution of the smaller particles, thereby producing larger particles. Thus, a low primary amine ligand-precursor ratio produces larger particles.

The phenomenon of the decrease in particle size with the increase in ligand- precursor ratio can be described by classic nucleation theory[132]. In the homogeneous nucleation case, the nucleation rate, J, decays exponentially with free energy change, ΔG

(equation 2-1):

, (2-1)

63 where A is a proportionality constant, k is Boltzmann constant, and T is temperature. ΔG is proportional to the reciprocal of ligand concentration (equation 2-2) since the decomposition rate of precursor is directly determined by the degree of supersaturation,

S:

, (2-2)

where v is the molecular volume and σ is the interfacial tension of nuclei in the solution.

Moreover, the volume of nanoparticles, V, should be proportional to J-1, which will determine the total number of particles, N, when the individual nanoparticles are growing uniformly (equation 2-3):

, (2-3)

where r is the radius of the particles formed, assuming spherical growth. In other words, as shown in Figure 2-7, V will decrease exponentially with (ln S)2, where S could be increased through addition of aliphatic primary amine ligand to the solution, thereby prompting the decomposition process. In contrast, if the functional group on the ligand is carboxyl (-COOH) rather than amine (-NH2), no prompting of the decomposition process has been reported. Instead, the nucleation rate will be inhibited because the excess ligand in the solution acts as a diffusion barrier.

Now that we have control over the size, the next step is to verify whether or not the change in nucleation rate affects the crystallinity of the IONPs, since the crystallinity directly affects the magnetic properties[27, 133]. From the X-ray Diffraction (XRD)

64 patterns on the as-synthesized NPs (Figure 2-11), an inverse spinel structure is observed for all the particles. When compared to a bulk iron oxide references (magnetite, PDF

No.:01-088-0315 and maghemite PDF No.:01-089-5892), these particles display a much higher relative intensity ratio between (220) peak and (311) peak than the index reference. This increase in intensity likely indicates that (110) is a preferred plane for growth. Since the broadening of the peak is related to the crystalline (i.e. coherence length of a single crystal) within the IONPs, the full width at half maximum peak height

[134] was measured at the (220), (311), and (511) peaks, respectively (Table 2-4). The ratio of the coherence length (CL) to the diameter determined from TEM can now be used to determine which sample had higher crystallinity, where smaller ratios indicate more polycrystalline behavior. By this measure, samples with a higher ligand-precursor ratio had higher crystallinity.

Figure 2-14 XRD patterns for six different iron oxide nanoparticles samples, as a function of type and molar amount of ligand used during the synthesis. Peak locations correspond with that of a ferrite inverse spinel. Data are offset for clarity.

65 mean diameter ligand- coherence CL/Parti ±Std Dev. ligand type precursor length cle from TEM molar ratio (nm) Diameter (nm) 9 ± 3 hexylamine 1:1 3.3 0.4

5 ± 1 hexylamine 60:1 2.8 0.6 octadecylamin 11 ± 3 1:1 3.2 0.3 e octadecylamin 5 ± 1 60:1 3.1 0.6 e 12 ± 3 oleylamine 1:1 3.3 0.3

7 ± 1 oleylamine 60:1 3.1 0.5

Table 2-4 Coherence length (CL) and ratio of coherence length to particle diameter of the IONPs synthesized with different types and ligand-precursor molar ratio.

Finally, the magnetic properties of the IONPs were characterized as a function of molar ratio. The results of room temperature (300 K) hysteresis loops are shown in

Figure 2-12. As expected, the saturation magnetization, Ms, increases with increasing particle size for the octadecylamine series nanoparticles. In contrast, this trend is not present in the oleylamine series, where the largest and smallest particles have MS within error of each other. This lack of a trend could arise in part due to reduced crystallinity in the oleylamine samples compared to the octadecylamine samples (as determined by the

XRD coherence lengths shown in Table 4). Another possible contributor to MS is the unknown but variable ratio of Fe3O4/γ-Fe2O3 present in each sample. This variation

66 would result from the higher concentrations of ligand increasing the reduction of Fe3+ to

Fe2+, changing the chemical composition without significant change in the crystallographic structure. The change in ligand itself may also affect the magnetic properties through surface chemistry, as can be seen in the positive slope at high magnetic fields. Specifically, the other prominent feature of these loops is the positive slope of the data above 500 kA/m (6,250 Oe). This slope has often been attributed to surface spin canting, perhaps from an amorphous layer on the surface of the nanoparticles[135]. Given that the coherence length determined from XRD is less than

60% of the size of the particle, an amorphous layer is possible. Therefore, synthesizing nanoparticles with both a desired size and optimal magnetic properties will require tuning the reaction kinetics to balance reduction caused by the aliphatic primary amine with growth rate for high crystallinity.

Figure 2-15 Hysteresis loop on the synthesized IONPs measured at 300 K in powder form with Kapton tape background removed and normalized to concentration of iron. (a) IONPs synthesized with octadecylamine, (b) IONPs synthesized with oleylamine.

67 2.4 Conclusion

In summary, we successfully synthesized IONPs with aliphatic primary amines having tunable size and narrow size distribution through a thermodecomposition procedure. The size of the synthesized IONPs decreased with increasing ligand-precursor ratio, in contrast with previous reports. This discrepancy could potentially be due to the presence of the primary aliphatic amine ligand working as a mild reducing agent and promoting the nucleation rate. The higher nucleation rate helped form smaller nanoparticles by consuming more precursor in the solution compared to a sample with a lower nucleation rate. Quantifying the functional relation between reaction conditions and IONP size may provide requisite knowledge for precise control and prediction of the size, crystallinity, and magnetic properties of IONPs. Finally, the synthesis of narrow polydispersity IONPs capped with aliphatic primary amine, which allows for easy surface modification for binding biomolecules, might broaden the applications for these particles in the future.

68 CHAPTER 3 EXPERIMENTAL OBSERVATION AND MONTE

CARLO SIMULATION OF LIGAND MOLECULE LENGTH EFFECT

ON THE IRON OXIDE NANOPARTICLES SIZE DISTRIBUTION

3.1 Introduction

Due to their unique magnetic properties and biocompatibility, iron oxide nanoparticles (IONPs) has been intensively investigated in past decades for their potential applications in the area of therapeutic treatments and imaging[136]. Specifically, IONPs have been promoted in the field of magnetic resonance image (MRI) contrast enhancement agents[122, 137], drug delivery vessels[1], hyperthermia treatment media[138] and bioseparation carriers[111, 139]. Investigation of kinetics mechanism of nanoparticles formation process is critical due to the magnetic properties, such as anisotropy constants, coercivity, and saturation magnetization are significantly affected by the particles size and shape and will further influence particles performance in biomedical area[140-141]. The thermal decomposition of chelated iron precursors at high temperature is a commonly used method to tailor the particle size by simply changing reagent and reaction conditions.[124] One of the most effective approaches to control particles size is changing the ratio of capping ligand/precursor in the reaction which could tune the particles size in 1 nm scale[31, 63]. We have previously demonstrated that IONPs size could be adjusted reversely by increasing ligand/precursor ratio since the amine group on the ligand works as a mild reducing agent and will promote the nucleation rate[142]. While the molar ratio of ligand to precursor has been

69 explored, there has been little discussion on the effect of the molecular weight of the ligand on the synthesis of iron oxide nanoparticles. The goal of this work was to explore the ligand molecule length effect on the size of IONPs and investigate the kinetic mechanisms in the nucleation and grow stages.

In this work we utilized Monte Carlo simulations to predict the nucleation and growth of the iron oxide nanoparticles with different molecular weight ligands. Monte

Carlo (MC) simulation is an effective method of calculating and tracking molecules and particles motion with applied field and random force in the system.[86] This simulation algorithm can account for discrete multi-body contributions with the massive calculation of all contributing particle interactions. It is a stochastic approach, which could account for probing timescale separation with tracked motions occurring at rapid time scales[143]. Monte Carlo modeling has been applied to simulate systems of high particle volume fraction close to the phase transition conditions under the influence of particle interactions precisely[143]. This technique can sketch the system containing particles and/or molecules accurately in general circumstance as well as in much more complicated even extreme conditions. Incorporation of Monte Carlo modeling method to current nanoparticles nucleation hierarchy endeavors holds promising potential for improving the accuracy of numerical simulation and broadening the range of applicability for investigating the mechanism of nuclei diffusion. A classical method to performing

MC begins by casting the process as being Markovian[100]. The system progresses from one state to the next by aid of its transition probability matrix. Metropolis et. al.[101] present the now standard Metropolis method for conducting MC simulations using the

70 energy of the system as the criterion to evaluate the acceptance or rejection of a MC step.

Additional details about Monte Carlo modeling are given by Allen[89] and Frankel et. al.[87, 144] While there have been some investigations into the ligand/precursor ratio as well as molecule length effect on the particles size. However, the investigation of nucleation process by modeling the nuclei movements using Monte Carlo simulation has been rarely reported[86].

Herein, we report how the size distribution of iron oxide nanoparticles is affected by the molecule chain length of aliphatic primary amine ligand. Both the experimental methods as well as Monte Carlo simulation investigated the kinetic mechanism of these changes at the nucleation stage. This was accomplished by utilizing three different length ligands (1-hexylamine, 1-dodecylamine and 1-octadecylamine) in a series of ligand/precursor ratio by thermodecomposing Fe(acac)3 at 300 C. Particle size were tracked by analyzing TEM images of samples and simulations were conducted to reproduce experimental procedure and provide insight into crystal growth route of this reaction.

3.2 Experimental

3.2.1 Materials

The materials used were: Iron(III) acetylacetonate (Fe(acac)3) (99%) was purchased from Alfa Aesar, benzyl ether (99%) was purchased from Acros Organics, hexylamine (99%) was purchased from TCI America, 1-dodecylamine (98%+) and 1-

71 octadecylamine (98%) were purchased from Alfa Aesar, ethanol (anhydrous, 95%) was purchased from Fisher Scientific and hexanes (HPLC grade) was purchased from EMD.

All chemicals were used directly without further purification.

3.2.2 Synthesis of Iron Oxide Nanoparticles

In order to investigate the role of ligand length on particle formation, 0.5 mmol constant Fe(acac)3 was used as the iron precursor, and 1-hexylamine, 1-dodecylamine and 1-octadecylamine with five different molar ratio of the ligand/precursor (1:2, 3:2,

2:1, 5:1, 30:1) were studied for each of the ligands, yielding a total of twelve reactions.

10 ml of benzyl ether was used as solvent to keep the consistency of the reaction. In a typical reaction, 0.5 mmol Fe(acac)3 were added into a three-neck flask with 0.5 mmol of hexylamine and 10 ml benzyl ether and stirred vigorously via a magnet stir bar. The solution was initially heated to 110 ºC and held for 20 min with a nitrogen purge. It was further heated at 3 ºC / min to 300 ºC, and kept at reflux for an hour under a nitrogen blanket. After cooling to room temperature, the resulting nanoparticles were precipitated by adding 50 ml ethanol into the black prepared solution, isolated by centrifugation at

10000 RPM for 15 min and washed by the mixture of ethanol and hexane (volume ratio

2:1) three times. After washing, the product was redispersed in hexane for long term storage.

72 3.2.3 Characterization

Samples for transmission electron microscopy (TEM) were obtained by dropping diluted solution onto a copper grid with carbon film. High-resolution TEM images were acquired at an accelerating voltage of 300 kV on Hitachi H-9500, and the mean average size of nanoparticles were acquired from TEM using FovaPro image analysis software.

Approximately 400 nanoparticles for each sample were measured for each size measurement. Hydrodynamic diameter of IONPs was obtained by the Zetasizer NanoZS dynamic light scattering (DLS) instrument from Malvern Instruments. X-ray powder diffraction (XRD) was performed at 40 kV and 40 mA using a Rigaku Ultima IV X-ray diffractometer with CuKα radiation (λ =0.154056 nm) at a scanning rate of 0.5°/min. A/C susceptibility was characterized by IMEGO DynoMag AC susceptometor.

3.3 Modeling

A simulation model was built using Mathworks Matlab 2012b version. This modeling study simulated the motion in a 2-D area of iron nuclei where these nuclei are the localized particles formed by precursor immediately following decomposition. 2-D coordinate simulation is presented since the results include nuclei position and size can be visualized facilely and distinctly without overlapping. Spherical iron oxide nanoparticles were chosen as computational candidates. To test the size dependency to the ligand length, 400 iron oxide nuclei and corresponding number (200 and 12000) of ligand molecules were placed in a 1400×1400 (Angstrom2) area. The nucleation process start at 127 ºC (400 K), and all precursor molecules were assumed decomposed at the

73 same time. The decomposition temperature of Fe(acac)3 along with primary amine was verified in the previous study[142]. In this system, it was assumed there would be no magnetic interactions as it is above the Curie temperature of the iron oxide nanoparticles[145-146], and the driving forces controlling the diffusion of the nuclei were thermal fluctuation, van der Waals interaction, and electrostatic interaction.

To give a better understanding of our modeling, herein we demonstrate the algorithm of the computation in detail. To probe how the particle size would be affected by the ligand with different molecule length, particles distribution data were measured at

3 min, 10 min and 30 min after the beginning of the decomposition. The particles motion in the heat bath follow the Langevin Equation[147]:

(3-1)

where m is the particle mass, v is the velocity of the particle, f is the systematic force, x is the particles’ moving distance, t is the time R is the random force caused by the thermal fluctuation and  is the friction coefficient.

Electrostatic interaction consists ion-ion(Eii), ion-dipole(Eid) and dipole- dipole(Edd) interaction three components:

(3-2)

(3-3)

(3-4)

74 where z is the charge of the candidate (ligand molecules and nuclei), 0 is the permittivity of the vacuum,  is the relative permittivity of the media (2.4398 for benzyl ether)[148], u is the dipole moment of candidate and R is the distance between two candidates in 2-D space,  is the angle between dipole length and ion-dipole center distance and c is the constant which relate to the relative angle between two dipoles. Here we set c as 2 to depict two collinear dipole vectors are the same spatial orientation. Van der Waals interactions include Keesom force (dipole-dipole), Debye force (dipole-induced dipole), and London dispersion force (induced dipole-induced dipole) and repulsion force.

(3-5)

(3-6)

(3-7)

(3-8) where  is the electronic polarizability of the candidate, v is the vibrational frequency of electrons,  is a constant which was candidate size related, n is an adjustable constant and was set at 1×10-23, h is the Planck constant, k is the Boltzmann constant, T is the Kelvin temperature. The electrostatic and van der Waals interactions were calculated in particle- particle, particle-ligand and ligand-ligand cases. Both of them would be accounted for in the systematic force f. Random force R is generated by the thermal fluctuation. The interactions between solvent molecule (benzyl ether) and candidate (ligand or particles)

75 were simplified as the part of the friction coefficient . All the force along the x and y directions are calculated separately.

To quantitatively simulate the particles growth process, each step was match with

100 ms in the actual reaction and there was up to 9×103 steps been computed in this

Monte Carlo simulation.

3.4 Results and discussion

Iron oxide nanoparticles with a matrix of reaction conditions of varying molecular length and ligand/precursor molar ratio were synthesized. TEM images (Figure.3-1) demonstrated that iron oxide nanoparticle size decreased with decreasing ligand molecule size when ligand/precursor ratio was 1:2. This trend (Figure.3-2) turned out to be less conspicuous when the ligand/precursor ratio increased, and the size of the nanoparticles obtained from 3 different ligands were nearly identical when ligand-precursor molar ratio reach 30:1. From the plot (Figure.3-2) we could see that nanoparticles synthesized in short ligand molecules are smaller than that of the longer ones when the ligand/precursor ratio is below 2:1. When ligand/precursor ratio was at a high level, say equal to or above

2:1, the size of nanoparticles synthesized by different ligands is similar and only sensitive to the concentration changes. DLS result in figure 3 matched results of TEM in figure 1.

76

Figure 3-1 TEM images of IONPs obtained with ligand-precursor molar ratio of 1:2 (a) hexylamine (b) dodecylamine (c) octadecylamine, and 30:1 of (d) hexylamine (e) dodecylamine (f) octadecylamine.

Figure 3-2 Mean size of IONPs changes with ligand molecule size in different ligand-precursor ratio.

77

Figure 3-3 DLS data of IONPs synthesized by hexylamine and octadecylamine with ligand-precursor ratio of 1:2 and 30:1

A time-size dependency study was employed to track the nuclei growth process with changing of time and temperature. Aliquot samples were picked every 25 ºC from

125ºC to 300 ºC and 7.5 min and 15 min after temperature reached 300 ºC. A/C susceptibility investigation of these samples in Figure 3-4 demonstrated there was no observable susceptibility increase, which is size dependent, till the temperature is above

200 ºC. Then the size of the nuclei grows significantly when ligand/precursor ratio is 1:2; and relatively slower when ligand concentration is high. This phenomenon could be due to the diffusion rate difference in two reaction solutions with different ligand concentration. Susceptibility of 1:2 hexylamine sample reached equilibrium before 300

ºC and had smaller value comparing to 1:2 octadecylamine sample. This is the evidence that the ligand molecule length can affect the initial nuclei numbers and further tune the particle size in grow step.

78

Figure 3-4 A/C susceptibility plot of IONPs at different reaction time with ligand-precuorsor in 1:2 and 30:1 molar ratio of hexylamine and octadecylamine.

a b

c

Figure 3-5 DLS curves of reagent solution containing 1:2 ligand-precursor ratio of a) hexylamine, b) dodecylamine and c) octadecylamine sampling at 125 °C, 175 °C, 200 °C, 250 °C, 300 °C, and 7.5 min, 15 min at 300 °C.

79

Figure 3-6 TEM images of IONPs synthesized by hexylamine and octadecylamine with 1:2 and 30:1 ligand-precursor molar ratio sampling at different temperature, scale bars represent 50 nm.

DLS data in Figure 3-5 successfully recorded the size evolution of the sample with 1:2 ligand/precursor ratio. TEM images (Figure 3-6) agreed well with magnetic susceptibility results. Large nanoparticle arisen earlier in lower ligand-precursor ratio sample at around 250 ºC comparing to higher ligand-precursor sample, and 1:2 octadecylamine-Fe(acac)3 molar ratio sample showed largest particle size in all four series. No image analysis was processed on these TEM result due to the lack of image contrast or insufficient stability of the nanoparticles sampled during the reaction.

80 In order to explain this relationship between resulting particles size and ligand molecule size, we hypothesis that ligands having longer chain lengths would decrease diffusion rate in the solution, as shown in Figure.3-7a, when ligand/precursor molar ratio was low. Therefore, it would take longer time for embryos which were the micro crystals decomposed from Fe(acac)3 forming nuclei via thermo fluctuation. As a result, ligands having longer chain lengths would restrain the nucleation rate in the particles synthesis process, which could lead to the formation of larger particles. Correspondingly, particles prepared with the shorter chain length molecules would result in small nanoparticles

(Figure. 3-7b). On the contrary, the ligand molecule length effect will be less significant to the nuclei formation when the ligand/precursor reach to a critical point (2:1), because the individual molecule length’s contribution to the diffusion rate was “diluted” by the number of ligand molecules in this case. Therefore, the nanoparticles size tends to be identical at the same ligand/precursor ratio (Figure 3-7c and 3-7d).

Figure 3-7 Schematic of effect of ligand length with different concentration on nanoparticle size. (a) long chain, low concentration, (b) short chain, low concentration, (c) long chain, high concentration, (d) short chain, high concentration.

81 We implemented Monte Carlo method to simulate the motion of nuclei of particle during the nucleation and growth stages in the reaction for building a mathematical mechanism interpretation of experimental results as well as theoretical hypothesis. The movements of 400 nuclei in 1400×1400 (Angstrom2) area were tracked inducing 200 and

12000 of octadecylamine and hexylamine two type of ligand molecules, which match both ligand/precursor ratio as well as absolute volume concentration to ensure the simulation data comparable to the experimental results. Four simulated reaction time slots were picked as investigation candidates (1.5 min, 4.5 min, 9 min and 15 min) to probe the evolution of the nuclei. Nuclei distribution in 2-D area in Figure 3-8 illustrates that there were fewer nucleuses left in the solution after 15 minutes growth in octadecylamine than in hexylamine at lower concentration, and this difference was less significant at higher concentration. The plot of the mean average size from the simulation data (Figure 3-9) and size distribution histogram (Figure 3-10) quantified the trends of the nuclei size evolution, which coincided our hypothesis as well as the experimental results. On the other hand, the mean average size of nuclei decrease with increasing ligand/precursor was explained in details by our previous study. The velocity distribution of nuclei in

Figure 3-11 further confirmed our hypothesis that ligand with different molecule length can tune the diffusion rate of nuclei and therefore change the velocity of nuclei in the solution. The ligand molecule length is the dominant factor that restrain the diffusion rate of nuclei at lower ligand-precursor ratio, and the diffusion rate of nuclei tend to be identical when ligand-precursor ratio is high.

82

Figure 3-8 Images represent simulation results of nanocrystal size and number after 1.5 min, 4.5 min, 9 min and 15 min growth since the beginning of nucleation of solutions with ligand-precursor 1:2 and 30:1 of hexylamine and octadecylamine by Monte Carlo modeling

83

Figure 3-9 Mean average diameter of nanoparticles acquired from simulation results.

Figure 3-10 Histogram of particles diameter acquired from simulation results

84

Figure 3-11 Histogram of particles velocity acquired from simulation results

85

Figure 3-12 Images represent simulation results of nanocrystal size and number after 1.5 min, 4.5 min, and 9 min growth since the beginning of nucleation of solutions with ligand-precursor 1:2 and 30:1 of oleic acid by Monte Carlo modeling

In order to test the universality of the simulation model in different reaction hierarchies, oleic acid was employed as an alternative ligand by tuning the electrostatic and van der Waals interactions between nuclei and ligand molecule. The simulation results in Figure 3-12, 3-13 and 3-14 did not match well with the Park’s[63] experimental data. This could be due to the intrinsic difference between two reaction hierarchies: precursor activation energy (Fe(acac)3 vs. iron oleate), nucleation temperature, and interactions of the solvent molecule, which could not be fully represented by replacing the ligand functional group alone. The may cause the activation energy changes, nucleation temperature shifting and crystal growth rate difference, which could significantly affect particles size in the reaction.

86 The character of the modeling study is to verify the results in Fe(acac)3-amine system mathematically, therefore this model skipped to discuss the behavior variety of precursor in the reaction by far. Iron oleate due to its own chemical property can produce much larger particles than Fe(acac)3 (250nm vs. up to 15 nm) and unique morphology

(spherical, cubic, triangular and rodlike) comparing to simple spherical particles, which is synthesized by Fe(acac)3. It is probable that iron oleate has a dominant effect on nucleation process comparing to the ligand, oleic acid, in the solution. Therefore it may not be suitable to apply this model mechanically to match the published results of iron oleate-oleic acid system. Moreover, sole use of oleic acid with Fe(acac)3 may lead to significant high polydisperse of the produced nanoparticles and even incomplete of the reaction[56].

In order to build a universal algorithm across the different reactant, the activation energy should be included in the calculation and a whole comprehensive model need to be build in future work. (please see future work chapter for more details)

87

Figure 3-13 Histogram of particles diameter synthesized by oleic acid acquired from simulation results

Figure 3-14 Histogram of particles velocity synthesized by oleic acid acquired from simulation results

88 X-ray diffraction patterns in Figure 3-15 indicated that nanoparticles in this procedure were Fe3O4/γ-Fe2O3 phase. The crystallinity was not distinctive between low and high concentration ligand synthesized nanoparticles since the larger nanoparticles contains more amorphous phase. Particles size calculated by Scherrer’s equation[149-

150]

(3-10)

from XRD patterns(Table 3-1) illustrated the similar trends of size evolution with ligand molecule size. The difference between the results of XRD and TEM while ligand- precursor ratio equal 1:2 may be due to the morphology of nanoparticles was no longer close to spherical which eventually effect the shape factor in Scherrer’s equation. From

Figure.3-6 we could found that nuclei formed at around 150 ºC and DLS data indicated the size evolution during the whole crystal growth history.

Figure 3-15 XRD patterns for IONPs samples as a function of molecule size and molar ratio of ligand used during the synthesis. Peak locations correspond with that of a ferrite inverse spinel. Data are offset for clarity.

89 ligand:precursor=1:2 ligand:precursor=30:1

hexylamine 7.5±1.2 nm 6.5±0.1 nm

dodecylamine 7.9±0.4 nm 6.6±0.9 nm

octadecylamine 8.4±0.5 nm 6.4±1.1 nm

Table 3-1 IONPs particles size acquired from XRD patterns and calculated by Scherrer’s equation.

The magnetization of nanoparticles (Figure 3-16) changing with ligand molecule length at 1:2 and 30:1 ligand/precursor ratios presented no remarkable hysteresis loop, which is probably due to the relative small size of IONPs. The saturation magnetization difference among the samples obtained from three ligand with 1:2 ligand/precursor ratios is contributed by the particles size difference of the different samples.

Figure 3-16 Hysteresis loop on the synthesized IONPs measured at 300 K in powder form with Kapton tape background removed and normalized to concentration of iron. IONPs synthesized with (a) ligand-precursor 1:2 molar ratio, (b) ligand-precursor 10:1 molar ratio.

90 3.5 Conclusions

In conclusion, the size of nanoparticles synthesized with the shorter chain length ligand is smaller than that produced with the longer ligands. The length of the ligand is related to the diffusion rate in nucleation stage to further affects the nucleation rate. As a result, mean average size of nanoparticles can be tuned when concentration of ligand is low. On the other hand, particles size control is less when the concentration of ligand is high, because the relative ligand concentration is saturated and the diffusion rates are nearly identical. Therefore, nucleation process is dominated by the amount of ligand molecules. The role of ligand length in the nucleation step were simulated and confirmed by Monte Carlo modeling by the computed size and distribution of nanoparticles within the field of thermal fluctuation, electrostatic interaction as well as van der Waal interaction. The results support our hypothesis of the role of ligand length on the formation mechanism of these particles, suggesting that additional size control is possible by changing the composition of ligands.

91 CHAPTER 4 COLLOIDAL ENVIRONMENT DEPENDENCY OF

SPECIFIC ABSORPTION RATE ON IRON OXIDE

NANOPARTICLES

4.1 Introduction

Iron oxide nanoparticles (IONPs) provide a favorable potential in the exploitation of biomedical area. Two dominate factors would change the theranostics performance are the magnetic core (including particles size, shape and crystallinity), and their colloidal environmental (such as type of dispersant, pH, concentration) The contribution of particles size and shape has been intensively studied because it is directly related to the magnetic properties of the materials and further affects the performance in biomedical applications. The colloidal environmental conditions effect on the properties of the

IONPs, likely due to the wide variability in possible media, has been rarely reported.

MRI performance for both iron[30] and gadolinium[151-152] based nanoparticles changing with pH has been reported. In addition, pH difference have been utilized as the driving force for releasing drugs from magnetic drug loading vessels for cancer treatment

[153-154]. Jayasundar and co-workers[155] investigated the magnetic hyperthermia would lower the local pH of the in and out of tumor cells when temperature increased.

However there is lacking of systematic study of how active adjust pH would affect magnetic hyperthermia.

Hyperthermia is a promising approach of therapy for tumor treatment, since it can

92 neutralize tumor cell as well as stimulate the immune system[107, 156]. This therapeutic method is based on the fact that magnetic nanoparticle generate heat in an alternating magnetic field, therefore the tumor cell will be killed once its temperature raise above 41

°C, due to lacking of temperature tolerance[155, 157-158]. The heat generate from

Brown relaxation, N el relaxation, and hysteresis losses[113, 159]. and the heating efficiency of magnetic nanoparticles under AC field is expressed by the specific absorption rate (SAR). SAR value could be influenced by (i) measurement instrumental setting including frequency and field amplitude, (ii) chemical composition, phase and morphology of the magnetic nanoparticle core, and (iii) colloidal environment such as solution pH, polarity, particle surface chemistry, colloidal dispersity[6, 122, 157, 160-

162]. Presa and coworkers[119] studied the dependency of SAR of IONPs synthesized by coprecipitation procedure to the concentration of IONPs and applied field intensity and frequency. They found the SAR value kept almost identical with increasing of IONPs concentration. This may be due to the relative wide distribution sizes and the larger aggregates that keep interparticle interaction nearly constant. Lima, et al.,[118] investigated the size SAR dependency by utilizing Fe(acac)3 as precursor synthesized

IONPs with the size from 3 nm to 26 nm in diameter via seed growth based method. In this work, maximum SAR was obtained by 12 nm IONPs in toluene. However, the high polydisperse distribution of IONPs, potential magnetic dead layer caused by seed growth method and using water heat capacity without normalizing by toluene in SAR calculation indicate further research is needed. Guardia, et al.,[138] reported the highest SAR over

2400 W/g among the published results basing on their own characterization instrument.

93 The intrinsic weakness of for hyperthermia treatment is to localize the heat to the malignant cells without destroying the healthy one[106]. In order to solve this problem it is important to investigate the interactions between the SAR and the colloidal environment including pH, concentration of IONPs, and type of solution. The pH value of the fluid in human body is not an individual number, which could vary from 1.5 to

8[33, 155] depending on the region.

Herein, we reported SAR value change with pH value of colloidal as well as concentration in both toluene and aqueous solution. SAR values were calculated by the first 30 seconds of the heating process to refrain from thermal dissipation involved.

IONPs with series sizes were synthesized by one-pot method, which was modified of

Park’s procedure. No seed growth step was induced due to the potential frustration layer may produce and reduce the heating efficiency[163].

4.2 Experimental

4.2.1 Materials

The materials used were: sodium oleate (97%) was purchased from TCI (Tokyo,

Japan), iron(III) chloride hexahydrate and divinylbenzene (80%) was purchased from

Alfa Aesar (Ward Hill, MA), oleic acid (95%) were purchased from Spectrum Chemicals

(Gardena, CA). Ethyl acetate (99.5%) was purchased from Macron Chemicals (St. Louis,

MO), sodium periodate (99%) and styrene (99%) was purchased from Acros Organics

(Summerville, NJ). Potassium permanganate (99%), tert-butanol (99%), acetonitrile

(99.8%) 1-octadecene (90%) and 2,2’-azobisisobutyronitrile (98%) was purchased from

94 Sigma-Aldrich (St. Louis, MO). Potassium carbonate anhydrous(99%) and hexanes

(HPLC grade) was purchased from EMD Chemicals Inc. (Gibbstown, NJ), acetone

(99.9%) was purchased from VWR (Westchester, PA), and ethanol (anhydrous, 95%) was purchased from Fisher Scientific (Springfield, NJ). All chemicals were used as received without further purification.

4.2.2 Nanoparticles synthesis

Oleic acid capped IONPs were synthesized according to the modified procedure reported by Park and coworkers. Iron-oleate complex was first synthesized as precursor:

10 mmol of iron (III) chloride and 30 mmol of sodium oleate were dissolved in a mixture solvent composed of 40 ml ethanol, 30 ml DI water and 70 ml hexane. The resulting solution was heated to 60 °C and kept refluxing at that temperature for 4 hours. Then the upper organic layer containing the iron oleate complex was washed with 30 ml DI water for three times. The aqueous phase at bottom was discarded and the upper organic layer was collected for further use. In a typical IONPs synthesis step, 5 mmol of the iron-oleate complex and 15 mmol of oleic acid were dissolved in 50 ml of 1-octadecene at room temperature. The reaction mixture was heated to 110 °C with a constant heating rate of 5

°C min-1 and kept at that temperature for 30 min with nitrogen bubbling. Then the solution was further heated to 320 °C with the same heating rate, and kept refluxing at that temperature for 1 hour still under a nitrogen blanket. The resulting solution containing the nanoparticles was cooled to room temperature afterwards, and was precipitated by adding 500 ml mixture of acetone and ethanol with 1:1 volume ratio and

95 centrifuged at 10000 rpm for 10 min. Then the IONPs particles were redispersed in hexane. The whole purification process was taken three times and the IONPs were stored in hexane for further use.

4.2.3 Encapsulation of IONPs into polymer matrix

In order to study the IONPs interaction changes with and without applying alternating magnetic field in low and high concentration of particles, 5 mg/ml and 1 mg/ml of 22 nm IONPs were suspended in 1 ml monomer solution of styrene and divinylbenzene mixture in 6.67:1 volume ratio by evaporating hexane off. Then 3.3 mg azobisisobutyronitrile (AIBN) was dissolved in each sample and the mixture were heated at 70 °C in the alternating magnetic field of 41 kA/m and 147 kHz for 4 hrs to complete the polymerization. Two reference samples were encapsulated with no field being applied as comparison.

4.2.4 Hydrophilic conversion of IONPs

In order to obtain hydrophoilic IONPs, a modified oxidation procedure[164-165] was employed to tailor the oleic acid molecules bonding on the surface of IONPs. 100 ml hexane dispersed IONPs (containing 200 mg IONPs) was mixed with 70 ml tert-butanol and 30 ml of 2 wt % K2CO3 aqueous solution and stirred for 10 min at room temperature.

Then 25 ml of Lemieux-von Rudloff reagent (5.7 mM KMnO4 and 0.105 M NaIO4 aqueous solution) was added drop wise into the IONPs solution. The solution was mechanical stirred in ice water with sonication applied for two hours. The bottom aqueous layer was collected and the IOPNs were magnetically separated and washed with

96 ethanol and DI water for three times. The hydrophilic IONPs were stabilized by electrostatic repulsion force and redispersed in DI water prior to use.

4.2.5 Characterization

Samples for transmission electron microscopy (TEM) were obtained by dropping a diluted hexane solution onto a carbon coated copper grid. High-resolution TEM images were acquired at an accelerating voltage of 300 kV on a Hitachi H-9500, and the mean average size of nanoparticles were acquired from TEM using FovaPro image analysis software. Images with magnification rate between 70000x and 150000x were selected as analysis candidates, in which one nanometer in length contains approximately 10 to 15 pixels across. Approximately 380 nanoparticles were measured for each sample and each reaction conducted. From this dataset, the resulting mean and standard deviation of the size of the nanoparticles was determined to describe the polydispersity of the sample.

Microtome process was employed by Ultracut E ultramicrotome from Reichert-Jung in

80 nm/step.

Hydrodynamic diameter of IONPs was acquired by a Zetasizer NanoZS dynamic light scattering (DLS) instrument from Malvern Instruments. X-ray powder diffraction

(XRD) was performed at 40 kV and 40 mA using a Rigaku Ultima IV X-ray diffractometer with CuKα radiation (λ =0.154056 nm) at a scanning rate of 0.5°/min. To track the concentration the loaded samples were digested via concentrated nitric acid over two days and then the sample was diluted to 2% /wt nitric acid. Inductively coupled plasma atomic emission spectroscopy (ICP-AES) (Perkin Elmer Optima 3100 RL) was

97 used to determine the amount of iron in the sample. The magnetization data was normalized by mass of iron in the sample. AC calorimetry measurement was performed by AMBRELL Easyheat induction heating systems with lab build five turn copper solenoid. Alternating magnetic field amplitude was measured using an AMF Life

Systems magnetic field probe by attached to a BK PRECISION digital storage oscilloscope 2532B.

4.3 Results and discussion

Iron oxide nanoparticles with mean average size ranges from 9 nm to 26 nm in diameter were synthesis by varying the ligand and precursor concentration, refluxing temperature in the reaction. From TEM images and their size distribution histograms of five samples (Figure 4-1 and 4-2), combining with reaction conditions in Table 4-1, there is signification dependency between particles size and refluxing temperature as well as the concentration of both ligand and precursor.

Figure 4-1 TEM images of IONPs of a) 9 nm, b) 14 nm, c) 17 nm, d) 22 nm and e) 26nm in diameter

98

Figure 4-2 Histograms of IONPs diameter in TEM images a) 9 nm, b) 14 nm, c) 17 nm, d) 22 nm and e) 26nm

In order to perform the hyperthermia study, 0.5 ml of sample colloid in 1 ml cylindered cuvette was placed in AC calorimeter system (Figure 4-3) with a water jacket with has been stabilized at 37 °C to simulate human body temperature. The applied alternating field with fixed amplitude of 41 kA/m and 147 kHz frequency was employed during the investigation.

Figure 4-3. Schematic of AC calorimeter apparatus

99 mean ligand precursor ligand refluxing

average concentration concentration precursor ratio temperature/°C

diameter /nm mol/L mol/L

9 0.32 0.16 2:1 320

14 0.6 0.2 3:1 365

17 0.1 0.1 1:1 320

22 0.48 0.16 3:1 320

26 0.9 0.2 4.5:1 365

Table 4-1 Reaction and purification conditions of obtained nanoparticles

SAR was calculated by the equation of:

(4-1)

where Cv is the volume heat capacity of the sample which was calibrated by assuming sample heat capacity change linearly with its component fraction. T is the temperature change in the t time slot. In this research, temperature changes in first 30 seconds was probed as valid data to minimize the heat dissipation effect on the measurement results. mFe is the mass of the nanoparticle in testing sample which has been normalized by iron.

Each sample was processed triplicate measurements and took the mean average number as investigation result. The SAR changes with size at different concentration were shown in Figure 4-4. IONPs with size of 22 nm reach their maximum SAR, below and above which the SAR will decrease.

100

Figure 4-4 Specific absorption rate of IONPs in toluene with different size and concentration in the applied field of 41 kA/m and 147 kHz.

This phenomena illustrated the existence of an optimized size for SAR measurement which has a similar trend in Rosensweig theory[115] due to the Brownian and Néel relaxation, which are the major energy source of SAR, occur at the same time , and the effective relaxation time is given by

(4-2)

where B and N are the Brownian and Néel relaxation time. The imaginary component of susceptibility in alternating field is therefore calculated by

(4-3)

where 0 is the susceptibility in DC field, f is frequency. And based on the linear responsive theory [117], SAR could be calculated from field by

101

(4-4) where 0 is the vacuum permeability, H0 is the field amplitude.

According to the relation above, there is a maximum value for img when

-1 eff =2f, and as a result, correspondingly SAR reached its maximum value. However, the optimum size in the experiment is different from the calculated data of Rosensweig

(22nm vs 14nm). This difference could be due to the diversity between the experimental alternating field(41kA, 149kHz) and theoretical conditions (72kA/m, 300kHz). The existence of optimized size vs. SAR is consequently due to both Brownian and Néel relaxation depends upon the size of particles[115].

The other phenomena we found in Figure 4-4 are that nanoparticles with lower concentration performed better heating efficiency comparing to the high concentration sample. It is opposite to the published results of SAR is independent with particle concentration, because the heating efficiency has been normalized by mass of particles in the sample in SAR calculation. However, that’s based the assumption that heat generated from individual nanoparticle at low concentration environment is identical to the high concentration one which is debatable. Saville and co-workers[166] recently reported the discrete nanoparticles in a applied alternating field can form chain structures which may decrease the MRI contrast agent performance due to the relaxation has been restrained.

Our hypothesis is the in absence of applied field, nanoparticles in both high and low concentration environment can move freely in the way of Brownian motion. When a

102 magnetic field has been applied, the relaxation of particles in low concentration environment will less likely affected by the magnetic moment of other particles due to the space barrier. However, particles in high concentration environment are intended to lock with the particles nearby and aligned with applied field physically. This short-range order structure (Figure 4-5) can significantly localize the magnetic nanopartices in the colloid and weaken the both Brownian and Néel relaxation and therefore lower the SAR value of sample with high concentration of iron.

Figure 4-5 Schematic diagrams of nanoparticles arrangement in different concentration with and without applied field a) low concentration, field off; b) low concentration, field on; c) high concentration, field off; d) high concentration, field on.

To verify IONPs arrangement behavior changes with concentration and alternating field, the microtomed four embedded IONPs samples were characterized by

TEM. IONPs were evenly spreaded in the polymer matrix without AC field of both 1 mg/ml and 5 mg/ml samples according to the TEM images (Figure 4-6 a, c). However,

IONPs of 5 mg/ml arranged to form ordered structure along the applied field and no such phenomena could be observed in 1 mg/ml sample in the field (Figure 4-6 b, d). This is the

103 evidence demonstrate the particles arrangement in the solution will be affected by the concentration and alternating magnetic field to form long range ordered structure.

Figure 4-6 TEM images of IONPs encapsulated in polymer matrix in different concentration with and without applied field a) low concentration, field off; b) low concentration, field on; c) high concentration, field off; d) high concentration, field on.

To investigate the IONPs heat efficiency changes by pH value and colloidal stability in aqueous solution, one step oxidization approach to transfer nanoparticles from hydrophobic to hydrophilic based on modified Kang’s procedure was employed. In SAR calculation, IONPs with 1, 3, and 5 mg/ml of iron, three concentrations in 5 different pH

(4, 6, 8, 10 and 12) were measured in triplicate. HCI and NaOH aqueous solution were utilized in pH adjustment.

104 The colloidal stability of different size IONPs at different pH of was characterized by hydrodynamic size and zeta potential using DLS.

Figure 4-7 Hydrodynamic size of different size IONPs change with pH in aqueous solution

The hydrodynamic size of all samples (Figure 4-6) increased remarkably when colloidal pH decrease to 6 and below. This is because the surface charges of nanoparticles, which work as repulsive interaction to keep colloidal stable, were neutralized when adjusting pH below 7. This was confirmed by the zeta potential changes of IONPs in Figure 4-7. Zeta potential of all samples was approaching to and passed the isoelectric point which indicated the colloidal lost its stability in neutral and acidic environment. There is the minimum number of hydrodynamic size at pH 10, and hydrodynamic size increases when pH was further raised to 12, which indicated the

105 agglomeration present again. There were no corresponding changes of zeta potential at pH 12 however, which could be due to the repression of the double layer of the particle charges when excessive hydroxide ions were induced in.

Figure 4-8 Zeta potential of different size IONPs change with pH in aqueous solution

There were the same correlations between particles size and heating efficiency in aqueous as in toluene been observed at high concentration of iron (Figure 4-8). However, the absolute SAR value was decrease significantly after oxidation. This could be due to the oxidation state changes from magnetite to maghemite lowered the magnetic properties of the nanoparticle, and/or the nanoparticles aggregation was processed in the oxidation comparing to the size in toluene (Figure 4-9).

106

Figure 4-9 Specific absorption rate changes of 5 mg/ml IONPs in aqueous solution with pH from 4 to 12 in the applied field of 41 kA/m and 147 kHz.

Figure 4-10 Hydrodynamic volume of different size IONPs in toluene.

107 No detectable heating changes of IONPs in 1 mg/ml and 3 mg/ml of Fe samples in the measurement.

Figure 4-11 Specific absorption rate changes with hydrodynamic diameter of 5 mg/ml IONPs in aqueous solution with pH from 4 to 12 in the applied field of 41 kA/m and 147 kHz.

The XRD patterns in Figure 4-11 clearly demonstrated the crystal structure of all

IONPs after oxidation remained as spinel phase, and the intensity of the peaks are correlated to the trend of the particle size changes.

108

Figure 4-12 X-ray diffraction patterns of oxidized IONPs with different sizes.

4.4 Conclusions

In summary, iron oxide nanoparticles with various sizes were obtained by thermodecomposing iron oleate with different ligand/precursor ratio at series of refluxing temperature. The maximum SAR value at particles diameter of 22nm could be due to the total relaxation time contributed by Brownian and Néel relaxation mechanisms. The SAR value is significantly affected by the concentration of iron in toluene, which opposite to the published report. This could be due to the iron oxide colloidal will interact with each other in the AC field and form the localized ordering structure which could restrain the relaxation of IONPs. A pH dependency of SAR was observed in aqueous solution, which confirms the pH will tune the surface charge of the nanoparticles and further affect the colloidal stability and SAR value. The results above have the implications for IONPs size

109 control and prediction in synthesis and optimization of IONPs colloidal performance in potential theranostics application.

110 CHAPTER 5 CONCLUSIONS

In this work, a potential kinetic mechanism describing the synthesis of IONPs was developed. In addition, the effects of the colloidal environmental conditions on the specific absorption rate of IONPs were investigated. With these developments it is possible to precisely predict and finely tune the IONPs mean average size and distribution to improve the magnetic properties as well as specify the colloidal stability and heating efficiency in aqueous suspension at varying pH. In theranostic applications, this colloidal environment will affect the IONPs especially in magnetic hyperthermia treatment as demonstrated by this work.

The ligand-precursor ratio effect on particles size of IONPs produced via the thermodecomposition synthesis approach was discussed in Chapter 2. The opposite trends of nanoparticles size change with the ligand-precursor ratio to the published results was attributed to the primary amine group of ligand acting as mild reducing agent which effectively increases the nucleation rate in the beginning of the precursor decomposition.

A comparative study was employed to clarify the promotion of the nucleation was contributed by the ligand-precursor molar ratio, not the increment of the total volume of the reaction solution. The crystallinity of synthesized IONPs could be improved by increasing the ligand-precursor molar ratio, which could further enhance the magnetic properties of IONPs and help their future biomedical applications.

Ligand molecule length effect on the kinetic mechanism of nucleation and crystal growth of IONPs was investigated in Chapter 3. Smaller nanoparticles will be obtained

111 by applying ligand with shorter chain length as the ligand works as diffusion barrier at the beginning of the nucleation. Ligands with longer chains restrain nucleation further.

This effect is not as significant at high concentration of ligand, due to the dominant factor of the diffusion barrier being migrated by the large amount of ligand in the solution. 2-D

Monte Carlo modeling was employed to simulate the nucleation scenarios of the solution by applying van der Waals and thermofluctuation in the system. The computed results match well with the experimental results and the theoretical hypothesis. This study can help better understanding kinetics mechanism in IONPs synthesis as well as perspective exploration of using computational modeling in nanoparticles size prediction.

Hyperthermia performance of IONPs was characterized in both toluene and aqueous suspensions in Chapter 4. The existence of optimized size (22 nm) in SAR measurement is different from that of in Rosensweig theory, which could be due to the applied field amplitude and frequency divergence. The dependency between concentration of iron and SAR value could be due to the long-range ordering at higher concentration restrained the Brownian and Néel relaxation in the alternating field.

Colloidal stability is significantly affected by pH value in aqueous solution. The most stabile colloids were obtained at pH 10. At pH greater or lower than 10, can result in further aggregation, which may be due to either neutralize of the charge on the particles or the repression of double layer. This study could avail of preparing adaptive nanoparticle to minimize heating loss at different pH environment in magnetic hyperthermia treatment.

112 CHAPTER 6 FUTURE WORK

Significant progress has been made in the studies above in understanding the kinetic mechanism and role of the reaction condition and reagent in IONPs synthesis as well as specific absorption rate changes with colloidal stability by varying the solution pH value. However, additional work needs to be carried out to achieve the goal in biomedical aspect of synthesis Safe, Simple, Stable and Strengthened “4S” iron oxide nanoparticles.

These “4S” are four individual required properties of an ideal iron oxide nanoparticles in theranostic applications, where each property has intrinsic correlations to each other. Safe requires synthesized IONPs are provided with high biocompatibility and low cytotoxicity. Simple refers to the easy procedure of preparing IONPs, and the ability to scale the reaction up for massive productions as well as minimizing the required modification steps. Stable means the stability of the produced IONPs colloidal preventing from oxidation and agglomerations, more importantly it means the consistency of physical and chemical properties including size and shape between IONPs from batches to batches. Last but not least, strengthened IONPs require the high efficiency and/or good performance in biomedical applications. This will not only enhance the curative effects but also minimize the dosage of particles required in patients.

It is not likely to achieve the goals in one action, but there could be some approachable work the be put into effect:

113 6.1 IONPs complex synthesis

IONPs could be applied in various biomedical aspect, however it is not easy to build multipurpose IONPs by only iron oxide one compound. Dual or even multi components are necessary for combining diagnostic and therapeutic functions into one nanoparticle. The secondary component can also improve the intrinsic property in practical applications. The perspective component candidates are silica, ferrite, titania and noble metals.

Silica could be performed as stabilization layer to transfer IONPs into aqueous without hesitation, moreover, it could work as a media due to its inert, amorphous phase.

It will be facilitated to process surface modification or heat treatment on IONPs with no consideration of crystal growth or agglomeration once coating with silica. It’s also approachable to build “onion” like nanodevice with different component out side of

IONPs by utilizing silica as intermedia. One potential obstacle is that silica layer can easily grow thick which will prohibit the resulted IONPs being applied in the body if they were oversized.

Ferrite has twice the saturation magnetization as magnetite, which is desired in almost all applications. The major drawback of ferrite materials is its oxidation tendency.

Therefore, build a core-shell structure with iron oxide can enhance the magnetic property without affecting biocompatibility verses something like cobalt.

114 Titania as a biomedical material has been widely applied as media in hydroxyapatite-titanium complex formation in osseous restore. It’s a promising approach to embed IONPs on the surface of titania film for long term diagnosis and/or in vivo recirculating treatment. The possible fabrication routes could be (1) Synthesis iron oxide or ferrite-titania nanoparticles first using iron or iron and titanium organic precursors by the thermodecomposition procedure. Then suspend the obtained nanoparticle in a titania sol which could be pre-hydrolysis from titanium isopropoxide for example. After the suspending process the nanoparticle sol could be dip-coated or spin coated onto a substrate like silica, silicon, and titanium depending on the needs. Coated substrate could be heat treated at 450 °C for anatase or above 500 °C for rutile. (2) Another possible approach is decomposing iron-titanium precursor directly in the presence of substrate. A pre-obtained intermediate porous amorphous layer on the substrate maybe necessary. The advantage of this is that the iron oxide nanoparticles could be preserved from growing bigger due to they are surrounded by titania. Moreover the synthesized material could be used as not only the in vivo recirculating treatment but also as recyclable photocatalyst in the application of water treatment.

It’s still a challenge to synthesize uniform coated IONPs with noble metal, such as gold, without aggregations. Gold is the top-pick material in synthesizing bi-mode IONPs due to its inert chemical properties as well as attractive optical performance. One shortcut of synthesizing evenly coated Au/Fe3O4 is utilizing silica as a median if there is adequate tolerance on the final particles size. The amorphous phase can sufficiently reduce the interfacial strain from lattice mismatch. If the final size of resulted nanoparticles is

115 critical to the further analysis or application, two-step coating is a promising method. In this method, IONPs should be transferred into aqueous solution first by modification.

Then a freshly synthesized gold seeds (2 nm-4 nm) solution will be mixed with IONPs and gold will attach on the IONPs by electrostatic force. Then additional gold salt will be added in the solution and further reduced by a reducing agent on the particles surface due to the surface energy. The drawback of this procedure is multiple steps could reduce the yield rate and raises the potential for contamination as well as reaction inconsistency.

Also, it’s only good for the IONPs of the size much larger than the size of the gold seed,

~ 10 nm. To overcome these challenges, combing seed attachment and layer growth as one step is worth of investigation

6.2 Kinetics modeling including precursor chemical reactivity

Nucleation and crystal growth simulation can help better understand the kinetic mechanism of reaction, predict nanoparticles morphology and confirm the experimental results. However, the simulations only included heat and mass transfer by applied force so far. The reactivity (chemical potential) of precursor in the reaction has yet to be considered. By switching precursors in the reaction, it could obtain IONPs with different size, shape and even crystallinity, which are the critical factors can change the magnetic properties and even the performance of IONPs in biomedical applications. Therefore, it could be conducive to depict and compare the nucleation process across different precursors at their own reaction conditions if activation energy of precursor and / or initial thermodynamic term can be induced into the simulation algorithm. Specifically,

116 the mystery of nanoparticles morphology difference between the precursor of iron oleate and Fe(acac)3 and the precursor-ligand-solvent matching relevance would be unveiled.

Moreover, it is expected to fully control and utilize driving force of crystal growth to synthesize size tunable triangular, rodlike or other high anisotropic iron oxide nanoparticles.

In order to achieve the goal above, one feasible approach is to add a chemical potential term in the Langevin equation and expand the start point of simulation time from right after to right before the decomposition process. It can contribute to represent the nucleation and growth process precisely if temperature could be considered as the functional of time in Langevin equation. Space boundary can be kept as continues condition whereas the initial chemical boundary, especially the velocity of the decomposition produces should be simulated as a function of chemical potential at that time. The interactions of solvent molecule to ligand and precursor should be included in the systematic force instead of simplifying as a part of friction coefficient when some high polarity solvent, such as ionic liquid, was discussed. The universal algorithm of simulating nucleation and crystal growth process across different reaction hierarchies is in sight, if most major kinetic and thermodynamic factors could be included in the model.

6.3 Anisotropic IONPs synthesis

High anisotropic IONPs can improve the SAR performance effectively. However, it is challenging to unveil the proper approach of synthesizing uniform anisotropic IONPs due to their thermodynamically unflavored morphology. It’s essential to discover the

117 media, which can occupy certain site of IONPs crystal and change their preferable growth directions.

6.4 One pot synthesis of hydrophilic IONPs by thermodecompostiton procedure with tunable size

Monodisperse nanoparticle with tunable size are mainly synthesized in non- aqueous solution, which are necessary to transferred into aqueous solution for biomedical application. The hydrophilic transition processes are time and money consuming, and/or lead oxidation of particle, and even may induce secondary contaminations. One pot synthesis of hydrophilic IONPs by thermodecompostiton procedure with tunable size are favored and may approachable by using stabilizing ligand with hydrophilic functional groups on both ends. The ideal molecules could be combination of amine-amine or amine-thiol, because the amine group can help improve the particles yield rate in the synthesis, and both amine and thiol group are favored for the succeed modifications like ligand exchange or layer coating. Catechol and carboxyl groups could be considered as well and the only obstacle is the further ligand exchange might not be simple.

6.5 SAR change with field and frequency

It is also worth of investigating the real relevance between applied field condition and SAR value of IONPs since the linear response theory has its own restrictions. This research can help build a standard conversion process to compare the research results between different characterization instruments.

118 REFERENCES

1. Laurent, S.; Forge, D.; Port, M.; Roch, A.; Robic, C.; Elst, L. V.; Muller, R. N. Magnetic Iron Oxide Nanoparticles: Synthesis, Stabilization, Vectorization, Physicochemical Characterizations, and Biological Applications Chem. Rev. 2008, 108, 2064.

2. Jun, Y. W.; Huh, Y. M.; Choi, J. S.; Lee, J. H.; Song, H. T.; Kim, S.; Yoon, S.; Kim, K. S.; Shin, J. S.; Suh, J. S., et al. Nanoscale Size Effect of Magnetic Nanocrystals and Their Utilization for Cancer Diagnosis Via Magnetic Resonance Imaging J. Am. Chem. Soc. 2005, 127 (16), 5732-5733.

3. Weimuller, M. G.; Zeisberger, M.; Krishnan, K. M. Size-Dependant Heating Rates of Iron Oxide Nanoparticles for Magnetic fluid Hyperthermia J. Magn. Magn. Mater. 2009, 321, 1947-1950.

4. Nath, K.; Nelson, D. S.; Ho, A. M.; Lee, S. C.; Darpolor, M. M.; Pickup, S.; Zhou, R.; Heitjan, D. F.; Leeper, D. B.; Glickson, J. D. P-31 and H-1 Mrs of Db-1 Melanoma Xenografts: Lonidamine Selectively Decreases Tumor Intracellular Ph and Energy Status and Sensitizes Tumors to Melphalan Nmr in Biomedicine 2013, 26 (1), 98-105.

5. Chen, Y.; Yin, Q.; Ji, X. F.; Zhang, S. J.; Chen, H. R.; Zheng, Y. Y.; Sun, Y.; Qu, H. Y.; Wang, Z.; Li, Y. P., et al. Manganese Oxide-Based Multifunctionalized Mesoporous Silica Nanoparticles for Ph-Responsive Mri, Ultrasonography and Circumvention of Mdr in Cancer Cells Biomaterials 2012, 33 (29), 7126-7137.

6. Sonvico, F.; Mornet, S.; Vasseur, S.; Dubernet, C.; Jaillard, D.; Degrouard, J.; Hoebeke, J.; Duguet, E.; Colombo, P.; Couvreur, P. Folate-Conjugated Iron Oxide Nanoparticles for Solid Tumor Targeting as Potential Specific Magnetic Hyperthermia Mediators: Synthesis, Physicochemical Characterization, and in Vitro Experiments Bioconjugate Chemistry 2005, 16 (5), 1181-1188.

7. Coffey, W. T.; Crothers, D. S. F.; Dormann, J. L.; Geoghegan, L. J.; Kalmykov, Y. P.; Waldron, J. T.; Wickstead, A. W. The Effect of an Oblique Magnetic Field on the Superparamagnetic Relaxation Time Journal of Magnetism and Magnetic Materials 1995, 145 (3), L263-L267.

119 8. Mørup, S.; Hansen, M. F.; Frandsen, C., 1.14 - Magnetic Nanoparticles. In Comprehensive Nanoscience and Technology, Editors-in-Chief: David, L. A.; Gregory, D. S.; Gary, P. W., Eds. Academic Press: Amsterdam, 2011; pp 437-491.

9. E.R. Cohen, T. C., J.G. Frey, B. Holmström, K. Kuchitsu, R. Marquardt, I. Mills, F. Pavese, M. Quack, J. Stohner, H.L. Strauss, M. Takami, and A.J. Thor, Quantities, Units and Symbols in Physical Chemistry. In IUPAC Physical and Biophysical Chemistry Division [Online] RSC Publishing: Cambridge 2007.

10. S Saini; RB Frankel; DD Stark; JT Ferrucci, J. Magnetism: A Primer and Review American Journal of Roentgenology 1988, 150 (4), 735-743.

11. Lyo, S. K. Magnetic Susceptibility in a Layered Superlattice Solid State Communications 1984, 51 (9), 709-712.

12. Handschuh, S.; Landes, J.; Köbler, U.; Sauer, C.; Kisters, G.; Fuss, A.; Zinn, W. Magnetic Properties of Amorphous Fe in Fe/Y Layered Structures Journal of Magnetism and Magnetic Materials 1993, 119 (3), 254-260.

13. Shamonin, M.; Snarskii, A.; Zhenirovskyy, M. Effective Magnetic Permeability of Ferromagnetic Composites. Theoretical Description and Comparison with Experiment NDT & E International 2004, 37 (1), 35-40.

14. Bretcanu, O.; Spriano, S.; Verné, E.; Cöisson, M.; Tiberto, P.; Allia, P. The Influence of Crystallised Fe3o4 on the Magnetic Properties of Coprecipitation-Derived Ferrimagnetic Glass–Ceramics Acta Biomaterialia 2005, 1 (4), 421-429.

15. Ehrhart, P.; Schönfeld, B.; Ettwig, H. H.; Pepperhoff, W. The Lattice Structure of Antiferromagnetic Γ-Iron Journal of Magnetism and Magnetic Materials 1980, 22 (1), 79-85.

16. Josephson, L.; Groman, E. V.; Menz, E.; Lewis, J. M.; Bengele, H. A Functionalized Superparamagnetic Iron Oxide Colloid as a Receptor Directed Mr Contrast Agent Magnetic Resonance Imaging 1990, 8 (5), 637-646.

120 17. Maleki, H.; Simchi, A.; Imani, M.; Costa, B. F. O. Size-Controlled Synthesis of Superparamagnetic Iron Oxide Nanoparticles and Their Surface Coating by Gold for Biomedical Applications Journal of Magnetism and Magnetic Materials 2012, 324 (23), 3997-4005.

18. Wei, X.-K.; Su, Y.; Sui, Y.; Zhou, Z.; Yao, Y.; Jin, C.; Yu, R. Ferromagnetic Antiphase Domain Boundary in Mn-Doped Hexagonal Batio[Sub 3] Multiferroics Applied Physics Letters 2013, 102 (24), 242910.

19. Coffey, W. T. The Low Damping Néel Relaxation Time for Fine Ferromagnetic Particles from the Energy Diffusion Method of Kramers Journal of Molecular Structure 1999, 479 (2–3), 261-269.

20. Gittleman, J. I.; Abeles, B.; Bozowski, S. Superparamagnetism and Relaxation Effects in Granular Ni-SiO2 and Ni-Al2O3 Films Physical Review B 1974, 9 (9), 3891- 3897.

21. Mørup, S.; Tronc, E. Superparamagnetic Relaxation of Weakly Interacting Particles Physical Review Letters 1994, 72 (20), 3278-3281.

22. Darby, M. I. Tables of the Brillouin Function and of the Related Function for the Spontaneous Magnetization British Journal of Applied Physics 1967, 18 (10), 1415.

23. Arbain, R.; Othman, M.; Palaniandy, S. Preparation of Iron Oxide Nanoparticles by Mechanical Milling Minerals Engineering 2011, 24 (1), 1-9.

24. Tartaj, P.; Morales, M. d. P.; Veintemillas-Verdaguer, S.; Gonzalez-Carreno, T.; Serna, C. J. The Preparation of Magnetic Nanoparticles for Applications in Biomedicine JOURNAL OF PHYSICS D: APPLIED PHYSICS 2003, 36.

25. Mooney, J. B.; Radding, S. B. Spray Pyrolysis Processing Annual Review of Materials Science 1982, 12, 81-101.

121 26. Ge, J.; Hu, Y.; Biasini, M.; Dong, C.; Guo, J.; Beyermann, W. P.; Yin, Y. One-Step Synthesis of Highly Water-Soluble Magnetite Colloidal Nanocrystals Chem. Eur. J. 2007, 13, 7153.

27. Rao, P. M.; Zheng, X. L. Unique Magnetic Properties of Single Crystal Gamma- Fe2O3 Nanowires Synthesized by Flame Vapor Deposition Nano Letters 2011, 11 (6), 2390-2395.

28. Zhang, Z. J.; Wei, B. Q.; Ajayan, P. M. Self-Assembled Patterns of Iron Oxide Nanoparticles by Hydrothermal Chemical-Vapor Deposition Applied Physics Letters 2001, 79 (25), 4207-4209.

29. Lefebure, S.; Dubois, E.; Cabuil, V.; Neveu, S.; Massart, R. Monodisperse Magnetic Nanoparticles: Preparation and Dispersion in Water and Oils J. Mater. Res. 1998, 13, 7.

30. Gao, G. H.; Im, G. H.; Kim, M. S.; Lee, J. W.; Yang, J.; Jeon, H.; Lee, J. H.; Lee, D. S. Magnetite-Nanoparticle-Encapsulated Ph-Responsive Polymeric Micelle as an Mri Probe for Detecting Acidic Pathologic Areas Small 2010, 6 (11), 1201-1204.

31. Hyeon, T. Chemical Synthesis of Magnetic Nanoparticles Chem. Commun. 2003, 8, 927.

32. Lian, S. Y.; Wang, E.; Kang, Z. H.; Bai, Y. P.; Gao, L.; Jiang, M.; Hu, C. W.; Xu, L. Synthesis of Magnetite Nanorods and Porous Hematite Nanrods Solid State Commun 2004, 129 (8), 485-490.

33. Salazar, J. S.; Perez, L.; de Abril, O.; Lai, T. P.; Ihiawakrim, D.; Vazquez, M.; Greneche, J. M.; Begin-Colin, S.; Pourroy, G. Magnetic Iron Oxide Nanoparticles in 10- 40 Nm Range: Composition in Terms of Magnetite/Maghemite Ratio and Effect on the Magnetic Properties Chem. Mat. 2011, 23 (6), 1379-1386.

34. Croker, D.; Loan, M.; Hodnett, B. K. Kinetics and Mechanisms of the Hydrothermal Crystallization of Calcium Titanate Species Crystal Growth & Design 2009, 9 (5), 2207- 2213.

122 35. Shevchenko, E. V.; Talapin, D. V.; Schnablegger, H.; Kornowski, A.; Festin, O.; Svedlindh, P.; Haase, M.; Weller, H. Study of Nucleation and Growth in the Organometallic Synthesis of Magnetic Alloy Nanocrystals: The Role of Nucleation Rate in Size Control of Copt3 Nanocrystals J. AM. CHEM. SOC. 2003, 125.

36. Sun, C.; Lee, J.; Zhang, M. Magnetic Nanoparticles in Mr Imaging and Drug Delivery Adv. Drug Delivery Rev. 2008, 60, 1252-1265.

37. Kim, S. B.; Cai, C.; Sun, S.; Sweigart, D. A. Incorporation of Fe3o4 Nanoparticles into Organometallic Coordination Polymers by Nanoparticle Surface Modification Angew. Chem.-Int. Edit. 2009, 48 (16), 2907-2910.

38. Bersani, D.; Lottici, P. P.; Montenero, A. Micro-Raman Investigation of Iron Oxide Films and Powders Produced by Sol-Gel Syntheses Journal of Raman Spectroscopy 1999, 30 (5), 355-360.

39. Wang, L.; Luo, J.; Maye, M. M.; Fan, Q.; Rendeng, Q.; Engelhard, M. H.; Wang, C.; Lin, Y.; Zhong, C.-J. Iron Oxide-Gold Core-Shell Nanoparticles and Thin Film Assembly J. Mater. Chem 2005, 15.

40. Guardia, P.; Perez, N.; Labarta, A.; Batlle, X. Controlled Synthesis of Iron Oxide Nanoparticles over a Wide Size Range Langmuir 2010, 26 (8), 5843-5847.

41. Iskandar, F. Nanoparticle Processing for Optical Applications – a Review Advanced Powder Technology 2009, 20 (4), 283-292.

42. Park, J. B.; Jeong, S. H.; Jeong, M. S.; Kim, J. Y.; Cho, B. K. Synthesis of Carbon- Encapsulated Magnetic Nanoparticles by Pulsed Laser Irradiation of Solution Carbon 2008, 46 (11), 1369-1377.

43. Ogi, T.; Kaihatsu, Y.; Iskandar, F.; Tanabe, E.; Okuyama, K. Synthesis of Nanocrystalline Gan from Ga2o3 Nanoparticles Derived from Salt-Assisted Spray Pyrolysis Advanced Powder Technology 2009, 20 (1), 29-34.

123 44. Huang, C. C.; Liu, T. Y.; Su, C. H.; Lo, Y. W.; Chen, J. H.; Yeh, C. S. Superparamagnetic Hollow and Paramagnetic Porous Gd2o3 Particles Chem. Mat. 2008, 20 (12), 3840-3848.

45. Kang, Y.; Risbud, S.; Rabolt, J. F.; Stroeve, P. Synthesis and Characterization of Nanometer-Size Fe3 O4 And-Fe2 O3 Particles Chem. Mater. 1996, 8, 2209-2211.

46. Babes, L.; Denizot, B.; Tanguy, G.; Le Jeune, J. J.; Jallet, P. Synthesis of Iron Oxide Nanoparticles Used as Mri Contrast Agents: A Parametric Study Journal of Colloid and Interface Science 1999, 212 (2), 474-482.

47. Tronc, E.; Belleville, P.; Jolivet, J. P.; Livage, J. Transformation of Ferric Hydroxide into Spinel by Iron(Ii) Adsorption Langmuir 1992, 8 (1), 313-319.

48. Iida, H.; Takayanagi, K.; Nakanishi, T.; Osaka, T. Synthesis of Fe3o4 Nanoparticles with Various Sizes and Magnetic Properties by Controlled Hydrolysis Journal of Colloid and Interface Science 2007, 314 (1), 274-280.

49. Gupta, A. K.; Wells, S. Surface-Modified Superparamagnetic Nanoparticles for Drug Delivery: Preparation, Characterization, and Cytotoxicity Studies IEEE transactions on nanobioscience 2004, 3 (1), 66-73.

50. Kim, D. K.; Zhang, Y.; Voit, W.; Rao, K. V.; Muhammed, M. Synthesis and Characterization of Surfactant-Coated Superparamagnetic Monodispersed Iron Oxide Nanoparticles Journal of Magnetism and Magnetic Materials 2001, 225 (1–2), 30-36.

51. Xu, Z. C.; Shen, C. M.; Hou, Y. L.; Gao, H. J.; Sun, S. S. Oleylamine as Both Reducing Agent and Stabilizer in a Facile Synthesis of Magnetite Nanoparticles Chem. Mat. 2009, 21 (9), 1778-1780.

52. Liang, X.; Wang, X.; Zhuang, J.; Chen, Y. T.; Wang, D. S.; Li, Y. D. Synthesis of Nearly Monodisperse Iron Oxide and Oxyhydroxide Nanocrystals Advanced Functional Materials 2006, 16 (14), 1805-1813.

124 53. Jia, C. J.; Sun, L. D.; Yan, Z. G.; You, L. P.; Luo, F.; Han, X. D.; Pang, Y. C.; Zhang, Z.; Yan, C. H. Iron Oxide Nanotubes - Single-Crystalline Iron Oxide Nanotubes Angew. Chem.-Int. Edit. 2005, 44 (28), 4328-4333.

54. Rockenberger, J.; Scher, E. C.; Alivisatos, A. P. A New Nonhydrolytic Single- Precursor Approach to Surfactant-Capped Nanocrystals of Transition Metal Oxides Journal Name: Journal of the American Chemical Society; Journal Volume: 121; Journal Issue: 49; Other Information: PBD: 15 Dec 1999 1999, Medium: X; Size: page(s) 11595-11596.

55. Hyeon, T.; Lee, S.; Park, J.; Chung, Y.; Na, H. Synthesis of Highly Crystalline and Monodisperse Maghemite Nanocrystallites without a Size-Selection Process J. Am. Chem. Soc. 2001, 123, 4.

56. Sun, S.; Zeng, H.; Robinson, D. B.; Raoux, S.; Rice, P. M.; Wang, S. X.; Li, G. Monodisperse M Fe2 O4(M = Fe, Co, Mn) Nanoparticles J. Am. Chem. Soc. 2004, 126, 273-279.

57. Park, J.; An, K. J.; Hwang, Y. S.; Park, J. G.; Noh, H. J.; Kim, J. Y.; Park, J. H.; Hwang, N. M.; Hyeon, T. Ultra-Large-Scale Syntheses of Monodisperse Nanocrystals Nature Materials 2004, 3 (12), 891-895.

58. Li, Z.; Chen, H.; Bao, H.; Gao, M. One-Pot Reaction to Synthesize Water-Soluble Magnetite Nanocrystals Chem. Mat. 2004, 16 (8), 1391-1393.

59. Li, Z.; Wei, L.; Gao, M. Y.; Lei, H. One-Pot Reaction to Synthesize Biocompatible Magnetite Nanoparticles Advanced Materials 2005, 17 (8), 1001-1005.

60. Hu, F. Q.; Wei, L.; Zhou, Z.; Ran, Y. L.; Li, Z.; Gao, M. Y. Preparation of Biocompatible Magnetite Nanocrystals for in Vivo Magnetic Resonance Detection of Cancer Advanced Materials 2006, 18 (19), 2553-2556.

61. Cai, W.; Wan, J. Q. Facile Synthesis of Superparamagnetic Magnetite Nanoparticles in Liquid Polyols Journal of Colloid and Interface Science 2007, 305 (2), 366-370.

125 62. Young, K. L.; Xu, C. J.; Xie, J.; Sun, S. H. Conjugating Methotrexate to Magnetite (Fe3O4) Nanoparticles Via Trichloro-S-Triazine Journal of Materials Chemistry 2009, 19 (35), 6400-6406.

63. Park, J.; An, K.; Hwang, Y.; Park, J.; Noh, H.; Kim, J.; Park, J.; Hwang, N.; Hyeon, T. Ultra-Large-Scale Syntheses of Monodisperse Nanocrystals Nature Mater. 2004, 3, 891-895.

64. Zeng, H.; Sun, S.; Li, J.; Wang, Z. L.; Liu, J. P. Tailoring Magnetic Properties of Core/Shell Nanoparticles Applied Physics Letters 2004, 85 (5), 792-794.

65. Gyergyek, S.; Makovec, D.; Kodre, A.; Arčon, I.; Jagodič, M.; Drofenik, M. Influence of Synthesis Method on Structural and Magnetic Properties of Cobalt Ferrite Nanoparticles J Nanopart Res 2010, 12 (4), 1263-1273.

66. Capek, I. Preparation of Metal Nanoparticles in Water-in-Oil (W/O) Microemulsions Advances in colloid and interface science 2004, 110 (1–2), 49-74.

67. Santra, S.; Tapec, R.; Theodoropoulou, N.; Dobson, J.; Hebard, A.; Tan, W. H. Synthesis and Characterization of Silica-Coated Iron Oxide Nanoparticles in Microemulsion: The Effect of Nonionic Surfactants Langmuir 2001, 17 (10), 2900-2906.

68. Okoli, C.; Sanchez-Dominguez, M.; Boutonnet, M.; Jaras, S.; Civera, C.; Solans, C.; Kuttuva, G. R. Comparison and Functionalization Study of Microemulsion-Prepared Magnetic Iron Oxide Nanoparticles Langmuir 2012, 28 (22), 8479-8485.

69. Lu, Y.; Yin, Y. D.; Mayers, B. T.; Xia, Y. N. Modifying the Surface Properties of Superparamagnetic Iron Oxide Nanoparticles through a Sol-Gel Approach Nano Letters 2002, 2 (3), 183-186.

70. Cui, H. T.; Liu, Y.; Ren, W. Z. Structure Switch between Alpha-Fe2O3, Gamma- Fe2O3 and Fe3O4 During the Large Scale and Low Temperature Sol-Gel Synthesis of Nearly Monodispersed Iron Oxide Nanoparticles Advanced Powder Technology 2013, 24 (1), 93-97.

126 71. Sun, S.; Zeng, H. Size-Controlled Synthesis of Magnetite Nanoparticles J. Am. Chem. Soc. 2002, 124, 8204.

72. Jana, N.; Chen, Y.; Peng, X. Size- and Shape-Controlled Magnetic (Cr, Mn, Fe, Co, Ni) Oxide Nanocrystals Via a Simple and General Approach Chem. Mater. 2004, 16, 3931-3936.

73. Bronstein, L. M.; Huang, X.; Retrum, J.; Schmucher, A.; Pink, M.; Stein, B. D.; Dragnea, B. Influence of Iron Oleate Complex Structure on Iron Oxide Nanoparticle Formation Chem. Mater. 2007, 19, 3624.

74. Kim, D.; Lee, N.; Park, M.; Kim, B. H.; An, K.; Hyeon, T. Synthesis of Uniform Ferrimagnetic Magnetite Nanocubes Journal of the American Chemical Society 2009, 131 (2), 454.

75. Wang, Y.; Yang, H. Synthesis of Iron Oxide Nanorods and Nanocubes in an Imidazolium Ionic Liquid Chemical Engineering Journal 2009, 147 (1), 71-78.

76. LaMer, V. K.; Dinegar, R. H. Theory, Production and Mechanism of Formation of Monodispersed Hydrosols J. Am. Chem. Soc. 1950, 72, 4847-4855.

77. Murray, C. B.; Kagan, C. R.; Bawendi, M. G. Synthesis and Characterizetion of Monodisperse Nanocrystals and Close-Packed Nanocrystal Assemblies Annu. Rev. Mater. Sci. 2000, 30, 68.

78. Pileni, M. P. Nanosized Particles Made in Colloidal Assemblies Langmuir 1997, 13, 3266-3276.

79. Lee, H.; Yang, H.; Holloway, P. H. Single-Step Growth of Colloidal Ternary Zncdse Nanocrystals Journal of Luminescence 2007, 126, 314.

80. Avrami, M. Kinetics of Phase Change. I General Theory The Journal of Chemical Physics 1939, 7 (12), 1103-1112.

127 81. Tsukimura, K.; Suzuki, M.; Suzuki, Y.; Murakami, T. Kinetic Theory of Crystallization of Nanoparticles Crystal Growth & Design 2010, 10 (8), 3596-3607.

82. Kwon, S. G.; Piao, Y.; Park, J.; Angappane, S.; Jo, Y.; Hwang, N. M.; Park, J. G.; Hyeon, T. Kinetics of Monodisperse Iron Oxide Nanocrystal Formation by "Heating-up" Process Journal of the American Chemical Society 2007, 129 (41), 12571-12584.

83. Shubayev, V. I.; Pisanic, T. R.; Jin, S. H. Magnetic Nanoparticles for Theragnostics Adv. Drug Deliver Rev. 2009, 61 (6), 467-477.

84. Velasquez, E. A.; Mazo-Zuluaga, J.; Mejia-Lopez, J. Size Dependence Study of the Ordering Temperature in the Fast Monte Carlo Method J Nanopart Res 2013, 15 (2).

85. Northrup, S. H.; Allison, S. A.; McCammon, J. A. Brownian Dynamics Simulation of Diffusion-Influenced Bimolecular Reactions Journal of Chemical Physics 1984, 80 (4), 1517-1526.

86. Chen, J. C.; Kim, A. S. Brownian Dynamics, Molecular Dynamics, and Monte Carlo Modeling of Colloidal Systems Advances in colloid and interface science 2004, 112 (1- 3), 159-73.

87. Frenkel, D.; Eppenga, R. Monte Carlo Study of the Isotropic-Nematic Transition in a Fluid of Thin Hard Disks Physical Review Letters 1982, 49 (15), 1089-1092.

88. Lyshevski, M. A. In Brownian Dynamics: Molecular Systems Modeling and Control, Nanotechnology, 2004. 4th IEEE Conference on, 16-19 Aug. 2004; 2004; pp 225-227. 89. Allen, M. P. Brownian Dynamics Simulation of a Chemical Reaction in Solution Molecular Physics 1980, 40 (5), 1073-1087.

90. van Gunsteren, W. F.; Berendsen, H. J. C. Algorithms for Brownian Dynamics Molecular Physics 1982, 45 (3), 637-647.

91. Ermak, D. L. A Computer Simulation of Charged Particles in Solution. I. Technique and Equilibrium Properties The Journal of Chemical Physics 1975, 62 (10), 4189-4196.

128 92. Scala, A. Brownian Dynamics Simulation of Polydisperse Hard Spheres Eur. Phys. J.-Spec. Top. 2013, 216 (1), 21-29.

93. Allen, M. P. Algorithms for Brownian Dynamics Molecular Physics 1982, 47 (3), 599-601.

94. Turq, P.; Lantelme, F.; Friedman, H. L. Brownian Dynamics: Its Application to Ionic Solutions The Journal of Chemical Physics 1977, 66 (7), 3039-3044.

95. Alder, B. J.; Wainwright, T. E. Studies in Molecular Dynamics. I. General Method The Journal of Chemical Physics 1959, 31 (2), 459-466.

96. Rahman, A. Correlations in the Motion of Atoms in Liquid Argon Physical Review 1964, 136 (2A), A405-A411.

97. Verlet, L. Computer "Experiments" on Classical Fluids. I. Thermodynamical Properties of Lennard-Jones Molecules Physical Review 1967, 159 (1), 98-103.

98. Rustad, J. R.; Felmy, A. R.; Bylaska, E. J. Molecular Simulation of the Magnetite- Water Interface Molecular simulation of the magnetite-water interface 2003, 67.

99. Starr, F. W.; Schrøder, T. B.; Glotzer, S. C. Molecular Dynamics Simulation of a Polymer Melt with a Nanoscopic Particle Macromolecules 2002, 35 (11), 4481-4492.

100. Spall, J. C. Estimation Via Markov Chain Monte Carlo Control Systems, IEEE 2003, 23 (2), 34-45.

101. Metropolis, N.; Rosenbluth, A. W.; Rosenbluth, M. N.; Teller, A. H.; Teller, E. Equation of State Calculations by Fast Computing Machines The Journal of Chemical Physics 1953, 21 (6), 1087-1092.

102. Sanz, E.; Marenduzzo, D. Dynamic Monte Carlo Versus Brownian Dynamics: A Comparison for Self-Diffusion and Crystallization in Colloidal Fluids Journal of Chemical Physics 2010, 132 (19).

129 103. Guarnieri, F.; Still, W. C. A Rapidly Convergent Simulation Method: Mixed Monte Carlo/Stochastic Dynamics Journal of Computational Chemistry 1994, 15 (11), 1302- 1310.

104. Guarnieri, F. Theory and Algorithms for Mixed Monte Carlo-Stochastic Dynamics Simulations J Math Chem 1995, 18 (1), 25-35.

105. Hwang, H.; Schatz, G. C.; Ratner, M. A. Kinetic Lattice Grand Canonical Monte Carlo Simulation for Ion Current Calculations in a Model Ion Channel System Journal of Chemical Physics 2007, 127 (2).

106. van der Zee, J. Heating the Patient: A Promising Approach? Annals of Oncology 2002, 13 (8), 1173-1184.

107. Kumar, C.; Mohammad, F. Magnetic Nanomaterials for Hyperthermia-Based Therapy and Controlled Drug Delivery Advanced Drug Delivery Reviews 2011, 63 (9), 789-808.

108. Wust, P.; Hildebrandt, B.; Sreenivasa, G.; Rau, B.; Gellermann, J.; Riess, H.; Felix, R.; Schlag, P. M. Hyperthermia in Combined Treatment of Cancer The Lancet Oncology 2002, 3 (8), 487-497.

109. Chang, E.; Alexander, H. R.; Libutti, S. K.; Hurst, R.; Zhai, S.; Figg, W. D.; Bartlett, D. L. Laparoscopic Continuous Hyperthermic Peritoneal Perfusion1 1 No Competing Interests Declared Journal of the American College of Surgeons 2001, 193 (2), 225-229.

110. Falk, M. H.; Issels, R. D. Hyperthermia in Oncology International Journal of Hyperthermia 2001, 17 (1), 1-18.

111. Ahmed, N.; Fessi, H.; Elaissari, A. Theranostic Applications of Nanoparticles in Cancer Drug Discovery Today 2012, 17 (17-18), 928-934.

112. Gilchrist, R. K.; Medal, R.; Shorey, W. D.; Hanselman, R. C.; Parrott, J. C.; Taylor, C. B. Selective Inductive Heating of Lymph Nodes Annals of surgery 1957, 146 (4), 596- 606.

130 113. Glöckl, G.; Hergt, R.; Zeisberger, M.; Dutz, S.; Nagel, S.; Weitschies, W. The Effect of Field Parameters, Nanoparticle Properties and Immobilization on the Specific Heating Power in Magnetic Particle Hyperthermia Journal of Physics: Condensed Matter 2006, 18 (38), S2935.

114. Hergt, R.; Dutz, S.; Muller, R.; Zeisberger, M. Magnetic Particle Hyperthermia: Nanoparticle Magnetism and Materials Development for Cancer Therapy Journal of Physics-Condensed Matter 2006, 18 (38), S2919-S2934.

115. Rosensweig, R. E. Heating Magnetic Fluid with Alternating Magnetic Field Journal of Magnetism and Magnetic Materials 2002, 252 (1-3), 370-374.

116. Kallumadil, M.; Tada, M.; Nakagawa, T.; Abe, M.; Southern, P.; Pankhurst, Q. A. Corrigendum to “Suitability of Commercial Colloids for Magnetic Hyperthermia” [J. Magn. Magn. Mater. 321 (2009) 1509–1513] Journal of Magnetism and Magnetic Materials 2009, 321 (21), 3650-3651.

117. Carrey, J.; Mehdaoui, B.; Respaud, M. Simple Models for Dynamic Hysteresis Loop Calculations of Magnetic Single-Domain Nanoparticles: Application to Magnetic Hyperthermia Optimization Journal of Applied Physics 2011, 109 (8).

118. Lima, E.; Torres, T. E.; Rossi, L. M.; Rechenberg, H. R.; Berquo, T. S.; Ibarra, A.; Marquina, C.; Ibarra, M. R.; Goya, G. F. Size Dependence of the Magnetic Relaxation and Specific Power Absorption in Iron Oxide Nanoparticles J Nanopart Res 2013, 15 (5).

119. de la Presa, P.; Luengo, Y.; Multigner, M.; Costo, R.; Morales, M. P.; Rivero, G.; Hernando, A. Study of Heating Efficiency as a Function of Concentration, Size, and Applied Field in Γ-Fe2O3 Nanoparticles The Journal of Physical Chemistry C 2012, 116 (48), 25602-25610.

120. Lee, J. E.; Lee, N.; Kim, H.; Kim, J.; Choi, S. H.; Kim, J. H.; Kim, T.; Song, I. C.; Park, S. P.; Moon, W. K., et al. Uniform Mesoporous Dye-Doped Silica Nanoparticles Decorated with Multiple Magnetite Nanocrystals for Simultaneous Enhanced Magnetic Resonance Imaging, Fluorescence Imaging, and Drug Delivery J. Am. Chem. Soc. 2010, 132 (2), 552-557.

131 121. Zeng, H.; Sun, S. Syntheses, Properties, and Potential Applications of Multicomponent Magnetic Nanoparticles Adv. Funct. Mater. 2008, 18 (9), 391-400.

122. Stone, R.; Willi, T.; Rosen, Y.; Mefford, O. T.; Alexis, F. Targeted Magnetic Hyperthermia Therapeutic Delivery 2011, 2 (6), 815-838.

123. Kim, T. H.; Jang, E. Y.; Lee, N. J.; Choi, D. J.; Lee, K. J.; Jang, J. T.; Choi, J. S.; Moon, S. H.; Cheon, J. Nanoparticle Assemblies as Memristors Nano Lett. 2009, 9 (6), 2229-2233.

124. Park, J.; Lee, E.; Hwang, N. M.; Kang, M.; Kim, S. C.; Hwang, Y.; Park, J. G.; Noh, H. J.; Kim, J.; Park, J. H., et al. One-Nanometer-Scale Size-Controlled Synthesis of Monodisperse Magnetic Iron Oxide Nanoparticles Angew. Chem. Int. Ed. 2005, 44, 2872- 2877.

125. Yu, W. W.; Falkner, J. C.; Yavuz, C. T.; Colvin, V. L. Synthesis of Monodisperse Iron Oxide Nanocrystals by Thermal Decomposition of Iron Carboxylate Salts Chem. Commun. 2004, 20 (20), 2306-2307.

126. Huber, D. L.; Venturini, E. L.; Martin, J. E.; Provencio, P. P.; Patel, R. J. Synthesis of Highly Magnetic Iron Nanoparticles Suitable for Field Structuring Using a Beta- Diketone Surfactant J. Magn. Magn. Mater. 2004, 278 (3), 311-316.

127. Meledandri, C. J.; Stolarczyk, J. K.; Ghosh, S.; Brougham, D. F. Nonaqueous Magnetic Nanoparticle Suspensions with Controlled Particle Size and Nuclear Magnetic Resonance Properties Langmuir 2008, 24, 14159-14166.

128. Yin, M.; Willis, A.; Redl, F.; Turro, N. J.; O'Brien, S. P. Influence of Capping Groups on the Synthesis of Gamma-Fe2O3 Nanocrystals J. Mater. Res. 2004, 19 (4), 1208-1215.

129. Salazar-Alvarez, G.; Qin, J.; Sepelak, V.; Bergmann, I.; Vasilakaki, M.; Trohidou, K. N.; Ardisson, J. D.; Macedo, W. A. A.; Mikhaylova, M.; Muhammed, M., et al. Cubic Versus Spherical Magnetic Nanoparticles: The Role of Surface Anisotropy J. Am. Chem. Soc. 2008, 130 (40), 13234-13239.

132 130. Xie, J.; Xu, C.; Xu, Z.; Hou, Y.; Young, K. L.; Wang, S. X.; Pourmond, N.; Sun, S. Linking Hydrophilic Macromolecules to Monodisperse Magnetite (Fe3 O4) Nanoparticles Via Trichloro-S-Triazine Chem. Mater. 2006, 18, 5401-5403.

131. Chemistry WebBook, National Institute of Standards and Technology, Material Measure Laboratory.

132. Mullin, J. W., Crystallization. 4th ed.; Butterworth-Heinemann: 2001; p 181-210.

133. Zalich, M. A.; Vadala, M. L.; Riffle, J. S.; Saunders, M.; St Pierre, T. G. Structural and Magnetic Properties of Cobalt Nanoparticles Encased in Siliceous Shells Chem. Mat. 2007, 19 (26), 6597-6604.

134. Jimenez-Villacorta, F.; Munoz-Martin, A.; Prieto, C. X-Ray Diffraction and Extended X-Ray Absorption Fine-Structure Characterization of Nonspherical Crystallographic Grains in Iron Thin Films J. Appl. Phys. 2004, 96 (11), 6224-6229.

135. Krycka, K. L.; Booth, R. A.; Hogg, C. R.; Ijiri, Y.; Borchers, J. A.; Chen, W. C.; Watson, S. M.; Laver, M.; Gentile, T. R.; Dedon, L. R., et al. Core-Shell Magnetic Morphology of Structurally Uniform Magnetite Nanoparticles Phys. Rev. Lett. 2010, 104 (20), 2072031-2072034.

136. Lee, N.; Hyeon, T. Designed Synthesis of Uniformly Sized Iron Oxide Nanoparticles for Efficient Magnetic Resonance Imaging Contrast Agents Chem. Soc. Rev. 2012, 41 (7), 2575-2589.

137. Parveen, S.; Misra, R.; Sahoo, S. K. Nanoparticles: A Boon to Drug Delivery, Therapeutics, Diagnostics and Imaging Nanomedicine : nanotechnology, biology, and medicine 2012, 8 (2), 147-166.

138. Guardia, P.; Di Corato, R.; Lartigue, L.; Wilhelm, C.; Espinosa, A.; Garcia- Hernandez, M.; Gazeau, F.; Manna, L.; Pellegrino, T. Water-Soluble Iron Oxide Nanocubes with High Values of Specific Absorption Rate for Cancer Cell Hyperthermia Treatment ACS Nano 2012, 6 (4), 3080-3091.

133 139. Gu, H.; Xu, K.; Yang, Z.; Chang, C. K.; Xu, B. Synthesis and Cellular Uptake of Porphyrin Decorated Iron Oxide Nanoparticles-a Potential Candidate for Bimodal Anticancer Therapy Chem. Commun 2005.

140. Fan, Z.; Senapati, D.; Singh, A. K.; Ray, P. C. Theranostic Magnetic Core- Plasmonic Shell Star Shape Nanoparticle for the Isolation of Targeted Rare Tumor Cells from Whole Blood, Fluorescence Imaging, and Photothermal Destruction of Cancer Molecular Pharmaceutics 2013, 10 (3), 857-866.

141. Park, B. J.; Song, K. H.; Choi, H. J. Magnetic Carbonyl Iron Nanoparticle Based Magnetorheological Suspension and Its Characteristics Materials Letters 2009, 63 (15), 1350-1352.

142. Qi, B.; Ye, L.; Stone, R.; Dennis, C.; Crawford, T. M.; Mefford, O. T. Influence of Ligand–Precursor Molar Ratio on the Size Evolution of Modifiable Iron Oxide Nanoparticles J. Phys. Chem. C 2013, 117 (10), 5429-5435.

143. Velásquez, E. A.; Mazo-Zuluaga, J.; Mejía-López, J. Size Dependence Study of the Ordering Temperature in the Fast Monte Carlo Method J Nanopart Res 2013, 15 (2), 1- 12.

144. Frenkel, D.; Ladd, A. J. C. New Monte Carlo Method to Compute the Free Energy of Arbitrary Solids. Application to the Fcc and Hcp Phases of Hard Spheres The Journal of Chemical Physics 1984, 81 (7), 3188-3193.

145. Liu, S. L.; Zhou, J. P.; Zhang, L. N. Effects of Crystalline Phase and Particle Size on the Properties of Plate-Like Fe2O3 Nanoparticles During Gamma- to Alpha-Phase Transformation Journal of Physical Chemistry C 2011, 115 (9), 3602-3611.

146. Fleurier, R.; Bhattacharyya, S.; Saboungi, M. L.; Raimboux, N.; Simon, P.; Kliava, J.; Magrez, A.; Feher, T.; Forro, L.; Salvetat, J. P. Increase in the Curie Temperature and Magnetic Anisotropy in Fepd/Pt-Iron Oxide Core-Shell Nanoparticles Journal of Applied Physics 2009, 106 (7).

134 147. Gemmer, J.; Breuer, H. P. Projection Operator Techniques and Hilbert Space Averaging in the Quantum Theory of Nonequilibrium Systems Eur. Phys. J.-Spec. Top. 2007, 151, 1-12.

148. Shvets, G.; Urzhumov, Y. A. Electric and Magnetic Properties of Sub-Wavelength Plasmonic Crystals Journal of Optics a-Pure and Applied Optics 2005, 7 (2), S23-S31.

149. Monshi, A.; Foroughi, M. R.; Monshi, M. R. Modified Scherrer Equation to Estimate More Accurately Nano-Crystallite Size Using Xrd World J. Nano Sci. Eng. 2012, 2 (2), 154-160.

150. Scherrer, P. Bestimmung Der Größe Und Der Inneren Struktur Von Kolloidteilchen Mittels Röntgenstrahlen Nachrichten von der Gesellschaft der Wissenschaften zu Göttingen, Mathematisch-Physikalische Klasse 1918, 2 (2), 98-100.

151. Zhang, S. R.; Wu, K. C.; Sherry, A. D. A Novel Ph-Sensitive Mri Contrast Agent Angew. Chem.-Int. Edit. 1999, 38 (21), 3192-3194.

152. Watanabe, A.; Wada, Y.; Obata, T.; Ueda, T.; Tamura, M.; Ikehira, H.; Moriya, H. Delayed Gadolinium-Enhanced Mr to Determine Glycosaminoglycan Concentration in Reparative Cartilage after Autologous Chondrocyte Implantation: Preliminary Results Radiology 2006, 239 (1), 201-208.

153. Yuan, Q.; Venkatasubramanian, R.; Hein, S.; Misra, R. D. K. A Stimulus- Responsive Magnetic Nanoparticle Drug Carrier: Magnetite Encapsulated by Chitosan- Grafted-Copolymer Acta Biomaterialia 2008, 4 (4), 1024-1037. 154. Purushotham, S.; Chang, P. E. J.; Rumpel, H.; Kee, I. H. C.; Ng, R. T. H.; Chow, P. K. H.; Tan, C. K.; Ramanujan, R. V. Thermoresponsive Core–Shell Magnetic Nanoparticles for Combined Modalities of Cancer Therapy Nanotechnology 2009, 20 (30), 305101.

155. Jayasundar, R.; Honess, D.; Hall, L. D.; Bleehen, N. M. Simultaneous Evaluation of the Effects of Rf Hyperthermia on the Intra- and Extracellular Tumor Ph Magnetic Resonance in Medicine 2000, 43 (1), 1-8.

135 156. Hergt, R.; Andrä, W., Magnetic Hyperthermia and Thermoablation. In Magnetism in Medicine, Wiley-VCH Verlag GmbH & Co. KGaA: 2007; pp 550-570.

157. Barick, K. C.; Singh, S.; Jadhav, N. V.; Bahadur, D.; Pandey, B. N.; Hassan, P. A. Ph-Responsive Peptide Mimic Shell Cross-Linked Magnetic Nanocarriers for Combination Therapy Advanced Functional Materials 2012, 22 (23), 4975-4984.

158. Jon, W.; Matthew, D.; Bryce, M. W.; William, C.; Alex, S.; Christopher, G. R.; Marissa Nichole, R. 3d Viability Imaging of Tumor Phantoms Treated with Single- Walled Carbon Nanohorns and Photothermal Therapy Nanotechnology 2013, 24 (27), 275102.

159. Hergt, R.; Dutz, S.; Müller, R.; Zeisberger, M. Magnetic Particle Hyperthermia: Nanoparticle Magnetism and Materials Development for Cancer Therapy Journal of Physics: Condensed Matter 2006, 18 (38), S2919.

160. Shlyakhtin, O. A.; Leontiev, V. G.; Young-Jei, O.; Kuznetsov, A. A. New Manganite-Based Mediators for Self-Controlled Magnetic Heating Smart Materials and Structures 2007, 16 (5), N35.

161. Hung Manh, D.; Hong Nam, P.; Van Chien, N.; Thi Bich Hoa, P.; Dai Lam, T.; Anh Tuan, N.; Quoc Thong, P.; Van Hong, L.; Xuan Phuc, N. Magnetic Heating Characteristics of La 0.7 Sr X Ca 0.3- X Mno 3 Nanoparticles Fabricated by a High Energy Mechanical Milling Method Advances in Natural Sciences: Nanoscience and Nanotechnology 2011, 2 (3), 035003.

162. Behdadfar, B.; Kermanpur, A.; Sadeghi-Aliabadi, H.; Morales, M. D.; Mozaffari, M. Synthesis of Aqueous Ferrofluids of Znxfe3-Xo4 Nanoparticles by Citric Acid Assisted Hydrothermal-Reduction Route for Magnetic Hyperthermia Applications Journal of Magnetism and Magnetic Materials 2012, 324 (14), 2211-2217.

163. Levy, M.; Quarta, A.; Espinosa, A.; Figuerola, A.; Wilhelm, C.; Garcia-Hernandez, M.; Genovese, A.; Falqui, A.; Alloyeau, D.; Buonsanti, R., et al. Correlating Magneto- Structural Properties to Hyperthermia Performance of Highly Monodisperse Iron Oxide Nanoparticles Prepared by a Seeded-Growth Route Chem. Mat. 2011, 23 (18), 4170- 4180.

136 164. Wang, M. A.; Peng, M. L.; Cheng, W.; Cui, Y. L.; Chen, C. A Novel Approach for Transferring Oleic Acid Capped Iron Oxide Nanoparticles to Water Phase Journal of Nanoscience and Nanotechnology 2011, 11 (4), 3688-3691.

165. Chen, Z.; Chen, H.; Hu, H.; Yu, M.; Li, F.; Zhang, Q.; Zhou, Z.; Yi, T.; Huang, C. Versatile Synthesis Strategy for Carboxylic Acid−Functionalized Upconverting Nanophosphors as Biological Labels Journal of the American Chemical Society 2008, 130 (10), 3023-3029.

166. Saville, S. L.; Woodward, R. C.; House, M. J.; Tokarev, A.; Hammers, J.; Qi, B.; Shaw, J.; Saunders, M.; Varsani, R. R.; St Pierre, T. G., et al. The Effect of Magnetically Induced Linear Aggregates on Proton Transverse Relaxation Rates of Aqueous Suspensions of Polymer Coated Magnetic Nanoparticles Nanoscale 2013, 5 (5), 2152-63.

137