COLLOIDAL INORGANIC NANOCRYSTALS: NUCLEATION, GROWTH AND BIOLOGICAL APPLICATIONS

Transcription

1 COLLOIDAL INORGANIC NANOCRYSTALS: NUCLEATION, GROWTH AND BIOLOGICAL APPLICATIONS By JARED JAMES LYNCH A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA

2 2012 Jared James Lynch 2

3 To my older brother, for all his inspiration To my mother, with love 3

4 ACKNOWLEDGMENTS I would like to express my sincere gratitude to my research adviser, Dr. Y. Charles Cao, who has supported me throughout my graduate career at the University of Florida. If it were not for him, I would not be the chemist I am today. He has truly shaped me, not only as a scientist, but also as a person. I am a stronger, more driven individual due to his tutelage and guidance. I am also thankful to the rest of my Ph.D. committee members, Dr. Charles Martin, Dr. Weihong Tan, Dr. Valeria Kleiman, and Dr. Franky So. Your advice and willingness to talk helped me through many challenging times and my appreciation can never be fully shown. To Dr. Kerry Siebein at the Major Analytical Instrumentation Center (MAIC) a special thanks for all of the assistance measuring TEM samples and troubleshooting the instrument. You always had a kind word for me when I was working hard. I would also like to recognize Dr. Oleg Mateev for training me on the ICP-AES instrument. I would like to acknowledge Dr. Frank Zhao for collaborating with me on my experiments involving living cells. Your help was crucial to successfully completing the project. I would like to thank my fellow group mates for all of their assistance and valuable discussions. Special thanks goes to Dr. Jiaqi Zhuang for teaching me the basic skills needed to move forward in all of my research endeavors, as well as Marcus Tirado, Dr. Huimeng Wu, Yasutaka Nagaoka, Dr. Tie Wang, and Dr. Zhongliang Wang. I would like to acknowledge all of my friends in Gainesville. If it were not for you all, I might not have been able to complete my studies here at UF. You have kept my mind set positive and hopeful even through the toughest of times. Finally, I would like to recognize my family, who have supported and struggled with me during the past six years. Your love and tolerance have not been forgotten. 4

5 TABLE OF CONTENTS ACKNOWLEDGMENTS...4 LIST OF TABLES...9 LIST OF FIGURES...10 ABSTRACT...13 CHAPTER 1 INTRODUCTION...15 page 1.1 Nanocrystal Formation Synthetic Schemes Hot Injection Method Non-Injection Synthesis Nanocrystal Properties Optical and Electrical Properties Magnetic Properties Thermal Properties Catalytic Properties Surface Functionalization Nanocrystal Materials Discussed in this Work Iron Oxide Nanocrystals CdS/ZnS Core/Shell Nanocrystals Summary of Present Research GAS-BUBBLE EFFECTS ON THE FORMATION OF COLLOIDAL IRON OXIDE NANOCRYSTALS Prologue Experimental Section Chemicals Synthesis of Iron Oxide Nanocrystals Synthesis of iron-oleate precursor Synthesis in boiling solvents Synthesis in non-boiling solvents The synthesis of iron-oxide nanocrystals under reduced pressures Synthesis of iron-oxide nanocrystals using argon bubbling Synthesis of iron-oxide nanocrystals under controlled Ar bubbling The Determination of Iron-Oxide Crystallization Yield Kinetic Study of Iron-Oxide Syntheses Under boiling and non-boiling conditions Under controlled Ar bubbling

6 Calculation of average nanocrystal diameter and distribution Instrumentation TEM measurements TGA measurements Results and Discussion Iron-Oxide Nanocrystal Synthesis System The Effects of Solvent Gas Bubbles Generated From Boiling Solvents Iron Oxide Crystallization Yield The Effects of Solvent-Gas Bubbles Generated from Boiling Solvents with Reduced Pressure The Effects of Ar Bubbles Kinetic Studies on Gas-Bubble Effects General Discussion Nanocrystal Formation Semi-quantitative Analysis Comparisons with Classical Nucleation Theory Summary THE DOPING OF CDS/ZNS CORE-SHELL NANOCRYSTALS WITH COPPER AND SILVER Prologue Experimental Section Chemicals Three Step Synthesis of Cu- or Ag-Doped CdS/ZnS Core/Shell Nanocrystals Preparation of precursors Synthesis of Cu- or Ag-doped CdS/ZnS core/shell nanocrystals Characterization of Cu- or Ag-Doped CdS/ZnS Core/Shell Nanocrystals Absorption measurements Photoluminescence measurements TEM measurements X-ray powder diffraction (XRD) measurements Inductively-coupled plasma (ICP) atomic emission spectroscopy measurements Results and Discussion Cu-Doping of Thick Shelled (6 ML) CdS/ZnS Core/Shell Nanocrystals Cu-Doping of 1.6 ML CdS/ZnS Core/Shell Nanocrystals Jander Analysis of Cu-doped CdS/ZnS Core/Shell Nanocrystals CdS/ZnS core/shell with 4.1 nm CdS core CdS/ZnS nanocrystals with a 3.4 and 4.9 nm CdS core Non linearity of Jander plots Physical Processes Associated with Alloying of Cu-Doped CdS/ZnS Nanocrystals at Temperatures between 200 to 220 C Determination of Alloying Activation Energy Ag-Doping of CdS/ZnS Core/Shell Nanocrystals Ag doping of 1.6 ML CdS/ZnS core/shell nanocrystals Mechanism of Ag doping in CdS/ZnS core/shell nanocrystals

7 3.4 Summary SURFACE FUNCTIONALIZATION OF METAL OXIDE NANOCRYSTALS BY INTRODUCTION OF ELECTROSTATIC CHARGE VIA INORGANIC SALT IONS Prologue Experimental Section Chemicals Synthesis of Metal Oxide Nanocrystals Iron oxide nanocrystals Manganese oxide nanocrystals Indium oxide nanocrystals Zinc oxide nanocrystals Ligand Exchange of Metal Oxide Nanocrystals using AsO 2 -, HPO 4 2-, and OH - ions Characterization of Water Soluble Metal Oxide Nanocrystals Absorption measurements TEM measurements Energy dispersive spectroscopy (EDS) measurements Zeta (ζ) potential measurements Inductively-coupled plasma (ICP) atomic emission spectroscopy (AES) measurements Cellular Studies Using AsO 2 - Coated Iron Oxide Composite Nanocrystals (AICN) Materials for cellular studies Cell morphology study Cell viability study Hoechst staining assay DNA ladder assay Results and Discussion Ligand Exchange using AsO 2 -, HPO 4 2-, and OH - ions Arsenite Coated Iron Oxide Composite Nanocrystals (AICN) as Cancer Therapeutics Cell morphology study Cell viability studies Fluorescent DNA staining experiments Gel electrophoresis assay Summary CONCLUSIONS Summary of Present Research Perspectives APPENDIX: TABLES SUMMARIZING JANDER ANALYSIS AND ICP-AES DATA LIST OF REFERENCES

8 BIOGRAPHICAL SKETCH

9 LIST OF TABLES Table page 2-1 Summary of different solvent compositions and reaction temperatures used for the boiling and non-boiling reactions during the synthesis of iron oxide nanocrystals Numerical calculation results from average reaction yield determinations A-1 Jander analysis data for 27 Cu atoms per 1.4 ML CdS/ZnS core/shell NCs A-2 Jander analysis data for 48 Cu atoms per 1.6 ML CdS/ZnS core/shell NCs A-3 Jander analysis data for 60 Cu atoms per 1.9 ML CdS/ZnS core/shell NCs A-4 ICP-AES data for 5.2 nm Fe 3 O 4 -AsO 2 NCs in water

10 LIST OF FIGURES Figure page 1-1 LaMer diagram Illustration of the electronic structure of metal and semiconductor materials in their bulk and nanocrystal forms Schematic depicting how the idealized density of electronic states for one band of a semiconductor changes with the dimensionality of the material Size-dependent optical properties of CdSe nanocrystals in solution Absorption spectra calculated using Mie theory for spherical gold NCs with diameters varying from 5 nm to 100 nm Size effect of cobalt NCs on the magnetic coercivity (H c ) Graph of m s 1/3 vs 1/r for Fe 3 O 4 NCs with sizes ranging from 4 to 12 nm Unit cell of magnetite Unit cell of maghemite Unit cell of wüstite (Fe 1-x O) with x equal to Unit cell of zinc-blende cadmium sulfide Unit cell of cubic ZnS Representations of different core/shell nanocrystals showing their conduction and valence band offsets relative to both the core and shell TEM images with histograms of iron oxide nanocrystals from syntheses in boiling solvents and non-boiling solvents TGA measurements for iron oxide crystallization yield syntheses in boiling solvent and non-boiling solvent Reaction setup for reduced pressure experiments TEM images and histograms of iron oxide nanocrystals synthesized in boiling solvents under reduced pressures and in non-boiling solvents under 1 atm Reaction setup for Ar bubbling experiments TEM images and histograms of iron-oxide nanocrystals made from Ar bubbling experiments

11 2-7 Kinetics study of the syntheses of iron-oxide nanocrystals in boiling and non-boiling solvents Kinetics study of the synthesis of iron oxide nanocrystals under controlled Ar bubbling TEM images of iron oxide nanocrystals grown in pure ODE at 300 C for 60 minutes with 12 minutes or 18 min of Ar bubbling Reaction scheme illustrating the formation of thermal free radicals during the endothermic decomposition of iron oleate and the exothermic formation of primary iron-oxide clusters Reaction coordinate diagram describing the formation of iron-oxide nanocrystal nuclei from active monomers Scheme of three-step synthesis for synthesizing Cu-doped CdS/ZnS core/shell nanocrystals Absorption spectra of CdS/ZnS core/shell nanocrystals with 6 ML ZnS after addition of 48 Cu atoms per NC with increasing reaction time at 220 C Absorption spectra of CdS/ZnS core/shell nanocrystals with 1.6 ML ZnS after addition of 48 Cu atoms per NC with increasing reaction time at 220 C Absorption spectra of CdS/ZnS core/shell nanocrystals with 1.6 ML ZnS without addition of 48 Cu atoms per NC with increasing reaction time at 220 C Change in absorption spectra vs. time at different reaction temperatures for 48 Cu atoms per 1.6 ML CdS/ZnS NCs; squares Plot of percent solid-solution formation vs. measured change in absorption spectra at 220 C for 48 Cu atoms per 1.6 ML CdS/ZnS core/shell NC Graphs of Jander analysis of degree of solid-solution formation versus reaction time after Cu addition for 48 Cu atoms per 1.6 ML CdS/ZnS core/shell NCs at different alloying temperatures Graphs of Jander analysis of degree of solid-solution formation versus reaction time after Cu addition for 27 Cu atoms per 1.4 ML CdS/ZnS core/shell NCs at different alloying temperatures Graphs of Jander analysis of degree of solid-solution formation versus reaction time after Cu addition for 60 Cu atoms per 2.0 ML CdS/ZnS core/shell NCs at different alloying temperatures Absorption spectra of 3.4 nm diameter CdS core with 27 Cu atoms per core added in dropwise

12 3-11 Arrhenius plots of ln k versus 1000/T Absorption spectra of aliquots taken with time for the 48 Ag atoms per 1.6 ML CdS/ZnS core//shell NCs Scheme illustrating Ag doping of CdS/ZnS core/shell nanocrystals and its effects on the optical properties of the particles in solution Schematic illustrating use of nitrosonium tetrafluorborate salt to transfer colloidal inorganic nanocrystals into polar solvents Picture showing phase transfer of CdSe nanocrystals from toluene to formamide using potassium sulfide Scheme of metal oxide nanocrystal ligand exchange between oleate and arsenite Pictures of nanocrystal solutions before and after ligand exchange with AsO TEM images of iron oxide and indium oxide nanocrystals in water after ligand exchange with AsO 2 - ions EDS measurements of Fe 3 O 4 -AsO 2 and In 2 O 3 -AsO 2 NCs after ligand exchange into water TEM images of two-dimensional superlattices of 7.2 nm diameter iron oxide NCs Optical microscope images of Huh7 cells after 48 hours of incubation with AICN and pure NaAsO 2 both at 100 μm arsenite Summaries of cell viability tests comparing pure NaAsO 2 with AICN at inducing necrosis to Huh7 cells Fluorescent microscope images of Huh7 cells treated with 50 μm pure NaAsO 2 and AICN Gel electrophoresis assay showing DNA fragmentation patterns for healthy liver cells treated with different concentrations of AICN A-1 ICP-AES calibration curves for arsenic and iron used to calculate the concentration of arsenite on AICN A-2 Absorbance spectra of ZnO nanocrystals before and after exchange with arsenite ligand

13 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy COLLOIDAL INORGANIC NANOCRYSTALS: NUCLEATION, GROWTH AND BIOLOGICAL APPLICATIONS Chair: Y. Charles Cao Major: Chemistry By Jared James Lynch August 2012 Colloidal inorganic nanocrystals are a class of material whose size ranges from a few nanometers to a hundred nanometers in dimension. These nanocrystals have size dependent properties that differ significantly from the bulk material counterparts. Due to their unique physical properties colloidal inorganic nanocrystals have several promising applications in a diverse range of areas, such as biomedical diagnosis, catalysis, plasmonics, high-density data storage and solar energy conversion. This dissertation presents the study of the formation of iron oxide nanocrystals under the influence of solvent and Ar gas bubbles, the phase transfer of metal oxide nanocrystals into water using inorganic ions, and the doping of semiconductor CdS/ZnS core/shell nanocrystals with copper and silver ions. First, the formation of iron oxide nanocrystals is investigated in the presence of boiling solvent or Ar bubbles. Using a non-injection based synthesis method, the thermal decomposition of iron oleate was studied under various reaction conditions, and the role of the bubbles on the nucleation and growth of iron oxide nanocrystals was determined. Kinetics studies were used to elucidate how latent heat transfer from the bubbles allows for active monomers to form 13

14 preferentially from exothermic reactions taking place during nucleation. General insights into colloidal inorganic nanocrystal formation are discussed. Second, a non-injection based synthesis for CdS/ZnS core/shell nanocrystals is used to make high quality semiconductor particles which are intentionally doped with Cu or Ag ions. The Ag ions effect on the optical properties of the CdS/ZnS nanocrystals is investigated. The absorption and fluorescence of the samples is measured as a function of time and temperature. Proposed mechanisms for the observations are given and thoroughly discussed. Comparisons between previous results for Cu doped CdS/ZnS nanocrystals are also made to further understand how doping of semiconductor nanocrystals can be realized. Finally, a novel phase transfer process is demonstrated using inorganic salts, such as sodium arsenite, to make water soluble metal oxide nanocrystals. The water soluble iron oxide nanocrystals are fully characterized by several complementary techniques and then used in cellular studies. The arsenite-coated iron oxide composite nanocrystals (AICN) are shown to be effective cancer therapy agents. 14

15 CHAPTER 1 INTRODUCTION At present, the study of nanoscience and/or nanotechnology is being actively pursued across many different fields of research. One major area of focus is on the synthesis, characterization, and application of colloidal inorganic nanocrystals (NCs) that vary in type from metal oxides, to semiconductors, to pure metals. Nanocrystals can be simply defined as crystalline materials that have sizes between nm in dimension. These extremely small particles typically have hundreds to thousands of individual atoms that make up there crystalline domain. At these size regimes, the material properties of the nanocrystals differ significantly from their bulk counterparts. 1,2 Nanocrystals have the same crystalline structure as their bulk materials, but have unique optical, magnetic, catalytic, biological, and mechanical properties that are considered size dependent properties inherent to all matter. 2 Many examples already exist that demonstrate these size dependent properties of inorganic nanocrystals. Magnetic NCs have magnetization transition-temperatures, coercivities (H c ), and saturation magnetization ratios (M r /M s ) that are directly related to the size of the NCs. 3-5 Semiconductor nanocrystals show size-dependent absorption and emission spectra that result from quantum confinement effects. 2,6 Noble metal NCs vary in color depending on their size due to surface plasmons. 7 Several different types of metal NCs show significant enhancement in their catalytic properties with sizes of a few nanometers The band gap of CdS NCs can be tuned from 2.5 to 4.0 ev, the radiative rate for the lowest allowed optical excitation ranges from nanoseconds to tens of picoseconds, the melting temperature varies from 1600 to 400 C, and the pressure to induce the transformation from a four to a six coordinate phase increases from 2 to 9 GPa as the NC size decreases. 1 These unique, size dependent properties are being researched to further understand their fundamental 15

16 physical origins and to manipulate them for application in areas such as biological diagnostics, solar energy conversion, LED development, catalysis, 25 and high-density data storage. 26,27 In most cases, colloidal inorganic nanocrystals are coated with an organic or inorganic molecule to stabilize the particles in solution and prevent aggregation from occurring. This organic molecule, or ligand, creates challenges for the application of nanocrystals in areas related to drug delivery and electronic device construction. By exchanging the type of ligand on the nanocrystals surface, the NCs properties can be altered and thereby a new avenue for manipulating the NCs properties is presented. Several areas of nanocrystals research are not yet fully understood, and until progress is made in understanding the physical mechanisms by which NCs are formed, the true promise of these novel materials will not be possible. First, a deeper understanding of how colloidal inorganic nanocrystals form in solution is needed so that tailor-made, larger-scale reactions can be realized. Classical nucleation models are insufficient at explaining the observed kinetics of NC formation. 28,29 The two synthetic methods most widely used in synthesizing NCs are based on a rapid precursor injection 30,31 or a non-injection based synthesis (NIS) Both methods have theoretical frameworks that attempt to explain the NC formation mechanism, but both suffer from a lack of empirical data to support their proposed predictions. Methodologies that can give greater insight into the types of reactions that occur when forming nanocrystals and how the kinetics of those reactions affect the nucleation and growth of the resulting NCs are needed. Second, surface functionalization of inorganic NCs is quickly becoming a major technique to create hybrid systems that can be used for biomedical applications. 42,43 By creating novel surface ligands, the NCs can be made to bind with other ligands with complementary structure. 16

17 This allows for targeted delivery of NCs to areas where they can be most effective for their specific purpose. 44 By modifying NCs surfaces, the solubility and stability can be changed. 45 Furthermore, NCs chemical and physical properties can be altered by changing the ligands on their surface. 46 The following introduction gives an overview of the theoretical background explaining NC formation in solution and describes the two major synthetic schemes used to create inorganic nanocrystals. The size-dependent physical properties of nanocrystals are also discussed. A special emphasis will be given to iron oxide NCs as they are central to the second and third chapters outlined in this work. 1.1 Nanocrystal Formation Colloidal inorganic nanocrystals (CINs) are a class of materials that are characterized by their size dimensions being in the nanometer (10-9 m) range. The organic phase synthesis of CINs takes place in three distinct stages: prenucleation, nucleation, and growth. By controlling theses stages it is possible to tune the size, shape, and size distribution of the resulting inorganic nanocrystals. 31,47,48 The most well established theoretical framework for explaining how nucleation and growth affect nanocrystal formation is based on classical nucleation theory and the LaMer diagram 49 as shown in Figure 1-1. Classical nucleation theory is based on the liquid drop model and the Gibbs-Thompson equation dictating that nucleation occurs when supersaturation of an active monomer is reached in solution. 28,29 In this sense, active monomer is defined as the basic chemical subunit that directly forms nanocrystal nuclei. The LaMer diagram explains the formation of particles out of supersaturated solutions as taking place in three steps as shown in Figure 1.1. The diagram illustrates nucleation beginning when the concentration of the active monomer (AM) reaches a threshold or critical supersaturation point and finishing when the concentration deceases below this threshold

18 Concentration C * max Critical limiting supersaturation C * min Growth by diffusion or reaction C s Solubility I II III Rapid self nucleation Time Figure 1-1. LaMer diagram. In phase I, no nucleation or growth occurs; in phase II, rapid nucleation occurs and this is followed by particle growth in phase III. The monomer concentration is constantly increasing with time as shown in Figure 1-1, stage I. At this point, nucleation has not occurred even under supersaturated conditions (C > Cs), due to the high energy barrier that exists for spontaneous homogeneous nucleation. At stage II, the degree of supersaturation has increased to the point where the energy barrier, C > C* min, is overcome and nucleation takes place. Stage II results in the formation and buildup of stable nuclei. Afterwards, the nuclei begin to grow and consume monomers from the surrounding reaction media. When the rate of monomer consumption due to both the nucleation and growth reactions equals the rate of monomer formation, the monomer concentration reaches its highest point (C = C* max ). Thereafter, the concentration of monomer starts to decrease until it drops below C* min, at which point the overall nucleation rate is negligible and the second stage of the LaMer diagram ends. It is essential that the duration of stage II be short in order to effectively separate the nucleation from the growth of CINs. When C < C* min, nanocrystal formation enters the third stage in the LaMer diagram, also known as the growth stage, in which no further 18

19 nucleation occurs and the existing nanocrystals grow as long as the solution is still supersaturated. The growth rate is controlled by both diffusion and reaction kinetics. Finally, the nanocrystals are passivated with organic ligands which prevent them from aggregating and fusing together. The organic ligands also provide the nanocrystals with their solubility in either polar or nonpolar solvents depending on the functionality of the ligands. By effectively separating the nucleation and growth stages of particle formation the final product has a narrow size distribution which is crucial for studying and applying CIN s properties. LaMer s diagram gives a simplistic view of nanocrystal formation, but in reality the reaction system is more elaborate and in need of further study. The growth of NCs can be described by monomer diffusion towards the surface of a nanocrystal, followed by dissociation of the surface ligands from the nanocrystal s surface, and then subsequent reaction between the diffusing moieties onto the growing NCs. 37,38,40 Therefore, any parameters which alter the diffusion rate or interaction between the ligands and the NC surfaces (e.g. solvents, ligands, reaction temperature, pressure, reaction time, and concentration of precursors) can play a significant role in determining the final nanocrystal size and shape. While classical nucleation theory and the LaMer diagram are useful in describing the formation of CINs, they have limitations in predicting nanocrystal nucleation kinetics in solution. 28,39,50,51 Two major inadequacies are the lack of ability to measure the concentrations of the AM in solution and a general lack of knowledge as to the chemical structure of these active monomers. Neither the reactions forming AMs nor the reactions between AMs that lead to nuclei are fully understood. In general, it is hypothesized that these reactions involve multiple steps that could be reversible, parallel, or consecutive in nature. Theoretical analysis has been performed to better explain the formation of CINs and is generally modeled on reaction kinetics 19

20 with simplified assumptions. 39,51 Once a theoretical framework is established, it is compared with experimental evidence to support its validity. Many times the modeling is based on rate equations describing the conversion of precursor to monomer, the reversible growth of monomer units into larger clusters, the nucleation of nanocrystals from monomers, and then the depletion of monomer via growing attachment to the nanocrystals. 37,40,52,53 For simplicity it is usually assumed that there is only one rate-limiting step and that this step is nonreversible for the formation of AMs from precursor. 28,29,39,41 It is assumed that the depletion of precursor correlates with an increase in monomer concentration although it is known that several types of reactions occur after the precursor has decomposed from its primary structure. Yet, with every new attempt made at clarifying the reaction kinetics of nanocrystal formation the simplifying assumptions have little experimental evidence to support them. Hence, more ongoing research into the physical processes dictating CINs nucleation and growth is needed. A nanocrystal s mean size and distribution can also be affected by the difference in the chemical potential at the boundary between the surface of a growing particle and the growth solution. The chemical potential of a nanocrystal is inversely proportional to the nanocrystal s size. Therefore, the concentration of monomer at equilibrium with a small nanocrystal is significantly higher than from a larger particle. This leads to the formation of concentration gradients between growing nanocrystals with mass transport going from smaller particles to larger particles. This process is called Ostwald ripening or the coarsening effect, and so at equilibrium, there is a balance between dissolution and growth of CINs. The sizes and shapes of CINs can be guided by changing the number of nuclei and/or the chemical potentials of the components of the reaction system. Tuning the chemical potentials is typically achieved by varying the reaction parameters, such as the chemical structure or 20

21 concentrations of precursors and ligands, the reaction temperature, the reaction time, or the heating rates. Normally, the surface energy of spherical NCs is lower than that of non-spherical particles (e.g. cubes, tetrahedral pyramids, and triangular prisms) and this is governed by the fact that the surface chemical potentials on the individual facets are different. 58 Thus, for shape control, the surface chemical potential must be considered. To date, high quality CINs with cubic, rodlike, pyramidal, and numerous other shapes have been synthesized. 47, Synthetic Schemes Currently, there are two major reaction pathways for synthesizing CINs in the organic phase. The most widely used method for CIN synthesis is the hot injection method (HIM) that relies on a rapid injection of nanocrystal precursors into a hot growth solution that typically has both a high boiling point solvent and organic surfactants in it. The other method used in synthesizing CINs is based on a non-injection synthesis (NIS) where a precursor solution is heated continually in the presence of stabilizing ligands to the reaction temperature Hot Injection Method The first high quality CINs were formed using the hot-injection based method. Murray and Bawendi first demonstrated this strategy in 1993 and since then numerous other examples of nanocrystal syntheses based on the HIM have been developed. 30 In the hot injection method, reaction precursors are quickly injected into a hot solvent under active stirring. After the injection, nucleation occurs almost instantly, and is followed by a growth period that varies in duration depending on the specific material and synthesis being performed. After nucleation, the concentration of AM in solution drops below the critical supersaturation threshold, and further AM is grown onto the nuclei in a reversible fashion. The growth rate of these reactions is controlled by the rate of diffusion of monomers to the nanocrystals and/or by the kinetics of the governing growth reactions. Since nucleation time is established by both the rates of injection 21

22 and monomer diffusion, injection-based syntheses have poor reproducibility and are unsuitable for large scale production, making these methods unsuitable for industrial use Non-Injection Synthesis With limitations on the reproducibility and the scaling up of injection-based methods being difficult to bypass, efforts to create a non-injection-based synthetic method for nanocrystals were pursued. Recently, Cao s group reported on the synthesis of CdSe and CdTe nanocrystals using a non-injection based method. By tuning the reactivity of the cadmium precursor, a separation of the nucleation and growth was realized. 34 In another example, nucleation initiators were used to control the thermodynamics and reaction kinetics in the nucleation of CdS nanocrystals yielding high quality semiconductor particles 35 with optical properties similar in quality to the best particles made using injection based methods. During non-injection based syntheses, the reaction precursors, ligands, and solvents are mixed at room temperature and then heated at a predetermined rate to the desired reaction temperature. By changing the reaction temperature and time, the nanocrystal size and size distribution can be controlled. Park et al. clearly demonstrated this control by synthesizing iron oxide nanocrystals between 9 and 22 nm diameters using iron oleate as the precursor, oleic acid as the stabilizing ligand, and varying the solvent to control the reaction temperature from 274 to 365 C. 36 To date, several different types of materials have been synthesized using the non-injection based scheme, including metal oxides, 36,60,65 noble metals, 66,67 and semiconductors Nanocrystal Properties The current promise of nanocrystals in fields ranging from materials science to medical diagnostics is premised on using their size-dependent properties in applications that would not be possible with the same bulk material. By shrinking matter down to nanometer dimensions, two specific phenomena begin to play important roles in determining the properties of the material. 22

23 1) the surface to volume ratio becomes larger, and 2) quantum confinement of the atomic wavefunctions begins to occur. These effects alter the chemical and physical properties of the material including their optical, electrical, magnetic, catalytic, and mechanical characteristics Optical and Electrical Properties The electronic structure of metal and semiconductor materials changes as their size decreases. 1 Figure 1-2 illustrates this phenomenon where discrete energy levels form near the band edges of the material. 1 The Fermi level for metals is in the center of the band and the energy levels that form near the band edges do not vary significantly with the size of the nanocrystals. Only for extremely small crystallites (tens to hundreds of atoms) at exceedingly low temperatures (near zero Kelvin) will the thermal energy, kt, be smaller than the spacing between the energy levels allowing for their direct measurement. In contrast, a semiconductor s Fermi level is centered in between two bands, also known as the band gap. As the size of the semiconductor nanocrystal increases, the energy difference between the bands decreases, i.e. the band gap gets smaller. The discreteness at the band edges results in changes in transition energies as one or more dimensions of the crystals decrease to the nano-scale. The density of states due to quantum confinement varies with the degree of spatial dimensions that reach into the nano-scale for semiconductors. Figure 1-3 shows how the density of electronic states changes when the dimensionality of the semiconductor is reduced from 3d to the 0d, or the molecular limit. 1 For this idealized situation, the density of electronic states can be predicted by a simple particle in a box model derived from quantum mechanics. For the 3d case, the semiconductor is in an unconfined state typical of bulk materials with nearly continuous energy levels. In the highly confined state of the 0d quantum dot the energy levels have become quantized. These quantized states and the 23

24 Figure 1-2. Illustration of the electronic structure of A) metal and B) semiconductor materials in their bulk and nanocrystal forms. The density of states near the band edges becomes discrete, showing specific energy levels dependent on the type of material. The HOMO-LUMO band gap is inversely proportional to the semiconductor nanocrystal size. Reprinted with permission from Alivisatos, A. P. J. Phys. Chem.1996, 100,

25 transitions between them are dependent on the type of semiconductor and the size of the nanocrystal. It is these transitions that give rise to the size-dependent optical and electrical properties in semiconductor NCs. Figure 1-3. Schematic depicting how the idealized density of electronic states for one band of a semiconductor changes with the dimensionality of the material. The energy levels are continuous for the 3d case while for the 0d or molecular limit case the energy levels have become discrete. Reprinted with permission from Alivisatos, A. P. J. Phys. Chem.1996, 100, The electronic properties of semiconductor nanocrystals have been described and modeled using molecular orbital theory along with the effective mass approximation (EMA). 69,70 EMA contends that the electronic density of states depends on the radius (R) of the NC as compared to the Bohr radius, a b, given by (1-1) where μ* is the exciton reduced mass, ћ is Planck s constant divided by 2π and κ is the dielectric constant of the nanocrystal. An exciton is an electron-hole pair that is bound together via electrostatic interactions. On one hand, when R is much greater than a b, the amount of 25

26 confinement on the exciton is effectively zero, Coulombic forces dominate, and the exciton acts as a single particle. On the other hand, when R is less than a b, the electron-hole pair is decoupled and the electron and hole effective masses can be used in place of μ* to give ( ) (1-2) where m e * and m h * are the effective masses of the electron and hole, respectively. When R < a b, the exciton experiences a strong quantum confinement resulting in a widening band gap and discrete energy level formation at the edges of the valence and conduction bands. 1 The band gap shift relative to the bulk band gap of the semiconductor material can be calculated by 71 ( ) (1-3) where R is the radius of the semiconductor NC. The optical properties of semiconductor nanocrystals can be directly measured and visualized using absorption and fluorescence spectroscopes. As the size of a semiconductor NC increases, the first excitonic peak in the absorption spectra and the photoluminescence emission peak from the nanocrystal solution shift to longer wavelengths. CdSe nanocrystals offer a representative example of these optical properties, where the wavelength of the first excitonic peak shifts from 459 to 652 nm with a diameter increase from 2.3 to 5.5 nm. 72 This effect is clearly demonstrated in Figure 1-4. Noble metal nanocrystal solutions display colors that are not seen in the bulk materials. This absorption phenomenon is caused by the surface plasmon bands associated with the noble metal NCs. Surface plasmon bands originate from oscillations of surface conduction electrons on noble metal NCs, such as gold or silver, which occur when these materials are irradiated with electromagnetic fields of the proper wavelength. If the collective oscillation of the electrons has 26

27 A Under UV-irradiation B Figure 1-4. Size-dependent optical properties of CdSe nanocrystals in solution. A) Photo of CdSe nanocrystal solutions under UV-irradiation/excitation. B) As the size of the nanocrystals increases, the absorption (left) and fluorescence (right) spectra shift to longer wavelengths. Reprinted with permission from Yongan Andrew Yang, Huimeng Wu, Kathryn R. Williams, Y. Charles Cao. Angew. Chem. Int. Ed. 44, 2005, the same frequency as that of the incident photons, localized surface plasmon resonance (LSPR) occurs, resulting in unique absorption profiles in the visible or near-infrared regions. Figure 1-5 is an example of how, according to the Mie theory, the surface plasmon band for spherical gold NCs changes as the NC size is increased. 73 As the size of the spherical gold NCs increases from 5 to 100 nm, the surface plasmon peak shifts from ~520 to 570 nm and 27

28 becomes broader in peak width. 73 The position of the surface plasmon band peak depends not only on the type of noble metal but also on the density of electronic states on the particle surface, Figure 1-5. Absorption spectra calculated using Mie theory for spherical gold NCs with diameters varying from 5 nm to 100 nm. Reprinted with permission from Hu, M.; Chen, J. Y.; Li, Z. Y.; Au, L.; Hartland, G. V.; Li, X. D.; Marquez, M.; Xia, Y. N. Chem. Soc. Rev. 2006, 35, the particle shape, and the surrounding environment, e.g. temperature and solvent. 74 Gustay Mie was the first to propose a quantitative description of SPR by solving Maxwell s equations with appropriate boundary conditions for spherical particles. 75 In his model, Mie calculates the total extinction cross section, ext, as the sum of the absorption and scattering cross-sections of the metal NCs over all electric and magnetic multipole oscillations. 76 When the particle diameter is shorter than the wavelength of the incident light, the scattering part of the NCs and higher order extinction terms can be neglected, resulting in a simplified equation that can be approximated by the dipolar excitation model as seen in 77 ( ) (1-4) 28

29 where D is the diameter of the spherical NCs, is the wavelength of the incoming light, ε m is the dielectric constant of the surrounding medium (independent of frequency), and ε 1, ε 2 are the real and imaginary parts of the frequency-dependent dielectric constant of the NC material, respectively. It should be noted that the total extinction cross section, ext, reaches its maximum value when, and therefore, the observed SPR is strongly dependent on the diameter of the metal NC as well as the dielectric constant of the surrounding environment. Recently, successful syntheses of nearly monodispersed NCs with well-defined shapes (i.e. rods, cubes, triangular prisms, multipods, and polyhedra, 78 demonstrate that the shapes of NCs also affect the SPR. Where spherical gold NCs display only one SPR peak in their absorption spectra, nonspherical gold NCs display multiple absorption peaks in the visible and near-infrared regions due to localized surface plasmons Magnetic Properties Magnetic NCs, such as iron oxide, nickel, cobalt, and mixed metal oxides have unique magnetic properties that are size, shape, and composition dependent. Values for the coercivity (H c ) and the remanence to saturation magnetization ratio (M r /M s ) for magnetic NCs reach maximum values when the NCs reach a critical diameter where all magnetic spins align unidirectionally. 3,4,80 This size regime tends to be on the order of tens of nanometers with variation coming mostly from the crystal structure and the chemical composition of the NC. 81 For larger size NCs, H c and M r /M s decrease due to the formation of multiple domains which compose the crystalline structure as depicted in Figure 1-6A. 82 An illustrative example of this effect can be seen in Figure 1-6B for cobalt NCs, 83 where, as the diameter of the particles increases from 4 to 13 nm, the coercivity increases to a maximum of 1680 Oe at 8 nm in diameter before decreasing due to the formation of multiple magnetic domains inside the NC to a 29

30 value of 250 Oe. 82 The coercivity (H c ) change with volume (V) in the single domain regime can be modeled by [ ( ) ] (1-5) where m s is the saturation magnetization, K u is the anisotropic magnetization constant, k is Boltzmann s constant, and T is the temperature. 82 Figure 1-6. Size effect of cobalt NCs on the magnetic coercivity (H c ). A) H c as a function of NC diameter showing transition from single to multi magnetic domains. B) Change in measured coercivity of Co NCs vs. diameter. Reprinted with permission from Jun, Y. W.; Seo, J. W.; Cheon, A. Acc. Chem. Res. 2008, 41, 179. Another interesting size-dependent property of magnetic NCs is superparamagnetism where magnetic spin transitions between up and down states in sufficiently small NCs cause a net magnetization of zero to be measured. In this situation, the magnetic anisotropic energy barrier between spin up and spin down states is proportional to the product of the NC volume and the magnetic anisotropic constant K u. 82 In bulk materials, the energy barrier is much higher than the thermal energy, kt, while for magnetic NCs the energy barrier is greatly reduced. This reduction in the energy barrier for spin state change is adequate to allow the thermal energy of the NC to readily switch spin direction. 84 The transition temperature required for this switch to occur is called the blocking temperature, T b, and is calculated as 30

31 (1-6) These size dependent magnetic changes have been observed for γ-fe2o3 as well as Co NCs. 83,85 The NCs surface has also been shown to be a crucial aspect that determines the overall magnetic properties of the material. Due to defects on the particle surfaces, differences in the saturation magnetization from the bulk value can occur. Inherently, magnetic materials contain disordered layers near the surface caused by reduced spin-spin exchange coupling energies. 86,87 In macroscopic materials, this surface effect is negligible since the surface is only a small fraction of the overall magnetic material. In contrast, NCs are characterized by enormous surface to volume ratios that cause surface effects to dramatically contribute to the overall properties of the material. For this case, the saturation magnetization, m s, can be calculated by [ ( ) ] (1-7) where r is the radius of the NC, M S is the saturation magnetization of the bulk material, and d is the thickness of disordered surface layer. 87 This effect is clearly observed in the case of Fe 3 O 4 NCs with sizes ranging from 4 to 12 nm where a linear relationship between m 1/3 s as a function of 1/r is shown in Figure The shape of the NCs can also dictate the magnetic characteristics as reported by Song for cubic and spherical cobalt ferrite NCs. 60 The results state that saturation and remanant magnetization of CoFe 2 O 4 NCs are affected by the size, regardless of the shape (spherical or cubic). Furthermore, the NC shape strongly influences the coercivity of the nanocrystals due to the effect of surface anisotropy

32 Figure 1-7. Graph of m s 1/3 vs 1/r for Fe 3 O 4 NCs with sizes ranging from 4 to 12 nm as predicted by equation 1-7. Jun, Y. W.; Seo, J. W. Reprinted with permission from Cheon, A. Acc. Chem. Res. 2008, 41, Thermal Properties CINs possess unique thermal properties that are directly related to their nano-scale dimensions. Numerous accounts of size-dependent melting point depression have been reported for nanomaterials such as Au, Al, and In In the case of CdS NCs, a melting point depression of ~55 % was measured with particle sizes less than 15 nm. 89 The predicted and observed thermal properties of CINs are strongly related to the surface-to-volume ratio of nanocrystals which is significantly larger than for the bulk materials. Surface atoms of NCs constitute a relatively larger percentage of the total amount of atoms as compared to larger sized crystals. These surface atoms have associated surface energies caused by increased surface tension that contributes to an overall higher energy state for the surface of a NC then for the interior. The melting of a material occurs when the chemical potential of the solid and liquid phases becomes equal. Hence, when a NC with a significant amount of its atoms on the surface melts, the extra surface energy of these atoms makes it easier to melt the NC. Thought of another way, when a material is melted, interactions between the atoms, ions, or molecules must be disrupted by the 32

33 input of energy in the form of heat. Since the surface atoms have extra energy as compared to the interior atoms they are easier to disrupt and hence the total amount of heat required for melting is reduced Catalytic Properties Unique and unexpected catalytic properties are another consequence of CIN s higher surface-to-volume ratio. By increasing the surface of the NC relative to its overall size, the number of active sites available for catalytic reactions to occur is increased. Several reports of NCs having superior catalytic activity can be found in the literature for numerous different types of important reactions. Pd, Pt, and Rh NCs have superior activity for hydrogenation reactions, while metal oxide NCs have been reported for oxidation reactions and dehydrogenations. 98,99 Gold nanoparticles can be used as a catalyst for alcohol and CO oxidation. 10, The catalytic activity is size dependent and can also be influenced by interactions with the support material, the particle preparation method, and the activation procedure. TiO 2 NCs prepared with high energy (001) crystal facets are excellent catalysts for photodegradation of environmental organic pollutants and the photooxidation of hydroxyl radicals. 103,104 These high energy crystal planes also show superior water splitting capabilities for hydrogen production when compared to commercial Degussa P25 TiO 2 powder (P25). 105 The excellent catalytic activity of (001) crystal planes of TiO 2 NCs is in part due to characteristic surface configurations with increased amounts of dangling bonds and abundant surface defects Surface Functionalization Surface functionalization of CINs is not only required to stabilize them in solution and prevent aggregation, but can also be used as a means to introduce new properties to the NCs. In most organic phase syntheses of CINs, organic ligands containing electron-rich capping groups such as carboxylates, amines, phosphines, phosphine oxides, or thiolates can coordinate to the 33

34 electron-poor metal ions on the NCs surface (e.g. Fe 3+, Cd 2+, Zn 2+ ) and to elemental metals (e.g. Au and Ag). These ligands tend to be ampliphilic having both polar and nonpolar functionalities. The end unit that is free to interact with the surrounding solvent determines the solubility of the NCs being synthesized. When the end group is hydrophobic, like in alkyl groups, the NCs are soluble in low polarity solvents such as hexanes, toluene, or chloroform. When the end groups are hydrophilic in nature they endow the CINs with solubility in polar solvents such as water or ethanol. The stability of CINs in solution is correlated to the strength of ligand bonding to the NCs surface. Weaker bonds lead to lower stability and vice versa. But, with strong ligand bonding comes an inability of the NCs to grow properly because strong bonds will block the growth of NCs and result in a reduced size of the NCs. In most of the reported literature, the strength of the ligand binding is kept intermediary so as to allow a balance between controlling particle size and keeping stability high. This not too strong, not too weak bonding allows for numerous post-synthetic modifications of the surface ligands, also called surface engineering, through ligand exchange or coordination chemistry, which allows for the precise control of the NCs solubility, stability, and other properties. There are many reported methods to engineer the surface of CINs to make them soluble in polar solvents. 45,106,107 Nag demonstrated the use of inorganic ions, e.g. S 2-, Se 2-, Te 2-, or OH -, to replace the organic ligands and make semiconductor or metal NCs water soluble. 108 In another example, Dong showed how nitrosonium tetrafluoroborate (NOBF 4 ) can be used along with secondary ligands to sequentially transfer CINs from low polarity solvents to higher polarity solvent. 109 This method allows for several functionalities to be introduced onto the NCs surface using a hierarchy of ligands that interact with one another on the surface of the NCs

35 Surface functionalization of CINs can be used to tailor the properties of the NCs and allow for their use in many bioapplications. Bioconjugation can be controlled so that specific antibodies, enzymes, nucleic acids, or aptamers can be bonded to the surface CINs; this control provides the ability to use NCs as biosensors or in bioanalytical diagnostics. Gold NCs functionalized with oligonucleotides and proteins have been used to detect the specific binding events between enzymes and their substrates or between two complimentary strands of DNA. These binding events can be monitored in vivo or in vitro by measuring the change in the SPR band of the gold NCs in solution. 110 Iron oxide NCs can be used as MRI contrast agents, and when they have oligonucleotides or other biorecognition groups on their surface, they can be used as cancer therapeutics via the application of an alternating magnetic field to produce localized heat that can destroy tumor cells. 43, Nanocrystal Materials Discussed in this Work There exists a plethora of materials that can now be synthesized with nanocrystalline dimensions. Noble metals NCs such as gold, silver, platinum, and palladium have been made in both aqueous and organic solutions. Metal oxides have been extensively studied and synthesized, ranging from binary metal oxides such as Fe 3 O 4, γ-fe 2 O 3, In 2 O 3, CuO, ZrO 2, SnO 2, and ZnO to ternary metal oxides, e.g. MFe 2 O 4 where M can be Co, Ni, Mg, Zn, or MTiO 3 where M represents Ba or Sr to name a few. 60,65,114 Likewise numerous semiconductor nanocrystals have been created that span several different types and architectures. They include but are not limited to CdSe, CdS, CdTe, PbSe, PbS, PbTe, InAs, ZnS, ZnSe, and GaAs. In the following sections, specific emphasis will be given on the two nanomaterials used throughout this work. They include Fe 3 O 4 nanocrystals and CdS/ZnS core/shell nanocrystals that have been intentionally doped with trace amounts of impurities. 35

36 1.4.1 Iron Oxide Nanocrystals Iron oxide nanocrystals can be composed of different phases of iron oxide, which include magnetite (Fe 3 O 4 ), maghemite (γ-fe 2 O 3 ), or wüstite (Fe 1-x O). Magnetite has an inverse spinel crystal structure with a face-centered cubic unit cell that is shown in Figure The lattice parameter of magnetite is nm. Fe 3 O 4 is unique among the iron oxides in that it contains both Fe 2+ and Fe 3+ cations. Eight tetrahedral sites are divided between divalent and trivalent irons where Fe 3+ occupies both tetrahedral and octahedral sites. 116 Figure 1-8. Unit cell of magnetite. Grey spheres represent iron. Red spheres represent oxygen. Created with data from Wechsler, B. A.; Lindsley, D. H.; Prewitt, C. T. Amer. Mineral. 1984, 69, 754. Figure 1-9. Unit cell of maghemite. Brownish spheres are iron. Red spheres are oxygen. Created with data from Pecharroman, C.; Gonzalezcarreno, T.; Iglesias, J. E. Phys. Chem. Miner. 1995, 22,

37 Maghemite has a similar crystal structure to magnetite except that most if not all of the irons are in the +3 oxidation state. γ-fe 2 O 3 has a cubic unit cell with a lattice parameter a equal to nm. Eight cations lie in tetrahedral sites while the rest are randomly spaced over octahedral sites. Vacancies are confined to octahedral sites. 116 The unit cell of maghemite is shown in Figure Figure Unit cell of wüstite (Fe 1-x O) with x equal to The tan spheres are iron. The red spheres are oxygen. Created with data from Jette, E. R.; Foote, F. J. Chem. Phys. 1933, 1, 29. Wüstite (Fe 1-x O) iron oxide is unstable at low pressures and therefore exists as a nonstoichiometric material under standard conditions. For the stable, cation-deficient phase, x can vary from 0.83 to 0.95 at 0.1 MPa and temperatures greater than 567 C. If cooled quickly, the non-stoichiometric material can be obtained in a metastable phase at room temperature. 116 Wüstite (Fe 1-x O) has a defective NaCl crystal structure that consists of two interpenetrating face centered cubic structures of Fe 2+ and O 2-. The lattice parameter, a, is dependent on the number of vacancies in the crystal structure and ranges from to nm. The unit cell for wüstite (Fe 1-x O) is shown in Figure Iron oxide nanocrystals have been synthesized using organometallic precursors such as iron pentacarbonyl 119 and iron oleate 36 for organic syntheses. Water soluble iron oxides have 37

38 been formed by the hydrolysis and reduction of iron(iii) cations in diethylene glycol while rapidly injecting a solution of sodium hydroxide at an elevated temperature 120 or by microemulsion techniques. 121,122 Iron oxide NCs have been extensively studied for their use as MRI contrast agents due to their superior T-2 contrast characteristics and their low toxicity in vivo. 44,81 As biofunctionalization techniques have developed, iron oxide NCs have become an attractive platform for multifunctional, simultaneous drug delivery and bioimaging. 43,112,113 The ability to synthesize iron oxide NCs with controllable size and shape using single source precursors has also made them representative reaction systems to study the kinetics of NC formation CdS/ZnS Core/Shell Nanocrystals Cadmium sulfide (CdS) is a II-VI direct band gap semiconductor with a band gap of 2.42 ev at room temperature, and it is widely used in solar cells for photoelectric conversion, light emitting Figure Unit cell of zinc-blende cadmium sulfide. Blue spheres are cadmium. Yellow spheres are sulfur. Created with data from Skinner, B. J. Amer. Mineral. 1961, 46, diodes for flat-panel displays, and sensor applications. 22, CdS comes naturally in two crystal structures, the more stable hexagonal wüstite structure (a = nm) and the cubic zinc- 38

39 blende structure (a = nm). 131 The unit cell of zinc-blende cadmium sulfide is shown in Figure Zinc sulfide (ZnS) is a wide band gap semiconductor with band gap energy of 3.6 to 3.91 ev and a high refractive index. Zinc sulfide has high transmittance in the visible region and has been widely used in ultraviolet light-emitting diodes, phosphors in flat-panel displays, and thin film electroluminescence. 22, ZnS comes in two naturally occurring crystal structures similar to CdS. There is the cubic, sphalerite structure (a = nm) and the hexagonal, wüstite structure (a = nm). 131 The unit cell of cubic ZnS is shown in Figure Figure Unit cell of cubic ZnS. Cyan sphere are zinc. Yellow spheres are sulfur. Created with data from Xu, Y. N.; Ching, W. Y. Phys. Rev. B 1993, 48, ZnS is commonly used as an epitaxially grown, inorganic overcoating material for group II-VI semiconducting nanocrystals (e.g. CdSe, CdTe, CdS), forming a type-i core/shell nanocrystal. Type-I core/shell NCs are characterized by having the valence band (conduction band) of the shell semiconductor being lower (higher) than that of the core semiconductor. In this type of core/shell NC the electron and hole are confined to the core region of the particles The ZnS layer helps passivate the surface of the core material, thus reducing structural defects and increasing the fluorescence quantum yield. Growth of an inorganic shell 39

40 also improves the stability and the size distribution of the core/shell NCs when compared to the core by itself. Core/shell NCs exist in many different types, each defined by the differences in the conduction and valence bands of the core and shell materials. Figure 1-13 shows the relationship between the core s conduction and valance bands relative to the shell s. 141 Figure Representations of different core/shell nanocrystals showing their conduction and valence band offsets relative to both the core (center rectangles) and shell (outside rectangles). Reprinted with permission from Reiss, P.; Protiere, M.; Li, L. Small 2009, 5, Summary of Present Research The presented research was undertaken in order to study the nucleation and growth of iron oxide nanocrystals under new reaction conditions, to investigate the mechanisms of using ligand exchange to create surface charge on colloidal inorganic nanocrystals and to characterize the effects of metal dopants on type-i core/shell nanocrystals. In Chapter 2, the effects of gas bubbles on the nucleation and growth of iron oxide nanocrystals is presented. It was determined that gas bubbles, via latent heat transfer, affect the concentration of active monomer by shifting the equilibrium of a key exothermic reaction to favor monomer formation. A theoretical framework is constructed along with a general discussion on how such observations change the understanding of nanocrystal formation relating to the classical LaMer mechanism. 40

41 In Chapter 3, using our group s three-step doping method, the doping of CdS/ZnS core/shell nanocrystals with Cu and Ag ions is described. It is shown that Cu catalyzes the alloying of the CdS core with the ZnS shell. This alloying process correlates to the measureable blue shift in the absorption spectra of the core/shell nanocrystals in solution, and a method to determine the extent of structural change is introduced. The core size dependence of this alloying process is calculated and presented as well. Finally, the addition of Ag to the CdS/ZnS core/shell nanocrystals is shown to have a significant effect on the fluorescence intensity of the nanocrystals in solution. The doping process can be related to the measureable decrease and subsequent increase in the fluorescence intensity as the reaction temperature and time is increased. In Chapter 4, describes a new method of producing water soluble metal oxide nanocrystals using simple, inexpensive inorganic salt ions. The method relies on the introduction of electrostatic charge, via the inorganic ions, to give the metal oxide NCs their solubility in the aqueous phase. This method is shown to work for several different compositions of metal oxides using three different inorganic salts each shown to be effective at exchanging with the original long carbon chain ligands on the surface of the metal oxide nanocrystals. We further demonstrate the application of arsenite coated iron oxide composite nanocrystals (AICN) as an alternative cancer therapeutic that has multiple advantages over other arsenic based cancer treatments. It was found that when the arsenite molecule was delivered without the iron oxide nanocrystals that the toxicity significantly increased when compared with the surface functionalized iron oxide particles. The IC 50 concentration was determined and the toxicity of the AICN was shown to be less for healthy liver cells than for Huh7 cancer cells. 41

42 Chapter 5 includes overall conclusions on the research introduced and gives suggestions/predictions of future directions for studies on colloidal inorganic nanocrystals. 42

43 CHAPTER 2 GAS-BUBBLE EFFECTS ON THE FORMATION OF COLLOIDAL IRON OXIDE NANOCRYSTALS 2.1 Prologue Over the past two decades numerous advances have been made in synthesizing colloidal inorganic nanocrystals. 31 Throughout that time, two typical synthetic methods were developed for the synthesis of high quality nanocrystals. One is based on a rapid precursor injection into a hot solvent to induce nucleation, 30,142 while the other involves a non-injection based synthesis (NIS) methodology wherein a mixture of reaction precursors is heated continuously to an elevated reaction temperature. 26,34-36 Using both of these synthetic schemes has led to the synthesis on many different varieties of colloidal inorganic nanocrystals with a strong degree of control over their size, shape, and composition. 31 Using these two methods, metal, semiconductor, and metal oxide nanocrystals of varying shapes have been made, including spheres, cubes, plates, and rods. 31,143,144 Studying and understanding the nucleation and growth of colloidal inorganic nanocrystals is being actively pursued in many different fields of science. By understanding the chemical processes involved in nanocrystal formation, the ability to tailor the inorganic materials size dependent properties, such as size-dependent semiconductor band gap, radiative rate, solid-solid phase transition pressure, superparamagnetic transition temperature, and surface plasmon resonance frequency, is made possible. 1,2,89, Several mechanisms have been proposed to explain the evolution of the chemical precursors from the early stages of monomer formation to the chemical processes involved in the nucleation and growth of nanocrystals. 41,142,148,149 It remains a difficult challenge to connect the molecular mechanisms of precursor decomposition to the formation of nanocrystal nuclei. Most results to date rely on the LaMer diagram and theory of nucleation to explain the processes involved in nanocrystal formation

44 Classical nucleation theory is based on the liquid-drop model and the Gibbs-Thompson equation claiming that the supersaturation of an active monomer is the driving force behind nucleation and growth of crystals. 28,29 LaMer further used this theory to separate the formation of crystals into three stages: prenucleation, nucleation, and growth, which are illustrated in the LaMer diagram. 49 The LaMer diagram shows that crystal nucleation occurs when the active monomer reaches a threshold concentration (named the critical supersaturation point ) and ends when the active monomer concentration falls back below this point. In successful syntheses, the separation of the nucleation and growth stages leads to nanocrystals that are nearly monodispersed, which is highly desirable for their further characterization and application. While still the standard explanation of crystal formation, the liquid-drop model tied with classical nucleation theory breaks down when attempting to assign macroscopic thermodynamic properties (i.e. solubility) to nanoscale systems. Likewise, using this standard explanation fails to properly explain the observed kinetics of nanocrystal nucleation and growth reported thus far. 39,51 Difficulties arise when trying to verify predictions of classical nucleation theory due to the inability to measure active monomer concentrations or even to identify what the active monomer is on a molecular level. Furthermore, there is a gap in the understanding of the types of chemical reactions that are involved in the synthesis of active monomers and nuclei from the precursors used in nanocrystal syntheses. It is generally hypothesized that these chemical reactions are complex and may include numerous parallel, competing pathways, and/or consecutive steps to create nuclei. Indeed even a simple crystal unit cell has numerous atomic components (i.e. atoms and bonds) especially when considering the role of surface ligands on their formation. 44

45 To gain a better understanding of the types of chemical reactions involved in nanocrystal nucleation and growth, new insights must be made on the nature of the active monomers and the chemical reactions that make them. To address these questions, gas bubbles were used to perturb the nucleation and growth of colloidal iron oxide nanocrystals. The results demonstrate that exothermic reactions exist in the formation of active monomers that directly yield nuclei that can further grow into nanocrystals of different sizes. The results support the idea that gas bubbles can be used to control the nucleation rate of iron oxide nanocrystals. In Chapter 2, the use of gas bubbles from boiling solvents under atmospheric (or reduced) pressure to facilitate the formation of high quality iron oxide nanocrystal is described first. Second, artificial bubbles from an argon flow are shown to be able to replace the solvent gas bubbles in controlling the nucleation and growth of the nanocrystals. Third, a mechanistic study on the separation of nucleation and growth of iron oxide nanocrystals using a controlled argon flow is reported. Finally, some general discussions on the nature of the chemical reactions involved in the formation of iron oxide nanocrystals are presented. 2.2 Experimental Section Chemicals Iron chloride hexahydrate (FeCl3 6H 2 O, 99 %) was purchased from Acros Organic; oleic acid (OLA, 90 %), 1-octadecene (ODE, 90 %), and 1-tetradecene (TDE, 92 %) were purchased from Aldrich. n-docosane (DCA, 98 %) and n-tetracosane (TCA, 99 %) were purchased from Alfa Aesar. Sodium oleate (97 %) was purchased from Tokyo Chemical Industry. Nanopure water (18 MΩ cm) was made by a Barnstead Nanopure Diamond system. All the other solvents were purchased from Fisher Scientific International, Inc. 45

46 2.2.2 Synthesis of Iron Oxide Nanocrystals Synthesis of iron-oleate precursor Iron oleate was synthesized using a modified literature method. 36 In a typical synthesis, iron chloride (10.8 g, 40 mmol) was first dissolved in a mixture of 80 ml ethanol and 60 ml Nanopure water in a three-neck flask (500 ml). Afterwards, sodium oleate (36.5 g, 120 mmol) was quickly added to the iron chloride solution along with 140 ml hexane. The resulting solution was allowed to stir until the sodium oleate was completely dissolved. Next, the reaction solution was heated to ~ 58 C for a 4-hour reflux. When the reaction was finished and cooled to room temperature, the upper organic layer containing the iron oleate was washed three times with 30 ml of Nanopure water in a 250 ml separatory funnel. Excess hexane was then evaporated using a rotovap. The resulting iron oleate was transferred into a 100 ml round bottom flask and connected to a Schlenk line where it was placed under a vacuum of ~ 50 mtorr overnight, and then the iron oleate was well sealed in a glass vial and stored in a desiccator for two days of aging. The reaction yield of iron oleate is ~ 90 % Synthesis in boiling solvents For a boiling temperature of 290 C: iron oleate (1 mmol, 0.9 g) and oleic acid (0.55 mmol, g) were added into a three-neck flask (25 ml) with a solvent mixture of TDE/ODE (1.75 g/3.25 g). The mixture was stirred under argon flow at room temperature. After 10 minutes, the reaction solution was heated to 290 C at a heating rate of ~18 C/min. Reaction time was counted from the moment when 290 C was reached. After one hour, the reaction solution was quickly cooled to room temperature by blowing air across the reaction flask. The resulting ironoxide nanocrystals were precipitated using acetone, and then were purified using three rounds of precipitation/redispersion cycles with acetone and hexane as the solvents. After purification, the product was dispersed in nonpolar solvents such as hexane or toluene. 46

47 Similar procedures were used to conduct the syntheses in boiling solvents at 300, 320, 340, and 365 C, except that the reaction solvents were used as follows: TDE/ODE (1 g/4 g), pure ODE (5 g), ODE/ DCA (2 g/3 g) and DCA (5 g), respectively Synthesis in non-boiling solvents For a non-boiling temperature of 290 C: iron oleate (1 mmol, 0.9 g) and oleic acid (0.55 mmol, g) were added into a three-neck flask (25 ml) with a solvent mixture of TDE/ODE (1 g/4 g). The mixture was stirred under Argon flow at room temperature. After 10 minutes, the reaction solution was heated to 290 C at a heating rate of ~18 C/min. Reaction time was counted from the moment when 290 C was reached. After one hour, the reaction solution was quickly cooled to room temperature by blowing air across the reaction flask. The resulting ironoxide nanocrystals were precipitated using acetone, and then were purified using three rounds of precipitation/redispersion cycles with acetone and hexane as the solvents. After purification, the product was dispersed in nonpolar solvents such as hexane or toluene. Table 2-1. Summary of different solvent compositions and reaction temperatures used for the boiling and non-boiling reactions during the synthesis of iron oxide nanocrystals. Temperature ( C) Boiling Reaction Non-boiling Reaction 290 C 1.75 g TDE/ 3.25 g ODE 1 g TDE/ 4 g ODE 300 C 1 g TDE/ 4 g ODE 5 g ODE 320 C 5 g ODE 2 g ODE/ 3 g DCA 340 C 2 g ODE/ 3 g DCA 5 g DCA 365 C 5 g DCA 5 g TCA Reprinted with permission from Lynch, J.; Zhuang, J. Q.; Wang, T.; LaMontagne, D.; Wu, H. M.; Cao, C. J. Am. Chem. Soc. 2011, 133, Similar procedures were used to conduct the syntheses in non-boiling solvents at 300, 320, 340, and 365 C, except that the reaction solvents were used as follows: pure ODE (5 g), ODE/DCA (2 g/3 g), DCA (5 g), and TCA (5 g), respectively (Table 2-1). 47

48 The synthesis of iron-oxide nanocrystals under reduced pressures In a typical experiment, iron oleate (1 mmol, 0.9 g) and oleic acid (0.55 mmol, g) were added into a three-neck flask (25 ml) with a solvent mixture of pure ODE (5 g) at a reaction temperature of 300 C or pure TCA (5 g) at a reaction temperature of 320 C. The mixture was stirred under argon flow at room temperature. After 10 minutes, the reaction solution was heated to the chosen reaction temperature at a heating rate of ~18 C/min. Once this temperature was reached, a vacuum pump was turned on to decrease the reaction pressure which was fine-tuned using the clamps on the vacuum tubing (Figure 2.3), and the reaction temperature dropped to the chosen temperature. Once boiling was observed at the chosen temperature, reaction time was counted as zero. Caution should be taken not to allow the vacuum to pull the reaction mixture up through the condenser and into the tubing. Each synthesis was made under reduced pressure and 1 atm. After one hour of reaction, the vacuum was shut off and the reaction mixture was quickly cooled to room temperature. The resulting iron-oxide nanocrystals were precipitated using acetone, and then were purified using three rounds of precipitation/redispersion cycles with acetone and hexane as the solvents Synthesis of iron-oxide nanocrystals using argon bubbling In a typical experiment, iron oleate (1 mmol, 0.9 g) and oleic acid (0.55 mmol, g) were added into a three-neck flask (25 ml) with solvents of pure ODE (5 g) at a reaction temperature of 300 C or pure TCA (5 g) at a reaction temperature of 300 C or 330 C. The mixture was stirred under Argon flow at room temperature. After 10 minutes, the reaction solution was heated at a rate of ~18 C/min to a temperature 10 C above the chosen reaction temperature. At this point, a long stainless steel needle connected to a plastic joint was inserted into the reaction mixture and connected to Ar flow tubing. The Ar flow was turned on to induce bubbling into the reaction mixture. After a one-hour reaction time the Ar bubbling was shut off 48

49 and the reaction mixture was quickly cooled to room temperature. The resulting iron-oxide nanocrystals were precipitated using acetone, and then were purified using three rounds of precipitation/redispersion cycles with acetone and hexane as the solvents Synthesis of iron-oxide nanocrystals under controlled Ar bubbling Two experiments were performed. In the first experiment, iron oleate (1 mmol, 0.9 g) and oleic acid (0.55 mmol, g) were added into a three-neck flask (25 ml) with pure ODE (5 g) as the solvent at a reaction temperature of 300 C. The reaction was allowed to stir for 10 minutes under argon flow and was then heated at a rate of 18 C/min to 10 C above the chosen reaction temperature. At this point, a long stainless steel needle connected to a plastic syringe housing was inserted into the reaction mixture and connected to Ar flow tubing. The Ar flow was slowly turned on to induce bubbling into the reaction mixture. The reaction solution cools to the chosen reaction temperature and the reaction time is counted as zero when the reaction solution reaches 300 C. The Ar bubbling was shut off at 12 minutes, and after one hour the reaction solution was cooled to room temperature. The resulting nanocrystals were purified using three precipitation/redispersion cycles with hexane and acetone as solvents. Similar procedures were used to conduct the second experiment, except that the Ar bubbling was turned off after 18 minutes The Determination of Iron-Oxide Crystallization Yield In a typical experiment, iron oleate (1 mmol, 0.9 g) and oleic acid (0.55 mmol, g) were added into a three-neck flask (25 ml) with a solvent mixture of TDE/ODE (1 g/4 g) for boiling syntheses or pure ODE (5 g) for non-boiling syntheses, both at a reaction temperature of 300 C. The mixture was stirred under argon flow at room temperature. After 10 minutes, the reaction solution was heated to 300 C at a heating rate of ~18 C/min. Reaction time was counted from the moment when 300 C was reached. After the reaction was cooled to room 49

50 temperature, the resulting nanocrystals were purified using one precipitation/redispersion cycle with hexane and ethanol as solvents, and then the nanocrystals were further purified using three precipitation/redispersion cycles with hexane and acetone as solvents. Special care was used to minimize the loss of iron-oxide nanocrystal products. After purification, the sample was dried in a vacuum oven overnight at 50 C with a reduced pressure (30 mmhg), and then the resulting sample was used for thermogravimetric analysis (TGA). The mass percent of ligand molecules on the nanocrystals surface was measured by TGA, and then this mass percent was subtracted from the dried nanocrystal weight previously recorded. The experimental iron-oxide crystallization yield was calculated via dividing the actual weight of the ligand-free iron oxides by the theoretical reaction yield as shown in Eq These boiling and non-boiling experiments were carried out three times to eliminate the experimental errors in the determination of iron-oxide crystallization yield. The Student s t-test was performed using Eq. 2-2 along with data from Table 2-2, and the t calculated is This t calculated value is corresponding to a 99.5 % confidence level. 150 RRY 100 (2-1) where RRY is the relative reaction yield. (2-2) where is the mean reaction yield for boiling syntheses, is the mean reaction yield for non-boiling syntheses, n 1 is the number of boiling syntheses made, n 2 is the number of nonboiling syntheses made and s pooled is expressed as follows, ( ) ( ) (2-3) 50

51 where s 1 is the boiling syntheses reaction yield standard deviation and s 2 is the non-boiling syntheses reaction yield standard deviation. Table 2-2. Numerical calculation results from average reaction yield determinations. For boiling and non-boiling reactions at 300 C run in triplicate. All measurements are shown along with their respective uncertainties. Non-boiling Boiling 16d 16d 2 16d 3 16e 16e 2 16e 3 Weight of Iron Oleate (mg) 899±2 904±2 902±2 898±2 909±2 911±2 Theoretical yield at 100 % (mg) 77.1± ± ± ± ± ±0.2 Weight of crude product (mg) 55.2± ± ± ± ± ±0.1 Percent ligand coverage from TGA 21.3±0.1 15± ± ± ± ±0.1 Weight final product (mg) 43.4± ± ± ± ± ±0.1 Percent reaction yield 56.3± ± ± ± ± ±0.3 Average ligand coverage Average reaction yield 16.5±0.2 % Standard deviation= ±0.5 % Standard deviation= ±0.2 % Standard deviation= ±0.5 % Standard deviation=4.1 Reprinted with permission from Lynch, J.; Zhuang, J. Q.; Wang, T.; LaMontagne, D.; Wu, H. M.; Cao, C. J. Am. Chem. Soc. 2011, 133, Kinetic Study of Iron-Oxide Syntheses Under boiling and non-boiling conditions In a typical experiment, iron oleate (1 mmol, 0.9 g) and oleic acid (0.55 mmol, g) were added into a three-neck flask (25 ml) with a solvent mixture of TDE/ODE (1 g/4 g) for boiling syntheses or pure ODE (5 g) for non-boiling syntheses, both at a reaction temperature of 300 C. The mixture was stirred under argon flow at room temperature. After 10 minutes, the reaction solution was heated to 300 C at a heating rate of ~18 C/min. Reaction time was 51

52 counted from the moment when 300 C was reached. Aliquots (~0.5 ml) of the reaction solution were taken periodically with a glass syringe at different times during the course of the reaction and allowed to cool. After one hour, the reaction solution was cooled to room temperature. The resulting nanocrystals from both the aliquots and final products were purified using one precipitation/redispersion cycle with hexane and ethanol as solvents, and then the nanocrystals were further purified using three precipitation/redispersion cycles with hexane and acetone as solvents. Special care was used to minimize the loss of iron-oxide nanocrystal products Under controlled Ar bubbling Two sets of experiments were performed. In the first set of experiments, iron oleate (1 mmol, 0.9 g) and oleic acid (0.55 mmol, g) were added into a three-neck flask (25 ml) with a solvent of pure ODE (5 g) at a reaction temperature of 300 C. The reaction was allowed to stir for 10 minutes under argon flow and was then heated at a rate of 18 C/min to 10 C above the chosen reaction temperature. At this point, a long stainless steel needle connected to a plastic joint was inserted into the reaction mixture and connected to Ar flow tubing. The Ar flow was slowly turned on to induce bubbling into the reaction mixture. The reaction solution was cooled to the chosen reaction temperature and the reaction time was counted when the reaction solution reached 300 C. Aliquots (~0.5 ml) of the reaction solution were taken periodically with a glass syringe at different times during the course of the reaction and allowed to cool. After one hour, the reaction solution was cooled to room temperature. The resulting nanocrystals from both the aliquots and final products were purified using one precipitation/redispersion cycle with hexane and ethanol as solvents, and then the nanocrystals were further purified using three precipitation/redispersion cycles with hexane and acetone as solvents. Special care was used to minimize the loss of iron-oxide nanocrystal products. 52

53 Similar procedures were used to conduct the second set of experiments, except that the Ar bubbling was turned off after 15 minutes Calculation of average nanocrystal diameter and distribution For all reactions outlined above, the average diameters were calculated using a one of two methods. For reactions with bubbles present, assuming spherical nanocrystals, Image J software was used to determine ensemble population distributions of the areas of the circular two dimensional projections of the spherical nanocrystals. Afterwards, using OriginLab Origin 8.5 graphing software, the area populations was converted into diameter populations that were several hundred nanocrystals in number. Then statistical analysis was performed on the diameter populations with the average value plus or minus the standard deviation being reported with the TEM image. Furthermore, histograms of the diameter populations were created to verify Gaussian distributions for the monodispersed samples. For reactions with no bubbles present, average diameters were reported but with assumptions that the actual shape of the nanocrystals were not spherical. The goal was to give relative comparisons of the size difference between reactions with bubbles and without bubbles. Using Adobe Photoshop CS software, the average cross sectional length of the two dimensional projection of the 3D nanocrystals were measured in pixel units. Between 50 and 100 nanocrystals were measured and populations were converted into nanometer units using the length of the scale bar in pixels as a conversion factor. Then statistical analysis was performed on the diameter populations with the average value plus or minus the standard deviation being reported with the TEM image. Furthermore, histograms of the diameter populations were created to verify Gaussian distributions for the monodispersed samples. 53

54 2.2.5 Instrumentation TEM measurements Transmission electron microscopy (TEM) measurements were performed on a JEOL 200X operated at 200 kv. The specimens were prepared as follows: a particle solution in toluene (10 μl) was dropped onto a 200-mesh copper grid and was dried overnight at ambient conditions TGA measurements Thermogravimetric analysis (TGA) measurements were made on a Hi-Res TGA 2950 from TA Instruments. Samples were prepared as follows: mg of dried sample were weighed and placed on a platinum pan. Samples were run from C at 5 C/min under N 2 flow at ml/min. 2.3 Results and Discussion Iron-Oxide Nanocrystal Synthesis System The synthesis strategy used in this work is a NIS system modified from Hyeon and coworkers methodology. 36 The synthesis involves only 3 components: the iron oleate precursor, oleic acid as the ligand, and a long, hydrocarbon solvent, such as 1-octadecene (ODE), whose boiling point rises with increasing hydrocarbon chain length. After the reaction solution is heated to an elevated temperature (ie C) the iron oleate precursor begins to decompose and the formation of iron oxide nanocrystals takes place. The iron oxide nanocrystals consist of a mixture of magnetite (Fe 3 O 4 ) and maghemite (γ-fe 2 O 3 ) phases whose composition can be written as (Fe 3 O 4 ) x (Fe 2 O 3 ) 1-x where x ranges from 0.5 to 0.7 depending on nanocrystal size. 36 The major chemical reactions that are involved in the decomposition of iron oleate have yet to be fully indentified, but major byproducts of metal carboxylate decomposition have been determined to be CO 2 and H 2 in previous reports. 41,

55 The chemical nature and reactivity of the iron oleate precursor prepared via Hyeon s method is known to have a substantial effect on the nanocrystals that are formed. By placing the precursor under a vacuum of 50 mtorr overnight and then aging it in a desiccator at room temperature for two days, the heating rate of the reaction was able to be increased from 10 C/min, as used in Hyeon s method, to 25 C/min in our syntheses 154. The resulting iron oxide particles had size distributions that were as good as those reported using the slower heating rate and untreated precursor (Figure 2-1). The extra iron oleate precursor treatment alters the chemical reactivity by removing trace amounts of solvent molecules (e.g. ethanol and water) from the iron oleate complexes, while the extra aging time allows the precursor to become homogenous and polymerized to allow for a more uniform reactivity to develop. Reaction precursors with uniform reactivity have been found to be important in promoting the burst of nucleation in the synthesis of monodispersed group II-VI semiconductor nanocrystals. 34 In addition, aging effects on precursor reactivity were also observed by Peng et al. in the synthesis of CdSe nanorods. 155 Other examples of the nature of the iron oleate precursor affecting the outcome of the final iron oxide nanocrystals are shown by Bronstein. 156 They found that thermal treatment of the iron oleate precursor could improve the quality of the iron oxide nanocrystals that formed The Effects of Solvent Gas Bubbles Generated From Boiling Solvents Using a boiling solvent to synthesize nearly monodispersed iron oxide nanocrystals has been demonstrated previously in the literature. 36,119 To determine the effect of solvent gas bubbles on the formation of iron oxide nanocrystals, two sets of experiments were carried out (boiling vs. non-boiling) at five different temperatures ranging from 290 to 365 C. The concentration of iron oleate precursor and oleic acid ligand were kept constant in these syntheses. Four types of long-chain hydrocarbon solvents were used to create the boiling or non- 55

56 boiling reaction environments at these reaction temperatures: n-docosane (DCA), 1-octadecene (ODE), n-tetracosane (TCA), and 1-tetradecene (TDE). Pure ODE (or DCA) was used as a boiling solvent at 320 o C (or 365 o C). Two-solvent mixtures TDE/ODE at a mass ratio of 7:13 or 1:4, and ODE/DCA at 2:3 were used to create boiling environments at 290, 300, and 340 C, respectively (Table 2-1). The non-boiling systems were also created using either a pure solvent or a mixture of two solvents. In a typical experiment, iron oleate (1.0 mmol), oleic acid (0.55 mmol), and a pure or mixed long-chain hydrocarbon solvent (5 g) were added in a three-neck flask, and the mixture was heated to an elevated reaction temperature under stirring at a heating rate of ~18 C/min. This reaction temperature was maintained for 60 minutes, and the synthesis was terminated by cooling the reaction solution to room temperature. The resulting iron-oxide nanocrystals were purified by a triple precipitation/redispersion treatment using acetone and hexane. TEM measurements of the iron oxide nanocrystals from the two sets of experiments (boiling vs. non-boiling) showed marked differences in the final size, shape, and size distribution of the iron oxide nanocrystals (Figure 2-1). The boiling reactions resulted in spherical iron oxide nanocrystals with mean diameters of 5.2, 6.5, 9.9, 13.5 and 16.7 nm at respective reaction temperatures of 290, 300, 320, 340, and 365 C (Figure 2-1a-e). These boiling syntheses had typical size distributions of less than 6 %. In contrast to the boiling experiments, the non-boiling experiments resulted in irregular polyhedrons of iron oxide with an average size of approximately three times that of the spherical nanocrystals made from boiling reactions at the same reaction temperature. Furthermore, the observed size distributions for the non-boiling experiments were always above 10 %. 56

57 Figure 2-1. TEM images with histograms of iron oxide nanocrystals from syntheses in boiling solvents (a-e) and non-boiling solvents (f-j) at reaction temperatures of 290 C (a,f); 300 C (b,g); 320 C (c,h); 340 C (d,i); and 365 C (e,j). The size of the resulting particles is 5.2±0.4 nm in (a), 6.5±0.3 nm in (b), 9.9±0.4 nm in (c), 13.5±1.0 nm in (d), 16.7±1.1 nm in (e), 12.3±2.1 nm in (f), 27.1±5.6 nm in (g), 33.9±10.4 nm in (h), 41.9±12 nm in (i), and 44 ±8.5 nm in (j). All scale bars are 50 nm. Reprinted with permission from Lynch, J.; Zhuang, J. Q.; Wang, T.; LaMontagne, D.; Wu, H. M.; Cao, C. J. Am. Chem. Soc. 2011, 133, All other things being equal, final larger sized nanocrystals are a direct consequence of fewer numbers of initial nuclei being formed in the non-boiling syntheses when compared to the boiling syntheses. 28,29 Therefore, for a given amount of precursor, a larger amount of nuclei will 57

58 result in smaller sized nanocrystals and vice versa. It was hypothesized that the presence of solvent gas bubbles in the boiling reactions promoted the nucleation of iron oxide nanocrystals. To determine whether or not the solvent gas bubbles affected the growth as well as the nucleation of iron oxide nanocrystals, experiments were run to investigate how the overall crystallization ( or reaction yield) differed between boiling and non-boiling reactions. The crystallization yield is a direct outcome of both nucleation and growth of the nanocrystals Iron Oxide Crystallization Yield Three parallel syntheses in boiling or non-boiling conditions at 300 C using an ODE/TDE mixture as the boiling solvent and pure ODE as the non-boiling reaction solvent were conducted to compare the effects of solvent gas bubbles of the iron oxide crystallization yield. These syntheses were carried out using an identical concentration of iron oleate and oleic acid with a heating rate of 18 C/min. After one hour at 300 C, the resulting iron oxide nanocrystals were thoroughly purified and dried at 50 C in a vacuum oven. Afterwards, the mass percent of ligand molecules on the particles surface was measured via thermogravimetric analysis (TGA, Figure 2-2). The experimental crystallization yield was calculated by dividing the actual weight of the ligand-free iron oxide nanoparticles by the theoretical reaction yield. An average yield of (77±4 %) was found for the iron oxide syntheses under the boiling condition, whereas a yield of (57±1 %) was determined for the synthesis under the non-boiling condition (Table 2-2). The significance of the differences in the reaction yields between the boiling and non-boiling syntheses was supported by the small standard deviations for each set of experiments (less than 5 %) and by the confidence level calculated by the student t-test method (99.5 %) between the differences in the two results. This data supports the hypothesis that solvent gas bubbles affect the crystallization yield in the synthesis of iron oxide nanocrystals. 58

59 Figure 2-2. TGA measurements for iron oxide crystallization yield syntheses in boiling solvent (a) and non-boiling solvent (b) used to determine mass percent ligand on nanocrystal surface. Reprinted with permission from Lynch, J.; Zhuang, J. Q.; Wang, T.; LaMontagne, D.; Wu, H. M.; Cao, C. J. Am. Chem. Soc. 2011, 133, Taken together the crystallization yield results further support the claim that the solvent gas bubbles not only affect the nucleation, but also the growth of iron oxide nanocrystals. One explanation that might explain these effects is the slight difference is solvent composition between the boiling and non-boiling syntheses. To address this possibility, experiments were performed using identical solvent composition and reaction temperatures under reduced pressure conditions The Effects of Solvent-Gas Bubbles Generated from Boiling Solvents with Reduced Pressure Two sets of iron-oxide nanocrystal syntheses (boiling vs. non-boiling) with identical solvents, reaction temperatures, and concentrations of iron oleate and oleic acid were run. ODE and TCA were used as solvents to perform the iron oxide syntheses at 300 and 320 C, respectively. To induce boiling below the normal boiling point of the solvents used, a vacuum pump was employed to reduce the pressure of the reaction system in a precise, controllable way (Figure 2-3). The temperature of the reaction solution fluctuated by ± 3 C. After performing the experiments, TEM measurements were made and showed that significant differences existed between the boiling and non-boiling syntheses. The boiling reactions yielded smaller spherical nanocrystals with a narrow size distribution, while the non- 59

60 boiling counterpart reaction resulted in larger iron oxide polyhedrons with a poor size distribution (Figure 2-4b and d). The size and size distribution for the reduced pressure boiling syntheses were similar to the iron oxide nanocrystals made using naturally boiling conditions at the same reaction temperatures (Figure 2-4a and c, Figure 2-1b and c). These results rule out the possibility that the solvent composition plays a major role in controlling the nucleation and growth of iron oxide nanocrystals. Figure 2-3. Reaction setup for reduced pressure experiments. (Left panel) a picture of the actual reaction setup for reduced pressure experiments, and (right panel) the scheme of reaction setup. Reprinted with permission from Lynch, J.; Zhuang, J. Q.; Wang, T.; LaMontagne, D.; Wu, H. M.; Cao, C. J. Am. Chem. Soc. 2011, 133, When trying to discern the role that boiling solvents have on the formation of iron oxide nanocrystals, three main effects were considered. The bubbles could (1) provide low free-energy solvent-gas interfaces, (2) promote mass transfer, and (3) increase the heat transfer coefficient of the reaction solutions. 157,158 Low free-energy interfaces created by solvent-gas bubbles have 60

61 Figure 2-4. TEM images and histograms of iron oxide nanocrystals synthesized in boiling solvents under reduced pressures (a, c) and in non-boiling solvents under 1 atm (b, d). The syntheses with pure ODE as the solvent and reaction temperature at 300 C (a, b): the resulting particles exhibit 7.3±0.3 nm in diameter (a) and 18.3 ± 3.4 nm in (b). The syntheses with pure TCA as the solvent and reaction temperature at 320 C (c, d): the resulting particles exhibit 9.4±0.4 nm in diameter in (c), and 36.8±5.6 nm in (d). All scale bars are 50 nm. Reprinted with permission from Lynch, J.; Zhuang, J. Q.; Wang, T.; LaMontagne, D.; Wu, H. M.; Cao, C. J. Am. Chem. Soc. 2011, 133, been observed to act as preferential sites for the heterogeneous nucleation of additional gas bubbles. 28,29 The free energy of bubble gas/liquid interfaces is determined by the surface tension of the liquid or in the case of the iron oxide syntheses the solvent being used. For long, hydrocarbon solvents like the ones used in these experiments the surface tension is approximately 50 times smaller than that of solid iron oxides. 159,160 For this reason the bubbles acting as nucleation sites was disregarded as a major effect on the formation of iron oxide nanocrystals. The second role gas bubbles can play on the formation of iron oxide nanocrystals is to increase the amount of mass transfer that is occurring in the reaction solution. Solvent gas bubbles can increase the amount of convection occurring in solution by spontaneously forming 61

62 and growing throughout the course of the reaction. Yet, this increased mass convection is significantly less than the amount of convection that is already occurring in the vigorously stirred (~1100 rpm) reaction solution from the magnetic stir bar. While mass transfer is increased by the presence of the solvent gas bubbles, it is unlikely to be a major cause of the observed differences between the boiling and non-boiling syntheses. By process of elimination, increased heat transfer was hypothesized to be the major reason for the observed effects on the nucleation and growth of iron oxide nanocrystals. Latent heat transfer a process of giving off or absorbing heat without changing temperature, is a well-known phenomenon that can be significantly increased through the process of boiling. 157,158,161,162 Boiling-induced latent heat transfer giving off and absorbing heat simultaneously has been used in many technological applications related to energy production such as the cooling of nuclear reactors. 162 It is understood that during the nucleation and growth of iron oxide nanocrystals, both endothermic and exothermic reactions take place. The process of boiling-induced latent heat transfer could act to promote such endothermic and exothermic reactions. The decomposition reactions related to the iron oleate precursor are endothermic processes, while the iron oxide crystal lattice formation processes are exothermic reactions. It is crucial to determine whether the endothermic decomposition reactions or the exothermic crystallization reactions influence the formation of iron oxide nanocrystals. To answer this question, experiments were designed using a room temperature Ar flow as a source of cold bubbles (Figure 2-5). Although the cold Ar bubbles can absorb only limited amounts of heat from the surrounding hot reaction solution due to their small molar heat capacity, 131 the presence of these bubbles can substantially facilitate the evaporation of the surrounding hot solvent molecules, 157,158 which can absorb a larger amount of heat from the 62

63 reaction solution. These two processes, the absorption of latent heat from the reaction solution and the evaporation of the solvent molecules, can act on the local microenvironment around the bubbles to promote the exothermic reactions involved in iron oxide nanocrystal formation. Figure 2-5. Reaction setup for Ar bubbling experiments. (Left panel) A picture of actual reaction setup used to flow Ar bubbles through the reaction solution. (Right panel) Scheme of reaction setup. Reprinted with permission from Lynch, J.; Zhuang, J. Q.; Wang, T.; LaMontagne, D.; Wu, H. M.; Cao, C. J. Am. Chem. Soc. 2011, 133, The Effects of Ar Bubbles Three iron-oxide nanocrystal syntheses with identical concentrations of iron oleate and oleic acid were performed to identify the effects that Ar bubbles had on iron oxide nanocrystal formation. Both ODE and TCA were used as the solvent for the synthesis at 300 C, and TCA was used as the solvent for the synthesis at 330 C. In these syntheses, the reaction solutions were first heated to an elevated temperature that is 10 C higher than the target reaction temperature. Once this temperature was reached, the Ar bubbles were turned on and the temperature of the solution was lowered to the target temperature for a one hour reaction (Figure 2-5). 63

64 Interestingly, the Ar-bubbling syntheses yielded nearly monodispersed, spherical iron oxide nanocrystals similar to the case when boiling bubbles were used (Figure 2-6 and Figure 2-1a-e). Figure 2-6. TEM images and histograms of iron-oxide nanocrystals made from Ar bubbling experiments: particles made in pure ODE as solvent at 300 C (a) with 6.7±0.4 nm diameter; in pure TCA as solvent at 300 C (b) with 7.0±0.3 nm in diameter; in pure TCA as solvent at 330 C (c) with 11.6±0.7 nm diameter. All scale bars are 50 nm. Reprinted with permission from Lynch, J.; Zhuang, J. Q.; Wang, T.; LaMontagne, D.; Wu, H. M.; Cao, C. J. Am. Chem. Soc. 2011, 133, The two syntheses at 300 C gave iron oxide nanocrystals with similar diameter: 6.7±0.4 nm in ODE and 7.0±0.3 in TCA (Figure 2-6a and b). These measured average diameters are similar to those determined for the boiling syntheses at atmospheric and reduced pressures at the same reaction temperatures (Figure 2-1b and 2.4a). The Ar-bubbling synthesis at 330 C produced iron oxide nanocrystals with an 11.6±0.7 nm diameter (Figure 2-6c). This size lies in between the sizes of those particles made by boiling bubbles at 320 and 340 C (Figure 2.1c and d). Taken together, the results from the Ar-bubbling experiments illustrate that Ar bubbles can effectively mimic the function of boiling bubbles in controlling the formation of iron-oxide nanocrystals. 64

65 Another important observation to take from this set of experiments is that the exothermic reactions (but not the endothermic reactions) play a dominant role in controlling the formation of iron oxide nanocrystals, since the Ar bubbles can only promote the exothermic reactions taking place during iron oxide nanocrystal formation. The demonstration that Ar-bubbling can replace a boiling solvent in forming nearly monodispersed iron oxide particles allows the use of only one high boiling solvent to produce a large size range of nanocrystals by simply tuning the reaction temperature. Additionally, Ar-bubbling opens the door to a new investigation technique for nanocrystal nucleation and growth, namely by turning on/off the Ar bubbles at different times while maintaining identical conditions of reaction temperature, solvent type, and concentrations of precursors and ligands Kinetic Studies on Gas-Bubble Effects To gain more information on the effects of Ar-bubbling on nanocrystal nucleation and growth, kinetics studies were performed on two sets of syntheses at 300 C (boiling vs. nonboiling, and bubbling vs. non-bubbling). Both sets of syntheses used identical concentrations of iron oleate (1 mmol) and oleic acid (0.55 mmol). For the boiling synthesis, ODE/TDE (4 g /1 g) was used as the solvent, while for the rest of the syntheses pure ODE was used as the solvent. In these experiments, the reaction solutions were heated to the reaction temperature at a rate of 18 C/min. The moment when the reaction solution was stabilized at 300 C was set as reaction time zero, and then serial aliquots were taken for kinetic studies using TEM analysis. No nucleation was observed for the boiling syntheses until ~13 min reaction time resulting in 5.1 nm iron oxide nanocrystals at 15 min (Figure 2-7). With further time, the iron oxide nanocrystals grew and both their size distribution and growth rate decreased during the course of the reaction. After one hour of reaction, the final particles had an average diameter of 6.5±0.3 nm (Figure 2-7d). Afterwards, no substantial increase in nanocrystal size was observed, 65

66 Figure 2-7. Kinetics study of the syntheses of iron-oxide nanocrystals in boiling (a-d) and nonboiling solvents (e-h). TEM images with histograms of iron-oxide nanocrystals were taken at different reaction times during the synthesis. The moment when the solution temperature reached 300 C was counted as the zero reaction time. The resulting particles were taken from the synthesis in boiling solvent at: (a) 15 min (5.1±0.4 nm in diameter), (b) 20 min (5.6±0.4 nm in diameter), (c) 30 min (6.0±0.4 nm in diameter), and (d) 60 min (6.5 ± 0.3 nm in diameter). The resulting particles were taken from the synthesis in the non-boiling solvent at: (e) 18 min (5.1±0.5 nm in diameter), (f) 25 min (15.7±5.8 nm in diameter), (g) 30 min (17.2±6.5 nm in diameter), and (h) 60 min (19.9±4.9 nm in diameter). All scale bars are 50 nm. Reprinted with permission from Lynch, J.; Zhuang, J. Q.; Wang, T.; LaMontagne, D.; Wu, H. M.; Cao, C. J. Am. Chem. Soc. 2011, 133,

67 indicating that the nanocrystal growth was complete and the reaction precursors were almost consumed during the 1-hr reaction. These results show that while iron oleate can decompose at 300 C, 36,41 reaching this temperature is not the sole requirement for nanocrystal nucleation to occur in the reaction solution. A similar delayed-nucleation phenomenon was previously observed by Alivisatos et al. in the preparation of iron-oxide nanocrystals using an injection-based synthesis. 148 Delayednucleation supports the hypothesis that nucleation does not strongly depend on the concentration of the iron oleate precursor itself. Instead, nucleation is likely dependent on the concentration of the iron oleate precursor decomposition products known as active monomers. This nucleation still follows the LaMer diagram, only in this case the iron oleate decomposition products form through combination reactions and build up the active monomer concentration to the nucleation threshold. 49 Another observation from the kinetics experiments is that the growth rate of the iron oxide nanocrystals becomes extremely rapid directly after nucleation occurs. The nanocrystals are already 5.1 nm in diameter two minutes after nucleation has taken place (Figure 2-7a). In addition, no secondary nucleation is observed in the boiling synthesis, indicating an effective separation of the nucleation and growth stages during iron oxide nanocrystal formation. In contrast to the boiling synthesis kinetics study, the non-boiling synthesis displays completely different nanocrystal formation kinetics (Figure 2-7e-h). First, the nucleation of ironoxide nanocrystals took place at a later time than that in the boiling synthesis. No nanocrystals were observed until ~18 min at 300 C, and both the size- and shape-distributions of the newly formed particles were poor (Figure 2-7e). These results are significant in that they show bubbles can facilitate one of two processes: the formation of active monomers and/or the formation of nuclei from active monomers. 49 This phenomenon would suggest that at least one of these two 67

68 processes involves exothermic reactions. Second, the iron-oxide nanocrystals in the non-boiling synthesis displayed a faster growth rate than those in the boiling synthesis, and their shapes appeared as irregular polyhedrons (Figure 2-7f-h). Also observed in the non-boiling synthesis were the appearance and disappearance of smaller sized nanocrystals during the growth stage of the iron oxide, demonstrating that secondary nucleation and Ostwald ripening was taking place (Figure 2-7f-h). These results, along with the results from the non-boiling crystallization yield studies which show lower overall product yields when compared to boiling syntheses, further indicate that the non-boiling syntheses form lower numbers of iron oxide nuclei than the boiling syntheses. These results also support the idea that the formation of iron oxide nanocrystals follows the LaMer diagram. 49 When compared with the synthesis using boiling solvents, the non-boiling synthesis produced fewer iron-oxide nuclei, and thus the subsequent growth of these nuclei cannot consume an adequate amount of active monomers and/or the chemical species that can produce these active monomers in order to decrease the concentration of active monomers below the nucleation threshold. As a consequence, secondary nucleation took place in the nonboiling synthesis experiments (Figure 2-7e-h). To further investigate the mechanism of iron oxide nanocrystal formation, two more kinetics experiments were performed at 300 C using controlled Ar-bubble flow. In the first synthesis, the Ar bubbles flowed continuously during the entire reaction period, and this synthesis displayed nearly identical kinetics to the boiling synthesis (Figure 2-8a-d and Figure 2-7a-d). After 15 minutes, iron oxide nanocrystals with a 4.3 nm diameter and a size distribution of 9.3 % had appeared. During further growth, the size distribution and growth rate of the iron oxide nanocrystals decreased monotonically, and 6.7-nm nanocrystals were obtained after one 68

69 hour of reaction (Figure 2-8a-d). These results establish that the Ar bubbles can replace the boiling solvent bubbles in controlling the colloidal synthesis of nearly monodispersed iron oxide nanocrystals, advocating the idea that an exothermic reaction should be the rate limiting step in the formation of iron oxide nanocrystals. In the second synthesis, the Ar bubble flow was shut off after 15 minutes. At that instant, the iron oxide nanocrystals that had formed exhibited nearly identical size and size distribution when compared to the nanocrystals formed at the same time in the synthesis with constant Ar bubbling (Figure 2-8e-h). Yet, after only five more minutes of reaction without Ar bubble flow, a noticeable difference in the kinetics of nanocrystal formation had occurred. At this point, the mean size of the resulting nanocrystals (4.7 nm) was slightly smaller than those particles (5.6 nm) formed in the synthesis with continuous Ar-bubble flow, while their size distribution widened due to the presence of smaller-sized particles (3-4 nm). During the early reaction times (15-20 min), a large amount of unreacted precursors still exist in solution; therefore, the smallersized particles that are observed should have been formed via secondary nucleation and not via the dissolution of larger-sized particles through an Ostwald ripening process. As the reaction progressed, the iron oxide nanocrystals average size gradually grew, while small particles (3-4 nm) continuously formed and then grew. After reaction for one hour, the final nanocrystals exhibit a broad Gaussian-shaped size distribution centered at ~6.8 nm with a continuous tail down to 3 nm in diameter (Figure 2-8h). These observations clearly demonstrate that continuous secondary nucleation events took place after the Ar bubble flow was turned off, indicating that the primary nucleation of iron oxide nanocrystals was incomplete at 15 minutes of Ar bubbling. Likewise, the initial nucleation did not produce a sufficient amount of iron oxide nuclei whose further growth could bring the 69

70 active monomer concentration below the nucleation threshold, and therefore nearly continuous secondary nucleation was observed after the Ar bubbles were turned off (Figure 2-8f-h). Figure 2-8. Kinetics study of the synthesis of iron oxide nanocrystals under controlled Ar bubbling. TEM images of iron oxide nanocrystals taken from the synthesis in pure ODE at 300 C under Ar bubbling through the entire synthesis at (a) 15 min (4.3±0.4 nm in diameter), (b) 20 min (5.6±0.4 nm in diameter), (c) 30 min (6.4±0.4 nm in diameter), and (d) 60 min (6.7±0.4 nm in diameter). TEM images of iron oxide nanocrystals taken from the synthesis in pure ODE at 300 C with 15-min Ar bubbling at (e) 15 min (4.2±0.3 nm in diameter), (f) 20 min (4.7±0.7 nm in diameter), (g) 30 min (5.6±0.8 nm in diameter), and (h) 60 min (6.8±1.5 nm in diameter). The corresponding particle size distribution histograms are shown in the right side in these panels. All scale bars are 50 nm. Reprinted with permission from Lynch, J.; Zhuang, J. Q.; Wang, T.; LaMontagne, D.; Wu, H. M.; Cao, C. J. Am. Chem. Soc. 2011, 133,

71 To further study the kinetics of iron oxide nanocrystal nucleation, two additional iron oxide syntheses were carried out at 300 C in which the Ar bubble flow was turned off at 12-min or 18-min (Figure 2-9). Surprisingly, the synthesis with 18 minutes of Ar bubbling yielded monodispersed iron oxide nanocrystals with a 7.0 nm (Figure 2-9b) diameter, which is just slightly larger than those particles made in the synthesis under continuous Ar bubbling (Figure 2-8d). This result indicates that the reaction with 18 minutes of Ar bubbling produced a sufficient Figure 2-9. TEM images with histograms of iron oxide nanocrystals grown in pure ODE at 300 C for 60 minutes with (a) 12 minutes of Ar bubbling resulting in 17.9±2.9 nm diameters, and (b) 18 minutes of Ar bubbling resulting in 7.0±0.5 nm diameters. Scale bars are 50 nm. Reprinted with permission from Lynch, J.; Zhuang, J. Q.; Wang, T.; LaMontagne, D.; Wu, H. M.; Cao, C. J. Am. Chem. Soc. 2011, 133, amount of iron oxide nuclei whose growth was enough to prevent the active monomer concentration from increasing to the point where secondary nucleation would occur. The result also suggests that Ar bubbles have a larger effect on the nucleation of iron oxide nanocrystals than their growth, because the growth of iron oxide nanocrystals was not substantially affected in 71

72 the absence of Ar bubbles (Figure 2-9b). Significantly, these results show that the chemical reactions associated with the nucleation of iron oxide are inherently different than those reactions related to the growth of the nanocrystals. This difference might be the chemical basis for the efficient separation of nucleation and growth of iron oxide nanocrystals in solution-phase synthesis (vida infra). For the second synthesis with 12 minutes of Ar bubble flow, irregular shaped iron oxide nanocrystals that are similar to those made in the non-boiling synthesis were formed (Figure 2-9a and 2-1g), suggesting that these two syntheses had similar nanocrystal-formation kinetics. When considering both results, it is further supported that Ar bubbles play an important role in promoting iron oxide nanocrystal nucleation. Additionally, these results indicate that the separation of the nucleation and growth of iron oxide nanocrystals is under a delicate kinetic balance (unlike a permanent separation suggested by the original LaMer diagram). 49 The balance is easily thrown off when the growth of the initially nucleated nanocrystals cannot keep the active monomer concentration below the nucleation threshold. Therefore, there exists a prerequisite for an effective separation of these two stages: namely, that the primary nucleation should produce a critical (or sufficient) amount of nuclei whose subsequent growth can prevent secondary nucleation events. 2.4 General Discussion Nanocrystal Formation The formation of nanocrystals is a crystallization process that occurs on a nanometer scale. By definition, a crystal unit cell is the smallest possible crystal that can be formed, and thus it is understood that the nuclei of nanocrystals has a similar size to that of its unit cell. When looking at a magnetite (Fe 3 O 4 ) nanocrystal, the unit cell contains 62 oxygen atoms and 39 iron atoms with more than 100 bonds. 115 With this many atoms and bonds, the formation of a magnetite 72

73 unit cell should involve numerous, complex reaction steps that successively build on one another through simple monomer addition that is generated through iron oleate precursor decomposition. These addition reactions along with other types of reactions such as elimination, substitution, and/or rearrangement reactions should yield various types of interconvertible iron oxide clusters that can exhibit 1-, 2-, and 3-dimesional structures. It is important to understand that active monomers as used in these syntheses and discussion refer to the clusters involved in the reactions that directly yield nuclei. When the active monomer concentration reaches a specific threshold, nucleation takes place. While the molecular mechanism behind nanocrystal nucleation is not fully understood, the results discussed above indicate that the nucleation of iron oxide nanocrystals is strongly dependent on an exothermic reaction because this process can be promoted by gas bubbles through latent heat absorption. 157,158,161 Furthermore, these results demonstrate that gas bubbles have a greater effect on the nucleation of nanocrystals than on their growth (Figure 2-8d and 2-9b), which shows that the nucleation and growth of iron oxide nanocrystals may depend on different types of iron oxide clusters which undergo chemical reactions with different heat releasing capacities. The primary step in the formation of iron oxide nanocrystals is the decomposition of the iron oleate precursor. This process includes endothermic reactions which most likely include the formation of thermal free radicals (Figure 2-10). The next step in the iron oxide nanocrystal formation occurs when the primary free radicals undergo addition and/or substitution reactions with iron oleate complexes creating larger iron oxide clusters that can further decompose to form more free radicals. The combination of these iron oleate containing free radicals includes exothermic bond forming processes, and the inter- and intra-molecular combination of the free radicals can result in iron oxide clusters with 1-D, 2-D, or 3-D structures

74 The nucleation of nanocrystals is a phase transition process that creates crystalline nuclei from amorphous active monomer clusters. 28,29 The chemical reactions directly associated with the nucleation should largely depend on multiple-bond forming processes that can release an Figure Reaction scheme illustrating (a) the formation of thermal free radicals during the endothermic decomposition of iron oleate and the exothermic formation of primary iron-oxide clusters, (b) the formation of 1-D iron-oxide clusters via exothermic, single-bond forming reactions, and (c) the creation of 2-D iron-oxide clusters via highly exothermic, multiple-bond forming reactions. Reprinted with permission from Lynch, J.; Zhuang, J. Q.; Wang, T.; LaMontagne, D.; Wu, H. M.; Cao, C. J. Am. Chem. Soc. 2011, 133, extremely large amount of heat locally. As an example, the formation of three Fe-O bonds can release 1221 kj/mol. 164 The release of such a large amount of heat on a nanometer scale can promote the reverse reaction which would involve the thermal decomposition of the transition species (clusters) in the formation of nuclei and decrease the yield of the forward nucleation reaction. The act of solvent boiling or Ar bubbling can absorb this local heat through latent heat 74

75 transfer and this leads to a significant increase in the number of nanocrystal nuclei that forms during iron oxide nanocrystal nucleation. While nanocrystal nucleation is a phase transition process, nanocrystal growth is not and therefore most likely takes place on the nanocrystal surface. Single-bond forming reactions can lead to nanocrystals growth, so multiple-bond forming reactions are not necessary. Single-bond forming processes release less local heat than multiple-bond forming processes. This indicates that these single-bond forming processes can occur between iron-oxide clusters other than those that make up the active monomer. The formation free energies of the iron-oxide clusters involved in the nanocrystal growth are lower than those of the active monomers, and this indicates that the concentration of these iron-oxide clusters is higher than that of the active monomers in the reaction solution during an iron oxide nanocrystal synthesis. The growth of iron oxide nanocrystals should therefore involve these low heat releasing, single-bond forming reactions. This supports the observations that growth of nanocrystals is not strongly influenced by the gas bubble latent heat transfer occurring in the reaction solution (Figure 2-9b). The identification of exothermic reactions also supports the results from the iron oxide nanocrystal syntheses in boiling solvents at temperatures ranging from 290 to 365 C. These experiments resulted in iron oxide nanocrystals whose final size depended on the reaction temperature. The higher reaction temperatures lead to a larger final size of the iron oxide nanocrystals and vice versa (Figure 2-1). This would indicate that the higher reaction temperature syntheses lead to fewer nuclei because all of the reactions had the same amount of starting precursor. This observation shows that a higher reaction temperature hinders the nucleation of iron oxide nanocrystals. This temperature phenomenon further supports the hypothesis that the rate determining step in the nucleation of iron oxide nanocrystals is a 75

76 reversible (or a quasi-reversible) reaction in which the forward nucleation reaction is an exothermic process. 165,166 This result is also supported by Le Chatelier s principle, where a decreasing reaction temperature can cause such a reversible reaction to go toward the exothermic process. 165, Semi-quantitative Analysis The experimental results discussed thus far allow for a semi-quantitative analysis of the reaction enthalpy for the reversible reaction that leads to iron oxide nanocrystal nucleation. Three assumptions are needed to calculate the reaction enthalpy: (1) the equilibrium of the reversible reaction is achieved at the end of the initial nucleation stage, 37,38,167 (2) the concentration of the active monomers (the reactants) is at a nucleation threshold that is independent of reaction temperatures in the range of C, and (3) the crystallization yield is consistent in these boiling syntheses. With these three assumptions, the concentration of the nuclei formed during the initial burst of nucleation is represented by Equation 2-1, [ ] [ ] (2-4) where [N] denotes the nucleus concentration, [AM] is the concentration of active monomers, a represents the reaction order, and K is the equilibrium constant of the reaction. If the forward and backward reactions obey Arrhenius kinetics, K can be written as a function of the reaction temperature, the difference between the activation energies of the forward and backward reactions ( G = E a1 - E a-1 ), and the ratio of their pre-exponential factor (A = A 1 /A -1 ). By substituting these values in for K, Equation 2-4 becomes Equation 2-5. or (2-5) Based on the third assumption that the crystallization yields are consistent between the syntheses, the nucleus concentration can be described using Equation 2-6, 76

77 [ ] [ ] { ( ) } (2-6) where [P] stands for the initial precursor concentration, Φ is crystallization yield, d is the diameter of the nanocrystal product, D is the density of iron oxide, MW is the molar molecular weight of iron oxide, and F is the number of iron atoms in the molecular formula of iron oxide. In this calculation, the composition of the final iron oxide product is assumed to be constant in terms of the amount of magnetite and maghemite phases that are formed in these syntheses at the reaction temperatures used. Therefore, D and MW are constant in Equation 2-6. When Equations 2-4, 2-5, and 2-6 are combined they result in Equation 2-7, [ ] (2-7) where ( Φ[ ] ) and is constant in all the experiments. The concentration of the active monomers is above the nucleation threshold during the initial nucleation stage, and throughout this early time new active monomer is being formed and being consumed via further decomposition of the iron oleate precursor and nucleation, respectively. If the concentration of the active monomers changes only slightly during the initial nucleation stage, Equation 2-4 shows that the natural logarithm of nanocrystal diameter (d) is inversely proportional to reaction temperature if G is constant. When a plot is made relating ln(d) as a function of 1/T, a straight line should result. Using the diameters and reaction temperatures from Figure 2-1, a linear relationship is clearly seen in Figure The percentage variance in the linear fitting is 0.98 which further suggests that the data fits Equation 2-4 with a high degree of certainty. The slope of the linear fitting gives a reaction enthalpy of nucleation formation value of -142 ± 12 kj/mol. The highly negative heat of reaction 77

78 supports the observation that gas bubbles can significantly promote the nucleation of iron oxide nanocrystals. The result also demonstrates that the nucleation of iron oxide nanocrystals is determined by the chemical reaction kinetics. Figure (a) Reaction coordinate diagram describing the formation of iron-oxide nanocrystal nuclei from active monomers, and (b) a plot of the natural logarithm of nanocrystal diameter as a function of inverse reaction temperature in Kelvin. Reprinted with permission from Lynch, J.; Zhuang, J. Q.; Wang, T.; LaMontagne, D.; Wu, H. M.; Cao, C. J. Am. Chem. Soc. 2011, 133, Comparisons with Classical Nucleation Theory Classical nucleation theory predicts that the number of nuclei formed during nucleation can increase with an increase in nucleation temperature. 49,168 This prediction contradicts the temperature dependent nucleation effect observed throughout this work. It is understood that classical nucleation theory does not include the effects of chemical reaction kinetics occurring between the precursors and therefore does not accurately predict the observations shown in Figure Moreover, the temperature effect is also different from the recent observation by Peng et al. in the synthesis of InP nanocrystals, where they found the number of nuclei increased with an increase in reaction temperature with a fairly small activation energy of ~11 kj/mol. 39 This difference may arise from the difference in the two nanocrystal synthesis systems. 78

79 Taken together, the results from this study give valuable new insight into the separation of nucleation and growth of nanocrystals. There is a competition between the nucleation and growth of nanocrystals in which both processes consume the active monomers as well as the iron-oxide clusters that can form these active monomers. This results in the concentration of active monomers decreasing below the nucleation threshold and terminating the nucleation stage of nanocrystal formation. These observations suggest that the separation of nucleation and growth of iron oxide nanocrystals is under a delicate balance, which can be easily broken during the subsequent nanocrystal growth, leading to secondary nucleation (Figure 2-8h). Furthermore, these data demonstrate that a critical number of nuclei must form during the initial nucleation stage in order to effectively eliminate secondary nucleation from occurring in a closed synthesis system (Figure 2-9b). If the primary nucleus number is smaller than the critical number, than the initial nucleation can still be halted via the growth of these newly formed nanocrystals. But, since the growth rate of the newly formed nanocrystals is inversely proportional to their diameter, the consumption of the active monomers by these nanocrystals is insufficient to prevent the concentration of active monomers from reaching the nucleation threshold again, which results in secondary nucleation as observed in Figure 2-7e-h and 2-8f-h. These observations further support the conclusion that the control of the primary nucleation rate is critical in the synthesis of monodispersed colloidal nanocrystals. 2.5 Summary In summary, Chapter 2 reports a mechanistic study of colloidal iron oxide nanocrystal synthesis by comparing boiling versus non-boiling solvents as well as the effects of Ar-bubbling. It was shown that solvent boiling bubbles or Ar bubbles can significantly promote the initial nucleation of iron oxide nanocrystals and can therefore successfully suppress secondary nucleation from happening, yielding nearly monodispersed nanocrystals in the final product. It 79

80 was shown that the bubbling effects take place through local latent heat released from the exothermic reactions that occur during the nucleation and growth of the iron oxide nanocrystals. The types of chemical reactions occurring during nucleation may well be different from those involved in the growth of nanocrystals. The growth of the iron oxide nanocrystals may primarily depend on single-bond forming reactions that occur on the surface of the newly formed nanocrystals while nucleation more likely involves multiple-bond forming exothermic reactions that release larger amounts of heat locally in the reaction solution. The observation that exothermic reactions play a substantial role in the nucleation of iron oxide nanocrystals is further supported by our results from the synthesis of iron oxide nanocrystals with boiling solvents at temperatures ranging from 290 to 365 C. Based on the experimental data, the reaction enthalpy of the nucleation was calculated to be -142±12 kj/mol. The results also show the separation of nucleation and growth of iron oxide nanocrystals is under a delicate balance that can be easily broken during subsequent growth of nanocrystals. These results suggest that a prerequisite for effectively suppressing secondary nucleation is that the primary nucleation must produce a critical amount of nuclei, and this finding is important for a priori design of colloidal synthesis of monodispersed nanocrystals in general. 80

81 CHAPTER 3 THE DOPING OF CDS/ZNS CORE-SHELL NANOCRYSTALS WITH COPPER AND SILVER 3.1 Prologue Doped semiconductor nanocrystals are an emerging class of materials with important applications including wavelength tunable lasers, bioimaging, and solar cells. 169 Doping refers to the intentional inclusion of trace impurities into pure semiconductors. By adding impurities (dopants) into semiconducting NCs, the optical, magnetic, and electrical properties can be finetuned In one example, the doping of magnetic impurities into semiconductor NCs leads to the formation of dilute magnetic semiconductors (DMS). 174 Magnetic impurities can significantly increase the Zeeman splitting, creating splittings greater than 100 times that of normal semicondutors, allowing for potential applications in optical gating and spin-based electronics (i.e. spintronics). 171 In traditional bulk semiconductors, doping with impurities can change the number of charge carriers (electrons or holes) that flow through the material. On one hand, when the dopant has an extra valence electron compared to the host atom it is able to donate that electron creating an n-type semiconductor. On the other hand, when the dopant has one fewer valence electron it can add an extra hole to the semiconductor making a p-type semiconductor. 175 By combining both types of doped semiconductors, p-n junctions can be produced that are at the core of many devices used in the computer industry, such as microprocessors. 175 The introduction of dopants into semiconducting NCs (quantum dots) has attracted great interest due to the possibility of discovering new properties not seen in the corresponding bulk material or undoped quantum dots (QDs). Yet, there exists difficult synthetic challenges to doping QDs. For bulk CdS and CdSe II-VI semiconductors, Mn atoms have an appreciable solubility (greater than 10 %), but when in nanocrystalline form, the same Mn impurities are not 81

82 easily incorporated into the materials using the standard conditions at either high or low temperatures. 123,176,177 but not inside. 123,176,177 In several cases, the dopant atoms survive only at the surface of the NCs By only being able to attach the dopant onto the surface of the semiconducting NCs rather than introducing it into the inner core, the effects of the impurity are minimized or nonexistent. The difficulty in doping QD materials has various explanations, including the high surface-to-volume ratio of CINs, the noticeable difference between the surface and inner core of the materials, and the size and charge differences between the impurity atoms and the host material. 174,176,178,179 One method used to overcome the inherent difficulties in doping QDs is called the isocrystalline core/shell growth method, developed by Gamelin et al. 178 This method attaches impurity atoms to the surface of host QDs and then isolates and purifies these NCs before growing additional isocrystalline layers to trap the dopant inside the semiconducting material. 143 This procedure has also been applied to systems with differing core and shell materials, e.g heterocrystalline systems Our research group has synthesized CdS/ZnS core-shell NCs with controlled position doping of Mn atoms. 68,186 By using successive ZnS shell layer growth, the position of the Mn impurities can be tailored along with the phosphorescence quantum yield from the Mn emission. This system has increased the quantum yield of the Mn emission up to 44 % at a doping amount of 0.36 %. 68 Chapter 3 uses synthetic techniques developed in our laboratory for Mn doping to explore the effect of Cu and Ag dopants on CdS/ZnS core/shell NCs. Using CdS/ZnS NCs as the starting material, we found that the presence of Cu dopants facilitates the alloying of the core-shell system into CdZnS NCs. The alloying process is accompanied by no observable change in the NC s size or size distribution as well as a constant final doping concentration of Cu. This 82

83 alloying effect is correlated to the measured blue shift in the first excitonic absorption peak of the NCs in solution and is used to estimate the amount of alloying that occurs for other sized CdS cores. Furthermore, the kinetics of this alloying process is investigated using Jander analysis to determine the activation energy associated with the observed alloying effect. It was determined that two kinetic processes control the Cu doping process: 1) the surface adsorption of the dopant onto the ZnS shell and 2) the tunneling and stirring effect of the Cu dopant once inside the inner core of the CdS/ZnS NCs. In a separate case, Ag dopants are added at low reaction temperatures and shown to have a position-dependent effect on the fluorescence quantum yield of the CdS/ZnS NCs. It was observed that during surface attachment and early tunneling that the emission intensity decreases with increased time and temperature, while after full incorporation of the Ag impurity into the inner core of the CdS/ZnS NCs a measured increase back to the original intensity of the fluorescence was observed. 3.2 Experimental Section Chemicals Sulfur powder ( %), 1-octadecence (ODE, tech. 90 %), oleylamine (OAm, tech. 70 %), copper (II) acetate (98 %), and silver (I) acetate (99.99 %) were purchased from Aldrich. All solvents were purchased from Fisher Scientific Company. Nitric acid ( 69.5 %, TraceSELECT) was purchased from Fluka. Cadmium nitrate tetrahydrate (Cd(NO 3 ) 2 4H 2 O, %), zinc stearate (count as ZnO 14 %) and myristic acid (MA, 99 %) were purchased from Alfa Aesar. The chemicals were used as received without further purification. 83

84 3.2.2 Three Step Synthesis of Cu- or Ag-Doped CdS/ZnS Core/Shell Nanocrystals Preparation of precursors Cu(OAc) 2 solution. Oleylamine (4 ml) was added to a 25 ml flask and heated at 120 C for 10 min under a vacuum of 20 mtorr. After the OAm solvent was cooled to room temperature, Cu(OAc) 2 (10 mg, 55.5 μmole) was quickly added to the flask. The mixture was degassed at room temperature and at 120 C for 10 min for each step. After a clear solution was obtained, the solution was cooled to room temperature and was ready for use. Note that the Cu precursor solution should be freshly made before the synthesis. Ag(OAc) solution. Oleylamine (4 ml) was added to a 25 ml flask and heated at 120 C for 10 min under a vacuum of 20 mtorr. After the OAm solvent was cooled to room temperature, Ag(OAc) (10 mg, 60 μmole) was quickly added to the flask. The mixture was degassed at room temperature and at 50 C for 10 min for each step. After a clear solution was obtained, the solution was cooled to room temperature and was ready for use. Note that the Ag precursor solution should be freshly made immediately before the synthesis. Sulfur solution. Sulfur powder (12.8 mg, 0.4 mmol) was added to a flask with ODE (10 ml). After degassing at room temperature for 10 min, the solution was heated to 130 C under Ar flow. The temperature was maintained for 5 min, and then the resulting sulfur solution was cooled to room temperature for use. Caution should be taken to avoid heating the solution to a higher temperature. Zinc stearate solution. Zinc stearate powder (0.4 mmol) was added to a flask with ODE (10 ml). After degassing at room temperature for 10 min, the mixture was heated to 200 C to dissolve zinc stearate. The solution was cooled to room temperature and a slurry formed. The slurry was directly used for ZnS shell growth. 84

85 Synthesis of Cu- or Ag-doped CdS/ZnS core/shell nanocrystals (1) Synthesis of starting host particles CdS nanocrystals. The synthesis of CdS nanocrystals was based on a modification of a literature method. In a typical synthesis, cadmium myristate (1.0 mmol) and sulfur (0.5 mmol) were loaded into a three-neck flask with 1-octadecene (ODE, 50 g). After degassing under vacuum (~20 mtorr) for 10 min, the vacuum was removed. Under argon flow, the temperature was raised to 240 C. The growth was monitored by taking the absorption spectra of aliquots extracted from the reaction solution. After reaching the desired size, the reaction mixture was allowed to cool to room temperature and the nanocrystals were precipitated with acetone and redispersed in hexane. The as-prepared CdS crystals have a zinc-blende crystal structure. CdS/ZnS core/shell nanocrystals. First, zinc-blend CdS nanocrystals were prepared as described above. Second, ZnS shells were grown onto the resulting CdS nanocrystals at 220 C in a solvent mixture of ODE and OAm with a volume ratio of 3:1. ZnS shells were added to the reaction solution with CdS nanocrystals by alternate, dropwise injections of a solution of zinc stearate in ODE (40 mm) and sulfur in ODE (40 mm). Growth time was 10 min after each injection. When the desired shell thickness was achieved, the growth solution was cooled to room temperature. The resulting CdS/ZnS core/shell nanocrystals were precipitated by acetone, and redispersed in hexane. (2) Addition of dopant Cu doping. In a typical experiment, a hexane solution of host particles (2 ml, 50.4 nmol) was added to a solvent mixture of ODE and OAm (8.0 ml, ODE/OAm: 3:1), and then hexane was removed under vacuum. Under argon flow, the nanocrystal solution (CdS/ZnS core/shell nanocrystals as starting host particles) was heated to the growth temperature (160 to 280 C) and then the Cu(OAc) 2 solution and the sulfur solution at a molar ratio of 1:10 were introduced into 85

86 the hot solution by direct injection. The doping progress was monitored by taking the absorption spectra of 50 μl aliquots extracted from the reaction solution. After the absorption spectra peak ceased to shift, the synthesis was stopped, and the nanocrystals were precipitated by adding acetone and were redispersed in hexane as a high-concentration solution for further analysis. Ag doping. In a typical experiment, a hexane solution of host particles (2 ml, 50.4 nmol) was added to a solvent mixture of ODE and OAm (8.0 ml, ODE/OAm: 3:1), and then hexane was removed under vacuum. Under argon flow, the Ag(OAc) solution and the sulfur solution at a molar ratio of 1:10 were introduced into the 50 C nanocrystal solution (CdS/ZnS core/shell nanocrystals as starting host particles) by direct injection. The reaction solution was heated at a rate of 10 C/min in 50 C intervals (50 to 300 C) with 10 min growth times at each interval until the final growth temperature of 300 C. The doping progress was monitored by taking the absorption and fluorescence spectra of aliquots extracted from the reaction solution. After the measured change in the fluorescence ceased, the synthesis was stopped, and the nanocrystals were precipitated by adding acetone and were redispersed in hexane as a high-concentration solution for further analysis Characterization of Cu- or Ag-Doped CdS/ZnS Core/Shell Nanocrystals Absorption measurements UV-Vis absorption spectra were measured using a Shimadzu UV1800. Nanocrystals were dissolved in toluene for the measurement Photoluminescence measurements Nanocrystals were dissolved in toluene for the measurement. Photoluminescence (PL) experiments were performed on a fluorometer (Fluorolog-3, Horiba Jobin-Yvon, Irvine, CA). 86

87 TEM measurements TEM measurements were performed on a JEOL 200X operated at 200 kv, or a JEOL 2010F TEM operated at 200 kv. The specimens were prepared as follows: a nanoparticle solution (10 μl) was dropped onto a 200-mesh copper grid and was dried overnight under ambient conditions X-ray powder diffraction (XRD) measurements XRD measurements were performed on a Philips XRD 3720 spectrometer. The specimens were prepared as follows: about 15 mg of the purified nanocrystals were dissolved in about 0.5 ml of toluene and then dropped onto a low-scattering quartz sample-holder. The sample was dried in air and kept overnight in a vacuum desiccator Inductively-coupled plasma (ICP) atomic emission spectroscopy measurements The ICP measurements were performed on a Vista RL CCD Simultaneous ICP-AES (Varian, Inc.). The purified nanocrystal samples were digested with nitric acid (69.5 %). The digestion was performed at about 100 C until the solution became colorless. The digestion solutions were further diluted with a nitric acid solution to obtain a final nitric acid concentration of about 1 2 %. The concentrations of Cu, Ag, Cd, and Zn in solutions were determined by data from ICP measurements as compared with the corresponding working calibration curves. The Cu or Ag doping levels were defined as: [ ] [ ] [ ] [ ] (3-1) 3.3 Results and Discussion Cu-Doping of Thick Shelled (6 ML) CdS/ZnS Core/Shell Nanocrystals Following previously reported synthetic schemes developed in our group, 68,186 CdS/ZnS core/shell NCs were synthesized using a three-step method (Figure 3-1). The overall process 87

88 Figure 3-1. Scheme of three-step synthesis for synthesizing Cu-doped CdS/ZnS core/shell nanocrystals. (1) Synthesis of host particles, (2) ZnS shell growth, and (3) Cu-dopant growth. starts with the synthesis of CdS core particles with an approximate diameter 4.1 nm as determined from their extinction coefficients and TEM measurements. Next, the CdS core NCs are overgrown with a ZnS shell whose thickness can be controlled by the number of sequential Zn and S additions that are allowed to react. Initially, six monolayers (MLs) of ZnS shell were grown onto the CdS core with a first excitonic absorption peak measured near 420 nm. Once synthesized, the core/shell NCs were isolated, purified of excess Zn and S reagents, and prepared for doping with copper ions. The amount of Cu grown on was directly related to the amount of Cu and S growth precursors (Cu(OAc) 2 and S) injected into the reaction solution. Upon direct injection of 48 Cu atoms per NC, a marked blue shift in the absorption spectra (Figure 3-2A) of the CdS/ZnS core/shell NCs was measured, and the photoluminescence was completely quenched. TEM analysis showed no change in the average diameter (6.3 nm) or the size distribution (6.2 %) of the resulting Cu-doped CdS/ZnS core/shell NCs (Figure 3-2C,D). With no change in the average size or size distribution of the core/shell NCs being observed, Ostwald ripening could be ruled out as the mechanism explaining the measured blue shift in the absorbance spectra of the Cu-doped CdS/ZnS NCs. ICP-AES measurements at different times during Cu growth indicate a near constant 2 % atomic concentration of Cu inside the core/shell NCs (Figure 3-2B). 88

89 Figure 3-2. (A) Absorption spectra of CdS/ZnS core/shell nanocrystals with 6 ML ZnS after addition of 48 Cu atoms per NC with increasing reaction time at 220 C. (B) Copper doping amounts measured from ICP-AES at different reaction times. (C) TEM of CdS/ZnS core/shell nanocrystals with 6 ML ZnS before addition of Cu atoms. (D) TEM of CdS/ZnS core/shell nanocrystals with 6 ML ZnS after addition of Cu atoms. This data indicates that not only is the doping of Cu inside CdS/ZnS successful but that after only two minutes the amount of Cu that is grown onto the NCs remains the same from that time forward. Therefore, Cu growth and diffusion are both rapid processes that do not change the size or shape of the resulting particles. Furthermore, since the amount of copper remains constant after two hours of growth, any processes that might eject the Cu dopants from the core/shell particles are not significant in Cu-doping of CdS/ZnS NCs. It has been reported that structural rearrangement of CdSe NCs after incorporation of Zn and Se precursors at high temperatures can result in changes to the absorption of the now alloyed Zn x Cd 1-x Se NCs. 187 Similar results have been reported for alloyed Zn x Cd 1-x S NCs at high reaction temperatures. 188 In both cases, changes in the absorption were measured as blue shifts in the first excitonic peak of the semiconductor NCs. X-ray diffraction is a strong tool that can be used to determine 89

90 whether alloying of a system has occurred by measuring shifts in the diffraction peaks relative to a non-alloyed sample. Having ruled out ripening processes or changes in Cu concentrations as being responsible for the observed change in the absorption spectra, XRD measurements were made on the 6 ML CdS/ZnS core/shell NCs to determine whether shifts in the diffraction peaks were observed for the Cu-doped particles. It was quickly realized that due to the large thickness of the ZnS shell that the XRD peaks measured were dominated by the cubic ZnS 111, 220, and 311 peaks (data not shown) Cu-Doping of 1.6 ML CdS/ZnS Core/Shell Nanocrystals To overcome the difficulty in analysizing the XRD spectra of the thick-shelled Cu-doped CdS/ZnS NCs, we devised a reaction scheme where during the ZnS shell growth stage we added only enough Zn and S precursors to grow 1.6 ML of ZnS shell onto the 4.1 nm CdS core NCs. With only a thin ZnS shell, Cu-doping proceeded as in the case with 6 ML CdS/ZnS NCs. Similar to previous results, after we injected Cu and S growth precursors at 220 C, a marked blue shift in the absorption spectra of the 1.6 ML CdS/ZnS core/shell NCs was observed (Figure 3-3A). This process occurred at an increased rate compared to the 6 ML CdS/ZnS core/shell NCs presumably because of the thinner ZnS shell through which the Cu ions would have to diffuse. To keep both reactions comparable, 48 Cu atoms per NC were added to the 1.6 ML CdS/ZnS NCs to match the number added to the 6 ML CdS/ZnS particles. Using ICP-AES analysis, the atomic percentage of Cu successfully added into the thin-shelled NCs was measured at 3.2 %, nearly 1.5 times the amount found in the thick-shelled NCs. This indicates that the thickness of the ZnS shell acts as a barrier to Cu diffusion into the core/shell NCs. 90

91 Figure 3-3. (A) Absorption spectra of CdS/ZnS core/shell nanocrystals with 1.6 ML ZnS after addition of 48 Cu atoms per NC with increasing reaction time at 220 C. (B) Copper doping amounts measured from ICP-AES at different reaction times. (C) TEM of CdS/ZnS core/shell nanocrystals with 1.6 ML ZnS before addition of Cu atoms. (D) TEM of CdS/ZnS core/shell nanocrystals with 1.6 ML ZnS after addition of Cu atoms. TEM measurements confirm that the average diameter (4.5 nm) and size distribution (7.4 %) do not change significantly after growth of the copper dopant onto the 1.6 ML CdS/ZnS core/shell NCs (Figure 3-3C, D). Observing similar trends for both thick and thin shelled CdS/ZnS NCs, a control experiment was run to verify copper s responsibility for the changes to the absorption spectra of the core/shell NCs. In the control experiment, no copper was added to 1.6 ML CdS/ZnS NCs at 220 C and the reaction time was set to over 195 minutes (Figure 3-4). Without the addition of the Cu growth precursors no changes in the absorption spectra were observed after three hours reaction at 220 C. 91

92 Figure 3-4. Absorption spectra of CdS/ZnS core/shell nanocrystals with 1.6 ML ZnS without addition of 48 Cu atoms per NC with increasing reaction time at 220 C. Red line emphasizes no measured shift in first absorption peak of samples. The control experiment verifies that Cu caused the changes to the absorption properties of the CdS/ZnS core/shell NCs. The dependency of the change in the absorption spectra on the reaction temperature and amount of Cu added was also determined (Figure 3-5A, B). In both situations, a direct relationship was uncovered between the rate of blue shift versus either the reaction temperature or the Cu doping amount. These observations indicate that a kinetic process is occurring when Cu is doped into CdS/ZnS core/shell NCs. It has been proposed by Norris et al. that doping of semiconductor NCs is a kinetically driven process where the incorporation of dopants is governed by dopant-growth kinetics, a model known as the trapped dopant model. 189 XRD measurements of the Cu-doped thin-shelled CdS/ZnS NCs at different reaction times clearly shows the CdS core peaks corresponding to the 111, 220, and 311 crystal planes. As the 92

93 Figure 3-5. (A) Change in absorption spectra (blue shift) vs. time at different reaction temperatures for 48 Cu atoms per 1.6 ML CdS/ZnS NCs; squares: 200 C, circles: 220 C, diamonds: 240 C, triangles: 260 C, stars: 280 C. (B) Change in absorption spectra (blue shift) vs. time for different Cu amounts at 220 C for 1.6 ML CdS/ZnS NCs; squares: 24 Cu atoms, circles: 48 Cu atoms, triangles: 60 Cu atoms. (C) Shift in the 111 XRD peak for 48 Cu per 1.6 ML CdS/ZnS NCs at different reaction times. (D) Percent alloying vs. 111 XRD peak shift calculated using Vegard s Law. reaction time increased, all of the XRD peaks shifted to larger 2θ values (Figure 3-5C). This is indicative of an alloying process occurring where the CdS core mixes with the ZnS shell to form a Zn x Cd 1-x S nanocrystal. Since CdS and ZnS have different lattice parameters, 0.58 and 0.54 nm, respectively, when they alloy the resulting crystal has a lattice parameter that has a value between that of pure CdS and pure ZnS. When looking at CdS XRD peaks, as alloying proceeds the lattice parameter will increase relative to pure CdS. An increase in the lattice parameter is measured as a higher 2θ value in the XRD peak positions. It has been previously reported that Cu 2+ ions can assist in the alloying of Au-Pd NCs. 190 The amount of alloying can be calculated 93

94 using Vegard s Law coupled with ICP-AES analysis of the concentrations of Zn and Cd in the alloyed nanocrystals. Using such analysis, the percentage of solid-solution formation was calculated and plotted versus the shift in the 111 XRD peak. A near-linear relationship was observed, further indicating a process that proceeds once the Cu has been doped into the core/shell NCs. The process of using the XRD peak position to determine the extent of alloying is time consuming, requiring careful analysis of the peaks. Since the measured XRD shift was on the same time scale as the measured blue shift in the absorption spectra, attempts were made Figure 3-6. Plot of percent solid-solution formation (alloying) vs. measured change in absorption spectra (blue shift) at 220 C for 48 Cu atoms per 1.6 ML CdS/ZnS core/shell NC. Red line plots linear regression of data points. to correlate the change in the absorption spectra to the extent of alloying. This allows for quick, accurate determinations of the amount of alloying taking place without the need to directly measure the change in the XRD peak positions. Using the blue shift that was measured at 220 C along with the alloying amounts calculated from the change in the XRD peak positions, a near-linear relationship was discovered (Figure 3-6). While not perfectly linear, the regression line between the percent alloying and measured blue shift in the absorption spectra may be used as a predictive tool to estimate the extent that the CdS core mixes with the ZnS shell. When 94

95 considering time and cost restraints, such a tool is a welcomed addition for semiconductor nanocrystal synthesis Jander Analysis of Cu-doped CdS/ZnS Core/Shell Nanocrystals CdS/ZnS core/shell with 4.1 nm CdS core Park et al. was able to determine an activation energy for the alloying of CdSe/ZnSe core/shell NCs 191 by using Jander analysis 192 along with the Arrhenius equation. By investigating the kinetics of the alloying process, they concluded that the diffusion of Zn 2+ ions via the dissociation of ZnSe bonds was the active mechanism for the alloying process. It was our desire to run a similar analysis to help elucidate the mechanism for alloying of the CdS/ZnS core/shell NCs when Cu 2+ ions are present. Alloying reactions were run using the same batch of 1.6 ML CdS/ZnS NCs, each having 48 Cu atoms added per NC. The reaction temperature was varied between 200 and 280 C. For each reaction, the extent of alloying was estimated using the regression equations from the percent solid-solution formation versus blue shift data (Figure 3-6). We used Jander analysis, based on parabolic kinetics, to investigate the rates of the alloying process. 192 Jander analysis is based on three-dimensional diffusion through a spherical particle. It assumes a planar surface relative to the movement of constituents into and out of the solid particle. 193,194 By applying Jander analysis to each of the alloying reactions at five different temperatures, the reaction rate could be estimated based on linear fitting to the Jander data using Equation 3-2 for the analysis. [ ( ) ] (3-2) Here, x is the degree of solid-solution formation, k is the reaction rate constant, and t is the reaction time after injection of the Cu dopant. By extracting the slope of the linear fitting to the 95

96 data (Figure 3-7) for each reaction temperature between 200 to 280 C, it becomes possible to extract an activation energy for the alloying process induced from Cu doping. Using Jander analysis for the alloying process induced via Cu doping of a 4.1 nm CdS core with 1.6 ML of ZnS shell led to a highly linear relationship for reaction temperatures at 240, 260, and 280 C. Yet, for the lower reaction temperatures of 200 and 220 C the data did not produce a linear relationship between the extent of solid-solution formation and the reaction time after Cu addition. Initial attempts to explain this phenomenon relied on applying different models of solid state reaction kinetics in hopes that a different theoretical model would lead to better fittings for all the reaction temperatures used. We used several different models, but no one framework allowed for the extraction of the reaction rates with a high degree of accuracy at all the reaction temperatures. The decision was made to determine if there were different physical processes occurring during the doping and alloying of the CdS/ZnS particles at lower reaction temperatures as compared to higher temperatures CdS/ZnS nanocrystals with a 3.4 and 4.9 nm CdS core Initial designs of experiments to investigate the role that the lower and higher reactions temperatures played on the physical processes involved in the Cu doping of CdS/ZnS core/shell NCs with a 4.1 nm diameter core lead to realizations that, for this system, either alloying occurred to slowly or too quickly to collect data points in an efficient manner. Therefore, the decision was made to investigate how the CdS core size would affect the alloying process. It was hypothesized that due to the smaller core size, lower reaction temperatures would be sufficient to induce the alloying process in a time scale that was conducive for collecting data points in an effective manner. Efforts were made to keep the reaction variables constant so that fair comparisons could be made between the doping and alloying processes for each size core. We chose two CdS core sizes, 3.4 and 4.9 nm, with first 96

97 Figure 3-7. Graphs of Jander analysis of degree of solid-solution formation versus reaction time after Cu addition for 48 Cu atoms per 1.6 ML CdS/ZnS core/shell NCs at different alloying temperatures. (A) 200 C, regime 1, early reaction times, (B) 200 C, regime 2, late reaction times, (C) 220 C, regime 1, early reaction times, (D) 220 C, regime 2, late reaction times, (E) 240 C, (F) 260 C, (G) 280 C, (H) schematic of doping process leading to formation of alloyed Zn x Cd 1-x S nanocrystals. 97

98 excitonic peaks centered at 400 nm and 440 nm, respectively. To keep the atomic ratio between the Cd and Zn nearly constant, the ZnS shell grown onto the 3.4 nm CdS core was slightly thinner (~1.4 ML) while for the 4.9 nm CdS core it was slightly thicker (~2.0 ML) when compared to the 4.1 nm core used in the original experiments. Furthermore, from previous data (Figure 3-5B) we chose Cu doping amounts that would allow for alloying to occur on similar time scales as for the 4.1 nm CdS core. In the cases of the 3.4 and 4.9 nm diameter CdS cores, we determined that 27 and 60 Cu atoms per core/shell NC, respectively, would be appropriate. Reaction temperatures were chosen that would investigate the lower and higher temperatures used in all the alloying experiments. For the smaller sized CdS core, reaction temperatures ranging from 160 to 215 C were chosen, while for the larger sized core, temperatures from 220 to 260 C were used. Figures 3-8 and 3-9 show the data points along with linear regressions that correspond to the 3.4 nm diameter CdS cores and the 4.9 nm diameter CdS cores, respectively. For the smaller sized 3.4 nm CdS core with 1.4 ML of ZnS shell, the Jander analysis produces highly linear fits for alloying temperatures of 160 and 180 C. Upon closer examination of the 170, 205 and 215 C plots we quickly realized that, similar to the Cu-doped 1.6 ML CdS/ZnS core/shell NCs alloyed at 200 and 220 C, the Jander analysis did not produce linear regressions with a highly linear relationship. For the Cu-doped 2.0 ML CdS/ZnS core/shell NCs with the 4.9 nm diameter CdS, all of the reaction temperatures used gave linear fittings with R 2 values all > It is worth noting though, that the 220 C alloying temperature did give the lowest R 2 value of all the linear regressions for the 4.9 nm diameter CdS core. By considering all of the Jander analysis plots 98

99 Figure 3-8. Graphs of Jander analysis of degree of solid-solution formation versus reaction time after Cu addition for 27 Cu atoms per 1.4 ML CdS/ZnS core/shell NCs at different alloying temperatures. (A) Schematic of doping process leading to formation of alloyed Zn x Cd 1-x S nanocrystals. (B) 160 C, (C) 170 C, (D) 180 C, (E) 205 C, regime 1, early reaction times, (F) 205 C, regime 2, late reaction times, (G) 215 C, regime 1, early reaction times, (H) 215 C, regime 2, late reaction times. for each sized CdS core, it was concluded that for most of the alloying temperatures used in all of the experiments Jander analysis as a theoretical model to extract reaction rates was a valid 99

100 Figure 3-9. Graphs of Jander analysis of degree of solid-solution formation versus reaction time after Cu addition for 60 Cu atoms per 2.0 ML CdS/ZnS core/shell NCs at different alloying temperatures. (A) Schematic of doping process leading to formation of alloyed Zn x Cd 1-x S nanocrystals. (B) 220 C, (C) 230 C, (D) 240 C, (E) 250 C, (F) 260 C. approximation. But, for the reaction temperatures specified above, Jander analysis broke down as an accurate theoretical model for the alloying process. 100

101 Non linearity of Jander plots It should be noted that while linear regressions were used for all Jander plots, as mentioned above, there are clear cases where Jander analysis did not produce linear fittings. It is quite feasible that Jander analysis as a model for the processes occurring in the Cu doped CdS/ZnS system being studied do not accurately describe the physical mechanism behind the alloying observed. Likewise, experimental error might also explain the deviation from linearity although this seems less likely due to tight controls over all reaction variables such as temperature, doping procedure, and nanocrystals used from the same batch synthesis in all Jander experiments. Clearly, further study is required for a complete explanation of the observed phenomenon in the Jander experiments. For the time being, slopes extracted from the linear regressions were used for the activation energy determination, but with an understanding that the actual importance and validity of such calculations needs more research to be verified as true for the system being studied. The following section attempts to set a framework for the physical processes that must be occurring for any doping experiments in hope that this underlying theory will allow for the rational design of future experiments which will delve further into developing an accurate model for the solid solution formation mechanism that is occurring Physical Processes Associated with Alloying of Cu-Doped CdS/ZnS Nanocrystals at Temperatures between 200 to 220 C When doping CdS/ZnS core/shell NCs, two processes must take place for the incorporation of the Cu 2+ ions into the core/shell nanocrystals. First, the Cu 2+ must adsorb onto the NCs surface. This process is highly dynamic with an equilibrium being established rapidly but resulting in dopant growth yields less than unity. This has been seen in other successful semiconductors nanocrystal doping experiments. In our research group, Mn-doped CdS/ZnS NCs were synthesized with a 17 % growth yield for Mn. The amount of Mn was affected by the 101

102 growth temperature, the strength of the Mn binding to the core/shell particle s surface, the location of the dopant on the ZnS shell relative to the core, and the chemical reactivity of the Mn precursor. 186,195 In another example, Chelikowsky et al. argued that self-purification is an intrinsic property of defects in semiconductor NCs. 196 This argument leads to a size dependence on the likelihood of successful doping of semiconductor NCs with the smaller the NC the more difficult the doping. Taken together, this information suggests that one major process involved in the doping of semiconductor NCs is the adsorption of the doping ion onto the surface on the NC. The second process that is crucial for successful doping of semiconductor NCs is the diffusion of the dopant atom into the interior of the nanocrystal. In our system, where alloying is an overall result of the doping process, we argue that migration of the Cu 2+ ion leads to a mixing effect between the CdS core and ZnS shell layers of the core/shell system. It is demonstrated that with no copper addition the alloying process does not occur (Figure 3-4). These data support the idea that Cu atoms act as mixers to induce alloying between the core and shell. One might counter that Cu could induce the observed change (blue-shift) in the absorption spectra of the semiconductor NCs without the ZnS shell layer and hence no alloying is needed to measure a change in the absorption spectra. With this in mind, a control experiment was run at 160 C with a 3.4 nm CdS solution with no ZnS shell grown onto the core. With no shell to passivate the CdS core, special care was taken to prevent ripening of the particles, because ripening would obscure the ability to observe the blue shift if indeed it occurred. To account for this, a change in the doping procedure from direct injection of the Cu growth precursor to dropwise addition of the growth solution was enacted. Keeping as many variables consistent as 102

103 possible, the doping amount was set to 27 Cu atoms per 3.4 nm CdS core. The reaction temperature was set to 160 C and the reaction was held at that temperature for 200 minutes a sufficient duration to observe the change in absorption if it was induced. Figure 3-10 shows the absorption spectra of aliquots taken after the dropwise addition of the dopant growth solution at various time points. The spectra show no measureable blue shift in the first excitonic absorption peak of the 3.4 nm diameter CdS core. Indeed, no significant change in the absorption profile is measured. This phenomenon is typical of purified semiconductor NCs that are heated for extended periods at high reaction temperatures in the absence of ligand, as was seen for CdS NCs. 34 The control experiment (Figure 3-10) supports the argument that Cu acts to induce alloying between the CdS core and the ZnS shell in the CdS/ ZnS core/shell NCs. When Cu is added to only CdS core particles there is no measured blue shift in the absorption spectra indicating that alloying is contingent on there being both a core and a shell material. In a successful doping synthesis of semiconductor nanocrystals, both surface adsorption and dopant diffusion, occur simultaneously. Likewise, both processes should have an associated energy barrier that must be overcome for them to occur spontaneously. By examining the Jander analysis regressions for the 3.4 and 4.1 nm diameter CdS cores at the reaction temperatures where the linear regression gave nonideal fittings, specific kinetic regimes were set to represent either the surface adsorption process for Cu doping or the tunneling/ mixing process. The early time regimes represent the surface adsorption process. This makes logical sense because surface adsorption must occur first before tunneling can proceed. The surface adsorption process has an energy barrier that depends on the reaction temperature used and the diameter of the CdS core being doped. For the 3.4 nm diameter CdS core, the Jander analysis produces two kinetic regimes at the high end of the reaction 103

104 temperatures used in this set of experiments (Figure 3-8E-H). For the 4.1 nm diameter CdS core, the Jander analysis produces two kinetic regimes at the lower end of the reaction temperatures used for that set of experiments (Figure 3-7A-D). This observation should mean that for the smaller sized, 3.4 nm diameter CdS core at the lower temperature range (160 and 180 C), one process is kinetically favored over the other. It is hypothesized that at these low temperatures the surface adsorption of the Cu dopants dominates the kinetics of the alloying process. There is insufficient energy in the form of heat for the tunneling process to occur rapidly at these reaction temperatures for the 3.4 nm diameter CdS core. But, once the reaction temperature is increased to between 200 and 220 C both process Figure Absorption spectra of 3.4 nm diameter CdS core with 27 Cu atoms per core added in dropwise. Red line emphasizes red-shift in first absorption peak of samples. 104

105 have sufficient energy to overcome their respective energy barriers and proceed. Hence, a situation is observed where both processes show up in the Jander analysis for the CdS/ZnS core/shell NCs with a 3.4 nm diameter core. For the 4.1 nm diameter CdS core, we fit highly linear regressions for reaction temperatures from 240 to 280 C. At these high reaction temperatures, surface adsorption should be near instantaneous and therefore the tunneling process is the kinetically dominate one. Similar to the smaller sized core, at the lower reaction temperatures, both processes become evident in the Jander analysis and therefore two kinetic regimes show up in linear regressions. With two kinetic regimes from which to choose for the further kinetic analysis to determine activation energies, we decided to use regime 2 (later times in Jander analysis) for the 3.4 nm CdS core and regime 1 (early reaction times in the Jander analysis) for the 4.1 nm CdS cores, both with reaction temperatures between 200 and 220 C. By choosing these regimes, we kept the observed trend in the reaction rates constant (Table A-1, 2). We calculated the reaction rates from the slopes of the linear regressions, and they were consistent with those set forth by the rest of the experiments for the 3.4 and 4.1 nm diameter core CdS/ZnS core/shell particles. In other words, we assume that as temperature increases, the reaction rate calculated from the slope of the linear regressions should also increase Determination of Alloying Activation Energy Jander analysis of solid-state solution formation allows for the reaction rates (k) of the alloying process to be estimated at different temperatures. By using Arrhenius-type analysis (Equation 3-3) and the values of k we were able to estimate activation energies for the formation of Zn x Cd 1-x S alloyed NCs for each size CdS core. 105

106 (3-3) Here, k o is a constant, R is the gas constant in J/mol K, Q is the activation energy for solidsolution formation, and T is the heat treatment temperature. Figure 3-11 shows the Arrhenius plots for each size CdS core along with the linear regression fit to the data. All three Arrhenius plots show a high degree of linearity and were used to estimate the activation energies for the alloying reaction for each size CdS core. From the slope of the Arrenhius plots we calculated activation energies of 89 ± 20 kj/mol, 119 ± 7 kj/mol, and 144 ± 20 kj/mol for the 3.4, 4.1, and 4.9 nm diameter CdS cores respectively. To our surprise, the activation energy for the solid-solution formation between the CdS core and the ZnS shell to form Zn x Cd 1-x S alloyed NCs is dependent on the diameter of the CdS core. This observation could be explained in part by the fact that as the CdS core size increases the amount of surface defects that occur between the CdS core and the thin ZnS shell increases. These defects are due to the lattice mismatch between the zinc-blende CdS core and the cubic ZnS shell. It is likely that higher amounts of structural surface defects create a higher energy barrier for the surface adsorption of Cu ions onto the core/shell particles. This effect leads to overall larger activation energy for the larger sized CdS cores. Similar surface defect effects have been observed for Mn doped CdS/ZnS core/shell NCs. 186,195 Combining the data from Figure 3-4 and the Jander analysis to calculate activation energies, we propose that the Cu dopants act as catalysts for the alloying between the CdS core and the ZnS shell. Without the Cu dopants, no change in the absorption spectra is measured and hence no alloying takes place. It is worth noting that the new findings from this study will allow for the building of structures with precisely controlled arrangements between their nanoscopic semiconductor components. This is profoundly important for semiconductor-based electronics with new atomic structures, such as 106

107 the next generation spin-based electronic devices and high power electronics like those used in computer processors. Figure Arrhenius plots of ln k versus 1000/T. (A) For 3.4 nm diameter CdS core. (B) For 4.1 nm diameter CdS core. (C) For 4.9 nm diameter CdS core. Red lines are linear regressions of the data points used to calculate activation energies for the solidsolution formation reaction (alloying) Ag-Doping of CdS/ZnS Core/Shell Nanocrystals The initial success of copper doping of CdS/ZnS core/shell nanocrystals encouraged further studies into other possible dopants that could be incorporated into our particles. There have been numerous reports of Ag + doping of semiconductor nanocrystals and the effects that are induced by such doping Therefore, we attempted to dope Ag ions into CdS/ZnS nanocrystals using a similar approach as was applied for the Cu doping. The first attempts at doping Ag into the thin shelled CdS/ZnS core/shell nanocrystals at high reaction temperatures (> 220 C) lead to a sudden color change in the reaction solution and an immediate loss of absorption and photoluminescence properties. It was inferred that at these reaction temperatures, the addition of the silver and sulfur growth solutions lead to Ostwald ripening of the core/shell particles. Similar effects were seen when Cu dopants were added too quickly to pure CdS core nanocrystals in previous experiments (data not shown). It was hypothesized that at high temperatures, the silver acetate was reduced by oleylamine to Ag 0 and that this lead to the observed loss in optical properties for the CdS/ZnS nanocrystals. To overcome this difficulty, we designed a novel method for doping the silver into the CdS/ZnS NCs. The temperature was 107

108 set to 50 C for doping because at this temperature the reducing activity of the oleylamine was low enough for silver ions to adsorb onto the surface of the particles before they were reduced. After silver addition, the reaction temperature was raised slowly to the desired final temperature, while aliquots of the reaction solution were taken for absorption and photoluminescence measurements Ag doping of 1.6 ML CdS/ZnS core/shell nanocrystals Using the same thin-shelled CdS/ZnS NCs with a 4.1 nm diameter CdS core, we added 48 Ag atoms per nanocrystal at 50 C and measured the change in the absorption and photoluminescence (PL) spectra as the reaction temperature was raised in 50 C intervals to a final temperature of 200 C. Figure 3-12 shows the absorption and PL spectra of aliquots taken from the Ag-doped 1.6 ML CdS/ZnS core/shell NCs. Two important observations become apparent in the optical properties of these core/shell particles. First, the absorption spectra (Figure 3-12A) remain essentially unchanged with both time and increase in reaction temperature. Since no shift in the absorption peaks is measured, we can assume that, unlike Cu doping, Ag doping does not lead to an alloying effect between the CdS core and the ZnS shell. Furthermore, the sharpness and number of absorption peaks remains consistent throughout the course of the reaction after Ag is introduced. This data suggests that doping silver ions into the CdS/ZnS core/shell NCs does not induce Ostwald ripening that would otherwise diminish the quality of the absorption profile. When measuring the PL spectra (Figure 3-12B) of the Ag-doped core/shell aliquots, the observed trend is significantly different than for the absorption spectra. Recall, that for Cu- 108

109 Figure (A) Absorption spectra of aliquots taken with time for the 48 Ag atoms per 1.6 ML CdS/ZnS core//shell NCs. (B) PL spectra of aliquots taken with time for the 48 Ag atoms per 1.6 ML CdS/ZnS core//shell NCs. Reaction temperature reaches 200 C at 36 minutes. doping of 1.6 ML CdS/ZnS core/shell NCs, the PL was quenched immediately upon addition of the Cu growth solution (data not shown). But for the addition of the Ag growth solution, the PL spectrum follows a completely different trend. At early times and lower temperatures, the PL spectrum slowly decreases in intensity all the way to the point that at 26 min and 150 C there is essentially a complete quenching of the PL for the Ag-doped 1.6 ML CdS/ZnS core/shell NCs. Initially we assumed that this was a similar process as the Cu doping experiments, except that due to the low reaction temperature that was used for Ag doping the quenching needed longer reaction times and higher reaction temperatures to occur. But, to our surprise, with further heating to 200 C, the PL intensity began to slowly return to > 90 % of the intensity of the 109

110 undoped sample. This observation is the first time such kinetically distinct observations in the PL of doped semiconductor NCs have been observed. We hypothesized that this trend could be due to the surface adsorption and tunneling of Ag + into the 1.6 ML CdS/ZnS NCs Mechanism of Ag doping in CdS/ZnS core/shell nanocrystals To confirm that the Ag-doping was successful, we ran two more Ag-doping experiments where 48 Ag per 1.6 ML CdS/ZnS core/shell NCs were added at 50 C. For one experiment the reaction was stopped at the point where the PL intensity had decreased to a minimum (near complete quenching, at ~150 C), and for the other experiment we stopped the reaction once the PL intensity had returned to near the original intensity of the undoped sample (~200 C). Using ICP-AES, we determined the doping amount of silver in the purified NCs for each reaction. For the doping experiment where the reaction was stopped at the point where the PL was nearly completely quenched, near 150 C, the ICP data gave an Ag doping amount of approximately 2.0 mol %. This data confirms that silver was successfully doped into the core/shell particles at the point where the PL has been quenched significantly. This would suggest that as the Ag dopant goes through the two processes that need to occur for successful doping (i.e. surface adsorption and tunneling) the introduction of defects that increase the likelihood of nonradiative recombination must be taking place (Figure 3-13). Combine this with the fact that the absorption profile remains constant throughout the course of the reaction, and it becomes apparent that the physical structure of the core/shell particles is retained, but new pathways for carrier recombination are introduced. We hypothesized that during the initial stages of Ag doping at temperatures between 50 to 150 C, the Ag ions are adsorbing to the surface of the CdS/ZnS core/shell NCs. At these low temperatures, more time is required to reach the surface adsorption equilibrium before tunneling can take place. Hence, we observe a decrease in the PL spectrum 110

111 that is measured due to the surface adsorption of Ag ions which act as sites for nonradiative carrier recombination. Figure Scheme illustrating Ag doping of CdS/ZnS core/shell nanocrystals and its effects on the optical properties of the particles in solution. Once the reaction temperature has reached a high enough value, the energy barrier associated with the tunneling process can be overcome and the Ag ions begin to tunnel into the CdS/ZnS NCs. This process leads to fewer and fewer Ag ions on the surface of the NCs as they begin to enter the interior of the core/shell NCs. When the PL intensity returns to near the value of the original undoped NCs, we assumed that two events could be occurring, 1) most of the Ag dopants were incorporated into the interior of the CdS/ZnS NCs, and therefore few, if any, surface adsorbed ions are available to act as surface defect sites for nonradiative carrier recombination, or 2) the Ag ions completely ejected from the system and so the NCs returned to what they were before doping took place. By measuring the ICP-AES of the sample in which PL intensity had returned to near original values we gained insight into which mechanism is taking place. The ICP data for that reaction gave an Ag concentration of 0.55 mol % Ag in the 111

112 core/shell NCs. This data indicates that there is a difference in the Ag doping concentration between when the sample s PL is quenched and when it returns. This phenomenon seems similar to previous models for doping of semiconductor NCs where self-purification can take place between the dopant and the core/shell system. 196 It would seem that as the Ag ions tunnel into the CdS/ZnS core/shell particles that an equilibrium is established between the silver and the CdS/ZnS core/shell system. It is possible that due to the large size of the Ag ions compared to Cu that a smaller amount can be incorporated into the particle system. Larger ions would be more likely to introduce larger sized inclusions into the semiconductor lattice. When a specific threshold is reached, the NCs begin to eject the excess silver from their interior. This mechanism has similarities to other doping systems investigated in the Cao research group. It has been previously reported that for Mn dopants on CdS/ZnS core/shell NCs that there exists weakly and strongly bound Mn on the surface of the particles. 186 The strongly bound Mn leads to successful doping inside the core/shell particles while the weakly bound Mn is less likely to be incorporated inside the NCs. It is possible that a similar process occurs for Ag ions where some are more strongly bound to the ZnS shell surface than others. In this case, as the initial equilibrium for the surface adsorbed silver is being established, there exists a mixture of strongly and weakly bound silver. This creates a situation where the surface concentration of Ag is highest and so the PL intensity is lowest. As the reaction temperature is increased, the strongly bound Ag is successfully doped inside the 1.6 ML CdS/ZnS core/shell NCs while the weakly bound Ag is washed away during purification. This mechanism would explain the observed difference between the Ag concentration when the PL intensity is quenched and when (and how) it returns. 112

113 These observations for the Ag doping experiments are the first to our knowledge to demonstrate how the kinetic evolution of Ag ions as they are doped into a semiconductor system can be monitored using optical measurements. At each step, the PL spectra can describe the physical processes that are taking place during the doping experiment, almost like a snap shot of the doping process as it is occurring. In the early stages, when the growth solutions are added into the core/shell NCs at 50 C and during heating up to 150 C, the Ag ions are establishing equilibrium for the surface adsorbed species. This process leads to higher concentrations of Ag on the particles surface and a clear, measureable decrease in the PL spectra intensity. These surface adsorbed species are made up of strongly and weakly bound silver species. As the temperature is increased, the energy barrier for silver tunneling is surpassed and the Ag ions begin to enter into the interior of the CdS/ZnS NCs. This process gives preferential incorparation to the strongly bound Ag and hence less silver is doped into the particle than was originally adsorbed onto the surface. As the Ag ions tunnel inward, there is less and less silver on the surface to act as sites for nonradiative carrier recombination. Hence, we see a measured increase in the PL spectrum intensity and a significant decrease in the ICP concentration of silver between the quenched sample at 150 C and the sample at 200 C where the PL intensity has returned. 3.4 Summary In conclusion, using doping procedures developed in our lab, we have successfully doped copper and silver into CdS/ZnS core/shell nanocrystals. It was demonstrated that copper dopants act as catalysts to promote the alloying of the CdS core with the ZnS shell. To confirm that Cu dopants were acting as catalysts in the observed nanoscopic alloying, the activation energy (E A ) of the process was determined. It was shown that Cu can substantially lower the E A of the alloying process. Interestingly, it was also found that the E A for the alloying of nanocrystals is 113

114 size dependent. The E A ranged from 89 to 144 kj/mol for CdS cores with diameters between 3.4 to 4.9 nm. These results will have a significant impact of the design and production of new semiconductor components where the specific arrangement of the constituent atoms is of the utmost importance. Furthermore, Ag ions were also successfully doped into CdS/ZnS core/shell nanocrystals. Here, we demonstrated the first doping experiments where the optical properties of the host particles change in time with the physical doping processes occurring in semiconductor nanocrystals, i.e surface adsorption and tunneling. The PL spectra can give accurate insight into the relative location of the dopants, and through the observed mechanisms there is further support for the idea that both strongly and weakly bound dopants adsorb onto the surface of semiconductor nanocrystals. 114

115 CHAPTER 4 SURFACE FUNCTIONALIZATION OF METAL OXIDE NANOCRYSTALS BY INTRODUCTION OF ELECTROSTATIC CHARGE VIA INORGANIC SALT IONS 4.1 Prologue Surface engineering of nanocrystals is a key step in the advancement of nanocrystal-based applications for commercial use. New functionalization methods need to be designed that allow for better solubility in polar and nonpolar solvents as well as the processing of nanomaterials for use in the medical and electronics industries. Previous research indicates that nanocrystal properties are very sensitive to their surface states. Modification of nanocrystal surfaces directly improves their stability and solubility, as well as their chemical and physical properties. Since NCs have exhibited a number of potential bioapplications, one goal of surface engineering is to achieve robust and biocompatible nanocrystals for use in vivo. By modifying NC surfaces, (e.g. ligand exchange 12 or coating with silica or amphiphilic polymers 6,42,202 ) the NCs can be made water-soluble. Conjugation with proteins, DNA, and other biological molecules presents ways to use QDs for biosensors or in diagnostic studies. Presently, two major methods have been developed to modify the coatings of hydrophobic NCs using organic ligands. The first approach is based on coordinate bonding. Functional groups (such as thiol, 12 dithiol, 106 phosphine 203 and dopamine 204 ) are used to link hydrophilic groups directly to the surface of hydrophobic NCs by replacing the original hydrophobic ligands. The second approach uses hydrophobic van der Waals interactions, through which the hydrophobic tails of amphiphilic ligands interact with (but do not replace) the hydrophobic ligands on the NCs, 205 leading to the formation of nanocrystalline-micelles. Many types of water-soluble NCs made by these two approaches suffer from low stability and/or high nonspecific binding with non-target biomolecules. Water-soluble NCs coated with PEGylated amphiphilic polymers have very high stability and low non-specific-adsorption levels, 106,205 but 115

116 the PEGylated polymer shells often produce large hydrodynamic diameters (HDs), on the order of 30 to 40 nm, which can limit the use of these NCs in applications such as in vivo cell imaging. More recently, methodologies for making various types of NCs soluble in polar solvents have relied on using small, charged ions to impart surface charge on their surface. 108,109 Using a nitrosonium tetrafluorborate salt (Figure 4-1), Dong et. al. were able to transfer various types of colloidal inorganic nanocrystals into DMF, DMSO, or acetonitrile. 109 Figure 4-1. Schematic illustrating use of nitrosonium tetrafluorborate salt (NOBF 4 ) to transfer colloidal inorganic nanocrystals into polar solvents. Reprinted with permission from Dong, A. G.; Ye, X. C.; Chen, J.; Kang, Y. J.; Gordon, T.; Kikkawa, J. M.; Murray, C. B. J. Am. Chem. Soc. 2011, 133, 998. In another strategy, Nag et. al. used metal-free inorganic ions to transfer metallic and semiconductor nanocrystals into polar solvents such as formamide and DMSO (Figure 4-2). 108 These small inorganic ions provide stability for several months when processed in an inert environment and are shown to facilitate charge transport in films of the coated nanocrystals. The overall process is explained on the basis of Pearson s hard and soft acid/base principles. 108 In both cases the transferred nanocrystals are most stable in non-aqueous polar solvents, limiting 116

117 their application in cellular studies. Furthermore, in both cases the ligand exchange process is applied mostly to metallic or semiconducting nanocrystals with less emphasis on metal oxides. Figure 4-2. (A) Picture showing phase transfer of CdSe nanocrystals from toluene to formamide (FA) using potassium sulfide. (B) Schematic illustrating the binding of the metal-free inorganic ligands onto the surface of a nanocrystal. Reprinted with permission from Nag, A.; Kovalenko, M. V.; Lee, J. S.; Liu, W. Y.; Spokoyny, B.; Talapin, D. V. J. Am. Chem. Soc. 2011, 133, Herein, we describe a new method of producing water soluble metal oxide nanocrystals using simple, inexpensive inorganic salt ions. This method relies on the introduction of electrostatic charge, via the inorganic ions, to give the metal oxide NCs their solubility in the aqueous phase. The method is shown to work for several different compositions of metal oxides, including iron oxide, manganese oxide, indium oxide, and zinc oxide. Different inorganic salts (sodium arsenite, sodium phosphate, and sodium hydroxide) are shown to be effective at exchanging with the original long carbon chain ligands on the surface of the metal oxide nanocrystals. We further demonstrate the application of arsenite-coated iron oxide composite nanocrystals (AICN) as an alternative cancer therapeutic that has multiple advantages over other arsenic based cancer treatments. The coating of arsenite onto the surface of iron oxide 117

118 nanocrystals allows for the simultaneous delivery of arsenic (III) into specific target locations while also monitoring the 3-dimensional in vivo biodistribution of the cancer therapeutics using MRI. It was found that when the arsenite molecule was delivered without the iron oxide nanocrystals that the toxicity significantly increased when compared with the surfacefunctionalized iron oxide particles. The half maximal inhibitory concentration (IC 50 ) for our cell experiments was shown to be lower for healthy liver cells, but not for hepatocellular carcinoma cells when comparing the effectiveness of the arsenite-coated iron oxide nanocrystals towards inducing cellular death via necrosis. This is important for reducing the side effects from chemotherapy for cancer patients. 4.2 Experimental Section Chemicals Iron chloride hexahydrate (FeCl 3 6H 2 O, 99 %) was purchased from Acros Organic; zinc (II) stearate (Zn[CH 3 (CH 2 ) 16 CO 2 ] 2, 1-octadeconal (97 %) and stearic acid (CH 3 (CH 2 ) 16 CO 2 H, 90 %) was purchased from Alfa Aesar; hexadecane (HDA, 99 %), manganese chloride tetrahydrate (MnCl 2 4H 2 O, 99 %), oleic acid (OLA, 90 %), oleylamine (OAm, 70 %), 1-octadecene (ODE, 90 %), 1-tetradecene (TDE, 92 %), indium (III) acetate (InC 6 H 9 O 6, %), trimethylamine N- oxide (TMNO, 98 %), and sodium (meta) arsenite (NaAsO 2, 99 %) were purchased from Aldrich. Sodium oleate (97 %) was purchased from Tokyo Chemical Industry. Nanopure water (18 MΩ cm) was made by a Barnstead Nanopure Diamond system. Sodium hydroxide (99 %), sodium phosphate heptahydrate (Na 2 HPO 4 7H 2 O, 99 %), and all other solvents were purchased from Fisher Scientific International, Inc. 118

119 4.2.2 Synthesis of Metal Oxide Nanocrystals Iron oxide nanocrystals From a previous report, 154 for a reaction temperature of 290 C: iron oleate (1 mmol, 0.9 g) and oleic acid (0.55 mmol, g) were added into a three-neck flask (25 ml) with a solvent mixture of TDE/ODE (1.75 g/3.25 g). The mixture was stirred under argon flow at room temperature. After 10 minutes, the reaction solution was heated to 290 C at a heating rate of ~18 C/min. Reaction time was counted from the moment when 290 C was reached. After one hour, the reaction solution was quickly cooled to room temperature by blowing air across the reaction flask. The resulting iron-oxide nanocrystals were isolated using acetone, and then were purified using three rounds of precipitation/redispersion cycles with acetone and hexane as the solvents. After purification, the product was dispersed in nonpolar solvents such as chloroform or toluene Manganese oxide nanocrystals From a previous report, 206 for a reaction temperature of 250 C: manganese oleate (1 mmol, 0.62 g) was added into a three-neck flask (25 ml) with TDE (5 g) as the solvent. The mixture was stirred under vacuum at 70 C for one hour. Then the reaction solution was heated to 250 C at a heating rate of ~15 C/min. Reaction time was counted from the moment when 250 C was reached. After one hour, the reaction solution was quickly cooled to room temperature by blowing air across the reaction flask. The resulting manganese oxide nanocrystals were isolated using ethanol, and then were purified using three rounds of precipitation/redispersion cycles with ethanol and hexane as the solvents. After purification, the product was dispersed in nonpolar solvents such as chloroform or toluene. 119

120 Indium oxide nanocrystals From a previous report, 207 indium acetate (165 mg, 0.4 mmol), oleylamine (0.55 ml), and oleic acid (0.6 ml) were mixed with ODE (7 ml) in a three-neck flask. The mixture was degassed under vacuum at 110 C for 30 minutes to form a clear light-green solution. TMNO (161 mg, 1.45 mmol) was added to the vigorously stirring hot mixture under Ar. Then, the reaction mixture was further degassed at 120 C for 1 hour. The resultant yellow/orange solution was heated to 290 C at a heating rate of ~10 C/min. After 35 minutes, the reaction solution was quickly cooled to room temperature by blowing air across the reaction flask. The resulting indium oxide nanocrystals were precipitated using ethanol, and the yielded precipitate was redispersed in hexane followed by centrifugation to remove the very small amount of insoluble aggregates. Then the nanocrystals were purified using two rounds of precipitation/redispersion cycles with ethanol and hexane as the solvents. After purification, the product was dispersed in nonpolar solvents such as chloroform or toluene Zinc oxide nanocrystals From a previous report, 208 zinc stearate (0.2 mmol) and 4 g of 1-octadecene (ODE) were loaded in a 25 ml three-necked flask. The mixture was heated to 280 C under Ar atmosphere. 1-Octadecanol (1 mmol) dissolved in 1 g of ODE at 200 C was quickly injected into the zinc stearate solution, and the reaction temperature was then set at 250 C throughout the entire synthesis. To synthesize ZnO nanopyramids, 8 minutes after the injection of 1-octadecanol, stearic acid (0.2 mmol) dissolved in 0.5 g of ODE at 120 C was injected into the ZnO nanocrystals solution and incubated for 2 hours. To convert the ZnO nanopyramids back to spherical particles, stearic acid (0.2 mmol) dissolved in 0.5 g of ODE at 120 C was injected into the ZnO nanopyramids solution, and during the first few minutes after acid injection, the shape of ZnO nanocrystals became spherical. The zinc oxide nanocrystal solution was quickly cooled 120

121 to room temperature by blowing air across the reaction flask, and ethyl acetate was used to isolate the zinc oxide nanocrystals. Then the nanocrystals were dispersed in toluene, and any insoluble residue was removed by centrifugation. The zinc oxide nanocrystals were purified using two rounds of precipitation/redispersion cycles with ethanol and hexane as the solvents. After purification, the product was dispersed in nonpolar solvents such as chloroform or toluene Ligand Exchange of Metal Oxide Nanocrystals using AsO - 2, HPO 2-4, and OH - ions The ligand exchange process was typically carried out in air. Solutions of inorganic ligands were prepared in water immiscible with chloroform or toluene. For a typical ligand exchange using AsO - 2 ions, 1 ml of iron oxide NC solution (10-15 mg/ml) was mixed with 1 ml of NaAsO 2 solution (0.1 M). The mixture was stirred for several hours leading to a complete phase transfer of the iron oxide NCs from chloroform to the water phase. The water phase was allowed to separate out and the NCs isolated using ethanol followed by centrifugation. The iron oxide NCs were purified using two rounds of precipitation/redispersion cycles with ethanol and water as the solvents. After purification, the product was dispersed in water and used for further studies. The NC dispersion in water was stable for weeks when the exchange was performed in air and for months when performed under inert atmosphere. Ligand exchange with HPO 2-4 and OH - ligands were carried out in a similar manner, except using 1.0 M salt solutions and increasing exchange time to overnight. Ligand exchange for MnO, In 2 O 3, and ZnO NCs were carried out in a similar manner. Not all metal oxide NCs could be exchanged with all salts Characterization of Water Soluble Metal Oxide Nanocrystals Absorption measurements UV-Vis absorption spectra were measured using a Shimadzu UV1800. Nanocrystals were dissolved and diluted in toluene or water for the measurement. 121

122 TEM measurements TEM measurements were performed on a JEOL 200X operated at 200 kv, or a JEOL 2010F TEM operated at 200 kv. The specimens were prepared as follows: a nanoparticle solution (10 μl) was dropped onto a 200-mesh copper grid and was dried overnight under ambient conditions Energy dispersive spectroscopy (EDS) measurements Energy dispersive spectroscopy measurements were performed on a JEOL 2010F TEM operated at 200 kv. The specimens were prepared as follows: a concentrated nanoparticle solution (10 μl) was dropped onto a 200-mesh copper grid and was dried overnight under ambient conditions Zeta (ζ) potential measurements ζ- potential measurements were performed on a Brookhaven ZetaPlus. ζ- potential was calculated from the electrophoretic mobility using Henry s equation in the Smoluchowski limit Inductively-coupled plasma (ICP) atomic emission spectroscopy (AES) measurements The ICP measurements were performed on a Vista RL CCD Simultaneous ICP-AES (Varian, Inc.). The purified nanocrystal samples were digested with concentrated (69.5 %) nitric acid. The digestion was performed at about 100 C until the solution became colorless. The digestion solutions were further diluted with a nitric acid solution to obtain a final nitric acid concentration of about 1 2 %. The concentrations of As and Fe in the solutions were determined by data from ICP measurements as compared with the corresponding working calibration curves (Figure A-1). 122

123 4.2.5 Cellular Studies Using AsO - 2 Coated Iron Oxide Composite Nanocrystals (AICN) Materials for cellular studies Human hepatocellular carcinoma cell lines, Huh7, were grown in Dulbecco's Modified Eagle's Medium (DMEM) with 10 % fetal bovine serum (FBS) and antibiotics (100 U/mL penicillin and 100 µg/ml streptomycin) at 37 C in 5 % CO 2 and were purchased from Life Technologies Corporation. Hoechst 33258, phenazine methosulfate (PMS), phosphate buffered saline (PBS), lysis buffer (1 % NP-40 in 20 mm EDTA, 50 mm Tris-HCl, ph 7.5), sodium dodecyl sulfate (SDS), RNase A, agarose gel, and proteinase K were obtained from Sigma. 3- (4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxy-phenyl)-2-(4-sulfophenyl)-2H-tetrazolium (MTS) cell proliferation assay was purchased from Promega Corporation Cell morphology study Huh7 cells were seeded onto 35-mm wells of a six-well cell culture plate and cultured overnight. The culture medium was then removed and replaced with new medium (1 ml/plate) along with different concentrations of AICN (0, 10, 25, 50, 100 μm As) and pure NaAsO 2 (0, 10, 25, 50, 100 μm As). After 48 hours treatment, morphological changes were detected using an IX51 Olympus fluorescent microscope (Olympus America, Inc.). Control cells were incubated only with the culture medium Cell viability study Huh7 cells were dispensed into 96-well plates at a final concentration of ~ cells/well in a culture medium (100 μl), and incubated overnight before treatment. The culture medium was then removed and replaced with new medium along with different concentrations of AICN (0, 10, 25, 50, 100 μm As) and pure NaAsO 2 (0, 10, 25, 50, 100 μm As). After 72 hours treatment, cell viability was measured using the 3-(4,5-dimethylthiazol-2-yl)-5-(3- carboxymethoxyphenyl)-2-(4-sulfophenyl)-2h-tetrazolium (MTS) cell proliferation assay. The 123

124 cells were incubated for 3 hours after addition of MTS/PMS solution to allow for color development, and then the absorbance values were read at 492 nm using a Multiscan plate reader Hoechst staining assay Apoptosis was determined through nuclear morphology change. After treatment with different concentrations of AICN (0, 10, 25, 50, 100 μm arsenite) and pure NaAsO 2 (0, 10, 25, 50, 100 μm arsenite), Huh7 cells were stained with Hoechst at 37 C for 10 minutes. Cellular DNA fragmentation/nucleus condensation was detected using an IX51 Olympus fluorescent microscope (Olympus America, Inc.) DNA ladder assay A DNA ladder assay was performed as previously described 209 with modification. Briefly, cells treated under different concentrations of AICN (0, 10, 25, 50, 100 μm arsenite) were harvested and washed one time with phosphate buffered saline (PBS). Cell pellets were resuspended with lysis buffer (1 % NP-40 in 20 mm EDTA, 50 mm Tris-HCl, ph 7.5). The samples were centrifuged at 1,600 g for 5 minutes and supernatants were transferred to new tubes. SDS (final concentration 1 %) and RNase A (final concentration 5 mg/ml) were added to the supernatants, and the mixture was incubated at 56 C for 2 hours. After addition of proteinase K (final concentration 2.5 mg/ml), the samples were incubated at 37 C for 2 hours. DNA was precipitated with the addition of 10 M ammonium acetate and ethanol, washed once with 70 % ethanol, and then dissolved in water and separated by electrophoresis in 1 % agarose gel. 124

125 4.3 Results and Discussion Ligand Exchange using AsO - 2, HPO 2-4, and OH - ions Metal oxide nanocrystals (Fe 3 O 4, In 2 O 3, ZnO, MnO) were synthesized according to previously reported methods 154, and transferred into water using a ligand exchange procedure with AsO - 2, HPO 2-4, and OH - ions as depicted in Figure 4.3. Figure 4-3. Scheme of metal oxide nanocrystal ligand exchange between oleate and arsenite. After the initial synthesis of the metal oxide NCs, the purified particles were suspended in chloroform or toluene at concentrations ranging from 10 to 15 mg/ml. Solutions of the inorganic salts (1 ml, 0.1 M or 1.0 M) were added to each of the NC solutions (1 ml) and allowed to exchange in air or under inert atmosphere for up to 5 hours. Afterwards the phases were separated via centrifugation and the lower aqueous phase was isolated for further purification via precipitation with ethanol. After two rounds of purification the nanocrystals were suspended in fresh Nanopure water and were ready for characterization. Figure 4-4 shows pictures of the nanocrystals solutions before and after the ligand exchange 125

126 Figure 4-4. Pictures of nanocrystal solutions (Fe 3 O 4, MnO, In 2 O 3, ZnO,) before (A) and after (B) ligand exchange with AsO 2 -. (A) Upper phases are solutions of purified nanocrystals in hexane, lower phases are pure water. (B) Upper phases are pure hexane and lower phases are purified nanocrystals with AsO 2 - ligand on the surface. with AsO 2 - ions (arsenite). It can be seen that the nanocrystals are completely transferred into the aqueous phase after the exchange with the inorganic ions. These NCs are stable for several months depending on the atmosphere used to do the exchange and the size of the initial nanocrystals. In air, the particles are stable for a few weeks to months. Larger sized NCs (> 10 nm in diameter) were more unstable than smaller (< 7 nm in diameter) sized ones. In inert atmosphere, the exchanged NCs could last several months irrespective of the size. To determine whether or not the ligand exchange had been successful, several techniques were used to characterize the water soluble metal oxide nanocrystals. Initially, TEM measurements were made to ensure that the size, shape, and size distribution were consistent with the original NCs synthesized in the nonpolar solvents. Figure 4-5 shows TEM images of the iron oxide (Fe 3 O 4 -AsO 2 ) and indium oxide (In 2 O 3 -AsO 2 ) nanocrystals after exchange with the AsO 2 - ions. Noticeably, the particles are nearly monodispersed and spherical in shape with average diameters similar to those of the originally synthesized NCs. It is evident from the TEMs that size is not an important factor for the successful exchange of metal oxide nanocrystals into the aqueous phase using inorganic ions as ligands. 126

127 Figure 4-5. TEM images of iron oxide and indium oxide nanocrystals in water after ligand exchange with AsO 2 - ions. (A) Fe 3 O 4 with a 5.2±0.4 nm diameter. (B) Fe 3 O 4 with a 7.2±0.6 nm diameter. (C) Fe 3 O 4 with a 12.1±1.0 nm diameter. (D) In 2 O 3 with a 10.2 ±1.1 nm diameter. Using the TEM, EDS measurements were made on the Fe 3 O 4 -AsO 2 and In 2 O 3 -AsO 2 NCs to qualitatively determine whether or not arsenic was present in the sample. Figure 4-6 shows the results of the EDS analysis, and these results clearly show the x-ray peaks for the arsenic, iron, indium, and oxygen that are characteristic of the NCs after they have been exchanged with the arsenite ligands. EDS can only show the existence of elements in a sample and therefore it gives no evidence to the position of the ligands on the nanocrystals. While EDS is supportive of the hypothesis that the inorganic ions have exchanged with the bulky, long alkyl chain oleate ligands, other techniques are needed to confirm an exchange on the surface of the metal oxide nanocrystals. DLS and ζ-potential measurements have been used in many studies to estimate the hydrodynamic radius and surface charge of particles and confirm successful ligand exchanges. 45,108,109 DLS and ζ-potential measurements were made on two sizes of Fe 3 O 4 -AsO 2 (5.2 and 7.2 nm diameters) and the results showed that the surface of the NCs had negative 127

128 Figure 4-6. EDS measurements of Fe 3 O 4 -AsO 2 and In 2 O 3 -AsO 2 NCs after ligand exchange into water. (A) Spectra of iron oxide NCs with arsenite ions showing characteristic iron, arsenic, and oxygen peaks. (B) Spectra of indium oxide NCs with arsenite ions showing characteristic indium, arsenic, and oxygen peaks. Cu and Si peaks are from the TEM grid. charges of -48 and -45 mv. The results from DLS analysis were inconclusive due to the small size of the NCs and the relative inaccuracy of the instrument that was employed and available. Since DLS analysis was unable to be performed successfully, the interparticle spacing of the 7.2 nm diameter Fe 3 O 4 -AsO 2 was used to demonstrate the exchange of the long oleate ligand (~2.2 nm) with the significantly shorter arsenite ion. Figure 4-7 shows TEM images of twodimensional superlattices of the Fe 3 O 4 -AsO 2 NCs in water and the original Fe 3 O 4 NCs in toluene. Interparticle spacing analysis showed that for the Fe 3 O 4 -AsO 2 NCs in water the spacing was 1.7±0.1 nm while for the Fe 3 O 4 -oleate NCs in toluene the spacing was 3.4±0.4 nm. These results further confirm the successful ligand exchange from oleate to arsenite initiating the phase transfer of the metal oxide nanocrystals from nonpolar solvents to water. 128

129 Figure 4-7. TEM images of two-dimensional superlattices of 7.2 nm diameter iron oxide NCs. (A) Fe 3 O 4 -AsO 2 NCs in water. (B) Fe 3 O 4 -oleate NCs in toluene. Scale bars are 50 nm. The analysis up to this point had qualitatively confirmed the ligand exchange procedure for the metal oxide NCs using inorganic ions. To quantitatively determine the amount of the ligand in the NCs solutions, we performed ICP-AES analysis on the purified samples of Fe 3 O 4 -AsO 2 NCs in water. The ICP-AES results were compared to calibration curves (Figure A-1) for iron and arsenic, and the final concentration of the arsenic was calculated as well as the Fe to As ratio. For the 5.2 nm Fe 3 O 4 -AsO 2 NCs in water, the concentration of arsenic in the solution was ~20.2 mm. The ICP-AES data is summarized in Table A-4. Taken together, the results from the characterization of the Fe 3 O 4 -AsO 2 NCs strongly support that the ligand exchange was successful and that the arsenite ions are bound to the surface of the metal oxide nanocrystals introducing electrostatic charge which gives the NCs their solubility in water. After using the arsenite to exchange the metal oxide nanocrystals, attempts were made to find other inorganic ligands that could transfer metal oxide NCs into water. Since previous reports 108 had shown that the hydroxide ion (OH - ) can be used with indium oxide nanocrystals to transfer them into polar solvents, it was chosen along with the phosphate ion (HPO 2-4 ) to attempt ligand exchanges. Using similar procedures as before, it was 129

130 found that the hydroxide ion can be used for Fe 3 O 4, In 2 O 3, and MnO NCs. The phosphate ions were shown to be successful for Fe 3 O 4 and In 2 O 3 NCs. The reason behind the inability to transfer all metal oxides using all of the inorganic ligands is still under investigation Arsenite Coated Iron Oxide Composite Nanocrystals (AICN) as Cancer Therapeutics It has been shown by O Halloran and his group that arsenic oxide in nanocrystal form can be used as an effective cancer therapeutic. 210,211 One disadvantage of such treatments is the toxicity to both cancer and healthy cells. Nanocrystals offer several advantages to traditional cancer treatments because of the ability to simultaneously allow for targeted drug delivery and monitoring the distribution throughout the body. 43,113 Iron oxide nanocrystals can be used as effective MRI contrast agents with minimal side effects. 44,81,112 Arsenite-coated iron oxide composite nanocrystals (AICN) offer a unique opportunity to both deliver arsenite ions and monitor their distribution via MRI contrast imaging. To test the efficacy of the AICN, several in vitro cellular experiments were performed and the results were compared to control experiments and to the pure sodium arsenite salt used for the ligand exchange Cell morphology study Initial experiments were run to determine if the AICN could be effective at inducing cellular death to human hepatocellular carcinoma cell lines (Huh7). Using the concentration of arsenic determined from ICP-AES analysis, 5.2 nm AICN were prepared at a concentration of 100 μm arsenic with similar concentrations prepared using the pure NaAsO 2 salt. Both solutions were added to cell cultures along with control experiments where only fresh growth medium was added. Figure 4-8 shows the optical microscope images of the Huh7 cells after 48 hours incubation with the pure salt, the AICN, and the controls. 130

131 Figure 4-8. Optical microscope images of Huh7 cells after 48 hours of incubation with AICN and pure NaAsO 2 both at 100 μm arsenite. (A,C) Control experiments with only fresh growth media added. (B) Pure NaAsO 2 (D) AICN. For the control experiments, the Huh7 cells were still attached to the culture dish showing elongated, interconnected cells that are hallmarks of healthy growing organisms (Figure 4-8A, C). For the same cell line treated with pure NaAsO 2 or AICN there are clear differences in the cells morphology and location (Figure 4-8B,D). The Huh7 cells have detached from the culture dish indicated by the white contrast that is observed for the cells, showing the difficulty in focusing on the cells due to their different depth profiles. Furthermore, the shape of the cells is clustered and more globular in nature. These are clear signs of cellular death via necrosis. The experiments support the hypothesis that AICN can be used to induce cellular death to cancer cells. But, it is also apparent that any source of arsenic, like that from pure sodium arsenite, can be used to kill cancer cells. One of the advantages AICN should provide is a lower toxicity to cells due to the NCs having only a surface coating of arsenic containing ligand rather than the entire particle composition containing arsenic. 131

132 Cell viability studies To determine how the toxicity of the pure NaAsO 2 compares to that of the AICN, cell viability experiments were performed to determine the half maximal inhibitory concentration (IC 50 ), a measure of the efficacy of a compound at inhibiting a biological or biochemical process. The IC 50 is a quantitative indicator of how much of a particular substance is required to inhibit a given biological function by half. In the experiments, the concentration of arsenic needed to induce necrosis in 50 % of the Huh7 cells was determined. Using a standard cell proliferation assay via MTS conversion to formazan, the amount of cell death can be estimated by measuring the absorbance of formazan at 490 nm. Healthy living cells convert MTS to formazan at a known rate and therefore calibration curves can be made estimating the amount of living cells in a culture plate. If the original amount of Huh7 cells is known, then the amount of cells that died in a given period of time can be estimated. Figure 4-9 summarizes the results from experiments comparing pure NaAsO 2 to AICN at arsenite concentrations ranging from μm. It is clear that at the lower concentrations of arsenite, both the sodium arsenite and the AICN have similar capacities to induce necrosis in the Huh7 cells up to a 25 μm arsenite concentration. When the concentration of arsenite was increased to 50 μm, the results show significant differences in the ability of the sodium arsenite versus AICN to induce necrosis. This trend continues up to 100 μm arsenite. These results suggest that up to some critical concentration of arsenite, the cancer efficacy is minimal, most likely because there is not enough active material to significantly alter the metabolism of the living cells. But, after crossing the 50 μm threshold, there is enough arsenite to induce cellular death in Huh7 cancer cells. It is also apparent that once the threshold is passed, the pure NaAsO 2 becomes more toxic to the cancer cells than the AICN. This is expected because the pure sodium arsenite in fully dissolved in solution and therefore the ions can all be quickly and efficiently up taken by the Huh7 cells while for the AICN the arsenite 132

133 ligand is attached to an iron oxide nanocrystal. The nanocrystal is much larger than individual ions and therefore should take longer to be metabolized by the Huh7 cells. While being less toxic at 50 μm when compared to pure NaAsO 2, the AICN can induce necrosis upon ~57 % of the cells in the experiments, which is near the IC 50 limit. The cell viability experiments allow for further investigation into the toxicity of the AICN when compared to pure NaAsO 2 at concentrations that would be required for a cancer therapeutic to be considered viable for actual cancer treatments. Figure 4-9. Summaries of cell viability tests comparing pure NaAsO 2 (blue) with AICN (red) at inducing necrosis to Huh7 cells. The dotted black line represents the IC 50 limit where half of the cells have undergone necrosis Fluorescent DNA staining experiments To further test the efficacy of AICN as a cancer therapeutic, DNA staining with a fluorescent dye was employed. When cells undergo necrosis their inner organelles are disrupted and the contents ejected into the cytoplasm. When this occurs in the nucleus, the DNA is fragmented into many smaller length oligonucleotides. The Hoechst fluorescent dye 133

134 binds to DNA, and therefore the more strands of DNA the more binding that occurs. This leads to a stronger fluorescent signal when cellular death is induced compared to healthy control cells. Figure 4-10 shows experiments run with pure NaAsO 2 and AICN at 50 μm arsenite which is near the IC 50 concentration that would be typically required for acceptance as medical treatments. Looking at the control experiments, (Figure 4-10A, C) the cells show minimal fluorescence and have shapes that are similar in size and shape to each other. This indicates cell that are healthy and functional. The lack of fluorescence indicates insignificant amounts of binding by the Hoechst fluorescent dye to the nuclear DNA of the Huh7 cells which would be expected for control cells that have not had any arsenite added into their growth media. In the experiments with 50 μm arsenite added either from pure NaAsO 2 (Figure 4-10B) or AICN (Figure 4-10D), there is a high level of fluorescent signal emanating from the cells compared to the control experiments. Furthermore, the shape and size of the cells has become irregular with signs of cellular disruption and fragmentation, all indicative of necrosis. These observations show that upon addition of the arsenite, the Huh7 cells experience stress from the treatment and their nuclear DNA fragments allow for more Hoechst fluorescent dye to bind to the shorter DNA strands. With more binding comes stronger fluorescence leading to the observed increase in intensity. Another trademark of necrosis is seen from the irregular, smaller shaped cells highlighted by the white arrows shown in the zoomed in area of each image (Figure 4-10B, D). These cells are undergoing necrosis, in which their inner organelles are destroyed and ejected from the cytoplasm, thus altering the cells shape and size Gel electrophoresis assay With sufficient evidence to support that AICN can be used as a cancer therapeutic in vitro, the next goal was to show that the delivery of arsenite from the AICN would be less toxic to healthy tissue cells while still being toxic to cancer cells. Side effects of cancer therapies are one 134

135 of the major obstacles patients face when getting cancer treatments. These treatments can severely damage healthy tissue and/or the immune system, thereby making recovery times longer and more difficult. Developing cancer therapeutics that are effective at destroying tumors while limiting the damage to the surrounding healthy tissue would have major advantages over traditional cancer treatments. To test AICN toxicity on healthy liver cells, gel electrophoresis Figure Fluorescent microscope images of Huh7 cells treated with 50 μm pure NaAsO 2 and AICN. (A,C) Control experiments with only fresh growth media added. (B) 50 μm pure NaAsO 2. (D) AICN with 50 μm arsenite. The upper outlined section of the figure is zoomed in directly below with red arrows indicating healthy cells and white arrows indicating cells undergoing necrosis. was run with arsenite concentrations of 10, 25, and 100 μm arsenite while simultaneously running two control experiments. Figure 4-11 shows the results of the gel electrophoresis assay. Gel electrophoresis can easily help in visualizing DNA fragmentation that occurs when cells experience necrosis. The gel experiments illustrate the fragmentation of DNA by having set markers that correspond to specific lengths of DNA oligonucleotides. When cells undergo necrosis, their internucleosomal DNA fragments randomly give rise to many different length DNA strands. This can be observed in the gel electrophoresis experiments as a streak in the lane showing cellular death, as seen in Figure 4-11E for Huh7 cancer cells treated with 100 μm 135

136 arsenite from AICN. The other lanes in Figure 4-11 correspond to healthy liver cells treated with increasing concentrations of arsenite delivered via AICN. The absence of any streak or individual marks in these lanes (Figure 4-11B, C, D) suggests that at these concentrations AICN cannot induce significant levels of necrosis in healthy liver cells in vitro. These observations support the idea that AICN is indeed likely to be less toxic to healthy tissue than to cancer tissue. Figure Gel electrophoresis assay showing DNA fragmentation patterns for healthy liver cells treated with different concentrations of AICN. (A) Cells treated with only fresh growth media. (B) 10 μm arsenite. (C) 25 μm arsenite. (D) 100 μm arsenite. (E) Positive control with Huh7 cells treated with 100 μm arsenite. Left side of image shows DNA markers that correspond to specific DNA fragment lengths. This could lead to fewer side effects from AICN as a cancer therapeutic when compared to other cancer treatments. 4.4 Summary In conclusion, a new technique to transfer metal oxide nanocrystals from nonpolar to polar solvents was outlined. Using the inorganic ions AsO 2 -, HPO 4 2-, and OH - as surface ligands, we introduced surface charge onto metal oxide nanocrystals allowing for their transfer into water. 136