WASHINGTON UNIVERSITY Department of Chemistry

Dissertation Examination Committee: William E. Buhro, Chairman Daniel J. Leopold Patrick C. Gibbons Sophia E. Hayes Richard A. Loomis John R. Bleeke

HIGH-QUALITY COLLOIDAL II-VI AND IV-VI SEMICONDUCTOR NANOWIRES: DIAMETER-CONTROLLED SYNTHESES, QUANTUM-CONFINEMENT-EFFECT STUDIES, AND ELECTRONIC STRUCTURE

by Jianwei Sun

A dissertation presented to the Graduate School of Arts and Sciences of Washington University in partial fulfillments of the requirements for the degree of Doctor of Philosophy

May 2008 Saint Louis, Missouri

UMI Number: 3316683

INFORMATION TO USERS

The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleed-through, substandard margins, and improper alignment can adversely affect reproduction. In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion.

UMI
UMI Microform 3316683 Copyright 2008 by ProQuest LLC. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. ProQuest LLC 789 E. Eisenhower Parkway PO Box 1346 Ann Arbor, Ml 48106-1346

Copyright by Jianwei Sun 2008

ABSTRACT OF THE DISSERTATION

High-Quality Colloidal II-VI and IV-VI Semiconductor Nanowires: Diameter-Controlled Syntheses, Quantum-Confinement-Effect Studies, and Electronic Structure by Jianwei Sun Doctor of Philosophy in Chemistry Washington University in St. Louis, 2008 Professor William E. Buhro, Chairman

This project aims to grow colloidal II-VI and IV-VI semiconductor nanowires employing the solution-liquid-solid (SLS) mechanism first developed in our group. Bi nanoparticles are used to catalyze the wire growth in the coordinating solvent tri-noctylphosphine oxide. The morphologies, compositions, crystal structures, and optical properties of the nanowires are characterized by transmission electron microscopy (TEM), powder X-ray diffraction analysis (XRD), energy-dispersive X-ray spectroscopy (EDS), and ultraviolet-visible-near-infrared (UV-Vis-NIR) absorption spectroscopy. CdTe nanowires having purposefully controlled diameters in the range of 5-11 nm are grown using cadmium n-octadecylphosphonate and tri-n-octylphosphine telluride as precursors. The nanowires adopt the wurtzite crystal structure, and grow along the [002] direction (parallel to the c axis). Blue shifts of the absorption edges relative to the bulk band-edge absorption confirm that the wires are quantum wires. The size dependence of the effective band gaps in the quantum wires are determined from the absorption spectra, and compared to the experimental results for CdTe quantum dots. In contrast to the

ii

predictions of an effective-mass approximation, particle-in-a-box model, and previous experimental results from CdSe and InP dot-wire comparisons, the effective band gaps of CdTe dots and wires of like diameter (5-11 nm) are found to be experimentally indistinguishable. The present results are explained with the assistance of density functional theory under the local-density approximation by implementing a chargepatching method (DFT-LDA-CPM). The higher-level theoretical analysis finds the general existence of a threshold diameter, above which dot and wire effective band gaps converge. The origin and magnitude of the threshold diameter is discussed. The size-dependent electronic structure of wurtzite CdTe quantum wires is determined by DFT-LDA-CPM calculations. The results of the calculations are then used to assign the size-dependent absorption spectrum of the SLS-grown CdTe quantum wires. Quantitative agreement between experiment and theory is achieved. The

absorption features comprise transitions involving the highest 25-30 valence-band states and lowest 15 conduction-band states. Individual transitions are not resolved; rather, the absorption features consist of clusters of transitions that are determined by the conduction-band energy-level spacings. The sequence, character, and spacing of the conduction-band states are strikingly consistent with the predictions of the simple effective-mass-approximation, particle-in-a-cylinder model. The model is used to

calculate the size dependence of the electron effective mass in CdTe quantum wires. Rock-salt PbSe nanowires having diameters in a wide range of 6-18 nm are grown by the Bi-catalyzed SLS mechanism. The diameter control of the nanowires is achieved by varying the reaction temperatures, the precursors and their ratios, and the Bi nanoparticles. Small diameter wires (< 10 nm) are produced at low reaction temperatures (< 180 C),

using lead carboxylate and tri-n-alkylphosphine selenide precursors with short carbon chains, and employing small-diameter Bi-catalyst nanoparticles. A thermally more stable lead alkylphosphonate precursor is required to grow large diameter (> 15 nm) wires at high reaction temperatures (~ 270 C). The mean diameter of the resulting nanowires is found to be larger than that of the corresponding initial Bi nanoparticles, suggesting significant swelling of the catalyst nanoparticles. The EDS analysis reveals the alloy nature of the catalyst nanoparticle attached to the wire end, consistent with the phase separation and recrystallization process observed in the catalyst nanoparticle under prolonged irradiation of TEM electron beam. The NIR absorption at 2280 nm observed from the 6.8-nm diameter PbSe nanowires dispersed in carbon tetrachloride shows a strong quantum confinement of AEg = 266 meV. High-quality rock-salt PbS nanowires having diameters in the range of 9-31 nm and wurtzite CdS nanowires having diameters in the range of 6.5-11 nm are successfully grown using single-source precursors (metal diethyldithiocarbamates) instead of dualsource precursors. Well-resolved NIR absorption features are observed from the

colloidal PbS nanowires, confirming strong quantum confinement. Compared with the CdS nanowires grown by the vapor-liquid-solid mechanism using the same single-source precursor, the SLS CdS nanowires produced at much lower temperatures have smaller diameters and narrower diameter distributions. The present results strongly suggest that the single-source SLS strategy may extend to other colloidal semiconductor nanowire systems that are not easily prepared by a dual-source precursor approach.

IV

ACKNOWLEDGMENTS

This dissertation would not have been completed without the help of many people. First of all, I would like to thank my research advisor, Professor William E. Buhro, for his warm encouragement and invaluable support throughout these years. During my research, I encountered various severe problems related to chemicals, which made me feel frustrated. Professor Buhro impressively encouraged me to solve problems and find new research directions independently. Moreover, Professor Buhro helped me a lot to understand the theoretical results related to the quantum-confinement-effect studies and electronic structure of CdTe quantum wires. In addition, Professor Buhro provided much-needed assistance in preparing me to find a postdoctoral position. I also express my appreciation to other members of my dissertation examination committee, Professor Daniel J. Leopold, Professor Patrick C. Gibbons, Professor Sophia E. Hayes, Professor Richard A. Loomis, and Professor John R. Bleeke, for taking time from their busy schedules to read my dissertation and participate in my defense. Professor Loomis and Professor Hayes also served on my annual research advisory committee and contributed helpful discussions and suggestions for my research. Special thanks go to Dr. Lin-Wang Wang and Dr. Joshua Schrier at Lawrence Berkeley National Lab. They performed higher-level theoretical calculations for the electronic structure of CdTe quantum wires. Special thanks are extended to Dr. John G. Glennon, who provided training in collecting near-infrared absorption spectra from colloidal dispersions of narrow band-gap

IV-VI semiconductor nanowires and contributed helpful discussions in data analyses, and Dr. Tyrone L. Daulton for the high-resolution TEM training. I would also like to thank my friends, the current and past members in the Buhro group for their experimental assistance and helpful discussions. Dr. Fudong Wang and Mr. Rui Tang provided Bi catalyst nanoparticles for the SLS growth of all the II-VI and IV-VI semiconductor nanowires investigated in this dissertation. Thanks to Dr. Ed. Hiss, Ms. Norma Taylor, and other department staff for help in many issues. The encouragement and support from my parents is of great significance for me to complete this dissertation. Otherwise, I may not have the opportunity to do interesting research at Washington University. Finally, very special thanks go to my wife, Zhichun Lu, for always providing much emotional support and believing that what I did is potentially good stuff.

VI

TABLE OF CONTENTS
Abstract of the Dissertation Acknowledgments Table of Contents List of Figures List of Tables ii v vii ix xviii

Introduction to the Dissertation Refenences Chapter 1: SLS Growth of High-Quality CdTe Quantum Wires Introduction Results Discussion Conclusion Experimental Section References Chapter 2: Similarity of the Effective Band Gaps of CdTe Quantum Wires to Those of Equidiameter CdTe Quantum Dots Introduction Results Discussion Conclusion References Chapter 3: Electronic Structure of Colloidal CdTe Quantum Wires Introduction Computational Methods Results and Discussion Conclusion References

1 15 20 21 27 53 57 58 64

67 68 69 77 93 94 96 97 98 99 114 114

vn

Chapter 4: SLS Growth of Diameter-Controlled PbSe Nanowires Introduction Results Discussion Conclusion Experimental Section References Chapter 5: SLS Growth of II-VI and IV-VI Metal Sulfide Semiconductor Nanowires: A Single-Source Precursor Approach Introduction Results Discussion Conclusion Experimental Section References

117 118 120 147 150 151 156

159 160 163 188 189 190 197

viii

LIST OF FIGURES
Introduction: Figure 1-1 Number of publications containing the concept of 'nanowires' and 'semiconductor nanowires' by year, determined from a CAS SciFinder search

Figure 1-2 Figure 1-3 Figure 1-4 Figure 1-5

Schematic diagrams illustrating the nanowire growth via VLS and SLS mechanisms from catalyst nanoparticles 5 Schematic diagram illustrating the diameter control of nanowires from catalyst nanoparticles with various sizes Experimental setup for the SLS growth of semiconductor nanowires Plot of AEg vs. l/d2 for various quantum systems having the same composition 7 8 12

Chapter 1; Figure 1-1 Figure 1-2 Plot of AEg vs. l/d2 for CdTe quantum dots and self-assembled CdTe quantum wires reported by Kotov and coauthors 23

Representative low- and high-magnification TEM images of selfassembled CdTe nanostructures synthesized by self-organization of CdTe quantum dots 24 Representative optical properties of the self assembled CdTe nanostructures 25

Figure 1-3 Figure 1-4 Figure 1-5 Figure 1-6 Figure 1-7 Figure 1-8

Schematic illustration of the designed synthetic route for the solutionliquid-solid (SLS) growth of CdTe quantum wires 27 Bulk Bi-Cd binary phase diagram 28

Representative TEM images of CdTe nanostructures synthesized in 1octadecene (ODE) 29 TEM images of CdTe nanostructures synthesized using various Cd/Te ratios in TOPO 30 TEM images of CdTe nanostructures synthesized using high concentration of TOPTe and various Cd/Te ratios in TOPO 31
IX

Figure 1-9

TEM images of CdTe nanowires synthesized using low concentration of TOPTe and various Cd/Te ratios in TOPO 32

Figure 1-10 TEM images of CdTe nanostructures synthesized using cadmium stearate and various Cd/Te ratios in TOPO 33 Figure 1-11 Representative low- and high-magnification TEM images of CdTe nanowires synthesized using Cd(OA)2 and TOPTe precursors in TOPO..34 Figure 1-12 Representative diameter-distribution histograms for CdTe nanowires synthesized using Cd(OA>2 and TOPTe as precursors in TOPO 35

Figure 1-13 TEM images of CdTe nanowires synthesized using Cd(OA)2 and TBPTe (0.02mmol) as precursors in TOPO 36 Figure 1-14 Figure 1-15 Figure 1-16 Figure 1-17 Representative low-magnification TEM images of CdTe nanowires synthesized using Cd(OA>2 and TBPTe as precursors in TOPO 37

TEM images of 11.1-nm CdTe nanowires synthesized using 15.9-nm Bi nanoparticles in TOPO at 320 C 37 Representative low-magnification TEM images of high-quality CdTe quantum wires synthesized from Cd(ODPA) and TOPTe in TOPO 39

Representative high-magnification TEM images of high-quality CdTe quantum wires synthesized from Cd(ODPA) and TOPTe in TOPO 40

Figure 1-18 Representative diameter-distribution histograms for CdTe quantum wires synthesized from Cd(ODPA) and TOPTe in TOPO 41 Figure 1-19 Figure 1-20 High-resolution TEM image from a CdTe nanowire bundle Plot of CdTe quantum-wire diameter vs. initial Bi nanoparticle diameter 42 43

Figure 1-21 TEM images and diameter-distribution histogram for large-diameter CdTe nanowires synthesized from ~40-nm Bi nanoparticles 43 Figure 1-22 Powder X-ray diffraction (XRD) pattern of CdTe quantum wires 44

Figure 1-23 Lattice-resolved high-resolution TEM (HRTEM) image of a 10-nm CdTe quantum wire 45 Figure 1-24 Lattice-resolved high-resolution TEM image of a single CdTe quantum wire having a catalyst nanoparticle attached to the wire tip 46

Figure 1-25

Energy-dispersive X-ray spectra (EDS) collected from the catalyst nanoparticle and the wire

47

Figure 1-26

Arrangement of atoms projected to the (001) planes in rhombohedral Bi and wurtzite CdTe 48 Absorption spectrum of 7.3-nm CdTe quantum wires diluted in toluene..48 Representative fitting of the absorption spectrum of 7.3-nm CdTe quantum wires 50 Representative absorption spectra of CdTe quantum wires of various diameters in the range of 5.3-10.2 nm and their lowest-energy excitonic peaks extracted by nonlinear least-squares fitting 51 Room temperature photoluminescence spectra of the 5.4-nm and 7.4-nm CdTe quantum wires synthesized from Cd(ODPA), and diluted with toluene 52 Representative room temperature photoluminescence spectra of the asprepared and methanol-washed 5.5-nm CdTe quantum wires 53 Schematic illustration of the proposed four-stage SLS growth mechanism for CdTe quantum wires 56

Figure 1-27 Figure 1-28

Figure 1-29

Figure 1-30

Figure 1-31

Figure 1-32

Chapter 2; Figure 2-1 Experimental CdTe quantum wire and quantum dot data plotted as AE% vs. d2 71 Diameter dependence of quantum confinement energies of CdTe quantum wires 72 Experimental CdTe quantum-wire and quantum-dot data, and theoretical CdTe quantum-wire and quantum-dot curves plotted as AEg vs. d2 74 Linear fits to theoretical CdTe quantum-wire and quantum-dot data 76

Figure 2-2

Figure 2-3

Figure 2-4 Figure 2-5 Figure 2-6

Diameter dependence of quantum confinement energies of various II-VI quantum dots and quantum wires 79 Diameter dependence of quantum confinement energies of various III-V quantum dots and quantum wires 80

XI

Figure 2-7

Plot of theoretical quantum confinement energy (AEg) vs. d2 for various II-VI quantum dots and quantum wires 81 Plot of theoretical quantum confinement energy (AEg) vs. d2 for various III-V quantum dots and quantum wires 82 Plot of threshold diameter vs. /WAore at various (<%0t- QWire) values 83

Figure 2-8

Figure 2-9 Figure 2-10

Diameter dependence of the estimated exciton binding energies in CdTe quantum dots and quantum wires 85 Theoretical ZnTe quantum-wire and quantum-dot curves plotted as AEg vs d2 87 Log-normal plot of threshold diameter vs. calculated bulk exciton binding energy for various II-VI and III-V semiconductors 89 Diameter dependence of the experimental effective band-gap energy (AE&) of CdTe quantum dots and quantum wires, and a plot of AEg vs. d2 for CdTe dots and wires 92

Figure 2-11

Figure 2-12

Figure 2-13

Chapter 3; Figure 3-1 Representative experimental absorption spectra of CdTe quantum wires with various diameters 100 Plots of experimentally observed absorption features (relative to energy of the first excited state) vs. energy of the first excited state for CdTe quantum wires of various diameters 101 Calculated imaginary dielectric constant (optical absorption) as a function of energy, for 5.3 nm, 7.3 nm, and 10.2 nm diameter wurtzite CdTe quantum wires 102 Experimental absorption spectra and calculated absorption spectra with average polarization of CdTe quantum wires with diameters of 5.3 nm, 7.3 nm, and 10.2 nm 103 Comparison of the experimentally observed absorption features and the calculated absorption features 104 Calculated optical absorption for transitions at the T-point for the 5.3 nm diameter wurtzite CdTe quantum wire 105

Figure 3-2

Figure 3-3

Figure 3-4

Figure 3-5

Figure 3-6

xii

Figure 3-7

The energy-level diagram for the 5.3 nm diameter wurtzite CdTe quantum wire and plots of the electron-and hole-density distributions for the conduction band and valence band states 106 Plots of the electron-density distributions for the CB7 state and the (CB10, CB11) degenerate pair of states for the 5.3-nm diameter quantum wire shown in Figure 3-7 109 Plots of the CB-state energy (Enj) of CdTe quantum wires of various diameters at the T point vs. ^

Figure 3-8

Figure 3-9

111

Figure 3-10

Plot of (lPe - lSe), (lUe - XLe) vs. diameter of CdTe quantum dots and quantum wires 113

Chapter 4: Figure 4-1 Schematic illustration of the designed synthetic route for the solutionliquid-solid (SLS) growth of PbSe nanowires 120 Bulk Bi-Pb binary phase diagram 121

Figure 4-2 Figure 4-3

Representative TEM images of PbSe nanostructures synthesized in poly-1decene and phenyl ether 123 Representative low-magnification TEM images of high-quality PbSe nanowires of various diameters

Figure 4-4

124

Figure 4-5

Representative high-magnification TEM images of high-quality PbSe nanowires of various diameters 125 Representative diameter-distribution histograms for PbSe nanowires of various diameters 126 127

Figure 4-6

Figure 4-7 Figure 4-8 Figure 4-9 Figure 4-10

Low-magnification TEM images of PbSe nanowire bundles

TEM images of PbSe nanostructures synthesized using 99% TOPO with different batch numbers purchased from Aldrich 128 TEM images of PbSe nanostructures synthesized from lead stearate and TOPSe at 244 C in TOPO 128 TEM images of PbSe nanowires synthesized using Bi nanoparticles of various diameters at 240-250 C in TOPO 129

xiii

Figure 4-11

TEM images of PbSe nanowires synthesized from lead oleate, TOPSe, and 4.3-nm Bi nanoparticles at different reaction temperatures in TOPO 131 TEM images of PbSe nanowires synthesized from 4.3-nm Bi nanoparticles at -180 C in TOPO 132 TEM images of PbSe nanowires synthesized from lead oleate and TBPSe at low reaction temperatures 133 TEM images of PbSe nanowires synthesized from 4.3-nm Bi nanoparticles at 160 C in TOPO 134 TEM images of PbSe nanowires synthesized from 4.3-nm Bi nanoparticles, lead octanoate, and TBPSe at different reaction temperatures in TOPO TEM images of PbSe nanostructures synthesized from 6.4-nm Bi nanoparticles at 170 C in TOPO TEM images of PbSe nanowires synthesized from 13.5-nm Bi nanoparticles at 275 C in TOPO

Figure 4-12

Figure 4-13

Figure 4-14

Figure 4-15

135

Figure 4-16

136

Figure 4-17

137

Figure 4-18

TEM images of PbSe nanowires synthesized from large Bi nanoparticles of various diameters at temperatures in the range of 270-275 C in TOPO 138 TEM images of PbSe nanowires synthesized at high precursor concentrations at 272 C in TOPO

Figure 4-19

139

Figure 4-20

TEM images of PbSe nanowires synthesized using various amount of 13.5-nm Bi stock solutions at 275 C in TOPO 140 Representative TEM images of PbSe nanostructures synthesized in the absence of Bi nanoparticles in TOPO 142 Powder X-ray diffraction (XRD) pattern of PbSe nanowires 143

Figure 4-21

Figure 4-22 Figure 4-23

High-resolution TEM image of a 15-nm PbSe nanowire having a catalyst nanoparticle attached to the wire tip 143 High-resolution TEM images of a 15-nm PbSe nanowire, recorded with prolonged exposure to the electron beam 144 Energy-dispersive X-ray spectra (EDS) recorded from the catalyst nanoparticle and the wire

Figure 4-24 Figure 4-25

145

xiv

Figure 4-26 Near-infrared spectrum of 6.8-nm PbSe nanowires dispersed in carbon tetrachloride 146 Figure 4-27 Plot of the PbSe nanowire diameter vs. the initial Bi nanoparticle diameter 148 149

Figure 4-28 Plot of PbSe nanowire diameter vs. reaction temperature for various precursor sets Chapter 5: Figure 5-1 Figure 5-2 Figure 5-3 Figure 5-4 Figure 5-5 Figure 5-6 Figure 5-7 PbS nanostructures synthesized from dual-source precursors CdS nanostructures synthesized from dual-source precursors

161 162

Schematic illustration of the designed synthetic route for the solutionliquid-solid (SLS) growth of PbS nanowires 163 TEM images of PbS nanostructures synthesized in 1-octadecene and phenyl ether 164 Representative low-magnification TEM images of PbS nanowires of various diameters 165

Representative high-magnification TEM images of PbS nanowires of various diameters 166 Representative diameter-distribution histograms for PbS nanowires of various diameters 167 168

Figure 5-8 Figure 5-9

Additional high-magnification TEM image of PbS nanowires

TEM images of PbS nanostructures synthesized in the absence of Bi nanoparticles Figure 5-10 TEM images of PbS nanowires synthesized in a mixed TOPO/TOP solvent Figure 5-11 Diameter control of PbS nanowires synthesized using 10.3-nm Bi nanoparticles at various reaction temperatures

168 169 169

Figure 5-12 TEM images of PbS nanowires synthesized using 4.3-nm Bi nanoparticles at various reaction temperatures 170

xv

Figure 5-13

TEM images of PbS nanostructures synthesized using large Bi nanoparticles having various mean diameters TEM images of PbS nanostructures synthesized in the presence of TDPA

171

Figure 5-14

172

Figure 5-15 Figure 5-16

TEM images of PbS nanostructures synthesized in the presence HDA ..173 Representative TEM images of PbS nanowires synthesized using the TOP dispersion of Pb(S2CNEt2>2 as the injection mixture 173 TEM images of PbS nanostructures synthesized using 99% TOPO with different batch numbers purchased from Aldrich 174 175 175

Figure 5-17

Figure 5-18 Figure 5-19 Figure 5-20 Figure 5-21

Powder X-ray diffraction (XRD) pattern of PbS nanowires Lattice-resolved HRTEM image of a 15-nm diameter PbS nanowire

Energy-dispersive X-ray spectra (EDS) recorded from the catalyst nanoparticle and the wire 176 Near-infrared absorption spectra of colloidal PbS nanowires of various diameters dispersed in carbon tetrachloride 177 Nonlinear least-squares fitting of the near-infrared absorption spectra of PbS nanowires with diameters of 10.2 nm and 11.2-nm 178 Representative low-magnification TEM images of CdS nanowires of various diameters

Figure 5-22

Figure 5-23

179

Figure 5-24

Representative high-magnification TEM images of CdS nanowires of various diameters 180 Representative diameter-distribution histograms for CdS nanowires of various diameters 181 TEM images and UV-Vis absorption spectra of CdS nanostructures synthesized using various amount of TDPA in TOPO TEM images of CdS nanostructures synthesized by injecting an oleylamine solution of Cd(S2CNEt2)2 into TOPO 182 183 183

Figure 5-25

Figure 5-26

181

Figure 5-27

Figure 5-28 Figure 5-29

Powder X-ray diffraction (XRD) pattern of CdS nanowires Lattice-resolved HRTEM image of a 9-nm diameter CdS nanowire

xvi

Figure 5-30 Representative TEM images of ZnS nanostructures synthesized using single-source precursor Zn(S2CNEt2)2 in TOPO 184 Figure 5-31 EM images of PbTe nanostructures synthesized using lead tetradecyphosphonate and TOPTe as precursors in TOPO Figure 5-32 Representative TEM images of PbTe nanostructures synthesized using singe-source precursor PbTe03 Figure 5-33 HRTEM image of a 16-nmPbTenanowire 186 187 185

Figure 5-34 HRTEM images of a ~15-nm PbTe nanowire recorded with prolonged exposure to the electron beam 187

xvu

LIST OF TABLES
Introduction; Table 1-1 Size and shape dependences of quantum confinement energy for various model quantum systems predicted by EMA-PIB model 11

Chapter 1: Table 1-1 Table 1-2 Table 1-3 Bulk band gaps of wurtzite cadmium chalcogenide semiconductors Cell parameters of rhombohedral Bi and wurtzite CdTe 21 48

Band gap energies and quantum confinement energies for CdTe quantum wires of various mean diameters 50 Synthetic conditions and results for CdTe quantum wires and other morphologies

Table 1-4

61

Chapter 2; Table 2-1 First excitonic absorption peak positions and quantum confinement energies of CdTe quantum dots of various diameters

70

Table 2-2

a and fi values for various II-VI and III-V semiconductor quantum dots and quantum wires 77 (Otar-aw), /WArne values, and threshold diameters for various II-VI and III-V quantum systems 83 Bohr radii, threshold diameters, and calculated bulk exciton-binding energies for various II-VI and III-V systems

Table 2-3

Table 2-4

88

Chapter 3: Table 3-1 Center energies of the absorption features (E\-E) of CdTe quantum wires of various mean diameters 99 Zeros (roots) of the cylindrical Bessel functions Zeros (roots) of the spherical Bessel functions 108 112

Table 3-2 Table 3-3

xviii

Chapter 4: Table 4-1 Synthetic conditions and results for PbSe nanowires and other morphologies 153

Chapter 5: Table 5-1 Table 5-2 Table 5-3 Band gaps and quantum confinement energies of PbS nanowires of various diameters and bulk PbS Synthetic conditions and results for PbS nanowires and other morphologies Synthetic conditions and results for CdS nanowires and other morphologies 178 193 195

xix

Introduction

Scope of the dissertation. This dissertation reports the solution-liquid-solid (SLS) growth of colloidal, diameter-controlled, single-crystalline CdTe (Chapter 1), PbSe (Chapter 4), PbS (Chapter 5), and CdS (Chapter 5) nanowires (or quantum wires) from nearly-monodisperse Bi catalyst nanoparticles and the quantum-confmement-effect studies and electronic structure calculations associated with the optical characterization of these nanowires. High-quality CdTe and PbSe nanowires are produced using the dualsource precursor approach, which is commonly employed for the SLS growth of binary semiconductor nanowires, but unfortunately demonstrated to be problematic for PbS and CdS nanowires due to the difficulty of finding dual precursors with appropriately balanced reactivities. Alternatively, a single-source precursor approach for the SLS growth of high-quality PbS and CdS nanowires is described. The room-temperature absorption spectra of CdTe quantum wires are richly structured. We first investigate the band-edge absorption features through a detailed comparison of the size dependence of the effective band gap, which is defined to be the energy necessary to excite the electron and hole into their lowest-energy confinement states, in colloidal CdTe quantum wires and quantum dots (Chapter 2). In contrast to the effective-mass-approximation, particle-in-a-box (EMA-PIB) prediction and the wire-dot comparisons from the CdSe and InP systems, we report experimentally indistinguishable effective band gaps of CdTe quantum wires and quantum dots having the same diameters in the range of 5-11 nm. A higher-level theoretical analysis provides new insights into the comparative effective band gaps of quantum dots and wires. We then report for the first time the assignment of the theoretical excitonic transitions predicted by higher-level electronic-structure calculations to the multiple absorption

features observed in the experimental absorption spectra of the colloidal, SLS-grown CdTe quantum wires (Chapter 3). We show that the conduction-band energy levels follow the sequence predicted by the simple effective-mass-approximation, particle-in-acylinder (PIC) model. The close correspondence of the higher-level electronic structure and the simple PIC model provides a conceptually attractive framework for analyzing the electronic structure of semiconductor quantum wires. Why semiconductor nanowires? Nanoscience and nanotechnology are typically relevant to fundamental studies and practical applications of materials on the length scale between one nanometer and a few tens to perhaps one hundred nanometers. Scientific historians often refer the beginning of this currently very active field back to 1959, the year that physicist Richard P. Feynman gave a famous lecture entitled 'There's Plenty of Room at the Bottom' to the American Physical Society at Caltech.1 He predicted that materials would behave differently as they were scaled down into the nanoscale, and people could eventually take advantages of the novel properties in broad fields. After that, especially over the past two decades, tremendous efforts have been made by chemists to synthesize nanostructures with tunable sizes, controllable morphologies, and various compositions through the so-called "bottom-up" chemical approaches starting from molecules.2"19 It has been well established that the physical properties of nanoscale materials may significantly differ from those of the bulk counterparts, and the engineering of the material properties is possible by controlling the size, shape, and composition of materials.20"23 More importantly, it has been demonstrated that various nanostructures are potentially promising key building blocks for the construction of nanoscale devices.24"29

a Nanowires Semiconductornanowires 4500 -A

in 4000 a

c o

3500

.2 3000 Y 3 2500

a.
* : 2000 1500- 7

1000 - / 500-/

~ff
MM111

Tit

O r - N W n U l l p N C p q i O r N W ^ t f l l p h . O O O* Of) 0)0)0)0>wO)0)Cllo)0>0

Year

Figure 1-1. Number of publications containing the concept of 'nanowires' and 'semiconductor nanowires' by year, determined from a CAS SciFinder search.

Among various morphologies currently investigated, nanowires represent a very important category of nanostructures due to their unique one-dimensionality (ID) and possible quantum confinement in the other two dimensions.15,23 The increasing research interest is reflected by the rapid growth of the number of publications related to nanowires (Figure 1-1). Among various categories of nanowires, semiconductor nanowires are of special interest due to their attractive electric and optical properties, which make them potentially ideal building blocks for fabrication of various nanoscale devices including lightemitting-diodes (LEDs),30 field-effect-transistors (FETs),30 logic gates,31 lasers,32,33

photodetectors,34 sensors for gases and bio-molecules,35 solar cells,36,37 and waveguides.38 Moreover, semiconductor nanowires are one of the ideal model quantum systems for investigating fundamental physical properties such as quantum confinement in two dimensions.

Catalyst-based growth mechanisms for semiconductor nanowires: VLS vs. SLS. The first essential step toward reliable fundamental studies and ultimate applications of semiconductor nanowires is to synthesize high-quality wires in a controllable way. To this end, several main strategies have been established to produce semiconductor nanowires with various compositions, including (i) the hard-template method, which makes use of the ID nanostructures (usually pores) of the template to direct the formation of nanowires, (ii) the growth-directing ligand approach that takes advantage of the selective binding of the ligand molecules to specific crystal facets to preferentially direct the ID growth of nanostructures, (iii) the oriented-attachment approach, by which the ID nanowires are formed via the self-assembly and reorganization of OD nanostructures, and (iv) the catalyst-assisted growth strategies that employ catalyst nanoparticles to create a liquid-solid interface where the nanowire grows.15
(a) Gaseous precursor(s) By-products (b)

Metallo-organic precursors) By-products

Growth direction

Nanowire

%J

',

substrate

Figure 1-2. Schematic diagrams illustrating the nanowire growth via VLS (a) and SLS (b) mechanisms from catalyst nanoparticles. Among these synthetic strategies, the catalyst-assisted vapor-liquid-solid (VLS) approach seems to be the most extensively employed method to produce semiconductor nanowires in a wide range of compositions so far. 19 By employing an in situ high-

temperature TEM technique, three stages for the VLS growth of semiconductor nanowires were clearly identified to be metal alloying, crystal nucleation and axial growth.39 During the metal alloying stage, semiconductor-metal alloy nanodroplets are formed with increasing amount of precursor decomposition and dissolution in metal catalyst nanoparticles. When the alloy nanodroplets reach a supersaturation point,

continued feeding of semiconductor precursor(s) results in nucleation of solid semiconductor phase, thus a liquid/solid interface forms. Because it's energetically more favorable for the incoming semiconductor precursors to diffuse to and react at the existing solid/liquid interface than to form a secondary nucleation site, ID axial growth of semiconductor nanowires can be readily achieved in a controllable way. The VLS growth mechanism is schematically illustrated in Figure I-2a. The VLS mechanism was proposed as early as 1964.40 At that time, no practical method was developed to purposefully produce metal catalyst nanoparticles. As a consequence, the minimum radius of semiconductor ID structures that may be achieved by the VLS mechanism is mainly governed by equilibrium thermodynamics shown in equation 1-1, where OLv is the liquid-vapor surface free energy, VL is the molar volume, R is the molar gas constant, T is the temperature, and a is the vapor phase supersaturation. By substituting typical values into equation 1-1, r^n is determined to be on the order of -200 nm, suggesting that small-diameter nanowires exhibiting interesting novel properties should not be expected based on the VLS approach. However, this limitation of equilibrium thermodynamics has been overcome by using laser ablation technique and metal catalyst nanoparticles having well-controlled diameters.41 Nevertheless, the VLS

approach generally affords nanowires with diameters larger than 10 nm,19'42 which is above the strong quantum confinement regime for most semiconductor compositions.
r. =
2<JLVVL

(1-1)

To systematically study the quantum confinement effects in two dimensions, it is desirable to produce nanowires having controlled diameters within the intermediate to strong quantum-confinement regime (usually smaller than 10 nm). As noted above, the VLS approach only produces relatively large diameter nanowires, which should be related to very high reaction temperatures employed by this method.19 In the early 1990s, an analogous solution-based wire-growth mechanism named SLS mechanism (Figure I2b) was initially reported by our group.43 The reaction temperatures employed in the SLS approach are typically much lower than those in the VLS approach. Nevertheless, the SLS approach did not produce small-diameter, uniform nanowires until a synthetic method for preparing nearly monodisperse catalyst nanoparticles, such as In and Bi, was developed.44 It has been subsequently found that Bi nanoparticles seem to be the most generally useful catalyst for the diameter-controlled SLS growth of semiconductor nanowires of various compositions.6

Figure 1-3. Schematic diagram illustrating the diameter control of nanowires from catalyst nanoparticles with various sizes.

Ideally, for a given semiconductor composition the diameter control of the nanowires by the SLS mechanism can be achieved by employing catalyst nanoparticles with various sizes (Figure 1-3); that is, the diameter of the resulting nanowires is determined by the initial diameter of the catalyst nanoparticles. However, the actual situations may be complex due to several reasons. First, the catalyst nanoparticles with different sizes have different physical properties, such as melting point,45 which may require different reaction conditions. Second, agglomeration of catalyst nanoparticles may happen,

especially for small-diameter catalyst nanoparticles, which have higher tendency to agglomerate at relatively high reaction temperatures. This is believed to be the main limitation for producing nanowires having diameters less than ~5 nm. Third, catalyst nanoparticles with large diameters may also be problematic due to their low catalytic activities. Nevertheless, the SLS syntheses generally afford high-quality nanowires with diameters in the range of 5-20 nm, which facilitates the quantum confinement studies for many semiconductor compositions.

Syringe r Catalyst nanoparticles I Precursor Septum stopper

Reaction tube

Salt bath Hot plate stirrer -

Solvent (Precursor) (Surfactants)

Figure 1-4. Experimental setup for the SLS growth of semiconductor nanowires.

Compared with the VLS approach, the SLS synthesis of nanowires also offers other advantages, including narrow diameter distributions, control of surface passivation, and nanowire solubility.6 Additionally, the experimental setup for the SLS growth of

semiconductor nanowires is very simple, thus allows ease of implementation (Figure 1-4). SLS growth of CdTe, CdS, PbSe, and PbS nanowires: Dual-source precursor approach vs. single-source precursor approach. Among cadmium chalcogenide

nanowires, the SLS growth of high-quality CdSe nanowires has been previously reported.46'47 However, it was believed that the SLS growth of CdS and CdTe nanowires was more challenging from the synthetic point of view due to lack of appropriate precursors.48 In Chapter 1 and Chapter 5 we report our efforts to produce high-quality CdTe and CdS nanowires by the dual-source and single-source precursor SLS approach, respectively. Recently, narrow band-gap IV-VI semiconductor nanostructures are highly desirable due to their potential for various photovoltaic, thermovoltaic, and thermoelectric devices as well as infrared-based optical applications.3,12 The recent discovery of efficient carrier multiplication in lead chalcogenide semiconductor nanocrystals has triggered special scientific and industrial interest in the study of narrow-gap nanostructures because they are potentially the most promising materials for thin-film photovoltaics.49'50 In Chapter 4 and Chapter 5 we report the dual-source and single-source precursor approach for the SLS growth of PbSe and PbS nanowires, respectively. The high quality of these wires allows us to report the near infrared absorption spectra from colloidal dispersions of PbSe and PbS nanowires with relatively small diameters.

Although the SLS growth has been applied to a wide range of

TJ_VI, 46 ' 4/ ' M " M

m-

y 43,55-59 ^ j jy_yj60 ^ a j y nanowire compositions, the method is not yet generally applicable, as each new composition requires extensive trial-and-error experimentation to identify appropriate precursors and reaction conditions. In several cases, such as for metal-sulfide semiconductor nanowires, the dual-source-precursor SLS approach has failed to produce highly crystalline, diameter-controlled nanowires. Alternatively, we report the use of single-source precursors, in which the metallic and nonmetallic semiconductor constituents are combined in a single molecule, to solve the reactivitybalance problem, resulting in the successful SLS growth of high-quality PbS and CdS nanowires. The single-source approach has the potential to greatly extend the generality of the SLS method for semiconductor-nanowire synthesis. Quantum confinement of excitons in model quantum systems: EMA-PIB model. The excitation of a given semiconductor material generates excitons, which are electronhole pairs. In bulk semiconductors, there is an equilibrium distance called the Bohr radius between the oppositely charged electron and hole. When the size of a

semiconductor falls in the length scale less than or comparable to twice its exciton Bohr radius, quantum confinement of the excitons takes place, which is one of the most compelling properties associated with semiconductor nanostructures. There are three model quantum systems: the quantum well, quantum wire (cylindrical), and quantum dot (spherical), which are ID-, 2D-, and 3D-confined nanostructures, respectively. It has been well-known that the effective band gap (Eg) in semiconductor nanostructures increases as size decreases.20 The simple EMA-PIB model predicts that the quantum confinement energy (Ag), which is the increase in the effective

10

band-gap energy over the bulk value, is proportional to the inverse square of the diameter or thickness (l/a ) of the quantum systems.61"65 Importantly, the magnitude of Ag is different for quantum wells, wires, and dots having the same diameters or thickness. The shape and size dependencies of quantum confinement energy for various quantum systems are collected in Table 1-1, and illustrated in Figure 1-5. According to the Ag expressions (Table 1-1), the slope ratio between the well, wire, and dot lines in the AESvs.-l/d2 plot is AWeE:^wire^dot = 1.00/2.34/4.00 for a given semiconductor. In other words, at a given diameter or thickness, the 3D-confined quantum dot shows the largest quantum confinement, whereas the 2D-confined quantum wire exhibits less quantum confinement due to loss of one confinement dimension, and the ID-confined quantum well shows the smallest quantum confinement due to loss of two confinement dimensions.

Table 1-1. Shape and size dependences of quantum confinement energy (AEg) for various model quantum systems predicted by EMA-PIB model.

Quantum system Quantum well Geometrical dimensions Confinement dimensions Ag

2 1 2( + )
%d

Quantum wire
1 2 2.34 h2 .1
** &d
(

Quantum dot
0 3

1.
+

rri

r\

* h2 , 1 45-( + U2 me TTL

Note: In the AEg expressions, d is the diameter or thickness of the quantum systems, h is the Planck's constant, and me and mh are the effective masses of the electron and hole, respectively.

11

Figure 1-5. Plot of AEg vs. 1/d2 for various quantum systems having the same composition. The relative slopes of the dot, wire and well lines are based on the EMAPIB predictions (see Table 1-1). The slope ratio is determined to be Aweii^4wire:Adot = 1.00/2.34/4.00.

The EMA-PIB model is overly simple due to its assumptions of size-independent bulk effective masses, parabolic band shapes, and infinite potential barriers at the surfaces of quantum systems. Additionally, the model ignores the Coulomb interaction within electron-hole pairs. Thus, this simple model does not correctly predict the

absolute slope ratios and non-zero intercepts. Nevertheless, the predicted wire-dot slope ratio Amie'Adot of 0.585 has been experimentally confirmed in InP and CdSe quantum systems,61"63 and the predicted well-wire slope ratio ^weiMwke of 0.427 has been experimentally confirmed in GaAs quantum systems.66 In Chapter 2, we extend the wire-dot comparison to the CdTe quantum systems. However, an Awire:A<iot slope ratio near unity was obtained, indicating a convergence of the dot and wire effective band gaps, over the diameter regime studied (5-11 nm). The result contrasts to the EMA-PIB prediction and the wire-dot comparisons from the CdSe and InP systems. This is the first direct wire-dot comparison that suggests the breakdown

12

of the EMA-PIB model. To understand the near-unity wire-dot slope ratio for CdTe system, we thus sought to interpret the experimental results with the assistance of a higher-level theoretical analysis. Higher-level theoretical calculations of quantum confinement energies for quantum dots and quantum wires: DFT-LDA-CPM. Recently, Li and Wang

employed the density functional theory under the local-density approximation by implementing a charge-patching method (DFT-LDA-CPM) to calculate the size dependence of the effective band gaps in a series of quantum dots and wires.65 The DFTLDA-CPM is a combined method, but essentially the DFT calculation with ab initio accuracy. In this method, partially charged pseudohydrogen atoms are used to ideally passivate the surfaces of quantum systems so that the surface dangling bond states are removed. The bulk effective masses, which are critical to the quantum confinement effect, are corrected to agree with experimental values by modifying the LDA band structures. The newly developed charge-patching method is employed to reproduce the LDA charge densities without doing direct LDA calculations, but giving essentially the same results as a traditional self-consistent first-principles LDA calculation. Thus, it dramatically saves computational time for thousand-atom calculations. Importantly, the electron-hole Coulomb interactions in quantum dots of various compositions have been calculated to match experimental results. Although the DFT-LDA-CPM calculations for quantum wires completely neglected the Coulomb contributions, good agreement between theoretical and experimental effective band gaps has been achieved, indicating significantly smaller electron-hole Coulomb energies in wires than those in the corresponding dots ,65

13

The DFT-LDA-CPM method calculated the effective band gaps of dots and wires over a range of specific diameters. The results were fit with a general form of AEg = fi/da, where d is the diameter of the quantum dots or wires, a and 0 are fitting parameters depending on both the composition and the shape of quantum systems.65 The DFT-LDACPM results indicate that the dot and wire quantum confinement energies do not scale with lid2, but rather with \lda where oris in the range of 1-2,65 differing from the value determined by the simple EMA-PIB model. In Chapter 2, we report that the DFT-LDA-CPM results for CdTe quantum systems afford a threshold diameter d^ of 5.4 nm, which divides the AEs-vs.-(l/d2) plot into two regimes: a convergence regime above do, in which dot and wire effective band gaps are effectively indistinguishable, and a divergence regime below dth in which the Awn-e^dot slope ratios approximate the EMA-PIB value of 0.6. As noted above, our experimental wire-dot comparison for the CdTe quantum systems is limited to the diameter range of 511 nm, which is above the threshold diameter, thus, the dot-wire comparison for the CdTe systems are made in the convergence regime. We further discuss that this tworegime behavior may be general for semiconductor quantum dots and wires, and propose that the convergence regime results from differing electron-hole Coulomb energies in dots and wires. Electronic structure of colloidal quantum wires: theoretical calculations vs. experimental data. Studies of the electronic structure of colloidal semiconductor

quantum dots have benefited greatly from comparisons of calculated energy-level diagrams and calculated optical spectra with experimental spectroscopic data, generally from photoluminescence excitation spectroscopy.67"72 In contrast, theory-experiment

14

comparative studies of the electronic structure of colloidal semiconductor quantum wires, to our knowledge, have not appeared yet due to lack of high-quality quantum-wire specimen and much lower quantum yield of quantum wires than that of quantum dots. Interestingly, we observed highly structured absorption spectra of the high-quality SLS-grown CdTe quantum wires, which allow us to study the electronic structure of colloidal CdTe quantum wires by comparing calculated optical spectra with experimental absorption spectra. In Chapter 3, we employ the DFT-LDA-CPM (see above) method to calculate the optical spectra of CdTe quantum wires. The results of the calculations are then used to assign the size-dependent absorption spectrum of colloidal CdTe quantum wires. Quantitative agreement between experimental and theoretical spectra corroborates the electronic structure of CdTe quantum wires. We report that the sequence, character, and spacing of the conduction-band states are strikingly consistent with the predictions of the simple effective-mass-approximation, particle-in-a-cylinder (PIC) model. The close correspondence of the DFT-LDA-CPM results and the simple PIC model provides a conceptually attractive framework for analyzing the electronic structure of semiconductor quantum wires.

References (1) (2) Feynman, R. P. Eng. Sci. 1960, 23, 22-36 (transcript of the lecture given in Decemeber 1959). Park, J.; Joo, J.; Kwon, S. G.; Jang, Y.; Hyeon, T. Angew. Chem. Int. Ed. 2007, 46, 4630-4660. Rogach, A. L.; Eychmueller, A.; Hickey, S. G.; Kershaw, S. V. Small 2007, 3, 536-557.

(3)

15

(4) (5)

Wang, X.; Li, Y. Chem. Comm. 2007,2901-2910. Rao, C. N. R.; Vivekchand, S. R. C ; Biswas, K.; Govindaraj, A. Dalton Trans. 2007, 3728-3749. Wang, R; Dong, A.; Sun, J.; Tang, R.; Yu, H.; Buhro, W. E. Inorg. Chem. 2006, 45,7511-7521. Jun, Y.-w.; Choi, J.-s.; Cheon, J. Angew. Chem. Int. Ed. 2006, 45, 3414-3439. Cozzoli, P. D.; Pellegrino, T.; Manna, L. Chem. Soc. Rev. 2006, 35, 1195-1208. Kumar, S.; Nann, T. Small 2006, 2, 316-329. Yin, Y.; Alivisatos, A. P. Nature 2005,437, 664-670. Jun, Y. w.; Lee, J. H.; Choi, J. s.; Cheon, J. J. Phys. Chem. B 2005, 109, 1479514806. Sargent, E. H. Adv. Mater. 2005, 17, 515-522. Lisiecki, I. J. Phys. Chem. B 2005,109,12231-12244. Shah, P. S.; Hanrath, T.; Johnston, K. P.; Korgel, B. A. J. Phys. Chem. B 2004, 108, 9574-9587. Xia, Y.; Yang, P.; Sun, Y.; Wu, Y.; Mayers, B.; Gates, B.; Yin, Y.; Kim, R; Yan, H. Adv. Mater. 2003, 15, 353-389. Sun, Y.; Xia, Y. Science 2002, 298, 2176-2179. Puntes, V. R; Krishnan, K. M.; Alivisatos, A. P. Science 2001, 291,2115-2117. Peng, X.; Manna, L.; Yang, W.; Wickham, J.; Scher, E.; Kadavanich, A.; Alivisatos, A. P. Nature 2000, 404, 59-61. Duan, X.; Lieber, C. M. Adv. Mater. 2000, 12, 298-302. Yoffe, A. D. Adv. Phys. 1993, 42, 173-262. Alivisatos, A. P. Science 1996, 271, 933-937. Alivisatos, A. P. /. Phys. Chem. 1996, 100, 13226-13239. Law, M.; Goldberger, J.; Yang, P. Annu. Rev. Mater. Res. 2004, 34, 83-122.

(6)

16

Sun, Y.; Rogers, J. A. Adv. Mater. 2007, 19, 1897-1916. Li, Y.; Qian, F.; Xiang, J.; Lieber, C. M. Materials Today 2006, 9, 18-27. Pauzauskie, P. J.; Yang, P. Materials Today 2006, 9, 36-45. Thelander, C ; Agarwal, P.; Brongersma, S.; Eymery, J.; Feiner, L. F.; Forchel, A.; Scheffler, M.; Riess, W.; Ohlsson, B. J.; Gosele, U.; Samuelson, L. Materials Today 2006, 9, 28-35. Talapin, D. V.; Murray, C. B. Science 2005, 310, 86-89. Bhattacharya, P.; Ghosh, S.; Stiff-Roberts, A. D. Annu. Rev. Mater. Res. 2004, 34, 1-40. Duan, X.; Huang, Y.; Cui, Y.; Wang, J.; Lieber, C. M. Nature 2001, 409, 66-69. Huang, Y.; Duan, X.; Cui, Y.; Lauhon, L. J.; Kim, K.-H.; Lieber, C. M. Science 2001, 294, 1313-1317. Duan, X.; Huang, Y.; Agarwal, R.; Lieber, C. M. Nature 2003, 421, 241-245. Huang, M. H.; Mao, S.; Feick, H.; Yan, H.; Wu, Y.; Kind, H.; Weber, E.; Russo, R.; Yang, P. Science 2001, 292,1897-1899. Hayden, O.; Agarwal, R.; Lieber, C. M. Nature Mater. 2006, 5, 352-356. Cui, Y.; Wei, Q.; Park, H.; Lieber, C. M. Science 2001, 293,1289-1292. Huynh, W. U.; Dittmer, J. J.; Alivisatos, A. P. Science 2002, 295, 2425-2427. Gur, I.; Fromer, N. A.; Geier, M. L.; Alivisatos, A. P. Science 2005, 310, 462-465. Law, M.; Sirbuly, D. J.; Johnson, J. C ; Goldberger, J.; Saykally, R. J.; Yang, P. Science 2004, 305, 1269-1273. Wu, Y.; Yang, P. J. Am. Chem. Soc. 2001, 123, 3165-3166. Wagner, R. S.; Ellis, W. C. Appl. Phys. Lett. 1964, 4, 89-90. Morales, A. M.; Lieber, C. M. Science 1998, 279, 208-211. Barrelet, C. J.; Wu, Y.; Bell, D. C ; Lieber, C. M. /. Am. Chem. Soc. 2003, 725, 11498-11499.

17

Trentler, T. J.; Hickman, K. M.; Goel, S. C ; Viano, A. M.; Gibbons, P. C ; Buhro, W. E. Science 1995, 270, 1791-1794. Yu, H.; Gibbons, P. C ; Kelton, K. F.; Buhro, W. E. /. Am. Chem. Soc. 2001, 123, 9198-9199. Olson, E. A.; Efremov, M. Y.; Zhang, M.; Zhang, Z.; Allen, L. H. /. Appl. Phys. 2005, 97, 034304. Yu, H.; Li, J.; Loomis, R. A.; Gibbons, P. C ; Wang, L. W.; Buhro, W. E. J. Am. Chem. Soc. 2003,125,16168-16169. Grebinski, J. W.; Hull, K. L.; Zhang, J.; Kosel, T. H.; Kuno, M. Chem. Mater. 2004, 16, 5260-5272. Yu, H., PhD Dissertation, Washington University, 2003. Schaller, R. D.; Sykora, M.; Pietryga, J. M.; Klimov, V. I. Nano Lett. 2006, 6, 424-429. Ellingson, R. J.; Beard, M. C ; Johnson, J. C ; Yu, P.; Micic, O. I.; Nozik, A. J.; Shabaev, A.; Efros, A. L. Nano Lett. 2005, 5, 865-871. Dong, A.; Wang, F.; Daulton, T. L.; Buhro, W. E. Nano Lett. 2007, 7,1308-1313. Kuno, M.; Ahmad, O.; Protasenko, V.; Bacinello, D.; Kosel, T. H. Chem. Mater. 2006, 18, 5722-5732. Ouyang, L.; Maher, K. N.; Yu, C. L.; McCarty, J.; Park, H. J. Am. Chem. Soc. 2007, 129, 133-138. Fanfair, D. D.; Korgel, B. A. Chem. Mater. 2007, 19, 4943-4948. Trentler, T. J.; Goel, S. C ; Hickman, K. M.; Viano, A. M.; Chiang, M. Y.; Beatty, A. M.; Gibbons, P. C ; Buhro, W. E. /. Am. Chem. Soc. 1997, 119, 2172-2181. Dingman, S. D.; Rath, N. P.; Markowitz, P. D.; Gibbons, P. C.; Buhro, W. E. Angew. Chem. 2000, 112, 1530-1532. Yu, H.; Buhro, W. E. Adv. Mater. 2003, 15, 416-419. Yu, H.; Li, J.; Loomis, R. A.; Wang, L.-W.; Buhro, W. E. Nature Mater. 2003, 2, 517-520. Fanfair, D. D.; Korgel, B. A. Cryst. Growth Des. 2005, 5, 1971-1976.

18

Hull, K. L.; Grebinski, J. W.; Kosel, T. H.; Kuno, M. Chem. Mater. 2005, 17, 4416-4425. Yu, H.; Li, J.; Loomis, R. A.; Wang, L.-W.; Buhro, W. E. Nat. Mater. 2003, 2, 517-520. Yu, H.; Li, J.; Loomis, R. A.; Gibbons, P. C ; Wang, L.-W.; Buhro, W. E. J. Am. Chem. Soc. 2003,125,16168-16169. Wang, F.; Yu, H.; Li, J.; Hang, Q.; Zemlyanov, D.; Gibbons, P. C ; Wang, L.-W.; Janes, D. B.; Buhro, W. E. J. Am. Chem. Soc. 2007, 129, 14327-14335. Li, J.; Wang, L.-W. Chem. Mater. 2004,16, 4012-4015. Li, J.; Wang, L.-W. Phys. Rev. B 2005, 72, 125325. Dong, A., PhD Dissertation, Washington University, 2007. Norris, D. J.; Bawendi, M. G. Phys. Rev. B 1996,53,16338-16346. Masumoto, Y.; Sonobe, K. Phys. Rev. B 1997,56, 9734-9737. Banin, U.; Lee, C. J.; Guzelian, A. A.; Kadavanich, A. V.; Alivisatos, A. P.; Jaskolski, W.; Bryant, G. W.; Al, L. E.; Rosen, M. J. Chem. Phys. 1998, 109, 2306-2309. Bertram, D.; Micic, O. I.; Nozik, A. J. Phys. Rev. B 1998,57, R4265-R4268. Nikesh, V. V.; Lad, A. D.; Kimura, S.; Nozaki, S.; Mahamuni, S. J. Appl. Phys. 2006,700,113520. Peterson, J. J.; Huang, L.; Delerue, C ; Allan, G.; Krauss, T. D. Nano Lett. 2007, 7, 3827-3831.

19

Chapter 1 SLS Growth of High-Quality CdTe Quantum Wires

20

Introduction CdTe is an important II-VI semiconductor with the bulk band gap of 1.5 eV at room temperature (Table 1-1). Compared with the well-studied CdSe nanostructures, CdTe nanostructures are potentially better candidates in certain applications, such as solar cells1'2 and biological imaging,3'4 due to their additional near-infrared absorption and photoluminescence.

Table 1-1. Bulk band gaps of wurtzite cadmium chalcogenide semiconductors. Semiconductor CdS CdSe Room temperature band gap (eV)a 2.42 Room temperature band gap (nm) 512.3 712.6 826.6 Band-gap type direct direct direct

1.74 1.50 CdTe a Data are taken from reference 5.

Since

the

pioneering

organometallic

hot-injection

synthesis

of

cadmium

chalcogenides nanocrystals reported by Murray and coworker in 1993,6 advances have been made to grow high-quality CdTe nanocrystals of various morphologies including quantum dots, rods, and tetrapods,7"9 with the quality comparable to, or even better than, the quality of CdSe nanocrystals. However, successful growth of high-quality CdTe quantum wires (or nanowires, these two terms are used interchangeably herein) with purposefully controlled diameters was rare. Although hard-template approach10"13 and solvothermal method14 have been applied to produce CdTe nanowires, the wire quality was poor. Kotov and coworkers reported an oriented-attachment approach to grow onedimensional (ID) CdTe quantum wires through self-assembly of zero-dimensional (0D) CdTe quantum dots upon partial removal of the surface ligand.15 The diameter of the

21

resulting quantum wires was measured to be the same as that of the initial quantum dots, associated with a small red shift of the photoluminescence peak of 10-20 nm. The driving force for this remarkable self-assembly process was believed to be strong dipoledipole interactions. The band gaps in quantum dots and true quantum wires of the same diameter should be quite different due to different confinement dimensionality of excitons.16"19 A simple effective-mass-approximation, particle-in-a-box (EMA-PIB) model predicts that the band-gap increase relative to bulk value (AEg) should scale linearly with the inverse square of the diameter {IIcf) of the nanostructure, and a slope ratio A^JA^ of 0.585 for

such a wire line and dot line in plots of AES vs.l/d2 should be observed. The previous quantum-confinement-effect studies on InP and CdSe quantum systems confirmed the EMA-PIB predictions.17'20'21 We similarly plotted the AE$ vs. \lS graph based on Kotov's CdTe quantum-dot and self-assembled quantum-wire data extracted from their photoluminescence spectra (Figure 1-1). Surprisingly, the slope ratio AwkeMdot was measured to be 0.94, suggesting similar magnitude of quantum-confinement energies in CdTe quantum dots and selfassembled CdTe quantum wires having similar diameters. Note that the available

diameter range investigated by Kotov and coworker is only 2.5-5.5 nm. For the CdTe nanostructures having such small diameters, it's hard to believe that the self-assembled quantum wires could be true wires. In a following paper, a 35-nm red shift of the photoluminescence peaks was observed for ID aggregates of CdTe quantum dots relative to the initial dots,22 which provided additional evidence that the small red shift of the

22

photoluminescence peaks of self-assembled CdTe quantum wires may not be due to the formation of true wires.

1.0

0.9 0.8

o a

CdTe QDs (Kotov) Self-assembled CdTe QWs (Kotov)

0.4

T-

0.02

0.04

0.06

0.08

0.10

0.12
2

0.14

0.16

0.18

Met (nm ) Figure 1-1. Plot of AEg vs. l/d2 for CdTe quantum dots and self-assembled CdTe quantum wires reported by Kotov and coauthors.15 The data points were extracted from their photoluminescence spectra. The dashed and solid lines are linear least-squares fits to the dot and wire data, respectively, yielding the slopes Awire = 2.64 eV nm2 and Adot = 2.82 eV nm2, and the slope ratio A^A^t = 0.94. The CdTe project was initiated due to our suspicion over Kotov's explanation of their photoluminescence data of self-assembled CdTe quantum wires. We then followed their synthetic procedure15 to produce CdTe nanowires. Representative low- and high-

magnification TEM images of self-assembled CdTe nanostructures after the wire growth for 7 days (Figure 1-2) show that the CdTe wires are typically 100-400 nm long. Additionally, pearl-necklace shaped nanostructures and nanoparticles could also be observed, indicating that the self-assembly process was incomplete.

23

Figure 1-2. Representative low- (A) and high-magnification (B) TEM images of selfassembled CdTe nanostructures synthesized by self-organization of CdTe quantum dots. The UV-Vis absorption and the corresponding photoluminescence spectra of the initial CdTe quantum dots and the self-assembled CdTe nanostructures were shown in Figure 1-3A. For the photoluminescence spectra, a red shift of 13.2 nm was measured for the self-assembled nanostructures relatively to the initial dots, which is consistent with the values reported in the literature.1^ However, we observed two features in the

absorption spectrum of the self-assembled CdTe nanostructures. The first one is at -525 nm, which also exists in the absorption spectrum of the initial dots. The second one centered at -650 nm is very broad, but absent in the dot absorption spectrum. By comparing the photoluminescence and the absorption spectra of the self-assembled CdTe nanostructures, we noticed that the photoluminescence peak position was at the shortwavelength side of the absorption feature at -650 nm.

24

<

350

400

450

500

550

600

650

700

750

800

Wavelength (nm)

Figure 1-3. Representative optical properties of the self assembled CdTe nanostructures. (A) UV-Vis absorption and normalized photoluminescence spectra of the initial CdTe quantum dots (dotted lines) and the self-assembled CdTe nanostructures (solid lines) after 7 days. (B) TEM image of the self-assembled CdTe sample. The scale bar is 100 nm. (C) Diameter-distribution histogram of the self-assembled CdTe nanowires. The mean diameter of the initial CdTe quantum dots should be -3.4 nm based on the photoluminescence peak position.15 However, the mean diameter of the self-assembled wires was measured to be 7.6 nm (Figure 1-3C), which is about two times of the initial dot diameter. This diameter increase-phenomenon in the self-organization process was also observed in other studies employing a similar synthetic strategy to produce CdTe nanowires.23"26 Based on the above observations, it's reasonable to assign the absorption feature at -650 nm of the self-assembled CdTe nanostructures to nanowires, and another absorption feature at -525 nm to the remaining nanoparticles and the pearl-necklace aggregates.22 In 25

this regard, the observed photoluminescence of the self-assembled CdTe nanostructures should be mainly from the pearl-necklace aggregates instead of the self-assembled nanowires. There are disadvantages for the synthesis of CdTe nanowires via the oriented attachment approach. First, it is performed in aqueous solution. Generally, the quality of the CdTe nanostructures grown in aqueous solution27 is less than that grown in organic solvents, such as tri-n-octylphosphine oxide (TOPO)7 and 1-octadecene (ODE),8 at high reaction temperatures. Second, it's hard to avoid the diameter increase, which was also observed in other semiconductors, such as ZnO.28 Therefore, it's desirable to employ other approaches to grow high-quality CdTe nanowires. The SLS growth of nanowires offers several advantages over other methods, including purposeful control of nanowire mean diameters, narrow diameter distributions, small diameters in the quantum-confinement regime, control of surface passivation, and nanowire solubility.29 Dr. Heng Yu in our group did preliminary work on the synthesis of CdTe nanowires using the SLS mechanism.30 Unfortunately, it was found very difficult to produce wires using common precursors known to grow CdTe quantum dots, which was likely due to the inappropriately balanced reactivities of the precursors. In this chapter, we report the SLS growth of high-quality, diameter-controlled, singlecrystalline CdTe quantum wires from Bi-catalyst nanoparticles. The quantum wires exhibit size-dependent quantum-confinement effects, as revealed in their UV-Vis absorption spectra, which are interestingly rich in detail, consistent with the high quality of the wires. The detailed study of the electronic structure of colloidal CdTe quantum wires will be the topic of Chapter 2.

26

Results Synthesis. A key precursor development used for the solution-based synthesis of high-quality CdTe nanocrystals6'7 enabled us to design a synthetic route for the SLS growth of CdTe quantum wires, which is shown in Figure 1-4. Cadmium oxide (CdO) was used as the initial cadmium source, which is much more environmentally friendly than the extremely toxic dimethylcadmium precursor.7 The actual cadmium precursor is generated in-situ through the reaction between CdO and long-chain organic acids, such as carboxylic acids or alkylphosphonic acids, forming a soluble Cd complex at elevated temperatures (>300 C) in TOPO. The tellurium precursor is a tri-ra-alkylphosphine telluride (TOPTe and TBPTe stand for tri-n-octylphosphine telluride and tri-nbutylphosphine telluride, respectively), which can be easily prepared by the reaction between tri-n-alkylphosphine (TOP and TBP stand for tri-ra-octylphosphine and tri-nbutylphosphine, respectively) and elemental tellurium and stored for routine use. Bi nanoparticles were employed as the catalysts for the growth of CdTe quantum wires because of the low eutectic point of 146 C shown in the bulk Bi-Cd binary phase diagram (Figure 1-5). Importantly, it has been demonstrated by Yu that Bi could serve as a good catalyst to grow CdTe nanowires.30 Therefore, Bi nanoparticles were our choice in designing the synthetic route.

A Solvent * Acid " Carboxylic acid or Alkylphosphonic acid

CdO + Acid*

Cadmium carboxylate or alkylphosphonate

Bi seeds

+ R3P=Te
Precursors (R - M-CH, or n-CH17)

Solvent 240-300 C CdTe nanowire

Figure 1-4. Schematic illustration of the designed synthetic route for the solution-liquidsolid (SLS) growth of CdTe quantum wires.

27

Weight P e r c e n t C a d m i u m 20 so <p op

70

80

90

100

|IIIIIM|HI

|*-<M)

(Cd)-

10

SO

3Jo" " '" l i*'o" l "' L tfd "" llouc 70 Atomic Percent Cadmium

80

90

ido

Cd

Figure 1-5. Bulk Bi-Cd binary phase diagram. The eutectic point is 146 C at ~55 atomic percent cadmium. Most of the optimization experiments were conducted using the cadmium oleate precursor Cd(OA>2. Our initial attempts to synthesize CdTe quantum wires were conducted in the solvent 1-octadecene (ODE), which is a noncoordinating solvent used for the synthesis of high-quality CdTe nanocrystals of various shapes, including dots, rods, and tetrapods.8 Unfortunately, only poor wires in very low yield could be grown in this solvent under various reaction conditions investigated (Figure 1-6). Note that the precursor concentration employed (0.02 mmol of Te in 4 g of solvent) was much lower than that used for the synthesis of CdTe quantum dots (0.05 mmol of Te in 4 g of solvent).7,8 Nevertheless, dots were still observed to be predominant in the TEM images. Given the presence of Bi catalyst nanoparticles in the synthesis, the results suggest very high precursor reactivities in ODE, which favor the nucleation and growth of CdTe quantum dots instead of the SLS growth of wires from Bi nanoparticles.

28

Figure 1-6. Representative TEM images of CdTe nanostructures synthesized in 1octadecene (ODE). Cd(OA)2 and TOPTe (0.02 mmol) were used as precursors. The Cd/Te ratio was 9. Bi nanoparticles with the mean diameter of 9.4 nm were used as catalysts. The reaction was conducted at 260 C. We thus employed TOPO, a commonly used coordinating solvent, for the SLS growth of CdTe quantum wires. The Cd/Te ratio played a critical role for successful wire growth. Figure 1-7 shows the TEM images of CdTe nanostructures synthesized using various Cd/Te ratios in the range of 0.05 to 10. When an excess amount of Te precursor was used, dots and a minority of short rods were produced (Figure 1-7A, B). We did not observe any catalyst nanoparticles attached to either end of the rods, indicating that the rods were likely not grown by the SLS mechanism. These results are different from the SLS growth of CdSe nanowires developed in our group, which showed that high-quality CdSe wires could be produced when an excess amount of Se precursor was used (the Cd/Se ratio could be as low as 1:70).17 Fortunately and excitingly, when an excess amount of Cd precursor was used, CdTe nanowires were successfully produced (Figure 1-7C, D), although nanoparticles could also be observed in the TEM images. We did observe the catalyst nanoparticles attached to one end of individual CdTe nanowires, confirming the SLS growth of the wires. The nanoparticle formation suggested that the precursor reactivities were still too high.

29

Figure 1-7. TEM images of CdTe nanostructures synthesized using various CdVTe ratios in TOPO. (A) Cd/Te = 0.05, (B) Cd/Te = 0.10, (C) Cd/Te = 5, and (D) Cd/Te = 10. For (A) and (B), cadmium stearate and TOPTe were used as precursors. For (C) and (D), Cd(OA>2 and TOPTe were used as precursors. The amount of the limiting precursor was held at 0.03 mmol. Bi nanoparticles with the mean diameter of 10 nm were used as catalysts. The reaction was conducted at 260 C. The dotted red circles shown in (C) highlight the catalyst nanoparticles attached to one end of individual wires, confirming the SLS wire growth. To further optimize the reaction conditions, the TOPTe concentration and Cd/Te ratio were varied. When 0.05 mmol of TOPTe was employed, we produced both wires and star-like large nanoparticles, which were observed to be aggregates of small elongated nanoparticles (Figure 1-8). Experimentally, we observed precipitate formation during the reaction. These results suggested higher precursor reactivities compared with the However, the nanoparticle

synthesis using 0.03 mmol of TOPTe (Figure 1-7C).

formation was significantly reduced when 0.02 mmol of TOPTe was employed (Figure 1-

30

9). From the TEM images, we can clearly see long straight wires, which tend to form bundles. In the range of Cd/Te ratios investigated, CdTe nanowires of similar quality were produced, indicating that the synthesis could be conducted in a relatively broad range of Cd/Te ratios. Unfortunately, a minority population of nanoparticles could also be observed in the TEM images. We believe that the nanoparticle formation should be further suppressed by using a smaller amount of TOPTe (< 0.02 mmol). Alternatively, we employed a less reactive cadmium precursor to solve this problem, as will be described later.

Figure 1-8. TEM images of CdTe nanostructures synthesized using high concentration of TOPTe (0.05 mmol) and various Cd/Te ratios in TOPO. (A) Cd/Te = 3, (B) Cd/Te = 5, (C) Cd/Te = 7, and (D) Cd/Te = 9. Cd(OA)2 was the Cd precursor. Bi nanoparticles with the mean diameter of 9.4 nm were used as catalysts. The reaction was conducted at 260 C.

31

Figure 1-9. TEM images of CdTe nanowires synthesized using low concentration of TOPTe (0.02 mmol) and various Cd/Te ratios in TOPO. (A) Cd/Te = 3, (B) Cd/Te = 5, (C) Cd/Te = 7, and (D) Cd/Te = 9. Cd(OA)2 was the Cd precursor. Bi nanoparticles with the mean diameter of 9.4 nm were used as catalysts. The reaction was conducted at 260 C. Because commercially available cadmium stearate has been employed to grow highquality CdSe nanowires,17 we also made efforts to grow CdTe nanowires using this cadmium precursor. From the TEM images (Figure 1-10) we found that nanowires were predominantly produced only when Cd/Te = 3 was used. However, many wires were found to be not straight, thus they could not form bundles effectively. When the Cd/Te ratio was increased to 5 or above, both wires and dots were produced. Compared with the results shown in Figure 1-9, it was clearly shown that cadmium stearate was not a good cadmium precursor for the SLS growth of CdTe nanowires under the synthetic conditions investigated.

32

Figure 1-10. TEM images of CdTe nanostructures synthesized using cadmium stearate and various Cd/Te ratios in TOPO. (A) Cd/Te = 3, the inset low-magnification image shows that many wires are not straight. (B) Cd/Te = 5, (C) Cd/Te = 7, and (D) Cd/Te = 9. TOPTe (0.02 mmol) was the Te precursor. Bi nanoparticles with the mean diameter of 9.4 nm were used as catalysts. The reaction was conducted at 260 C. Under optimized reaction conditions, CdTe nanowires with tunable mean diameters in the range of 5-8 nm could be grown using Cd(OA)2 and TOPTe as precursors (Figure 1-11). The wires are typically micrometers long. The straight wires tend to form bundles. The standard deviation in the diameter distributions was less than 20 % of the mean diameter (Figure 1-12). The wire diameter could be controlled by the diameter of the Bi nanoparticles and the reaction temperature (Table 1-4). For example, 6.7-nm and 7.9-nm CdTe nanowires were produced from 7.8-nm and 12.2-nm Bi nanoparticles, respectively, at 260 C. At lower temperature of 255 C, 5.5-nm wires were grown using 7.8-nm Bi nanoparticles.

33

Low magnification

High magnification

Figure 1-11. Representative low- and high-magnification TEM images of CdTe nanowires synthesized using Cd(OA)2 and TOPTe (0.02 mmol) as precursors in TOPO. (A) Jwire = 5.5 + 0.8 nm ( 15%), (B) dwire = 6.7 1.0 nm (+ 15%), (C) dwire = 7.4 + 1.0 nm ( 14%), and (D) dwire = 7.9 1.0 nm (+ 13%). The Cd/Te ratio was held in the range of 6-9. The reactions were conducted in the temperature range of 255-260 C.

34

10

11 CdTe nanowire diameter (nm)

CdTe nanowire diameter (nm)

Figure 1-12. Representative diameter-distribution histograms for CdTe nanowires synthesized using Cd(OA>2 and TOPTe as precursors in TOPO. (A) dwm = 5.5 0.8 nm ( 15%), (B) cUe = 6.7 1.0 nm ( 15%), (C) dwire = 7.4 1.0 nm ( 14%), and (D) d^e = 7.91.0nm(13%). Efforts have been made to grow small-diameter CdTe nanowires using TBPTe as tellurium precursor and small Bi nanoparticles (Figure 1-13). Generally, small Bi nanoparticles exhibit strong agglomeration tendencies. Therefore, we hoped to minimize the nanoparticle agglomeration by using TBPTe with a short carbon chain. Experimentally, we found that TBPTe is more reactive than TOPTe, as indicated by a faster color change of the reaction mixture after injection of Bi nanoparticles and Te precursor. The use of TBPTe allowed us to grow 5.0-nm CdTe nanowires from 6.1-nm Bi nanoparticles. Unfortunately, we were not able to produce wires with smaller diameters. When 4.1-nm Bi nanoparticles were used, the mean diameter of the resulting nanowires was measured to be 5.1 nm, which is larger than the initial diameter of the Bi

35

nanoparticles, suggesting agglomeration of the small catalyst nanoparticles. Additionally, for the CdTe nanowire specimen prepared using TBPTe, we observed many wires shorter than 500 nm (Figure 1-14). The reason for the short wires remains unclear.

Figure 1-13. TEM images of CdTe nanowires synthesized using Cd(OA)2 and TBPTe (0.02mmol) as precursors in TOPO. (A) dBi = 6.1 0.7 nm (+ 11 %), dc<ne = 5.0 0.7 nm (+ 14 %), and (B) dBi = 4.1 0.8 nm (+ 20 %), dCsre = 5.1 + 0.6 nm (+ 12 %). The Cd/Te ratio was held at 6. The reactions were conducted at 260 C. To grow large-diameter CdTe nanowires, large Bi nanoparticles were employed at reaction temperatures higher than 300 C. Under these harsh synthetic conditions, we often observed nanostructures with up to three arms grown from one catalyst nanoparticle (Figure 1-15). Considering the large size of Bi nanoparticles and the high reaction temperatures, a high possibility for a secondary nucleation site on the catalyst alloy surface should account for the formation of nanostructures with multiple arms.

36

Figure 1-14. Representative low-magnification TEM images of CdTe nanowires synthesized using Cd(OA>2 and TBPTe as precursors in TOPO.

Figure 1-15. TEM images of 11.1-nm CdTe nanowires synthesized using 15.9-nm Bi nanoparticles in TOPO at 320 C. Cd(OA)2 and TOPTe were used as precursors. The Cd/Te ratio was 9. The dotted circles highlight the catalyst center. As noted above, although CdTe nanowires of various mean diameters were successfully produced using Cd(OA)2 as the cadmium precursor, the formation of

37

quantum dots in small amount could not be avoided. The dots were presumably formed by homogenous nucleation in a process competing with SLS growth. Peng and

coworkers have previously demonstrated that cadmium alkylphosphonate precursors are much less reactive than cadmium carboxylates.8 In this regard, we expected that the problem of dot formation could be solved by employing less reactive cadmium alkylphosphonates. The reaction conditions optimized for the Cd(OA)2 precursor were employed with the cadmium n-octadecylphosphonate Cd(ODPA) precursor. Indeed, we found that the use of Cd(ODPA) dramatically suppressed homogeneous nucleation and dot formation under optimized experimental conditions. Representative low- and high-magnification TEM images of CdTe quantum wires (Figure 1-16 and 1-17) synthesized from Cd(ODPA) and TOPTe show that dot formation was nearly eliminated, as indicated by the clean image backgrounds. The diameters of the wires were purposely varied within the range of 5-11 nm. The standard deviation in the diameter distributions was less than 20% of the mean diameter (Figure 1-18). TEM images obtained at the lowest magnification (Figure 1-16A) established that the wires were typically several micrometers long, and tended to form large bundles having widths up to the micrometer scale. The separations between the individual wires in the bundles were very uniform, and presumably corresponded to the volume occupied by the surfactant coatings on the wire surfaces. The high-magnification TEM images (Figure 1-17) showed that the nanowires possessed near-constant diameters along their lengths. The high-resolution TEM (HRTEM) image (Figure 1-19) from a wire bundle indicated that the wires were single crystalline over large domains, and crystallographically oriented. The CdTe quantum wires prepared from Cd(ODPA)

38

represent the best-quality semiconductor nanowires that can be achieved by the SLS mechanism to date.

Figure 1-16. Representative low-magnification TEM images of high-quality CdTe quantum wires synthesized from Cd(ODPA) and TOPTe (0.02 mmol) in TOPO. (A) 7.3 + 1.0 nm (+ 13.7%). This image at very low magnification shows that the lengths of the wires are typically several micrometers and the long straight wires form bundles with width up to micrometer scale. (B) 5.3 + 1.1 nm (+ 20.8%), (C) 5.8 + 1.0 nm ( 17.2%), (D) 7.4 0.9 nm ( 12.2%), (E) 7.9 1.3 nm (+ 16.5%), (F) 8.7 1.5 nm ( 17.2%), (G) 9.3 1.6 nm ( 17.2%), (H) 9.7 2.0 nm (+ 20.6%), and (I) 10.2 + 1.7 nm ( 16.7%). Individual nanowires in bundles can be clearly seen in images (B) through (I). The Cd/Te ratio was held at 5. The reaction was conducted in the temperature range of 250285 C.

39

Figure 1-17. Representative high-magnification TEM images of high-quality CdTe quantum wires synthesized from Cd(ODPA) and TOPTe (0.02 mmol) in TOPO. (A) 5.3 1.1 nm ( 20.8%), (B) 5.8 1.0 nm ( 17.2%), (C) 7.3 1.0 nm ( 13.7%), (D) 7.4 0.9 nm ( 12.2%), (E) 7.9 1.3 nm ( 16.5%), (F) 8.7 1.5 nm ( 17.2%), (G) 9.3 1.6 nm ( 17.2%), (H) 9.7 2.0 nm ( 20.6%), and (I) 10.2 1.7 nm ( 16.7%). The Cd/Te ratio was kept at 5. The reaction was conducted in the temperature range of 250-285 C.

40

-1rT*

1 '

120 10080

(b)

I
3 60 40

10

CdTe nanowire diameter (nm)

J
4

|MPT^1 9 10 11 5 6 7 CdTe nanowire diameter (nm)

12

r* 100-

T*pr-

(c)

_,,,r,,,

80-

Counts

ro-

JL
.^MlH IlIlM-^
CdTe nanowire diameter (nm)
10 11 12 13

S
ll-

CdTe nanowire diameter (nm)

140120100-

(e)

Counts

40200-

11
^MMIM
4 5 6 7 8 9 10

wwwpw^^iT1|,
11 12 13 14

15

9 10 11 12 13 14 15 16 17

CdTe nanowire diameter (nm)

CdTe nanowire diameter (nm)

Figure 1-18. Representative diameter-distribution histograms for CdTe quantum wires synthesized from Cd(ODPA) and TOPTe in TOPO. (a) 5.3 1.1 nm (+ 20.8%), (b) 5.8 1.0 nm ( 17.2%), (c) 7.3 1.0 nm ( 13.7%), (d) 7.9 1.3 nm ( 16.5%), (e) 9.3 1.6 nm ( 17.2 %), and (f) 10.2 1.7 nm ( 16.7%).

41

Figure 1-19. Representative high-resolution TEM image from a CdTe nanowire bundle. Diameter control was achieved by varying the diameters of the Bi catalyst nanoparticles. Although the size of the catalyst nanoparticles was the most important factor for controlling wire diameters, the reaction temperature was also adjusted in conjunction with variations in catalyst nanoparticle size (Table 1-4). These temperature variations optimized the wire-diameter control in response to the different melting points and agglomeration tendencies of the variously sized Bi nanoparticles.31 Thus, a relatively low reaction temperature was used with small Bi nanoparticles, because they have low melting points and a high tendency to agglomerate. However, relatively high reaction temperatures were optimal for larger Bi nanoparticles because of their comparatively higher melting points (and lower agglomeration tendencies). The diameters of the CdTe quantum wires (dcaje) were found to scale linearly with the initial diameters of the Bi nanoparticles (J B 0, revealing the relationship in equation 11 (see Figure 1-20). The slope value of 0.54 is comparable to the slope values measured in other II-VI and III-V semiconductor-nanowire syntheses by SLS growth.29 dcdTe = (0.54 + 0.06)dBi + (1.92 + 0.68) (1-1)

42

13 12 11 H
d

10 9H

CdT. = < 0 - 5 4 *

0 0 6

d B 1

9 2

0 6 8

8 7 6
5 4 3
||.||||iJ.|i!.r

10 11 12 13 14 15 16

Initial Bi nanoparticle diameter (nm)

Figure 1-20. Plot of CdTe quantum-wire diameter vs. initial Bi nanoparticle diameter. The error bars represent the standard deviations of the measured diameters. The red line is a linear fit to the data. Thefitafforded a slope of (0.54 0.06) and a y intercept of (1.92 0.68).

10

15

20

25

CdTe nanowire diameter (nm)

Figure 1-21. TEM images and diameter-distribution histogram for large-diameter CdTe nanowires synthesized from ~40-nm Bi nanoparticles. (A)-(C) TEM images with different magnifications, and (D) diameter-distribution histogram, giving the statistical data of 17.3 7.5 nm ( 43%). For this sample, Cd(ODPA) and TOPTe were used as precursors at the Cd/Te ratio of 5. The reaction was conducted at 310 C in TOPO.

43

Equation 1-1 applies only to Bi nanoparticles having diameters within the range of -5-17 nm. The growth of high-quality CdTe quantum wires having mean diameters of less than 5 nm has not yet been achieved because of the rapid, uncontrolled agglomeration of very small Bi nanoparticles. Very large Bi nanoparticles were also problematic. For example, CdTe wires having a 17.3-nm mean diameter were produced using ~40-nm Bi nanoparticles, but the standard deviation in the diameter distributions was measured to be 43.4% of the mean diameter, indicating that the diameter control was very poor (Figure 1-21).

CdTe Nanowires

3 id
(0

Bulk CdTe (Wurtzite) Bulk CdTe (Zinc Blende) Bulk BI (Rhombohedral)

0>

i i 'i

|' i

20

25

30 26(degree)

Figure 1-22. Powder X-ray diffraction (XRD) pattern of CdTe quantum wires. The standard patterns of the bulk wurtzite (ICDD-PDF file 04-003-4983) and zinc blende (ICDD-PDF file 01-070-8041) CdTe structures and the bulk rhombohedral Bi (ICDDPDFfile00-044-1246) structure are also shown in the graph. The observed reflections of CdTe nanowires are compared to those of wurtzite CdTe. For large-diameter CdTe nanowires, large non-spherical catalyst nanoparticles could be observed at the wire ends (Figure 1-21C), which may suggest a recrystallization process during cooling of the reaction mixture, as will be described later. However, we

44

usually find nearly spherical catalyst nanoparticles attached to the ends of small-diameter wires. Because the development of facets is energetically favorable for large crystals, it's reasonable to observe non-spherical catalyst nanoparticles attached to the ends of large wires. Structure. The crystal structure of the CdTe quantum wires was investigated by powder X-ray diffraction (XRD, Figure 1-22) and HRTEM (Figure 1-23). By comparing the experimental XRD pattern of the wires with the standard patterns of bulk CdTe having wurtzite and zinc blende structures, the wurtzite structure of the CdTe quantum wires was confirmed. The missing of the 002 reflection suggests that the wurtzite nanowires preferentially lie with their c axes parallel to the substrate. A very weak reflection at -27 was indexed to the 012 reflection of rhombohedral Bi.

Figure 1-23. Lattice-resolved high-resolution TEM (HRTEM) image of a 10-nm CdTe

quantum wire. The inset shows the indexed fast-Fourier-transform (FFT) pattern

obtained from the well-resolved CdTe wire lattice, indicating the [002] growth direction of thenanowire.

45

The fast-Fourier-transform (FFT) pattern obtained from the well-resolved HRTEM lattice image was also indexed to the wurtzite structure (inset, Figure 1-23), indicating the [002] growth direction for the wires. The well-resolved lattice fringes perpendicular to the growth direction in the HRTEM image corresponded to a spacing of 0.38 nm, consistent with the d-spacing between (002) planes in hexagonal CdTe.32

Figure 1-24. Lattice-resolved high-resolution TEM image of a single CdTe quantum wire having a catalyst nanoparticle attached to the wire tip. Figure 1-24 shows a lattice-resolved HRTEM image of a single CdTe quantum wire having a catalyst nanoparticle attached to the wire tip. The catalyst nanoparticle appears darker in the image because of Z contrast: ZBi > ZTe = Zed- Energy-dispersive X-ray spectra (EDS) collected from other wires with attached catalyst nanoparticles (Figure 125) contained only Cd and Te signals from the wires, and only Bi signals from the catalyst nanoparticles. Well-resolved lattice fringes in the Bi nanoparticle tip are evident in the Figure-1-24 HRTEM image, indicating a recrystallization of Bi upon cooling after the synthesis procedure. Interestingly, the Bi fringe pattern is aligned parallel to the

46

corresponding lattice fringes in the attached CdTe wire, suggesting an epitaxial relationship. The measured lattice spacing in the Bi nanoparticle is 0.398 nm, which matches the 003 d-spacing of rhombohedral Bi.33 Thus, the results indicate an epitaxial junction between Bi (003) and CdTe (002) crystal faces, which is reasonable due to the very small lattice mismatch of 0.5% (Table 1-2 and Figure 1-26).

0.000

2.000

4.000

6.000

8.000 10.000 12.000 14.000

Energy (keV)

Cu

(b)

c
3

0.000

2.000

4.000

6.000

8.000 10.000 12.000 14.000

Energy (keV) Figure 1-25. Energy-dispersive X-ray spectra (EDS) collected from the catalyst nanoparticle (a) and the wire (b).

47

Table 1-2. Cell parameters of rhombohedral Bi and wurtzite CdTe. Materials Bi CdTe Crystal structure Rhombohedral Wurtzite Cell parameters a = 6 = 4.547 A, c= 11.862 A a= ,0=90,7=120 a = b = 4.572 k,c = 7.485 A ^=1=90,^=120

Figure 1-26. Arrangement of atoms projected to the (001) planes in rhombohedral Bi (left) and wurtzite CdTe (right).

(0 J3

o c

o
(0

<

Wavelength (nm)

Figure 1-27. Absorption spectrum of 7.3-nm CdTe quantum wires diluted in toluene, showing five well-resolved absorption features indicated by black arrows.

48

UV-Vis absorption spectra and size-dependent quantum-confinement effects in CdTe quantum wires. The spectroscopic studies and band-gap determinations were conducted with CdTe quantum wires synthesized from Cd(ODPA). A representative absorption spectrum of 7.3-nm CdTe quantum wires (Figure 1-27) shows five wellresolved absorption features. Other wire specimen show at least four resolved absorption features, consistent with the high-quality of the wires. Note that the originally recorded absorption spectra were in a wavelength scale. To extract the exact positions of the first excitonic peak and the higher transition peaks, we employed a modified fitting method reported previously.17 We first converted each original spectrum in a wavelength scale to the corresponding spectrum in an energy scale, and then fitted the converted spectrum using non-linear curve-fitting software (Origin 7.5). The reason for using the spectrum in an energy scale is that a Gaussian fit better represents absorption peaks in an energy scale than in a wavelength scale. We used one exponential growth function for the spectrum background and multiple Gaussian functions for the absorption features. The general fitting function is shown in equation 1-2.
(x-xc)2

y = y0+\e*"

JT(_A^ ~^)

(1-2)

where y and x represent the absorbance and the energy, respectively; yo is the offset constant; A0 and to represent the amplitude and growth constant of the exponential function, respectively; A,, w and xc, represent the amplitude, the standard deviation, and the peak center of the i* Gaussian function, respectively. The fitting procedure yielded the center energy of the excitonic peaks and the error in the center energy. A representative fitting of the converted absorption spectrum of 7.3-nm CdTe quantum

49

wires (Figure 1-28) shows that the fitted absorption spectrum is very close to the experimental spectrum, indicating the effectiveness of our fitting procedure.

-~i

r~

1.6 1.7 1.8 1.9 2.0 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9

Photon energy (eV)

Figure 1-28. Representative fitting of the absorption spectrum of 7.3-nm CdTe quantum wires. The black solid curve and the red dotted curve are the experimental spectrum and the fitted spectrum, respectively. The fitting model is composed of one exponential (for the background) and five Gaussian functions (for the features). The fitting parameters are shown in the graph. Individual fitting functions including the exponential background (gray dotted curve) and multiple Gaussian peaks (colored solid curves) are also plotted. Table 1-3. Band gap energies and quantum confinement energies (AEg) for CdTe quantum wires of various mean diameters (dcare)done (nm) 5.3 5.4 5.8 7.3 7.4 7.9 8.7 9.3 10.2
oo (bulk)

Band gap (eV) 1.850 1.816 1.786 1.721 1.707 1.684 1.675 1.665 1.650 1.50

Band gap (nm) 670.2 682.7 694.2 720.4 726.3 736.2 740.2 744.6 751.4 826.6

AEZ (eV) 0.350 0.316 0.286 0.221 0.207 0.184 0.175 0.165 0.150

50

The converted absorption spectra and the corresponding lowest-energy absorption features extracted by the fitting procedure described above for CdTe quantum wires having diameters in the range of 5.3-10.2 nm are shown in Figure 1-29. The band-gap energy for wires of each diameter studied was assigned as the center of the resulting Gaussian-fitted peak and summarized in Table 1-3. All the band-gap absorptions of the CdTe quantum wires investigated show blue shifts relative to the bulk band-edge absorption at 1.50 eV. In addition, when the diameter of the wires decreased, we are able to observe systematic increases of the quantum-confinement energy, indicating the conventional size-dependent quantum-confinement effects.

Wavelength (nm)
850 750 650
111111111 i i | i i i

550
i | i i i i

450
11

750 650

o a c
CO Q

o
(A fi

<

~iiiii'ii

"-"i'i'r

1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 1.6 1.8 2.0
Photon energy (eV)

Figure 1-29. Representative absorption spectra of CdTe quantum wires of various diameters in the range of 5.3-10.2 nm (a) and their lowest-energy excitonic peaks extracted by nonlinear least-squares fitting (b).

51

o> u
l_

3 3

(0 .Q

c o
(0

0)

<
400 450 500 550 600 650 700 750 800 850

Wavelength (nm)

Figure 1-30. Room temperature photoluminescence spectra of the 5.4-nm and 7.4-nm CdTe quantum wires synthesized from Cd(ODPA), and diluted with toluene. The corresponding UV-Vis absorption spectra (dotted curves) are also shown in the graph. Room-temperature photoluminescence. We also collected room-temperature

photoluminescence spectra for CdTe quantum wires synthesized from Cd(ODPA). However, we observed only weak room-temperature photoluminescence from the smalldiameter wires; the photoluminescence from the large-diameter wires was weak and noisy (Figure 1-30). Note that the spectra shown in Figure 1-30 were collected from the toluene dilution of the as-synthesized wires. When small-diameter wires were

washed/purified with methanol, the wire photoluminescence disappeared (Figure 1-31). There are two possible reasons for the photoluminescence quenching upon wire purification. First, the purification step partially removes the surface ligands, and thus creates surface trap sites, which may significantly lower the photoluminescence. Second, the bundling of long straight wires is more severe when free ligands around the wires are washed off, which may also play a quenching role for wire photoluminescence.

52

1.0-r

0.8- \ O O
nj

r*
0.6<D 3 W

<
0.2-

as-synthesized

0.0400 450 500 550 600 650 700 750 800 850

Wavelength (nm)

Figure 1-31. Representative room temperature photoluminescence spectra of the asprepared and methanol-washed 5.5-nm CdTe quantum wires.

Discussion Balanced precursor reactivities for high-quality CdTe quantum wires. For the SLS growth of binary compound semiconductor nanowires from dual-source precursors, the reactivities of the precursors should not be too high and have to be carefully balanced in order to produce well-controlled nanowires. In this regard, the successful synthesis of CdSe nanowires in our group17 did not directly lead to success in the SLS growth of CdTe nanowires due to different precursor reactivities between TOPSe and TOPTe. For the tri-n-octylphosphine chalcogenide (TOPE, E = S, Se, Te) precursors, their reactivities are shown in an increasing order as follows: Reactivity: TOPS < TOPSe < TOPTe Note that TOPTe is the precursor with the highest reactivity in the series, which should be the reason that the SLS growth of CdTe nanowires was previously found to be very difficult.30 The precursor TOPS is also problematic due to its low reactivity, as will be described in Chapter 4.

53

Solvent may influence the reactivity of a given precursor. Generally, the precursor reactivity in noncoordinating solvents should be higher than that in coordinating solvents. Indeed, our experiments showed the failure of the SLS growth of CdTe nanowires in noncoordinating solvent ODE, and the success in coordinating solvent TOPO. Therefore, we attribute the success to lowered precursor reactivities in TOPO. One may argue that the failure of the wire growth in ODE could be possibly due to the Bi catalyst nanoparticles because they are likely not protected well. However, the successful SLS growth of ZnSe and ZnTe nanowires in ODE34 can rule out this possibility. For the TOPO-based synthesis, the wire quality was experimentally demonstrated to be influenced by several synthetic parameters. We now relate them to the context of the reactivities of dual-source precursors and their reactivity balance. (1) Cd/Te ratio. As shown in the Results section, CdTe nanowires could not be produced when excess amount of TOPTe was used (Figure 1-6). However, for CdSe wires, excess amount of TOPSe was preferred.17 As noted above, TOPTe is more reactive than TOPSe, therefore, excess amount of TOPTe represents inappropriately balanced precursor reactivities, which favors the homogeneous nucleation and growth of CdTe nanocrystals. (2) Precursor concentration. Generally, the higher the precursor concentration is, the higher the precursor reactivity is. Experimentally, we observed more CdTe nanocrystals when higher precursor concentrations were employed (Figure 1-7), which should be due to higher precursor reactivities. (3) Less-reactive cadmium precursor. As mentioned in the Results section, a minority population of dots were often observed in the CdTe nanowire specimen grown from

54

Cd(OA)2 precursor; and the best CdTe nanowires were produced from less reactive Cd(ODPA). These results indicated that appropriately balanced precursor reactivities were achieved when Cd(ODPA) and TOPTe were employed as precursors. The optimization of experimental conditions for the SLS growth of CdTe nanowires, which was previously found to be very difficult, may serve as helpful guides for the SLS growth of other nanowire compositions. Epitaxial recrystallization of catalyst nanoparticles and the SLS growth mechanism for CdTe quantum wires. We observed by HRTEM (Figure 1-23) and EDS (Figure 1-24) that the catalyst nanoparticle attached to the wire tip was composed of Bi only and there was an epitaxial junction between Bi (003) and CdTe (002) crystal faces. Considering the alloy nature of the catalyst nanodroplet during the SLS wire growth, the HRTEM and EDS results strongly suggested a phase separation process in the Bi-Cd-Te ternary alloy nanodroplet and an epitaxial recrystallization of Bi upon cooling. It's believed that the SLS mechanism is analogous to the well-known VLS mechanism. Because it's much more difficult to directly observe the SLS growth of nanowires in solution compared to the corresponding VLS wire growth in gas phase due to technical limitations, we thus sought to borrow the results from the mechanism study for the VLS growth of nanowires. By employing an in situ high-temperature TEM technique, Wu and Yang were able to clearly identify three stages for the VLS growth of nanowires from metal catalyst nanoparticles: metal alloying, crystal nucleation, and axial growth.35

55

Combined with what we observed in HRTEM and EDS for CdTe quantum wires, we proposed the following four-stage mechanism for a complete SLS growth process of CdTe quantum wires (Figure 1-32).

Cd precursor

Te precursor Bi Bi-Cd-Te alloy (Stage I) Cd precursor Nucleation of CdTe phase (Stage II)

Te precursor

Axial growth of CdTe QW (Stage III)

cooling

Growth of CdTe QW segment upon cooling

Phase separation in Bi-Cd-Te alloy and epitaxial recrystallization of Bi (Stage IV)

Figure 1-32. Schematic illustration of the proposed four-stage SLS growth mechanism for CdTe quantum wires. Stage I: Formation of liquid Bi-Cd-Te ternary alloy nanodroplet; Stage II: Nucleation of CdTe phase; Stage III: Axial growth of CdTe quantum wire; Stage IV: Phase separation in Bi-Cd-Te ternary alloy nanodroplet and epitaxial recrystallization of Bi. The dotted box is a guide to the eye. Stage I: Formation of liquid Bi-Cd-Te ternary alloy nanodroplet. A Bi-Cd-Te ternary alloy nanodroplet is formed with increasing amount of Cd and Te-precursor decomposition and dissolution in Bi catalyst nanoparticle. The concentrations of Cd and Te in the droplet will keep increasing until reaching a supersaturation point. During this process, the diameter of the alloy nanodroplet should be larger than that of the initial Bi nanoparticle.35

56

Stage II: Nucleation of CdTe phase.

When the alloy nanodroplet reaches a

supersaturation point, continued feeding of Cd and Te-precursor results in nucleation of a solid CdTe phase, forming a liquid/solid interface. Ideally, for small alloy nanodroplets, only one liquid/solid interface can be formed due to limited volume. However, the possibility of forming multiple interfaces increases for large alloy nanodroplets, leading to multiple wire arms (Figure 1-15). Stage III: Axial growth of CdTe quantum wire. It's energetically more favorable for the incoming Cd- and Te-precursor to diffuse to and react at the existing solid/liquid interface than to form a secondary nucleation site. Consequently, the wire will keep growing until the precursors are consumed. Stage IV: Phase separation in Bi-Cd-Te ternary alloy nanodroplet and epitaxial recrystallization of Bi. Upon cooling, the solubilities of Cd and Te in the alloy droplet should decrease, which reasonably results in a phase-separation process accompanied by a further growth of CdTe quantum wire segment. The wire length that can be achieved from this final growth should be very short due to lack of Cd and Te precursors. When Cd and Te in the droplet are nearly used up, the recrystallization of Bi will occur upon further cooling to room temperature.

Conclusion Bi-catalyzed SLS growth of wurtzite CdTe quantum wires of the highest quality was achieved from Cd(ODPA) and TOPTe precursors (Cd/Te = 5) due to appropriately balanced precursor reactivities. The mean diameters of the wires could be purposefully controlled in the range of 5-11 nm by varying the diameters of the Bi catalyst

57

nanoparticles. The standard deviation in the diameter distributions was less than 20 % of the mean diameter. At least four well-resolved features were evident in the UV-Vis absorption spectra, consistent with the high-quality of the wires. Diameter-dependent quantum-confinement effects were demonstrated by the diameter dependence of the quantum-wire band gaps. On one hand, these quantum wires can serve as potentially excellent candidates for fabrication of nanoscale devices, such as solar cells.2 On the other hand, the high quality of the wires allows us to study their fundamental electronic structure, as will be described in Chapter 2 and Chapter 3.

Experimental Section (1) CdTe nanowires prepared by orientated attachment approach Materials. Thioglycolic acid (TGA, 98%), tellurium powder (Te, -200 mesh, 99.8%), sodium borohydride (NaBH4, 99%), sodium hydroxide (NaOH, pellets, ACS reagent grade, 97+%), methanol (99.8+%, ACS reagent grade) were purchased from Aldrich and used as received. Cadmium perchlorate (Cd(C104)2-6H20) was purchased from Alfa Aesar and used as received. Preparation of CdTe nanocrystals. CdTe nanocrystals were prepared by a literature method.36 Typically, Cd(C104)2-6H20 (1.80 g, 4.3 mmol) and TGA (1.20 g, 13.0 mmol) were mixed in de-ionized (DI) water (250 mL). The solution pH was adjusted to 11.3 by adding an aqueous solution of NaOH (1 M), and then the solution was continuously stirred for about 30 min at room temperature with nitrogen purging to eliminate dissolved oxygen. Separately, Te powder (254 mg, 2 mmol) and NaBH4 (165 mg, 4.4 mmol) were loaded into a small flask (5 mL) with a small stir bar in the glovebox, followed by

58

injecting air-free DI water (2 mL). The flask was then immersed in an ice-water bath at ~0 C with stirring. In ~1 hr the black Te powder disappeared and the color of the solution was changed to pinkish. This freshly prepared pinkish sodium hydrotelluride (NaHTe) solution was then transferred by a cannula into the solution of Cd(C104)2-6H20 containing TGA at room temperature under vigorous stirring. The reaction mixture was refluxed and aliquots were taken out at different time intervals for optical measurement and the self-assembly of CdTe nanocrystals. Preparation of CdTe nanowires. CdTe nanowires were prepared by a literature method.15 Briefly, CdTe nanocrystals were precipitated by adding excess amount of methanol (~5 mL) to the as-prepared CdTe nanocrystals solution (2 mL). The precipitate was collected by centrifugation and then redispersed in DI water (2 mL) at pH 9.0. The resultant solution was allowed to age in dark at room temperature for 7 days. (2) CdTe nanowires prepared by SLS growth mechanism Materials. Cadmium oxide (CdO, 99.99+%), tellurium (Te, -5+50 mesh, 99.99%), oleic acid (OA, tech., 90%), palmitic acid (PA, 99%), trioctylphosphine (TOP, tech., 90%), trioctylphosphine oxide (TOPO, tech., 90%), methanol (99.8+%, ACS reagent), 2propanol (99.5%, HPLC grade), and toluene (99.8%, HPLC grade) were purchased from Aldrich and used as received except for TOPO, which was vacuum distilled before use. Cadmium stearate and n-Octadecylphosphonic acid (ODPA) were purchased from Strem and PolyCarbon Industries, respectively, and used as received. Preparation of Bi-nanoparticles. Bi catalyst nanoparticles were prepared by a modified literature method17 and detailed syntheses of nanoparticles over a range of sizes will be reported elsewhere.

59

Preparation of a stock TOPTe solution.

The precursor tri-n-octylphosphine

telluride (TOPTe) was prepared from a mixture of elemental Te (0.256 g, 2 mmol) and TOP (79.744 g). The mixture was stirred at about 240 C under dry, 02-free N2(g) until all elemental Te disappeared (-15 min) to form a transparent yellowish solution. The resulting solution was cooled to room temperature and stored in the glovebox for use. General preparation of CdTe quantum wires. All procedures were conducted under dry, 02-free N2(g). The quantities of reagents used are recorded in Table 1-4. In a typical preparation, CdO, ODPA, and TOPO were loaded into a Schlenk reaction tube. In a separate vial, the Bi-nanoparticle stock solution and TOPTe stock solution were mixed, and the vial was septum capped. The reaction tube was then inserted into a preheated salt bath (NaN03/KN03, 46:54 by weight) at 300-330 C to obtain a clear solution, and then transferred to another preheated salt bath at a desired temperature (see Table 1-4). The reaction mixture was allowed equilibrate to the salt bath temperature for several minutes, and then the contents of the vial were quickly injected into the reaction tube by a syringe. Within 10 s the color of the reaction mixture was observed to change to an apparent black (actually a dark red-brown) as the nanowires grew. The reaction tube was withdrawn from the salt bath 5 min after the injection and allowed to cool. Before the reaction mixture was solidified, toluene (~ 5 mL) was added to the reaction tube to prevent solidification. The CdTe-nanowire product was indefinitely stable in this form, and was typically stored in this form. The CdTe nanowires were separated from the reaction mixture by adding a minimum amount of 2-propanol (~ 5 mL) to the mixture, whereupon the TOPO and soluble byproducts dissolved and the nanowires precipitated. The precipitated wires were

60

collected by centrifugation. The precipitated wires could be redispersed in toluene, hexanes, and chloroform. Table 1-4. Synthetic conditions and results for CdTe quantum wires and other morphologies.
CdO (mmol) 0.19 0.03" 0.03" 0.15 0.30 0.15 0.25 0.35 0.45 0.06 0.10 0.14 0.18 0.06a 0.10a 0.14" 0.18" 0.12 0.12 Acid* (mmol) OA 0.59 OA 0.48 OA 0.88 OA 0.46 OA 0.77 OA 1.09 OA 1.36 OA 0.18 OA 0.30 OA 0.42 OA 0.54 -OA 0.60 OA 0.64 TOPO (g) 4" 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 TOPTe (mmol) 0.02 0.30 0.60 0.03 0.03 0.05 0.05 0.05 0.05 0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.02* 0.02* T (C) 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 260 Time (min) 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 3 3 Bi stock solution (mg) 30 30 30 30 30 30 30 30 30 30 30 30 30 30 30 30 30 20 20
"wire

(nm) 9.4 10 10 10 10 9.4 9.4 9.4 9.4 9.4 9.4 9.4 9.4 9.4 9.4 9.4 9.4 4.1 6.1

(nm) -~ 5.1 5.0

Notes dots mainly dots dots and rods wires and dots wires and NPs wires and NPs wires and NPs wires and NPs wires and NPs mainly wires mainly wires mainly wires mainly wires mainly wires wires and dots wires and dots
wires

and dots wires wires

61

0.12 0.18 0.18 0.18 0.18 0.10 0.11 0.10 0.11 0.10 0.11 0.11 0.12 0.11 0.12 0.10
a

OA 0.71 OA 0.54 OA 0.55 OA 0.56 OA 0.56 ODPA 0.15 ODPA 0.20 ODPA 0.20 ODPA 0.20 ODPA 0.20 ODPA 0.20 ODPA 0.20 ODPA 0.20 ODPA 0.20 ODPA 0.20 TDPA 0.20

4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4

0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.02

255 260 260 260 320 250 274 273 272 275 277 280 285 281 288 310

4 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5

40 30 30 30 30 20 20 20 28 18 20 20 20 20 12 13

7.8 7.8 9.4 12.2 15.9 5.8 7.8 9.4 8.5 12.4 12.9 13.5 15.8 14.3 16.5 -40

5.5 6.7 7.4 7.9 11.1 5.3 5.8 7.3 7.4 7.9 8.7 9.3 9.7 10.2 10.3 -19

wires wires wires wires wires wires wires wires wires wires wires wires wires wires wires wires

Cadmium stearate. b OA = oleic acid; ODPA = n-octadecylphosphonic acid; TDPA = n-tetradecylphosphonic acid. c Additional palmitic acid (PA, 0.10 mmol) was added. d 1-Octadecene (ODE). e TBPTe Low-resolution TEM, energy-dispersive X-ray spectroscopy (EDS), and high resolution TEM (HRTEM). All the wire samples were purified by isopropanol/toluene precipitation from the reaction solution. Carbon-coated copper grids were prepared by evaporating one drop of the nanowire toluene solutions on them. Low-resolution TEM images and EDS analysis of CdTe quantum wires were collected using a JEOL 2000 FX

62

microscope operating at 200 kV. HRTEM images were collected using a JEOL JEM21 OOF microscope operating at 200 kV. Statistics of wire diameters. The wire diameters were measured by a commercial software Image-Pro plus (version 5.0) at the 2x zoom from -400 wires in TEM images taken at a magnification of 500K. The measured diameter data were then transferred to standard spreadsheet software, such as Microsoft Excel and Origin 7.5, for statistical analysis. The statistical result was expressed as the mean wire diameter one standard deviation. X-ray powder diffraction (XRD). The isolated CdTe nanowires were redispersed in a small amount of toluene (~1 mL). The wire dispersion was then dropped to a glass sample holder and dried in air. XRD data were collected using a Rigaku Dmax A vertical powder diffractometer with Cu Ka radiation {X = 1.5418 A), and analyzed using Materials Data Incorporated (MDI) automation and Jade software. Absorption and fluorescence spectroscopy. The UV-Vis absorption and the

photoluminescence spectra of CdTe samples were collected at room temperature using a Varian Cary 100 Bio UV-visible spectrophotometer and a SPEX Fluoromax spectrophotometer, respectively. Nonlinear least-squares fitting of the UV-Vis absorption spectra. The UV-Vis absorption spectra were fit using a modification of a method reported previously.17 Briefly, each absorption spectrum was converted from a wavelength to an energy scale, and then fit with one exponential (for the background) and multiple Gaussian functions (for the features) using Origin 7.5 software (www.OriginLab.com). The number of Gaussian functions used was determined by the number of absorption features present in

63

the spectra. The fitting procedure yielded the center energy of the excitonic peaks and the error in the center energy. The quantum-wire band gaps were assigned to the centers of the lowest-energy Gaussian peaks.

References (1) Shah, A.; Torres, P.; Tscharner, R.; Wyrsch, N.; Keppner, H. Science 1999, 285, 692-698. Gur, I.; Fromer, N. A.; Geier, M. L.; Alivisatos, A. P. Science 2005, 310, 462-465. Lim, Y. T.; Kim, S.; Nakayama, A.; Stott, N. E.; Bawendi, M. G.; Frangioni, J. V. Mol. Imaging 2003, 2, 50-64. Michalet, X.; Pinaud, F. F.; Bentolila, L. A.; Tsay, J. M.; Doose, S.; Li, J. J.; Sundaresan, G.; Wu, A. M.; Gambhir, S. S.; Weiss, S. Science 2005, 307, 538544. Lide, D. R. Handbook of Chemistry and Physics; 88th ed.; CRC press: Boca Raton, 2007. Murray, C. B.; Norris, D. J.; Bawendi, M. G. J. Am. Chem. Soc. 1993, 115, 870615. Peng, Z. A.; Peng, X. J. Am. Chem. Soc. 2001, 123, 183-184. Yu, W. W.; Wang, Y. A.; Peng, X. Chem. Mater. 2003, 75,4300-4308. Manna, L.; Milliron, D. J.; Meisel, A.; Scher, E. C ; Alivisatos, A. P. Nature Mater. 2003, 2, 382-385. Xu, D.; Guo, Y.; Yu, D.; Guo, G.; Tang, Y.; Yu, D. P. /. Mater. Res. 2002, 17, 1711-1714. Zhao, A. W.; Meng, G. W.; Zhang, L. D.; Gao, T.; Sun, S. H.; Pang, Y. T. Appl. Phys. A 2003, 76, 537-539. Xu, D.; Chen, D.; Xu, Y.; Shi, X.; Guo, G.; Gui, L.; Tang, Y. Pure Appl. Chem. 2000, 72, 127-135.

(2) (3)

(4)

(5)

(6)

(7) (8) (9)

(10)

(11) (12)

64

Enculescu, I.; Sima, M.; Ghiordanescu, V.; Secu, M. Chalcogenide Lett. 2005, 2, 9-15. Yang, Q.; Tang, K.; Wang, C ; Qian, Y.; Zhang, S. J. Phys. Chem. B 2002,106, 9227-9230. Tang, Z.; Kotov, N. A.; Giersig, M. Science 2002,297, 237-240. Yu, H.; Li, J.; Loomis, R. A.; Wang, L.-W.; Buhro, W. E. Nat Mater 2003, 2, 517-520. Yu, H.; Li, J.; Loomis, R. A.; Gibbons, P. C ; Wang, L. W.; Buhro, W. E. J. Am. Chem. Soc. 2003,125,16168-16169. Li, J.; Wang, L. W. Chem. Mater. 2004, 16, 4012-4015. Li, J.; Wang, L.-W. Phys. Rev. B 2005, 72, 125325. Yu, H.; Li, J.; Loomis, R. A.; Wang, L.-W.; Buhro, W. E. Nature Mater. 2003, 2, 517-520. Wang, F.; Yu, H.; Li, J.; Hang, Q.; Zemlyanov, D.; Gibbons, P. C ; Wang, L. W.; Janes, D. B.; Buhro, W. E. J. Am. Chem. Soc. 2007, 129, 14327-14335. Tang, Z.; Ozturk, B.; Wang, Y.; Kotov, N. A. J. Phys. Chem. B 2004, 108, 69276931. Zhang, H.; Wang, D.; Yang, B.; Mohwald, H. J. Am. Chem. Soc. 2006, 128, 10171-10180. Zhang, H.; Wang, D.; Mohwald, H. Angew. Chem. Int. Ed. 2006, 45, 748-751. Niu, H.; Zhang, L.; Gao, M.; Chen, Y. Langmuir 2005, 21,4205-4210. Wang, Y.; Tang, Z.; Liang, X.; Liz-Marzan, L. M.; Kotov, N. A. Nano Lett. 2004, 4,225-231. Gaponik, N.; Talapin, D. V.; Rogach, A. L.; Hoppe, K.; Shevchenko, E. V.; Kornowski, A.; Eychmuller, A.; Weller, H. J. Phys. Chem. B 2002,106, 71777185. Pacholski, C ; Kornowski, A.; Weller, H. Angew. Chem. Int. Ed. 2002, 41, 11881191. Wang, F.; Dong, A.; Sun, J.; Tang, R.; Yu, H.; Buhro, W. E. Inorg. Chem. 2006, 45,7511-7521.

65

(30) (31)

Yu, H., PhD Dissertation, Washington University, 2003. Olson, E. A.; Efremov, M. Y.; Zhang, M.; Zhang, Z.; Allen, L. H. J. Appl. Phys. 2005, 97, 034304/1-034304/9. ICDD-PDF No. 01-080-0019. ICDD-PDF No. 00-044-1246. Dong, A.; Wang, F.; Daulton, T. L.; Buhro, W. E. Nano Lett. 2007, 7,1308-1313. Wu, Y.; Yang, P. J. Am. Chem. Soc. 2001, 123, 3165-3166. Rogach, A. L.; Nagesha, D.; Ostrander, J. W.; Giersig, M.; Kotov, N. A. Chem. Mater. 2000, 12, 2676-2685.

(32) (33) (34) (35) (36)

66

Chapter 2 Similarity of the Effective Band Gaps of CdTe Quantum Wires to Those of Equidiameter CdTe Quantum Dots

67

Introduction Over the past two decades, semiconductor quantum nanostructures such as quantum dots, rods, wires, and wells have drawn increasing scientific and technological interest.1"7 Theoretical8"10 and experimental studies11"16 have begun to elucidate how the geometric dimensionality of confinement influences the electronic structures of quantum-confined systems. The increasing availability of semiconductor quantum wires has allowed us to draw experimental comparisons between wires and corresponding sets of quantum dots,13"15 rods,14'16 and wells.17 We have previously compared the confinement in quantum wires and dots using an overly simple effective-mass-approximation, particle-in-a-box (EMA-PIB) model to estimate the kinetic confinement energies of electron-hole pairs. According to this model, plots of Ag, the increase in the effective band gap over the bulk value, vs. cf2, the inverse-square diameter, should yield straight lines having slopes that depend on the confinement dimensionality and thus shape of the quantum nanostructure. The slope ratio for dots and wires of the same composition is predicted by the EMA-PIB model to be Awire^dot = 0.585, and the measured experimental values for InP (Awire:Adot = 0.62 + 0.03 and 0.66 + 0.03)13'15 and CdSe (Awke:A6ot = 0.53 + 0.05)14 were close to this value. These prior results have confirmed that confinement in the wires is weakened to the expected extent relative to that in dots by the loss of one confinement dimension. In this Chapter, we report that the AWire:Ad0t slope ratio extracted from the AEg-vs.-d"2 plots for CdTe quantum wires and dots was 1.09, near unity, indicating a convergence of the dot and wire effective band gaps, over the diameter regime studied. The result stands in contrast to the EMA-PIB prediction and the wire-dot comparisons from the CdSe and

68

InP systems.

To investigate this disparity, a higher-level theoretical analysis was

conducted using the results of density-functional-theory calculations.10 The theoretical analysis provided here found that the AEg-vs.-d~2 plots for semiconductor wires and dots generally consist of two regimes: a convergence regime above a threshold diameter (d^) in which dot and wire effective band gaps are effectively indistinguishable, and a divergence regime below d^ in which the AWjre^Adot slope ratios approximate the EMA-PIB value of 0.6. The dot-wire comparisons for the CdTe system presented here were thus made in the convergence regime. The origin and magnitude of the threshold diameter d& are discussed. We propose that the convergence regime results from differing electron-hole Coulomb energies in dots and wires, and that d^ increases with decreasing electron-hole Coulomb energy. Thus, as should be expected, the dot-wire systems having comparatively small Coulomb energies conform most closely to the overly simple EMA-PIB model.

Results Experimental Awire:Adot slope ratio for CdTe quantum wires and quantum dots in the AEg-ys.-d~2 plot. The quantum confinement energies (AEg) of high-quality CdTe quantum wires of various mean diameters have been summarized in Table 1-3 (see Chapter 1). To compare those data to the corresponding CdTe quantum-dot data, we extracted the size-dependent first excitonic absorption peak positions of high-quality CdTe quantum dots reported by Peng and coworkers18 using Digitizer in Origin 7.5 software. The AEg values for CdTe quantum dots were then calculated and summarized in Table 2-1.

69

Table 2-1. First excitonic absorption peak positions (A) and quantum confinement energies (AES) of CdTe quantum dots of various diameters (d).
d(nm) 3.42 3.82 4.01 4.12 4.19 4.58 5.31 5.71 6.60 7.00 7.38 7.72 7.86 8.44 8.52 9.12

A(nm)
569.4 607.1 605.1 641.1 648.2 654.9 682.8 692.9 713.2 717.9 721.9 726.9 733.1 736.8 741.8 744.9

AEe (eV) 0.677 0.542 0.548 0.434 0.413 0.393 0.316 0.289 0.239 0.227 0.218 0.206 0.191 0.183 0.171 0.164

As noted in the introduction, we have previously compared the confinement in corresponding sets of quantum wires and dots using the EMA-PIB model. This model predicts straight-line plots of AEg vs. d~2 (inverse-square diameter) for dots and wires, with a slope ratio of AwkeAdot = 0.585. The measured experimental values for InP (Awire^ot = 0.62 + 0.03 and 0.66 + 0.03)13,15 and CdSe (A^A^t = 0.53 0.05)14 were consistent with this prediction. We then similarly compared the confinement in CdTe quantum wires and dots by graphing AEg vs. cT2 (Figure 2-1) using the data from Table 1-3 for CdTe wires, and the data from Table 2-1 for CdTe dots. Linear fits to the wire and dot data yielded the slopes Awire = 7.2 + 0.3 eV nm2 and A<j0t = 6.6 0.3 eV ran2, respectively. The experimental
70

slope ratio was thus found to be Awiie^ot = 1.09 0.07, which to our initial surprise was far from the theoretical value of 0.585 predicted by the EMA-PIB model.9'10'13"15 Indeed, the experimental slope ratio near unity indicated that the size dependences of the effective band gaps in CdTe wires and dots were essentially indistinguishable, over the diameter regime investigated. d(nm)

5f

0.04

0.06

0.08

0.10

d2(nm2) Figure 2-1. Experimental CdTe quantum wire (red open squares) and quantum dot (blue open circles) data plotted as AE% vs. d2. The red and blue solid lines are the linear leastsquares fits to the wire and dot data, respectively, yielding the slopes AWjre = 7.2 0.3 eV-nm2 and Adot = 6.6 0.3 eV-nm2 and the ratio of the slopes Awire/Adot = 1.09 0.07. Threshold diameter for CdTe quantum wires and dots. As noted above and previously,9'10'13"15 the EMA-PIB model is an overly simple approximation, and the linear fits described above do not fully correspond to the expected confinement behavior. For example, both the dot and wire lines should extrapolate through the Figure 2-1 origin, as quantum confinement disappears at large diameters. The EMA-PIB model assumes

71

parabolic band shapes, bulk (size-independent) effective masses, and infinite potential barriers at the dot and wire surfaces, which are severe approximations. Additionally, the model ignores the Coulomb interaction within electron-hole pairs. We will argue later that these Coulomb interactions are very important for determining the relative confinement behaviors of dots and wires.
1.8

1.6-1
1.4 1.2 > 1.0

CdTe QWs

a,01 o.8
0.6 0.4 0.2
1 ' 1
r

AE =2.05/d'
8

1.0

1.5

2.0

nr 2.5 3.0

Tiir-

3.5

4.0

4.5

5.0

Diameter (nm)

Figure 2-2. Diameter dependence of quantum confinement energies of CdTe quantum wires. The solid line is the fit to the calculated data points using the equation of AEg =
pcT01.

We thus sought to interpret the experimental results in Figure 2-1 with the assistance of a higher-level theoretical analysis. Li and Wang have recently calculated the size dependence of the effective band gaps in a series of quantum dots and wires using density functional theory under the local-density approximation by implementing a chargepatching method (DFT-LDA-CPM) and with LDA band-gap corrections,10 which allows ab initio accuracy for thousand-atom calculations of any given semiconductor materials. By this method, the effective band gaps of dots and wires were calculated over a range of

72

specific diameters, and the results were fit with equation 2-1, where or and /?are fitting parameters that were found to depend on both nanostructure composition and shape. AEg = j3dTa (2-1)

For CdTe quantum dots, the fitting parameters were < % o t = 1.69 and /?dot = 4.38, which were directly given by Li and Wang.10 To obtain the fitting parameters for CdTe quantum wires, we fitted their calculated diameter-dependent quantum confinement energies of CdTe quantum wires to the general form shown in equation 2-1, and determined the fitting parameters to be a^ire = 1-24 and yfiUre = 2.05 (Figure 2-2). With all the fitting parameters available, we then redrew the zlEg-vs.-d"2 plot for CdTe quantum dots and wires (Figure 2-3), which allows us to make a close comparison between the experimental data and the theoretical results. Several aspects of the DFTLDA-CPM curves in Figure 2-3 merit consideration. First, these curves lie below the experimental data for CdTe dots and wires, indicating that the theoretical results have underestimated the experimental confinement energies. However, the theoretical curves were extrapolated from calculations on ranges of specific diameters that were smaller than the diameter ranges employed in the experimental studies. Thus, if additional calculations were completed for larger diameters contained within the experimental diameter ranges, the absolute agreement between the theoretical and experimental data should improve, as was previously found for InP quantum wires.15 Therefore, the curves plotted in Figure 2-3 will be useful here for relative rather than absolute comparisons to the experimental data.

73

d(nm)
8 7 6 5 4

-i

0.00

0.02

0.04
2

0.06

0.08
2

0.10

0.12

d (nm )

Figure 2-3. Experimental CdTe quantum-wire (red open squares) and quantum-dot (blue open circles) data, and theoretical CdTe quantum-wire (red dashed curve) and quantumdot curves (blue dashed curve) plotted as AE% vs. d~2. More interestingly, the rvalues (o&>t = 1-69, awire = 1-24) from the DFT-LDA-CPM results differ from the value of 2 determined by the EMA-PIB model. At higher levels of theory the dot and wire effective band gaps do not scale with d~2, but rather with <Ta where a is in the range of 1-2.10 These differences are obviously responsible for the gentle curvature of the theoretical plots in Figure 2-3. As a result of the curvature, the theoretical fits do extrapolate through the Figure-2-3 origin, as they should. Note also that the differences in the a and fi values for dots and wires cause the theoretical curves to cross one another. We will refer to this crossover point as the threshold diameter, d&, which for CdTe dots and wires is found to be 5.40 nm. We do not believe that the curve crossing in Figure 2-3 is physically real, as 2Dconfinement in wires should not exceed the 3D-confinement in the corresponding dots.9,10

74

Rather, the curve crossing is likely an artifact of the curve extrapolations from the effective band-gap calculations for smaller diameter dots and wires (see above). Consequently, here we characterize the crossover point as the threshold diameter (d^), above which the dot and wire curves approach one another closely (see Figure 2-3). That is, we will use dth as the minimum approximate diameter at which the dot and wire curves effectively converge. As a consequence of this effective convergence, for diameters greater than d^, the dot and wire effective band gaps become, in practice, experimentally indistinguishable. The experimental comparisons of CdTe dot and wire effective band gaps in Figure 23 were limited to the diameter range of 5-10 nm, because we are presently unable to synthesize CdTe quantum wires having diameters smaller than 5 nm (see Chapter 1). As this diameter range lies above the theoretical d& for CdTe dots and wires, we were limited to comparisons in the regime where dot and wire effective band gaps are indistinguishable. Consequently, we measured an experimental slope ratio within this size regime of Awn-e^dot = 1-09 (see above), near unity, in accord with the higher-level theoretical predictions. Interestingly, we applied linear fits to the theoretical curves in Figure 2-3 in the other regime, in which the diameters were less than d^ (see Figure 2-4). We then calculated a "theoretical" slope ratio of /We^dm = 0.63 0.01 for this small-diameter regime. The value calculated is close to the slope ratio of 0.585 predicted by the simple EMA-PIB model. The result suggests that if we are ultimately able to collect effective-band-gap data from smaller-diameter CdTe wires, a slope ratio close to the EMA-PIB prediction may yet be observed. Consequently, two regimes apparently exist in effective-band-gap

75

comparisons of CdTe dots and wires, for diameters above and below Jth- In the largediameter regime the effective band gaps converge and are thus indistinguishable, leading to a slope ratio near unity. In the small-diameter regime the effective band gaps diverge and the wire-dot slope ratios should be close to 0.6, in accord with our prior EMA-PIB rule of thumb.13"15 d(nm)
6 5 4 3

0.70.6C- 0.5Hi
TI

0.40.3-

0.2-1iiiiii|i|i|iiiiiii

0.03

0.04

0.05

0.06

0.07
2

0.08
2

0.09

0.10

0.11

0.12

of (nm )

Figure 2-4. Linear fits to theoretical CdTe quantum-wire (red open squares) and quantum-dot (blue open circles) data calculated from -4E,gtWjre = 2.05d~i24 and AE&!Aoi = 4.3M169, respectively, yielding the slopes A^e = 3.56 0.07 eV nm2 and Adot = 5.62 0.04 eV nm2 and the ratio of the slopes Awire:Adot = 0.63 0.01. Note that the diameter range is 3.0-5.4 nm.

Therefore, we may anticipate this two-regime behavior to be general. A threshold diameter, d^, may exist in other similar comparisons of semiconductor dots and wires, above which dot and wire effective band gaps are in practice indistinguishable. Our previous comparisons of InP13'15 and CdSe14 dots and wires found slope ratios near 0.6, suggesting that the d& values are comparatively large in those cases. In the CdTe case

76

reported here, the d^ value is shown theoretically and experimentally to be smaller. An obvious question arises. What determines whether d^ will be relatively large or small? In the following discussion the physical origin of and factors influencing the value of d^ are explored. Table 2-2. a and j8 values for various II-VI and III-V semiconductor quantum dots and quantum wires.
Material ZnO ZnS ZnSe II-VI ZnTe CdS CdSe CdTe A1N GaN HI-V GaAs InN InP InAs
a

Structure0 W W ZB ZB W

Quantum dot a&ot 1.35 1.29* 1.28 1.49 1.40 1.38* 1.45 1.18 1.69 1.29 1.23* 1.36 1.32* 1.02 1.06* 1.08 1.17* 1.31 1.21

Quantum wire
flwire Pwire

Aot
0.96 1.90* 1.36 2.54 3.00 3.81* 2.08 2.12 4.38 0.57 0.92* 1.98 2.75* 2.95 3.54* 2.29 2.92* 3.36 4.11

1.85 1.16 1.29 1.19 1.27 1.24 1.24 1.12 1.15 0.96 1.04 1.16 1.02

1.58 1.24 1.79 1.69 1.72 1.95 2.05 0.89 1.57 1.83 1.86 2.00 2.38

w
ZB

w w
ZB W ZB ZB

W = wurtzite; ZB = zinc blende. * No Coulomb energy.

Discussion We first address how the theoretical threshold diameter d^ for semiconductor dots and wires may be extracted from the relevant DFT-LDA-CPM data. The a and /3 values

77

of quantum dots and wires for several II-VI and III-V compositions were obtained directly from Li and Wang,10 or by fitting their theoretical effective band-gap data using the general form shown in equation 2-1 (see Figure 2-5 and Figure 2-6). These a and /3 values are summarized in Table 2-2. The value of d& was determined by setting wire and dot effective band gaps equal at d^ (equation 2-2), and solving for d^ (equation 2-3). The dth values calculated from equation 2-3 are shown in Figure 2-7 and Figure 2-8, and also recorded in Table 2-3.

/L<C'=/LA^
A -(R
"th
-

(2-2)
(2-3)

IR

V'teda-ow.)

vA'dot ' HvAa)

Equation 2-3 establishes that d^ depends on two new, combined parameters, (o&or fl^ire) and y#d0t/Avire- The dependence of d& on these combined parameters is plotted in Figure 2-9. If (fifor-as^ire) is small (< 0.10) and positive, then /WAvire - 1-3 is sufficient to produce a large dth (^ 10 nm). As (efoot-Owire) becomes larger, then larger / W A r o values are required to produce a large dth- For CdTe quantum dots and quantum wires, the (cfoot-flwire) value is very large (0.45), however, the /WArire value is only 2.14, which is not large enough to obtain a large d^, leading to theoretically indistinguishable quantum-dot and quantum-wire band gaps. In two cases (Table 2-3), negative values of (efoot-tfwire) are found, indicating that at diameters less than dth the confinement in wires is stronger than that in dots, which is initially counterintuitive. For one such case, CdSe, we will argue below that the negative (<%0rwire) value is likely to be incorrect. Although we have catalogued here the relative values of (o&ot-flwire) and /WAire necessary to produce a large d^, the physical origin of these dependences remains to be elucidated.

78

0.90.80.7> 0.6V \ a

ZnS Q W s

1.81.41.2> 1.0-

ZnSe qWs

\ \ \
\ AE S1.79/CJ 1 "

J 0.50.40.30.2' 1 ' T

N v

AE=1.24/rf '
9

11

U,*0.8. 0.6-

" ^ " " " ^ " " " ^

0.40.24.0 1.0 1.S

^ ^ ^ ^^^-^^^^
2.0 2.5 3.0 3.5 4.0 4.5

2.0

2.5

3.0

3.S

Diameter (nm) 1.6 gnTe QPs a no Coulomb energy including Coulomb energy

Diameter (nm)

1l

2.0

2.5

3.0

3.5

4.0

4.5 Diameter (nm)

Diameter (nm)

1.21.00.80.60.40.21 '1

Cd (SW?

\JI

If

AE = 1 . 7 2 / d 1 "

*-

..._.,.,,...,. . _ , _ , - . . . _ ,

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

Diameter (nm)

Figure 2-5. Diameter dependence of quantum confinement energies of various II-VI quantum dots and quantum wires. The solid and open squares represent the calculated
quantum confinement energies for quantum dots and quantum wires with various

diameters. The solid and dotted curves are the fits to the calculated data points using the equation of AE& = J3tfa.

79

AIN QDs

0.6 A

AIN QWs

0.5 A

5 0.4

AE = 0.89/d" 0.3

1.5

2.0

2.S

3.0

3.5

1.0

1.6

2.0

2.5

3.0

3.5

4.0

Diameter (nm)
2.0-1 1.81.61.4,.1.2
<3

Diameter (nm)

GaAs QDs

1.81.6-

GaAs QWs

'-...
El,

n no Coulomb energy including Coulomb energy

1.4-

%
1.0 AE 0.8 0.6 2.0 2.5
8

i
1 > 1

1.21.0N ^

2MkfM*

^ ^ O , 1 1

0.80.60.4-

AE - 1.83/</"" a

a 2.0 2.5 3.0 Diameter (nm) 3.5 4.0 4.5

3.0

3.5

Diameter (nm)

2.2 2.0 1.8 1.6-

InAs QWs

% 1-2 1.0 0.8 0.6 0.4 1.0 1.5 2.0 2.5

AE 2.38/cf

3.0

3.5

4.0

4.5

Diameter (nm)

Figure 2-6. Diameter dependence of quantum confinement energies of various III-V quantum dots and quantum wires. The solid and open squares represent the calculated quantum confinement energies for quantum dots and quantum wires with various diameters. The solid and dotted curves are the fits to the calculated data points using the equation of AEg = J3d~a.

80

cL = 0.72 nm

0.0

0.2

0.4

0.6

0.8

1.0

1.2
2

1.4

1.6

1.8

2.0

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0

(nm- )
0.6

d2 (nm-2)

ZnTe
0.5 Dot curve (no Coulomb energy) 0.4 0.3 0.2 0.1
i11Ii(IirTiIIII|'I Dot curve (including Coulomb energy)

0.00 0.0S 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50

0.02

0.03

0.05

d"2 (nm-2)

d2 (nnr2)

CdSe
108Dot curve ^^^

dm 0.25 nrn,^^

5f
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.8 1.8 2.0

6420- i ~ r ~

^ ^ ^

Wire curve

10

12

14

16

d"2 (nm-2)

d2 (nm-2)

Figure 2-7. Plot of theoretical quantum confinement energy (AES) vs. d for various IIVI quantum dots (black curves) and quantum wires (red curves). The threshold diameters for each material are shown in the graph.

81

1.41.21.00.80.60.40.20.0

3.5 AIN
Dot curve "'i^^^ (no Coulomb energy) ^^s^*^ ...^^'^Wire curve d|h = 1,35nm ^ j f jr ^~-^ ^ ^^^"^^ ^^-~-"" Dot curve (including Coulomb energy)

3.0 2.5

OaN > Dot curve (no Coulomb energy) Dot curve (including Coulomb energy)

^
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0

0.4

0.6

d2 (ran"2)
4.S 4.0GaAs

d 2 (nm2)
0.35
InN

0.30
Dot curve (no Coulomb energy) Dot curve (Including Coulomb energy)

3.5 3.0 2.52.0 1.51.0 0.5

0.25-

Dot curve (no Coulomb energy) Dot curve (Including Coulomb energy)..

0.4

0.6

0.005 d
4

0.010 (niti" )
2

0.015

d2 (nnv2)

InP
0.25Dot curve

0.70.6

InAs

0.20-

0.5Wire curve

Dot curve^ Wire curve

0.15 0.100.05/ 0.00dm = 31.8 nm

. 1
i

if
1 1

0.40.30.20.1-

0.010

0.015
2

0.020

0.00.00

0.01

0.02

0.03

0.04

0.05

d' (nm- )

(nm )

Figure 2-8. Plot of theoretical quantum confinement energy {AE) vs. d for various IIIV quantum dots (black curves) and quantum wires (red curves). The threshold diameters

for each material are shown in the graph.

82

Table 2-3. (o&ot-Owire), / W A r o values, and threshold diameters (Jth) for various II-VI and III-V quantum systems. Material ZnO ZnS ZnSe II-VI ZnTe CdS CdSe CdTe A1N GaN III-V GaAs InN InP
a

Structure0 W W ZB ZB W W ZB W W ZB

Q&orOwire

Pdot'Pwke

dth (nm)

-0.50 -0.56* 0.12 0.20 0.21 0.19* 0.18 -0.06 0.45 0.17 0.11* 0.21 0.17* 0.06 0.10* 0.04 0.13* 0.15

0.61 1.20* 1.10 1.42 1.78 2.25* 1.21 1.09 2.14 0.64 1.03* 1.26 1.75* 1.61 1.93* 1.23 1.57* 1.68

2.71 0.72* 2.16 5.75 15.4 72.1* 2.87 0.25 5.40 0.07 1.35* 3.02 27.0* 2859 733.7* 181 32.1* 31.8 17.7

w
ZB

ZB 0.19 InAs 1.73 W = vvurtzite; 23B s zinc bletide. *NoCoulom o energy.

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0

2.2

"dot "wire

Figure 2-9. Plot of threshold diameter vs. /WAvire at various (a<iot-awire) values indicated by the numbers shown in the graph. The red dashed lines are guides to the eye.

83

The simple EMA-PIB model referred to above, which accounts only for the kinetic confinement energies of electrons and holes and ignores the Coulomb energies associated with their interactions, predicts that dot and wire effective band gaps should converge only at infinite d, which in practice would be at 2-3 times the bulk exciton Bohr radius. According to Table 2-3, the DFT-LDA-CPM calculations indicate that convergence may occur at smaller d^ values, but what physical effect(s) should drive such convergence? We note here that the DFT-LDA-CPM calculations for dots may either include or omit the Coulomb energies resulting from interactions of electrons and holes. In contrast, an ab initio theoretical method for estimating the Coulomb interactions in wires has not yet to our knowledge been developed.19"21 Thus the Coulomb contributions are not included in the Table-2-3 data for wires. We will argue immediately below that the Coulomb interactions in the experimentally studied wires are significantly smaller than those in the corresponding dots. We will then propose that the comparative Coulomb interactions in dots and wires are responsible, at least in part, for their effective band-gap convergence at relatively small diameters. Although the electron-hole Coulomb energies in quantum wires have not been calculated by ab initio methods, they have been investigated by effective-massapproximation theory.19"21 Unlike the case for a quantum dot, the Coulomb energy for an electron-hole pair in a quantum rod or wire is significantly influenced by the polarizability of the surrounding medium.19,22'23 Brus's estimation of the Coulomb energy in a quantum dot is -3.572e2/4^s3esemid (fisemi = semiconductor dielectric constant), which does not depend on the dielectric constant of the medium, ned-24 In contrast, e^e<i must be taken into account to calculate the Coulomb energy in a rod or wire.

84

160-I 140120100CdTe

>"
| 80Uj* 604020oi i
1

QDs

QWs

1
1

i
1

"
i

'

8 d(nm)

10

11

Figure 2-10. Diameter dependence of the estimated exciton binding energies in CdTe quantum dots (QDs) and quantum wires (QWs). Shabaev and Efros have provided an analytical approximation for the Coulomb energy in a quantum wire as a function of ^,ed (see equation 2-4, where | z I is the distance between electron and hole, which we assume to be equal to the bulk CdTe exiton Bohr radius <3B = 7.5 nm, and /?eff = 035d.).23 Using this approximation and the dielectric constant of the capping ligands for ^ed (Sued is set equal to that of diethyl octadecylphosphonate, ^e<i = 4.05,25 an ester of the presumptive surface-capping ligand for the quantum wires26'27), we estimate electron-hole Coulomb energies of 39-32 meV in CdTe quantum wires having diameters in our range of 5-11 nm, respectively. The Coulomb energies of CdTe quantum dots obtained from the Brus estimation over the same diameter range are 145-66 meV (Figure 2-10). Therefore, the Coulomb energies in the wires are suggested to be only 27-48% of the corresponding values for dots. As noted above, the DFT-LDA-CPM calculations for quantum wires completely neglected the Coulomb contributions, and yet achieved good agreement between theoretical and

85

experimental effective band gaps.

Both of these arguments support significantly

smaller electron-hole Coulomb energies in wires (capped with synthetically relevant ligands) than in the corresponding dots.
b,wire = 2 / [ 4 ^ b ^ e d ( l z l + A f f ) ] (2-4)

One means of theoretically assessing the influence of electron-hole Coulomb energies on dot-wire effective band-gap convergence is to compare calculated d^ values when the Coulomb interactions in the dots have been included and omitted, respectively. Ideally the comparison here would use theoretical data for CdTe dots and wires. However, effective-band-gap data for CdTe quantum dots with omission of the Coulomb interactions were not reported.10 Consequently, we make the comparison using

theoretical data for the closely related ZnTe dots and wires. Figure 2-11 plots the theoretical diameter dependence of the effective band gaps in ZnTe wires, ZnTe dots without Coulomb interactions, and ZnTe dots with Coulomb interactions, using the data from Li and Wang.10 As expected, the Coulomb interactions decrease the effective band gaps in the dots, such that the curve for dots with Coulomb interactions lies below the curve for dots without them. A consequence of this curve lowering is an apparent decrease in d^; that is, the crossing point of the dot and wire curves moves to a smaller diameter. The d^ calculated from equation 2-3 with the dot curve omitting Coulomb interactions is 72.1 nm. (This finite d^ value is likely an artifact of fitting the theoretical results to the equation 2-1 approximation; the value probably should be infinite because the kinetic confinement energy in a dot should always be larger than that in a wire of equal diameter.) This d^ value shrinks to 15.4 nm when the

86

dot curve including Coulomb interactions is used.

Therefore, strong Coulomb

interactions in the dots relative to the wires will generally enforce a smaller dth.
cf(nm)
1211 10
0.30-1iiiiiiii i|i i iii 1

9
1

8
1

7
1 -p

0.25-

0.20-

> *

0.15-

V
0.100.05-

0.00-| 0.000

-T 0.005

, 0.010

1 0.015

1 0.020

1 0.025

d2 (nm'2)

Figure 2-11. Theoretical ZnTe quantum-wire (red) and quantum-dot (dashed blue, no Coulomb energy; solid blue, including Coulomb energy) curves plotted as AES vs d~2. These curves were obtained by fitting DFT-LDA-CPM effective-band-gap calculations to AE% = 0d~a (see text). The threshold diameters (d^ values) were determined to be 72.1 nm (not shown) and 15.4 nm (gray line) when the Coulomb energies in the dots were omitted and included, respectively. The comparison above suggests a rationale for the relative magnitude of d&. For compositions in which the Coulomb interactions are relatively small, they should be small in both dots and wires. Consequently, relatively large dth values should obtain, and the Awire^dot slope ratios (as defined herein) over limited diameter ranges below d& may approach the rule-of-thumb value of 0.6 predicted by the EMA-PIB model. In contrast, for compositions in which the Coulomb interactions are relatively large, they should be larger in dots than in the corresponding wires according to the argument above.

87

Consequently, in these cases the <iih values should be comparatively smaller and convergence of dot and wire effective band gaps thus achieved at smaller threshold diameters.

Table 2-4. Bohr radii (an), threshold diameters (dth), and calculated bulk exciton-binding energies (b,ex) for various II-VI and ffl-V systems. Material ZnO ZnS ZnSe II-VI ZnTe CdS CdSe CdTe A1N GaN III-V GaAs InN do? (nm) 2.71 2.16 5.75 15.4 2.87 0.25 5.40 0.07 3.02 2859 181
b,ex

(nm) 2.3 2.5 3.8 5.2 2.8 5.6 7.5 1.9 2.8 12.5 5.1

(unit in me) 0.19 0.20 0.13 0.10 0.16 0.094 0.085 0.26 0.18 0.062 0.097

semi

(meV) 38.9 29.5 24.3 13.0 30.3 14.0 10.7 42.5 24.1 5.1 15.2

8.1 9.6 8.6 10.3 8.43 9.57 10.4 9.14 10.1 12.80 9.3

InP 31.8 5.9 10 0.069 12.56 InAs 17.7 0.025 1.5 35 15.15 a Coulomb energ ies in quantum dots were inchided for the ca culation of d*. b Reduced exciton mass, calculated from the equation ju = me*mb*/( me* + mC), where me* and nth are the effective masses of electron and hole, respectively. c Dielectric constant of the semiconductor material. d Bulk exciton-binding energy, calculated from //^(S/j^emiW)" 1 - 8 Note: The we , m^ , and ,emi data were taken from (1) Semiconductors: Data Handbook, 3 rd ed; Madelung, O., Ed.; Springer: Berlin, 2004 and (2) Semiconductors: Basic Data, 2nd ed; Madelung, O., Ed.; Springer: Berlin, 1996.

Therefore, we now look for an empirical correlation between the calculated dth values and the Coulomb energies in various semiconductors (Table 2-4). We represent the Coulomb energies with calculated bulk exciton binding energies (Eb,ex), which in turn depend on electron and hole effective masses in and the bulk dielectric constants of

semiconductors. These values are also reported in Table 2-4. Figure 2-12 is a lognormal plot of dth vs. b,ex, in which an empirical correlation is clearly evident. In general, the smaller bulk exciton binding energies correspond to the larger threshold diameters, and visa versa. (Note that the da, values in Figure 2-12 were obtained from equation 2-3 using a and /? values from the dot calculations that included Coulomb interactions.) For example, the III-V GaAs, InN, InP, and InAs systems have small exciton binding energies; they correspondingly show large threshold diameters. For the II-VI zinc chalcogenides, we noticed the following consistent trends:
b,ex: ZnS > ZnSe > ZnTe fa ZnS < ZnSe < ZnTe
10" 7

bM(meV) Figure 2-12. Log-normal plot of threshold diameter vs. calculated bulk exciton binding energy for various II-VI and III-V semiconductors. The red line is the linear leastsquares fit to the solid data points. The green dashed line shows a threshold diameter of 10 nm; systems having threshold diameters below this value will exhibit dot and wire effective-band-gap energies that are not currently experimentally distinguishable (see text). Consequently, with the exception of CdSe (see text), we should not be able to measure EMA-PIB Awire:Adot slope ratios (approaching 0.6) for semiconductor systems that fall below the green dashed line. The arrow indicates what we believe to be the actual placement of the CdSe point. 89

The green dashed line drawn in Figure 2-12 at a threshold diameter of 10 nm estimates our current experimental limitation for characterizing the divergence regime for a set of quantum dots and wires; that is, for measuring an Awire:Adot slope ratio approaching the EMA-PIB rule-of-thumb value of 0.6. We estimate this value of 10 nm by considering that a slope determination would ideally span a wire diameter range of at least 5 nm, and that 5 nm is the approximate lower limit to the diameters of the wires we can presently synthesize. Recall that the EMA-PIB slope ratio should be observed only within diameter ranges below d^- Consequently, for semiconductors having da < 10 nm, we will not have access to a sufficiently wide diameter range for a reliable Ay/ae determination within the divergence regime. Therefore, for semiconductors below the green line in Figure 2-12, we should measure experimental A^eAioi slope ratios approaching unity, for the diameter range we have general access to (5-12 nm). For those above the green line we should measure experimental Awire:Adot slope ratios approaching 0.6. The semiconductor CdTe falls below the line, and we report herein an experimental slope ratio near unity. In contrast, InP falls above the line and we previously reported j4wire:.Adot slope ratios of 0.62 and 0.66.13'15 The semiconductor GaAs also falls above the line, and we have measured A^e for this system.17 Unfortunately, experimental effective-band-gap data for GaAs

quantum dots are not available, and so Adot and AWire/Adot are not known. However, the EMA-PIB model predicts a slope ratio for corresponding sets of quantum wells and quantum wires of Aweu:AWire = 0.427, and we determined experimental Awen:AWjre values of 0.41-0.49.17 The very large wire-dot d^ value (Table 2-4) for GaAs is a consequence of the very small Coulomb energy, and a similarly large well-wire do> value can be safely

90

assumed. Thus, the results for the CdTe, InP, and GaAs systems are consistent with the analysis proposed here and depicted in Figure 2-12. A contradiction to the analysis above is presented by the experimental results for the CdSe system. The position of CdSe in Figure 2-12, far below the green line, suggests that we should find an experimental Awire:Adot slope ratio near unity, whereas we have measured Awire:Adot to be 0.53,14 close to the EMA-PIB value. That is, our experimental CdSe dot-wire comparisons are clearly in the divergence regime, whereas Figure 2-12 predicts they should be in the convergence regime. As previously indicated, CdSe is one of only two cases in Table 2-3 for which theory finds a negative (<%0t-flwire) value. A negative (Oaot-ctwire) requires that in the divergence diameter regime, the wires will have larger effective band gaps than the dots. Our experimental results show the opposite;14 CdSe wires exhibit smaller effective band gaps than the dots of like diameter, as is normally expected. We note the possibility of significant errors in the theoretical or and J3 values reported in Table 2-2, as a consequence of the fittings and extrapolations used in their determination. Therefore, we believe that equation 2-3 should not be used to calculate d& in cases having negative (a0rwire) values. extrapolated theoretical and/or experimental data should be used. Consequently, we have used experimental effective-band-gap data for CdTe dots28 and wires14 to obtain experimental (flaot-fiwire) and dth values (Figure 2-13). The fitting of the experimental data affords (<%0rWe) = 0.70, which is a positive but quite large value by comparison to those listed in Table 2-3. We note that the fitting of the experimental data may also lead to significant errors associated with the limited diameter ranges studied. The experimental d^ value is found to be 9.54. The experimental findings of a Instead, precise non-

91

normal divergence regime for CdSe dots and wires in the diameter range of 5-11 nm, a positive (o&ot-Owire), and a d^ near 10 strongly suggest the placement of the CdSe point at > 10 nm, on or above the green line in Figure 2-12.

0.60.50.40.3Hi 0.2-

(a)
\*
CdSe QDs: * * - " 1 - 7 1 0.15
dot

/?dot - 4,32 0.87

A = 4.32d 1.71
0.10.0

0.08 r
I'
1

10

d(nm)

0.00

0.02

0.04

0.06
2

0.08

d" (nm )

Figure 2-13. Diameter dependence of the experimental effective band-gap energy (AE&) of (a) CdTe quantum dots and (b) quantum wires, and (c) a plot of AEg vs. d2 for CdTe dots and wires. The curves in (a) and (b) are fits to the experimental CdTe quantum dot28 and quantum wire14 data using equation AE% = J3d~a. The fitted a and J3 values from (a) and (b) are used with equation 2-3 to determine the experimental threshold diameter d& = 9.54 5.89 nm. The experimental (ocdot-otWire) value derived from the fitted values is 0.70 0.16. For the ZnTe, InN, and InAs quantum systems, the DFT-LDA-CPM calculations predict that we should measure experimental Awire:Adot slope ratios approaching 0.6.

92

Unfortunately, reliable comparisons between experimental quantum-wire and quantumdot effective-band-gap data for these systems are not currently available, which should be directions of future research. We do not argue that the electron-hole Coulomb energy is the only factor influencing the magnitude of the threshold diameter d^. Figure 2-12 contains considerable scatter, and shows several pair-wise relationships that are not fully consistent with the Coulombenergy explanation. For example, the calculated bulk exciton-binding energy in CdTe is smaller than that in InN, but a smaller threshold diameter is determined for CdTe than for InN (Table 2-4). Analysis of other factors also contributing to Jth will require further theoretical study. We also note that the theoretical d^ values calculated by equation 2-3 contain uncertainties resulting from the use of fitted, extrapolated a and J3 values. Even so, the empirical relationship between d& and the exciton-binding energy revealed in Figure 2-12 is strongly suggestive of a significant influence by electron-hole Coulomb energies.

Conclusion The rule-of-thumb wire-dot slope ratio of o.69'10'13"15 in the AEg-vs.-d~2 plot derived from the simple EMA-PIB model was found to break down for CdTe quantum wires and dots. The analysis herein reached the conclusion that the EMA-PIB slope ratio should be observed only within a divergence diameter regime below a threshold diameter dth, which decreases with increasing electron-hole Coulomb energy. In retrospect, the conclusion is almost obvious. As the EMA-PIB model does not account for Coulomb energies, it should break down whenever Coulomb energies become significant in comparison to the

93

kinetic confinement energies of electrons (or holes). Our analysis also proposes that in general there may be a diameter regime in which quantum-confined dot and wire effective band gaps converge, producing slope ratios near unity. To date, the CdTe system provides our only known example. However, as there are many semiconductors below the green line in Figure 2-12, other potential systems exist for testing this proposal, when we are able to synthesize quantum wires having smaller diameters. Observing both the convergence and divergence regimes for a single system should also then be possible.

References (1) (2) (3) Bhattacharya, P. K.; Dutta, N. K. Annu. Rev. Mater. Sci. 1993, 23, 79-123. Alivisatos, A. P. Science 1996, 271, 933-937. Xia, Y.; Yang, P.; Sun, Y.; Wu, Y.; Mayers, B.; Gates, B.; Yin, Y.; Kim, R; Yan, H. Adv. Mater. 2003, 15, 353-389. Buhro, W. E.; Colvin, V. L. Nat. Mater. 2003, 2, 138-139. Law, M.; Goldberger, J.; Yang, P. Annu. Rev. Mater. Res. 2004, 34, 83-122. Bhattacharya, P.; Ghosh, S.; Stiff-Roberts, A. D. Annu. Rev. Mater. Res. 2004, 34, 1-40. Michalet, X.; Pinaud, F. F.; Bentolila, L. A.; Tsay, J. M.; Doose, S.; Li, J. J.; Sundaresan, G.; Wu, A. M.; Gambhir, S. S.; Weiss, S. Science 2005, 307, 538544. Yoffe, A. D. Adv. Phys. 2002, 57, 799-890. Li, J.; Wang, L.-W. Chem. Mater. 2004, 16, 4012-4015. Li, J.; Wang, L.-W. Phys. Rev. B 2005, 72, 125325. Tang, Z.; Kotov, N. A.; Giersig, M. Science 2002,297, 237-240.

(4) (5) (6)

(7)

(8) (9) (10) (11)

94

Kan, S.; Mokari, T.; Rothenberg, E.; Banin, U. Nat. Mater. 2003, 2,155-158. Yu, H.; Li, J.; Loomis, R. A.; Wang, L.-W.; Buhro, W. E. Nat. Mater. 2003, 2, 517-520. Yu, H.; Li, J.; Loomis, R. A.; Gibbons, P. C ; Wang, L.-W.; Buhro, W. E. J. Am. Chem. Soc. 2003,125,16168-16169. Wang, R; Yu, H.; Li, J.; Hang, Q.; Zemlyanov, D.; Gibbons, P. C ; Wang, L.-W.; Janes, D. B.; Buhro, W. E. J. Am. Chem. Soc. 2007, 129, 14327-14335. Wang, F.; Buhro, W. E. /. Am. Chem. Soc. 2007, 129, 14381-14387. Dong, A., PhD Dissertation, Washington University, 2007. Yu, W. W.; Wang, Y. A.; Peng, X. Chem. Mater. 2003, 15, 4300-4308. Slachmuylders, A. F.; Partoens, B.; Magnus, W.; Peeters, F. M. Phys. Rev. B 2006, 74,235321. Xia, J.-B.; Cheah, K. W. Phys. Rev. B 1997, 55,1596-1601. Banyai, L.; Galbraith, I.; Ell, C ; Haug, H. Phys. Rev. B 1987, 36, 6099-6104. Muljarov, E. A.; Zhukov, E. A.; Dneprovskii, V. S.; Masumoto, Y. Phys. Rev. B 2000, 62, 7420-7432. Shabaev, A.; Efros, A. L. Nano Lett. 2004, 4, 1821-1825. Brus, L. E. J. Chem. Phys. 1984, 80, 4403-4409. Kosolapoff, G. M. J. Am. Chem. Soc. 1954, 76, 615-617. Wang, W.; Banerjee, S.; Jia, S.; Steigerwald, M. L.; Herman, I. P. Chem. Mater. 2007, 19, 2573-2580. Peng, Z. A.; Peng, X. J. Am. Chem. Soc. 2002, 124, 3343-3353. Murray, C. B.; Norris, D. J.; Bawendi, M. G. J. Am. Chem. Soc. 1993, 115, 87068715.

95

Chapter 3 Electronic Structure of Colloidal CdTe Quantum Wires

96

Introduction In recent years, several theoretical studies of the electronic and optical properties of II-VI1"3 and HI-V4"12 semiconductor quantum wires have appeared. Some of these have reported calculated optical spectra; however, to our knowledge only one direct comparison of calculated and experimental spectra has been previously reported.1 In that case, the theoretical spectra of CdSe quantum wires were compared to the experimental spectra obtained from long CdSe quantum rods. In contrast, studies of the electronic structure of semiconductor quantum dots have benefited greatly from comparisons of calculated energy-level diagrams and calculated optical spectra with experimental spectroscopic data, generally from photoluminescence excitation spectroscopy.13"18 Furthermore, related comparative studies for CdSe semiconductor quantum rods have revealed, among other insights, the geometric requirements for the crossover from a dark to bright exciton ground state.19"21 Consequently, efforts to correlate experimental and theoretical studies of quantum wires will be useful for characterizing their electronic structures and the size dependences of their optical properties. The lack of direct comparison of calculated and experimental spectra of quantum wires suffers from the synthetic challenge of producing high-quality quantum wires exhibiting highly structured optical spectra. Among various wire-growth approaches currently available, the solution-liquid-solid (SLS) method has proven to be the most successful approach to purposefully produce nanowires having diameters less than -10 nm,22 which is within the strong quantum confinement regime for many semiconductor compositions. In Chapter 1, we have demonstrated the SLS growth of colloidal CdTe quantum wires having lengths of several micrometers and well-controlled diameters of

97

5.3-10.2 nm, within the strong confinement regime. Interestingly and importantly, the absorption spectra of all the wires are richly structured, and contain several higher-energy absorption features. In Chapter 2, we analyzed only the diameter-dependence of the lowest-energy feature, and therefore the effective band gap, in the wires. In this Chapter, we report that the experimental absorption spectra of CdTe quantum wires are dominated by the first 15 conduction-band (CB) energy levels determined by DFT-LDA-CPM calculations (see Chapter 2).23 The results establish that the absorption features observed experimentally are not discrete transitions, but rather clusters of closely spaced transitions that are grouped by the energy spacing of the CB levels. Furthermore, we show that these CB energy levels follow the sequence predicted by the simple effective-mass-approximation, particle-in-a-cylinder (PIC) model, by assigning

symmetries to the DFT-LDA-CPM electron-density distributions. The PIC model is used to determine the size dependence of the electron effective mass, and to confirm that the low resolution of the experimental spectra is due to the quantum-wire size distribution and spectroscopic method employed rather than to closely spaced quantum-wire energy levels. The close correspondence of the DFT-LDA-CPM results and the simple PIC model provides a conceptually attractive framework for analyzing the electronic structure of semiconductor quantum wires.

Computational Methods The absorption spectrum was calculated using the DFT-LDA-CPM method as described previously for CdTe quantum dots and zinc-blende quantum wires,23 performed with a 35 Ryd planewave energy cutoff. The DFT-LDA-CPM approach attempts to

98

obtain the correct effective masses, and as a result the calculated band gap is smaller as is commonly observed in LDA calculations.23 We find that a constant shift of the bandgap by 0.38 eV agrees well wim all of the experimental quantum-wire results. To sample the Brillouin zone of the quantum wire, we explicitly calculated 5 ^-points along the c-axis of the quantum wire, and then linearly interpolated the energies and transition dipole matrix elements for 27 intermediate points. The imaginary part of the dielectric function, which is proportional to the optical absorption spectrum by a volume factor, was calculated as in Ref. 24. Table 3-1. Center energies of the absorption features (E\-E^) of CdTe quantum wires of various mean diameters (d).
d(nm) 5.3 5.4 5.8 7.3 7.4 7.9 8.7 9.3 10.2
Ei (eV)

E2 (eV) 2.065 2.056 1.967 1.856 1.806 1.773 1.759 1.738 1.711

E3 (eV) 2.418 2.409 2.264 2.089 2.020 1.957 1.922 1.900 1.856

E4 (eV) 2.794 2.741 2.659 2.354 2.309 2.213 2.154 2.108 2.051

E5 (eV)
~

E6 (eV)
-

1.850 1.816 1.786 1.721 1.707 1.684 1.675 1.665 1.650

2.660 2.652 2.630 2.349 2.281 2.199

2.638 2.646 2.635

Results and Discussion Diameter dependence of experimental absorption features. It has been shown in Chapter 1 that the UV-Vis absorption spectra collected from the toluene dilutions of the high-quality CdTe quantum wires typically exhibited multiple resolved absorption features (see Figure 1-29). Using the nonlinear least-squares fitting established in Chapter 1, we decomposed each spectrum to at least four absorption envelopes (Figure 3-

99

1), whose positions (center energies) are summarized in Table 3-1. In all cases, the lowest-energy feature was fit as a reasonably narrow Gaussian, whereas the higherenergy features were generally much broader.

Photon energy (eV)

Photon energy (V)

Figure 3-1. Representative experimental absorption spectra of CdTe quantum wires with diameters of (a) 5.3 nm, (b) 5.8 nm, (c) 7.3 nm, (d) 8.7 nm, (e) 9.3 nm, and (f) 10.2 nm. Fitted spectra (barely visible orange curves) are shown behind the corresponding experimental spectra. Individual fitting functions (one exponential function for the background and multiple Gaussian functions for the absorption features) of each fitted spectrum are also plotted (the absorption features are color-coded). 100

Based on the data in Table 3-1, we plotted the experimentally observed absorption features (relative to the energy of the first excited state) vs. energy of the first excited state for CdTe quantum wires of various diameters in Figure 3-2, which clearly shows the diameter-dependent evolution of the absorption features of CdTe quantum wires. We note that the x axis is chosen to be energy of the first excited state instead of the mean diameter of CdTe quantum wires because energy can be more precisely measured than the wire diameter. Thus CdTe quantum wires are better described by the energy of the first excited state than by the wire diameter.13 We also note that the sixth absorption features and some of the fifth absorption features may be unreliable (see the circled data in Figure 3-2) because the diameter dependence of those absorption features disappears, which is inconsistent with the other absorption features. We next sought to compare the experimental data with calculated absorption spectra.
Decreasing quantum-wire diameter

Energy of 1st Excited State (eV)

Figure 3-2. Plots of experimentally observed absorption features (relative to energy of the first excited state) vs. energy of the first excited state for CdTe quantum wires of various diameters. The numbers in parentheses represent the nth absorption features as indicated in Table 3-1. The data points for the 5th and 6th absorption features shown in the dotted circle may be unreliable because the diameter dependence of those absorption features disappears. 101

(a) 5.3 nm

(4a) (4b)(4c)

(b) 7.3 nm
(2)

(3)

1
tf
(1)

/\

M
/->>.. /"/" :;/"" *\.
i i i

/ \ /''<
1.8 2.0 2.2 2.4 2.6 2.8 1.8 2.0

P" "'?'< ,'

2.2

2.4

Energy (eV)

Energy (eV)

1.8

2.0

Energy (eV)

Figure 3-3. Calculated imaginary dielectric constant (optical absorption) as a function of energy, for (a) 5.3 nm, (b) 7.3 nm, and (c) 10.2 nm diameter wurtzite CdTe quantum wires. The solid line shows the average (disordered) polarization, the dotted and dashed lines show the contribution from the momentum matrix elements perpendicular and parallel to the wurtzite c-axis (i.e., the growth axis). The numbered arrows indicate the proposed correspondence to the experimental absorption envelopes depicted in Figure 3-4, and plotted in Figure 3-5. Calculated absorption data. Figure 3-3 shows the calculated imaginary dielectric constant (proportional to optical absorption) as a function of energy for 5.3 nm, 7.3 nm, and 10.2 nm diameter wurtzite CdTe quantum wires. The value of the imaginary part of the dielectric constant in the calculated energy range was comparable to the bulk measurements of Cardona.25 Excellent agreement between the experimental spectra and the corresponding calculated spectra with average polarization is evident in Figure 3-4, for the first four fitted absorption envelopes at each diameter. The higher resolution of

102

the calculated spectra resulted from two factors: (1) Only a finite number of k points along the wire axis were sampled, thus artificially limiting the broadening of the peak energies; and (2) the experimental spectra were further broadened by the diameter distributions of the samples. However, the clusters of features evident in the calculated spectra fell nicely within the broad Gaussian functions used to fit the experimental spectra. As the quantum-wire diameter increased, the resolution of the calculated peaks decreased, leading to broadened absorption features.

Energy-Energy of 1st Excited State (eV)

Energy-Energy of 1 st Excited State (eV)

Energy-Energy of 1st Excited State (eV)

Figure 3-4. Experimental absorption spectra (upper panel) and calculated absorption spectra with average polarization (lower panel) of CdTe quantum wires with diameters of (a) 5.3 nm, (b) 7.3 nm, and (c) 10.2 nm. Fitted spectra (barely visible orange curves) are shown behind the corresponding experimental spectra. Individual fitting functions (one exponential function for the background and multiple Gaussian functions for the absorption features) of each fitted spectrum are also plotted (the absorption features are color-coded).

103

To quantitatively compare the calculated and experimental results, the relative energies of the excited-state envelopes from experiment (open symbols) and theory (filled symbols) were plotted vs. the first excited-state energy (Figure 3-5). The resolved peaks in a relevant cluster of peaks in the calculated spectra were averaged for comparison to the experimental Gaussian-fitted envelopes. Peaks subjected to averaging are indicated by the designations in Figure 3-3 (e.g., 4a, 4b, 4c, etc.). The near overlap of the calculated and experimental curves in Figure 3-4 confirmed the excellent agreement. The theoretical and experimental curves were separated by only 35-90 meV, with the greatest separation for the highest-energy absorption envelope (the blue envelope labeled 4 in Figure 3-3). An assignment of the quantum-wire transitions contributing to these absorption envelopes was next undertaken.

1.65

1.70.

1.75

1.80

1.85

Energy of 1 st Excited State (eV)

Figure 3-5. Comparison of the experimentally observed absorption features (open symbols) and the calculated absorption features (solid symbols). The same color scheme is used as in Figure 3-3 and Figure 3-4.

104

CM CM COCO

" s

VT CD CD

__

99

99 \ I9

COlf> COCO

as
o 9 6?!
CM O) CO

I
CO n CD CD

CO CQ COCO

o mm V o to o !8 > coco
CD '

S"S9 ?>S8\
,

51 9
i> .

.>> CM ' V

.,* ;>'
-'OO
e>cd CD CO

5A
CM" O>

9.
CQ

9 i\ \
CD J \ TZ l 1

>1 1 7 1
Jl 1

'

1 *> 1 m

s 9x

A " " 71 *' 55 o > / \ >> >> m / '*- *' 1 \ A 5/ \

v

53
CO

>

I >

>/

'7
/I

O
i

|\
1

1 \

0.00

0.25

0.50

X\
0.75

JL

r 1.00

Energy-Energy of 1st Excited State (eV) Figure 3-6. Calculated optical absorption for transitions at the T-point (solid curve) for the 5.3 nm diameter wurtzite CdTe quantum wire. The vertical lines indicate the individual transitions, which are summed and broadened to produce the total curve. We have color-coded the transitions making the largest contributions, matching the colorcoding used in Figure 3-3, 3-4, and 3-5, and labeled the states involved in these transitions. The dotted curve shows the sum of the contributions from all 31 ^-points, for comparison. Assignment of absorption features. Using the T point as a reference, the strongest contributions to each of the peaks were identified for quantum wires of the three diameters theoretically investigated. The explicit case of the 5.3-nm diameter quantum wire is presented in more detail in Figures 3-6 and 3-7. For all three sizes of the quantum wires, we found that the first excitation into the CB1 state more substantially involved the VB2 state than the VB1 state. The second excited peak was found to be a contribution predominantly from the VB5 and VB6 states into the degenerate CB2 and CB3 states (as well as a few other smaller contributions that varied for the three sizes examined here). These first two sets of peaks are consistent with the results of previous empirical pseudopotential calculations of transitions in CdSe quantum wires.1 The third excited

105

peak always contained a contribution into the degenerate CB4 and CB5 states, and to a lesser extent into the CB6 state. Finally, the fourth excited peak was the sum of many small peaks, without any particularly dominant contribution for all sizes of the quantum wires.
(a)
2.7 2.6
_CB1415(ir e )

^
CB1213 ( 2 r y CB10.11 _CB9(10 e ) CB8<1* CB7 CB6 (2Ze '[' |

oo
CB14 CB15
CB8

(b)

2.52.4

ww

) '

CB4

a
CB5 CB2 VB2

CB9

CB12

CB13

CB6

2.32.22.1- CB2.3 2.01.9CB1 (1Ee) 1.8-~ / i i ~ "

7i~ "tt
CB4.5 ' " ;" "rf (1Ae)

n
CB1 CB3

VB1

VB3

VB4

VB5

VB6

VB7

VB8

"'mm: wi
VB9 VB10 VB11 VB12

1.7-

>
C D c LU

VB13

>. E>
VB17

mtm*
VB19 VB20

VB14

VB15

VB16

0.0-

-0.1-0.2

VB23

VB24

Figure 3-7. (a) The energy-level diagram for the 5.3 nm diameter wurtzite CdTe quantum wire, with the transitions contributing to the broad absorption features identified by color-coded arrows, (b) Plots of the electron-and hole-density distributions for the conduction band (CB) and valence band (VB) states. Boxes indicate pairs of energetically degenerate levels. States that do not make major contributions to the optical absorption in Figure 3-6 are omitted.

106

While we considered a large number of states in our calculations, we noted that in all of the quantum wires the largest values of the momentum matrix elements (and hence the strongest contributions to the absorption spectrum) resulted from transitions between the highest 25-30 VB and lowest 15 CB states for the energy range considered here. We also note that the peak position obtained using only the T-point imaginary dielectric function (ignoring the other ^-points completely) was within 60 meV of the peak position obtained by including all of the 31fc-pointsas discussed above, as is revealed by comparing the solid (F-point only) and dashed (all ^-points) absorption spectra in Figure 3-6.. Both of these factors suggest that future semiquantitative assignments could be made on the basis of substantially fewer calculations than in the present work. Quantum-wire electronic structure. The DFT-LDA-CPM electron and hole density distributions for the CB and VB states participating in the transitions of the 5.3-nm quantum wire are shown in Figure 3-7b. The sequence of the (electron) CB energy levels is strikingly consistent with that predicted by the simple PIC model. According to this model, the CB-state energies Ej are expressed by equation 3-1, where d is the quantumwire diameter, m*e is the electron effective mass, .Eg.buik is the bulk CdTe band gap (1.606 eV at 2 K),26 and #,,/ is the nth root (zero) of the Zth-order cylindrical Bessel function.27 The number of the zero, n, functions as a principal (radial) quantum number. The order, /, functions as an angular-momentum quantum number. For the spherical symmetry of quantum dots, the values / = 0, 1, 2, 3, 4, etc. are assigned the conventional symmetry labels S, P, D, F, G, etc., respectively. However, for the cylindrical symmetry of quantum wires, these labels are converted to their Greek counterparts: X, n, A, O, T, etc, respectively.

107

n,l ~ ~

* .2 "l"-Cg,bulk

(3-D

2# mca The predicted PIC sequence is thus YLe < lTle < lAg < 7Le < l<Dg < 2Tle < lTe (see Table 3-2), which corresponds to CB1 < (CB2,CB3) < (CB4,CB5) < CB6 < (CB8, CB9) < (CB12,CB33) < (CB14,CB15). In the PIC model, E states are singly and non-E states doubly degenerate,28 which faithfully reproduces the DFT-LDA-CPM calculations, except for CB8 and CB9 (10 e ), which are energetically close (9 meV) but nondegenerate. Note that the expected radial nodes are evident in the 2Ee (CB6) and 211^ (CB12, CB13) density distributions. In general, the electron-density distributions are nearly idealized representations of the expected PIC CB states.

Table 3-2. Zeros (roots) of the cylindrical Bessel functions.' Function order, / [Ji(x)] (State Label) 0 Vo(x)] () 2.4048 5.5201 8.6537 11.7915 1 Ui(x)]

Number of zero (root) n 1 2 3 4

(n)
3.8317 7.0156 10.1735 13.3237

2 [J2(x)] (A) 5.1356 8.4172 11.6198

3 [J3(.x)] (O) 6.3802 9.7610 13.0152

4 [74(x)]

(H
7.5883 11.0647 14.3725

16.2235 14.7960 17.6160 5 19.4094 14.9309 16.4706 17.9598 20.8269 "Arfken, G. B .; Weber, H. J . Mathematical Models for Ph ysicists, 4th ed.; Academic: San Diego, 1995; p 635.

Interruptions in the predicted PIC sequence are evident in the insertions of the CB7 and CB10, CB11 states, which do not belong. Examination of the electron-density distributions for CB7 and the doubly degenerate CB10, CB11 states establishes their IE and i n symmetry, respectively (Figure 3-8). Their energetic placement requires a

108

different electron effective mass than that for the continuous series described above, and indicates that CB7, CB10, and CB11 are intruder states that have folded into the T-point sequence from a separate valley of the band structure. In the 7.3-nm diameter quantum wire, these intruder YLe and i n e states are intercalated into the T-point sequence at different positions, at CB11 and CB14, CB15 respectively, demonstrating their different energy-diameter scaling and confirming the different electron effective mass at their corresponding k point. Projections of the Bloch function for these states show that they originate along the T-M line in the bulk wurtzite Brillouin zone, analogous to previously calculated results for zinc-blende quantum wires.29 Consequently, having a non-r k value, they are optically inactive (note the absence of transition arrows involving CB7 and CB10, CB11 in Figure 3-7a), and thus the intruder states do not contribute to the calculated absorption spectra.

Figure 3-8. Plots of the electron-density distributions for the CB7 state and the (CB10, CB11) degenerate pair of states for the 5.3-nm diameter quantum wire shown in Figure 37b. These have IX and i n symmetry, respectively (see the text).

In comparison to the CB-state electron-density distributions, the VB-state holedensity distributions (Figure 3-7b) are more complex, as a result of VB mixing of the heavy-hole and light-hole states (derived from the bulk A and B VBs).13'21 Consequently, the VB (hole) states generally have a hybridized character, rather than being pure S, fl, A,

109

etc. Significantly, the VB states are much more closely spaced than are the CB states, because of the greater hole effective masses. Therefore, the comparatively large spacing of the CB levels dominates the absorption spectra, such that the first absorption envelope constitutes a cluster of transitions terminating in CB1, the second in (CB2, CB3), the third in (CB4, CB5) and CB6, and the fourth in (CB8, CB9), (CB12, CB13), and (CB14, CB15). The absorption envelopes are thus defined by the spacing of the CB states. Energy scaling of CB states. The PIC model asserts that the CB energy levels should scale linearly with ^2, (equation 3-1). Recognition of the non-parabolic CB dispersion adds a quadratic term ($,) to the energy dependence at higher levels of theory.30 The CB-state energies for the 5.3, 7.3, and 10.2-nm diameter quantum wires are plotted vs. $j in Figure 3-9, along with both linear and quadratic fits (equations 3-2, 3-3, and 3-4) to the data. Again, a close correspondence is evident between the results of the DFT-LDA-CPM calculations and the simple PIC model. Gentle curvature is present in all three sets of theoretical data (Figure 3-9), which is well fit by the quadratic term (equations 3-2, 3-3, and 3-4). The curvature (and thus the quadratic term) decreases with increasing quantum-wire diameter, indicating that equation 3-1 becomes a better approximation at larger diameters. The curves should extrapolate to the bulk band gap of 1.606 eV (2 K), and all come close to this value: 1.60, 1.57, and 1.58 eV for the 5.3, 7.3, and 10.2-nm diameter quantum wires, respectively. 5.3-nm QW: E,i = (-1.63xl0"4 e V ) $ + (2.73xl0"2 e V ) $ + 1.60 eV 7.3-nmQW: ,, = (-9.28xl0"5 e V ) ^ + (1.82xl0"2 e V ) $ + 1.57 eV 10.2-nm QW: EnJ = (-4.30xl0"5 eV) $* + (1.08xl0"2 eV) $ + 1.58 eV (3-2) (3-3) (3-4)

110

Figure 3-9. Plots of the CB-state energy (En<i) of CdTe quantum wires of various diameters at the T point vs. $l. The two 10g states for all three diameters, and the two 2n<, and two \Te states for the 10.2-nm diameter QW, are not strictly degenerate (in the DFT-LDA-CPM results). Consequently, two data points are plotted for each of these pairs, one open point and one closed point, which overlap significantly. The solid and dashed lines are the quadratic and linear fits to the energy levels, respectively. Inset: Plot of electron effective mass vs. quantum-confinement energy. The linear fit yields a yintercept of (0.102 + 0.006) mo, which is close to the bulk value of 0.095 m0. Size-dependent electron effective mass. Quantum confinement increases the

effective masses of carriers, and prior studies have confirmed that m* increases as quantum-wire diameters decrease.5'6'9'30'31 The size dependence of m* in CdTe quantum wires can be extracted from the slopes of the Enj-vs.-$j plots in Figure 3-9 (see equation 3-1). As noted above, these plots exhibit slight curvature, and so we have calculated m* values using the coefficients of the 0^ terms in equations 3-2, 3-3, and 3-4 in place of the slopes from the linear fits. These values range from 0.20 mo for the 5.3-nm diameter quantum wire to 0.14 mo for the 10.2-nm diameter quantum wire. Interestingly, the inset

111

of Figure 3-9 shows that the me values scale linearly with the quantum-wire confinement energy (Ag) and extrapolate to me = (0.102 0.006) mo at AES = 0, which is close to the bulk value of 0.095 m0 (at ~2 K).32 lPe-lSe spacing. As established above, the absorption features we observed

experimentally resulted from clusters of closely spaced transitions rather than from individual transitions. In contrast, prior studies of the optical spectra of CdSe13, CdTe14, InAs,15 and InP16 quantum dots successfully resolved the individual transitions. We suggested above that the CdTe quantum-wire absorption features were broadened by the size distributions, which are wider than those achieved for quantum dots. Additionally, the prior quantum-dot studies used photoluminescence excitation spectroscopy, which provides inherently higher spectral resolution than does the standard absorption spectroscopy used here. However, we now consider if the energy-level spacings in CdTe quantum wires are considerably smaller than those in the corresponding quantum dots, which would also result in lower spectral resolution. Table 3-3. Zeros (roots) of the spherical Bessel functions.0 Function order, / [//(*)] (State Label) 0 [7o(x)] (5) 3.141593 1 Ui(x)] (P) 4.493409 2 [J2(x)} (D) 5.763459 3 [73(*)] (F) 6.987932
4 [74(JC)]

Number of zero (root) n 1

(G) 8.182561

2 6.283185 7.725252 9.095011 10.417119 11.704907 3 9.424778 10.904122 12.322941 13.698023 15.039665 4 12.566370 14.066194 15.514603 16.923621 18.301256 5 15.707963 17.220755 18.689036 20.121806 21.525418 a Antosiewicz, H. A. In Handbook of Mathematical Functions; Abramowitz, M., Stegun, I. A., Eds.; National Bureau of Standards Applied Mathematics Series 55; U.S. Government Printing Office: Washington D.C., 1964; p 467. 112

0.5 O CdTe QDs CdTeQWs

? - H
w 0.3 A <o ^
1

(1P -1S)
e

= 0.270 eV

e ' Q D

0.2-1

( i n . - l l . ) ^ - 0 . 2 3 4 eV

0.1
I | I I I T ^ " T " T , " I " 1 " | "1 I I I |

i i ' i i i ' i '

6 7

10

11

d(nm)

Figure 3-10. Plot of (lPe - lSe), (in g - lEg) vs. diameter of CdTe quantum dots and quantum wires. The black curve is a fit to the DFT-LDA-CPM data for CdTe quantum dots,23 yielding a (lPe - lSe)QD value of 0.270 eV at a diameter of 5.3 nm. The (1IL. lXe)Qw value of 0.234 was obtained directly from the DFT-LDA-CPM calculations performed for this study. The DFT-LDA-CPM (1IL, - 12e)QW/(l/>e - 1>)QD ratio is thus calculated to be 0.87, consistent with 0.86 predicted by the PIC model. To investigate the comparative level spacings in CdTe quantum wires and dots, we use the difference in the YLe (lSe) and in<. (lPe) energies as representative. The simple PIC model above predicts a ratio (in e - lEe)Qw/(l^e - l>Se)Qp of 0.86, independent of diameter (assuming equal quantum-wire and quantum-dot diameters and me values; see equation 3-5, where i,i and i,0 are the 1st zeros (roots) of the 1st- and Otii-order spherical Bessel functions, respectively. The zeros (roots) of the spherical Bessel

functions can be found in Table 3-3). Correspondingly, the DFT-LDA-CPM calculations for CdTe quantum wires reported here and for CdTe quantum dots reported previously23 produce a ratio (Ul e - \Le)owl(lPe 1S)QD

of 0.87 at quantum-dot and quantum-wire

diameters of 5.3 nm (Figure 3-10). Therefore, the CB levels in CdTe quantum wires and quantum dots are similarly, although not equally, spaced. The result suggests that the 113

lower resolution of the current experimental results is due to the broader diameter distributions and the spectroscopic method employed.

(liyiS.)^ (lP.-lS.)^

frrXrf2 2x2med2 fi-fa ^3.8317 2 -2.4048 2 2 tftl h q0 "ff,-C"4.49342-3.14162" 2K2m\d2 2n2med2

'

K}

'

Conclusion The close agreement between the calculated and experimental absorption spectra for CdTe quantum wires corroborates the electronic structure depicted in Figure 3-7. The absorption features observed are largely determined by the energy spacing of the CB states. The sequence, character, and spacing of the CB states are remarkably consistent with the PIC model, which is the simplest model for 2D confinement in quantum wires. The VB states are complicated by VB mixing, and thus do not submit to this simple analysis. Prior theoretical studies have shown that the CB structure in quantum dots33'34 and quantum rods35 is analogously simple and well described by particle-in-a-sphere or PIC models, respectively. Here we demonstrate such a correlation explicitly for quantum wires, and present it in greater detail than the previous studies. The close correspondence of the DFT-LDA-CPM results and the PIC model confirms the latter as a useful firstorder picture for the CB structure in semiconductor quantum wires.

References (1) Li, J.; Wang, L.-W. Nano Lett. 2003, 3, 101-105.

114

(2) (3)

Xia, J. B.; Zhang, X. W. Eur. Phys. J. B 2006, 49, 415-420. Huang, S.-P.; Cheng, W.-D.; Wu, D.-S.; Hu, J.-M.; Shen, J.; Xie, Z.; Zhang, H.; Gong, Y.-J. Appl. Phys. Lett. 2007, 90, 031904. Redlinski, P.; Peeters, F. M. Phys. Rev. B 2008, 77, 075329. Persson, M. P.; Xu, H. Q. Nano Lett. 2004, 4, 2409-2414. Persson, M. P.; Xu, H. Q. Appl. Phys. Lett. 2002, 81,1309-1311. Persson, M. P.; Xu, H. Q. Phys. Rev. B 2006, 73, 125346. Persson, M. P.; Xu, H. Q. Phys. Rev. B 2004, 70, 161310. Lassen, B.; Willatzen, M.; Melnik, R.; Voon, L. J. Mater. Res. 2006, 21, 29272935. Niquet, Y. M.; Lherbier, A.; Quang, N. H.; Fernandez-Serra, M. V.; Blase, X.; Delerue, C. Phys. Rev. B 2006, 73, 165319. Zhang, X. W.; Xia, J. B. J. Phys.: Condens. Matter 2006, 3107-3115. Zhao, M.; Xia, Y.; Liu, X.; Tan, Z.; Huang, B.; Song, C ; Mei, L. J. Phys. Chem. B 2006, 110, 8764-8768. Norris, D. J.; Bawendi, M. G. Phys. Rev. B 1996,53,16338-16346. Masumoto, Y.; Sonobe, K. Phys. Rev. B 1997,56, 9734-9737. Banin, U.; Lee, C. J.; Guzelian, A. A.; Kadavanich, A. V.; Alivisatos, A. P.; Jaskolski, W.; Bryant, G. W.; Al, L. E.; Rosen, M. /. Chem. Phys. 1998, 109, 2306-2309. Bertram, D.; Micic, O. I.; Nozik, A. J. Phys. Rev. B 1998, 57, R4265-R4268. Nikesh, V. V.; Lad, A. D.; Kimura, S.; Nozaki, S.; Mahamuni, S. /. Appl. Phys. 2006,700,113520. Peterson, J. J.; Huang, L.; Delerue, C ; Allan, G.; Krauss, T. D. Nano Lett. 2007, 7, 3827-3831. Hu, J.; Li, L.-S.; Yang, W.; Manna, L.; Wang, L.-W.; Alivisatos, A. P. Science 2001, 292, 2060-2063.

(4) (5) (6) (7) (8) (9)

(10)

(11) (12)

(13) (14) (15)

(16) (17)

(18) (19)

115

Hu, J.; Wang, L.-W.; Li, L.-S.; Yang, W.; Alivisatos, A. P. J. Phys. Chem. B 2002, 106, 2447-2452. Zhao, Q.; Graf, P. A.; Jones, W. B.; Franceschetti, A.; Li, J.; Wang, L.-W.; Kim, K. Nano Lett. 2007, 7, 3274-3280. Wang, R; Dong, A.; Sun, J.; Tang, R.; Yu, H.; Buhro, W. E. Inorg. Chem. 2006, 45,7511-7521. Li, J.; Wang, L.-W. Phys. Rev. B 2005, 72, 125325. Schrier, J.; Demchenko, D. O.; Wang, L. W.; Alivisatos, A. P. Nano Lett. 2007, 7, 2377-2382. Cardona, M. J. Appl. Phys. 1965,36, 2181-2186. Madelung, O. Semiconductors: Basic Data; 2nd ed.; Springer: Berlin, 1996. Yoffe, A. D. Adv. Phys. 2002, 51, 799-890. Zhao, X.; Wei, C. M.; Yang, L.; Chou, M. Y. Phys. Rev. Lett. 2004, 92, 236805. Wang, L.-W.; Li, J. Phys. Rev. B 2004, 69,153302. Chen, R.; Bajaj, K. K. Phys. Rev. B 1994, 50, 1949-1952. Karanth, D.; Fu, H. Phys. Rev. B 2006, 74,155312. Madelung, O. Semiconductors: Data Handbook; 3rd ed.; Springer: Berlin, 2004. Chestnoy, N.; Hull, R.; Bras, L. E. J. Chem. Phys. 1986, 85, 2237-2242. Efros, A. L.; Rosen, M. Annu. Rev. Mater. Sci. 2000, 30, 475-521. Shabaev, A.; Efros, A. L. Nano Lett. 2004, 4, 1821-1825.

116

Chapter 4 SLS Growth of Diameter-Controlled PbSe Nanowires

117

Introduction The direct narrow band-gap of 0.278 eV (4.46 jum) at 300 K at the L point of the Brillouin zone1 and the large Bohr radius of 46 nm2 make PbSe an attractive material for fundamental quantum-confinement studies3 and potential infrared-based applications in photovoltaic devices,4"6 biological imaging,7 telecommunications,8 field-effect transistors (FET),9'10 and lasers.11 Recently, high-efficiency carrier multiplication in colloidal PbSe and PbS nanocrystals was discovered by different research groups.12"15 Specifically, up to seven excitons (quantum yield up to -700%) were experimentally generated per photon in PbSe nanocrystals.13 This finding may dramatically enhance the efficiency of photovoltaic energy-generation devices. Although the ultrafast nonradiative Auger decay of multiexcitons (leas than 150 ps) prevents the carriers from being harvested practically so far, this obstacle may be overcome by using elongated nanostructures, such as quantum rods and quantum wires.16 Additionally, it was demonstrated for the first time that both electrons and holes could be injected from an electrode into the quantumconfined states of PbSe quantum dot films, which may find novel applications.17 The first essential step toward successful applications of PbSe nanowires is to synthesize high-quality samples in a controllable and tunable way. Generally, it's more challenging to synthesize one-dimensional (ID) nanowires than zero-dimensional (0D) nanocrystals. Since the first report of the synthesis of high-quality colloidal PbSe

nanocrystals by Murray and coworkers in 2001, 18 significant advancement has been made to control the size and the shape of PbSe nanostructures through solution-based synthesis.19"28 However, only limited achievement has been made to control the

dimensionality. Several strategies have been developed to synthesize PbSe nanowires,

118

including the solvothermal approach,

the template-based method,

the oriented-

attachment method,32 the chemical-vapor-deposition (CVD) approach33"35 based on the vapor-liquid-solid (VLS) mechanism, and the solution-liquid-solid (SLS) wire growth approach.36 Among these strategies, the solvothermal approach, the template-based method, and the CVD approach usually generate PbSe nanowires with low quality, which may prevent practical applications. Although it was demonstrated that the orientatedattachment method could produce single-crystalline PbSe nanowires, the batch crystallinity is still questionable, for example, some segments of the wire may be composed of small domains of nanocrystals. However, this crystallinity issue does not exist for the wires grown by the SLS approach due to the continuous wire growth from metal-catalyst nanoparticles. Recently, Kuno and coworkers reported the SLS growth of PbSe nanowires from binary Au/Bi core/shell catalyst nanoparticles in mildly coordinating solvent; however, they were not able to finely tune the wire diameters in a systematic way.36 In this Chapter, we report the SLS growth of high-quality, diameter-controlled, single-crystalline PbSe nanowires from nearly monodisperse Bi catalyst nanoparticles in the strongly coordinating solvent tri-n-octylphosphine oxide (TOPO). We systematically optimized the synthetic conditions to achieve diameter control in the range of 6-18 nm. The well-resolved excitonic feature in the near-infrared absorption spectrum of smalldiameter wires demonstrated strong quantum confinement.

119

Results Synthesis. Contributions from various researchers to the synthesis of high-quality PbSe nanostructures helped us to find appropriate precursors and experimental conditions for the SLS growth of diameter-controlled PbSe nanowires. Our designed synthetic route is shown in Figure 4-1.

PbO + Acid*

Solvent

Lead carboxylate or alkylphosphonate

Biseeds Solvent 160-270 C PbSe nanowire

* Acid = Carboxylic acid or Alkylphosphonic acid

R3P=Se Precursors
(R = n-C 4 H 9 (rn-C s H n )

Figure 4-1. Schematic illustration of the designed synthetic route for the solution-liquidsolid (SLS) growth of PbSe nanowires. Both lead(II) acetate trihydrate (Pb(02CCH3)2-3H20) and lead(II) oxide (PbO) are common lead sources.18'19'22'25'32 When a carboxylic acid is used, the actual lead

carboxylate precursor is formed according to the following reactions: Pb(02CCH3)2-3H20 + 2 RCOOH - Pb(02CR)2 + 2 CH3COOH + 3 H20 PbO + 2 RCOOH - Pb(02CR)2 + H 2 0 (3-1) (3-2)

By comparing these two reactions, we can clearly see that acetic acid and more water will be released to the reaction mixture when lead(II) acetate trihydrate is used as the lead source. Therefore extra effort is usually necessary to dry the reaction mixture and remove acetic acid at elevated temperatures because even trace amount of acetate can influence the product morphologies.37 In this regard, we chose PbO as the lead source for convenience. The actual precursor for lead is lead carboxylate or alkylphosphonate generated in situ by the reaction between PbO and long-chain carboxylic acid or alkylphosphonic acid. The selenium precursors in our synthesis include tri-n-

120

octylphosphine selenide (TOPSe) and tri-n-butylphosphine selenide (TBPSe), which are common selenium precursors for the synthesis of a variety of selenium-containing nanostructures.18'38'39 There is a eutectic point of 125.5 C in the bulk Bi-Pb binary phase diagram (Figure 4-2), indicating that Bi may catalyze the SLS growth of PbSe nanowires at relatively low reaction temperatures. Experimentally, it has been demonstrated already by Kuno and coworkers that Au/Bi core/shell nanoparticles could be used as catalysts for the SLS growth of PbSe nanowires at reaction temperatures less than 200 C,36 which encouraged us to choose pure Bi nanoparticles as catalysts for the SLS growth of diameter-controlled PbSe nanowires.
Weight Percent Bismuth

30

40

SO

80

70

Atomic Percent Bismuth

Figure 4-2. Bulk Bi-Pb binary phase diagram. The eutectic point is 125.5 C at -45 atomic percent lead. Because the use of weakly coordinating and noncoordinating solvents are predominate in the literature for the synthesis of PbSe nanostructures, including nanowires, our initial attempts to grow PbSe nanowires were also conducted in these

121

types of solvents. Both phenyl ether (PE), which is a weakly coordinating solvent, and poly-1-decene (PDE), which is a noncoordinating solvent, were employed in our synthesis. Unfortunately, it was found very difficult to produce well-controlled wires. When the synthesis was conducted in PDE, short rods and a minority of dots were produced (Figure 4-3A). No catalyst nanoparticles were found at either ends of the rods, indicating the rods were not formed by the SLS mechanism. Although PbSe nanowires along with dots were observed in the TEM images (Figure 4-3B, C) when the synthesis was conducted in PE, we were only able to achieve poor control over the length and the diameter of the wires. In addition, we observed secondary nucleation and growth of nanocrystals along the wire surface, leading to apparently rough wires, which can be clearly seen in the magnified TEM images (Figure 4-3D). These results suggested very high precursor reactivities, especially for the synthesis in noncoordinating solvent. Importantly, for some wires with smooth surfaces, catalyst nanoparticles were observed at one end of the wires (Figure 4-3B, C), confirming that the wires were grown through the SLS mechanism. It should be pointed out that our findings are inconsistent with the results reported by Kuno and coworkers.36 In their synthesis, weakly coordinating solvents such as PE could be successfully used to grow good-quality PbSe nanowires. Although efforts were made to synthesize PbSe nanowires using reaction conditions similar to those used by Kuno, we could not produce well-controlled wires. This inconsistency will be discussed later.

122

Figure 4-3. Representative TEM images of PbSe nanostructures synthesized in poly-1decene (A) and phenyl ether (B, C, and D). The red dashed circles in the images highlight the catalyst nanoparticles attached to the end of wires, confirming the SLS wire growth.

The failure to produce well-controlled PbSe nanowires in noncoordinating and weakly coordinating solvents naturally guided us to use a strongly coordinating solvent, although it was previously reported that strongly coordinating solvents such as TOPO appeared to prevent the growth of PbSe nanostructures.36 Fortunately, high-quality PbSe nanowires of various diameters in the range of 6-18 nm were successfully grown from Bi catalyst nanoparticles in TOPO. Representative low- and high-magnification TEM images (Figure 4-4 and Figure 4-5) show that the wires are long, straight, smooth, and with uniform diameters along their lengths. The standard deviation in the diameter distributions was less than 20% of the mean diameter (Figure 4-6).

123

Figure 4-4. Representative low-magnification TEM images of high-quality PbSe nanowires of various diameters (d). (a) d = 6.8 1.0 nm ( 15%), (b) d = 8.8 1.6 nm (+ 18%), (c) d = 10.0 1.5 nm ( 15%), (d) d = 11.5 1.9 nm ( 17%), (e) d = 12.2 + 2.1 nm ( 17%), (f) d = 13.1 1.6 nm ( 12%), (g) d = 14.2 2.1 nm ( 15%), (h) d = 16.6 2.1 nm ( 13%), and (i) d = 17.7 + 2.8 nm ( 16%). These nanowires were synthesized under various synthetic conditions described in Table 4-1.

124

Figure 4-5. Representative high-magnification TEM images of high-quality PbSe nanowires of various diameters (d). (a) d = 6.8 1.0 nm ( 15%), (b) d = 8.8 1.6 nm ( 18%), (c) d = 10.0 1.5 nm ( 15%), (d) d=U.5 1.9 nm ( 17%), (e) d = 12.2 2.1 nm ( 17%), (f)d= 13.1 1.6 nm ( 12%), (g) d = 14.2 2.1 nm ( 15%), (h) d = 16.6 2.1 nm ( 13%), and (i) d = 17.7 2.8 nm ( 16%). These nanowires were synthesized under various synthetic conditions described in Table 4-1.

125

6 7 8 9 10 11 PbSe nanowire diameter (nm)

10

11

12

13

PbSe nanowire diameter (nm)

9 10 11 12 13 14 15 16 17

9 10 11 12 13 14 15 16 17 18 19 20 PbSe nanowire diameter (nm)

PbSe nanowire diameter (nm)

o O

14

16

18

20

22

24

10

12

14

16

18

20

22

24

PbSe nanowire diameter (nm)

PbSe nanowire diameter (nm)

Figure 4-6. Representative diameter-distribution histograms for PbSe nanowires of various diameters (d). (A) d = 6.8 1.0 nm ( 15%), (B) d = 8.8 1.6 nm ( 18%), (C) d = 10.8 1.6 nm ( 15%), (D) d = 13.1 1.6 nm ( 12%), (E) d = 15.8 2.3 nm ( 15%), and (F) d = 17.8 2.8 nm ( 16%).

126

The length of the PbSe nanowires was typically between one-half to several micrometers. When the same amount of precursors and Bi nanoparticles were used, the achievable length of PbSe wires was found to be shorter than that of CdTe nanowires reported in Chapter 1, as will be discussed later. To produce very long PbSe wires, the amount of Bi stock solution used for the synthesis of PbSe nanowires was typically less than that for CdTe nanowires (Table 4-1). The long straight wires tended to form bundles (Figure 4-7). However, it was found that the bundling tendencies depended on the wire length. The longer the wires are, the higher the bundling tendencies are, which should be due to the stronger total van der Waals forces between longer wires.

Figure 4-7. Low-magnification TEM images of PbSe nanowire bundles. (A) 11.5-nm wires with the length of several micrometers, and (B) 9.1-nm wires with the length of about one micrometer. These images indicate that longer wires have a higher tendency to form bundles. The diameter control of PbSe nanowires was not straightforward. Compared with CdTe nanowires reported in Chapter 1, it was more challenging to produce high-quality PbSe nanowires in a wide diameter range. It turned out that several synthetic parameters including solvent sources, precursors, diameters of the Bi catalyst nanoparticles, reaction temperatures, Se/Pb ratios, and precursor concentrations had significant and somewhat 127

complicated effects on the control of wire growth, which is the primary origin of the extensive empirical experimentation required to optimize the experimental conditions for PbSe nanowires.

Figure 4-8. TEM images of PbSe nanostructures synthesized using 99% TOPO with different batch numbers purchased from Aldrich. (A) Batch number 00529CD, and (B) batch number 03008CE.

Figure 4-9. TEM images of PbSe nanostructures synthesized from lead stearate and TOPSe at 244 C in TOPO. Bi nanoparticles with the mean diameter of 4.8 nm were used as catalysts. (A) Se/Pb = 0.5, (B) Se/Pb = 1.0, and (C) Se/Pb = 2.0. It should be pointed out that the successful synthesis of PbSe nanowires shown above (Figure 4-4 and Figure 4-5) and all the optimization of synthetic conditions were conducted using vacuum-distilled 90% technical grade TOPO purchased from Aldrich. However, our early syntheses employed 99% TOPO with the batch number 00529CD 128

purchased from the same company. This 99% TOPO also worked well (Figure 4-8A). Unfortunately, we were not able to reproduce the syntheses after the 99% TOPO was changed to another batch number 03008CE, although it was manufactured by the same company (Figure 4-8B). Thus, the researchers who want to repeat our results should take great care of the TOPO source.

Figure 4-10. TEM images of PbSe nanowires synthesized using Bi nanoparticles of various diameters (dm) at 240-250 C in TOPO. Lead oleate and TOPSe were used as precursors. The Se/Pb ratio was held at 1.2. (A) dm = 4.3 0.8 nm, dPbse = 14.2 2.1 nm, (B) dm = 7.1 0.9 nm, dme = 12.3 2.5 nm, (C) dBi = 8.9 0.8 nm, dPbSe = 13.1 1.6 nm, and (D) dm = 13.5 1.0 nm, dpbSe = 16.3 3.5 nm.

By borrowing experiences from the amazing success of the SLS growth of CdTe nanowires, it didn't take us very long to determine the optimal Se/Pb ratio when lead carboxylate and TOPSe precursors were employed. We varied the Se/Pb ratio from 0.5

129

to 2.0 when 4.8-nm Bi nanoparticles were used as catalysts at a reaction temperature of 244 C (Figure 4-9). When the Se/Pb ratio was 0.5, both wires and significant amount of dots were produced. However, when equal amount of Pb and Se precursors were employed, SLS growth of high-quality PbSe nanowires was achieved. The Se/Pb ratio could not be further increased to 2.0 because the resultant wires were observed to be short and thick. These results suggested that equal or a slight excess amount of Se precursor was desired to produce high-quality PbSe nanowires when lead carboxylate and TOPSe were used as precursors at reaction temperatures around 244 C. Efforts were then made to grow PbSe nanowires with various diameters from different Bi nanoparticles at the Se/Pb ratio of 1.2 and the reaction temperatures in the range of 240-250 C (Figure 4-10). Unlike the CdTe cases, it was found that the diameter control of the PbSe nanowires was very limited by varying the mean diameters of Bi nanoparticles in the range of 4-14 nm at similar temperatures. Unexpectedly, 14.2nm PbSe nanowires were produced when 4.3-nm Bi nanoparticles were employed. However, the standard deviation in the diameter distributions was measured to be 15% of the mean diameter, indicating that the wires were fairly uniform. When 7.1-nm and 8.9nm Bi nanoparticles were employed, 12.3-nm and 13.1-nm PbSe nanowires were produced, respectively. The fact that the mean diameter of the resultant uniform wires is larger than that of the initial Bi nanoparticles indicated swelling instead of agglomeration of the Bi nanoparticles because agglomeration usually results in broad diameter distributions. However, for larger Bi nanoparticles, such as 13.5 nm, we not only noticed swelling of the catalyst nanoparticles but also observed some wires with rough surfaces, which should be formed through a secondary nucleation and growth of nanocrystals on

130

the wire surface. These results raised two synthetic questions: (1) how to grow smallerdiameter PbSe nanowires, and (2) how to grow larger-diameter PbSe nanowires in a controllable way. The swelling degree of a given Bi nanoparticle should be determined by the solubility of PbSe in the liquid Bi-semiconductor-alloy nanodroplets. Generally, the higher the reaction temperature, the higher the solubility. Therefore, we expected that the

significant catalyst-nanoparticle swelling observed when small Bi nanoparticles were employed could be suppressed by lowering the reaction temperature, resulting in controlled growth of small-diameter PbSe nanowires.

Figure 4-11. TEM images of PbSe nanowires synthesized from lead oleate, TOPSe, and 4.3-nm Bi nanoparticles at different reaction temperatures (7) in TOPO. The Se/Pb ratio was held in the range of 1.6-2. (A) T = 201 C, d?hSe = 11.4 1.8 nm, (B) T = 190 C, <4bSe = 10.5 2.2 nm, and (C) T= 182 C, broad wire-diameter distribution.

Figure 4-11 shows the TEM images of PbSe nanowires synthesized from lead oleate, TOPSe, and 4.3-nm Bi nanoparticles at different reaction temperatures equal to or less than -200 C, which were achieved by using a ternary salt bath of NaNOs-KNCV Ca(N03>2 instead of the conventional binary salt bath of NaN03-KN03 with a low eutectic point at -240 C. Other common heating baths such as silicon oil also work

131

under 200 C; however, the salt bath provides extra convenience such as ease of clean-up. Compared with the 14.2-nm PbSe nanowires produced at 240 C using the same Bi nanoparticle stock solution, smaller wires were indeed grown at relatively low temperatures. For example, 11.4-nm and 10.5-nm PbSe nanowires were prepared at 201 C and 190 C, respectively. However, the wire diameter could not be further reduced at 180 C because the diameter distribution of the resultant wires was found to be very broad, indicating agglomeration of small Bi nanoparticles, which may be due to the low precursor reactivities at low temperatures.

Figure 4-12. TEM images of PbSe nanowires synthesized from 4.3-nm Bi nanoparticles at -180 C in TOPO. Lead oleate and TBPSe were used as precursors. (A) Se/Pb = 1.2, (B) Se/Pb = 5.6.

132

To conduct reactions at temperatures below 180 C, we first increased the reactivity of Se precursor by using TBPSe with short carbon chains instead of TOPSe. However, only short PbSe nanowires in low yield were produced when the previously optimized Se/Pb = 1.2 was used (Figure 4-12A). Importantly, we didn't observe agglomeration of small Bi nanoparticles, indicated by the uniform PbSe nanowires in the TEM images. When the Se/Pb ratio was further increased up to 5.6, long wires (> 1 jum) were produced (Figure 4-12B). These results suggested that large excess of Se precursor was necessary to grow high-quality long PbSe nanowires from lead carboxylate and TBPSe at temperatures below 180 C.

Figure 4-13. TEM images of PbSe nanowires synthesized from lead oleate and TBPSe at low reaction temperatures (T). The Se/Pb ratio was held at 5.6. Bi nanoparticles with the mean diameter of 4.3 nm were used as catalysts. (A) T = 180 C, dpbse =10.8 + 1.6 nm, (B) T = 170 C, JPbSe = 8.8 1.6 nm, and (C) T = 160 C, low wire-yield. The black arrows point to the catalyst nanoparticles, from which only very short rods were grown.

133

Under the newly optimized Se/Pb ratio, efforts were made to grow smaller-diameter PbSe nanowires at temperatures equal to or less than 180 C (Figure 4-13). Indeed, 10.8nm and 8.8-nm PbSe nanowires were produced at 180 C and 170 C, respectively. However, the reaction temperature could not be further lowered to 160 C because many short wires were produced in low yield, and some catalyst nanoparticles only initiated very limited growth, which could be due to the low reactivities of both precursors and Bi catalyst nanoparticles at temperatures as low as 160 C.

Figure 4-14. TEM images of PbSe nanowires synthesized from 4.3-nm Bi nanoparticles at 160 C in TOPO. Lead oleate and TBPSe were used as precursors. (A) Se/Pb = 12, dpbse = 8.8 1.4 nm, and (B) Se/Pb = 20, short wires with rough surfaces. As noted above, a higher Se/Pb ratio helped to produce longer wires in higher yield at low reaction temperatures around 180 C. Therefore, we employed very high Se/Pb ratio

134

to grow long PbSe nanowires at 160 C (Figure 4-14). Although long wires were produced when the Se/Pb ratio was increased to 12, it was noticed that the surface of some wires was slightly corrugated. Additionally, we were not able to grow wires with the mean diameter smaller than 8.8 nm under these synthetic conditions. When the Se/Pb ratio was further increased to 20, the produced wires were short, with rough surfaces, indicating that secondary nucleation and grow of nanocrystals on the wire surfaces appeared to be significant when very high Se/Pb ratio was employed.

Figure 4-15. TEM images of PbSe nanowires synthesized from 4.3-nm Bi nanoparticles, lead octanoate, and TBPSe at different reaction temperatures (7) in TOPO. The Se/Pb ratio was held at 6. (A) T = 170 C, dpbSe = 6.8 + 1.0 nm, and (B) T = 162 C, short and tapered wires, which can be clearly seen in a representative high-magnification TEM image (C). (D) Plot of local wire diameter of the single wire shown in (C) vs. the corresponding distance from the catalyst tip (d, defined by the double-headed arrow in C), indicating the gradual swelling of the catalyst nanoparticle. The dotted line is the linear least-squares fit to the data.

135

We then increased the reactivity of the Pb precursor by using lead octanoate with a shorter carbon chain than lead oleate. PbSe nanowires with the mean diameter of 6.8 nm were successfully produced from 4.3-nm Bi nanoparticles at 170 C (Figure 4-15A). However, only short wires (< 500 nm) were produced when the reaction was conducted at -160 C (Figure 4-15B). A representative high-magnification TEM image revealed that the wires were tapered (Figure 4-15C). The plot of local diameters of a single PbSe wire vs. the corresponding distances from the catalyst tip (Figure 4-15D) clearly showed the gradual swelling of the catalyst nanoparticle. At the initial stage of wire growth, the local diameter of the wire was less than 6 nm, which gradually increased along with the wire growth. This may prevent us from growing PbSe nanowires having diameters less than 6 nm.

Figure 4-16. TEM images of PbSe nanostructures synthesized from 6.4-nm Bi nanoparticles at 170 C in TOPO. Lead oleate and TBPSe was used as precursors. The Se/Pb ratio was 5.6. It should be pointed out that the SLS growth of PbSe nanowires at low reaction temperatures requires the use of small Bi nanoparticles only. If larger Bi nanoparticles

136

were employed, such as 6.4 nm, both wires and nanoparticles were produced (Figure 416), which should be due to the low catalytic activities of larger Bi nanoparticles. As mentioned above, when the PbSe nanowires were grown from lead carboxylate and TOPSe at -250 C, we encountered a problem for the 13.5-nm Bi nanoparticles due to its low catalytic activity. To increase the activity of large Bi nanoparticles, the reaction has to be conducted at higher temperatures. However, this will raise another problem because of the thermal instability of the lead carboxylate precursor, which has a high tendency to decompose at temperatures higher than 250C.

To overcome this experimental difficulty, we employed lead alkylphosphonate to produce larger-diameter PbSe nanowires. To our knowledge, the direct use of lead alkylphosphonate has not been reported in the literature for the synthesis of PbSe nanostructures. Thus, we first investigated the thermal stability of this type of lead precursor. It was found that a clear TOPO solution of lead -tetradecylphosphonate [Pb(TDPA)] could be formed by the reaction between PbO and n-tetradecylphosphonic

137

acid only at temperatures higher than -250 C, indicating that the thermal stability of lead alkylphosphonate is higher than that of lead carboxylate. We first investigated the influence of the Se/Pb ratio on the PbSe nanowires grown from Pb(TDPA) precursor at -270 C. TEM images (Figure 4-17) showed that higher Se/Pb ratio was preferred to produce longer wires. Thus, we typically used a high Se/Pb ratio (> 6) for our syntheses.

Figure 4-18. TEM images of PbSe nanowires synthesized from large Bi nanoparticles of various diameters (JBi) at temperatures in the range of 270-275 C in TOPO. Pb(TDPA) and TOPSe were used as precursors. The Se/Pb ratio was held in the range of 6-8. (A) dm = 7.1 0.9 nm, broad wire-diameter distribution, (B) dBi = 12.4 1.2 nm, dPbse = 15.8 2.3 nm, (C) dm = 13.5 1.0 nm, dPbSe = 17.7 2.8 nm, and (D) dm = 15.8 + 1.0 nm, both wires and cubes were produced. Large Bi nanoparticles of various diameters were employed to control the diameters of the PbSe nanowires at temperatures in the range of 270-275 C (Figure 4-18). When

138

12.4-nm and 13.5-nm Bi nanoparticles were used, 15.8-nm and 17.7-nm PbSe wires were produced, respectively. However, the wires grown from 7.1-nm Bi nanoparticles showed a broad diameter distribution, indicating agglomeration of Bi nanoparticles. When 15.5nm Bi nanoparticles were used, both wires and cubes were produced, which should be mainly due to the low catalytic activity of the very large Bi nanoparticles at the temperatures investigated. Further increase of reaction temperature did not help to produce larger-diameter wires because of the thermal instability of Pb(TDPA). These results suggested that only relatively large Bi nanoparticles could serve as good catalysts to produce high-quality PbSe nanowires from Pb(TDPA) at -270 C. As shown in Figure 4-17, although higher Se/Pb ratio helped to produce longer wires, the wire length was still not very long. We made efforts to deal with this problem with two strategies.

Figure 4-19. TEM images of PbSe nanowires synthesized at high precursor concentrations at 272 C in TOPO. Pb(TDPA) (0.10 mmol) and TOPSe were used as precursors. The Se/Pb ratio was 8. Bi nanoparticles with the mean diameter of 13.5 nm were used as catalysts. First, we used high precursor concentrations and kept all other synthetic parameters constant. In our typical syntheses of PbSe nanowires, 0.05 mmol of Pb precursor was 139

used. We then doubled the amount of Pb precursor, however, the produced wires showed rough surfaces (Figure 4-19), indicating secondary nucleation and growth of PbSe nanocrystals on the wire surfaces, which should be due to the high precursor reactivities. Experimentally, we observed precipitate during the wire growth.

Figure 4-20. TEM images of PbSe nanowires synthesized using various amount of 13.5nm Bi stock solutions (m) at 275 C in TOPO. Pb(TDPA) (0.05 mmol) and TOPSe were used as precursors. The Se/Pb ratio was ~6. (A) m = 25 mg, and (B) m = 16 mg. Second, we varied the amount of Bi stock solution used for the SLS growth of PbSe nanowires. TEM images (Figure 4-20) showed that the wire lengths could be increased from ~1 jum to several micrometers when the amount of Bi stock solution was decreased from 25 mg to 16 mg. The results suggested that the amount of Bi nanoparticles was an effective factor to control the wire lengths. From the above results we can see that optimizations of a broad range of synthetic conditions were investigated to achieve diameter and length control for the SLS growth of PbSe nanowires. We were also interested in the experimental results in the absence of Bi catalyst nanoparticles under various optimized synthetic conditions (Figure 4-21). When the reaction was conducted at temperature as low as 170 C, only large particles

140

with diameters up to 100 nm were produced from lead oleate and TBPSe precursors. The high-magnification TEM images showed that the large particles were actually assemblies of small nanoparticles. At medium reaction temperature of -250 C, both large particles and large wire-like structures were formed from lead oleate and TOPSe precursors. Again, these large structures appeared to be assemblies of small particles. These results are similar to those reported by Sashchiuk and coworkers.28 However, when Pb(TDPA) and TOPSe were used as precursors, nearly monodisperse PbSe nanocubes were produced at 274 C. In addition, a few rods or short wires could be occasionally observed in the TEM images. We noticed that the elongated nanostructures have almost the same diameter as that of the cubes, suggesting that they were likely formed by orientedattachment mechanism.32 Structure. The crystal structure of the PbSe nanowires was investigated by powder X-ray diffraction (XRD) and high-resolution TEM (HRTEM). The XRD pattern of PbSe nanowires matches well with the standard pattern of the bulk PbSe with the rock-salt structure (Figure 4-22). The single-crystalline nature of the PbSe nanowires is shown in the HRTEM images (Figure 4-23). The fast-Fourier-transform (FFT) pattern recorded from the well-resolved wire lattice was indexed to the rock-salt structure, indicating that the PbSe nanowires grow along the [200] direction. The well-resolved lattice fringes perpendicular to the growth direction in the HRTEM image corresponded to a spacing of 0.303 nm, consistent with the ^-spacing between (200) planes in rock-salt PbSe.40 A clear boundary between the PbSe wire and the catalyst nanoparticle was observed. However, we could not observe uniform lattice fringes in the catalyst nanoparticle attached to the PbSe nanowire tip.

141

Figure 4-21. Representative TEM images of PbSe nanostructures synthesized in the absence of Bi nanoparticles in TOPO. (A) Lead oleate and TBPSe were used as precursors. The Se/Pb ratio was 5.6. The reaction was conducted at 170 C. (B) Lead oleate and TOPSe were used as precursors. The Se/Pb ratio was 1.2. The reaction was conducted at 252 C. (C) Pb(TDPA) and TOPSe were used as precursors. The Se/Pb ratio was 6. The reaction was conducted at 274 C.

142

' '

T~T ' ' ' 1

-i

i-i-i

--*JWLNW*--' Vnt*'VAJi_ll_
200 220 111
I

B
311 222 400 3314|
i |" i i i
1

20

..I..
'

422
i i I i |

i "|

H .

i ' "| , S i ' i i ,

i' |

30

40

50

60

70

80

26 (degree) Figure 4-22. Powder X-ray diffraction (XRD) pattern of PbSe nanowires (A) and the standard pattern of the bulk PbSe with rock salt structure (B, ICDD-PDF file 00-0060354).

Figure 4-23. High-resolution TEM image of a 15-nm PbSe nanowire having a catalyst nanoparticle attached to the wire tip. The inset shows the indexed fast-Fourier-transform (FFT) pattern obtained from the well-resolved wire lattice, indicating the [200] growth direction of the nanowire.

143

Figure 4-24. High-resolution TEM images of a 15-nm PbSe nanowire, recorded with prolonged exposure to the electron beam. Note that this wire is the same as the one shown in Figure 4-23. The red dashed lines separate the wire from the catalyst nanoparticle, which undergoes phase separation and recrystallization. The inset in (C) shows the indexed fast-Fourier-transform (FFT) pattern obtained from the region defined by a dashed box. Interestingly, we directly observed a phase separation and recrystallization process of the catalyst nanoparticle when it was exposed to the electron beam in the HRTEM for a prolonged time (Figure 4-24). We noticed three separate regions (I, II, and III, Figure 424D) developed from the initial catalyst nanoparticle. The evolution of these regions was clearly shown in the HRTEM images. The fast-Fourier-transform (FFT) pattern obtained from region III was indexed to cubic Bi, with the measured lattice spacing of 0.268 nm (Figure 4-24C), consistent with the ^-spacing between (110) planes in cubic Bi.41 The lattice fringes in region I and II show epitaxial relationship with the wire lattice fringes,

144

and the measured lattice spacing matches with rock-salt PbSe, confirming that these two regions are PbSe. These results clearly show the alloy nature of the catalyst nanoparticle, which is different from the observed pure Bi nanoparticle attached to the CdTe nanowire tip (see Figure 1-24).

iyPtpL|iiiii|i i ii|iini|iiinlipiiipi4nii | J I | I iifli

0.000 2.000 4.000 6.000 8.000 10.000 12.000 14.000 0.000 2.000 4.000 6.000 B.000 10.000 12.000 14.000

Energy (keV)

Energy (keV)

Figure 4-25. Energy-dispersive X-ray spectra (EDS) recorded from the catalyst nanoparticle (A) and the wire (B). Energy-dispersive X-ray spectra (EDS) collected from other wires with attached catalyst nanoparticles (Figure 4-25) contained only Pb and Se signals from the wires, but Bi, Pb, and Se signals from the catalyst nanoparticles, which confirms the alloy nature of the catalyst nanoparticles. Near-infrared absorption and quantum confinement. The near-infrared (NIR) absorption spectra of PbSe nanowires were collected from carbon tetrachloride dilutions of the purified, dried PbSe nanowires. However, we were only able to observe wellresolved NIR absorption feature from the smallest wires with the mean diameter of 6.8 nm (Figure 4-26) so far. The absorption position was determined to be 2280 nm by a nonlinear least-squares fit to the spectrum using an exponential-decay function for the background and one Gaussian function for the absorption feature. This feature is strongly

145

blue-shifted in comparison to the PbSe bulk absorption edge at 4460 nm, corresponding to a strong quantum confinement of AEg = 266 meV. As expected, because the bulk exciton Bohr radius in PbSe is quite large (46 nm), 6.8-nm diameter PbSe nanowires are strongly confined.
^

1.00.8<D O C L-

Model ExpDecJ Gauss Weighting: No weighting y ChiA2/DoF 0.00006 R2 = 0.99899 yO A0 to A1 w1 xcl Experimental spectrum Exponential decay background - - - Gaussian peak 0.02467 18.61814 516.26269 25.28417 260.28457 2279.90552 0.00275 0.28243 2.92087 0.71667 5.39911 2.08104

0.60.4-

o <

.a

(0

^.

0.20.0I

"**. ** *.

^*v^^

" "" " ^ ^ " ^

'

'

1 i

"->

'

1600

1800

2000

2200

2400

2600

Wavelength (nm)

Figure 4-26. Near-infrared spectrum of 6.8-nm PbSe nanowires dispersed in carbon tetrachloride. The dashed and doted lines are the exponential decay background and the Gaussian peak, respectively, which were obtained by a nonlinear least-squares fit to the spectrum. The reason for the featureless absorption spectra of PbSe nanowires with diameters larger than ~9 nm may be twofold. First, collecting the NIR absorption spectrum of PbSe nanowires requires relatively high wire concentration, which is problematic for largediameter wires because of rapid precipitation. Second, the solvent absorption may interfere with the wire absorption when the wire absorption occurs at > 2500 nm. Systematic investigations of the NIR absorption and photoluminescence of our highquality colloidal PbSe nanowires having diameters in a broad range should be a direction of future research. 146

Discussion Syntheses of PbSe nanostructures: strongly coordinating solvent vs. weakly coordinating solvent. The strongly coordinating solvent TOPO was rarely used for the synthesis of PbSe nanocrystals due to formation of polycrystalline assemblies of PbSe nanocrystals.28 In addition, Kuno and coworkers reported that the use of TOPO solvent prevented wire growth for their SLS growth of PbSe nanowires from Au/Bi core/shell catalyst nanoparitcles.36 However, our syntheses clearly demonstrated that TOPO could serve as an excellent solvent for the SLS growth of high-quality, diameter-controlled PbSe nanowires. This inconsistency may be understood in the following ways. Generally, the precursor reactivities in weakly coordinating solvents should be higher than those in strongly coordinating solvents. Thus, it's essential to use catalyst

nanoparticles with very high activities to favor the SLS growth of PbSe nanowires instead of other competing reaction pathways in weakly coordinating solvents. In other words, small catalyst nanoparticles are preferred when weakly coordinating solvents are employed. We note that Au/Bi core/shell catalyst nanoparticles having very small

diameters (< 3 nm for the core/shell nanoparticles) were indeed used by Kuno and coworkers to catalyze the wire growth in weakly coordinating solvent mixture.36 However, our SLS growth of PbSe nanowires employed pure Bi nanoparticles having diameters > 4 nm, whose catalytic activities should be relatively lower compared with the very small Au/Bi core/shell nanoparticles. Consequently, we failed to produce wellcontrolled PbSe nanowires in the weakly coordinating phenyl ether solvent (see Figure 43); the successful synthesis was achieved by using less-reactive precursors in the strongly coordinating solvent TOPO, in accordance with our relatively large Bi nanoparticles.

147

There are several advantages for our relatively large Bi nanoparticles compared with the very small Au/Bi core/shell nanoparticles used by Kuno. First, relatively large nanoparticles have a low tendency to agglomerate, which benefits the growth of PbSe nanowires with narrow diameter distributions. Second, our Bi nanoparticles have a broad accessible diameter range, which helps to produce PbSe nanowires having controlled diameters in a broad range.

18-

I " L-

o
^nanowire
> d

Bi

jB 14

1 12T>

o o

2 io-

8 8

5 O
C (6 c
O

8-

c 6
A
"}

o
*

dnanowire < dBi D.

tft

r > T ' I 6 l 4

'

' I

10

12

14

16

18

Initial Bi nanoparticle diameter (nm) Figure 4-27. Plot of the PbSe nanowire diameter (dnanowire) vs. the initial Bi nanoparticle diameter (JB0- All our PbSe nanowires are in the dnanowire >fifeiregion. Swelling of catalyst nanoparticles and control of wire diameters. As mentioned in the Results section, the swelling of catalyst nanoparticles was evident, especially for small Bi nanoparticles at relatively high reaction temperatures. A plot of the PbSe nanowire diameter vs. the initial Bi nanoparticle diameter (Figure 4-27) clearly shows that all our high-quality PbSe nanowires are in the <^nanowire > d&i region, which is inconsistent with our previously reported results for currently available II-VI and III-V

148

semiconductor nanowires grown from Bi nanoparticles by SLS mechanism.42 This dnanowire > ^Bi behavior for PbSe nanowires may be reasonably due to the high solubility of the semiconductor elements in the liquid alloy nanodroplets during the wire growth, and/or the small contact angle of the liquid-solid interface. As noted above, the observed phase separation and recrystallization process of the catalyst nanoparticle under electron beam irradiation in the HRTEM strongly indicated the alloy nature of the catalyst nanoparticle, which is inconsistent with our results in CdTe nanowires shown in Chapter 1. To address this disparity, we now make a comparison between the bulk binary Bi-Pb and Bi-Cd alloy phase diagrams (see Figure 4-2 and Figure 1-5). It is clearly shown that the Bi-Pb alloy shows a lower eutectic point than does Bi-Cd alloy. We assume this is also true for the corresponding nanoscale phase diagrams. As a consequence, the low eutectic point in Bi-Pb alloy should help to keep the alloy nature of the catalyst nanoparticles upon cooling after the synthesis procedure.
17

2
E m g

16 151413121110

"5 O
c
CO

c co
Q.

8-i 7
6
1 1 1 1 1 1 1

D lead oleate, TOPSe O lead oleate, TBPSe A lead octanoate, TBPSe

1 1 1 1 1 1 1 1 r

160

170

180

190

200

210

220

230

240

250

Reaction temperature ( C )

Figure 4-28. Plot of PbSe nanowire diameter vs. reaction temperature for various precursor sets. The mean diameter of the Bi catalyst nanoparticles used for these syntheses was 4.3 nm. 149

There are advantages and disadvantages of the significant swelling of catalyst nanoparticles. We may be able to purposefully control the diameters of the PbSe nanowires by varying the reaction temperatures because the solubility of the semiconductor in the alloy nanodroplets should be temperature dependent. This

possibility has been demonstrated by our experimental results plotted in Figure 4-28, showing that the diameter of the PbSe nanowires generally decreases as the reaction temperature decreases for given precursor sets. However, the significant swelling of catalyst nanoparticles also makes it a synthetic challenge to grow small-diameter nanowires. So far, the growth of high-quality PbSe nanowires having mean diameters of less than 6.8 nm has not been achieved.

Conclusion High-quality rock-salt PbSe nanowires having diameters in the range of 6-18 nm were grown by the SLS mechanism using nearly monodisperse Bi catalyst nanoparticles in a strongly coordinating TOPO solvent. The wire-diameter control was achieved by varying the reaction temperatures, the precursors, and the Bi nanoparticles. At low reaction temperatures ^190 C, lead carboxylate and tri-n-alkylphosphine selenide precursors with short carbon chains and small Bi nanoparticles were required to produce wellcontrolled nanowires having diameters < 10 nm. At high reaction temperatures around 270 C, thermally more stable lead alkylphosphonate precursor and large Bi nanoparticles were preferred. The mean diameters of the PbSe nanowires were found to be always larger than those of the corresponding initial Bi catalyst nanoparticles due to significant swelling of the catalyst nanoparticles. Near-infrared absorption was observed from the

150

carbon tetrachloride suspension of 6.8-nm PbSe nanowires, indicating a strong quantumconfinement effect. To our knowledge, this is the first time that reliable near-infrared absorption from colloidal PbSe nanowires has been reported. These high-quality PbSe nanowires are excellent candidates for further fundamental studies, and for potential applications such as in solar cells,5'13 photodetectors,6 and field-effect transistors.9'10

Experimental Section Chemicals. Lead(II) oxide (99.9+%, Aldrich), octanoic acid (99.5+%, Acros), oleic acid (99%, Aldrich), stearic acid (98+%, Aldrich), n-tetradecylphosphonic acid (98.8%, Polycarbon Industries), tri-n-butylphosphine (TBP, 97%, Aldrich), tri-ra-octylphosphine (TOP, 97%, Strem), selenium shot (99.999%, Alfa Aesar) was used as purchased without further purification. Tri-n-octylphosphine oxide (TOPO, technical grade, 90%, Aldrich) was vacuum distilled before use. HPLC-grade toluene, methanol, hexane, acetone, and carbon tetrachloride were purchased from a variety of suppliers and used as received. All manipulations were carried out using standard Schlenk line techniques under dry, 0 2 -freeN 2 (g). Preparation of TOPSe and TBPSe stock solutions. TOPSe and TBPSe stock solutions with concentrations of 0.4 mmol/g and 2.0 mmol/g were prepared. General preparation procedure: Selenium shot and TOP (or TBP) were mixed at room temperature in a glove box. Then the mixture was heated to -150 C with stirring until all the selenium shot disappeared (< 30 min) to form a transparent homogeneous solution, which was cooled to room temperature and stored in the glove box for use.

151

In a typical preparation of 0.4 mmol/g TOPSe stock solution, selenium shot (0.632 g, 8.0 mmol) and TOP (19.368 g) were loaded into a Schlenk flask with a stir bar. The mixture was then heated to -150 C in a pre-heated salt bath (NaN03/KN03/Ca(N03)2, 7:45:48 by weight) to form a clear solution, which was allowed to cool, and transferred to a glove box for storage. The 0.4 mmol/g TBPSe stock solution and the 2.0 mmol/g TOP or TBP stock solution were prepared similarly. Synthesis of Bi nanoparticles. Bi catalyst nanoparticles were prepared by a

modified literature method43 and detailed syntheses of nanoparticles over a range of sizes will be reported elsewhere. General preparation of PbSe nanowires. In a typical preparation, lead(II) oxide, acid, and distilled TOPO were loaded into a Schlenk reaction tube. In a separate vial, the Bi-nanoparticle stock solution and the TOPSe (or TBPSe) stock solution were mixed, and the vial was septum capped. The reaction tube was then inserted into a preheated salt bath at a specific temperature to obtain a clear solution. The reaction tube was allowed equilibrate to the salt-bath temperature for several minutes, and then the contents of the vial were quickly injected into the reaction tube by a syringe. Within 10 s the color of the reaction mixture was observed to change to black as the nanowires grew. The reaction tube was withdrawn from the salt bath 5 min after the injection and allowed to cool. Before the reaction mixture was solidified, toluene (~5 mL) was added to the reaction tube to prevent solidification. The PbSe-nanowire product was indefinitely stable in this form, and was typically stored in this form. The quantities of chemicals used for the syntheses were summarized in Table 4-1.

152

Table 4-1. Synthetic conditions and results for PbSe nanowires and other morphologies.
PbO (mmol) 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 Acid" (mmol) OCA 0.23 OLA 0.21 OLA 0.23 OCA 0.21 OLA 0.22 OLA 0.20 OLA 0.21 OLA 0.22 OLA 0.22 OLA 0.21 TDPA 0.10 TDPA 0.10 OCA 0.21 OLA 0.24 OLA 0.20 OLA 0.21 OLA 0.21 OLA 0.22 TDPA 0.12 OLA 0.21 OLA 0.21 TDPA 0.10 TOPO (g) 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 R3P=Se* (mmol) TBPSe 0.30 TBPSe 0.60 TBPSe 0.28 TBPSe 0.30 TOPSe 0.10 TBPSe 0.28 TOPSe 0.08 TOPSe 0.06 TOPSe 0.06 TOPSe 0.06 TOPSe 0.30 TOPSe 0.30 TBPSe 0.30 TBPSe 0.28 TBPSe 0.28 TBPSe 1.0 TOPSe 0.08 TBPSe 0.06 TOPSe 0.06 TOPSe 0.06 TOPSe 0.06 TOPSe 0.30 Time (min) 5 5 5 4 4 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 Bi stock solution (mg) 16 13 12 16 12 11 12 24 25 21 17 16 16 12 12 16 11 12 25 24 22 14
^wire

CO
170 160 170 180 190 180 201 250 253 243 274 274 162 160 170 160 182 182 265 250 262 273

(nm) 4.3 4.3 4.3 4.3 4.3 4.3 4.3 7.1 8.9 4.3 12.4 13.5 4.3 4.3 6.4 3.3 4.3 4.3 14.3 13.5 14.3 15.8

(nm) 6.8 8.8 8.8 9.1 10.5 10.8 11.4 12.3 13.1 14.2 15.8 17.8 -

Morphologies wires wires wires wires wires wires wires wires wires wires wires wires short tapered wires wires and particles wires and particles rough wires wires (broad d distribution) short wires (low yield) short wires (low yield) Some wires w/ rough surface some wires w/ rough surface wires and cubes

153

TDPA 0.20 TDPA 0.10 0.21 OLA 0.05 0.20 OCA 0.05 0.23 OLA 0.05 0.21 TDPA 0.05 0.10 OLA 0.05 0.22 SA 0.05 0.19 SA 0.05 0.20 fl OCA = octanoi c 0.10

TOPSe 4 270 0.8 TOPSe 272 4.0 5 0.8 TBPSe 4.0 170 5 0.28 TBPSe 4.0 170 5 0.30 TOPSe 252 4.0 5 0.06 TOPSe 274 4.0 5 0.30 TBPSe 5.0" 180 4 0.12 TOPSe 4.0C 253 5 0.06 TOPSe 4.0d 253 5 0.06 acid; (3LA = ol<5ic acic ; TDPA 4.0

21 21 none none none none 23 30 30

8.9 13.5 -~ 4.3 7.1 7.1

--

wires and cubes wires and cubes large particles mainly large particles mainly large particles mainly cubes rough wires and NPs short rods short rods

= n-tetradecylp lospho nic acid; SA =

stearic acid. b Phenyl ether. c Octyl ether. d Poly-1-decene. e TBPSe = tri-n-butylphoshpine selenide; TOPSe = tri-n-octylphoshpine selenide. f A ternary salt bath (NaN03/KN03/Ca(N03)2, 7:45:48 by weight) was used for reactions conducted below 240C. A binary salt bath (NaN0 3 /KN0 3 , 46:54 by weight) was used for reactions conducted above 240C.

Isolation and characterization. The nanowires were separated from aliquots (~2 mL) of the cooled reaction mixture by addition of methanol (-10 mL). The precipitated wires were collected by centrifugation and the supernatant was decanted. The wires were then redispersed in hexane (~2 mL) and precipitated by adding acetone (~10 mL), followed by centrifugation and decantation. This process was repeated 2 or 3 times to remove the byproducts generated during the synthesis. The purified wires were

redispersed in toluene for TEM sample preparation. However, for the near-infrared study, the purified wires were dried under a stream of N2 to remove volatile organics, and then redispersed in carbon tetrachloride. Carbon-coated Cu TEM grids were prepared by evaporating one drop of the NW toluene solutions on them. Low-resolution TEM images

154

and energy-dispersive X-ray spectra (EDS) were collected using a JEOL 2000 FX microscope operating at 200 kV. High-resolution TEM images were collected using a JEOL JEM-2100F microscope operating at 200 kV. The XRD pattern was obtained using a Rigaku Dmax A vertical powder diffractometer with Cu Ka radiation (k = 1.5418 A). The near-infrared spectra were collected using a Perkin-ElmerLamda 950 UV-visNIR spectrometer. Determination of the diameter distributions of PbSe nanowires. For each sample, the diameter distributions were determined using several images at 500K magnification to ensure adequate resolution for accurate measurement. Diameters were measured and recorded using Image-Pro Plus software (version 5.0 for Windows). Digital images were zoom expanded within the software. The software-generated diameter lists were

transferred to and manipulated in a standard spreadsheet program for statistical analysis. The distributions were generally constructed from at least 300 diameter measurements for each specimen. Nonlinear least-squares fitting of the NIR absorption spectrum. The nearinfrared absorption spectrum was fit using a method reported previously.43 Briefly, the spectrum was fit with one exponential (for the background) and one Gaussian function (for the absorption feature) using Origin 7.5 software (www.OriginLab.com). The fitting procedure yielded the center energy of the excitonic peak and the error in the center energy. The wire band gap was assigned to the center of the Gaussian peak.

155

References (1) (2) (3) (4) Madelung, O. Semiconductors: Data Handbook; 3rd ed.; Springer: Berlin, 2004. Wise, F. W. Ace. Chem. Res. 2000, 33, 773-780. Allan, G.; Delerue, C. Phys. Rev. B 2004, 70,245321. Choudhury, K. R.; Sahoo, Y.; Ohulchanskyy, T. Y.; Prasad, P. N. Appl. Phys. Lett. 2005, 87,073110. Cui, D.; Xu, J.; Zhu, T.; Paradee, G.; Ashok, S.; Gerhold, M. Appl. Phys. Lett. 2006, 8,183111. Qi, D.; Fischbein, M.; Drndic, M.; Selmic, S. Appl. Phys. Lett. 2005, 86,093103. Michalet, X.; Pinaud, F. F.; Bentolila, L. A.; Tsay, J. M.; Doose, S.; Li, J. J.; Sundaresan, G.; Wu, A. M.; Gambhir, S. S.; Weiss, S. Science 2005, 307, 538544. Finlayson, C. E.; Amezcua, A.; Sazio, P. J. A.; Walker, P. S.; Grossel, M. C ; Curry, R. J.; Smith, D. C.; Baumberg, J. J. J. Mod. Opt. 2005, 52, 955-964. Talapin, D. V.; Murray, C. B. Science 2005, 310, 86-89. Talapin, D. V.; Black, C. T.; Kagan, C. R.; Shevchenko, E. V.; Afzali, A.; Murray, C. B. J. Phys. Chem. C2007, 111, 13244-13249. Schaller, R. D.; Petruska, M. A.; Klimov, V. I. J. Phys. Chem. B 2003, 107, 13765-13768. Schaller, R. D.; Klimov, V. I. Phys. Rev. Lett. 2004, 92,186601. Schaller, R. D.; Sykora, M.; Pietryga, J. M.; Klimov, V. I. Nano Lett. 2006, 6, 424-429. Ellingson, R. J.; Beard, M. C ; Johnson, J. C ; Yu, P.; Micic, O. I.; Nozik, A. J.; Shabaev, A.; Efros, A. L. Nano Lett. 2005, 5, 865-871. Luther, J. M.; Beard, M. C ; Song, Q.; Law, M.; Ellingson, R. J.; Nozik, A. J. Nano Lett. 2007, 7, 1779-1784. Klimov, V. I. J. Phys. Chem. B 2006, 110, 16827-16845. Wehrenberg, B. L.; Guyot-Sionnest, P. J. Am. Chem. Soc. 2003, 125, 7806-7807.

(5)

(6) (7)

(8)

(9) (10)

(11)

(12) (13)

(14)

(15) (16) (17)

156

Murray, C. B.; Sun, S.; Gaschler, W.; Doyle, H.; Betley, T. A.; Kagan, C. R. IBM J. of Res. Dev. 2001, 45, 47-56. Yu, W. W.; Falkner, J. C ; Shih, B. S.; Colvin, V. L. Chem. Mater. 2004, 16, 3318-3322. Pietryga, J. M.; Schaller, R. D.; Werder, D.; Stewart, M. H.; Klimov, V. I.; Hollingsworth, J. A. /. Am. Chem. Soc. 2004,126, 11752-11753. Lu, W.; Fang, J.; Ding, Y.; Wang, Z. L. J. Phys. Chem. B 2005, 109, 1921919222. Xu, J.; Ge, J. P.; Li, Y. D. J. Phys. Chem. B 2006, 110, 2497-2501. Lifshitz, E.; Bashouti, M.; Kloper, V.; Kigel, A.; Eisen, M. S.; Berger, S. Nano Lett. 2003, 3, 857-862. Gokarna, A.; Jun, K.-W.; Khanna, P. K.; Baeg, J.-O.; Seok, S. I. Bull. Korean Chem. Soc. 2005, 26,1803-1806. Yong, K. T.; Sahoo, Y.; Choudhury, K. R.; Swihart, M. T.; Minter, J. R.; Prasad, P. N. Nano Lett. 2006, 6, 709-714. Shi, W.; Sahoo, Y.; Zeng, H.; Ding, Y.; Swihart, M. T.; Prasad, P. N. Adv. Mater. 2006, 18, 1889-1894. Ji, T.; Jian, W. B.; Fang, J. J. Am. Chem. Soc. 2003, 125, 8448-8449. Sashchiuk, A.; Amirav, L.; Bashouti, M.; Krueger, M.; Sivan, U.; Lifshitz, E. Nano Lett. 2004, 4, 159-165. Wang, W.; Geng, Y.; Qian, Y.; Ji, M.; Liu, X. Adv. Mater. 1998, 10, 1479-1481. Liu, Y.; Cao, J.; Zeng, J.; Li, C ; Qian, Y.; Zhang, S. Eur. J. Inorg. Chem. 2003, 2003, 644-647. Zhao, W.-B.; Zhu, J.-J.; Chen, H.-Y. Scripta Mater. 2004, 50,1169-1173. Cho, K. S.; Talapin, D. V.; Gaschler, W.; Murray, C. B. J. Am. Chem. Soc. 2005, 727,7140-7147. Fardy, M.; Hochbaum, A. I.; Goldberger, J.; Zhang, M. M.; Yang, P. Adv. Mater. 2007,79,3047-3051. Zhu, J.; Peng, H.; Chan, C. K.; Jarausch, K.; Zhang, X. F.; Cui, Y. Nano Lett. 2007, 7, 1095-1099.

157

(35) (36)

Bierman, M. J.; Lau, Y. K. A.; Jin, S. Nano Lett. 2007, 7,2907-2912. Hull, K. L.; Grebinski, J. W.; Kosel, T. H.; Kuno, M. Chem. Mater. 2005, 17, 4416-4425. Houtepen, A. J.; Koole, R.; Vanmaekelbergh, D.; Meeldijk, J.; Hickey, S. G. J. Am. Chem. Soc. 2006, 128, 6792-6793. Murray, C. B.; Norris, D. J.; Bawendi, M. G. J. Am. Chem. Soc. 1993, 115, 87068715. Li, L. S.; Pradhan, N.; Wang, Y.; Peng, X. Nano Lett. 2004, 4, 2261-2264. ICDD-PDF No. 00-006-0354. ICDD-PDF No. 01-071-4088. Wang, R; Dong, A.; Sun, J.; Tang, R.; Yu, H.; Buhro, W. E. Inorg. Chem. 2006, 45,7511-7521. Yu, H.; Li, J.; Loomis, R. A.; Gibbons, P. C ; Wang, L. W.; Buhro, W. E. J. Am. Chem. Soc. 2003, 125, 16168-16169.

(37)

(38)

(39) (40) (41) (42)

(43)

158

Chapter 5 SLS Growth of II-VI and IV-VI Metal Sulfide Semiconductor Nanowires: A Single-Source Precursor Approach

159

Introduction II-VI and IV-VI metal sulfide semiconductors are important materials due to their attractive electrical and optical properties. For example, lead sulfide (PbS) is a direct, narrow band gap semiconductor with a bulk band gap of 0.42 eV (2952 nm) at 300K, while cadmium sulfide (CdS) is a direct, wide band gap semiconductor with a bulk band gap of 2.482 eV (500 nm) at 300K, which is about 6 times larger than that of PbS.1 Thus, II-VI and IV-VI metal sulfides nanostructures may find broad applications in spectral regions ranging from the infrared to the visible, and up to the ultraviolet due to the broad range of accessible band gaps. Recently multiple-exciton generation was discovered in nanocrystals of PbS and PbSe by different research groups.2'3 Specifically, up to four excitons (quantum yield up to 430%) were produced by absorption of a single photon in PbS nanocrystals.2 Although the ultrafast nonradiative Auger recombination of multi-excitons prevents the carriers from being harvested so far, this obstacle may be overcome by using elongated nanostructures, such as nanorods and nanowires.4 If the rate of Auger recombination is significantly slower in ID nanostructures, the efficiency of photovoltaic energy conversion might be greatly enhanced. Additionally, PbS has a relatively large exciton Bohr radius of 20 nm,5 which makes it one of the ideal candidates for investigating quantum-confinement effects in relatively large nanostructures. Although the exciton Bohr radius of CdS is only 2.8 nm,6 this material is one of the very important wide band gap semiconductors, having many applications including lightemitting diodes,7'8 photodetectors,9 and lasers.10

160

Tremendous efforts have been made to grow PbS and CdS nanowires using the hardtemplate approach,1112 CVD and related approaches,13"15 the solvothermal method,16 the nanoparticle-induced anisotropic growth,17 and surfactant-directed 1-D growth.18'19 However, for all of these synthetic strategies control over mean diameters and diameter distributions were limited.

Figure 5-1. PbS nanostructures synthesized from dual-source precursors. (A) The precursors were lead oleate and sulfur dissolved in 1-octadecene, and (B) the precursors were lead stearate and tri-n-octylphosphine sulfide (TOPS). Bi nanoparticles with the mean diameter of 7.1 nm were used as catalysts. The reactions were conducted in TOPO at ~250C. The SLS synthesis of nanowires offers several advantages over other methods, including purposeful control of nanowire mean diameters, narrow diameter distributions, small diameters in the quantum-confinement regime, control of surface passivation, nanowire solubility, and ease of implementation.20 However, although SLS growth has been applied to a wide range of II-VI,21"26 ni-V,2732 and IV-VI33 nanowire compositions,
the method is not yet generally applicable, as each new composition requires extensive

trial-and-error experimentation to identify appropriate precursors and reaction conditions. Our prior work has employed separate metallic-element and nonmetallic-element

161

precursors to synthesize high-quality, compound-semiconductor nanowires. ' ' ' Finding dual precursors with appropriately balanced reactivities has often been a primary origin of the extensive empirical experimentation required to develop a given, successful SLS nanowire synthesis, as described in Chapter 1 and Chapter 4. In several cases, such as for metal-sulfide semiconductors, the dual-precursor SLS approach has failed to produce highly crystalline, diameter-controlled nanowires (Figure 5-1 and Figure 5-2).

Figure 5-2. CdS nanostructures synthesized from dual-source precursors. (A) The precursors were cadmium tetradecylphosphonate and sulfur dissolved in 1-octadecene, and (B) the precursors were cadmium tetradecylphosphonate and TOPS. Bi nanoparticles with the mean diameter of 9.4 nm were used as catalysts. In this Chapter, we report that the use of single-source precursors, in which the metallic and nonmetallic semiconductor constituents are combined in a single molecule, solves the reactivity-balance problem, resulting in the successful SLS growth of PbS and CdS nanowires. Therefore, the single-source approach has the potential to greatly extend the generality of the SLS method for semiconductor-nanowire synthesis. As singlesource precursors have been extensively developed for use in the MOCVD growth of thin films,35"39 suitable precursors for the SLS growth of many new nanowire compositions are likely already available. 162

Results (1) PbS nanowires Synthesis. For the SLS growth of PbS nanowires, lead diethyldithiocarbamate [Pb(S2CNEt2>2] was chosen as the single-source precursor because it has been successfully employed for the MOCVD growth of PbS thin films40 and the solution-based growth of PbS nanostructures.41'42 Bi nanoparticles were used as the catalysts because of the low eutectic point of 125.5 C in the bulk Bi-Pb binary phase diagram (Figure 4-2). Additionally, we demonstrated in Chapter 4 that Bi catalyst nanoparticles were excellent seeds for the SLS growth of PbSe nanowires, which encouraged us to employ Bi nanoparticles for the SLS growth of PbS nanowires. Our designed synthetic route is shown in Figure 5-3. Compared with the dual-source precursor approach (see Figure 1-4 and Figure 4-1), the single-source precursor approach was more convenient in two ways. First, the formation of metal-precursor complex through the reaction between a metal oxide and an acid at elevated temperatures is not necessary. Second, the preparation of the precursor stock solution for the nonmetalic element is not necessary.

Solution

Liquid

Solid

Pb(S2CNEt2)2

t^^^m^^^^^^^m^^^^m^^^^i

By-products Bi PbS nanowire

Figure 5-3. Schematic illustration of the designed synthetic route for the solution-liquidsolid (SLS) growth of PbS nanowires.

163

Weakly coordinating or noncoordinating solvents are typically used for the synthesis of high-quality lead-salt nanocrystals; therefore, our initial attempts to grow PbS nanowires were conducted in the noncoordinating solvent 1-octadecene (ODE) and the weakly coordinating solvent phenyl ether (PE), in the presence of Bi nanoparticles. Only rods were produced when ODE was used as the solvent (Figure 5-4A). The catalyst nanoparticles could not be observed at either end of the rods, indicating that the rods may not have been produced by the SLS mechanism. When the reaction was conducted in PE, wires were observed in the TEM image (Figure 5-4B). However, the diameter

distribution of the wires was found to be very broad. In addition, the wire surface was rough. Furthermore, some large star-shaped nanoparticles were also shown in the image. The results indicated that the control of the PbS nanowire growth was very poor in these noncoordinating and weakly coordinating solvents.

Figure 5-4. TEM images of PbS nanostructures synthesized in 1-octadecene (A) and phenyl ether (B). When the reactions were conducted in the strongly coordinating solvent tri-noctylphosphine oxide (TOPO), high-quality PbS nanowires were successfully produced. In a typical synthesis of PbS nanowires, Pb(S2CNEt2>2 was first dissolved in phenyl ether 164

with warming, and combined with a tri-n-octylphosphine (TOP) solution of Bi catalyst nanoparticles. The mixture was then rapidly injected into a pre-heated TOPO solvent at a desired temperature (in the range of 230 - 300 C), held at this temperature for 5 min, and allowed to cool.

Figure 5-5. Representative low-magnification TEM images of PbS nanowires of various diameters (d). (A) d = 8.7 2.4 nm ( 28%). This image at very low magnification shows that the lengths of the wires are typically several micrometers and the long straight wires tended to form bundles. (B) d = 10.2 2.5 nm ( 25%), (C) d = 11.2 2.3 nm ( 21%), (D) d = 13.2 2.9 nm ( 22%), (E) d = 16.1 2.9 nm ( 18%), and (F) d = 31.2 3.9nm(13%).

165

Figure 5-6. Representative high-magnification TEM images of PbS nanowires of various diameters (d). (A) d = 8.7 2.4 nm ( 28%), (B) d = 10.2 2.5 nm ( 25%), (C) d = 11.2 2.3 nm ( 21%), (D) d = 13.3 2.8 nm ( 21%), (E) d = 16.1 2.9 nm ( 18%), and (F) d = 31.2 3.9 nm (+ 13%). Representative low- and high-magnification TEM images of the PbS nanowires shown in Figure 5-5 and Figure 5-6 indicate that the wire diameters could be varied over the range of 9-31 nm. The lengths of the nanowires prepared in TOPO solvent were typically several micrometers, and the long wires tended to form bundles (Figure 5-5A),

166

likely as a result of strong van der Waals attractions between the nanowires.43 The nanowires exhibited uniform diameters along their lengths (Figure 5-6), and the standard deviation in the diameter distributions was less than 30% of the mean diameter (Figure 57). For the high-quality PbS nanowires, catalyst nanoparticles were often observed at one end of the wires (Figure 5-8), confirming the SLS growth mechanism. Control

experiments were also conducted without Bi-catalyst nanoparticles; only nanoparticles with irregular shapes were produced (Figure 5-9).

140120100-

. . , .. .....

' '

Counts

4020-

10

12

14

0-

PbS nanowire diameter (nm)

JIIIll..
10 15

I III

1.
I | ipm 20 25

PbS nanowire diameter (nm)

8070605040302010-

ll

| |

0-

mill10 12 14 16 18

l l ll ll Mil.

!!!

20

22

24

10

12

14

16

18

20

22

24

26

PbS nanowire diameter (nm)

PbS nanowire diameter (nm)

Figure 5-7. Representative diameter-distribution histograms for PbS nanowires of various diameters (d). (A) d = 8.7 + 2.4 nm (+ 28%), (B) d = 11.2 + 2.3 nm (+ 21%), (C) d = 13.0 2.7 nm ( 21%), and (D) d = 16.1 2.9 nm ( 18%).

167

Figure 5-8. Additional high-magnification TEM image of PbS nanowires. The mean diameter of this sample is 11.0 nm. The solid red arrows point to the catalystnanoparticle tips, confirming the SLS growth mechanism.

Figure 5-9. TEM images of PbS nanostructures synthesized in the absence of Bi nanoparticles. The use of the TOPO/TOP mixed solvent enabled controlled growth of shorter nanowires (Figure 5-10), with mean lengths dependent on the TOPO/TOP ratio. Lower TOPO/TOP ratios produced shorter wires.

168

Figure 5-10. TEM images of PbS nanowires synthesized in a mixed TOPO/TOP solvent. (A) TOPOATOP = 3/1, (B) TOPO/TOP = 1/1, and (C) TOPO/TOP = 1/3. The total weight of the mixed solvent and the quantity of Bi nanoparticles were held constant.

Figure 5-11. Diameter control of PbS nanowires synthesized using 10.3-nm Bi nanoparticles at various reaction temperatures (T). (A) T = 212 C, (B) T = 232 C, and (C) T = 277 C. The mean diameters of wires shown in (B) and (C) are 11.0 and 13.3, respectively. The mean nanowire diameters were controlled by varying the reaction temperature, and by using an additional surfactant, n-tetradecylphosphonic acid (TDPA) (Table 5-2). Larger nanowire diameters were favored at higher reaction temperatures using Bi nanoparticles of a given diameter. Thus, 10.3-nm Bi nanoparticles produced 11.0 nm and 13.3 nm NW diameters at 232 C and 277 C, respectively (Figure 5-1 IB, C). Reaction temperatures at about 210 C were unsuccessful, which was true for both large (Figure 511 A) and small Bi nanoparticles (Figure 5-12A). The nanowire diameters were typically 169

larger than the initial diameters of the Bi nanoparticles, especially for smaller-diameter Bi nanoparticles. Surprisingly, the largest nanowires (31-nm diameter, Figure 5-12B) were obtained from the smallest (4.3 nm) Bi nanoparticles. However, a minority population of 8.9-nm nanowires was also produced under these conditions. The results indicated that the nanowires grew from larger Bi nanoparticles formed by nanoparticle agglomeration. Because the smallest Bi nanoparticles exhibit the strongest agglomeration tendencies, the best nanowire diameter control was achieved with the relatively large catalyst nanoparticles. Efforts have been made to synthesize large-diameter PbS nanowires using very large Bi nanoparticles. However, both wires and nanoparticles were produced when the initial diameter of the Bi catalyst nanoparticles was larger than ~12 nm (Figure 5-13), which should be due to the low catalytic activity of very large Bi nanoparticles.

Figure 5-12. TEM images of PbS nanowires synthesized using 4.3-nm Bi nanoparticles at various reaction temperatures (7). (A) T = 210 C, and (B) T = 224 C. The inset in (B) is the diameter-distribution histogram of this sample, showing a bimodal diameter distribution [d = 31.2 3.9 nm (majority) and 8.9 2.6 nm (minority)]. The black arrow in (B) points to a catalyst nanoparticle attached to a large nanowire.

170

Figure 5-13. TEM images of PbS nanostructures synthesized using large Bi nanoparticles having various mean diameters (JBO- (A) dm = 12.4 nm, and (B) d-&i = 13.5 nm. Although additional surfactants are not necessary to produce high-quality PbS nanowires, the use of a small amount of TDPA in the synthesis helped to produce smaller-diameter wires. For example, 8.7-nm PbS nanowires were grown using 10.3-nm Bi nanoparticles in the presence of TDPA (TDPA/Cd = 0.5, see Table 5-2). However, the amount of TDPA was critical; TDPA/Cd ratios larger than 1 produced short wires, cubes, and irregularly shaped particles (Figure 5-14). Interestingly, the catalyst

nanoparticle tips were at the thick ends of the wires when the TDPA/Cd ratio was 1.25, which should be due to the swelling of the catalyst particles during the wire growth. On the contrary, the catalyst tips were found at the narrow ends of the wires when the TDPA/Cd ratio was further increased to 2.50, which is uncommon in the SLS growth of various II-VI, III-V, and IV-VI semiconductor nanowires. At present, we don't know the 171

reason for this. We surmised that large amount of TDPA might interfere with the diffusion of Pb from the precursors into the catalyst nanoparticles because TDPA is a very strongly binding ligand for Pb.

Figure 5-14. TEM images of PbS nanostructures synthesized in the presence of TDPA. (A) TDPA/Cd = 1.25, and (B) TDPA/Cd = 2.50. The catalyst nanoparticle tips are highlighted by red dashed circles. Bi nanoparticles having the mean diameter of ~10 nm were used as catalysts. The reaction was conducted at 272 C. Another commonly used nanowire surfactant, n-hexadecylamine (HDA), was found to be synthetically detrimental in this system (Figure 5-15). When the HDA/Pb ratio was 0.8, only short rod-like nanostructures were produced. A further increase of the HDA/Pb ratio to 2.8 resulted in the formation of cubes.

172

Figure 5-15. TEM images of PbS nanostructures synthesized in the presence HDA. (A) HDA/Pb = 0.8, and (B) HDA/Pb = 2.8. Bi nanoparticles having the mean diameter of 7.1 nm were used as catalysts. The reaction was conducted at 254 C.

Figure 5-16. Representative TEM images of PbS nanowires synthesized using the TOP dispersion of Pb(S2CNEt2)2 as the injection mixture. Bi nanoparticles having the mean diameter of 8.9 nm were used as catalysts. The reaction was conducted at 244 C.

It should be pointed out that the solubility of Pb(S2CNEt2)2 in common organic solvents such as TOP is very low even at elevated temperatures. If the injection solution contains undissolved but well-dispersed Pb(S2CNEt2)2 precursor in TOP, the control of the wire growth could be poor because the precursor is not in a homogeneous molecular state. Representative TEM images of PbS nanowires synthesized using the TOP

dispersion of Pb(S2CNEt2)2 showed a very broad wire-diameter distribution (Figure 5-16),

173

which should be due to the inhomogeneous dissolution and decomposition of the solid state precursors. All the successful syntheses of high-quality PbS nanowires shown above were conducted in vacuum distilled 90-% technical grade TOPO and 99-% TOPO with the batch number of 00529CD. However, we found that the wire growth was dramatically influenced by the source of the 99-% TOPO (Figure 5-17). For example, when the 99-% TOPO with the batch number of 03008CE was employed, long wires could not be produced. Thus, the researchers who want to repeat our results should take great care of the TOPO source.

Figure 5-17. TEM images of PbS nanostructures synthesized using 99% TOPO with different batch numbers purchased from Aldrich. (A) Batch number 00529CD, and (B) batch number 03008CE. Structure. The expected rock-salt structure of the PbS nanowires was confirmed by powder X-ray diffraction (XRD, Figure 5-18) and high-resolution transmission electron microscopy (HRTEM, Figure 5-19). The single-crystalline nature of the nanowires was clearly evidenced in the HRTEM image, in which the (200) lattice fringes were well resolved and exhibited the correct spacing of 0.292 nm.44 The fast-Fourier-transform

174

(FFT) pattern of the PbS nanowire lattice was indexed to the rock-salt structure (inset, Figure 5-19), and indicated a [200] growth direction for the nanowires.

- PbS Nanowires
3 CO

Bulk PbS (Rock Salt)

I uUJU^JU
WMP 111 200

220

311 222
I T I I I i i i r| 11 r | 11 i-i | i r f i | i

420

400

331,

422

I'lTi'mri i i jii i I ' I I j I r r | r i rr \ M I r J i

20 25 30 35 40 45 50 55 60 65 70 75 80 26(degree)

Figure 5-18. Powder X-ray diffraction (XRD) pattern of PbS nanowires. The standard pattern of the bulk PbS with rock salt structure (ICDD-PDF No. 01-072-4873) is also shown in the graph.

Figure 5-19. Lattice-resolved HRTEM image of a 15-nm diameter PbS nanowire. The inset shows the indexed fast-Fourier-transform (FFT) pattern of the image, indicating that the nanowire grows along [200] direction.

175

Energy-dispersive X-ray spectra (EDS) collected from the wires with attached catalyst nanoparticles (Figure 5-20) contained only Pb and S signals (note that the Pb signal at 2.48 keV overlaps with the S signal at 2.47 keV) from the wires, and only Bi signals from the catalyst nanoparticles, which is different from the EDS results for PbSe nanowires.

Cu

(A)

.2.

a
c
3 O
Bi Bi

Cu

yiJkJly^
0.000 2.000 4.000 6.000

4^, i i T t i^l"iii^4
10.000 12.000 14.000

8.000

Energy (keV)

Pb S

(B)
Cu

I
L^i|Af;.T ,,
0.000 2.000 4.000 6.000 8.000

Pb

P"

Cu

ii)toi^iiiiiiiii(iiiiiiypitii|ttiiiiii|iiii<M^Mipii1ii^i 10.000 12.000

jpyJ^i.bA|Mi^Ml|iii,iiUtLilll
14.000

Energy (keV)

Figure 5-20. Energy-dispersive X-ray spectra (EDS) recorded from the catalyst nanoparticle (A) and the wire (B). Near-infrared absorption spectra and quantum confinement effects. Strong

quantum confinement effects in the PbS nanowires were confirmed by near-infrared

176

absorption spectroscopy. Well-resolved excitonic features of a 10.2-nm and an 11.2-nm wire specimen were clearly shown in Figure 5-21. Each first excitonic feature was determined by a nonlinear least-squares fitting of each spectrum (Figure 5-22). The wire band gaps were assigned to the centers of the lowest-energy Gaussian peaks and summarized in Table 5-1. For the 10.2-nm and 11.2-nm PbS nanowires, the first excitonic features were found to be at 2016 nm and 2343 nm, corresponding to quantum confinement of AEg = 205 meV and AEg =119 meV, respectively. The increase of the quantum confinement energy with decreasing wire diameters indicates the conventional size-dependent quantum confinement effects. As expected, because the bulk exciton Bohr radius in PbS is quite large (20 nm), PbS nanowires having diameters of 10-11 nm are strongly confined.

I
1600 1800 2000 2200 2400 2600

Wavelength (nm)

Figure 5-21. Near-infrared absorption spectra of colloidal PbS nanowires of various diameters (d), dispersed in carbon tetrachloride, (a) d = 10.2 + 2.5 nm (+ 25%), and (b) d = 11.22.3nm(21%).

177

Unfortunately, we were not able to collect near-infrared absorption spectra from PbS nanowires having larger diameters so far because they tended to precipitate very rapidly during spectrum collection. This remains to be a challenging problem because high wire concentration is usually required for the near-infrared absorption measurement of colloidal PbS nanowires.
Model: ExpDcJQaus$

1.2X .

1.0T>. Chl-ZVOoF 0.00003 R'2 0.99683 *^Sv vo "^"v A0 1D 'VN^ X X _ Al *'*./S^ wl '-..^fcfc,^ Cl ''.. ^ % ^ Ai '"*., ^ ^ w _ *J ''"-,. ^**Nfca 0.03743 13.4B2S7 878.4907 25.9958 3B8-18822 1798.899 25.121S3 142,74837 18.43*T2 +0.O38T3 *2 04892 *44.88774 122 97688 *139.2S1IJ1 *S4.6M 12.62884 30,98841 i(2,flB519

\
''CV '.Xv

/ K \

W e l B M l n o :
CW-2COF 4.3TMB^ R"2 0 M993 y O A0 10 Al wl "*' A2 w2 Xc2 0.01059 13.86408 881.SS504 57.81098 J93.UT03T 1911,43977 28.B3BW 258.10454 2343.30381 0.0101 *0.MM4 * 17,21283 '4.76182 *10772T 11.38988 10.7087) 12.31282 *1.2e448

1.0u c n .0 O (0 .a
fll

0.80.6-

0.80.60.40.20.01400
'

'.^S^ '. ^ N ^ _ '. 7 N N K _ ^ s . ^ V ^V. ^St, X ^ N ^

<

0.40.20.0-

1600

'

1800

'

2000

'

2200

2400

1600

1800

2000

2200

2400

2600

Wavelength (nm)

Wavelength (nm)

Figure 5-22. Nonlinear least-squares fitting of the near-infrared absorption spectra (black solid curves) of PbS nanowires with diameters of 10.2 nm (a) and 11.2-nm (b). Fitted spectra (orange curves) are shown behind the corresponding experimental spectra. The fitting model is composed of one exponential decay function for the spectrum background (black dotted curve) and two Gaussian functions for the absorption features (red and green curves). The fitting parameters are also shown in this graph.

Table 5-1. Band gaps (Eg) and quantum confinement energies (AEg) of PbS nanowires of various diameters and bulk PbS. Diameter (nm) 10.2 2.5 ( 25%) 11.2+ 2.3 (+20%) <*>(bulk) 1 Excitonic peak position (nm) 2016 2343 3024 Eg (meV) 615 529 410 AEg (meV) 205 119

178

(2) CdS nanowires. Synthesis. CdS nanowires were also similarly grown from the single-source

precursor Cd(S2CNEt2)2, which has been previously used for the MOCVD growth of CdS thin films,40 the VLS growth of CdS nanowires,45 and the solution-based growth of CdS nanostructures.46'47 TOPO and Bi nanoparticles were used as solvent and catalysts, respectively.

Figure 5-23. Representative low-magnification TEM images of CdS nanowires of various diameters (d). (A) J = 7.8 1.2 nm ( 15%). This image at very low magnification shows that the lengths of the wires are typically several micrometers and the long wires tended to form bundles. (B) d = 6.5 + 0.9 nm (+ 14%), (C) d = 7.8 1.5 nm ( 19%), and (D) d = 11.0 1.8 nm (+ 16%). Representative low- and high-magnification TEM images (Figure 5-23 and Figure 524) show that this SLS strategy afforded high-quality CdS nanowires having the mean

179

diameters in the range of 6.5-11 nm, which are smaller than those synthesized based on the VLS method using the same precursor.45 The lengths of the wires were typically several micrometers and the long wires tended to form bundles (Figure 5-23A). The nanowires exhibit uniform diameters along their lengths. The standard deviations in the diameter distributions of the SLS-grown CdS nanowires were below 20 % of the mean diameters (Figure 5-25), which are narrower than the distributions achieved by the VLS method.45 Catalyst nanoparticles could be observed at one end of the wires, confirming the SLS wire growth (Figure 5-24D).

Figure 5-24. Representative high-magnification TEM images of CdS nanowires of various diameters (d). (A) d = 6.5 0.9 nm ( 14%), (B) d = 6.9 1.3 nm ( 19%), (C) d = 7.8 1.2 nm ( 15%), and (D) d = 11.0 1.8 nm ( 16%).

180

CdS nanowire diameter {run) 70eo50Counts

CdS nanowire diameter (nm)

| |

mill rf

.1 1 I

2010 08 10 12 14 16 CdS nanowire diameter (nm)

iajffpa|pp,-r.r.. r ,
6 7 B S 10 11 12 CdS nanowire diameter (nm) 13

Figure 5-25. Representative diameter-distribution histograms for CdS nanowires of various diameters (d). (A) d = 6.5 0.9 nm ( 14%), (B) d = 6.9 1.3 nm (+ 19%), (C) d = 7.8 + 1.2 nm (+ 15%), and (D) d = 11.0 1.8 nm ( 16%).

I,.^ .

r T

. r . , i i i | i i,i.i t .l,ii. T i.y, l | i i | , . . r ,.nj

300 400 500 600 700

Wavelength (nm)

Figure 5-26. TEM images and UV-Vis absorption spectra of CdS nanostructures synthesized using various amount of TDPA in TOPO. (A) TDPA/Cd = 0, (B) TDPA/Cd = 0.5, (C) TDPA/Cd = 1.0, and (D) TDPA/Cd = 2.0. (E) UV-Vis absorption spectra of the CdS samples shown in (A), (B), (C), and (D) and labelled as (a), (b), (c), and (d), respectively.

181

The amount of additional surfactant TDPA was similarly varied (Figure 5-26). The use of small amounts of TDPA in the synthesis helped to produce smaller-diameter nanowires. For example, the mean diameter of the wires was measured to be 8.5 nm when TDPA was not used; while 6.9-nm CdS nanowires were produced when the TDPA/Cd ratio was 0.5. However, the amount of TDPA was critical; TDPA/Cd ratios larger than 1 produced short wires and dots. Although the dots were too small to be seen in the TEM images, they were evident in the absorption spectra (Figure 5-26E). In contrast to Pb(S2CNEt2>2, Cd(S2CNEt2)2 can be readily dissolved in organic solvents such as TOP and oleylamine. All successful wire growth was conducted by injecting the TOP solution of Cd(S2CNEt2)2 into TOPO. However, only nanoparticles were produced (Figure 5-27) when the oleylamine solution of Cd(S2CNEt2)2 was injected into TOPO.

Figure 5-27. TEM images of CdS nanostructures synthesized by injecting an oleylamine solution of Cd(S2CNEt2)2 into TOPO. Structure. The wurtzite crystal structure of the CdS nanowires was confirmed by XRD (Figure 5-28) and by the FFT pattern from an HRTEM image (Figure 5-29, inset). The indexed FFT pattern confirmed the [002] wire growth direction. The well-resolved 182

lattice fringes perpendicular to the growth direction in the HRTEM image corresponded to a spacing of 0.330 nm, consistent with the (002) planes in hexagonal CdS.48

CdS Nanowires

3
JO,

Bulk CdS (Wurtztte) Bulk CdS (Zinc Blende)

110

103

112

200|2,1

202

i | i ' i i i | i i i ' i | i i 'i i | 'i 'i ' i i | i i I i

20

25

30

35

40

45

50

55

60

26 (degree) Figure 5-28. Powder X-ray diffraction (XRD) pattern of CdS nanowires. The standard patterns of the bulk CdS with wurtzite (ICDD-PDF No. 01-075-1545) and zinc blende (ICDD-PDF No. 01-080-0019) structures are also shown in the graph.

Figure 5-29. Lattice-resolved HRTEM image of a 9-nm diameter CdS nanowire. The inset shows the indexed fast-Fourier-transform (FFT) pattern of the image, indicating that the nanowire grows along the [002] direction.

183

(3) Synthesis of ZnS nanowires. Our attempts to grow ZnS nanowires using a similar single-source precursor Zn(S2CNEt2)2 were unsuccessful so far, although ZnS nanowires have been produced by the VLS mechanism using the same precursor.45 Nevertheless, we did observe short wires with catalyst nanoparticle tips in the TEM images (Figure 5-30), confirming that the short wires were indeed produced by the SLS mechanism. This result suggested that long ZnS nanowires might be potentially grown by the SLS mechanism in the future. JT- -jr.:

it
\

Figure 5-30. Representative TEM images of ZnS nanostructures synthesized using single-source precursor Zn(S2CNEt2)2 in TOPO. (4) PbTe nanowires. Synthesis. The SLS growth of PbTe nanowires was found to be extremely difficult when common dual-source precursors like lead carboxylate, lead alkylphosphonate, and TOPTe49"51 were employed. Although we attempted to optimize the SLS syntheses by varying synthetic conditions in a broad range, we were never able to observe a single wire under TEM. This synthetic difficulty should be due to the high reactivities of both lead and tellurium precursors, which have high tendency to form nanocrystals (mainly cubes, Figure 5-31). 184

Figure 5-31. TEM images of PbTe nanostructures synthesized using lead tetradecyphosphonate and TOPTe as precursors in TOPO.

We then sought an appropriate single-source precursor to overcome the synthetic challenge. Unfortunately, unlike the stable single-source precursors used for the SLS growth of metal sulfide nanowires, it's hard to find a stable, organometallic, singlesource precursor for PbTe. Yu reported a single-source precursor, cadmium tellurite (CdTeCh), for the SLS growth of CdTe nanowires from Bi catalyst nanoparitcles. Although the precursor is

purely inorganic, it was demonstrated that CdTe nanowires with acceptable quality could be produced. We then used a similar purely inorganic lead tellurite (PbTeC<3) as the singe-source precursor to grow PbTe nanowires. The reaction may be written as follows: 4 PbTe0 3 -> PbTe + 3 PbTe0 4 (4-1)

TEM images shown in Figure 5-32 represent our best PbTe wire sample, revealing that both wires and cubes were produced. The wires were typically short, having very
broad diameter and length distributions. The poor wire quality should be due to the low

solubility of PbTeC>3 in common organic solvents. Thus, it seemed impossible to produce high-quality PbTe nanowires using this purely-inorganic single-source precursor.

185

Importantly, we could occasionally observe a catalyst nanoparticle attached to the wire ends, indicating that the wires were indeed grown by the SLS mechanism. The results reported here show that it should be possible to grow well-controlled PbTe nanowires from Bi catalyst nanoparticles if appropriate molecular single-source precursor could be found in the future.

Figure 5-32. Representative TEM images of PbTe nanostructures synthesized using singe-source precursor PbTeC>3. The dotted red circle highlights a catalyst nanoparticle attached to a PbTe wire. Structure. We didn't collect XRD patterns of the PbTe nanowires due to the coexistence of nanostructures of other morphologies (mainly cubes). However, the rock salt structure of the wires should be expected because it is the only crystal structure under normal conditions. The single-crystalline nature of the PbTe wires is clearly shown in a representative HRTEM image (Figure 5-33). The fast-Fourier transform (FFT) pattern recorded from the well-resolved wire lattice image was indexed to the rock salt structure, indicating that the PbTe nanowires grow along the [200] direction. The well-resolved lattice fringes perpendicular to the growth direction in the HRTEM image corresponded to a spacing of 0.316 nm, consistent with the d-spacing between (200) planes in rock salt PbTe.53

186

Figure 5-33. HRTEM image of a 16-nm PbTe nanowire. The inset shows the FFT pattern indexed to rock salt structure, indicating the [200] wire growth direction.

Figure 5-34. HRTEM images of a ~15-nm PbTe nanowire recorded with prolonged exposure to the electron beam, showing a phase separation and recrystallization process of the catalyst nanoparticle.

187

As for the PbTe nanowires, we also directly observed a phase separation and recrystallization process of the catalyst nanoparticle when it was exposed to the electron beam in HRTEM for a prolonged time (Figure 5-34), indicating the alloy nature of the catalyst nanoparticles.

Discussion CdS nanowire syntheses and quantum confinement effects. Because the bulk Bohr radius of CdS is very small (2.8 nm), the quantum confinement studies are thus very challenging due to synthetic difficulties. Although our single-source precursor SLS

approach afforded high-quality CdS nanowires with smaller diameters than those achieved by the VLS approach,45 it's still very hard, if not impossible, to produce CdS nanowires with diameters equal to or less than the Bohr radius. So far, the smallest wire diameter that could be achieved was 6.5 nm, which is more than two times larger than the Bohr radius. Consequently, all the absorption spectra of our CdS nanowires looked similar. The efforts to grow smaller CdS nanowires suffered from the higher tendencies of agglomeration of small Bi nanoparticles at high reaction temperatures. Nevertheless, we indeed observed small blue shifts (~20 nm) of the band gaps in our high-quality CdS nanowires relatively to the bulk band gap of 500 nm, indicating weak quantum confinement in large-diameter wires. The results shown here are consistent with the band-gap-shift results in CdSe21 and CdTe nanowires (see Chapter 1) having diameters larger than the corresponding bulk Bohr radius. PbTe nanowire syntheses. It was reported that PbTe has a very large transverse Bohr radius of 152 nm,50 which makes it one of the ideal candidates to study quantum

188

confinement effects in relatively large nanostructures.

Unfortunately, it has proven

difficult to produce well-controlled nanowires so far due to the lack of appropriate precursors. Common dual-source precursors used for the synthesis of high-quality PbTe nanocrystals seemed to be too reactive under the synthetic conditions investigated, which favors other reaction pathways such as homogenous nucleation competing with SLS growth. The use of a purely inorganic single-source precursor PbTe03 was incompatible with the current SLS synthetic conditions because of its low solubility. As single-source molecular precursors are often employed for the synthesis of metal telluride thin films,54 we believe that the successful SLS growth of high-quality PbTe nanowires should rely on finding a suitable single-source precursor which can be readily dissolved in certain organic solvents. Arnold and coworkers reported the synthesis of lead tellurolate (Pb[TeSi(SiMe3)3]2), which was found to be stable under N2 at room temperature.55 Interestingly, this single-source precursor can be readily dissolved in toluene. Moreover, it decomposed cleanly at 250 C according to eq. 4-2. Pb[TeSi(SiMe3)3]2 - PbTe + Te[Si(SiMe3)3]2 (4-2)

Based on these experimental observations, Pb[TeSi(SiMe3)3]2 may be a potential candidate for the SLS growth of PbTe nanowires, which should be a direction for future studies.

Conclusion High-quality colloidal rock-salt PbS nanowires having diameters in the range of 9-31 nm and wurtzite CdS nanowires having diameters in the range of 6.5-11 nm were grown by the SLS mechanism using single-source-precursor metal diethyldithiocarbamates

189

[M(S2CNEt2>2, M = Pb, Cd]. These single-source precursors also serve as common CVD precursors.40'45 As single-source precursors have been extensively developed for use in the CVD growth of semiconductors, our results strongly suggest that the single-source SLS strategy may extend to other colloidal semiconductor nanowire systems that are not easily prepared by a dual-source precursor approach. The high quality of the PbS nanowires enabled us to observe well-resolved nearinfrared excitonic absorption features, confirming strong quantum confinement. These PbS nanowires are candidates for further fundamental studies, and for potential applications in solar cells2 and photodetectors.56 Compared with the VLS-grown CdS nanowires using the same single-source precursor, our SLS-grown colloidal CdS nanowires could be produced at much lower reaction temperatures (-500 C lower than that employed in the VLS growth). Additionally, smaller diameters and narrower diameter distributions were achieved for our SLS-grown wires.

Experimental Section Materials. Lead(II) nitrate (99.999%, Aldrich), cadmium chloride (99.995%, Strem), zinc nitrate hexahydrate (98%, Aldrich), sodium diethyldithiocarbamate trihydrate (ACS grade, Alfa Aesar), lead tellurite (purified, City chemical), phenyl ether (99%, Acros), trin-octylphosphine (TOP, 90%, Aldrich), tri--octylphosphine oxide (TOPO, 99%, Batch #00529CD, Aldrich), and n-tetradecylphosphonic acid (TDPA, PolyCarbon Industries) were purchased and used as received. Technical-grade TOPO (90%, Aldrich) was vacuum distilled before use. The solvents of water (HPLC grade), methanol (> 99.9%,

190

HPLC), hexanes (98.5+%, HPLC), acetone (> 99.9%, HPLC), toluene (99.9% HPLC), and carbon tetrachloride (> 99.9%, HPLC) were purchased from Aldrich and used as received. The single-source precursors M(S2CNEt2)2 (M = Pb, Cd, Zn) were prepared by literature methods.41'57 The Bi-nanoparticle stock solutions were prepared as previously described.21 All procedures for the synthesis of nanowires were conducted under dry, 0 2 free N2(g). Synthesis of Pb(S2CNEt2)2. Sodium diethyldithiocarbamate trihydrate (4.506 g, 20 mmol) was dissolved in 200 ml of water, followed by filtration to remove any insolubles. Then lead(II) nitrate (3.313 g, 10 mmol) was dissolved in 100 ml of water. This Pb2+ solution was added to the sodium diethyldithiocarbamate solution dropwise over 1 hour. The yellowish precipitate was then collected by filtration, washed thoroughly with water, and dried in a vacuum oven at room temperature overnight. The dried powder was stored in a glove box. Synthesis of Cd(S2CNEt2)2. Sodium diethyldithiocarbamate trihydrate (4.506 g, 20 mmol) was dissolved in 200 ml of water, followed by filtration to remove any insolubles. Then cadmium chloride (1.835 g, 10 mmol) was dissolved in 100 ml of water. This Cd2+ solution was added to the sodium diethyldithiocarbamate solution dropwise over 1 hour. The white precipitate was then collected by filtration, washed thoroughly with water, and dried in a vacuum oven at room temperature overnight. The dried powder was redissolved in chloroform and recrystallized from it as large transparent crystals, which was ground to powder and stored in a glove box.

191

Synthesis of Zn(S2CNEt2)2. Sodium diethyldithiocarbamate trihydrate (4.506 g, 20 mmol) was dissolved in 200 ml of water, followed by filtration to remove any insolubles. Then zinc nitrate hexahydrate (2.977 g, 10 mmol) was dissolved in 100 ml of water. This Zn2+ solution was added to the sodium diethyldithiocarbamate solution dropwise over 1 hour. The white precipitate was then collected by filtration, washed thoroughly with water, and dried in a vacuum oven at room temperature overnight. The dried powder was redissolved in chloroform and recrystallized from it as crystals, which was ground to powder and stored in a glove box. Synthesis of PbS nanowires. In a typical synthesis, TOPO (90%-purity grade, distilled, 4.0 g) was loaded into a Schlenk reaction tube with a small magnetic stir bar. Separately, Pb(S2CNEt2)2 (20 mg, 0.04 mmol) and phenyl ether (0.8 g) were combined in a small vial. In a separate small vial, a Bi-nanoparticle stock solution (15 mg) was combined with TOP (0.5 g). The reaction tube was then inserted in a pre-heated salt bath (NaN03/KNC>3, 46:54 by weight) at 250 C for at least 5 min to allow the temperature to equilibrate. The Bi-nanoparticle mixture was loaded into a 2-mL syringe. The

Pb(S2CNEt2)2 mixture was gently heated with a heat gun until a homogeneous solution was obtained (< 1 min). This hot solution was loaded into the syringe containing the Bi nanoparticles. The contents of the syringe were rapidly injected into the pre-heated, stirred TOPO. The reaction mixture turned black within 10s. After the reaction mixture was stirred for 5 min, the reaction tube was withdrawn from the salt bath and allowed to cool. While cooling, toluene (~5 mL) was added to prevent solidification. The conditions and results for various PbS nanowire syntheses are summarized in Table 5-2.

192

Table 5-2. Synthetic conditions and results for PbS nanowires and other morphologies.
Pb(S2CNEt2)2fl (mmol) 0.05 0.04 0.04 0.04 0.04 0.05 0.04 0.04 0.04 0.04 0.05 0.05 0.04 0.04 0.05 0.05 0.05
a

Surfactant6 (mmol) TDPA 0.025

--

TOPOc T t (C) (min) (g) 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 5.0d 5.0* 4.0 262 232 250 277 291 252 256 224 210 212 254 254 272 272 252 252 250 4 5 5 5 5 5 5 4 5 5 5 5 5 5 5 5 5

Bi stock solution dm (nm) (mg) 10.3 10.3 10.3 10.3 10.3 6.4 7.1 4.3 4.3 10.3 7.1 7.1 -10 -10 7.1 7.1

"wire

(nm) 8.7 11.0 11.2 13.3 13.2 10.2 16.1 31.2

~ ~ -

Notes wires wires wires wires wires wires wires wires wires and particles wires and particles rods cubes and particles wires and particles wires and particles wires and particles rods particles

TDPA 0.022 OA 0.19

-~

HDA 0.04 HDA 0.14 TDPA 0.05 TDPA 0.10 OA 0.27 OA 0.27
-

Pb(S2CNEt2)2 was dissolved in phenyl ether (0.6-0.8 g) by gently heating using a heat gun. b TDPA = n-tetradecylphosphonic acid; OA = oleic acid; HDA = -hexadecylamine. c Distilled 90%-TOPO unless specified otherwise. d Phenyl ether. e 1-octadecyldecene. f Minor population of 8.9-nm nanowires. 193

Synthesis of CdS nanowires. In a typical synthesis, TDPA (7 mg, 0.025 mmol) and TOPO (99%-purity grade, 4.0 g) were loaded into a Schlenk reaction tube with a small magnetic stir bar. Separately, Cd(S2CNEt2)2 (40 mg, 0.1 mmol) and TOP (0.8 g) were combined in a small vial, into which a Bi-nanoparticle stock solution (25 mg) was added. The Cd(S2CNEt2)2 slowly dissolved in the mixture to form a homogeneous solution. The reaction tube was then inserted in a pre-heated salt bath (NaNCVKNOs, 46:54 by weight) at 250 C for at least 5 min to allow the temperature to equilibrate. The contents of the vial were loaded into a 2-mL syringe, and rapidly injected into the preheated, stirred reaction tube. The reaction mixture turned a yellowish color within 10 s. After the reaction mixture was stirred for 5 min, it was withdrawn from the salt bath and allowed to cool. While cooling, toluene (~5 mL) was added to prevent solidification. The conditions and results for various CdS nanowire syntheses are summarized in Table 5-3. Synthesis of ZnS nanowires. TOPO (99%-purity grade, 4.0 g) were loaded into a Schlenk reaction tube with a small magnetic stir bar. Separately, Zn(S2CNEt2)2 (36 mg, 0.1 mmol) and TOP (0.8 g) were combined in a small vial, into which a Bi-nanoparticle stock solution (24 mg, 7.1-nm) was added. The Zn(S2CNEt2)2 dissolved in the mixture to form a homogeneous solution. The reaction tube was then inserted in a pre-heated salt bath (NaN03/KN03, 46:54 by weight) at 300 C for at least 5 min to allow the temperature to equilibrate. The contents of the vial were loaded into a 2-mL syringe, and rapidly injected into the preheated, stirred reaction tube. After the reaction mixture was stirred for 5 min, it was withdrawn from the salt bath and allowed to cool. While cooling, toluene (~5 mL) was added to prevent solidification.

194

Table 5-3. Synthetic conditions and results for CdS nanowires and other morphologies. Cd(S2CNEt2)2fl (mmol) 0.05 0.05 0.05 0.10 0.10 0.05 0.05 0.10 0.05 0.05 0.10 0.106
a

TDPA TOPOc (mmol) (g) 0.025 0.022 0.025 0.029 0.029 0.025
-

T (C) 255 240 258 250 256 268 258 243 258 258 250 265

Time (min) 5 5 5 5 5 5 5 5 5 5 5 5

Bi stock solution (mg) 25 25 25 25 25 20 25 25 25 25 50 50

"wire

(nm) 4.8 4.8 7.1 6.4 7.1 13.5 7.1 4.8 7.1 7.1 7.1 9.4

(nm) 6.5 6.9 6.9 7.8 7.8 7.9 8.5 11.0

-~

Notes wires wires wires wires wires wires wires wires wires and dots wires and dots wires and dots dots

4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0

0.025 0.050 0.10 0.20

-

Cd(S2CNEt2)2 was dissolved in tri-n-octylphosphine (TOP, 0.8 g) unless specified otherwise. b Cd(S2CNEt2)2 was dissolved in oleylamine (0.8 g). c 99% TOPO (Batch # 00529CD).

Synthesis of PbTe nanowires. Lead tellurite (50 mg, 0.13 mmol) and TOPO (90%purity grade, distilled, 4.0 g) were loaded into a Schlenk reaction tube with a small magnetic stir bar. Separately, Bi stock solution (50 mg, 7.1-nm) and TOP (1.6 g) were combined in a small vial. The reaction tube was then inserted in a pre-heated salt bath (NaN0 3 /KN0 3 , 46:54 by weight) at 280 C. When TOPO was dissolved upon heating, the contents of the vial were loaded into a 2-mL syringe, and rapidly injected into the

195

stirred reaction tube. The mixture gradually turned to black. After the reaction mixture was stirred for 5 min, it was withdrawn from the salt bath and allowed to cool. While cooling, toluene (~5 mL) was added to prevent solidification. Some white precipitate was found in the reaction mixture, which could be the undecomposed precursor. Then the mixture was centrifuged to remove the precipitate, and the upper solution was used for future isolation and characterization. Nanowire isolation and characterization. The nanowires were separated from the cooled reaction mixtures by addition of methanol (-10 mL). The precipitated wires were collected by centrifugation and the supernatant was decanted. The wires were then redispersed in hexane (1 mL) and reprecipitated by addition of acetone (~10 mL), followed by centrifugation and decantation. This process was repeated 2 - 3 times to remove the byproducts generated during the reactions. The purified wires were

redispersed in toluene for TEM sample preparation. However, for the near-IR study the purified wires were dried under a stream of N2 to remove volatile organics, and then redispersed in carbon tetrachloride. Carbon-coated Cu TEM grids were prepared by evaporating one drop of the NW toluene solutions on them. TEM images were collected using a JEOL 2000 FX microscope at 200 kV. The HRTEM images were collected using a JEOL JEM-2100F microscope at 200 kV. The XRD pattern was obtained using a Rigaku Dmax A vertical powder diffractometer with Cu Ka radiation (X = 1.5418 A). The near-IR spectrum was collected using a Perkin-Elmer Lamda 950 UV-vis-NIR spectrometer. Determination of the diameter distributions of MS (M = Pb, Cd) nanowires. For each sample, the diameter distributions were determined using several images at 500K

196

magnification to ensure adequate resolution for accurate measurement. Diameters were measured and recorded using Image-Pro Plus software (version 5.0 for Windows). Digital images were zoom expanded within the software. The software generated

diameter lists which were transferred to and manipulated in a standard spreadsheet program for statistical analysis. The distributions were generally constructed from at lease 300 diameter measurements for each specimen. Nonlinear least-squares fitting of the NIR absorption spectrum. The near-

infrared absorption spectrum was fit using a method reported previously.21'31 Briefly, the spectrum was fit with one exponential (for the background) and multiple Gaussian functions (for the absorption features) using Origin 7.5 software (www.OriginLab.com). The number of Gaussian functions used was determined by the number of absorption features present in the spectra. The fitting procedure yielded the center energy of the excitonic peaks and the error in the center energy. The wire band gaps were assigned to the centers of the lowest-energy Gaussian peaks.

References (1) (2) Madelung, O. Semiconductors: Data Handbook; 3rd ed.; Springer: Berlin, 2004. Schaller, R. D.; Sykora, M.; Pietryga, J. M.; Klimov, V. I. Nano Lett. 2006, 6, 424-429. Ellingson, R. J.; Beard, M. C ; Johnson, J. C ; Yu, P.; Micic, O. I.; Nozik, A. J.; Shabaev, A.; Efros, A. L. Nano Lett. 2005, 5, 865-871. Klimov, V. I. /. Phys. Chem. B 2006, 110, 16827-16845. Wise, F. W. Ace. Chem. Res. 2000, 33, 773-780.

(3) (4) (5)

197

Landolt-Bornstein, New Series, Group III, Vol. 17(b), Springer, Berlin, 1982. Huang, Y.; Duan, X.; Lieber, C. M. Small 2005, 1, 142-147. Hayden, O.; Payne, C. K. Angew. Chem. Int. Ed. 2005, 44, 1395-1398. Hayden, O.; Agarwal, R.; Lieber, C. M. Nature Mater. 2006, 5, 352-356. Duan, X.; Huang, Y.; Agarwal, R.; Lieber, C. M. Nature 2003, 421, 241-245. Gao, R; Lu, Q.; Liu, X.; Yan, Y.; Zhao, D. Nano Lett. 2001, 1, 743-748. Thiruvengadathan, R.; Regev, O. Chem. Mater. 2005, 17, 3281-3287. Fardy, M.; Hochbaum, A. I.; Goldberger, J.; Zhang, M. M.; Yang, P. Adv. Mater. 2007,79,3047-3051. Bierman, M. J.; Lau, Y. K. A.; Jin, S. Nano Lett. 2007, 7, 2907-2912. Ge, J.-P.; Wang, J.; Zhang, H.-X.; Wang, X.; Peng, Q.; Li, Y.-D. Chem. Eur. J. 2005, 11, 1889-1894. Tang, K.-b.; Qian, Y.-t.; Zeng, J.-h.; Yang, X.-g. Adv. Mater. 2003, 15, 448-450. Yong, K. T.; Sahoo, Y.; Choudhury, K. R.; Swihart, M. T.; Minter, J. R.; Prasad, P. N. Chem. Mater. 2006, 18, 5965-5972. Patla, I.; Acharya, S.; Zeiri, L.; Israelachvili, J.; Efrima, S.; Golan, Y. Nano Lett. 2007, 7, 1459-1462. Acharya, S.; Patla, I.; Kost, J.; Efrima, S.; Golan, Y. J. Am. Chem. Soc. 2006, 128, 9294-9295. Wang, R; Dong, A.; Sun, J.; Tang, R.; Yu, H.; Buhro, W. E. Inorg. Chem. 2006, 45,7511-7521. Yu, H.; Li, J.; Loomis, R. A.; Gibbons, P. C ; Wang, L. W.; Buhro, W. E. J. Am. Chem. Soc. 2003,125,16168-16169. Dong, A.; Wang, F.; Daulton, T. L.; Buhro, W. E. Nano Lett. 2007, 7, 1308-1313. Grebinski, J. W.; Hull, K. L.; Zhang, J.; Kosel, T. H.; Kuno, M. Chem. Mater. 2004, 16, 5260-5272. Kuno, M.; Ahmad, O.; Protasenko, V.; Bacinello, D.; Kosel, T. H. Chem. Mater. 2006, 18, 5722-5732.

198

Ouyang, L.; Maher, K. N.; Yu, C. L.; McCarty, J.; Park, H. J. Am. Chem. Soc. 2007, 129, 133-138. Fanfair, D. D.; Korgel, B. A. Chem. Mater. 2007, 19, 4943-4948. Trentler, T. J.; Hickman, K. M.; Goel, S. C ; Viano, A. M.; Gibbons, P. C ; Buhro, W. E. Science 1995,270, 1791-1794. Trentler, T. J.; Goel, S. C ; Hickman, K. M.; Viano, A. M.; Chiang, M. Y.; Beatty, A. M.; Gibbons, P. C ; Buhro, W. E. J. Am. Chem. Soc. 1997, 119, 2172-2181. Dingman, S. D.; Rath, N. P.; Markowitz, P. D.; Gibbons, P. C ; Buhro, W. E. Angew. Chem. 2000, 112, 1530-1532. Yu, H.; Buhro, W. E. Adv. Mater. 2003, 15, 416-419. Yu, H.; Li, J.; Loomis, R. A.; Wang, L.-W.; Buhro, W. E. Nature Mater. 2003, 2, 517-520. Fanfair, D. D.; Korgel, B. A. Cryst. Growth Des. 2005, 5, 1971-1976. Hull, K. L.; Grebinski, J. W.; Kosel, T. H.; Kuno, M. Chem. Mater. 2005, 17, 4416-4425. Wang, F.; Yu, H.; Li, J.; Hang, Q.; Zemlyanov, D.; Gibbons, P. C.; Wang, L. W.; Janes, D. B.; Buhro, W. E. J. Am. Chem. Soc. 2007,129,14327-14335. Cowley, A. H.; Jones, R. A. Angew. Chem. Int. Ed. Engl. 1989, 28,1208-1215. O'Brien, P.; Nomura, R. /. Mater. Chem. 1995, 5,1761-73. Bochmann, M. Chem. Vap. Deposition 1996, 2, 85-96. Neumayer, D. A.; Ekerdt, J. G. Chem. Mater. 1996, 8, 9-25. Gleizes, A. N. Chem. Vap. Deposition 2000, 6, 155-173. Fainer, N. I.; Kosinova, M. L.; Rumyantsev, Y. M.; Salman, E. G.; Kuznetsov, F. A. Thin Solid Films 1996, 280, 16-19. Trindade, T.; O'Brien, P.; Zhang, X.-M.; Motevalli, M. J. Mater. Chem. 1997, 7, 1011-1016. Lee, S. M.; Jun, Y. w.; Cho, S. N.; Cheon, J. J. Am. Chem. Soc. 2002, 124, 1124411245.

199

(43)

Fan, S.; Chapline, M. G.; Franklin, N. R.; Tombler, T. W.; Cassell, A. M.; Dai, H. Science 1999, 283, 512-514. ICDD-PDF No. 01-072-4873. Barrelet, C. J.; Wu, Y.; Bell, D. C ; Lieber, C. M. J. Am. Chem. Soc. 2003, 125, 11498-11499. Yan, P.; Xie, Y.; Qian, Y.; Liu, X. Chem. Comm. 1999,1293-1294. Zhang, Y. C ; Wang, G. Y.; Hu, X. Y. J. Alloys Comp. 2007, 437, 47-52. ICDD-PDF No. 01-075-1545. Lu, W.; Fang, J.; Stokes, K. L.; Lin, J. J. Am. Chem. Soc. 2004, 126, 1179811799. Murphy, J. E.; Beard, M. C ; Norman, A. G.; Ahrenkiel, S. P.; Johnson, J. C ; Yu, P.; Micic, O. I.; Ellingson, R. J.; Nozik, A. J. J. Am. Chem. Soc. 2006, 128, 32413247. Urban, J. J.; Talapin, D. V.; Shevchenko, E. V.; Murray, C. B. J. Am. Chem. Soc. 2006, 128, 3248-3255. Yu, H., PhD Dissertation, Washington University, 2003. ICDD-PDF No. 01-072-6645. Ritch, J. S.; Chivers, T.; Afzaal, M.; O'Brien, P. Chem. Soc. Rev. 2007, 36, 16221631. Seligson, A. L.; Arnold, J. J. Am. Chem. Soc. 1993, 115, 8214-8220. McDonald, S. A.; Konstantatos, G.; Zhang, S.; Cyr, P. W.; Klem, E. J. D.; Levina, L.; Sargent, E. H. Nature Mater. 2005, 4, 138-142. Kahn, O. F. Z.; O'Brien, P. Polyhedron 1991, 10, 325-32.

(44) (45)

(46) (47) (48) (49)

(50)

(51)

(52) (53) (54)

(55) (56)

(57)

200