University of Southern California Dissertations and Theses

Some studies of nanocrystal quantum dots on chemically functionalized substrates (semiconductors) for novel biological sensing

(USC Thesis Other)

Some studies of nanocrystal quantum dots on chemically functionalized substrates (semiconductors) for novel biological sensing

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content SOME STUDIES of NANOCRYSTAL QUANTUM DOTS ON CHEMICALLY FUNCTIONALIZED SUBSTRATES (SEMICONDUCTORS) FOR NOVEL BIOLOGICAL SENSING by Siyuan Lu A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHEN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (PHYSICS) December 2006 Copyright 2006 Siyuan Lu ii Dedication To my teachers iii Acknowledgements I am deeply grateful to all of my teachers for their contributions to my intellectual progress. Among these, the three teachers of the greatest influence and contribution are my parents, Shunbao Lu and Xiurong Cai, who devoted over a decade of their life to my intellectual development in early ages, and my advisor Prof. Anupam Madhukar who inspired me and guided my progress toward being a professional in science. Prof. Madhukar’s devotion to education and science will be a moral example throughout my future professional life. I am thankful to all the colleagues in our research group, present and past, including Prof. Zhonghui Chen, Prof. Atul Konkar, Mary Francis, Dr. Euitae Kim, Max Ho, Hooman Akhavan, Yi Zhang, Anubhuti Bansal and Tetsuya Asano. I am thankful to them not only because of what I learned from them during the course of this work, but also for their friendship. Without these colleagues, this dissertation work will not be possible. I would like to thank Prof. Mark Humayun for his support of part of this dissertation work through the Biomimetic Microelectronic System Engineering Research Center (BMES-ERC). Prof. Humayun’s vision and style have also inspired me. I am thankful to Prof. A. Paul Alivisatos (University of California, Berkeley) for the great opportunity of working in his lab. I have benefited enormously from his group’s knowledge and culture. Special thanks are due to Steven Hughes and Bryce Sadtler who provide much essential support for the nanocrystal synthesis work. iv I would like to thank Prof. Theodore Berger and his student, Walid Soussou, and Prof. Ram Datar and his student, Henry Lin, for productive collaboration. I would like to thank Prof. Robert Hellwarth, Prof. Hans Bozler, Prof. Gerd Bergmann, and Prof. Mark Humayun, who kindly served on my dissertation committee. Finally, I acknowledge with gratitude the DARPA/AFOSR funded DURINT (Defense University Research Initiative in Nanotechnology) program (Program manager Dr. Genot Pomrenke) and NSF funded Engineering Research Center on Biomimetic Microelectronic Systems at the USC for providing financial support during this dissertation work. v Table of Contents Dedication ii Acknowledgement iii List of Figures ix List of Tables xxi Abbreviations xxii Abstract xxvi Chapter 1. Introduction 1 §1.1 New Horizon Rising With Nanotechnology 1 §1.2 Nanocrystals: History, Weaknesses, and Strengths 4 §1.3 Envisioned Application of Nanocrystal in Biological Sensing 9 §1.4 Outline of the dissertation 16 Chapter 1 References 20 Chapter 2. Colloidal Nanocrystals Synthesis and Electronic Structure: 24 The Basics §2.1 Solution Route to Nanocrystal Synthesis 24 §2.1.1 Historical Perspective 24 §2.1.2 Experimental Apparatus 26 §2.1.3 Zeroth Order Consideration: Precursors, Surfactants and Solvents 28 §2.1.4 Theory of Nanocrystal Growth Kinetics 30 §2.1.4(a) Growth Rate 31 §2.1.4(b) Nucleation Rate 35 §2.1.5 Strategy for Nanocrystal Growth: Two Step growth Approach & Size 36 Focusing §2.1.5(a) Conventional two-step growth strategy 36 §2.1.5(b) Focusing” of the size distribution 37 §2.1.6 Considerations for III-V nanocrystal growth 38 §2.2 InAs core nanocrystal synthesis 40 §2.2.1 Dehalosilylation Reaction 40 §2.2.2 Precursor and Stock Solution Preparation 40 §2.2.3 Real time PL monitoring of the InAs nanocrystal growth 42 §2.2.4 InAs growth procedure using the conventional two step growth 44 §2.2.5 Size Selective precipitation 45 §2.2.6 InAs nanocrystal growth results and discussions 47 vi §2.2.7 Enhancing Nanocrystal Average Diameter: 52 The “Punctuated” Growth Approach. §2.3 InAs/ZnSe shell/core nanocrystals. 55 §2.3.1 Introduction to core/shell nanocrystals 55 §2.3.2 Stock Solution for ZnSe shell 56 §2.3.3 InAs core Nanocrystal Concentration and Amount of Precursor Material 56 §2.3.4 ZnSe Shell Growth Procedure 57 §2.3.5 ZnSe Shell Growth Results and Discussion 58 §2.4 Electronic Structure of Nanocrystal Quantum Dots 62 §2.4.1 Weak and Strong Confinement Regimes 62 §2.4.2 Particle-in-a-Box model 66 §2.4.3 p k r r ⋅ calculation of the nanocrystal electronic structure 70 §2.4.3(a) The p k r r ⋅ framework 72 §2.4.3(b) Envelop Wavefunction Approximation and p k r r ⋅ in Heterostructures. 77 §2.4.3(c) The 8 band p k r r ⋅ formalism for spherical quantum dots and 80 calculation of InAs electronic structure. Chapter 2 References 92 Chapter 3. NCQDs for Cell Labeling, NSOM based Simultaneous 98 Cell Morphology and Optical Studies §3.1 Introduction 98 §3.2 Integrated Conventional/Near-Field Optical Microscopy 101 and Spectroscopy. §3.2.1 Overall Instrumentation and Far-Field Microscopy 101 §3.3 NSOM Instrumentation and experimental consideration 108 §3.3.1 Introduction to the NSOM technique 108 §3.3.2 NSOM Instrumentation 112 §3.3.3 NSOM Principle and Experimental Consideration 116 §3.3.3(a) Basic Theory of Near Field Optics and Generation 116 of Near Field Exciation §3.3.3(b) Experimental Consideration in NSOM measurement 127 §3.4 Simultaneous NSOM imaging of cancer cell morphology 139 and surface receptor distribution. §3.4.1 Her2/neu Surface Receptor and its role in cancer 139 §3.4.2 Nanocrystal Quantum Dot Labeling of Her2/neu Receptor 141 §3.4.3 NSOM based simultaneous morphological and optical imaging of cells 144 §3.4.4 Future direction — NIR QDs as fluorescent labels 150 Chapter 3 References 152 Chapter 4. Solid (Semiconductor) Surface Modification 156 and Adsorption of NCQDs §4.1 Near-surface quantum nanostructure and GaAs surface passivation 157 vii §4.1.1 Brief investigation of optical property of near-surface 157 quantum structure §4.1.2 Sulfur based GaAs surface passivation 163 §4.2 Adsorption of NCQDs on unmodified substrates 167 via controlled evaporation: dip-coating technique §4.2.1 Growth of ordered colloidal particle array 170 §4.2.2 Initiation of ordered colloidal particle array 177 §4.2.3 Dip-coating summary and future development 179 §4.3 SAM surface modification and SAM mediated adsorption 180 of small objects (nanocrystals, proteins) §4.3.1 SAM functionalization for Semiconductors 181 §4.3.1(a) Alkylsiloxane SAM functionalization of silicon 183 §4.3.1(b) Alkylthiol SAM functionalization of GaAs (001) 185 §4.3.2 SAM mediated nanocrystal adsorption 187 Chapter 4 References 192 Chapter 5. Bioconjugated SAM mediated cell adhesion on solid 197 surfaces via specific ligand-receptor interaction §5.1 SAM mediated protein adsorption 198 §5.2 Bioconjugated SAM (BSAM) mediated Cell adhesion 202 Chapter 5 References 216 Chapter 6. Hybrid Colloidal/Epitaxial Nanostructures 218 – A Study of Excitation Transfer §6.1 Photoluminescence of Nanocrystal Quantum Dots 220 §6.1.1 Overview of Photoluminescence Process 220 §6.1.1(a) Inhomogeneous Broadening Dominates the PL Width 222 §6.1.1(b) Relaxation in QDs 223 §6.1.1(c) PL Intermittence and “dark” charged QDs 229 §6.1.2 Photoluminescence Spectroscopy Methods 235 §6.2 Photoluminescence instrumentation 238 §6.2.1 Pre-existing Time-Integrated PL/PLE Setup 239 §6.2.2 The New PL/PLE/TRPL Setup: Overview 240 §6.2.2(a) Time integrated PL/PLE in the new setup 240 §6.2.2(b) TRPL in the new setup: overview and design consideration 243 §6.2.3 TRPL instrumentation: Optics 246 §6.2.4 TRPL instrumentation: Time correlated single photon 250 counting (TC/SPC) electronics §6.2.5 Time Resolution and Instrument Response Function 254 §6.2.6 TRPL data processing and deconvolution 256 §6.2.7 TRPL Future Upgrade 260 §6.3 Excitation transfer in NCQD/Semiconductor Hybrid structures 260 §6.3.1 Overview of the excitation transfer process. 262 §6.3.1(a) Förster Resonant Energy Transfer (FRET) 264 viii §6.3.1(b) Tunneling 266 §6.3.2 Sample Preparation and Optical Measurements in 268 Excitation Transfer Studies §6.3.2(a) Sample Preparation 268 §6.3.2(b) Optical Measurements 271 §6.3.3 Inter-NCQD energy transfer 272 §6.3.4 Excitation Transfer between NCQDs and Substrate 279 §6.3.4(a) Observation of excitation transfer 279 §6.3.4(b) Onset of PL Drop and Electronic Structure of 282 InAs/ZnSe on GaAs §6.3.4(c) Hole transfer from InAs/ZnSe NCQD to GaAs substrate 286 §6.3.4(d) Connection to QD PL Intermittence and Estimation of 290 Transfer Rate §6.3.4(e) PLE temperature dependence and detection energy dependence 294 §6.3.4(f) Hole transfer from GaAs substrate to the InAs/ZnSe NCQDs 299 §6.3.4(g) Conclusions 301 Chapter 6 References 305 Chapter 7. Conclusion and outlook 312 §7.1 Nanocrystal synthesis and electronic structure. (Chapter 2 summary) 312 §7.2 Quantum Dot based Cell labeling and Simultaneous 315 AFM/NSOM imaging. (Chapter 3 summary) §7.3 Semiconductor surface “passivation”/modification and 317 nanocrystal adsorption on untreated/modified semiconductor surfaces. (Chapter 4 summary) §7.4 Bioconjugated SAM mediated cell adhesion on solid 319 surfaces via specific ligand-receptor interaction (Chapter 5 summary) §7.5 Excitation transfer in nanocrystal quantum dot/ 322 crystalline semiconductor substrate hybrid structures (Chapter 6 summary) §7.6 Outlook – Nanocrystal heterostructure for functional 326 biological coupling Chapter 7 References 338 Bibliography 339 Appendix A: Synthesis of (TMS) 3 As 361 Appendix B: Angular part of the envelop wavefunction for 368 electron and hole states in spherical quantum dots. Appendix C: Breast Cancer Cell Line MDA-MB231 and 370 SK-BR-3 Cell Culture Protocol Appendix D: Rat Hippocampal Neurons Culture Protocols 374 Appendix E : CdSe Rods and Au/CdSe nanocrystal metal/semiconductor 379 junction synthesis. ix List of Figures Figure 1.1 Illustration of two categories of hybrid integrated colloidal/epitaxial nanostructures: (a) nanocrystals buried in crystalline semiconductor via epitaxial overgrowth to serve either as active quantum structures or as passive stressors for inducing vertically-aligned growth of epitaxial island SAQDs; (b) nanocrystals adsorbed on solid substrate containing near-surface quantum structures. 9 Figure 1.2 Schematic of nanocrystal quantum dot based labeling or probing of cells: (a) the living cell is adhered on a solid substrate with non-specific binding; (b) the target cell is selectively adhered on solid surface functionalized with ligand molecules that specifically bind to receptors on the target cell membrane. 10 Figure 1.3 Schematic of biological sensing using integrated colloidal/epitaxial hybrid nanostructure. 14 Figure 2.1 A typical reaction apparatus for nanocrystal synthesis 27 Figure 2.2 Diffusion controlled growth rate of nanocrystal vs. radius. r: radius of nanocrystal; r* critical radius; 34 Figure 2.3 The room temperature absorption peak and 1 st exciton PL peak wavelength vs. diameter calibrated for InAs nanocrystals. Data adopted from [2.22]. 43 Figure 2.4 (a) PL of InAs nanocrystal aliquots during conventional growth approach. Doted line: InAs nanocrystal grown to ~3.8nm average diameter (PL peak 1070nm) before one more injection of precursor. Solid line: after one more injection of precursor; note the merging of PL peak at 960nm indicating that new nucleation had occurred. (b) PL of InAs nanocrystal aliquots during punctuated growth approach. Dashed line: InAs nanocrystal of ~3.9nm diameter before injection of precursors. Solid line: immediately after one more injection of precursor, nanocrystals grow uniformly to ~5nm (PL peak 1250nm) diameter with little new nucleation occurring. 48 x Figure 2.5 (a) Size distribution statistics of largest InAs nanocrystals separated after a conventional two-phase growth as measured from TEM images of nanocrystals on C-coated grid. (b) Corresponding illustrative TEM image (courtesy of S. Hughes). (c) Size distribution statistics of largest InAs nanocrystals separated after a punctuated growth as measured from TEM images (d) Corresponding illustrative TEM image (courtesy of Steven Hughes, University of California, Berkeley). 49 Figure 2.6 Real time aliquots PL result during a conventional two-step growth as described in the text. Timing for precursor injections, TOP injections and PL measurements are marked on the right. 51 Figure 2.7 PL from size-selective-precipitated InAs nanocrystal samples; typical PL FWHM 100-150nm. The corresponding FWHM for nanocrystal size distribution is 20%-25%. 53 Figure 2.8 (a) InAs/ZnSe Aliquots PL evolution during growth as a function of calculated shell thickness (b) Integrated PL intensity vs. shell thickness (c) PL peak position vs. shell thickness. 60 Figure 2.9 TEM of InAs/ZnSe nanocrystals on carbon-coated grids which shows the InAs/ZnSe nanocrystal shape evolution as a function of ZnSe shell thickness: (a) 1.82ML (b)1.95ML (c) 2.47ML. Note the presence of dendrimer formation at 2.47ML shell thickness! In all three cases the same InAs core nanocrystals (~4nm diameter) are used as starting material. The same book keeping of shell thickness is followed as described in §2.3.3(TEM images courtesy of Atul Konkar, University of Southern California, and Steven Hughes, University of California, Berkeley). 61 Figure 2.10 (a) Schematic of a nanocrystal quantum dots structure. (b) Schematic of particle-in-a-box confinement potential arising from the alignment of the conduction and valence band edges of the core (A) and shell (B). The electron and hole energy states and wavefunction are schematically shown corresponding to the case of infinite confinement potential. 63 Figure 2.11 Schematic of the band structure near k=0 in a direct bandgap semiconductor of zinc-blende structure. 75 Figure 2.12(a) Calculated energy levels of states in InAs nanocrystal as a function of its radius. Solid line: energy of states of total angular momentum F=3/2, Dotted line: energy of states of total angular momentum F=1/2. 87 xi Figure 2.12 Continued (b): Same plot as figure 2.12(a) except zoomed into InAs nanocrystal radius range 2.0-4.5nm for clarity. 88 Figure 2.13 8 band p k r r ⋅ calculated InAs nanocrystals optical transition energy as a function of diameter and comparison to experimental data from Ref 2.65. 90 Figure 3.1 Schematic of the custom-built integrated conventional/near- field optical microscopy and spectroscopy setup. 103 Figure 3.2 Photograph of the custom build integrated conventional/near- field optical microscopy and spectroscopy setup. 104 Figure 3.3 Epi-fluorescent imaging of SKBR3 breast cancer cells with their Her2/neu surface receptors specifically labeled by CdSe/ZnS NCQDs (600nm emission). Panel (a), (b), (c) are images taken with the focal plane respectively at the top, middle and bottom (closest to the glass substrate) of the cells. Note in the three images only a fraction of the NCQDs are in good focus. Panel (d) shows a magnified value of a NCQD in focus, each pixel in this image corresponds to 107nm. A cross-section of the intensity profile of the NCQD is shown in panel (e). The Gaussian fitting of the intensity profile gives a width of 400nm demonstrating the ~ λ/2 diffraction limited resolution of this fluorescent image. 107 Figure 3.4 (a) SEM image of a tapered fiber tip. (b) Higher resolution SEM image showing the end of the tapered fiber coated with metal except for the aperture (c) Schematic of the metal coated fiber tip (d) optical image of the cantilevered fiber tip. (Image adapted from Nanonics product brochure). 111 Figure 3.5 Schematic of the NSOM setup based on Nanonics NSOM 100 head. 113 Figure 3.6 Schematic of the subwavelength aperture diffraction problem. 118 Figure 3.7 Normalized angular spectrum (Equ. 3.10) of a plane wave passing through a circular aperture of diameter 2a. λ=532nm. The white and gray areas are respectively the far-field and the near-field spectral components. 123 Figure 3.8 Calculated near field to far field energy ratio as a function of aperture diameter. 124 xii Figure 3.9 Diffracted light power density near an aperture of radius a=λ/150 (calculated). The light is polarized along x axis. See figure 3.6 for the definition of the coordinates. Taken from [3.28] 126 Figure 3.10 An NSOM image of single 600nm CdSe/ZnS QDs sparsely dispersed on a glass substrate. 133 Figure 3.11 Reduction of the quantum yield of CdSe QDs as a function of distance to an aluminum surface calculated using CPS theory. The geometry of the system and the parameters used are shown in the figure inset. Note the dipole of the CdSe QD is taken to be isotropic in the plane perpendicular to its c-axis. 135 Figure 3.12 Schematic showing three QDs respectively at the center (QD3) and closer to the edge of the aperture and the metallic coating (QD1). Hence QD1 will suffer a greater reduction in its quantum yield (QY). This quenching of the QY will effectively improve the resolution of NSOM imaging. 136 Figure 3.13: Solid line schematically represents the cross-section of a noisy image of an object. The two dotted lines define the upper and lower limit of this noisy cross-section. Note, the uncertainty in determining the location of this object on the image is proportional to both the noise level (or precisely inverse signal-to-noise-ratio) and the resolution of the image (FWHM of the cross-section). 138 Figure 3.14 (Top) Schematic of Her2/neu point mutation and dimerization. (Bottom) Intracelluar signalling pathway initiated by activation of growth factor receptor (including Her2/neu) that leads to cell growth. Figure adapted from [3.41] 140 Figure 3.15 Schematic of QD labeling of Her2/neu surface receptors on breast cancer cell SK-BR-3 and MDA-MB-231. 142 Figure 3.16 An illustration of the typical procedure for NSOM imaging. (a) Under conventional bright field microscopy the NSOM tip was brought to the area of interest on a 605nm QD labeled SKBR3 cell. (b) A fluorescence NSOM image of the selected area indicated by the white box in image a, obtained using 200nm diameter tip. (c) Higher resolution NSOM image zoomed into the selected area indicated by the white box in image b. (d) With the same tip stopped at the indicated position on image c, the fluorescent spectrum measured using excitation from the NSOM tip (488nm single mode Ar+ laser coupled). The broad peak in the spectrum corresponds to emission from multiple QDs, the two narrow peaks at 615nm and 625nm are the residual Ar + laser line. 145 xiii Figure 3.17 Simultaneously obtained AFM and fluorescent NSOM images of QD (CdSe/ZnS 605nm) labeled SKBR3 cells (panel a, b) and MDA cells (panel c, d). Image obtained with tip of 100nm diameter aperture. Insets (XSec1) and (XSec2) show two cross-sections on the NSOM image (panel b) to illustrate the smallest features on the image. The resolution of the image as demonstrated by the FWHM of such feature is <150nm. 147 Figure 3.18 (Upper) Fluorescence spectrum of a SKBR3 cell with no labeling. (Lower) Fluorescence spectrum of a QD labeled SKBR3 cell. The peak at 605nm corresponds to the emission from 605nm emission CdSe/ZnS QDs. Both spectra were taken at room temperature with 488nm Ar + laser excitation (far-field, through back port of IX71). The 500- 1000nm spectral range was measured with a Si CCD detector, and the 1000-1500nm spectral range was measured with an InGaAs array detector. Note the low fluorescence background from the SKBR3 cell in the NIR region (900-1500nm). 151 Figure 4.1 Schematic of biological sensing using integrated colloidal/epitaxial hybrid nanostructure. 157 Figure 4.2 Schematic of pyramid InAs SAQD with GaAs capping. 159 Figure 4.3 (a) PL of 2.5ML InAs SAQD with 170ML, 105ML, 74ML, 40ML GaAs capping (at 78K). Corresponding PL Peak Intensity (b), Peak Position(c), FWHM (d). 161 Figure 4.4 PL of as prepared 2.5ML InAs SAQD with 40ML GaAs capping (black) and after ferritin adsorption (red). 163 Figure 4.5 AFM image of Sulfur-passivated GaAs surface. (RMS roughness 0.15nm ) 165 Figure 4.6 (a) PL of untreated (black) and sulfur passivated (red) S-I GaAs substrate. After passivation, integrated PL intensity was enhanced 34 times. (b) PL of untreated (black) and sulfur passivated (red) near surface InAs SAQD sample (2.5ML InAs with 40ML GaAs capping). After passivation, integrated PL intensity from SAQD was enhanced 11 times. 166 Figure 4.7 Schematic of the apparatus for dip-coating. 169 Figure 4.8 Schematic depicts the meniscus of the suspension near the contact line where the leading edge of array grows by incorporating particles in the suspension. Figure adapted from [4.23]. 170 xiv Figure 4.9 Schematic of two spheres partially immersed in a liquid layer on a horizontal solid substrate. The deformation of the liquid meniscus gives rise to inter-particle capillary attraction. 173 Figure 4.10 AFM image of ordered, one-monolayer-high, 5nm diameter CdSe/TOPO nanocrystal array deposited on GaAs substrate using a dip- coating technique. 175 Figure 4.11 AFM images of 5nm diameter CdSe nanocrystals dip-coated on GaAs. Substrate drawn speed fixed at 2 μm/sec. Panels (a), (b), (c) correspond to nanocrystal toluene solution concentration 0.25mg/ml, 0.125mg/ml, 0.05mg/ml, respectively. The resulting nanocrystal coverages are, respectively, ~100%, 50%, 30% and scale linearly with the solution concentrations. 176 Figure 4.12 Schematic showing the initiation of the nanocrystal array on the surface. (a) the particles enter the flat film region due to solvent convective flow or Brownian motion (b) if the particles are of diameter on the order of but less than the film thickness, the partially immersed particles will experience lateral attractive capillary force (c) Initial ordered array formation due to capillary force. 178 Figure 4.13: Schematic of generic type of self-assembled monolayer” (a) SiCl 3 (CH 2 ) n R SAM on SiO 2 /Si. (b) HS(CH 2 ) n R SAM on GaAs. Here R represents the functional group on the free end of the molecule which could be: CH 3 -; HO-; HS-, NH 3 -, HO(CH 2 CH 2 O) n CH 2 - ……; (c) Structure of the four SAM molecules involved in our discussion: octadecyl-tricholorosilane (OTS); 3-bromopropyl-tricholoro-silane (BPTS); 1-octadecanethiol (ODT); 1, 6-Hexanedithiol (HDT) 182 Figure 4.14 AFM image of (a) octadecyl-tricholorosiloxane (ODS) SAM functionalized Si/SiO 2 surface. (RMS roughness: 0.16nm) (b) 3- bromopropyl-tricholoro-siloxane (BPTS) SAM functionalized Si/SiO 2 surface. (RMS roughness: 0.24nm); 185 Figure 4.15 AFM image of (a) 1-octadecanethiol (ODT) (b) 1, 6- hexanedithiol (HDT) SAM functionalized GaAs(001) surface. The RMS roughness for the two SAM functionalized surface are respectively: 0.28nm, 0.41nm. 186 Figure 4.16: (Left) AFM images of (a) CdSe (b) InAs NCQD adsorbed on HDT modified GaAs surface (Right) Cross-section analysis of the line as indicated in the corresponding AFM images. 189 xv Figure 4.17 Schematic of NCQD/SAM/Substrate hybrid structure. 191 Figure 4.18 PL of CdSe NC on Linked to HDT modified GaAs. Crresponding AFM image of the hybrid structure is shown in figure 4.16(a). 191 Figure 5.1 A scheme for biological sensing involving (a) selective adhesion of target cells on surface functionalized by peptide ligand based SAM, and (b) subsequent QD labeling for detection. 198 Figure 5.2 AFM image of ferritin protein adsorbed on ODS modified Si/SiO 2 surface. 199 Figure 5.3 AFM image of 3-bromopropyl-tricholoro-siloxane (BPTS) SAM functionalized Si/SiO 2 surface. (RMS roughness: 0.24nm); 200 Figure 5.4: AFM images show mutant Chaperonin adsorption on (a) bare Si surface (b) BPTS modified Si. Inset: higher resolution AFM image shows that disc-shaped ~30nm diameter features could be few-mers formed by aggregation of 3 or 4 donut-shaped chaperonins. 201 Figure 5.5 Schematic of conducting electrodes on nonconducting substrates employed in prosthesis for electrical stimulation of cells and tissue. 204 Figure 5.6. (a) Schematic of cell binding to the substrate via specific membrane receptor interaction with corresponding peptide ligands coated on the substrate. Note the undulation in the substrate surface is to depict a lateral scale (~200nm) of typical variation in the extra-cellular matrix topology and is on a length scale more than an order of magnitude smaller than the size of the cell (b) Magnified view of a laterally nanoscale (<20nm) region to reveal significance of the surface vertical position (called surface roughness) in relation to the employed bilinker length. 206 Figure 5.7: (a) Schematic illustrating the chemistry of the peptide IKVAV and KHIFSDDSSE immobilization onto glass substrates. (b) Schematic showing amino acids in IKVAV and KHIFSDDSSE peptide. 207 Figure 5.8: Optical fluorescence (left column) and AFM (right column) images of plain glass (top); APTS grafted glass treated with FITC dye (middle); and APTS grafted glass reacted with TAMRA labeled peptides ligands (bottom). 211 xvi Figure 5.9: Fluorescence microscopy images of neurons cultured on glass substrate functionalized as marked. Neuron nuclei were stained with propidium iodide dye for fluorescent imaging. 212 Figure 5.10: Neuron nuclei surface coverage histogram for glass substrate functionalized with APTS, PDL, IKVAV, and KHIFSDDSSE. 213 Figure 6.1 Schematic of sensing biological agents using integrated colloidal/epitaxial hybrid nanostructures chips. 218 Figure 6.2 Schematic showing the 3-step model of the PL process: (1) photon generation, (2) carrier relaxation, (3) radiative recombination. 221 Figure 6.3 Calculated electronic structure of InAs nanocrystal (diameter 5.2nm). Details of calculation are provided in chapter 2. 224 Figure 6.4 Transient absorption (TA) spectrum of CdSe/TOPO NCQD showing sub-ps electron relaxation to its 1S (LUMO) state. Data taken from [6.6] 225 Figure 6.5 Schematic showing the mechanism of carrier energy relaxation in nanocrystal quantum dots: (a) electron relaxation via Auger energy transfer to hole (b) hole relaxation via consecutive phonon emission. 227 Figure 6.6 Dynamics of ‘‘hot’’ PL detected at different spectral energies in CdSe NCQD (R=1.8nm, HOMO-LUMO PL emission 2.16eV). This measurement shows that the hole energy relax gradually from the high excited states to the HOMO level through a ladder of intermediate states, as depicted in figure 6.5(b). Data Taken from [6.14]. 228 Figure 6.7 Time trace of photoluminescence from a single ~29 Å radius CdSe QD. (Taken from 6.20). 230 Figure 6.8 (a) Schematic showing the capture and release of lone charge in a NCQD that cause the NCQD photoluminescence to be turned “off” and “on”. Schematic adapted from [6.23 Shimizu 2004] (b) Schematic showing the Auger energy transfer from an exciton to a lone carrier in the NCQD that quenches the photoluminescence of a charged NCQD. 231 Figure 6.9 Schematic of the new PL/PLE/TRPL setup. (This figure does not represent the actual layout of the optical components.) 241 Figure 6.10 Trigger circuit based on Hammatsu 5973 fast photodiode. 249 xvii Figure 6.11 The block diagram of the TC/SPC electronics using the inverted photon counting mode. 251 Figure 6.12 The timing configuration of TC/SPC in the inverted photon counting mode as being implemented in the electronics shown in figure 6.11 253 Figure 6.13 Measured instrument response function (IRF) of the TRPL setup. The FWHM, rise time and decay time of the IRF are, respectively, 27, 11 and 38ps. 255 Figure 6.14 Example of TRPL data deconvolution. Black dots show the measured TRPL response from a 2.0ML InAs SAQD sample (RG970221- 01-H) at 6K. Dotted line shows the IRF of the measurement (89ps FWHM). Green and red curves show, respectively, the deconvoluted TRPL response and the reconvoluted fitting to the measured data. The deconvoluted PL response of the 2.0ML InAs SAQD has a rise time of 87ps and a decay time of 705ps. 259 Figure 6.15: (a) Schematic of InAs/ZnSe core/shell nanocrystal quantum dots in direct contact with a GaAs substrate. (b) Shows the unknown nature of the relative alignment ( e V Δ and h V Δ ) of the electron and hole energies of the NCQDs and the substrate band edges. A common vacuum level and the simple electron-affinity (or photothreshold) rule from semiconductor heterojunctions is used to draw figure 6.16(b). Χ sub and Χ NCQD stand for, respectively, the electron-affinity of the substrate and the NCQD. 261 Figure 6.16 Room temperature photoluminescence of InAs/ZnSe nanocrystals in toluene solution. 270 Figure 6.17: Tapping mode AFM image of InAs/ZnSe NCQDs deposited on sulfur passivated GaAs (001) via dip-coating. 270 Figure 6.18 Schematic showing sample structure employed for inter-dot transfer study. 272 Figure 6.19 (a) Room temperature PL of InAs/ZnSe nanocrystals dip- coated on glass substrate. (b) Time resolved PL spectra detected at two energies 1.02 eV(black) and 1.13eV(red), corresponding to emission, respectively, from larger (~5nm diameter) and smaller (~4nm diameter) InAs/ZnSe nanocrystals. The single exponential fitted luminescence lifetimes are 3.58ns and 1.82ns, respectively. 274 xviii Figure 6.20 Measured temperature dependence of the luminescence lifetime detected at 1.02eV (black square) and 1.13eV (red square). Black Crosses show the luminescence lifetime of the 1.13eV emission nanocrystals estimated using the calculated FRET time constant and the assumption that non-FRET related recombination time of the smaller (1.13eV emission) nanocrystals is the same as the lager (1.02eV emission) nanocrystals. (See text for details). This estimation fits reasonably with the observed luminescence lifetime of the 1.13eV emission nanocrystals (red square). 275 Figure 6.21 Schematic showing a future direction of colloidal/epitaxial hybrid structure made of vertically aligned near surface SAQDs and adsorbed NCQDs. FRET coupling between the NCQDs and SAQDs can be exploited to establish communication link between the two. 278 Figure 6.22 Room temperature PL of NCQDs on different substrates for excitation at (a) below (1.38eV) and (b) above (1.476eV) GaAs band gap (1.42eV). In both panels, NCQDs on glass (black); NCQDs passivated GaAs (red). 281 Figure 6.23 Room temperature PLE of NCQDs deposited on: Glass (black), passivated GaAs (red), The detection is at 1.02eV (NCQD PL peak). Energies the of the most relevant optical transitions in the NCQDs are marked on the PLE data for reference and ease of discussion in the text. 283 Figure 6.24 Energy level diagram of InAs/ZnSe NCQDs on GaAs substrate drawn for a relative positioning of the substrate band edge and NCQD HOMO-LUMO energies consistent with the findings (see text for details). 284 Figure 6.25 TRPL measurement of NCQDs on passivated GaAs for excitation at 1.55eV (800nm black) and 1.37eV (900nm red) respectively above and below the GaAs bandgap. Detection is at 1.12eV (1100nm). Room temperature 287 Figure 6.26 Schematic depicting quenching of NCQD PL due to hole transfer to the substrate. Note that after the hole transfer, the NCQD remains “dark” due to the presence of the lone electron. 289 xix Figure 6.27 (a) PLE spectra of NCQDs on passivated GaAs substrate (as normalized to that on glass) at various temperatures. Detection is at the NCQD PL peak 1.02eV (which is temperature insensitive). (b) Corresponding PLE drop “onset” energies (red squares) and the inflexion point (blue squares) as a function of temperature; Solid curve: GaAs bandgap; Dotted red curve: GaAs bandgap-46meV, the best fit for PLE drop “onset” energies; Dotted blue curve: GaAs bandgap -30meV, the best fit for the PLE inflexion points. 295 Figure 6.28 (a) Room temperature PLE spectra of NCQDs on passivated GaAs (as normalized to that on glass) at various detection energies. (b) Excitation energies at the PLE inflexion points that mark the drop of each PLE curve are plotted as a function of detection energy. 297 Figure 6.29 The calculated energy diagram of InAs/ZnSe nanocrystals of 4.6nm, 5.2nm and 5.6nm core diameter on GaAs substrate. Emission energies of these three sizes correspond to the detection energies in the PLE measurements shown in figure 6.28(a) (1.13eV, 1.02eV and 0.94eV). On this plot the LUMO (1Se) levels of the nanocrystals of different sizes are assumed to be aligned in order to explain the observation (figure 6.28) that the “onset” excitation energy for PL reduction is independent of the detection energy. 298 Figure 6.30 Schematic depicting the quenching of NCQD PL due to hole transfer from a substrate to a NCQDs. Such hole transfer causes the NCQD to have a lone hole which quenches the PL of the NCQD. 300 Figure 6.31 Schematic of NCQD/SAM/Substrate hybrid structure. 304 Figure 7.1 Schematic of nanocrystal semiconductor/metal junction. 328 Figure 7.2 schematic band structure of a semiconductor/metal junction (Schottky junction) and its photovoltaic effect. 329 Figure 7.3 Illustrative TEM images ((a) low resolution, (b) high resolution) of as-synthesized Au/CdSe nanocrystal metal/semiconductor heterojunctions. (Courtesy of Bryce Sadtler and Steven Hughes, University of California, Berkeley) 331 Figure 7.4 Panel (a) (panel (b)) shows the rise (fall) of birefringence upon application (removal) of external electrical field of the following samples: CdSe rod (black), Au/CdSe NCHJ with light excitation off (blue) and light excitation on (red). 333 xx Figure 7.5 (a) Schematic of nanocrystal S/M junction embedded neuron membrane and (b) corresponding equivalent circuit diagram. 334 xxi List of Tables Table 2.1 Basic parameters for common III-V semiconductors (300K) 64 Table 2.2 Notation and zone center Bloch function of “most significant” bands. 76 Table 2.3: 0K p k r r ⋅ parameters for III-V semiconductors 84 Table 3.1 Nanonics NSOM tip nominal throughput as a function of aperture diameter. 130 xxii Abbreviations AFM: Atomic Force Microscopy APD: Avalanche Photodiode APTS: amino propyl triethoxysilane BSA: Bovine Serum Albumin BSAM: Bioconjugated Self Assembled Monolayer CFD: Constant Fraction Discriminator cw: Continuous Wave DI water: Deionized water DME: ethylene glycol dimethylether DOS: Density of States ECM: Extracelluar Matrix EDC: 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide EGFR: Epidermal Growth Factor Receptor FITC: fluorescein isothiocyanate FLN: Fluorescent Line Narrowing FRET: Föster Resonant Energy Transfer FT: Fourier Transformation HOMO: Highest Occupied Molecular Orbital HV: High Voltage IRF: Instrument Response Function LN2: Liquid Nitrogen xxiii LUMO: Lowest Unoccupied Molecular Orbital MBE: Molecular Beam Epitaxy MEE: Migration Enhance Epitaxy MCP-PMT: Microchannel Plate Photomultiplier Tube ML: Mode-Lock N.A.: Numerical Aperture NC: Nanocrystal NCQD: Nanocrystal Quantum Dot NIR: Near Infrared NSOM: Near Field Scanning Optical Microscope NSQD: Near Surface Quantum Dot NSQW: Near Surface Quantum Well ODS: octadecyltrichlorosiloxane OTS: octadecyltrichlorosilane PBS: Phosphate Buffered Saline PD: Photodiode PDL: Poly-D-Lysine PL: Photoluminescence PLE: Photoluminescence Excitation PMT: Photomultiplier Tube PR: Photoresist PSPD: Position Sensitive Photo Detector xxiv PV: Photovoltaic QD: Quantum Dot QE: Quantum Efficiency QW: Quantum Well RT: Room Temperature RTK: Receptor Tyrosine Kinase SAM: Self-Assembled Monolayer SAQD: Self-Assembled Quantum Dot S/M: Semiconductor/Metal SNR: Signal-to-Noise Ratio SS: Surface State TAC: Time to pulse Amplitude Convertor TAMRA: tetramethyl rhodamine TCE: Trichloroethylene TC/SPC: Time-correlated Single Photon Counting TEB: Transient Electrical Birefringence TEM: Transmission Electron Microscope TIRF: Total Internal Reflection Fluorescence Microscopy Ti:S: Titanium Sapphire TOP: Trioctylphosphine. TOPO: Trioctylphosphine oxide TRPL: Time-Resolved Photoluminescence xxv T.T.S: Transit Time Spread WD: Working Distance. xxvi Abstract Integration of epitaxial and colloidal semiconductor nanostructures into hybrid structures can potentially open unprecedented functionalities that combine the strengths of the epitaxial nanostructures as manifest in optoelectronics with the versatility of the colloidal nanocrystal and their application in solution environment. We envision that such hybrid structures can be a promising platform for chip-based, high-sensitivity detection of early disease and biohazard. This dissertation is devoted to several different research grounds that contribute to this envisioned application of colloidal/epitaxial hybrid structure. In the class of nanocrystal quantum dots (NCQDs), we investigated the synthesis of InAs based NCQDs that emit in the near infrared and thus their compatibility with the existing optoelectronics technology and significance for high- sensitivity biological detection owing to the significantly reduced autofluorescence of cells in this wavelength regime. A unique optical system was set up which enables simultaneous atomic- force-microscope and near-field-scanning-optical-microscope imaging of the NCQDs and of the morphology and fluorescence of the NCQD labeled cells. The setup is applied to examine the morphology of breast cancer cells and the distribution of their surface receptors Her2/neu. Solid surface functionalization using self-assembling molecules (SAMs) with appropriate end groups (linkers) was investigated. Hybrid structures of nanocrystals on semiconductor substrates without or with SAM bilinkers were constructed. xxvii Substrate surface functionalization with peptide-ligand conjugated SAM is used to achieve selective neuronal cell adhesion on prosthetic surfaces. As the first step towards signal transduction from a NCQD to an underlying substrate, excitation transfer in hybrid structures composed of nanocrystals in direct contact with a crystalline semiconductor substrate was studied using systematic photoluminescence (PL), PL-excitation and time-resolved PL spectroscopies. The study revealed that at even very low excitation power a unique type of “hot” charge transfer occurs from excited hole states in InAs/ZnSe NCQDs to interfacial states induced by the NCQD adsorption on the GaAs substrate. Such charge transfer may serve to establish the necessary communication between the NCQDs and substrates for chip based sensing of weak signals as expected for the earliest detection of disease or hazard signifying entities. 1 Chapter 1. Introduction §1.1 New Horizon Rising With Nanotechnology First Day Salv:. … Finally, we may say that, for every machine and structure, whether artificial or natural, there is set a necessary limit beyond which neither art nor nature can pass; it is here understood, of course, that the material is the same and the proportion preserved. -- Dialogues Concerning Two New Sciences (Galileo Galilei) As quoted above, Galileo started his book Dialogue Concerning Two New Science [1.1] with an argument concerning the property of machines made of constant material and geometry but with varying size. I guess that all great thinkers of that age would be amazed to see the size of artificial structures achieved 400 years later. Nevertheless it was certainly recognized even then that the available material and technology at any time limits the possible size of the artificial structures which in turn limits their applications. Indeed the size of artificial structures as a signature and the pride of technological achievements, from the great Pyramid to the Sears Tower, has always been pursued with enthusiasm by mankind. The latest excitement comes, however, at the opposite end of the spectrum: the ultra small artificial structures, pursued under the name of “nanotechnology”. For the first time, mankind has started to systematically investigate and acquire the capability to do controlled synthesis and characterization of materials and structures on the nanometer scale. Nanotechnology has opened a new horizon of research on properties of material structures of reduced sizes and their applications. For the majority of the last century, the research on condensed matter had been largely focused, without choice, 2 on the properties of bulk materials and their derivation from the atomistic nature of the material as largely limited by the available synthesis and characterization technology. Generally, each property of a material has its characteristic length on the order of 10-100nm, below which the property of material becomes, other than a function of its structure and composition, also a function of its size and shape. Some of the most beautiful cases of such finite size effect are found in semiconductor or metallic crystals on the order of 10nm size (nanocrystal), including their size- dependent electronic structure [1.2] , phase diagram [1.3], catalytic effect [1.4] etc. The finite-size properties of nanocrystals are exploited for improved or unprecedented performance of various applications spanning from optical (e.g. light emitting device [1.5], solar energy conversion [1.6]), chemical (catalysis [1.4]), to biology (fluorescent labeling [1.7, 1.8]). Other than the opportunity in fundamental studies, nanotechnology has in a sense brought us closer to the ultimate machinery of nature –life. We recognize that all great engineering achievements pale in front of the complexity and delicacy of the simplest form of life which is machines built upon the coordinated function of molecular components (proteins and ribosomes) on the order of 10nm size. For the first time mankind can build artificial structures of size comparable to those comprising life, so is it the time for artificially engineered material/structures to be applied in biology and medicine and interface with life at the fundamental molecular level? Looking back at the involvement of man-made structures in medicine, one recognized that for the longest time we can repair body function only at an organ 3 level using simple mechanics based artificial organs (bone, lung or heart etc.). Only recently, with the advancement of microelectronics technology, prosthetic devices (~10-100 μm) are now interfacing with the body at the level of single to a few cells. The latest developments are the retinal or cortical prostheses using micro-electrodes for electric stimulation of the neuronal cells. [1.9, 1.10] Now, with the arrival of nanotechnology, we can envision that in the future we may build the tools to probe and repair body function at a subcelluar and molecular level. We can also envision that the development of such tools will eventually help reveal the molecular nature of diseases. This will revolutionize medicine just like X-rays or endoscopy did 100 years ago. In this respect, semiconductor or metal colloidal nanocrystal structures seem to have an important role. Since the demonstration of the biological application of nanocrystals quantum dots [1.7, 1.8], the field has developed rapidly. As a super fluorophore, nanocrystal quantum dots have been applied to probe many intracellular processes [1.11], and are being investigated for ultra-sensitive disease detection [1.12]. Given the interesting properties and applications of colloidal nanocrystal (hereafter referred to as nanocrystals [1.13]) structures, especially at the exciting front of biological or medical application of nanotechnology as discussed above, we started a journey on its investigation some years ago. Part of excitement is summarized in this writing of the dissertation. 4 §1.2 Nanocrystals: History, Weaknesses, and Strengths Historically, research on nanocrystals dates back nearly 150 years to Faraday’s investigations of the optical properties of 3nm Gold Sol [1.14]. In fact, semiconductor nanocrystals (CdS x Se 1-x ) have been grown in glass matrix by Corning (USA) Corp. and Schott Corp. (Germany) for applications in photochromic glass since 1960’s. In the last two decades, attention has been redrawn to the field of semiconductor nanocrystals partly because of the pursuit of nanotechnology, and partly because their synthesis has been finally perfected. The two major developments are: (1) organometallic based solution route for nanocrystals [1.15] synthesis, (2) synthesis of core/shell nanocrystal structures [1.16]. These developments lead to high quality (high crystallinity, narrow size distribution, well surface passivated) nanocrystals whose intrinsic electric and optical properties related to its finite size are finally no longer masked by structural defects or surface imperfections. Semiconductor nanocrystals, because of their 3D confinement of carrier, are referred to as nanocrystal quantum dot (NCQDs). These NCQDs has near perfect fluorescent efficiency (>50% for common CdSe/ZnS NCQDs) [1.17] and inspired many prospective applications. The generally-held high expectations for applications of NCQDs currently fall in two categories: (a) in optoelectronic devices, and (b) as biological probes. Although the electronic structures and optical properties of nanocrystals have been understood quite well by now, no competitive optoelectronic device has been made 5 of NCQDs so far. The only few meaningful optoelectronic applications of NCQDs include hybrid organic/nanocrystal LEDs [1.5], wavelength converter [1.18], or sensitizer in low cost solar cells [1.6]. Stimulated emission from nanocrystals (dispersed in polymer matrix) has been demonstrated some years ago [1.19], but no meaningful nanocrystal laser has been achieved yet. In contrast the other class of quantum dots – epitaxially grown self-assembled quantum dots (SAQDs) [1.20, 1.21] is much more successful in optoelectronics application. SAQDs are well optimized and exploited to realize advanced devices such as ultra low threshold and high speed lasers [1.22], near to long wavelength infra-red detectors [1.23], solid state amplifiers, etc [1.24]. The inferior performance NCQDs in optoelectronic applications compared to SAQDs is quite understandable. The fault is not necessarily on the NCQD. Being synthesized by solution chemistry, NCQDs are conventionally only compatible with a polymer matrix to construct optoelectronic devices. The intrinsic electrical performance (such as conductivity) of the polymer matrix, the degree of control of the matrix and its interfering with NCQDs are nowhere close to what can be achieved with the crystalline semiconductor matrix in which the SAQDs are epitaxially grown. We recognize that if the NCQDs can be integrated coherently into a crystalline semiconductor matrix, one may completely change the role of NCQD in optoelectronics applications [1.25]. Indeed research on such integration of NCQD into epitaxial growth is an important direction for future research. 6 The fact that NCQDs are solution synthesized is however a great advantage if the application of NCQDs is to interact with a foreign environment, such as in the case of biological labeling (imaging) and probing [1.11]. Compared to conventional organic fluorescent dyes traditionally used for biological and medical applications, the commonly recognized advantages of NCQDs are resistance to photo-bleaching; broad excitation wavelength (single excitation wavelength for NCQDs having different emission wavelengths) etc. [1.7, 1.8]. Here we note two other advantages of NCQDs: (a) the flexibility in the NCQD design and manipulation, and (b) the similarity of NCQDs to epitaxial semiconductor nanostructures structurally and electronically. NCQDs have been grown of almost all semiconductor materials, the well studied ones include: CdSe, CdTe, PbS, PbSe, InP, InAs etc. The growth mechanism of NCQDs is reasonably understood [1.26]. Via appropriate growth control, the size of NCQDs is freely tailored providing almost continuously tunable electronic structures. Different shapes, such as rod, arrow head, tetrapods, also have been achieved [1.27]. More recently nanocrystal heterostructures involving different material combinations also have emerged [1.28, 1.29]. Some of the synthesis and characterization work of nanocrystal heterostructures done by this author is reported in §7.6. The electrical and optical properties of these NCQDs (including heterostructures) are also relatively simple. The surface functionalization of most type nanocrystals (with biological molecules) is of similar kind of chemistry. The 7 surface functionalization is also typically decoupled from the intrinsic property of the NCQDs. In comparison, there is huge library of organic molecules for different applications. However, tailoring of their structure can be challenging from a molecular synthesis view point. The prediction of the property and function, given the molecular structure is also difficult. Such problems of organic molecules pretty much have to be solved case-by-case. The difficulties are extremely severe in complex macro-molecules such as proteins. Predicted tailoring of their structure and function is currently impossible. As an example of this deficiency of molecules, we note that there are no effective organic dyes emitting in the near-infrared (1.1-1.5 μm), even though these can be very important for biological labeling and disease detection given the low autofluorescence (i.e. background) of cells and tissue at such wavelengths (see chapter 3). Yet nanocrystals emitting at this range of IR wavelength are easily available using narrow bandgap material such as InAs based nanocrystals. The other important advantage of colloidal nanocrystals is their similarity to the epitaxial quantum structures (quantum wells, wires and dots) and the possibility for integration. Semiconductor nanocrystals and the epitaxial quantum structures, if constructed using the same material system, will have similar lattice structure, electronic structure, and close alignment of their bandedge (or HOMO and LUMO levels for QDs). Therefore structural integration or electronic coupling between the two should be easier than, say, between organic molecules and the epitaxial quantum 8 structures. As the Madhukar group has been a pioneering contributor to the field of epitaxial quantum structures, we were particularly interested in the integration of the two independent fields of colloidal and epitaxial nanostructures. Many unprecedented functions have been envisioned [1.30] in hybrid structures integrating colloidal nanocrystals and epitaxial nanostructures (referred hereafter as hybrid colloidal/epitaxial nanostructures). One of our proposals [1.30] is to bury nanocrystals into a crystalline semiconductor matrix via epitaxial overgrowth to create active nanostructures, as schematically shown in figure 1.1(a) [1.25]. Given the flexibility of nanocrystal size, shape control and narrow size distribution, such integration can create more sophisticated nanostructures and electronic responses than can be achieved with epitaxial growth alone. The buried nanocrystals also may serve as seeds (stressors) for subsequent induced surface stress directed vertically aligned growth of ordered epitaxial island self-assembled quantum dot arrays (figure 1.1(a))[1.31]. Other than the buried nanocrystals, a different possibility of integration is the nanocrystal adsorbed on semiconductor surface with near-surface quantum structures buried underneath as schematically shown in figure 1.1(b). Such nanocrystals may serve as a communication link between the external environment and a solid-state chip for biological sensing or actuating application. Potential for realizing this type of integration is one of the main topics of investigation in this dissertation as discussed next. 9 Figure 1.1 Illustration of two categories of hybrid integrated colloidal/epitaxial nanostructures: (a) nanocrystals buried in crystalline semiconductor via epitaxial overgrowth to serve either as active quantum structures or as passive stressors for inducing vertically-aligned growth of epitaxial island SAQDs; (b) nanocrystals adsorbed on solid substrate containing near-surface quantum structures. §1.3 Envisioned Application of Nanocrystal in Biological Sensing. Given the above analysis of the strengths and the weaknesses of nanocrystals, we judge the most promising applications of nanocrystals are indeed in the solution environments encountered in biological studies such as labeling cells for imaging and sensing. The applications of nanocrystals can be even more fruitful if they can be combined with the epitaxial nanostructure based optoelectronics technology. We note that the usually conceived scheme of nanocrystal based labeling or probing of cells are as illustrated in figure 1.2(a) and (b). Conventionally the living cell to be probed is adhered on a solid substrate (such as glass slide) with non- specific (i.e. via electrostatic force etc.) binding (figure 1.2(a)). In order to enhance the detectivity of target cells, beyond the conventional implementation, we conceive 10 Figure 1.2 Schematic of nanocrystal quantum dot based labeling or probing of cells: (a) the living cell is adhered on a solid substrate with non-specific binding; (b) the target cell is selectively adhered on solid surface functionalized with ligand molecules that specifically bind to receptors on the target cell membrane. 11 that selective adhesion of target cells on solid surface can be achieved by surface functionalization with ligand molecules that specifically bind to the receptors on the target cell membrane (figure 1.2(b), discussed in chapter 5). In either case, the adhesion to a solid substrate (either nonspecific or specific) is typically required for the survival of most live cells other than metastatic cancer and blood cells. Nanocrystal quantum dots, surface functionalized with specific type of ligands, are attached to their targeted receptors (often genetically refereed to “biomarkers”) on the cell surfaces and in within the cell body. The relatively stable and strong fluorescence from nanocrystals (compared to a dye fluorophore) is advantageous in ultrasensitive detection of biomarkers at the single molecule level. Such ultrasensitivity is particularly valuable for early disease diagnosis based on protein biomarker detection since proteins, unlike DNAs, can not be amplified. The actual detection of the nanocrystals is typically via fluorescence microscopy, which reveals the presence and spatial distribution of the nanocrystals (hence their targeted molecules). Ultrasensitive detection at a single nanocrystal level requires fairly sophisticated optical microscopy facilities. While the above scheme of nanocrystal based biological probing is perfectly adequate to probe cellular processes at a laboratory level, since some related works are to be exponded in this dissertation, we note that the current trend is miniaturized biological detection system as demanded by economic, easy-to-use, and early-stage detection of disease or biohazard. Miniaturization typically means the realization of the sensing on a single chip without involving a laboratory. Avoiding use separate 12 reagents (reagentless) in the detection process is also a part of miniaturization. Some of the currently proposed and investigated miniaturized bio-sensing schemes include using ligand-functionalized micro-cantilevers [1.32] and ligand-functionalized nanowires [1.33, 1.34], etc.. These micro-cantilevers and nanowires are monolithically built into microelectronic chips. These devices are expected to have high sensitivity due to their relatively small size and large surface-to-volume ratio. On the aspect of miniaturization, even though nanocrystals have been very successfully applied as fluorescent labels in many laboratories for single molecule biological sensing, a chip-level nanocrystal-based biosensor has not been proposed. Compared to cantilevers ( μm size) or nanotubes (typically >10nm diameter and a few μm long), nanocrystals should offer truly single molecule detection sensitivity given their much smaller size (<10nm, comparable to protein sizes). Although nanocrystals are capable of interacting with biological environment, the hindrance is the connection between nanocrystals and the solid-state chip. Despite vast investigations on solution application of semiconductor nanocrystal quantum dots, fundamental studies on the interaction between such nanocrystals and solid substrates are virtually non-existent. However as we argued in the preceding, the similarity of nanocrystals in their structure and electric properties to the epitaxial quantum nanostructure is a strength of nanocrystals and the coupling between the nanocrystals and the mature epitaxial quantum structure based optoelectronics should be worth serious investigation. Indeed, this dissertation presents, to our knowledge, the first such systematic study. Successful realization of electronic communication 13 between NCQDs & epitaxial nanostructures will bring a major advancement to the biological sensing using nanocrystals. • The Proposition underlying this Dissertation We propose to realize chip based nanocrystals biosensing using an integrated hybrid colloidal/epitaxial nanostructure as schematically illustrated in figure 1.3. We note figure 1.3 is not a device schematic of any kind, but only an illustration of all the necessary components involved and thus the research bases that need to be investigated. In the envisioned application of the nanocrystal (in this section the nanocrystal refers to in general nanocrystal based nanostructures including nanocrystal heterostructures), nanocrystal functionalized with specific ligands (e.g. an antibody) molecules binds to the corresponding receptors (e.g. an antigen) on the specific type of cells to be detected via specific ligand-receptor binding. The nanocrystal structures needs to be appropriately designed to alter its optical properties upon such ligand-receptor binding. One example of such alteration is the recently investigated quenching of nanocrystal fluorescence via resonant energy transfer between a pair of nanocrystal and dye conjugated via biological molecules such as proteins and DNA [1.35, 1.36]. More generally the effect of environment on nanocrystal optical properties is found in the heavy dependence of nanocrystal optical response on its surfactant molecules [1.37]. 14 Figure 1.3 Schematic of biological sensing using integrated colloidal/epitaxial hybrid nanostructure. On the other end the nanocrystal is linked to and coupled to a solid state substrate containing near surface quantum nanostructures. We do not have a predetermined mechanism of coupling, some of the possible mechanism of coupling includes: charge or energy transfer between the nanocrystals and the substrate, the effect of nanocrystal induced fields on the substrate etc.. Nevertheless the purpose of the coupling is always to transduce the alteration of the nanocrystal property (signal) upon binding with target biological entity into the substrate, where such a weak signal can be amplified and processed. For the actual work exploring the electronic coupling in colloidal/epitaxial hybrid structure, we started with the material system 15 of InAs based nanocrystals and GaAs/InAs based epitaxial quantum structure, given a combination of the following considerations: the importance of GaAs/InAs based III-V epitaxial structure in optoelectronics, our group’s traditional strength in that field, as well as the importance of 1.1-1.5 μm wavelength regime provided by this material system to the future of biological detection (details argued in chapter 3). Clearly the actual realization of the proposed scheme of nanocrystal for biological detection as conceptually depicted above requires many knowledge bases to be covered. For the conventional nanocrystals based cell probing as illustrated in figure 1.2 the required bases are: (1) nanocrystal synthesis and ligand functionalization; (2) cell labeling using ligand functionalized nanocrystals via specific ligand- receptor binding; (3) optical microscopy and spectroscopy of nanocrystal labeled cells; (4) selective and non-selective cell adhesion on solid substrate. The proposed miniaturized biological sensors using integrated colloidal nanocrystal and epitaxial nanostructures (figure 1.3) requires the following additional bases to be covered: (5) understanding of nanocrystal electronic energy levels and their alignment with energy levels of the epitaxial quantum nanostructure contained in the substrate; 16 (6) construction of colloidal/epitaxial hybrid structure. This topic includes semiconductor surface modification and passivation, and nanocrystal absorption on substrate surface. (7) electronic interaction between nanocrystals and the substrate. This dissertation describes this author’s work as a part of our group’s initial investigation of the above noted bases. From the beginning we realize that reaching the end objective of miniaturized ultrasensitive biological detection using the proposed hybrid colloidal/epitaxial nanostructure involves a large research scope and demands long term, cooperative research effort. However, many of the component studies by themselves are challenging and ground-breaking. The journey towards the end objective will be fruitful enough and beneficial to many other applications as well. §1.4 Outline of the dissertation The dissertation is organized to cover the 7 bases noted above in the following way: Chapter 2 covers base (1) and base (5) – the nanocrystal synthesis and their electronic structure. Given our interest in the integration of the nanocrystals with the III-V compound semiconductor based epitaxial nanostructures and our interest in cell labeling in the infrared wavelength regime, we focus on the small bandgap (~0.4eV) InAs based nanocrystals, instead of the dominant large bandgap (~1.5eV) II-VI nanocrystals (such as CdSe). We present the procedures and results of InAs and 17 InAs/ZnSe core/shell nanocrystal synthesis. The control of nanocrystal growth kinetics to achieve narrow nanocrystal distribution is discussed. This understanding of growth kinetics led this author to introduce a non-conventional punctuated growth approach to achieve large InAs (>8nm diameter) nanocrystals. Also in chapter 2, we describe the nanocrystal electronic structure, first qualitatively using simple “particle-in-a-box” model, and then quantitatively using the p k r r ⋅ and envelop wavefunction framework. The electronic structure of spherical InAs nanocrystals calculated using eight band p k r r ⋅ method is presented and compared to the experimentally observed optical transitions. Chapter 3 covers base (2) and base (3) – cell labeling with surface functionalized nanocrystals and optical microscopy and spectroscopy of labeled cells. We introduce a unique integrated far-field and near-field optical microscopy and spectroscopy setup built as part of this dissertation work. This setup enables simultaneous atomic-force microscope (AFM) and near-field scanning optical microscope (NSOM) imaging of the morphology and fluorescence of the quantum dot labeled cells. The principle of near-field scanning optical microscopy (NSOM) and experimental considerations are discussed. In a clinically relevant case study, the combined AFM/NSOM is used to image the breast cancer cells SK-BR-3 and MDA- MB-231 with their cancer-inducing surface receptors labeled by nanocrystal quantum dots. The procedure and considerations for quantum dot labeling are presented. The results of simultaneous AFM/NSOM imaging are presented and discussed. 18 Chapter 4 covers base (6) -- construction of colloidal/epitaxial hybrid structures. In §4.1, we discuss an intrinsic challenge that the epitaxial quantum nanostructures face as a part of the hybrid structure – the degradation of optical performance due to surface recombination of excited carriers. Then we present results of our efforts to ease such challenge via sulfur based “surface passivation”. In §4.2, we discuss the simplest construction of colloidal/epitaxial hybrid structure based on deposition of nanocrystals on unmodified substrates using a refined evaporative assembly technique – dip-coating. The mechanism of dip-coating is presented, emphasizing its potential to create ordered array of nano-object on surfaces. In §4.2 we discuss the generic strategy to chemically modify the semiconductor surface via covalently grafted self-assembled monolayer (SAM). We present the procedure and result for SAM functionalization as well as the use of SAMs as bilinkers to direct chemisorption of nanocrystals on surface to create nanocrystal/SAM/substrate hybrid structures. Chapter 5 covers base (4) – cell adhesion on solid surfaces. We expand the application of SAM based surface modification to control the adsorption of biological entities. We first discuss SAM induced non-specific and specific protein adsorption on semiconductor surface. Then we discuss surface modification using peptide ligand conjugated SAM that we called Bio-SAM (BSAM), followed by some experimental results on BSAM induced specific cell adhesion. The specific case studied is hippocampal neuronal cell adhesion on ceramic (modified glass) substrates which is important problem in biomedical engineering relevant to the surface coating 19 of cortical prosthetic devices. Studies on systematic characterization of BSAM modified surface, as well as subsequent cell adhesion result are presented and discussed. Chapter 6 covers base (7), the electronic interaction between nanocrystals and underlying semiconductor substrate. We discuss the optical characterization of nanocrystals adsorbed on crystalline semiconductor substrates and report the discovery of excitation transfer between the nanocrystal and the substrate. A photoluminescence (PL) setup which includes PL, PL excitation, and time-resolved PL spectroscopy capability for this characterization work, built and test as a part of this dissertation work, is introduced in some detail. Systematic PL/PLE/TRPL studies of InAs/ZnSe core-shell nanocrystal adsorbed on GaAs substrates are presented. These studies reveal the existence of excitation transfer between the nanocrystal and the substrate at very low excitation power density as will be needed for the eventual objective of realizing high sensitivity sensors for early detection of chemical, biochemical and biological agents. The mechanism of excitation transfer is discussed. Chapter 7 summarizes the results presented in chapters 2 to 6, and suggests some directions for future research. 20 Chapter 1 References [1.1] G. Galilei, Dialogues Concerning Two New Sciences, translated by H. Crew and A. de Salvio, Prometheus Books, 1991. [1.2] A. I. Ekimov, A. L. Efros, A. A. Onushchenko, “Quantum size effects in semiconductor microcrystals”, Solid State Commun. 56, 921-924 (1985). [1.3] S. H. Tolbert, A. P. Alivisatos, “High-pressure structural transformation in semiconductor nanocrystals”, Ann. Rev. Phys. Chem. 46, 595-625 (1995). [1.4] J. D. Aiken III, R. G. Finke, “A review of modern transition-metal nanoclusters: their synthesis, characterization, and applications in catalysis”, J. Mol. Catal. A – Chem. 145, 1–44 (1999). [1.5] S. Coe, W. K. Woo, M. Bawendi, V. Bulovic, “ Electroluminescence from single monolayers of nanocrystals in molecular organic devices”, Nature, 420, 800- 803 (2002). [1.6] I. Gur, N. A. Fromer, M. L. Geier, A. P. Alivisatos, “Air-stable all-inorganic nanocrystal solar cells processed from solution”, Science 310, 462-465 (2005). [1.7] M. Bruchez Jr. M. Moronne, P. Gin, S. Weiss, A. P. Alivisatos, “Semiconductor Nanocrystals as Fluorescent Biological Labels”, Science 281, 2013- 2015 (1998). [1.8] W. C. W. Chan and S. Nie, “Quantum Dot Bioconjugates for Ultrasensitive Nonisotopic Detection”, Science, 281, 2016-2018 (1998) [1.9] J. W. Weiland, W. Liu, M. S. Humayun, “Retinal prosthesis”, Annu. Rev. Biomed. Eng. 7, 361-401 (2005). [1.10] Toward Replacement Parts for the Brain – Implantable Biomimetic Electronics as Neural Prostheses, Eds. T. W. Berger, L. Glanzman, MIT Press, Cambridge, MA (2005). [1.11] A. P. Alivisatos, W. W. Gu, C. Larabell, “Quantum dots as cellular probes”, Ann. Rev. of Biomed. Eng. 7, 55-76 (2005). [1.12] X. H. Gao, Y. Y. Cui, R. M. Levenson, L. W. K. Chung, S. M. Nie, “In vivo cancer targeting and imaging with semiconductor quantum dots”, Nature Biotechnol., 22, 969-976 (2004). 21 [1.13] Hereafter we use the phrase nanocrystal specifically for colloidal nanocrystals typical synthesized using solution chemistry in order to distinguish from the crystalline nanostructured material grown via vapor phase epitaxy and contained in solid substrates (for example the self-assembled quantum dots). [1.14] M. Faraday, “The Bakerian Lecture: Experimental Relation of Gold (and other Metals) to Light”, Philosophical Transactions of the Royal Society of London, 147, 145-181 (1857). [1.15] C. B. Murray, D. J. Norris, M. G. Bawendi, “Synthesis and characterization of nearly monodisperse CdE (E=S, Se, Te) semiconductor nanocrystallites.” J. Am. Chem. Soc. 115, 8706-8715 (1993). [1.16] M. A. Hines, P. Guyot-Sionnest, “Synthesis and Characterization of Strongly Luminescing ZnS-Capped CdSe Nanocrystals”, J. Phys. Chem. 100, 468-471 (1996). [1.17] B. O. Dabbousi, J. Rodriguez-Viejo, F. V. Mikulec, J. R. Heine, H. Mattoussi, R. Ober, K. F. Jensen, M. G. Bawendi, “(CdSe)ZnS core-shell quantum dots: Synthesis and characterization of a size series of highly luminescent nanocrystallites” J. Phys. Chem. B 101, 9463-9475 (1997). [1.18] W. G. J. H. M. van Sark, “Enhancement of solar cell performance by employing planar spectral converters”, Appl. Phys. Lett. 87, 151117 (2005). [1.19] V. I. Klimov, A. A. Mikhailovsky, S. Xu, J. A. Hollingsworth, C. A. Leatherdale, H. J. Eisler, and M. G. Bawendi, “Optical gain and stimulated emission in nanocrystal quantum dots”, Science, 290, 314-317 (2000). [1.20] A. Madhukar, Ch.2 in Nano-Optoelectronics, Ed. M. Grundmann, Springer- Verlag, Berlin, (2002) [1.21] S. Guha, A. Madhukar, K. C. Rajkumar, ”Onset of incoherency and defect introduction in the initial-stages of molecular-beam epitaxial-growth of highly strained InxGa1-xAs on GaAs (100)”, Appl. Phys. Lett. 57, 2110-2112 (1990). [1.22] D. Bimberg, M. Grundmann, F. Heinrichsdorff, N. N. Ledentsov, c, V. M. Ustinov, A. E. Zhukov, A. R. Kovsh, M. V. Maximov, Y. M. Shernyakov, B. V. Volovik, A. F. Tsatsul’nikov, P. S. Kop’ev and Zh. I. Alferov, “Quantum dot lasers: breakthrough in optoelectronics”, Thin Solid Films 367, 235-249 (2000). [1.23] E. Kim, A. Madhukar, Z. M. Ye, and J. C. Campbell, “High detectivity InAs quantum dot infrared photodetectors”, App. Phys. Lett. 84, 3277-3279 (2004). [1.24] Nano-Optoelectronics, Ed. M. Grundmann, Springer-Verlag, berlin (2002). 22 [1.25] A. Madhukar, S. Lu, A. Konkar, Y. Zhang, M. Ho, S. M. Hughes, A. P. Alivisatos, “Integrated semiconductor nanocrystal and epitaxical nanostructure systems: Structural and optical behavior”, Nano Lett. 5, 479-482 (2005). [1.26] D. V. Talapin, A. L. Rogach, M. Haase, H. Weller, ”Evolution of an Ensemble of Nanoparticles in a Colloidal Solution: Theoretical Study”, J. Phys. Chem. B 105, 12278-12285 (2001). [1.27] L. Manna, E. C. Scher, A. P. Alivisatos, ”Synthesis of soluble and processable rod-, arrow-, teardrop-, and tetrapod-shaped CdSe nanocrystals”, J. Am. Chem, Soc. 122, 12700-12706 (2000). [1.28] T. Mokari, E. Rothenberg, I. Popov, R. Costi, U. Banin, “Selective growth of metal tips onto semiconductor quantum rods and tetrapods”, Science 304, 1787-1790 (2004). [1.29] D. H. Son, S. M. Hughes, Y. D. Yin, A. P. Alivisatos, “Cation exchange reactions-in ionic nanocrystals” Science 306, 1009-1012 (2004). [1.30] A. Madhukar, “Surface Modification Engineered Assembly of Novel Quantum Dot Architectures For Advanced Electronics“, Proposal to DoD, (2000), DARPA/AFOSR Grant. No. F49620-01-1-0479. [1.31] Q. H. Xie, A. Madhukar, P. Chen. N. P. Kobayashi, “Vertically self-organized InAs quantum box island on GaAs (001)”, Phys. Rev. Lett. 75, 2542-2545 (1995). [1.32] G. H. Wu, R. H. Datar, K. M. Hansen, T. Thundat, R. J. Cote, A. Majumdar, “Bioassay of prostate-specific antigen (PSA) using microcantilevers”, Nature Biotechnol. 19, 856-860 (2001). [1.33] Y. Cui, Q. Q. Wei, H. K. Park. C. M. Lieber, Nanowire nanosensors for highly sensitive and selective detection of biological and chemical species, Science, 293, 1289-1292 (2001). [1.34] Y. P. Sun, K. F. Fu, Y. Lin, W. J. Huang, “Functionalized carbon nanotubes: Properties and applications”, Accounts Chem. Res. 35, 1096-1104 (2002). [1.35] I. L. Medintz, A. R.Clapp, H. Mattoussi, E. R. Goldman, B. Fiber, J. M. Mauro, “Self-assembled nanoscale biosensors based on quantum dot FRET donors”, Nature Materials, 2, 630-638 (2003). [1.36] J. H. Kim, D. Morikis, M. Ozkan, “Adaptation of inorganic quantum dots for stable molecular beacons”, Sensors and Actuat. B – Chem. 102, 315-319 (2004). 23 [1.37] D. S. Ginger and N. C. Greenham, "Electrical Properties of Semicondcutor Nanocrystals", Chap. 7 in Semiconductor and Metal Nanocrystals, Ed.V. Klimov, Marcel Dekker, Inc. (2004). 24 Chapter 2. Colloidal Nanocrystals Synthesis and Electronic Structure: The Basics In this chapter we discuss the basics of the nanocrystal quantum dot: their synthesis and electronic structure. Given our commitment in the NIR wavelength region, and their integration with the existing GaAs based optoelectronics, our work and discussion therefore is mainly on InAs based NIR nanocrystals. Commercially available CdSe based nanocrystals are also used in the work discussed in the next few chapters. We focus on nanocrystals prepared via solution routines, since they are the only type of nanocrystals that can be manipulated and are compatible with biological application, although historically lots of the conceptual issue related to nanocrystals property and synthesis had been developed contemporarily, if not earlier, in nanocrystals synthesized in inorganic matrix such as glass [2.1]. §2.1 Solution Route to Nanocrystal Synthesis §2.1.1 Historical Perspective Early investigations of solution route for nanocrystal synthesis centered on II- VI materials. Brus et al first prepared CdS nanocrystals using arrested precipitation method in aqueous solution [2.2]. The second general method for solution based nanocrystal synthesis was first reported by Meyer and coworkers who used inverse micelles as restricted matrices for nanocrystal growth [2.3]. The next two major advances came during nanocrystal growth using the inverse micelle method and can be attributed to, respectively, Steigerwald et al [2.4] and Bawendi et al [2.5]: (1) the 25 use of organometallic precursor (e.g. (Me 3 Si) 2 Se instead of inorganic precursor (e.g. Na 2 Se); (2) the annealing of nanocrystals in polar Lewis base such as trioctylphosphine (TOP) or trioctylphosphine oxide (TOPO). These two developments together lead to the third and so far the most successful synthesis route: organometallic precursors are thermolysed in polar Lewis base acting as both coordinating solvent and capping agent [2.6]. This methodology yields highly- crystalline, soluble nanocrystals with robust surface capping and high quantum yield (>10% at room temperature). Though the III-V nanocrystals had been investigated nearly contemporaneously with the II-V [2.7, 2.8], the synthesis of high quality III-V nanocrystals came a few years later. It involves the same two key advances as the II- VI: thermolysis of organometallic precursor and the polar Lewis base as coordinating solvent and surfactant. Guzelian et al [2.9] and Micic et al [2.10], respectively, reported the synthesis of InAs and InP nanocrystals. Both these syntheses were based on the same dehalosilylation reaction between InCl 3 and As(SiMe 3 ) 3 or P(SiMe 3 ) 3 in TOP or TOPO as the solvent. The dehalosilylation reaction for III-V synthesis itself had been employed by Alivisatos et al earlier [2.8]. The development of using TOP as solvent and surfactants allows not only high temperature reaction (at >260 C) for annealing of the nanocrystals but also a TOP/TOPO passivated surface, which yields highly crystalline and optically active nanocrystals. Another major development was the invention of core/shell structure in which the nanocrystal core is overcoated by a shell layer of higher bandgap 26 semiconductors. Hines et al first reported the synthesis of CdSe/ZnS core/shell nanocrystals using an all organometallic route [2.11], which was followed closely by Peng et al [2.12] and Dabbousi et al [2.13]. Later the core/shell method was applied to III-V nanocrystals by Banin et al [2.14] and Micic et al [2.15], for the InAs and InP nanocrystals, respectively. The shell provides a more effective passivation for the nanocrystals compared to the organic surfactants. This dramatically improves the quantum efficiency (QE) of the nanocrystals. The CdSe/ZnS core/shell nanocrystals typically have room temperature QE>50% and are thus now the most widely used nanocrystal quantum dots. Moreover, in a core/shell nanocrystal with thick (~ 3 atomic layers) shell layers, the carrier wavefunction is largely confined in the core and does not probe its surface, which makes the nanocrystal insensitive to perturbation in the surface status, no matter a designed surface modification or attacks (such as oxidation) from a hostile environment. This allows the most successful and the only commercial application of nanocrystal quantum dots so far: the nanocrystals are surface functionalized by biological reagent (such as antibodies) and applied as immunological fluorescent label which can be stable in biological solution environment for many weeks [2.16, 2.17]. §2.1.2 Experimental Apparatus The organometallic thermolysis based nanocrystal growth is typically done in a reaction apparatus depicted schematically in figure 2.1. Since oxidization needs always to be avoided for nanocrystals demanded for high optical quality, the reaction 27 apparatus is air-tight and kept under inert gas during the growth. The apparatus is connected to Schlenk line which allows the apparatus to be alternatively connected to either vacuum or inert gas to remove the residue air in it. For guidance in handling air-free chemical reaction and the use of Schlenk line, refer to [2.18]. Inject CVD type precursors E.g. InCl 3 and As(SiMe 3 ) 3 Cd(CH 3 ) 2 and Se:PBu 3 Hot anionic surfactant, “high T,” 250-350 C (size-dependent melting T!) P=O Heating Mantle Stirrer Air-free 3-neck flask To Shlenk Line Inject CVD type precursors E.g. InCl 3 and As(SiMe 3 ) 3 Cd(CH 3 ) 2 and Se:PBu 3 Hot anionic surfactant, “high T,” 250-350 C (size-dependent melting T!) P=O P=O P=O Heating Mantle Stirrer Air-free 3-neck flask To Shlenk Line Figure 2.1 A typical reaction apparatus for nanocrystal synthesis 28 §2.1.3 Zero th Order Consideration: Precursors, Surfactants and Solvents In a zero th order view, the synthesis of compound nanocrystals is straightforward. Precursor monomers (thereafter referred to as monomers) which contain the two component elements of nanocrystal are mixed and reacted in hot coordinating solvent under “appropriate growth conditions”. The monomers have high free energy. When they decompose in the right fashion, the reaction should move towards the most thermodynamically favorable direction, that is, the formation of the crystalline material. Often the coordinating solvent also acts as the surfactants that cover the formed nanocrystals to serve two purposes: (a) prevent them from agglomeration; (b) to passivate the surface states of the semiconductor nanocrystals. In other cases surfactants are added into the solvent before or during the reaction. In the most successful solution routes for II-V and III-V nanocrystals, highly reactive CVD type organometallics such as dimethyl cadium (Cd(Me) 2 ) or Trimethylsilyl Arsenide (TMS) 3 As are used as monomers [2.6, 2.9]. The reactivity of the monomers should be chosen such that they decompose to form nanocrystal at a reasonable rate in the growth condition, while being stable enough under storage and when mixed under room temperature before being injected into the reaction vessel. The reactivity of the metal organic precursor monomer is often tuned with different choices of appropriate organic group. In addition, the kind of monomer that tends to decompose into its component element should be avoided. The solvent/surfactant chosen should be stable to allow reaction to occur at high temperature (~300 C) for better nanocrystal crystallinity. To date, the nanocrystals of the highest quality are 29 produce in polar Lewis base TOP/TOPO as both solvent/surfactant [2.6, 2.9]. The binding between surfactant and the formed nanocrystals should have certain reversibility under growth temperature, which allows nanocrystals to be exposed to monomers and grow at a reasonable rate. The clever choices of surfactants can also allow independent control of growth rate on different crystalline faces of the nanocrystals, hence allowing nonspherical nanocrystals to be produced [2.19]. The fundamental difference between the growth of a nanocrystal and bulk material is the significance the surface energy of nanocrystals. Consider the Gibbs- Thomson equation [2.20] that decides the solubility of the colloidal crystals: ) / 2 exp( ) ( ) ( rRT V S r S m σ ∞ = (2.1) where ) (r S and ) ( ∞ S are respectively the solubility (solubility i d k k S / = , where i k and d k are, respectively, the rate constants of monomer incorporation and dissociation) of crystals of radius r and bulk crystal; σ is the specific surface energy; r is the nanocrystal radius; m V is the molar volume of the materials, R is gas constant; T is the absolute temperature. From the Gibbs Thomson equation, we would guess that the growth of nanocrystal has a characteristic length: RT V m σ 2 (which is often referred to as capillary length, and is usually on the order of 1nm). When the size of nanocrystal is comparable to the capillary length, its growth will be size dependent. (Our first encounter with of size dependent property and characteristic length!) In order to form crystals of narrow size distribution and as small as a few nanometers, by definition, the reaction should never happen at near equilibrium, 30 since large size crystals are always thermodynamically favorable given their low surface energy. Therefore nanocrystal growth is always controlled by the kinetics of the incorporation and dissociation of the monomers, whose evolution depends on the temperature and the monomer concentration as a function of time which is previously lumped under the phrase “the appropriate growth condition”. Below we discuss the growth kinetics of nanocrystals from which the general considerations to achieve “appropriate growth condition” can be appreciated. §2.1.4 Theory of Nanocrystal Growth Kinetics: Although nanocrystal growth kinetics is often divided into nucleation and continued growth stage and modeled separately (as we will do later), the microscopic view of the growth of an individual nanocrystal is unified: A cluster of (N units) is placed in a solution of monomers. The monomers have a certain probability to diffuse to and react with the surface of the cluster and add one more unit to it (the cluster grows), or a unit may leave the cluster and turn back into a monomer (the cluster dissolve). If the fluctuation of monomer concentration leads to cluster growth to such a size that statistically the cluster will tend to grow further larger given the average monomer concentration and temperature, we say that a nucleus of nanocrystal is formed. The evolution of the nanocrystal size from then on is referred to as the continued growth. The macroscopic concepts such as nucleation rate and growth rate are the statistical outcomes of the above microscopic process which is 31 decided by (a) the incorporation or dissociation of the monomer on the nanocrystal surface (b) the transport of the monomer to the nanocrystal surface. §2.1.4(a) Growth Rate The incorporation and the dissociation of the monomer can be described as chemical reaction involving activation barrier i μ Δ and d μ Δ . In the case of nanocrystals, since its chemical potential depends on the curvature of the surface by means of the Kelvin equation, the activation barrier is a function of the nanocrystal radius: r V r m i i σ α μ μ 2 ) ( ) ( + ∞ Δ = Δ (2.2) r V r m d d σ β μ μ 2 ) ( ) ( − ∞ Δ = Δ (2.3) where α and β are the transfer coefficients ( 1 = + β α ); ) ( ∞ Δ i μ and ) ( ∞ Δ d μ are respectively the activation energy for incorporation and dissociation the case of flat surface (or infinite radius). The rate constant of incorporation and dissociation can be expressed through the height of activation barrier: ) 2 exp( ) ( ) / ) ( exp( ) ( rRT V k RT r C r k m i i i i σ α μ − ∞ = Δ − = (2.4) ) 2 exp( ) ( ) / ) ( exp( ) ( rRT V k RT r C r k m d d d d σ β μ ∞ = Δ − = (2.5) where i C and d C are constants; ) ( ∞ i k and ) ( ∞ d k are the rate constant when the surface is flat. 32 The diffusion of the monomer to the surface is expressed by Fick’s first law: r x diff dx x M d D x J ≥ ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ = )] ( [ 4 2 π (2.6) where D is the diffusion constant of monomers (or more precisely a mass transfer coefficient under stir condition), and x is the distance from the center of the nanocrystal, M is the concentration of the monomer. At ∞ = x , the monomer is at a its solution bulk concentration () ∞ M At r x = , the growth rate of the nanocrystals is decided by the difference between the monomer incorporated and dissociated from the nanocrystal surface, which in turn have to be equal to the monomers diffusing in: diff d i m J k r r M k r dt dr V r = − = 2 2 2 4 )] ( [ 4 ) / 4 ( π π π (2.7) Combine Equ. 2.4-2.7 we obtain the growth rate of nanocrystals ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎦ ⎤ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎣ ⎡ ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ ∞ + ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ − ∞ ∞ ∞ = rRT V k D r rRT V S M DS V dt dr m i m m σ α σ 2 exp ) ( 2 exp ) ( )] ( [ ) ( (2.8) where ) ( / ) ( ) ( ∞ ∞ = ∞ i d k k S is the bulk solubility of the bulk crystal material. Observing Equ.2.8, we notice that for a given bulk monomer concentration there exists critical radius: )] ( / ) ( ln[( 1 2 * ∞ ∞ ⋅ = S M RT V r m σ (2.9) 33 at which nanocrystal has zero growth rate. On the average, nanocrystals smaller than critical radius dissolve and those larger than critical size grow. Therefore * r is the critical radius for nanocrystal nucleation. The critical radius itself increases with decreasing precursor concentration. Equ. 2.8 is too complicated and does not lead to an analytic model or intuitive understanding of the nanocrystal growth kinetic. Talapin et al [2.21] designed a Monte Carlo simulation of the evolution of the nanocrystal size distribution based on equ.(2.8) and a random generated Gaussian distribution of size at t=0. In order to gain a reasonably simple conceptual understanding of the nanocrystal growth that is able to guide the experimental work, the nanocrystal growth mechanism has been analyzed by Peng et al [2.22] using a classical model of diffusion limited colloidal growth by originally LaMer et al [2.23] and Reiss [2.24] which is later developed by Sugimoto [2.20] in the context of ~1 μm size colloidal particles. In Sugimoto’s model two assumptions are made: (a) the growth is diffusion limited, namely ) ( ∞ << i k D , (b) capillary length is smaller than the nanocrystal radius: r RT V m << σ 2 (consider RT V m σ 2 is usually on the order of 1nm, this assumption is very questionable for small nanocrystals). Under these assumptions the growth rate can be simplified to be: ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ − ∞ = r r rRT S DV dt dr m 1 1 ) ( 2 * 2 σ (2.10) 34 A representative calculated curve of size dependent growth rate in diffusion limited growth is shown in figure 2.2. The important qualitative feature to notice is that there exists a radius r max ~2r* corresponding to maximum growth rate. Reduction in growth rate is seen in nanocrystal beyond radius r max (figure 2.2). This is a manifestation of the reduction of precursor influx density at nanocrystal surface with increasing nanocrystal radius which is limited by the diffusion of precursor monomers. As we will see later this reduction in growth with increased nanocrystal radius as the signature of diffusion limited growth provides a way to keep the size distribution of nanocrystals narrow. Figure 2.2 Diffusion controlled growth rate of nanocrystal vs. radius. r: radius of nanocrystal; r*: critical radius; 35 §2.1.4(b) Nucleation Rate The details of nucleation kinetics in nanocrystal growth are not known to date. Here we borrow the result in [2.25] using the steady-state homogeneous nucleation theory, in which nucleation rate R can be expressed by: ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎝ ⎛ ⋅ − ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ ∞ Δ − = RT r RT A R i 2 * 4 exp ) ( exp πσ μ , (2.11) where ) ( ∞ Δ i μ and * r are respectively the activation energy for monomer incorporation and critical radius as defined in the §2.1.4(a). The radius distribution of nuclei ) (r P obeys the Gaussian function: ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎝ ⎛ − − = RT r r P r P 3 ) ( 4 exp ) ( 2 * 0 πσ (2.12) The dependence of nucleation rate and size distribution on the monomer concentration is implicitly contained in the critical radius * r (equ. 2.9) )] ( / ) ( ln[( 1 2 * ∞ ∞ ⋅ = S M RT V r m σ (2.9) Combining (2.9) and (2.11), we conclude that a higher monomer concentration leads to a smaller critical radius, hence accelerates nucleation process. If one intends to suppress new nucleation formation so that the precursor monomers can be efficiently used to grow nanocrystals to a large average size, lower monomer concentration is then the preferred condition to use. 36 §2.1.5 Strategy for Nanocrystal Growth: Two Step growth Approach & Size Focusing Early reported nanocrystal synthesis typically involves growth of nanocrystals at a moderate temperature (180 C-300 C) over an extended period of time [2.6], yielding a wide range of sizes. The broad size distribution is then sorted using post-growth techniques. It is clear in this case the nanocrystal size inhomogeneity comes from three sources (a) the inherent temporal distribution of the formation of critical nuclei (b) the size distribution of initial nuclei (c) the subsequent size-dependent growth rate of clusters beyond the critical size. §2.1.5(a) Conventional two-step growth strategy: As recognized by Reiss [2.24] in order to achieve narrow size distribution in colloidal growth, it is the key to temporally separate the colloidal nucleation and growth. Following this strategy, a two-step growth procedure (thereafter referred to as conventional two-step growth) involving separated nucleation and growth phase have been first designed [2.26] for nanocrystal (CdSe). In such a conventional two step growth, high concentration monomers are rapidly injected into the solvent at high temperature to induce instantaneous nucleation. Then the temperature is dropped for the growth step, multiple slow monomer injections are used at appropriate time to growth the nanocrystals to the desired size. The two step growth approach is straightforward given the understanding of nanocrystal nucleation and growth kinetics. The high monomer concentration 37 resulted from the first rapid injection leads to a small critical radius * r for nucleation (Equ. 2.9). The small * r and high temperature together causes fast formation of initial nuclei (equ. 2.9) of a certain size distribution. The initial nucleation drops the monomer concentration, while the reaction temperature is also dropped for the continued growth phase. The monomer concentration and temperature is controlled to be low enough to slow nucleation to negligible rate, but high enough to allow the formed nuclei to continue to grow. Often during the continued growth, further injections need to be made as guided by the growth kinetics to keep the nanocrystal size distribution narrow, referred to as “focusing” of the size distribution. §2.1.5(b) “Focusing” of the size distribution: During continued growth phase, in a scenario where there is no further supply of precursors, nanocrystals are competing for a finite amount of precursor monomers. The monomer concentration will eventually drop to such a level that the corresponding critical radius * r is larger than the nanocrystal’s most probable size, such that the smaller nanocrystal dissolves, and the precursor released is consumed by the larger one. Such growth, referred to as competitive growth or Ostwald ripening, generally lead to broadening of nanocrystal size distribution [2.27]. Nevertheless, we have the choice of further injection of precursor, which in diffusion limited growth can narrow nanocrystal size distribution. As discussed in §2.1.4(a), there exist a radius max r corresponding to the maximum grow rate in diffusion limited growth, which is dependent on the precursor concentration in the 38 reaction vessel. Via further injection of precursor monomer if one keep the max r a little smaller than the most probable size, a scenario can be created where the smaller nanocrystal grows faster and the larger one grows slow, therefore the nanocrystal size distribution is focused. The conditions to achieve focused nanocrystal size distribution during continued growth phase has been investigated by Peng et al [2.22] in the CdSe and InAs nanocrystal systems. §2.1.6 Considerations for III-V nanocrystal growth. In the previous two sections, we discussed the theoretical possibility to achieve narrow nanocrystal distribution using (a) separated nucleation and growth step; (b) focusing of size distribution during continued growth step. However the question is how well the above two methods can be implemented for real nanocrystal synthesis given the available parameter space, without conflicting with each other or another necessities. Indeed these two methods have been very successfully applied to II-VI nanocrystals growth as first applied by Katari et al [2.28] and is the standard method to grow II-VI nanocrystals. However, for the III-V nanocrystals the same methods do not work well due to the difference in nucleation and growth kinetics. The generally recognized difficulty is how to suppress the nucleation during the continued growth phase, as also our experimental result will show later in §2.2.5. As commented by Heath et al [2.29], the difficulty is a reflection of the covalent nature of the III-V compound material compared to the more ionic II-VI material. Unlike group II and VI species, the bare atom and ions of group III and V 39 are not chemically stable, which means the precursor monomers used for III-V synthesis have to be strongly complexed by the solvent to be stable. An example is the InCl 3 /TOP complex for InAs synthesis (see §2.2.1). Therefore the incorporation of precursor to the nanocrystal involves a high energy barrier to first decompose the precursor, which then requires higher temperature during continued growth phase that, undesirably, at the same time will favor nucleation to occur. A separate point is that, moving towards covalent semiconductors, amorphousness plays an increasingly important role. For III-V compounds, amorphous structure forms easily at lower growth temperature. Avoiding this also requires higher temperature to be maintained during the growth phase to generate highly-crystalline nanocrystals. But this again favors nucleation. We note from reported literature CdS or CdSe can be formed by aqueous routine at below 100 C, while the reported growth temperature for III-V InAs or InP is never below 250 C. The same trend of growth temperature requirement to form crystalline material is also seen in vapor phase techniques such as molecular beam epitaxy (MBE). The II-VI CdSe is typically grown at ~300 C, III- V GaAs typically at ~600 C, while IV group Ge typically at ~900 C. Given the above considerations, III-V nanocrystals are typically formed at moderate temperature ~250 C over extended period of time yielding a broad size distribution. It is regarded [2.29] impossible to generate narrow size distribution of III-V nanocrystal using the conventional two step growth and size focusing. In the next section we will discuss our experimental work on III-V InAs nanocrystals growth, as qualitatively guided by the understanding of nucleation and growth 40 kinetics discussed above. Our work shows with appropriate manipulation of the growth condition, the conventional two-step process is still applicable for InAs up to~5nm diameter. Beyond this size range, the concept of two-step growth process is still valid, although the implementation differs from conventional. §2.2 InAs core nanocrystal synthesis §2.2.1 Dehalosilylation Reaction The InAs nanocrystals are synthesized via dehalosilylation reaction of InCl 3 and As[Si(CH 3 ) 3 ] 3 at ~260 °C under argon atmosphere with trioctylphosphine (TOP) serving as both solvent and capping surfactants: InCl 3 + (TMS) 3 As Æ InAs + 3(CH 3 ) 3 Cl Si (gas) This specific reaction itself was first applied by Wells et al [2.30] for bulk InAs and GaAs crystal synthesis, and later applied by Guzelian et al [2.9] that lead to the most successful preparation of InAs nanocrystals so far. The synthesis procedure is adapted and modified based on [2.31]. §2.2.2 Precursor and Stock Solution Preparation. TOP Preparation: Trioctylphosphine (TOP) were purchased from Aldrich (90%) or Strem (97%). The main impurity in the as purchased TOP is its oxided form Trioctylphosphine Oxide (TOPO). The TOP needs to be distilled before used for InAs nanocrystal synthesis, as impurities in the TOP cause nanocrystal formation to be uncontrollable most like by serving as nucleation sites. The TOP was distilled 41 using standard vacuum distillation apparatus. TOP came out at ~170 C depending on vacuum strength. At above 50 C below 170 C, a white solid came out of the distillation preceding TOP. The white impurity must be collected before purified TOP is collected. Since the boiling point of TOP 170 C is much higher than room temperature, one may want to wrap the distillation column with glass wool or heating tape to ease the distillation. After distillation if small amount of white solid particles are still present, the TOP needs to be filtered with 0.22 μm syringe filter. (TMS) 3 As Preparation: Trimethylsilyl Arsenide ((TMS) 3 As) was synthesized following procedure from [2.32]. Briefly, Arsenic solid was reduced to -3 As using sodium-potassium alloy (NaK) in ethylene glycol dimethylether (DME). Then -3 As was reacted with distilled Trimethylsilyl Chloride ((TMS)Cl) to form (TMS) 3 As, which was then purified using a series of vacuum filtration and distillation steps. Note the component materials for this reaction are highly volatile, pyrophoric, and toxic, so this synthesis should be approached with great care. Detailed recipe for (TMS) 3 As synthesis is provided in Appendix A. The obtained (TMS) 3 As is a colorless liquid and should be stored in refrigerator (-30 C) inside the glovebox. Nevertheless (TMS) 3 As still decomposes over a few months. Whenever the stored (TMS) 3 As turns yellow or red in color, it should be redistilled. InCl 3 /TOP Preparation: The precursor for Indium is kind of ill-defined involving metal precursor InCl 3 complexed to TOP. InCl 3 power (99.999%) was obtained from Aldrich in 5g ampoules. (Note: After many trials, we concluded that InCl 3 obtained 42 from Strem does not work). 5.10g InCl 3 and 13.9g distilled TOP was mixed in the glovebox and taken out. The mixture was refluxed under Ar on the Schlenk line at 260 C for 2 hour. The final InCl 3 /TOP solution is a very viscous liquid, transparent or slight yellowish. No undissolved InCl 3 powers should present in the solution, since they serve as nucleation site during the InAs growth. Stock Solution: Stock solution for InAs synthesis were prepared by mixing (TMS) 3 As and InCl 3 /TOP precursor in the glovebox before the synthesis reaction. The two stock solution of In:As molar ration 2:1 and 1.2:1 were preparation, for nucleation and continued growth of the InAs nanocrystals respectively: Stock 1: (TMS) 3 As 0.150g (0.51x10 -3 mol) + InCl 3 /TOP(5.10g/13.9g) 0.85g (1.02x10 -3 mol of InCl 3 ). Stock 2: (TMS) 3 As 0.650g (2.21x10 -3 mol) + InCl 3 /TOP(5.10g/13.9g) 2.21g (2.65x10 -3 mol of InCl 3 ). Upon mixing the stock solution turns into reddish yellow immediately. §2.2.3 Real time PL monitoring of the InAs nanocrystal growth: The real time size distribution of InAs nanocrystals during growth was monitored using PL spectroscopy exploiting the nanocrystals’ size dependent band- edge luminescence. The PL wavelength vs. diameter calibration for InAs nanocrystals is shown in figure 2.3 as adopted from [2.22]. We translate the PL spectrum into nanocrystal size distribution by assuming: (a) a δ-function PL response for nanocrystal of certain size and (b) the same absorption cross-section and QE for 43 Figure 2.3 The room temperature absorption peak and 1 st exciton PL peak wavelength vs. diameter calibrated for InAs nanocrystals. Data adopted from [2.22]. nanocrystals of all sizes. Note that assumption (b) leads to systematic error in the estimated size distribution, since the small nanocrystals have a smaller absorption cross-section and the larger nanocrystal (>5nm diameter) have reduced QE [2.33]. During the growth, aliquots from the growth solution were taken and dissolved in anhydrous toluene in a 1cm optical path cuvette (ocean optics, glass, 5 windows) for PL measurements. The PL spectra were taken using a 5mW 532nm Nd:YAG laser as excitation, and a spectrometer (Acton SP300i) with LN2 cooled InGaAs photodiode 44 array (Princeton Instrument) for fast spectrum measurement. In order to minimize the systematic error due to re-absorption of emission from smaller nanocrystals, the solution for PL was diluted to optical density (OD) ~0.1 (estimate by eye observation). Such real-time PL should be done as quickly as possible to provide information to guide further growth. Finally, if the sole objective is to measure the nanocrystal size distribution during growth, compared to PL, the absorption spectra is a better choice if instrumentation allows. The PL characteristic (e.g. QE) of nanocrystals strongly depends on their surface status and is not very repeatable during the growth with the interference of all the excessive surfactants and precursors. By contrast, the absorption spectrum is insensitive to these factors and therefore a more reliable measurement. §2.2.4 InAs growth procedure using the conventional two step growth Insert 1g of TOP into a 25ml three neck flask, and heat to 300 C, stir vigorously. Avoid too much temperature overshoot, since TOP tends to polymerize at higher temperatures. To initiate nucleation of InAs nanocrystals, inject 0.5ml of Stock 1 into the flask as rapidly as possible (<0.2sec; use at least a 17gauge thick needle). After injection the reaction solution drops to ~280 C, but decrease the temperature further to 260 C for continued growth. From then on, it is important to work fast and follow the PL spectra of aliquots throughout the growth process (§2.2.4). Decisions regarding precursor injection should be made according to the 45 understanding of nanocrystal growth kinetics discussed in §2.1. In order to prevent Ostwald ripening whenever the nanocrystal ensemble size distribution stops to move forward to larger peak size and starts to broaden, new drop-wise injections of precursor should be made as suggested below. Sometimes in order to slow down nucleation during the growth, TOP can be injected to dilute the growth solution (detail provided in §2.2.6). As a rough guideline, 5mins after the initial injection, inject another 0.35ml of Stock 1 drop-wise (~0.5ml/min). Then inject 0.5ml of stock 2 drop-wise (0.5ml/min) roughly every 10mins, until the InAs nanocrystals reaches the desired size. §2.2.5 Size Selective precipitation: After the growth of InAs nanocrystals is done, the nanocrystals need to be precipitated from the growth solution and re-dissolved in the desired solvent for future use. The precipitation separates the nanocrystals from excessive TOP surfactant and precursor monomers in the growth solution. Additionally, due to the difficulty in keeping the InAs nanocrystal size distribution narrow during growth (especially for large average size nanocrystal distributions), such post-growth precipitation also serves the objective of selecting out InAs nanocrystals within a narrow range (~100-150nm PL FWHM typical), hence the term “size selective precipitation”. The principle of operation is that when a controlled amount of non- solvent is added into the InAs nanocrystal solution, statistically the larger 46 nanocrystals agglomerate more readily and fall out of the solution. The detailed procedure is described below: After the InAs synthesis is finished and solution cools to room temperature, seal the three neck flask and take it into a glovebox. Add ~15ml anhydrous toluene, the reaction solution should dissolve complete. Put the solution into a clean flask with stirrer, and add methonal while stirring. The amount of methonal needed depends on the concentration of the solution and the size of the nanocrystals. Keep adding methonal till the InAs nanocrystals start to precipitate, the solution becomes turbid (diffusive light scattering) and has a greenish color on top of the dark brown. (The color of the InAs nanocrystal solution in the flask is difficult to tell due to its high optical density. One needs to suck out some solution using a standard glass pipette and observe its color.) Inside the glovebox, vacuum filter the precipitated solution using a Buckner flask and 0.22 μm Teflon filter. The precipitated nanocrystals will be collected on the filter as brownish deposits. Before the filter dries, remove the filter to a vial and dissolve the precipitate in anhydrous toluene. By this stage the largest fraction of InAs nanocrystals are separated out. Repeat the above precipitation/filtration procedure to collect fractions of sequentially smaller nanocrystals, until the solution is almost colorless. If necessary, further rounds of size-selective precipitation can be performed to further narrow the size distribution of the nanocrystals at the cost of losing more along the way. 47 §2.2.6 InAs nanocrystal growth results and discussions: InAs nanocrystal growth was performed with the conventional two step growth approach described in §2.2.4. As a typical problem for III-V nanocrystals discussed in §2.1.6, we also encounter the broadening of the size distribution mainly due to nucleation during continued growth phase. Figure 2.4(a) shows a real-time PL of nanocrystals grown to ~3.8nm diameter (PL peak ~1070nm, dotted curve), and the PL right after one more injection of precursor (solid curve). The nanocrystals are still of reasonable size distribution up to ~3.8nm, however the additional injection of precursor causes new nanocrystals to nucleate and grow as is evident from the emerging PL peak at 960nm (solid curve, figure 2.4(a)) corresponding to ~2.9nm diameter nanocrystals. After several more injection of the precursor, as a consequence of nucleation and continued growth occurring simultaneously, the nanocrystal size distribution eventually becomes unacceptably broad in the steady state and the average size stops growing. Figure 2.5(a) shows the size-distribution of the largest fraction of nanocrystals extracted from the steady state nanocrystal ensemble after four sequential rounds of size-selective-precipitation. The sizes are derived from transmission electron microscopy images, an illustrative image being shown in figure 2.5(b). The average diameter of such extracted largest fraction of nanocrystals is ~6.5nm. A similar value is reported for the largest InAs nanocrystals which can be isolated from synthesis done at a fixed temperature allowing nucleation and growth to happen at the same time [2.9]. 48 (a) (b) Conventional Two Step Growth Punctuated Growth (a) (b) Conventional Two Step Growth Punctuated Growth Figure 2.4 (a) PL of InAs nanocrystal aliquots during conventional growth approach. Doted line: InAs nanocrystal grown to ~3.8nm average diameter (PL peak 1070nm) before one more injection of precursor. Solid line: after one more injection of precursor; note the merging of PL peak at 960nm indicating that new nucleation had occurred. (b) PL of InAs nanocrystal aliquots during punctuated growth approach. Dashed line: InAs nanocrystal of ~3.9nm diameter before injection of precursors. Solid line: immediately after one more injection of precursor, nanocrystals grow uniformly to ~5nm (PL peak 1250nm) diameter with little new nucleation occurring. 49 (a) (c) (b) (d) Conventional Two Step Growth Punctuated Growth (a) (c) (b) (d) Conventional Two Step Growth Punctuated Growth Figure 2.5 (a) Size distribution statistics of largest InAs nanocrystals separated after a conventional two-phase growth as measured from TEM images of nanocrystals on C-coated grid. (b) Corresponding illustrative TEM image (courtesy of S. Hughes). (c) Size distribution statistics of largest InAs nanocrystals separated after a punctuated growth as measured from TEM images (d) Corresponding illustrative TEM image (courtesy of Steven Hughes, University of California, Berkeley). 50 Since the temperature during continued growth can not be lowered arbitrarily while still maintaining the InAs nanocrystal quality (refer to §2.1.6), in order to suppress nucleation during continued growth the only natural countermeasure is to keep the precursor concentration low, but yet not too low for Ostwald ripening to happen. One possibility is to inject TOP solvent to dilute the growth solution when nanocrystals grow large. Indeed this measure eases the difficulty of nucleation during growth and leads to an improved procedure of two step growth. Shown in figure 2.6 is a series of aliquots PL during the growth using the same conventional 2 step growth procedure other than TOP solvent injection at different stages. Right after initial fast precursor injection, formed nanocrystal nucleation of ~2.7nm diameter (PL1 black, peak at 940nm). With two more slow injections, the nanocrystals grew to ~3.9nm (PL6 magenta, peak at 1080nm, FWHM ~150nm). Anticipating that further injection of precursor would have caused new nucleation formation, at this point 1ml of TOP was injected to dilute the growth solution. In such diluted growth solution, one more injection of precursor was made, the aliquot PL right after is curve PL8 (violent). Compare to before injection (PL7 yellow), PL8 is broadened towards smaller wavelength, indicating some small nanocrystals were still formed due to new nucleation. However, the damage of nucleation to size distribution in this case is obviously much smaller in comparison with the nucleation resulted from the same amount of precursor injection around the same stage of the growth but without TOP dilution (figure 2.4(a)). As shown in figure 2.6 PL11 (dark yellow), using further injection of precursor material and TOP, 51 Figure 2.6 Real time aliquots PL result during a conventional two-step growth as described in the text. Timing for precursor injections, TOP injections and PL measurements are marked on the right. 52 the nanocrystals can be growth to most probable diameter ~5nm (PL peak at 1220nm) with still reasonable size distribution (~200nm PL FWHM). However no attempt to push the nanocrystal beyond 5nm average diameter with more injection was successful without compromising size distribution. The average diameter of the largest isolatable fraction of nanocrystals was again limited to ~6.5nm diameter. For different applications, the InAs nanocrystal growth of course was stopped when desired size or emission wavelength is reach, and was size-selective precipitated using the procedure described in §2.2.5. PL curves of three illustrative precipitated samples are shown in figure 2.7. Typical PL FWHM 100nm to 150nm can be achieved, which translates into 20% to 25% FWHM for the nanocrystal size distribution. §2.2.7 Enhancing Nanocrystal Average Diameter: The “Punctuated” Growth Approach. Although TOP injection during continued growth offers a possibility to manipulate the precursor concentration and helps to lessen nucleation, the freedom of manipulation is limited. Because in a conventional two-step growth the initial nucleation and continued growth are coupled to each other in a single growth cycle. A reasonable volume of high concentration precursors needs to be injected for any reasonable initial nucleation rate which results in a large number of nanocrystal nuclei. To grow these nuclei using low precursor concentration means inversely 53 Figure 2.7 PL from size-selective-precipitated InAs nanocrystal samples; typical PL FWHM 100-150nm. The corresponding FWHM for nanocrystal size distribution is 20%-25%. proportional large volume of precursor and TOP solvent need to be injected during growth, which is not convenient and sometimes impractical. For the purpose of integration of colloidal nanocrystals in epitaxial structures, InAs nanocrystals as large as 8nm diameter are desired [2.34]. To create isolatable nanocrystals of this size, the suppression of nucleation during the growth is essential to not allow smaller nanocrystal to dominate the size distribution and to consume most of the injected precursor material. To achieve this objective, we have designed an alternative implementation of two step growth which we call the “punctuated 54 growth” approach. In this approach the nucleation phase and the continued growth phase are separated into two different growth cycles. In the first growth cycle, InAs nanocrystals were first nucleated and growth continued to ~4nm average diameter, same as in the conventional growth approach described in §2.2.6. The synthesized nanocrystals were precipitated and dissolved in TOP. These nanocrystals are used as nucleation seeds in a second growth cycle. Since there is now no need for initial nucleation, the second growth cycle can be carried out at a significantly lower nanocrystal nuclei concentration and precursor concentration so that the critical nucleus size is kept large and nucleation during continued growth can be better suppressed. As a typical example, our second growth cycle starts with 6ml 1mg/ml ~4nm diameter InAs nanocrystal/TOP solution. Growth was done at temperature 260 C with stirring. Precursor stock solution used was ~10 times diluted with respect to that described in §2.2.2 for the conventional two step growth (made by mixing (TMS) 3 As:InCl3:TOP at a molar ratio of 1:1.2:19). Several injections of stock solution were made (drop-wise 0.2ml a time) to grow InAs to the largest possible size [2.35]. As shown by the PL spectrum (figure. 2.4(b)) of aliquots from the growth solution, in the second growth cycle the initial InAs nanocrystals with average ~4nm diameter after one injection of the precursor grow uniformly to average ~5nm in diameter with little small nanocrystals appearing, indicating the desired suppression of nucleation during growth of existing nanocrystals. After the second growth cycle, the nanocrystals were separated by one round of size-selective- precipitation. The size distribution statistics of the largest portion is shown in figure 55 2.5(c). It is obtained from TEM images as discussed in the preceding, a typical example of which is shown in figure 2.5(d). The average diameter of the largest portion is 8.5nm which is 2nm larger than the largest obtained [2.9] using the conventional growth approach. §2.3 InAs/ZnSe shell/core nanocrystals. §2.3.1 Introduction to core/shell nanocrystals The as-synthesized InAs core nanocrystals has a relatively low quantum efficiency (QE) ~1% presumably due to imperfect surface passivation by the TOP surfactants. Generally for II-VI and III-V nanocrystals, it is well known that the luminescence efficiency, spectrum and lifetime were very strongly affected by the surface status of the nanocrystals. This is generally assigned to the presence of the mid-gap surface states and rising from surface nonstoichiometry and unsaturated bonds. (Detailed discussion of the surface effects on the optical property of the nanocrystal will be provided in chapter 5). Indeed the organic surfactants (such as TOPO or TOP) cannot be expected to fully passivate the surface by covering every unsaturated bond, due to the large cone angle of such molecules (e.g. TOPO cone angle 130 ) and the resulting geometric interference [2.28]. In addition, the passivation by surfactants is not reliable and may degrade over time due to the weak bonding between the surfactants and the nanocrystals surface atoms. A generally proven strategy to lessen the surface influence and to boost the QE is to grow a shell layer of high bandgap semiconductor on the core nanocrystals 56 to form a robust inorganic passivation of the nanocrystal surface. In such core/shell nanocrystal with thick shell layers, the carrier wavefunction is largely confined in the core and does not probe its surface, which makes the nanocrystal insensitive to perturbation in surface status Specifically for InAs core nanocrystals, systematic studies was carried out by Cao et al [2.31, 2.14] in which different choices of shell material on the InAs including InP, GaAs, CdSe, ZnSe and ZnS were tested. This study concluded InAs/ZnSe core/shell nanocrystal offers the best QE ~20%. Below we discuss the procedure for ZnSe shell growth on InAs core. §2.3.2 Stock Solution for ZnSe shell: TOPSe solution (1mmol) was prepared by dissolving Selenium in distilled TOP (see §2.2.1). Selenium shots ((99.99% Strem) was used. It needs to be stirred in TOP for a few hours inside a glovebox to dissolve completely. Stock solution for ZnSe shell growth was prepared by mixing TOPSe solution and 2M Zn(CH 3 ) 2 toluene solution (Aldrich) with equal molar amounts. §2.3.3 InAs core Nanocrystal Concentration and Amount of Precursor Material: The total mass/molar number of InAs core nanocrystal in a given solution sample needs to be estimated to determine the amount of ZnSe monomers to be used for shell growth. One way to do it is to take a measured fraction of the solution sample and evaporate all the solvent by passing inert gas (Ar or N 2 ) through it and then weigh the remaining material (assuming all the remaining is InAs nanocrystals). 57 More conveniently the InAs nanocrystal concentration can also be estimated by measuring its absorption. Take a measured amount of the InAs solution and dilute it with a measured amount of the solvent until the OD of the first exciton peak as measured by spectrometer is 1.0 (using a 1cm optical path cuvette). The concentration of an OD 1 solution is roughly 1 mg/ml [2.32]. Around such range, the OD and concentration can be assumed to be linear. One can convert the measured mass of InAs nanocrystals in a sample into molar number using the nanocrystal size distribution as determined by PL spectroscopy or TEM. One can further calculate the total surface area of the core InAs. Then assuming (a) all the injected Se and Zn monomer reacts to form shell on the core InAs, and (b) the ZnSe shell lattice constant remains the same as the bulk, one can calculate the molar number of Zn and Se precursor monomers needed to form one monolayer of shell. §2.3.4 ZnSe Shell Growth Procedure: ~0.4 μmol of TOP capped InAs core toluene solution (as prepared using the protocol in §2.2.2) was mixed with 2g of TOP and 2g of TOPO in a 25ml 3neck flask in the glovebox. The mixture was taken out, connected to the Schlenk line, and heated to 60 C under Ar. The TOPO will liquefy and nanocrystal dissolves. Toluene can be removed at this point by vacuum the reaction solution. Continue to heat the solution to 260 C under Ar. While stirring, a calculated amount of ZnSe stock solution corresponding to ¼ to ½ monolayer shell growth was injected dropwise into 58 the hot nanocrystal solution roughly every 10mins. After each injection, aliquots was taken from the growth solution and carefully diluted in toluene to the same optical density (~OD 0.1) by eye observation. (For convenience, the first diluted aliquots were kept aside as a standard for side-by-side comparison to each subsequent aliquots dilution.) PL of each diluted aliquots was measured, and the evolution of the PL intensity is monitored. If absorption spectrometer access had been convenient and could be done timely, the PL intensity should be normalized by the measured OD of the aliquots. Stop the injection of precursor until the PL intensity stop increasing, or the desired shell thickness as calculated is reached. Then the reaction solution was cooled to room temperature, transferred to the glovebox. The InAs/ZnSe core shell nanocrystals were separated from the growth solution and redissolved in toluene using a similar size-selective precipitation procedure as described in §2.2.5 using toluene and methanol as solvent and non-solvent respectively. The synthesized InAs/ZnSe core/shell nanocrystals are expected to have both TOP and TOPO as their surfactant. §2.3.5 ZnSe Shell Growth Results and Discussion Following the above procedure, ZnSe shell was coated on TOP capped InAs core nanocrystal (~4nm in diameter, PL peak 1094nm). Figure 2.8(a) shows the real- time aliquots PL evolution as a function of shell thickness during a typical growth. Figure 2.8(b) shows the evolution of integrated PL intensity. Figure 2.8(c) shows the evolution of PL peak position. Note here the shell thickness is just a book keeping number as calculated using the assumption described in §2.2.3, and has only relative 59 meanings. In taking the PL, all aliquots were diluted to the same optical density (see §2.2.3), so that the PL intensity is proportional to the QE of the nanocrystals. From figure 2.8(b), we see that as ZnSe shell growth thicker, the PL Intensity (and QE) of the nanocrystals reaches its maximum at ~2ML shell thickness, then with further shell growth, the PL intensity (and QE) decreases. The QE at ~2ML shell thickness is ~20 times higher than the QE without shell growth. In absolute terms, QE of InAs core without shell is ~1%. QE of InAs/ZnSe with 2ML shell is ~20% consisted with reported by Cao et al [2.31]. The improvement in QE most likely is due to the improved surface passivation by the ZnSe layer, hence the decrease in surface mediated nonradiative recombination. In addition, with the growth of ZnSe shell, the nanocrystal PL peak position remained essentially the same. This is not surprising, since the ZnSe shell bandgap 2.71eV is much larger than HOMO-LUMO gap (1.13eV) of the InAs core, therefore, as far as the ground state electron and hole are concern, the ZnSe is still essentially the same an infinite barrier as the organic surfactants. Figure 2.9(a)-(c) shows the TEM images of InAs/ZnSe core/shell nanocrystal samples from three different growths that ended with three different final shell thicknesses: 1.82ML, 1.95ML, and 2.47ML. (The starting InAs core for these three growth are the same ~4nm diameter ones as used the in the real-time PL study above. The book keeping of shell thicknesses is identical too.) The shape evolution of the InAs/ZnSe nanocrystal with increased shell coverage is evident from these TEM images. At 1.82ML shell thickness, all the nanocrystals are essentially round shaped. At 1.95ML shell thickness, a fraction of the nanocrystals starts to become non- 60 spherical, though precise description of the nanocrystal shape at this stage is difficult due to the quality of the TEM. (a) (a) (b) (b) (c) (c) Figure 2.8 (a) InAs/ZnSe Aliquots PL evolution during growth as a function of calculated shell thickness (b) Integrated PL intensity vs. shell thickness (c) PL peak position vs. shell thickness 61 ZnSe shell 1.82ML (a) ZnSe shell 1.82ML (a) ZnSe shell 1.95ML 20nm ZnSe shell 1.95ML 20nm ZnSe shell 2.47ML 20nm ZnSe shell 2.47ML 20nm Figure 2.9 TEM of InAs/ZnSe nanocrystals on carbon-coated grids which shows the InAs/ZnSe nanocrystal shape evolution as a function of ZnSe shell thickness: (a) 1.82ML (b)1.95ML (c) 2.47ML. Note the presence of dendrimer formation at 2.47ML shell thickness! In all three cases the same InAs core nanocrystals (~4nm diameter) are used as starting material. The same book keeping of shell thickness is followed as described in §2.3.3 (TEM images courtesy of Atul Konkar, University of Southern California, and Steven Hughes, University of California, Berkeley). (c) (b) 62 At 2.47ML shell thickness, essentially all the nano-crystals are developed into dendrimers, the TEM image of which is some what similar to that of CdSe tetrapods with short arms [2.19]. The ZnSe has a bulk lattice constant 5.6A, ~7% smaller than InAs (6A). Therefore the ZnSe shell on InAs core is under tensile stress, which is potentially the driving force for the dendrimerization of the nanocrystals. In addition, the onset of dendrimer formation coincides with the drop of nanocrystal QE >2ML shell thickness, suggesting that such dendrimerization is likely to be accompanied by defects formation in the shell layers. This is to author’s knowledge the first observation of such nanocrystal dendrimerization during stressed shell overgrowth. The detailed mechanism will required further investigation. If the process can be understood and properly control, it may lead to a new methodology to create complex nanostructures. §2.4 Electronic Structure of Nanocrystal Quantum Dots. §2.4.1 Weak and Strong Confinement Regimes In the previous section we discussed the synthesis of nanocrystals. A nanocrystal quantum structure is a nanometer sized material A (the nanocrystal core) embedded in a higher bandgap material B (organic surfactants or shell layer). Such a structure is schematically shown in figure 2.10. Here we explore the basic electronic structure and optical transition of these nanocrystal quantum structures. Like their growth behavior discussed before, the optical transitions also show finite-size effect which is anyway probably the most well-known finite size effect of any kind, 63 reflecting itself vividly via the colors exhibited by the nanocrystals. The characteristic wavelength for the optical transition is by definition related to the light emitting entity in the semiconductor: the electron-hole exciton. In bulk semiconductors an exciton is composed of an electron and a hole that weakly bind to each other by coulomb interaction and forms a hydrogen-like pair (Mott-Wannier exciton [2.36]). Two characteristic length scales of the exciton are: the exciton’s center of mass diffusion length and the exciton Bohr radius. The exciton diffusion length in the semiconductor is on the order of a micron depending on the temperature. E c (B) E c (A) E V (B) E V (A) Core Shell or Organic Surfactant E g (A) E g (B) Electron Energy Levels Hole Energy Levels 1S e 1P e 1S h 1P h (b) (a) A B E c (B) E c (A) E V (B) E V (A) Core Shell or Organic Surfactant E g (A) E g (B) Electron Energy Levels Hole Energy Levels 1S e 1P e 1S h 1P h (b) (a) A B Figure 2.10 (a) Schematic of a nanocrystal quantum dots structure. (b) Schematic of particle-in-a-box confinement potential arising from the alignment of the conduction and valence band edges of the core (A) and shell (B). The electron and hole energy states and wavefunction are schematically shown corresponding to the case of infinite confinement potential. 64 The Bohr radius of the exciton can be derived similar to the Bohr radius of a hydrogen atom (under the assumption that the energy surface for the electron and hole are spherical and nondegenerate): 053 . 0 0 2 2 × = = μ ε μ ε m e a B h nm (2.13) where the B a is the exciton Bohr radius, ε is the appropriate dielectric constant of the semiconductor, 0 m is the free electron mass, μ is the electron and hole reduced mass: 1 1 1 ) ( − − − + = h e m m μ , where e m and h m are respectively the electron and hole effective mass. A list of basic parameters and exciton radius of selected direct bandgap semiconductors is shown in Table 2.1. Table 2.1 Basic parameters for common III-V semiconductors (300K) [2.37] Lattice Constant (A) Band gap (eV) Electron effective mass ( ) / 0 m m e Hole effective mass ) / ( 0 m m h Exciton Bohr Radius (nm) Calculated Dielectric Constant (static) InP 5.869 1.35 0.08 0.6(hh) 0.089(lh) 9.4 12.5 InAs 6.0584 0.36 0.023 0.41(hh) 0.026(lh) 36.8 15.15 InSb 6.479 0.17 0.014 0.43(hh) 0.015(lh) 65.5 16.8 GaAs 5.6533 1.42 0.063 0.51(hh) 0.082(lh) 12.2 12.9 Note: Exciton Bohr Radius is calculated using (2.13) and heavy hole effective mass With respect to the exciton radius and diffusion length, the nanocrystal can be divided into two size regimes. In the first regime, the nanocrystal size is smaller than the exciton diffusion length but a few times larger than its Bohr radius, which is 65 referred to as “Weak Confinement” regime. In this case a hydrogen-like exciton can be considered as the zero th order approximation to the exciton in the nanocrystal. The confined center of mass motion of the exciton and the corresponding energy shift can be added as a perturbation. In the weak confinement regime the ground state exciton transition in the nanocrystal experiences a high energy shift: 2 2 , 2 2Ma E l n χ h = Δ (2.14) where M is the sum of electron and hole effective mass, a is the nanocrystal radius. l n, χ are the n th roots of spherical Bessel function ) (x j l . n and l describes the states connected with the exciton center-of-mass motion in the nanocrystal (1s, 1p… 2s, 2p etc.) The exciton energy shift in the weak confine region is rather small, typically less than a few meV. In the second regime, the nanocrystal size is comparable or smaller than the exciton Bohr radius. This situation is typically referred to as “Strong Confinement” regime. In the strong confinement regime, the bulk hydrogen-like exciton as a starting approximation to describe the exciton in the nanocrystals breaks down. In fact as calculation will show in §2.3.2 kinetic energy of the electron and the hole due to confinement now becomes larger than the coulomb binding between them. Therefore the individual motion of an electron and a hole should be considered as the zero th order approximation, and the coulomb interaction between them can be added as a perturbation. The strong confinement significantly changes the optical properties of the nanocrystal and gives rise to effects such as size-tunable optical transition and atom-like δ-function energy states, which of course are the primary reasons that 66 started the whole field of nanocrystal quantum dots. The rest of this chapter is then devoted to a discussion of the electronic structure of the nanocrystals in the strong confinement regime. §2.4.2 Particle-in-a-Box model The nanocrystals in the strong confinement regime are often called quantum dots (QD) or quantum boxes. The name itself implies a commonly applied physical model to depict the electronic structure of the nanocrystals: the “particle-in-a-box” model [2.38, 2.39]. At first glance, such a model seems inappropriate, since there is no empty box, the electrons are facing a complex potential created by the atoms in the nanocrystal. The “particle-in-a-box” picture is in fact tied to a basic method to treat the semiconductor electronic structure: the effective mass approximation [2.36]. In a periodic bulk solid, according to Bloch theorem, the electronic wavefunctions take the form: ) ( ) exp( ) ( , , r u r k i r k n k n r r r r r r ⋅ = ψ (2.15) where n is the index of the band and k r is the wave vector which identifies these wavefunctions. ) ( , r u k n r r is the Bloch function that has the same translational symmetry as the solid. Generally the energy (E) of these wavefunctions as a function of k r in each band is not a simple function of k r . However within the effective mass approximation, the energy bands are approximated to have simple parabolic form near their extrema (band edge). Consider a simple direct bandgap semiconductor with only one conduction band and one valence band, both with their band edge at 67 0 = k r . In the effective mass approximation, the conduction band energy C k E r and valence band energy V k E r are approximated as, e eff g C k m k E E 2 2 2 h r + = (2.16) h eff V k m k E 2 2 2 h r − = (2.17) where g E is the bandgap, e eff m is called the effective mass of the electron in the conduction band, and h eff m the effective mass of the hole in the valence band. Quantitatively e eff m , h eff m are the curvatures of the conduction band and the valence band at near 0 = k r . The effective mass approximation incorporates the effect of the complex crystal potential felt by the carriers in the bulk semiconductor into the effective mass of the electron and the hole and is expected to be reasonable for small departures in k r from the extrema of the true band energies. Now consider the case of the nanocrystal as schematically shown in figure 2.10(a), which comprises a small bandgap core semiconductor A surrounded by a high bandgap semiconductor shell B. We assume that the effective mass of the electron and hole in the finite-size core and the shell are the same as in ideal infinite crystal (since the dimension of the core and shell are still much large than a unit cell). Since as given in (2.16) and (2.17), effective mass approximation treat the electrons and holes effectively as free particles whose total energy is their kinetic energy displaced by the relevant band 68 edge energy, the motion of electron and hole in the core/shell nanocrystal (figure 2.10(a)) then can be reduced to “particle-in-a-box”. The potential of the box is defined by the conduction band edge for electrons (or the valence band edge for holes) of the core A and shell B as schematically shown in figure 2.10(b). Take the InAs core nanocrystal for example. They are spherically shaped. The bulk InAs has a small bandgap 0.36eV (300K) which is embedded in the much larger bandgap of the surfactants (typically ~4.5eV) [2.40]. For simplicity we assume for a moment the electron and hole can be treated as in an infinite spherical potential well of radius a (a is the nanocrystal radius), ⎩ ⎨ ⎧ > ∞ < = a r a r r V 0 ) ( (2.18) Given the symmetry of potential, the Schrödinger equation solved for this potential leads to simple atomic-like states labeled by the principal, angular momentum, and magnetic quantum numbers n, l, m. The corresponding energy levels and wavefunctions for the electrons and the holes are: 2 2 , 2 2 , 2 , 2 2 a m E m k E E e eff l n g e eff l n g e l n χ h h + = + = , r Y r k j C r m l l n l e nlm ) , ( ) ( ) , , ( , φ θ φ θ ψ = (2.19) 2 2 , 2 2 , 2 , 2 2 a m m k E h eff l n h eff l n h l n χ h h − = − = , r Y r k j C r m l l n l h nlm ) , ( ) ( ) , , ( , φ θ φ θ ψ = (2.20) where a k l n l n , , χ = is the wave vector quantized by the spherical potential boundary condition. l n, χ is the n th root of the spherical Bessel function ) (x j l . ) , ( φ θ m l Y is the spherical harmonic function. The ground and first excited electron and hole energy 69 levels and the nature of the attendant wavefunctions are schematically shown in figure 2.10(b). The optical transition energy or the total energy of an electron-hole exciton is: coulomb h e exciton E E E E + − = (2.21) where coulomb E is the coulomb interaction between the electron and holes which so far we have ignored. In a strong confinement regime, the coulomb energy is smaller than the confinement related energy and therefore can be added as a first order energy correction: r d r d r r e r r E h l m n e nml coulomb ′ − = ∫ r r r r r r ' ) ' ( ) ( 2 2 ' ' ' 2 ε ψ ψ (2.22) where ε is the dielectric constant of the nanocrystal and ) (r e nml r ψ , ) ( ' ' ' r h l m n r ψ are, respectively, the electron and hole wavefunctions. For the ground state exciton (1S electron-1S hole), the integration gives [2.39]: 2 2 8 . 1 ) 1 1 ( a e S S E h e coulomb ε ≈ (2.23) As a numerical example, for a 5nm diameter InAs nanocrystal, the confinement induced kinetic energy for a ground state exciton (1S e -1S h ) is on the order of 1eV, much larger than the coulomb energy 68meV estimated from Equ. (2.23). This justifies the treatment of the coulomb interaction as a perturbation. Nevertheless, the coulomb energy is by no means negligible. In fact its contribution to such a strong confined exciton is far greater than the coulomb contribution to a 70 hydrogen-like exciton in bulk InAs (1.3meV). This is understandable given the large overlap of electron and hole wavefunctions due to the confinement. Although the “particle-in-a-box” model provides a straightforward understanding of the nanocrystal quantum dots and qualitatively explains many of their properties such as size-tunable optical transition and atom-like δ-function energy states, etc, it is too crude for any quantitative use. The 1 st exciton (1S e -1S h ) transition energy in InAs nanocrystals (300K) according to the “particle-in-a-box” model corrected for coulomb interaction is (combining equ 2.19-2.23): 2 2 2 2 2 2 2 2 2 2 . 17 168 . 0 36 . 0 8 . 1 2 2 ) 1 1 ( a a a e a m a m E S S E h eff e eff g h e + − = − + + = ε π π h h eV (2.24) where a is the nanocrystal radius in nm. Compared to the experimentally measured 1S e -1S h transition energy, the calculated value is significantly overestimated, not even the order of magnitude is right. The deficiency lies partly in the oversimplified band structure and the neglect of an essential conceptual consideration – the mixing of different bulk bands due to the confinement potential. In the next section, we discussed the refined quantitative models for nanocrystal electronic structure. §2.4.3 p k r r ⋅ calculation of the nanocrystal electronic structure Beyond the “particle-in-a-box”, a few advanced methods have been applied to calculate the electronic structure of semiconductor quantum dots (nanocrystal or self-assembled) including: p k r r ⋅ method [e.g. 2.41, 2.40, 2.42], tight-binding method [e.g. 2.43-2.45], empirical pseudopotential [e.g. 2.46, 2.47] method. We discuss the 71 eight band p k r r ⋅ calculation of the nanocrystal quantum dots (NCQD) electronic structure motivated by the following reasons: (1) The majority of the optical and electric properties of the NCQDs are related to electronic structure near the conduction band and the valence band extrema, for which the p k r r ⋅ method can indeed give reasonable description. (2) p k r r ⋅ forms the basis on which the effective mass approximation stands, so some of the concepts discussed previously will be extended and clarified. Similar to effective mass approximation, p k r r ⋅ describes the system with only a few parameters, and is helpful for qualitatively understanding of experimental results. (3) Finally, the p k r r ⋅ method can fully exploit the symmetry of the nanocrystals and therefore can be handled with minimal computational power. A good detailed description of the p k r r ⋅ method as applied to bulk semiconductors and low dimension heterostructures can be found in Dr. Li Chen’s dissertation [2.48] in our group. Here we first briefly recapture the key concepts of this method and its application towards calculation of electronic structure of quantum nanostructures (within the envelop wavefunction approximation). Then we discuss a specific implementation of p k r r ⋅ suitable for the spherical narrow band NCQDs – the 8 band Pidgeon and Brown model [2.49]. Finally, as a numerical example, results for InAs nanocrystals are presented and discussed as a prelude to chapter 5 where the optical properties of these InAs nanocrystals will be discussed. 72 §2.4.3(a) The p k r r ⋅ framework The p k r r ⋅ method was developed by Luttinger and Kohn [2.50, 2.51], and Kane [2.52]. The essence of the method is: for a crystalline solid, knowing the 0 = k r (zone center) wavefunction 0 , n u and energy 0 , n E for all bands, calculate the wavefunction k n r , ψ and energy k n E r , of states for which k r is near 0 using perturbation method. The wavefunctions of electrons in a periodic semiconductor are eigenfunctions of Schrödinger equation for single particle Hamiltonian: ) ( ) ( ) ˆ ( 4 ) ( 2 ˆ ) ( ˆ , , , 2 2 0 0 2 , r E r p V c m r V m p r H k n k n k n k n r r r r h r r r r r r ψ ψ σ ψ = ⎥ ⎦ ⎤ ⎢ ⎣ ⎡ ⋅ × ∇ + + = (2.25) where ) (r V r is the periodical crystal potential, and the σ r r h ⋅ × ∇ ) ˆ ( 4 2 2 0 p V c m term in the Hamiltonian arises from the electron’s spin-orbit coupling. ∇ − = r h i p ˆ is the momentum operator, σ r is the Pauli spin matrix. Substituting ) ( ) exp( ) ( , , r u r k i r k n k n r r r r r r ⋅ = ψ into (2.21) and considering that the crystal momentum k r h is small compared to the atomic momentum ) ( ˆ , r u p k n r r in the atomic interior, we obtain: ) ( ) ( ) ˆ ( 4 ˆ ) ( 2 2 ˆ , , , 2 2 0 0 0 2 2 0 2 r u E r u p V c m p k m r V m k m p k n k n k n r r r r h r h r h r r r = ⎥ ⎥ ⎦ ⎤ ⎢ ⎢ ⎣ ⎡ ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎝ ⎛ ⋅ × ∇ + ⋅ + ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎝ ⎛ + + σ (2.26) 73 The Hamiltonian in (2.26) can be separated into ) ( 2 2 ˆ ˆ 0 2 2 0 2 0 r V m k m p H r h + + = and (2.27) σ r r h r h ⋅ × ∇ + ⋅ = ) ˆ ( 4 ˆ ˆ 2 2 0 0 1 p c m p k m H (2.28) Note 0 , n E and ) ( 0 , r u n r are the eigenvalues and eigenfunctions of 0 ˆ H : ) ( ) ( ˆ 0 , 0 , 0 , 0 r u E r u H n n n r r = (2.29) Near 0 = k r , 1 ˆ H is small compared to 0 ˆ H and hence can be considered as a perturbation to 0 ˆ H . k n E r , and ) ( , r u k n r r in (2.26) can be solved using perturbation theory. ) ( , r u k n r r can be expanded using zone center Bloch function )} ( { 0 , r u n r from all bands as the zeroth order basis: ) ( ) ( ) ( 0 , , , r u k C r u j j j n k n r r r r ∑ = (2.30) or ) ( ) exp( ) ( ) ( 0 , , , r u r k i k C r j j j n k n r r r r r r ∑ ⋅ = ψ (2.31) where ) ( , k C j n r are the expansion coefficients. In fact it can be strictly proved that an arbitrary Bloch function ) ( , r u k n r r can always be expanded in terms of the set )} ( { 0 , r u n r since )} ( { 0 , r u n r is complete for functions of the same translational symmetry as the solid [2.50]. 74 In an actual calculation for a specific ) ( , r u k n r r however, the zero th order perturbation basis )} ( { 0 , r u n r does not necessarily have to cover all bands, but only those m bands that contribute to ) ( , r u k n r r ”most significantly”. Using the corresponding m zone center Bloch functions as the basis, the Hamiltonian in (2.26) can be written as an m m × matrix: [] m j i j i i ij j p c m p k m m k E i H .. 1 , 2 2 0 0 , 0 2 2 0 , 0 , ) ˆ ( 4 ˆ ) 2 ' ( 0 , = ⎥ ⎦ ⎤ ⎢ ⎣ ⎡ ⋅ × ∇ + ⋅ + + = σ δ r r h r h h (2.32) which is referred to as the p k r r ⋅ Hamiltonian hereafter. According to degenerate perturbation theory, the first order solution of (2.26) reduces to the eigenvalue problem of the p k r r ⋅ Hamiltonian. Furthermore the contributions of the rest of the bands (referred to as remote bands) that is not included in the m perturbation basis are often incorporated into the p k r r ⋅ Hamiltonian via second order perturbation. These second order perturbation terms often appear in the diagonal matrix elements unless the bands they couple are degenerate at 0 = k r . For direct bandgap III-V zinc-blende semiconductors, their schematic band diagram near zone-center is shown in figure 2.11. Each of E-k curve in figure 2.11 corresponds to two degenerated bands (with respect to z angular momentum, see table 2.2). The two lowest conduction bands C 6 Γ arise from the group III s-orbital. The six highest valence bands V 8 Γ (light hole and heavy hole, four fold degenerate at k=0) and V 7 Γ (spin split-off hole double degenerate) arise from group V p-orbital. 75 The above 8 bands are often the “most significant” bands to form the p k r r ⋅ Hamiltonian. The rest of the bands such as V 6 Γ arising from group V s-orbital and C 7 Γ and C 8 Γ arising from group III p-orbital are often incorporated via second order perturbation. The Bloch functions and other properties of the eight most significant bands (2 conduction and 6 valence see above) are summarized in Table 2.2. v 7 Γ v 8 Γ v 6 Γ c 7 Γ c 8 Γ c 6 Γ 0 Δ g E E k v 7 Γ v 8 Γ v 6 Γ c 7 Γ c 8 Γ c 6 Γ 0 Δ g E E k Figure 2.11 Schematic of the band structure near k=0 in a direct bandgap semiconductor of zinc-blende structure. 76 Table 2.2 Notation and zone center Bloch function of “most significant” bands. Band Z J J, V C Jz J u , , Group Notation ( ) 0 = k E r 2 1 , 2 1 ↑ S Conduction Band 2 1 , 2 1 − ↓ S 6 Γ E g 2 3 , 2 3 ↑ + ) ( 2 1 iY X Heavy Hole (hh) or (v2) 2 3 , 2 3 − ↓ − ) ( 2 1 iY X 2 1 , 2 3 ↑ − ↓ + Z iY X 3 2 ) ( 6 1 Light Hole (lh) or (v1) 2 1 , 2 3 − ↓ − ↑ − − Z iY X 3 2 ) ( 6 1 8 Γ 0 2 1 , 2 1 ↑ + ↓ + Z iY X 3 1 ) ( 3 1 Spin Split- off Hole (so) or (v3) 2 1 , 2 1 − ↓ + ↑ − − Z iY X 3 1 ) ( 3 1 7 Γ Δ − 77 Given the different choices of the “most significant” bands and “remote” bands, different p k r r ⋅ Hamiltonians can be constructed. For example Luttinger’s 6x6 Hamiltonian [2.50, 2.51] that includes the 6 highest valence bands and counts other bands using second order perturbation (Notation adapted from Ref 2.53 Chang 1988) is: . ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎦ ⎤ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎣ ⎡ Δ − − − Δ − − − + − − − − + = P L Q L M P M L Q L L M Q P L M Q L L Q P M L Q M Q P L M L M L Q P H 0 2 / 2 2 / 3 2 0 2 2 / 3 2 2 / 2 / 2 0 2 2 / 3 0 2 / 3 2 0 2 2 / 0 ' * * * * * * * * * (2.33) where Δ is the spin-orbit splitting 0 2 1 2 2 / m k P L γ h − = , 0 2 2 2 2 2 2 / ) 2 ( m k k k Q z y x L − + − = γ h , (2.34) 0 3 2 / ) ( 3 m k ik k L z y x L − = γ h , 0 3 2 2 2 2 2 2 / ] 3 2 ) ( 3 [ m k k i k k M y x L x y L γ γ h h + − − = and L 1 γ , L 2 γ , L 3 γ are the Luttinger parameters, typically treated as fitting parameters to fit experimental results such as the zone center hole effective mass. The Luttinger model gives respectable valence band structure of III-V semiconductors. §2.4.3(b) Envelop Wavefunction Approximation and p k r r ⋅ in Heterostructures. In a NCQD (or heterostructures in general), however, the wave vector k r is no longer a good quantum number. The p k r r ⋅ method has thus to be applied in real space using a so-called envelop wavefunction approach. An arbitrary wavefunction 78 ) (r r Φ in the periodic solid can be expanded in terms of the Bloch form eigenfunctions )} ( { , r k n r r ψ : ∑ = Φ k n k n n r k b r r r r r r , , ) ( ) ( ) ( ψ (2.35) where ) (k b n r are the expansion coefficients. Combine (2.31) and (2.35), ) (r sp r Φ can be expanded in terms of zone center Bloch function from all bands: ) ( ) ( ) ( ) exp( ) ( ) ( ) ( 0 , 0 , , , r u r f r u r k i k C k b r j j j j k n j j n n sp r r r r r r r r ∑ ∑∑ = ⋅ ⋅ = Φ (2.36) where ∑ ⋅ ⋅ = k n j n n j r k i k C k b r f r r r r r , , ) exp( ) ( ) ( ) ( is the envelop wavefunction. Again typically the expansion only uses a limited number of “most significant” m bands. Following the work of Bastard et al [2.54, 2.55], two key assumptions are made for heterostructures such as an NCQD made of two component materials, core A and shell B (figure 2.10(a)): (1) The wavefunction in both core and shell can be expand as: ∑ = Φ ) ( ) ( ) ( , 0 , , , r u r f r B A n B A n B A sp r r (2.37) There must be a one to one correspondence between each band and Bloch function in the material A and material B. (2) The zone center Bloch function in A and B are assumed to be the same: ) ( ) ( 0 , 0 , r u r u B n A n r r = (2.38) 79 Under these two assumptions, p k r r ⋅ Hamiltonian for the NCQDs (or other heterostructures) can be written in real space as a m m × matrix of operators similar to the k space p k r r ⋅ Hamiltonian (such as the Luttinger Hamiltonian) for bulk material except: (1) z y x k , , r is replaced by momentum operator z y x z y x i p , , , , ˆ 1 ∇ − = r h with anti- symmetrization consideration and (2) all parameters (such as Δ, L 1 γ ...) in the Hamiltonian can become space dependent as ) (r r Δ ,) ( 1 r L r γ in the presence of chemical composition or strain variation. Before finally moving to the calculation of nanocrystal electronic levels, a few comments on the applicability of the p k r r ⋅ method and envelop wavefunction approach are in order. Since the p k r r ⋅ approach is a perturbative method using } { 0 , n u as the basis, it obviously provides the best description of electronic structure near the zone center 0 ≈ k . (Here we do not consider the type of p k r r ⋅ calculation generalized to indirect bandgap semiconductors.) For heterostructures it thus requires that the envelop wavefunction varies on a spatial scale much larger than the dimension of the unit cell. Thus the p k r r ⋅ works better for larger NCQDs and for slowly varying states having low energy close to the band edges. The envelop wavefunction approach assumes certain similarity between the materials involved in the heterostructures. The different materials in the 80 heterostructures should have one-to-one correspondence between Bloch functions, which means that the different materials should have the same crystal structure and chemical bond. Equ. 2.38 also requires the zone center Bloch function in different materials to be the same. These assumptions are reasonable in epitaxially grown semiconductor nanostructures. In the typical colloidal nanocrystal with organic surfactants these assumptions sound ridiculous. The relief however is that the organic surfactants typically have a larger bandgap (or, better to say HOMO-LUMO gap), typically >4eV. Qualitatively the electron wavefunctions of the lower energy states do not penetrate into the surfactants too much, therefore their properties become unimportant for the calculation of energy level in the core. Nevertheless the treatment of this semiconductor/organic interface is one of the most challenging problems in the calculation of (especially small) NCQD electronic structure and will be further commented upon as the calculation is presented. §2.4.3(c) The 8 band p k r r ⋅ formalism for spherical quantum dots and calculation of InAs electronic structure. The “particle-in-a-box” model in §2.4.2 can be viewed as a simplified version of the p k r r ⋅ calculation. The wavefunction calculated in Equ. 2.19 and 2.20 are actually the envelop part of the carrier’s wavefunction. Compared to the one conduction and one valence band semiconductor assumed in §2.4.2, a refined version of “particle-in-a-box” can be obtained by including all of the three bulk valence bands (heavy and light hole, split-off hole) as simple parabolic bands of 81 corresponding effective masses. Thus each valence band will lead to a ladder of “particle-in-a-box” states for the holes. However simple calculation will show even such a refined model does not fit the experimental optical transition data at all. Compared to the p k r r ⋅ , the fundamental feature missing in such a model is the non- diagonal elements in p k r r ⋅ Hamiltonian that represent the coupling (or mixing) between different bands under the spherical confinement potential in the quantum dot. Indeed under the spherical confinement, the angular momentum of the zone center Bloch functions i J ( ½ for conduction band, 3/2 for heavy, and light hole and ½ for split-off hole) solved for each bulk band, and the angular momentum of envelop wavefunction i L solved for each band are no longer good quantum numbers [2.56, 2.57]. It can be shown under “the spherical approximation” (nanocrystal have a spherical shape and the wrapping of valence band connected to the cubic symmetry of the zinc blende structure is neglected), that each electron or hole state can be described by the following good quantum numbers: the total angular momentum i i i L J F + = , its z projection Fz and the parity [2.56, 2.57]. Each of these states generally has contribution from all of the conduction and valence subbands. The general formalism for the application of the p k r r ⋅ method in symmetric heterostructures (cylindrical and spherical) was proposed by Sercel and Vahala [2.41, 2.57]. Specifically for CdSe nanocrystals, Ekimov et al [2.58] have applied the 6 band Luttinger Hamiltonian which includes the mixing between the “most significant” highest 6-valence bands to calculate the hole energy levels, and the 82 single band effective mass to calculate the electron energy levels, generating good agreement with optical data. The mixing between conduction and valence band states in CdSe is negligible given their large energy separation (CdSe bandgap 2.4eV). Therefore the electron energy level can be calculated separately from hole levels. However, for narrow bandgap III-V nanocrystals such as InAs, the mixing between conduction band and valence band is expected to be significant and needs to be accounted. To calculate the energy levels in InAs nanocrystals Banin et al [2.40] applied the 8 band Luttinger-Kohn (LK) Hamiltonian originally formulated by Sercel et al [2.41] for spherical heterostructures. Compared to the Luttinger Hamiltonian, the LK Hamiltonian adds the 2 lowest conduction bands in the “most significant” bands, thus accounting for the mixing between conduction and valence bands. The use of LK Hamiltonian in narrow bandgap materials was first proposed by Pidgeon and Brown in their calculation of Landau levels in InSb [2.49], and is thus often referred to as the PB model. Using the same PB model Efros et al have studied the effect of band mixing [2.59] in a range of narrow bandgap to wide bandgap nanocrystals (InSb, InP, CdTe, CdS). Below we apply the PB model to solve the electronic structure of the InAs nanocrystals. The 8 band LK Hamiltonian under spherical approximation has the form: (following the notation of Efros et al [2.59]) 83 c u 2 1 v u 2 3 2 3 , c u 2 1 − v u 2 1 2 3 , v u 2 1 2 3 , − v u 2 3 2 3 , − v u 2 1 2 1 , v u 2 1 2 1 , − c u 2 1 v u 2 3 2 3 , c u 2 1 − v u 2 1 2 3 , v u 2 1 2 3 , − v u 2 3 2 3 , − v u 2 1 2 1 , v u 2 1 2 1 , − 2 0 2 p m E g α + 2 0 2 p m E g α + 0 0 0 0 + p V i K 2 − p V i K 6 z K p V i 3 − p V K 3 1 0 z K p V 3 2 + − p V K 6 1 z K p V i 3 2 − − p V K 2 1 + p V i K 3 z K p V 3 1 − − − p V i K 2 ) ( Q P + − L − M − 0 L i 2 1 − Q i 2 z K p V 3 2 − − p V K 6 1 * L − ) ( Q P − − 0 M − M i 2 L i 2 3 − + − p V i K 6 z K p V i 3 2 − * M − 0 ) ( Q P − − L * 2 3 L i Q i 2 0 + − p V K 2 1 0 * M − * L ) ( Q P + − * 2M i * 2 1 L i z K p V i 3 − − − p V i K 3 * 2 1 L i Q i 2 − L i 2 3 − M i 2 − P − Δ − P − Δ − 0 0 L i 2 1 − Q i 2 − * 2 3 L i * 2M i − z K p V 3 1 − + p V K 3 1 c u 2 1 v u 2 3 2 3 , c u 2 1 − v u 2 1 2 3 , v u 2 1 2 3 , − v u 2 3 2 3 , − v u 2 1 2 1 , v u 2 1 2 1 , − c u 2 1 v u 2 3 2 3 , c u 2 1 − v u 2 1 2 3 , v u 2 1 2 3 , − v u 2 3 2 3 , − v u 2 1 2 1 , v u 2 1 2 1 , − 2 0 2 p m E g α + 2 0 2 p m E g α + 0 0 0 0 + p V i K 2 − p V i K 6 z K p V i 3 − p V K 3 1 0 z K p V 3 2 + − p V K 6 1 z K p V i 3 2 − − p V K 2 1 + p V i K 3 z K p V 3 1 − − − p V i K 2 ) ( Q P + − L − M − 0 L i 2 1 − Q i 2 z K p V 3 2 − − p V K 6 1 * L − ) ( Q P − − 0 M − M i 2 L i 2 3 − + − p V i K 6 z K p V i 3 2 − * M − 0 ) ( Q P − − L * 2 3 L i Q i 2 0 + − p V K 2 1 0 * M − * L ) ( Q P + − * 2M i * 2 1 L i z K p V i 3 − − − p V i K 3 * 2 1 L i Q i 2 − L i 2 3 − M i 2 − P − Δ − P − Δ − 0 0 L i 2 1 − Q i 2 − * 2 3 L i * 2M i − z K p V 3 1 − + p V K 3 1 (2.39) where the form of the basis zone center Bloch function of the conduction band and valence band V C Jz J u , , is provided in Table 2.2. The Hamiltonian contains momentum operators: z y x z y x i p , , , , ∇ − = h : y x ip p p ± = ± , 2 2 2 y x p p p + = ⊥ 0 2 1 2 / m p P γ = , 0 2 2 2 / ) 2 ( m p p Q z − = ⊥ γ , (2.40) 0 2 / 3 m p p i L − ± − = γ , 0 2 2 / 3 m p M − = γ The relevant p k r r ⋅ parameters in the Hamiltonian are summarized in Table 2.3. In the LK Hamiltonian (2.39) the constant Y p y V p x V X c m i x y ∂ ∂ − ∂ ∂ = Δ 2 2 0 4 3 h 84 Table 2.3: 0K p k r r ⋅ parameters for III-V semiconductors InAs (a) InP (b) InSb (c) GaAs (d) g E (eV) 0.418 1.42 0.237 1.52 p E (eV) 22.2 20.4 23.4 25.7 Δ (eV) 0.38 0.13 0.810 0.34 * c m () 0 m 0.023 0.080 0.014 0.067 L 1 γ 19.67 6.28 36.41 7.65 L 2 γ 8.37 2.08 15.96 2.41 L 3 γ 9.29 2.76 16.99 3.28 1 γ 2.53 1.49 3.44 2.01 γ 0.35 0.094 0.196 0.11 α 1.65 -1.46 -0.36 -0.85 (a) From [2.62] (b) from (c)from [2.59] (d) from [2.63] represents valence band spin-orbit splitting [2.52]. The magnitude of coupling between the conduction band and valence band is described by the Kane matrix element 0 0 / 2 m Z p S i m E V z P K − = = [2.52]. Just like Luttinger Hamiltonian, in LK Hamiltonian the contribution from remote band is included as second order perturbation. Parameter α describes the remote band contribution to the electron effective mass. This contribution and contribution from valence bands (expressed by Kane matrix element V and bandgap) together decide the electron effective mass c m : ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎝ ⎛ ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎝ ⎛ Δ + + + = g g p c E E E m m 1 2 3 1 1 0 α (2.41) Therefore α can be calculated knowing c m and other p k r r ⋅ parameters. 85 The parameters 1 γ and γ account for the contribution from the remote bands to the hole effective mass. In this case the notion of remote band does not include the two lowest conduction band. (In contrast the usual 6 band Luttinger Hamiltonian count the two lowest conduction bands as remote bands). The parameters 1 γ and γ are connected to the commonly used Luttinger parameter to 1 st order approximation by [2.60, 2.61]: g P L E E 3 1 1 + = γ γ , g P L E E 6 5 ) 3 2 ( 3 2 + + = γ γ γ (2.42) The wavefunction of these electron and hole states in the spherical nanocrystal can be expanded into a product of the eight zone center Bloch functions and corresponding envelop wavefunctions. Such states are characterized by their total angular momentum L J F + = , its z projection Z F and parity, in which J and L are the angular momenta of the zone center Bloch function and the envelop wavefunction: ∑ ∑∑ ∑ − = = − = ± − = ± ± Ω + Ω + Ω = 2 1 2 1 , 2 1 2 , 1 2 3 2 3 , 2 3 2 1 2 1 , ) ( ) ( ) ( Jz v Jz S Jz s i Jz v Jz hi Jz hi Jz C Jz C Jz C Fz F u r R u r R u r R ψ (2.43) where + Fz F , ψ and − Fz F , ψ are, respectively, the wavefunctions for even and odd states. ) ( , 1 , 1 , r R S h h C r ± are the radial part of the envelop functions. S h h C Jz , 2 , 1 , Ω are the angular part of the envelop function whose analytical expression is derived in [2.64] and listed in Appendix B. Substituting the above wavefunction into the Luttinger-Kohn Hamiltonian (equ. 2.39) one obtains two sets of four coupled one-dimensional second order 86 differential equations respectively for the radial function for even and odd states: ) ( , 1 , 1 , r R S h h C + , ) ( , 1 , 1 , r R S h h C − . We set the boundary condition to be 0 ) ( , 1 , 1 , = ± a R S h h C , (2.44) namely the wavefunction go to zero at interface between the InAs and the TOP surfactant. Note the treatment of boundary condition is different from [2.40] in which bandgap and p k r r ⋅ parameters are assumed for the organic layer and boundary condition is chosen to connect the wavefunction at the interface. Our choice of the boundary condition, though crude, avoids the arbitrariness in the choice of the parameter values. Results: Numerically solving the sets of differential equation for even and odd states using the above noted boundary condition we arrive at the energy levels shown in figure 2.12. Here we use the standard notation for atomic energy levels: nQ F , where n is the main quantum number, Q is the lowest L in wavefunction, and F is the total angular momentum. As a prelude to the discussion of our optical measurement in chapter 5, we note that the lowest electron level is 1S ½ state. The two next higher levels are 1P electron states: 1P 3/2 and 1P ½ close to each other. The lowest hole level 1VB is 1S 3/2 states. Without going into the detail of assigning the hole excited states, we simply note that compared to the electron, the excited states for hole levels are much denser in energy, which is not surprising due to the larger hole effective mass and the complex hole band structure. 87 (a) (a) Figure 2.12(a) Calculated energy levels of states in InAs nanocrystal as a function of its radius. Solid line: energy of states of total angular momentum F=3/2, Dotted line: energy of states of total angular momentum F=1/2. 88 (b) (b) Figure 2.12 Continued (b): Same plot as figure 2.12(a) except zoomed into InAs nanocrystal radius range 2.0-4.5nm for clarity. 89 The 0K optical transition energy of ground state exciton 1VB (1S 3/2 (h)) to 1S 1/2 (e) and excited exciton 1VB (1S 3/2 (h)) to 1P 3/2 (e) is calculated by subtracting corresponding electron and hole energy. Further the coulomb energy is also corrected to the first order using equ. (2.23) 2 2 8 . 1 a e E coulomb ε ≈ . The calculated results are plotted in figure 2.13 and compared against the measured optical transition energies as adapted from [2.65]. From this comparison, our calculated optical transitions fit the measured 1VB-1Se and 1VB-1Pe transition reasonably for nanocrystal >5nm in diameter with error of tens of meV. However when the nanocrystals get smaller, the calculation deviates significantly and fails to predict the optical transition. This deviation represents a typical failure of p k r r ⋅ applied to small nanocrystals in the literature. The possible reasons are the noted below. First is the limitation of the applicability of the p k r r ⋅ method. As a perturbative method at k~0, p k r r ⋅ works best for small k values and thus requires the envelop wavefunction to vary very slowly on the scale of the lattice constant. For a nanocrystal of radius 2 to 3 times the lattice constant, we may have already hit the limitation of the p k r r ⋅ method by involving too large k components. Therefore a method that can describe the whole Brillioun zone such as tight-binding should be used for the electronic structure of the small nanocrystals. The second problem is the treatment of the interface. Here our boundary condition demands the wavefunction goes to zero at the interface. In reality of course, the surfactants represents only a finite barrier, and a portion of the electron and hole 90 Figure 2.13 8 band p k r r ⋅ calculated InAs nanocrystals optical transition energy as a function of diameter and comparison to experimental data from Ref 2.65. wavefunction will penetrate. It can be expected that the deviation caused by our treatment of the boundary condition will be more significant for electron levels (small effective mass) and for high excited states. Indeed in figure 2.13, the calculated 1VB-1Pe transition deviates more from the experiment compared to the ground state transition. One refinement could be to assume a finite barrier for the surfactant (or in general the surrounding of the nanocrystals) and use the continuity of the wavefunction and its derivative (or better yet, electron current) as the boundary condition. In fact, as commented by Norris [2.66], often in the literature 91 the barrier height of the surrounding is treated as a fitting parameter to achieve the best compliance of the calculation to the experiments. Furthermore, there is no reason why the interface should be treated as an abrupt potential change when the nanocrystal is just a few times larger than the lattice constant itself. More sophisticated “general boundary condition theory” has been developed to resolve the issue [2.67]. Nevertheless, so far the correct treatment of the interface is by and large an open question. The above calculation is of course for energy levels and optical transitions at 0K. The finite temperature counterparts, in the first order can be obtained by simply correcting for the InAs bandgap change with temperature: E g (InAs) = 0.415 - 2.76·10 -4 ·T 2 /(T+83) (eV) [2.68] Such derived room temperature energy levels and optical transition will be used in the discussion of optical measurement of energy transfer in and out of the nanocrystal in chapter 5 (of course only for larger nanocrystals (>5nm) where p k r r ⋅ works reasonable well). 92 Chapter 2 References: [2.1] A. Ekimov, A. L.Efros, A. A. Onushchenko, ”Quantum size effects in semiconductor microcrystals”, Solid State Commun. 56, 921-924 (1985). [2.2] R. Rossetti, L. Brus, “Electron-hole recombination emission as a probe of surface chemistry in aqueous CdS colloids”, J. Phys. Chem. 86, 4470-4472 (1982). [2.3] M. Meyer, C. Wallberg, K. Kurihara, J. H. Fendler, “Photosensitized charge separation and hydrogen production in reversed micelle entrapped platinized colloidal cadium sulfide. J. Chem. Soc. Commun. 88, 969-977 (1984). [2.4] M. G. Bawendi, A. R. Kortan, M. L. Steigerwald, L. E. Brus,“X-ray structural characterization of larger CdSe semiconductor clusters.“ J. Chem. Phys. 91, 7282- 7290 (1989). [2.5] M. L. Steigerwald, A. P. Alivisatos, J. M. Gibson, T. D. Harris, R. Kortan, A. J. Muller, A. M. Thayer, T.M. Duncan, D. C. Douglass, L. E. Brus, “Surface derivatization and isolation of semiconductor cluster molecules.” J. Am. Chem. Soc. 110, 3046-3050 (1988). [2.6] C. B. Murray, D. J. Norris, M. G. Bawendi, “Synthesis and characterization of nearly monodisperse CdE (E=S, Se, Te) semiconductor nanocrystallites.” J. Am. Chem. Soc. 115, 8706-8715 (1993). [2.7] M. A. Olshavsky, A. N. Goldstein, A. P. Alivisatos, ”Organometallic synthesis of GaAs crystallites exhibiting quantum confinement”, J. Am. Chem. Soc. 112, 9438- 9439 (1990). [2.8] H. Uchida, C. J. Curtis, A. J. Nozik, “GaAs nanocrystals prepared in quinoline.” J. Phys. Chem. 95, 5382-5384 (1991). [2.9] A. A. Guzelian, U. Banin, A. V. Kadavanich, X. Peng, and A. P. Alivisatos, ‘Colloidal chemical synthesis and characterization of InAs nanocrystals quantum dot”, App. Phys. Lett. 69, 1432-1434 (1996). [2.10] O. I. Micic and A. J. Nozik, “Synthesis and characterization of binary and ternary III-V quantum dots. J. Lumin. 70, 95-107 (1996). [2.11] M. A. Hines, P. Guyot-Sionnest, “Synthesis and Characterization of Strongly Luminescing ZnS-Capped CdSe Nanocrystals”, J. Phys. Chem. 100, 468-471 (1996) [2.12] X. Peng, M. C. Schlamp, A. V. Kadavanich, A. P. Alivisatos, “ Epitaxial growth of highly luminescent CdSe/CdS core/shell nanocrystals with photostability and electronic accessibility”, J. Am. Chem. Soc. 119, 7019-7029 (1997). 93 [2.13] B. O. Dabbousi, J. Rodriguez-Viejo, F. V. Mikulec, J. R. Heine, H. Mattoussi, R. Ober, K. F. Jensen, M. G. Bawendi, “(CdSe)ZnS core-shell quantum dots: Synthesis and characterization of a size series of highly luminescent nanocrystallites” J. Phys. Chem. B 101, 9463-9475 (1997) [2.14] Y. Cao, U. Banin, “Synthesis and characterization of InAs/InP and InAs/CdSe core/shell nanocrystals.” Angew. Chem. Int. Ed. 38, 3692-3694 (1999). [2.15] O. I. Micic, B. B Smith, A. J. Nozik, “Core-shell quantum dots of lattice mathced ZnCdSe 2 shells on InP cores: experiments and theory”, J. Phys. Chem. B, 104, 12149-12156 (2000). [2.16] M. Bruchez Jr. M. Moronne, P. Gin, S. Weiss, A. P. Alivisatos, “Semiconductor Nanocrystals as Fluorescent Biological Labels”, Science 281, 2013- 2015 (1998). [2.17] W. C. W. Chan and S. Nie, “Quantum Dot Bioconjugates for Ultrasensitive Nonisotopic Detection”, Science, 281, 2016-2018 (1998). [2.18] R. J. Errington, Advanced practical inorganic and metalorganic chemistry, Chapman & Hall, London (1997). [2.19] L. Manna, E. C. Scher, A. P. Alivisatos, ”Synthesis of soluble and processable rod-, arrow-, teardrop-, and tetrapod-shaped CdSe nanocrystals”, J. Am. Chem, Soc. 122, 12700-12706 (2000). [2.20] T. Sugimoto, “Preparation of Monodisperse Colloidal Particles”, Adv. Colloidal Interfac. Sci, 28, 65-108, (1987). [2.21] D. V. Talapin, A. L. Rogach, M. Haase, H. Weller, ”Evolution of an Ensemble of Nanoparticles in a Colloidal Solution: Theoretical Study”, J. Phys. Chem. B 105, 12278-12285 (2001). [2.22] X. Peng, J. Wickham, A. P. Alivisatos, “Kinetics of II-VI and III-V Colloidal Semiconductor Nanocrystal growth: “Focusing” of Size Distribution”, J. Am. Chem. Soc. 120, 5343 (1998). [2.23] V. K. LaMer, R. H. Dinegar, “Theory, Production and Mechanism of Formation of Monodispersed Hydrosols”, J. Am. Chem. Soc. 72, 4847-4854 (1950). [2.24] H. Reiss, “The Growth of Uniform Colloidal Dispersions”, J. Chem. Phys. 19, 482-487 (1951). [2.25] L. C. Liu, S. H. Bisbud, “Quantum dot size distribution analysis and precipitation stages in semiconductor doped glass”, J. Appl. Phys. 68, 28-32 (1990). 94 [2.26] J. E. B. Katari, V. L. Colvin, A. P. Alivisatos, J. Phys. Chem. 98, 4109-4117, (1994). [2.27] An elegant asymptotic solution had been developed by Lifshitz, Slyozov and Wagner (LSW theory) to describe the evolution of an ensemble of particles during Ostwald ripening: I. M. Lifshitz V. V. Slyozov, “Kinetics of precipitation from super saturated solid solutions”, J. of Phys. and Chem. of Solids, 19, 35-50 (1961). [2.28] J. E. B Katari, V. L. Colvin, A. P. Alivisatos, “X-ray photoelectron- spectroscopy of CdSe nanocrystals with application to studies of the nanocrystal surface”, J. Phys. Chem. 98, 4109-4117 (1994). [2.29] J. Heath, J. J. Shiang, “Covalency in semiconductor quantum dots”, Chem. Soc. Rev. 27, 65-71 (1998). [2.30] R. L. Wells, C. G. Pitt, A. T. McPhail, A. P. Purdy, S. Shafieezad, and R. B. Hallock, “Use of Tri(trimethylsilyl)arsine To Prepare Gllium Arsenide and Indium Arsenide”, Chem. Mater., 1, 4-6 (1989). [2.31] Y. Cao and U. Banin, “Growth and Properites of Semiconductor Core/Shell Nanocrystals with InAs Cores”, J. Am. Chem. Soc. 122, 9692-9702 (2000). [2.32] J. Wickham, “Investigation of the Mechanism of Pressure-Induced Solid-Solid Structural Transformations in Semiconductor Nanocrystals”, Ph. D. Dissertation, University of California at Berkeley (2001). [2.33] U. Banin, Private Communication. [2.34] A. Madhukar, S. Lu, A. Konkar, Y. Zhang, M. Ho, S. Hughes, A. P. Alivisatos, “Integrated Semiconductor Nanocrystal and Epitaxical Nanostructure Systems: Structural and Optical Behavior”, Nano. Lett. 5, 479-482 (2005). [2.35] The second growth cycle is stopped when large mount of insoluble precipitation of nanocrystal is observed in the aliquots. [2.36] C. Kittle, Introduction to Solid State Physics, Wiley, New York, (1995). [2.37] Handbook Series on Semiconductor Parameters, Vol. 1, M. Levinshtein, S. Rumyantsev and M. Shur, ed., World Scientific, London, (1996). [2.38] L. Brus, “A simple model for the ionization potential, electron affinity, and aqueous redox potentials of small semiconductor crystallites”, J. Chem. Phys. 79, 5566-5571 (1983). 95 [2.39] L. Brus, “Electron–electron and electron-hole interactions in small semiconductor crystallites: The size dependence of the lowest excited electronic state”, J. Chem. Phys. 80, 4403-4409 (1984). [2.40] U. Banin, C. J. Lee, A. A. Guzelian, A. V. Kadavanich, A. P. Alivisatos, W. Jaskolski, G. W. Bryant, Al. L. Efros, M. Rosen, “Size-dependent electronic level structure of InAs nanocrystal quantum dots: Test of multiband effective mass theory”, J. Phys. Chem., 109, 2306-2309 (1998). [2.41] P. C. Sercel, K. J. Vahala, “Analytical formalism for determining quantum- wire and quantum-dot band structure in the multiband envelop-function approximation”, Phys. Rev. B 42, 3690-3710 (1990). [2.42] O. Stier, M. Grundmann, D. Bimberg “Electronic and optical properties of strained quantum dots modeled by 8-band p k r r ⋅ theory”, Phys. Rev. B 59, 5688-5701 (1999). [2.43] S. V. Nair, L. M. Ramaniah, K. C. Rustagi, “Electron-states in a quantum dot in an effective-bond-orbital model”, Phys. Rev. B 45, 5969-5979 (1992). [2.44] H. H. vonGrunberg, “Energy levels of CdSe quantum dots: Wurtzite versus zinc-blende structure”, Phys. Rev. B. 55, 2293-2302 (1997). [2.45] J. Perez-Conde, A. K. Bhattacharjee, “Electronic structure of CdTe nanocrystals: a tight-binding study”, Solid Stat. Comm. 110, 259-264 (1999). [2.46] L. W. Wang, A. Zunger, “ High-energy excitonic transitions in CdSe quantum dots”, J. Phys. Chem. B, 102, 6449-6454 (1998). [2.47] A. J. Williamson, A. Zunger, “InAs quantum dots: Predicted electronic structure of free-standing versus GaAs-embedded structures”, Phys. Rev. B 59, 15819-15824 (1999). [2.48] L. Chen, “Optical Properties of InGaAs/GaAs quantum well structures: Application to light modulators and switches”, Ph. D. Disseratation, University of Southern California (1993). [2.49] C. R. Pidgeon, R. N. Brown, “Interband magneto-absorption and Faraday rotation in InSb”, Phys. Rev. 146, 575-& (1966). [2.50] J. M. Luttinger, W. Kohn, “Motion of Electrons and Holes in Perturbed Periodic Fields”, Phys. Rev. 97, 869 (1955). [2.51] J. M. Luttinger, “Quantum Theory of Cyclotron Resonance in Semiconductors: General Theory”, Phys. Rev. 102, 1030 (1956). 96 [2.52] E. O. Kane “Band structure of Indium Antimonide“, J. Phys. Chem. Solids, 1, 249 (1957). [2.53] Y. C. Chang, G. D. Sanders, D. Z. –Y. Ting, in Exciton in Confined System, Eds. R. Del Sole, A. D’Andrea, and A. Lapiccirella, Spring-Verlog, Berlin, (1988). [2.54] G. Bastard, “Super-lattice band-structure in the envelop-function approximation”, Phys. Rev. B. 24, 5693-5697 (1981). [2.55] G. Bastard, J. A. Brum, “Electronic states in semiconductor heterostructures”, IEEE J. Quantum Electron. 22, 1625-1644 (1986). [2.56] J. B. Xia, “Electronic structure of zero-dimensional quantum wells”, Phys. Rev. B 40, 8500-8507 (1989). [2.57] K. J. Vahala, P. C. Sercel, “Application of a total-angular-momentum bassis to quantum-dot bandstructure”, Phys. Rev. Lett. 65: 239-242 (1990). [2.58] A. I. Ekimov, F. Hache, M. C. Schanne-Klein, D. Ricard, C. Flytzanis, I. A. Kudryavtsev, T. V. Yazeva, Al. L. Efros, “Absorption and intensity-dependent photoluminescence measurements on CdSe quantum dots: assignment of the first electronic transitions”, J. Opt. Soc. Am. B, 10, 100-107 (1993). [2.59] Al. L. Efros, M. Rosen, “Quantum size level structure of narrow-gap semiconductor nanocrystals: effect of band coupling”, Phys. Rev. B 58, 7120-7135 (1998). [2.60] G. L. Bfr, G. E. Pikus, Sov. Phys. Solid State 3, 2221 (1962). [2.61] B. L. Gel’mont, M. I. Dyakonov, “Acceptor levels in diamond type semiconductors ” Sov. Phys. Semicond. 5, 1905 (1972). [2.62] Y. H. Zhu, X. W. Zhang, J. B. Xia, “Electronic States in InAs quantum spheres and ellipsoids”, Phys. Rev. B 73, 165326 (2006). [2.63] P. Lawaetz, “Valence band parameters in cubic semiconductors”, Phys. Rev. B, 4, 3460-3467 (1971). [2.64] V. I. Sheka, D. I. Sheka, “Local states in a semiconductor with a narrow forbidden band”, Sov. Phys. JETP 24, 975 (1967). [2.65] U. Banin, Y. W. Cao, D. Katz, O. Millo, “Identification of atomic-like electronic states in indium arsenide nanocrystal quantum dots”, Nature, 400, 542-544 (1999). 97 [2.66] D. J. Norris, “Electronic structure in semiconductor nanocrystals”, Chap. 2 in Semiconductor and Metal Nanocrystals, Ed. V. I. Klimov, Marcel Dekker Inc., New York (2004). [2.67] A. V. Rodina, A. Y. Alekseev, A. L. Efros, M. Rosen, B. K. Meyer, “General boundary conditions for the envelope function in the multiband p k r r ⋅ model”, Phys. Rev. B 65, 125302 (2002). [2.68] Z. M. Fang, K. Y. Ma, D. H. Jaw, R. M. Cohen, G. B. Stringfellow, “Photoluminescence of InSb, InAs, and InAsSb Grown by organometallic vapor- phase epitaxy.” J. Appl. Phys. 67, 11 (1990) 7034-7039. 98 Chapter 3. NCQDs for Cell Labeling, NSOM based Simultaneous Cell Morphology and Optical Studies. §3.1 Introduction In the last chapter we discussed the synthesis of nanocrystals and its basic electronic and optical properties. In our roadmap towards the biological application of nanocrystals as discussed in the introduction chapter (refer to figure 1.2(a) and (b)), the second essential element is the conjugation and interaction between nanocrystals and biological entities. The as-synthesized nanocrystals are covered with surface chemical moieties that are generally incompatible with harmonious co-existence with live cells. Consequently, surface functionalization is necessary to introduce nanocrystals into a biological system and endow it a role. Typically the chosen surface species is based on the biological ligand-receptor interaction, in which a nanocrystal is conjugated to biochemical ligands (e.g. antibodies) that are appropriate for binding to their corresponding receptors (e.g. antigens). Such binding, since the inception of the field of application of nanocrystals to cells [3.1, 3.2], is largely the only relevant interaction between the nanocrystals and the biological entity. The semiconductor nanocrystal plays its natural role as a light emitter. It is the specificity of the particular ligand-receptor interaction that attaches the nanocrystals to their target sites and allows the existence or position of the target site preferentially to be detected via the fluorescence from the nanocrystals. For a recent review of 99 application of nanocrystal quantum dots as biological probes see ref. [3.3]. In spite of the immense popularity that the NCQDs have gained as biological labels in a relatively short time of about eight year, only recently a few studies have gone beyond the use of nanocrystals as fluorescent labels by probing resonant energy transfer between a pair of nanocrystal and fluorescent dye conjugated via biological molecules such as protein and DNA [3.4, 3.5]. Although exploiting nanocrystal as fluorescent label is not intellectually too exciting, enormous research has been called by the needs of modern biomedical applications in areas such as early disease detection. Traditionally disease detection at cell level typically means that a pathologist examines the dissociated cells or tissue section under optical microscope, and a judgment is made based on the morphology (shape, size, etc) of the cells and tissue. Such process is time-consuming and is at a phenomenological level largely depending on the pathologist’s experience. Today, with understanding of the disease at the molecular level, methods to detect the intrinsic biochemical changes attendant to the disease is clearly the direction to pursue for early detection. Many non-invasive optical (spectroscopy or imaging) based detection approaches have been investigated, which either exploit the intrinsic optical property of the cell or tissue (elastic light scattering, autofluorescence…) [e.g. 3.6, 3.7] or rely on extrinsic fluorescent markers [e.g. 3.8, 3.9]. In this respect, the subcomponent of our proposed colloidal/epitaxial heterostructures – the bioconjugated NCQDs themselves as robust and bright emitters – has an important role to play in disease detection at the molecular level. 100 In this chapter, we discuss our initial study of QD labeled cells and their optical characterization as driven by their application in disease diagnosis. In this arena an important question that has received little attention is how a diseased cell’s change in morphology and biochemical process as reflected in optical response (intrinsic or extrinsically labeled) are connected. Such knowledge that bridges the biochemical understanding and the pathologist’s observation, if acquired, will have significant impact on the early diagnosis of disease. For investigations along this line, we have established a unique instrumentation that integrates measurements of cell morphology and optical response in conventional far-field and near-field optical microscopy and spectroscopy. Part of the uniqueness is indeed the ability to simultaneously image the cell morphology and surface biochemistry via near field scanning optical microscopy (NSOM). As opportunity for collaboration arose, a clinically relevant study was carried out on the morphology of the SKBR-3 and MDA breast cancer cells and their surface receptor Her2/neu (an oncoprotein) distribution marked by QD fluorescent labeling. On the organization of the chapter, we first discuss the conventional/near- field optical microscopy and spectroscopy instrumentation and operational principle, with the emphasis on the NSOM. Second the QD labeling of breast cancer cells surface receptor and the result for simultaneous morphological and optical imaging will be presented. 101 §3.2 Integrated Conventional/Near-Field Optical Microscopy and Spectroscopy. To enable nanoscale spatially-resolved examination of labeled and unlabelled cells as well as probe certain aspects of cell adhesion to solid substrates, a unique instrumental capability was established as part of this dissertation. While a detailed description of this facility is provided in the following, we first simply catalogue the functions it provides and the additional functions that can be readily added: (1) Transmission imaging (including phase contrast, DIC imaging) (2) Reflection imaging (3) Fluorescent imaging with lamp or laser excitation (4) Total Internal Reflection Fluorescence Imaging (5) Transmission NSOM Imaging (6) Fluorescent NSOM imaging (7) Transmission spectroscopy (8) Reflection Spectroscopy (9) Broad area fluorescence spectroscopy (10) Micro-PL with far field laser excitation (11) Micro-PL with near-field laser excitation §3.2.1 Overall Instrumentation and Far-Field Microscopy The integrated conventional/near-field optical microscopy and spectroscopy setup is schematically shown in Figure 3.1. A photograph of the setup is shown in Figure 3.2. The setup is built around an Olympus IX71 inverted optical microscope. 102 The IX71 is a reasonable microscope with its optics (including the chosen objective lens) compatible with visible to the NIR range up to 1500nm. Here we introduce this setup describing its excitation/illumination, image formation, and imaging/spectroscopic detection capabilities. The conventional bright field transmission microscopy is illuminated by a Tungsten Halogen (THG) lamp and a long working distance (WD) condenser (NA 0.55 WD=27mm). For normal bright field imaging, an IR blocking filter is placed before the THG lamp for better imaging quality. For fluorescence imaging/spectroscopy, excitation is introduced from the upper backside port of the IX71 through a straight illuminator. Using a sliding mirror switch, excitation can be switched between a 100W Mercury lamp and an optical fiber coupled laser source. A cw water cooled visible Ar + laser (Lexel 95, single wavelength operation, ~2W maximum at 488nm) or a 5mW 532nm NY:YAG laser was coupled into the optical fiber for excitation. A set of objective lenses are available for bright field/phase contrast or fluorescence imaging. High resolution fluorescent imaging is performed with a high numerical aperture (NA) oil immersion objective lens (Olympus 60X, NA 1.45, infinite-corrected). The objective lens can also focus the laser excitation to diffraction limited spots ( diameter 2 / λ ), thus enabling far-field micro-PL measurements. Moreover the NA of this objective is so high that the excitation light can enter the glass/water interface at an incident angle larger than the critical angle 103 Cooled CCD Camera Mercury Lamp Spectrometer And Array Detector EG&G Photon Counting APD on xyz stage Tube Lens f=160mm Eye Pieces 60x NA 1.45 Objective Lens Field Diaphragm Field Diaphragm Long WD Condenser Phase Ring Stage Fiber Coupler Additional Filter Insert DIC Prism Insert Fluorescent Cube Turret Mirror Cubes Halogen Lamp Ar+ Laser or Nd:YAG Laser ND filter Laser Line Filter Optical Fiber Fiber Coupler Nanonics NSOM 100 Infrared Blocking Filter M M M M M M M M Lens on xyz stage L Sample L M M NSOM Cantilever Tip M Cooled CCD Camera Mercury Lamp Spectrometer And Array Detector EG&G Photon Counting APD on xyz stage Tube Lens f=160mm Eye Pieces 60x NA 1.45 Objective Lens Field Diaphragm Field Diaphragm Long WD Condenser Phase Ring Stage Fiber Coupler Additional Filter Insert DIC Prism Insert Fluorescent Cube Turret Mirror Cubes Halogen Lamp Ar+ Laser or Nd:YAG Laser ND filter Laser Line Filter Optical Fiber Fiber Coupler Nanonics NSOM 100 Infrared Blocking Filter M M M M M M M M Lens on xyz stage L Sample L M M NSOM Cantilever Tip M Figure 3.1 Schematic of the custom-built integrated conventional/near-field optical microscopy and spectroscopy setup. 104 for total internal reflection, which enables a special interface sensitive imaging technique – Total Internal Reflection Fluorescence Microscopy (TIRF) [3.10, 3.11]. Light beam formed by the objective lens passes through fluorescent mirror cube turret. By rotating the turret and inserting different mirror cubes in the turret into the light path, one can switch between transmission, reflection, and fluorescent imaging modes. For fluorescent imaging, the dichromatic mirror cube for the corresponding fluorophore is inserted. Figure 3.2 Photograph of the custom build integrated conventional/near-field optical microscopy and spectroscopy setup. The filtered light beam is refocused by the tube lens (f=160mm) and routed to four different exit ports of the IX71 for eye/imaging/spectroscopy/single channel 105 photon-counting. A Peltier cooled mono-chromatic CCD camera (Olympus DP30BW) is installed on the right side port for imaging purpose. The camera has 2/3 inch 1360x1024 pixel CCD chip of pixel size 6.45 μm x 6.45 μm which translates into a 107nm pixel limited resolution when X60 objective is used. The camera has a spectral range of 400-1000nm (5% QE at 1000nm), and therefore has limited imaging capability in near infrared range. A single photon counting avalanche photodiode (APD) module (EG&G SPCM-ARQ-1-5) is installed on the left side port 1 for NSOM imaging. The APD module has a 300ps time resolution and in the future can be converted into time resolved application. Left side port 2 is used to connect to a spectrometer for spectroscopic application. Two spectrometers, (a) Acton SP275i spectrograph equipped with LN 2 cooled CCD array (Princeton Instrument 256x1024 pixel, 400-1050nm), or (b) Acton SP300i spectrograph equipped with LN 2 cooled linear InGaAs array (Princeton Instrument 1x512 pixel, 800-1600nm) can be interchangeably used, covering, respectively, visible and NIR wavelength regions. A near field scanning optical microscope (Nanonics NSOM100) is integrated into the microscopy setup without affecting normal far-field optical microscope operation. The NSOM head is installed on the IX71 sample stage, fitted underneath the long WD condenser. In a typical NSOM experiment, we first image the sample under bright field or fluorescent mode, and then bring the NSOM tip onto the area of interest for high resolution NSOM imaging. The details of NSOM instrumentation will be discussed in §3.3. 106 The custom-built microscopy and spectroscopy setup as described above has great flexibility in its application. It can be switched back and forth without reassembling between the following imaging/spectroscopy modes: (1) Transmission imaging (including phase contrast); (2) Reflection imaging; (3) Fluorescent imaging with lamp or laser excitation; (4) Total Internal Reflection Imaging (5) Transmission NSOM Imaging; (6) Fluorescent NSOM imaging; (7) Transmission spectroscopy; (8) Reflection Spectroscopy; (9) Broad area fluorescence spectroscopy; (10) Micro-PL with far field laser excitation; (11) Micro-PL with near-field laser excitation. As an example of the far field microscopy, in figure 3.3(a)-(c) are shown a set of epi- fluorescent images of a SKBR3 breast cancer cells on a glass coverslip. The cell membrane receptor Her2/neu is labeled by appropriately conjugated 605nm emission CdSe/ZnS quantum dots. The sample details are provided in §3.4. The images (a)-(c) were taken when focused at different focal planes, respectively: (a) at the top, (b) middle, and (a) bottom of the cell. Figure 3.3(d) shows a magnified view of one of the single quantum dots in good focus. Single QD imaging is verified by the blinking behavior of the QD and defocused imaging. Commonly for both far-field and near- field microscopy, the FWHM of the image of the single QD (as a point light sources) is used as a good estimation for the resolution of the microscopy system [3.12, 3.13]. A cross-section of the intensity profile of the NCQD is shown in figure 3.3(e) with the Gaussian gives FWHM of 400nm. This demonstrates the ~ λ/2 diffraction limited resolution of the far field fluorescent microscopy. 107 20 μm 20 μm 20 μm (a) (b) (c) 20 μm 20 μm 20 μm 20 μm 20 μm 20 μm (a) (b) (c) (d) (d) (e) (e) Figure 3.3 Epi-fluorescent imaging of SKBR3 breast cancer cells with their Her2/neu surface receptors specifically labeled by CdSe/ZnS NCQDs (600nm emission). Panel (a), (b), (c) are images taken with the focal plane respectively at the top, middle and bottom (closest to the glass substrate) of the cells. Note in the three images only a fraction of the NCQDs are in good focus. Panel (d) shows a magnified value of a NCQD in focus, each pixel in this image corresponds to 107nm. A cross-section of the intensity profile of the NCQD is shown in panel (e). The Gaussian fitting of the intensity profile gives a width of 400nm demonstrating the ~ λ/2 diffraction limited resolution of this fluorescent image. 108 Even though the epifuorescence imaging gives a reasonable resolution, its limitation is also clear. Given the thickness of the cell ~10 μm and the smaller focal depth of the objective ~1 μm, in figure 3.3(a)-(c) only a fraction of the QDs in each image is in focus and can be imaged properly. Obviously far field fluorescent imaging can not provide information on the cell surface (membrane) morphology and cannot track and image the fluorescence from only the cell surface. The same basic limitation also exists in advanced far-field techniques such as confocal fluorescent microscopy. In order to fulfill our objective to study the morphology of the cell surface and its possible connection with the membrane biochemistry change, we have to invest in a surface sensitive imaging technique – the near field scanning optical microscope (NSOM), as is discussed next. §3.3 NSOM Instrumentation and experimental consideration §3.3.1 Introduction to the NSOM technique As described in the last section, conventional optical microscopy cannot track fluorescent labels on cell surface, or provide enough depth(z)-resolution to image the morphology of the cell. To overcome this shortcoming we added the NSOM capability to our microscopy system which is by and large the only means to obtain simultaneously precise morphology and optical information from cell surface. The idea of NSOM, as cited in many books, was first proposed in 1928 by E. H. Synge [3.14]. It is based upon the use of the presence of evanescent waves with 109 high spatial frequency (> λ/2) in diffraction phenomena, rather than the commonly employed propagating waves with low spatial frequency (< λ/2) which give rise to the conventional far-field diffraction and thus the well-known Abbe limit [3.15] of ~ λ/2 spatial resolution. He envisioned that by scanning a microscopic aperture in the near-field regime the Abbe diffraction limit could be surpassed significantly. The evanescent wave field intensity damps rapidly as the fourth power of the inverse of the distance away from the diffracting object in contrast to the much slower inverse quadratic decay of the field intensity for far field diffraction. Consequently, strict requirements on the distance of the aperture from the object need to be satisfied, typically less than 10nm, and with apertures on the nanometer scale spatial resolutions ~ λ/20 can be realized. However, Synge’s vision could not be realized for nearly 60 years, mainly due to two technological obstacles: (a) the difficulty in reliably manufacturing a subwavelength optical aperture (b) the difficulty in controlling relative motion between the aperture and the sample with nm precision and keeping the aperture/sample distance constant within nm range. Today with different technological advances that overcome these very difficulties, many different embodiments of NSOM have been designed and implemented [3.16, 3.17]. A Detailed discussion of these is beyond the scope of this dissertation. Here we will focus on our specific NSOM instrumentation and experiments. A popular choice of subwavelength optical aperture is a tapered optical fiber (either etched or pulled) which is coated with a metal except for the aperture (50nm diameter or larger) at the end of the fiber’s tip. Light can be conveniently coupled 110 into the fiber and is then emitted at the aperture to create near-field excitation. An SEM image of the type of fiber tip used for our experiments is shown in figure 3.4(a)-(b). The control of fiber tip / sample distance and scanning of tip over sample in NSOM is typically realized by some type of atomic force microscopy (AFM). In most of the NSOM systems, a straight fiber tip is used. The tip-sample distance is controlled via a tuning fork based shear force feedback loop [3.18]. Typically the sample position is fixed while the tip is mounted on a scanning tube and scanned on the sample. Our NSOM system from Nanonics (Israel) is quite unique. In this system, the fiber tip is bent at its end to form a cantilever that can essentially function as a normal AFM tip, but a tip in which one can pass light through (figure 3.4(d)). The cantilever fiber tip measures the sample morphology using conventional normal force feedback tapping mode AFM [3.19, 3.20], while at the same time the tip with a sub-wavelength aperture is used for near-field optical excitation or detection to form a corresponding optical image. In our NSOM system the fiber tip is fixed and the sample is scanned using a flat piezo scanner. Compared to the straight fiber based NSOM instruments, our cantilever fiber based NSOM has two distinct advantages: (1) compared to the shear force AFM, the normal force AFM allows better topology resolution. Its data is also more straightforward given the mechanism of shear force AFM is still ambiguous; (2) compared to the tuning fork and scanning tube that totally occupies one side of the sample, the use of a flat piezo scanner provides greater flexibility by allowing optical access from both sides of the sample. 111 100nm (a) (b) (d) (c) 100nm (a) (b) (d) (c) Figure 3.4 (a) SEM image of a tapered fiber tip. (b) Higher resolution SEM image showing the end of the tapered fiber coated with metal except for the aperture (c) Schematic of the metal coated fiber tip (d) optical image of the cantilevered fiber tip. (Image adapted from Nanonics product brochure). 112 §3.3.2 NSOM Instrumentation Our NSOM setup was built by combining a NSOM 100 system (Nanonics, Israel) and an IX71 inverted optical microscope (Olympus). The setup is schematically shown in figure 3.5. At the heart of it is the NSOM 100 head (hereafter referred to as NSOM head). The NSOM head comprises an upper and a lower part hinged together on their left end. The cantilevered fiber tip is magnetically attached to a tip holder which is mounted in the middle of the upper part. A piezo driving element is installed above the tip holder to vibrate the tip for tapping mode imaging. The right and left compartments of the upper part contain a focused 670nm laser diode and a Si position sensitive photo detector (PSPD) that together constitute the light lever to monitor the tip bending or vibration. The right compartment also houses a stepper motor that controls the spacing between the upper and lower parts and allows coarse approach of the tip to the sample. The lower part of the NSOM head (~7mm thick) contains a 3D flat piezo scanner of 42 μm(x), 42 μm(y), 23 μm(z) range. The sample resides on three diamond pins on the scanner due to its own gravity. Normally there is not relative motion between the sample and diamond pin so that for AFM/NSOM imaging the sample can be scanned with respect to the fiber tip within the range of the flat scanner. For the tip to access different areas on the sample, scanner is moved in fast strokes to cause the sample to slip on the scanner’s diamond pins in steps of a few μm given the sample’s inertia. By repeating such motion (called by Nanonics “inertia transfer”) the sample can be displaced a few millimeters with respect to the tip. The instrumentation details of NSOM 100 can be found in [3.21]. 113 ~100nm aperture Laser input Cantilevered fiber tip PSPD LD Topaz Controller (Cavendish Instrument) APD Controller 60x NA1.45 Objective lens 532 or 488nm Notch Filter 600nm Band Pass Filter 500 Long Pass Filter 620 Long Pass Filter EG&G Photon Counting APD SPCM-AQR-1-5 670nm LD 670nm BP Filter 640nm LP Filter PSPD Stepper Motor Planar Piezo Scanner M M Sample Fiber Tip Tip Holder 532nm Nd:YAG Laser or 488nm single line Ar+ Laser or ND 532 or 488nm Laser Line Filter Fiber Coupler NSOM Fiber Probe (exciting sample in The near-field) Cell 600nm CdSe QD Label Topaz PC Interface Topaz Nanonics Interface Power 5V Signal PC Quatz NSOM Control Program (Cavendish Instrument) Optional Integrator Tube Lens f=160mm Nanonics NSOM 100 Head ~100nm aperture Laser input Cantilevered fiber tip PSPD LD Topaz Controller (Cavendish Instrument) APD Controller 60x NA1.45 Objective lens 532 or 488nm Notch Filter 600nm Band Pass Filter 500 Long Pass Filter 620 Long Pass Filter EG&G Photon Counting APD SPCM-AQR-1-5 670nm LD 670nm BP Filter 640nm LP Filter PSPD Stepper Motor Planar Piezo Scanner M M Sample Fiber Tip Tip Holder 532nm Nd:YAG Laser or 488nm single line Ar+ Laser or ND 532 or 488nm Laser Line Filter Fiber Coupler NSOM Fiber Probe (exciting sample in The near-field) Cell 600nm CdSe QD Label Topaz PC Interface Topaz Nanonics Interface Power 5V Signal PC Quatz NSOM Control Program (Cavendish Instrument) Optional Integrator Tube Lens f=160mm Nanonics NSOM 100 Head ~100nm aperture Laser input Cantilevered fiber tip PSPD LD Topaz Controller (Cavendish Instrument) APD Controller 60x NA1.45 Objective lens 532 or 488nm Notch Filter 600nm Band Pass Filter 500 Long Pass Filter 620 Long Pass Filter EG&G Photon Counting APD SPCM-AQR-1-5 670nm LD 670nm BP Filter 640nm LP Filter PSPD Stepper Motor Planar Piezo Scanner M M Sample Fiber Tip Tip Holder 532nm Nd:YAG Laser or 488nm single line Ar+ Laser or ND 532 or 488nm Laser Line Filter Fiber Coupler NSOM Fiber Probe (exciting sample in The near-field) Cell 600nm CdSe QD Label Topaz PC Interface Topaz Nanonics Interface Power 5V Signal PC Quatz NSOM Control Program (Cavendish Instrument) Optional Integrator Tube Lens f=160mm Nanonics NSOM 100 Head Figure 3.5 Schematics of the NSOM setup based on Nanonics NSOM 100 head. 114 In operation, the NSOM head is installed on the IX71 translational stage underneath the long working distance condenser (refer to figure 3.1) with the fiber tip roughly aligned with the optical axis of the microscope (as limited by the translational stage precision ~5 μm) and in the focal plane of the objective. Like in conventional tapping mode AFM, the scanner brings the sample surface to a controlled distance to the tip. In its current configuration (as shown in figure 3.5) and the discussion to come, the NSOM is setup in “illumination” mode although “collection mode” and “excitation/collection” mode imaging are possible by reconfiguring the instrument. In the illumination mode, an Ar + 488nm or Nd:YAG 532nm laser is coupled to the optical fiber tip and creates a near-field illumination localized on the sample around the subwavelength aperture of the tip. The fluorophores on the sample surface are excited by the near-field. The fluorescent emission is detected while the sample surface is scanned under the fiber tip in a step-reside-step fashion (in our instrumentation the tip is fixed, sample moves). The integrated fluorescence measured while the tip resides at different locations on the sample is used to construct the fluorescence NSOM image. Specifically the fluorescence is collected by the high NA oil immersion objective lens (Olympus x60 NA1.45) of the IX71 from the back of the sample, and filtered by a high-discriminating ratio bandpass filter set. The bandpass filter set consists of (1) laser rejection filter (488nm rejection blocking>10 3 or 532nm rejection, blocking>10 5 , Omega Optics), (2) Band pass filter (600nm central 115 wavelength, 65nm FWHM, blocking >10 3 , Coherent) (3) 560nm Long pass filter (532nm blocking >10 2 , 488nm blocking>10 4 , Corning) (4) 620nm short pass (Blocking >10 3 , Earling). The 620nm short pass filter is for rejecting the 670nm laser for AFM control. The filter set provides 9 10 ≥ blocking for the excitation laser which is essential for the detection of the weak fluorescence signal. The filtered fluorescence was refocused by the tube lens (f=160mm) onto a single-photon counting APD (EG&G SPCM-AQR-1-5) which measures the fluorescence intensity. The APD has a small diameter of 195 μm. The small size is advantageous since the APD serves as a spatial filter that rejects the background noise, but it requires the NSOM tip aperture (hence the fluorescence generated around it) to be precisely focused onto the APD. Since the IX71 stages on which the NSOM head and tip sit do not have enough precision, this alignment has to be done by translating the position of APD on an XYZ stage while looking for the maximum fluorescence signal. If one cannot find the weak fluorescent light at all due to APD misalignment, then for coarse alignment one should first remove the fluorescent bandpass filter and translate the APD to find the excitation laser from the fiber tip, then restore the fluorescence filter and maximize the fluorescence signal. This tip- APD alignment is the most time consuming part of the NSOM experiment and should be performed patiently for many rounds for the best results. A few custom modifications have been made on the original Nanonics NSOM 100 system to improve its performance. These are: 116 (1) For the x60 oil immersion lens (WD ~200 μm) to get close enough to the sample, the position of the sample and thus the fiber tip has to be lower. Tip holder and sample holder were redesigned and custom made for this purpose. (2) The 670nm laser diode (a part of the light lever) turns out to have a broad spectrum extending + 50nm and interferes with fluorescence NSOM measurements. Therefore a 670+3nm bandpass filter has been installed before the laser diode to narrow its spectrum. (3) A 630nm long pass filter has been installed before the PSPD which prevents the PSPD from being interfered with by the excitation light when doing far- field fluorescence microscopy. This allows far-field and NSOM imaging to be switched back-and-forth. The filter also blocks the PSPD from ambient room light and thus increases the tapping mode AFM stability. (4) An integrating circuit box is made and can be inserted to freely change the APD integration time. §3.3.3 NSOM Principle and Experimental Consideration §3.3.3 (a) Basic Theory of Near Field Optics and Generation of Near Field Excitation Our fiber optic cantilever tip based implementation of the NSOM is aimed at two major application aspects: 117 (a) measurement of cell morphology using contact or tapping mode AFM, (b) measurement of fluorescence from labeled cell surface using near field optical excitation and far-field detection. AFM has been a well studied subject. Dr. T. R. Ramachandran’s dissertation [3.22] from our group provides a good review of the principles of AFM. A recent review of application of AFM in cell biology can be found in [3.23]. Below we focus the discussion on the basic theory of near field optics and the generation of near field excitation. In our implementation of the NSOM, the near field optical excitation is generated through the use of the tapered fiber tip with an aperture defined by metal coating. The excitation phenomenon involves physics of electromagnetic wave diffraction through a subwavelength aperture (as schematically shown in figure 3.6). A plane wave of wavelength λ comes from − ∞ = z and reaches a screen at z=0 with an opening of diameter 2a. Let us assume the screen is “opaque” and infinitely thin and ask what is the wave in the half space z>0? The answer, to the zero th order approximation, can be obtained using Born’s angular spectrum theory [3.24] as follows. 118 x y z 2a screen Normal Incident Plane Wave x y z 2a x y z 2a screen Normal Incident Plane Wave Figure 3.6 Schematic of the subwavelength aperture diffraction problem. Any arbitrary wavefront ) 0 , , ( y x U at plane z=0 can be expanded by the superposition of elementary plane waves ( ) ) ( exp ) , , ( z y x ik z y x B γ β α + + = at z=0 via Fourier Transformation (FT): [] ∫∫ +∞ ∞ − + − = ) ( ) ( ) ( exp ) 0 , , ( 2 1 ) 0 , , ( k d k d ky kx i k k A y x U β α β α β α π (3.1) or inversely [] ∫∫ +∞ ∞ − + = dxdy ky kx i y x U k k A ) ( exp ) 0 , , ( ) 0 , , ( β α β α (3.2) 119 where λ π 2 = k is the wave vector. α , β , γ are the direction cosines: k k x α = , k k y β = , k k z γ = , and 1 2 2 2 = + + γ β α . ) 0 , , ( k k A β α is the amplitude of the corresponding plane wave, which is called the angular spectrum associated with wavefront ) 0 , , ( y x U . Suppose the incident plane wave is normal to the screen, and the aperture is defined by function ⎩ ⎨ ⎧ > + ≤ + = ) ( , 0 ) ( , 1 ) 0 , , ( 2 2 2 2 a y x a y x y x t . If we assume that the usual Kirchhoff’s approximation [3.25] hold, i.e. that, (1) on the screen, 0 = ∂ ∂ = n u u (2) in the aperture the wave just like in the free space, ) exp( ikz u − = (3.3) then the wavefront at z=0 through the aperture is: ) 0 , , ( ) exp( ) 0 , , ( y x t ikz y x U × − = (3.4) Substituting into (3.2), ) , ( ) , ( ) , ( ) 0 , , ( k k T k k T k k k k A β α β α β α δ β α = ⊗ = (3.5) where ) , ( k k T β α is the FT of the aperture. Therefore the corresponding angular spectrum at plane z=0 where the aperture is located is: [] ∫∫ +∞ ∞ − + − = = ) ( exp ) 0 , , ( ) , ( ) 0 , , ( y x ik y x t k k T k k A β α β α β α dxdy (3.6) Introducing the conversion, 120 ⎪ ⎩ ⎪ ⎨ ⎧ ′ + = + − ′ + = + y y x x y x 2 2 2 2 β α α β β α β α (3.7) and substituting into (3.6) ϕ ϕ ρ β α ρ ρ β α β α π d ik d y d x d x ik y x t k k A a )] cos ( exp[ )] ( exp[ ) 0 , , ( ) 0 , , ( 2 2 2 0 0 2 2 + − = ′ ′ ′ + − = ∫ ∫ ∫∫ +∞ ∞ − (3.8) The last step in (3.8) above converts the integral into polar coordinates ) , ( ϕ ρ Consider the properties of Bessel functions: [] ∫ + − − = π ϕ ϕ ϕ π 2 ) cos ( exp 2 ) ( ) ( a a n n d n x i i x J and [] ) ( ) ( 0 1 x xJ x xJ = ′ (3.9) Then the integral in Equ 3.8 can be rewritten as, ( ) 2 2 2 2 1 2 2 ) 0 , , ( β α β α π β α + + = ka ka J a k k A (3.10) The wave front propagates to plane z can be expanded in the same fashion into its component plane waves: [] ∫∫ +∞ ∞ − + = ) ( ) ( ) ( exp ) , , ( 2 1 ) , , ( k d k d y x ik z k k A z y x U β α β α β α π (3.11) The evolution of the angular spectrum ) , , ( z k k A β α can be obtained using the Helmholtz equation of the wave: 0 2 2 = + ∇ U k U (3.12) Inserting (3.11) into (3.12), we obtain: () () 0 ) ( exp ) , , ( 1 ) , , ( 2 2 2 2 2 = + ⎥ ⎦ ⎤ ⎢ ⎣ ⎡ − − + ∂ ∂ ∫∫ ∞ + ∞ − dxdy y x ik z k k A k z k k A z β α β α β α β α (3.13) 121 which is satisfied only if for all α , β : () 0 ) , , ( 1 ) , , ( 2 2 2 2 2 = − − + ∂ ∂ z k k A k z k k A z β α β α β α (3.14) Solving linear differential equation (3.14) leads to, ( ) 2 2 1 exp ) 0 , , ( ) , , ( β α β α β α − − ⋅ × = ikz k k A z k k A (3.15) Namely, the angular spectrum of the wavefront that propagates to plane z in free space is merely that at plane z=0 multiplied by an exponential term: ( ) 2 2 1 exp β α − − ⋅ ikz . Inverse Fourier transforming ) , , ( z k k A β α , one can obtain the wavefront at plane z. ( ) [ ] 2 2 1 1 exp ) 0 , , ( ) , , ( β α − − ⋅ ⊗ = − ikz FT y x U z y x U (3.16) Close to the aperture plane, π λ 2 1 = << k z , and hence 1 < < kz . The FT term should be close to a delta function. The wavefront should not be broadened too much from ) 0 , , ( y x U , therefore still should be localized laterally in an area ~ the size of the aperture. For spectral components ) , , ( z k k A β α whose 1 2 2 < + β α , ( ) 2 2 1 exp β α − − ⋅ ikz only introduces a phase factor, so the corresponding plane waves propagate without attenuation in their amplitude. These spectral components can be detected using a collection system (such as an objective lens) placed away from the aperture and therefore is referred to as far-field components. 122 On the other hand for spectral components whose 1 2 2 ≥ + β α , (3.15) becomes: ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎝ ⎛ − × = p d z k k A z k k A exp ) 0 , , ( ) , , ( β α β α (3.17) where 1 2 1 1 2 2 2 2 − + = − + = β α π λ β α k d P is the penetration depth. So the corresponding plane wave is non-propagating, decaying exponentially away from z=0 which is similar to the evanescent wave generated by total internal reflection. Such spectral component is referred to as near-field components. Returning to diffraction through the circular aperture, we notice at zeroth order the amplitude of the angular spectrum at z=0 (3.10) is proportional to 2 a π area of the aperture, or the energy carried by each plane wave is proportional to 4 a . The angular spectrum obtained from Equ. (3.10) normalized to the aperture area is plotted in figure 3.7. We notice that the near-field components ( 1 2 2 ≥ + β α , in the shadowed area of figure 3.7) arise significantly only when the diffraction aperture width is smaller than the wavelength. The near field components go asymptotically towards zero with increased aperture size. The far field components ( 1 2 2 < + β α , in the white area of figure 3.7) however always exists at a finite value, whatever the aperture size. Namely, to correct a common misconception, the near-field components do not exist by themselves, but are always accompanied by the propagating far-field components. 123 2 2 β α + Normalized Angular Spectrum 2a=50nm 2a=100nm 2a=200nm 2a=500nm 2a=1000nm 2a=2000nm λ=532nm 2 2 β α + Normalized Angular Spectrum 2a=50nm 2a=100nm 2a=200nm 2a=500nm 2a=1000nm 2a=2000nm λ=532nm Figure 3.7 Normalized angular spectrum (Equ. 3.10) of a plane wave passing through a circular aperture of diameter 2a. λ=532nm. The white and gray areas are respectively the far-field and the near-field spectral components. Using angular spectrum of the aperture, Equ. (3.10), by integrating the modulus square of the amplitude components corresponding to 1 2 2 ≥ + β α and that corresponding to 1 2 2 < + β α , we obtain the ratio of energy associated with near- field to that associated with far-field wave as a function of the aperture diameter a 2 (figure 3.8). Note the fast turning of this curve at 2 / ~ 2 λ a . The practical value of this ratio is that, for a given NSOM tip aperture diameter and light wavelength, we 124 can estimate the near-field energy using the measured far-field transmission. As a numerical example for 532nm incident light and 100nm, 500nm diameter aperture, the calculated ratio of near-field energy to far field energy is respectively ~10 and ~0.25. Figure 3.8 Calculated near field to far field energy ratio as a function of aperture diameter. 125 The above model qualitatively depicts the generation of the near-field excitation in our NSOM implementation -- via diffraction through a subwavelength size aperture on a fiber tip. The model provides guidance for NSOM measurements to be presented. It gives two main qualitative results: (1) the diffracted wave close to the aperture plane ) ( λ << z laterally is localized ~ the aperture size, which means the resolution of the NSOM is roughly the aperture size, and, (2) the near-field components dominate when aperture diameter is less than around half of the wavelength. While its qualitative consequences hold, the model is not quantitatively accurate. The main deficiency of the model is the assumed Krichhoff approximation (Equ. 3.3). We never explicitly detailed what is the screen, which typically is made of a metal. The solution derived above based on Krichhoff approximation and angular spectral expansion does not satisfy the correct boundary condition on the metal screen at z=0, the tangential component of the electrical field is non-zero. Considering the boundary condition on the screen, the problem of diffraction through a small circular aperture on a perfect conducting screen has been studied by many authors. At the long wavelength limit ( 1 < < ka ), the far field and near field distribution has been analytically derived respectively by Bethe [3.26] and Bouwkamp [3.27]. As a widely used result proposed by Bethe, the diffraction through the small aperture (long wavelength limit) is often approximated as a superposition of emission from an electric dipole and a magnetic dipole placed at the 126 aperture. Later the same problem is revisited by Leviatan [3.28] and Durig et al [3.29] in the context of NSOM, in which numerical data are presented. Here we quote Leviatan’s result in figure 3.9, in which transmitted power density distribution near the aperture ( 150 / 2 λ = a ) is plotted. This calculation shows as we have argued before, the transmitted field close to the aperture is indeed confined to the aperture size (rather than the wavelength), and the depth of this confinement deep is also ~ the aperture size. x/a y/a z/a z/a Normalized Power Density Normalized Power Density 0.0 0.00 0.25 0.50 0.75 1.00 0.00 0.25 0.50 0.75 1.50 1.00 1.25 0.5 1.0 2.0 1.5 0.0 0.5 1.0 2.0 1.5 -3 3 1 -1 -3 3 1 -1 x/a y/a z/a z/a Normalized Power Density Normalized Power Density 0.0 0.00 0.25 0.50 0.75 1.00 0.00 0.25 0.50 0.75 1.50 1.00 1.25 0.5 1.0 2.0 1.5 0.0 0.5 1.0 2.0 1.5 -3 3 1 -1 -3 3 1 -1 Figure 3.9 Diffracted light power density near an aperture of radius a= λ/150 (calculated). The light is polarized along x axis. See figure 3.6 for the definition of the coordinates. Taken from [3.28] 127 Unfortunately, all the above analytical theories deals with diffraction through a small aperture ( 1 << ka ) on a perfect conducting screen. Most experimental situations, however, are more complicated since the tip aperture sizes are at most a few times smaller than the wavelength; tip has a non-planar geometry; the metal coating has a finite conductivity and thickness; a sample is nearby and interacts with the tip, etc.. Although many sophisticated models have been developed to describe the tip/sample interaction and NSOM imaging [3.17], these theories provide only quantitative guidance for specific measurements, their description is beyond the scope of this dissertation. Below, we focus on our NSOM instrumentation and its intended application for the fluorescent imaging of QD and QD labeled cells and discuss the related experimental considerations. §3.3.3(b) Experimental Consideration in NSOM measurement For successful usage of NSOM, three basic considerations relating to instrumentation and signal measured are critical: (i) NSOM tip transmission and excitation flux, (ii) fluorescence signal (iii) noise. In the following we describe these. (i) NSOM tip transmission and excitation photon flux In all the above discussion, the screen (therefore the aperture) is assumed to be made of an ideal conductor and infinitely thin. In real life, as recognized by Betzig et al [3.30], the metal (of finite conductivity) deposited on the fiber tip needs 128 a finite thickness to attenuate the light. The energy flux of light attenuation in a metal film is [3.25]: ) / exp( 0 δ d E E − = , (3.18) where ) / ( 4 ~ 4 0 0 2 ε ωμ μσ π λ π λ δ n = is the attenuation length, 2 n is the imaginary part of the refractive index, and σ is the conductivity. For the aluminum coating used in our fiber tips, the attenuation length is 5 . 6 = δ nm for wavelength 500nm. Aluminum is chosen largely for its high conductivity and small attenuation length. But still the necessary coating thickness of a few tens of nm is quite comparable to the aperture diameter. Therefore, the aperture needs to be considered actually as a waveguide and hence light attenuation going through a waveguide must be taken into account. The least attenuated mode in a circular waveguide is the TE11 mode [3.31], for which the energy flux decays at a rate of, ) / 682 . 3 exp( 0 a d E E − = (3.19) where a is the radius of the waveguide, d is the length. Comparing (3.18) and (3.19) gives the minimum aperture diameter δ 364 . 7 2 min = a , below which aperture has no more throughput than just the metal coating and the tip will be effectively apertureless. For aluminum coating ( 5 . 6 = δ nm) min 2a is ~48nm. (We note that although the apertureless tips are the tip of choice to achieve very high spatial resolution, their low throughput is not suitable for fluorescent imaging where signal is weak). 129 Since the transmitted energy flux in the zeroth order is approximately proportional to the aperture area, considering the attenuation in passing the finite- thickness aperture (3.19), the final throughput of the aperture as a function of its radius can be written as: 2 1 ) / 682 . 3 exp( a a d E ⋅ − ∝ , (3.20) Since even with the state-of-the-art technology for NSOM tips the maximum light power input allowed is typically ~10mW before the metal coating will melt due to local heating by the light, the fast decrease in throughput for small apertures becomes an essential limiting factor to be considered. It means if one intends to improve the resolution of the NSOM image by decreasing the tip diameter, a heavy penalty will be paid due to decreasing signal level. Cantilever Fiber Tips: Our NSOM experiments are performed with cantilever fiber tips from Nanonics. The tips are made of tapered multimode glass optical fiber (50 μm core diameter) with Cr/Al coating. According to our test, the damage threshold is ~25mW beyond which the coating melts. The tip’s nominal transmission efficiency as specified by Nanonics is listed in Table 3.1. The transmission decreases ~10 times when aperture diameter goes to half. 130 Table 3.1 Nanonics NSOM tip nominal throughput as a function of aperture diameter. Aperture Diameter (nm) Transmission Efficiency 500 10 -1 300 10 -2 200 10 -3 100 10 -4 50 10 -5 Before each fluorescent NSOM experiment, using the same setup (figure 3.5), the tip’s far-field transmission is always checked by switching the filter set to the corresponding laser line filter, and focusing the illumination (attenuated) from the NSOM tip into the APD detector. The transmitted photon flux is then calculated accounting for the APD quantum efficiency and the transmission coefficient on the detection side. For 5mW, 532nm light input, for “reasonable tips” the far field photon flux FF I for 100nm and 500nm diameter aperture tips are, respectively, ~10 10 hv/sec (40nW) and 10 13 hv/sec(40 μW). These are in fact ~2 orders of magnitude lower than calculated using Nanonics’ nominal transmission coefficient (table 3.1). The discrepancy is not yet solved. Further, among all received tips, only ~50% are in the “reasonable tips” category, the photon flux of other tips can be lower by 10-100 times the number quoted above. Clearly the control on the NSOM tip transmission is a difficult task, given its exponential dependence on the aperture diameter (Equ. 3.20). This uncertainty, coupled with the uncertainty for all the different optical alignments in the setup makes the absolute signal level in NSOM measurements 131 difficult to be quantified. If such quantification is necessary, a built-in calibration is needed. (ii) Fluorescence Signal The signal level is the primary struggle in a fluorescence NSOM measurement, given the limited excitation power allowed by the tip. The detected fluorescent signal is a product of the following factors: the EM field energy density (photon density) near the tip aperture, probability of absorption by the emitting entity of interest (QD or autofluorescencing protein), conversion of such absorbed photon into photons emitted at wavelength of interest, and the efficiency of the detection instruments. The fluorescent signal from the APD detector (in photon counts per unit time) Fluo S can be estimated to be: APD Obj IX Filter Obj QD QD QD hv Fluo QE T T T QE n c N S ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ = η σ 71 (3.21) In (3.21) hv N is the photon density (# of photon per unit volume), c is the light speed. We note that for both near-field and far-field the absorption of the QD is proportional to the energy density of the EM field (or equivalently the photon density hv N ). For far field c N hv ⋅ is the photon flux (# of photon per unit area and unit time). Though for near-field the photon flux is zero (non-propagating field), the photon density hv N is still a valid quantity; In (3.21) QD n is the number of QDs in the illumination area of the aperture (approximately the aperture area); QD σ is the QD size-dependent absorption cross-section of the QDs (~1nm 2 at 532nm for the 600nm 132 emission CdSe/ZnS QDs, Extinction Coefficient 2.6x10 5 M -1 cm -1 ); QD QE is the quantum efficiency of the QDs, which we assume for the moment is ~50%, the same as that measured in solution; Obj η is the collection efficiency of the objective lens for fluorescence emission (50% for the NA1.45 oil immersion objective lens), Obj T , Filter T , 71 IX T are, respectively, the transmission coefficients of the objective lens, the filter set and the rest of the IX71 microscope. These are, respectively, 85%, 40%, 90% at 600nm; APD QE is the quantum efficiency of the APD detector (60% at 600nm). The photon density hv N can be estimated using the measured total far-field photon fluence and the ratio of near-field to far-field energy F N R / as estimated in §3.3.3(a) as a function of aperture diameter (figure 3.8): ) 1 ( ) /( / 2 F N FF hv R a I c N + ⋅ = ⋅ π (3.22) where FF I is the measured far-field photon fluence at the aperture, F N R / is the ratio of near-field to far-field energy, 2 a π is the area of the aperture. Experimental Determination: In order to check the estimated fluorescence signal, we perform the following single QD imaging experiment. CdSe/ZnS QDs (600nm emission Evident Technology) are sparsely dispersed on pretreated glass cover slip (procedure adapted from [3.32]). A 500nm diameter aperture tip and 5mW 532nm Nd:YAG excitation was used for excitation. The corresponding measured NSOM image is shown in figure 3.10. The bright yellow spot on the image corresponds to the fluorescence 133 from single QDs. The measured far-field photon fluence FF I is 5x10 12 hv/sec. Using Equ. (3.22) c N hv ⋅ is ~2x10 7 hv/(sec·nm 2 ). The estimated fluorescent signal at the APD using Equ. (3.21) is 10 6 count/sec. However the measured fluorescent signal from these single QD is ~1.5x10 4 count/sec, which is ~50 times lower than the estimation. 2 μm Fluorescence (Counts) 1.7x10 4 5.4x10 3 2 μm Fluorescence (Counts) 1.7x10 4 5.4x10 3 Figure 3.10 An NSOM image of single 600nm CdSe/ZnS QDs sparsely dispersed on a glass substrate. The cause for this vast discrepancy is somewhat a mystery as it can not be accounted for only by the uncertainties in the experiment. A potential candidate is 134 the quantum yield of the nanocrystals QD QE . Is it reasonable to assume the same quantum efficiency of the QDs as in solution? One strong perturbation to the QDs is in fact the presence of the metal coated NSOM tip (200nm Al and 20nm Cr coating). In tapping mode imaging condition, tip-surface distance is at most a few nm. Therefore the QDs are in proximity to a metallic surface during imaging! The quenching of fluorophore luminescence near a metallic surface due excitation transfer is well known experimentally and theoretically. There is an established theory which is actually a subset of the more general theme of excitation transfer between NCQDs and substrate to be discussed in lot more detail in chapter 5. Here, in figure 3.11, we simply show the calculated reduction of the apparent quantum yield of CdSe/ZnS (600nm emission) QD as a function of its distance to an aluminum surface using the CPS theory developed by Chance, Prock and Silbey [3.33, 3.34]. The parameters used are shown in the figure inset. Note the lowest optical transition in CdSe nanocrystals has an isotopic 2D transition dipole in the plane perpendicular to its c-axis [3.35, 3.36]. The calculation shows that when the QD is within ~10nm of the aluminum surface, its quantum yield decreases ~10 times. In fact the decrease of fluorophore QY due to interaction with metal coated tip has been experimentally observed by Ambrose and coworkers [3.37]. However this issue seems not to have received the deserved attention in the application of NSOM since then. The degree of QY decrease will certainly dependent on the tip geometry and design. As schematically shown in figure 3.12, three QDs respectively at the center of the aperture and closer to the edge (thus closer to the metallic coating) 135 will suffer different reduction in their QY, with the least reduction for the one at the center. While the presence of the metal coating may decrease the fluorescent signal, this author thinks that an unintended benefit is that the fluorescent quenching by the metallic coat may enable imaging resolution beyond the aperture diameter. In fact given the steep 1/d 3 dependence of the QY reduction for small d<< λ [3.33] (also evident by observing figure 3.11), the author speculates that such fluorescent quenching alone can serve as the contrast mechanism for ultra-high resolution optical microscope. 10 0 10 1 10 2 10 -5 10 -4 10 -3 10 -2 10 -1 10 0 Reduction of Apparent QY: QY(d)/QY( ∞) Distance to Aluminum Surface d (nm) c-axis Parallel to surface c-axis Vertical to surface Random orientation averaged CdSe NC emission 600nm Standalone QY: 50% Media: air (n 1 =1) Aluminum: n 2 =1.20, k 2 =7.26 Aluminum CdSe NC 2D isotopic dipole ┴ c-axis Air c-axis d 10 0 10 1 10 2 10 -5 10 -4 10 -3 10 -2 10 -1 10 0 Reduction of Apparent QY: QY(d)/QY( ∞) Distance to Aluminum Surface d (nm) 10 0 10 1 10 2 10 -5 10 -4 10 -3 10 -2 10 -1 10 0 Reduction of Apparent QY: QY(d)/QY( ∞) Distance to Aluminum Surface d (nm) c-axis Parallel to surface c-axis Vertical to surface Random orientation averaged CdSe NC emission 600nm Standalone QY: 50% Media: air (n 1 =1) Aluminum: n 2 =1.20, k 2 =7.26 Aluminum CdSe NC 2D isotopic dipole ┴ c-axis Air c-axis d Figure 3.11 Reduction of the quantum yield of CdSe QDs as a function of distance to an aluminum surface calculated using CPS theory. The geometry of the system and the parameters used are shown in the figure inset. Note the dipole of the CdSe QD is taken to be isotropic in the plane perpendicular to its c-axis. 136 1 2 3 Substrate Tapered Glass fiber Metallic Coating Metallic Coating QY3>QY2>QY1 Aperture 1 2 3 Substrate Tapered Glass fiber Metallic Coating Metallic Coating QY3>QY2>QY1 1 2 3 Substrate Tapered Glass fiber Metallic Coating Metallic Coating QY3>QY2>QY1 Aperture Figure 3.12 Schematic showing three QDs respectively at the center (QD3) and closer to the edge of the aperture and the metallic coating (QD1). Hence QD1 will suffer a greater reduction in its quantum yield (QY). This quenching of the QY will effectively improve the resolution of NSOM imaging. (iii) Noise and Signal-to-Noise Ratio The potential sources of noise in our NSOM system include: (1) unfiltered excitation light (2) unfiltered AFM control laser (3) ambient light (4) dark count of the APD detector (5) fluctuation of the laser power (6) shot noise. Noise sources (1)-(3) have been minimized by appropriate spectral and spatial filtering as mentioned in the §3.3.2. The APD has ~100 dark counts/sec. These noise sources add up to ~10 3 counts as background on the image. The 532nm Nd:YAG laser has a power fluctuation ~10% on ms time scale, which is the primary reason that lately we switched to water-cooled Ar + laser whose power fluctuation is negligible. In our NSOM imaging system, as in fact in most NSOM systems where signal level (typically means photon counts) is weak, the dominant source of noise is 137 in fact the intrinsic shot noise resulting from the Poisson statistics of the photon counting process: n n e n n n P − = = ! ) ( ξ (3.23) where ) ( n P = ξ is the probability of observing n photon counts during a specific time interval, n is the mean number of photon counts averaged over many time intervals (expectation value). The root mean square deviation of observed photon count from the expectation value is n which is referred to as the shot noise in photon counting. The signal-to-noise ratio (SNR) thus is n , proportional to the square root of the signal. In our typical fluorescence NSOM measurement, the signal is ~10 4 count/sec, the shot noise is thus (10 4 ) 1/2 =100 count in 1Hz bandwidth, or SNR is 100 (Hz 1/2 ). Given a typical integration interval of 5ms during which the tip residing at one position on the sample on average 50 photon counts are gathered for one pixel. The shot noise is 7 counts, hence SNR is limited to 7 by just shot noise alone. Given the stability of our AFM control, such SNR allows images with 256x256 (512x512) pixel image in ~7mins (~28mins). The foremost struggle in our fluorescence NSOM for better image quality is a struggle for higher signal level. Unfortunately, the resolution and signal level are inversely related to the tip aperture diameter. With decreased aperture diameter and the benefit of improved resolution, a heavy penalty is paid for exponentially decreasing signal and SNR. Ultimately it is the requirement for SNR, rather than the 138 available technology to make small aperture at the tip, which limits the resolution. It is worth stressing that the ability to decide the precise location of an object in an image, (schematically shown in figure 3.13) or the ability to resolve different objects depends on both the resolution and SNR. Therefore, depending on the type of sample and the purpose of imaging, an experimentalist needs to optimize the product of resolution and SNR by choosing the appropriate aperture size. Uncertainty in object location Noise Level or SNR -1 Cross section of the image Noise Level Position Intensity 1 FWHM Normalized Signal Uncertainty in object location Noise Level or SNR -1 Cross section of the image Noise Level Position Intensity 1 FWHM Normalized Signal Figure 3.13: Solid line schematically represents the cross-section of a noisy image of an object. The two dotted lines define the upper and lower limit of this noisy cross- section. Note, the uncertainty in determining the location of this object on the image is proportional to both the noise level (or precisely inverse signal-to-noise-ratio) and the resolution of the image (FWHM of the cross-section). 139 §3.4 Simultaneous NSOM imaging of cancer cell morphology and surface receptor distribution. §3.4.1 Her2/neu Surface Receptor and its role in cancer The NSOM system discussed above has been employed to image the morphology and the membrane receptor distribution on cells. Specifically such studies are performed on: breast cancer cells with their Her2/neu surface receptors labeled by appropriately surface functionalized NCQDs. The oncoprotein Her2/neu (also known as ErbB2) is a mutation of normal Her2 receptor which is a transmembrane receptor tyrosine kinase (RTK) of the epidermal growth factor receptor (EGFR) family[3.38, 3.39]. The terminology Her2/neu refers to both the oncogene located on the human chromosome 17 and its product -- a 185 KD transmembrane glycoprotein. Her2 receptor, like other RTKs, only upon ligand binding, turns active and initiates an intracellular signalling pathway (by phosphorylating tyrosine residue on the RTK themselves or other target proteins) which leads to a change in gene expression, cell growth and multiplication. Compared to the normal Her2, Her2/neu oncoprotein has a point mutation that replaces a Valine amino acid into a Glutamine. Such a mutant is constantly active for kinase activity in the absence of ligands which lead to abnormal cell growth and cancerous behavior[3.40, 3.41] as shown in schematic figure 3.14. The mutated Her2/neu are found to be dimerized/oligermized on the cell surface as characterized by FRET based optical imaging techniques [3.42, 3.41]. Such dimerization is thought to be the cause of their malignant effect. Her2/neu has important clinical relevance 140 Oncogenic Mutation Val Æ Glu Dimerization of Her2/Neu Lead to constant activation Oncogenic Mutation Val Æ Glu Dimerization of Her2/Neu Lead to constant activation Figure 3.14 (Top) Schematic of Her2/neu point mutation and dimerization. (Bottom) Intracelluar signalling pathway initiated by activation of growth factor receptor (including Her2/Neu) that leads to cell growth. Figure adapted from [3.41] 141 since overexpression of Her2/neu in tumor cells occurs in 30% of early-stage breast cancers [3.42]. The monoclonal antibody against Her2/neu (trade marked as Herceptin) is now used to treat Her2/neu positive breast cancers and has made a important clinical impact [3.10] even though exact condition of the cancer for Herceptin to be effective is still not clear. In our studies reported in the following, two different representive breast cancer cell lines are used: SK-BR-3 (referred to as SKBR3) and MDA-MB-231 (referred to as MDA hereafter). The SKBR3 cell has an overexpressed Her2/neu receptor density of ~1x10 6 copy per cell [3.45], while MDA has a the lower density of ~2x10 4 copy per cell, similar to the Her2 receptor density on a normal cell. As discussed next, we study the distribution of of Her/neu receptors on these two cell lines using QD labeling and NSOM imaging. §3.4.2 Nanocrystal Quantum Dot Labeling of Her2/neu Receptor The cell labelling is done as shown schematically in figure 3.15. The SKBR3 and MDA cells are first labeled by primary antibody (mouse monoclonal anti- Her2/neu). Then secondary antibody antimouse IgG conjugated CdSe/ZnS QDs are used to label the cells through specific binding to primary antibody. Below we discuss the detailed procedure for QD labelling and the considerations involved. 142 Secondary Ab (AntiMouse IgG) conjugated Quantum Dot (CdSe 605nm) Primary Ab (Mouse Monoclonal Anti-Her2/neu) Her2/neu Surface Receptor Breast Cancer Cell (SK-BR-3, MDA-MB-231) QD QD QD Secondary Ab (AntiMouse IgG) conjugated Quantum Dot (CdSe 605nm) Primary Ab (Mouse Monoclonal Anti-Her2/neu) Her2/neu Surface Receptor Breast Cancer Cell (SK-BR-3, MDA-MB-231) QD QD QD Breast Cancer Cell (SK-BR-3, MDA-MB-231) QD QD QD QD QD QD Figure 3.15 Schematic of QD labeling of Her2/neu surface receptors on breast cancer cell SK-BR-3 and MDA-MB-231. Cell Culture: Breast cancer cell lines SKBR3 and MDA were obtained from American Type Tissue Culture Collection (ATCC). The subcultures were done according to ATCC specified protocols (see Appendix C) on unmodified 25mm diameter glass cover slips (by Mr. Henry Lin, Department of Pathology, University of Southern California). Fixation: The cultured SKBR3 and MDA cells on glass coverslip were fixed by dipping in acetone for 5-10 minutes at room temperature and then let them air dry. The 25mm coverslip was cut into two equal halves to generate two samples for QD labeling and imaging. We note that among different fixation reagents, acetone is excellent in preserving the immonoreactive site [3.46]. Also acetone is a poor penetrator of the cell [3.46]. This property of acetone is an advantage in our case, since we only intend to label the cell membrane receptors. Wash and Blocking: The fixed cell samples were washed by gently rinsed with ~2ml of phosphate buffered saline (PBS) and then incubated in PBS for 2 minutes. 143 Such wash process was repeated 3 times. After wash, the cell samples were incubated for 20mins at room temperature (RT) in 1% bovine serum albumin (BSA) PBS solution preserved by 0.1% sodium azide (referred to hereafter as BSA solution). BSA is the most widely used blocking reagent in immohistological labeling. The blocking reagent has a higher affinity for the hydrophobic area than typical antibodies, therefore prevents antibodies to adsorb via nonspecific hydrophobic interaction. In addition to this initial blocking, it is desirable to have the blocking reagent BSA present in all the subsequent antibody incubation and washing steps to reduce the nonspecific antibody adsorption, as done in our work. Primary Antibody Labeling: Excessive BSA solution was wicked off the samples. The sample was incubated at RT for an hour in 75 μl Mouse Monoclonal Anti- Her2/neu primary antibody solution at saturating concentration ~0.65 μg/ml (Biogenex, concentrated anti-Her2/neu MU134-UC 1:20 diluted in 1% BSA solution). The antibody is directed against and conjugates to the internal domain of Her2/neu oncoprotein. After the incubation, the sample was gently washed with BSA solution (rinse with ~2ml of BSA, and incubation in BSA for 5 minutes) for 3 times to remove the excessive unbound antibody. Labeling with Secondary Antibody Conjugated NCQD: Excessive BSA solution was wicked off the primary antibody labeled cell sample using filter paper. The sample was incubated at RT for an hour in 75 μl 20nM secondary antimouse IgG antibody conjugated CdSe/ZnS NCQD (emission 605nm, from Invitrogen Corp.) solution with 1% BSA. The secondary antibodies conjugated with QDs specifically 144 bind to the primary antibody Mouse Monoclonal Anti-Her2/neu. Typically more than one QD (other dye) conjugated secondary antibodies can binds to one primary antibody subject to the limitation of geometrical interference, of course. In fact the two-step labeling often serve as a way to amplify the fluorescent signal of the desired antigen. (If necessary further amplification can be achieved by adding yet one more labeling step using tertiary antibodies targeting the secondary antibodies.) After the incubation with QD conjugated secondary antibody, the sample was gently washed with PBS solution (rinse with ~2ml PBS, incubate in PBS for 5 minutes) 3 times to remove the unbound conjugated NCQDs. Excessive liquid was wicked of the sample as much as possible and then the sample was air-dried and stored in a glovebox for later imaging studies. §3.4.3 NSOM based simultaneous morphological and optical imaging of cells The 605nm emission QD labeled SKBR3 and MDA breast cancer cells prepared as discussed above were imaged using the AFM based NSOM technique. A typical imaging procedure is illustrated in figure 3.16. First, under conventional bright field or fluorescent microscopy (figure 3.16(a)), the NSOM tip was brought to the area of interest on a QD labeled SKBR3 cell. NSOM was employed to simultaneously image the morphology and the fluorescence from this selected area (figure 3.16(b) only NSOM image shown) as discussed in §3.3. Higher resolution 145 4 μm 1 μm Fiber Tip Aperture ~200nm 20 μm Optical Microscopy NSOM NSOM PL Spectrum (a) (b) (c) (d) NSOM Tip Fluorescence (Counts) 2.6x10 3 Fluorescence (Counts) 6.14x10 4 488nm Ar+ excitaton 7.4x10 4 2.2x10 3 4 μm 4 μm 1 μm Fiber Tip Aperture ~200nm 20 μm Optical Microscopy NSOM NSOM PL Spectrum (a) (b) (c) (d) NSOM Tip Fluorescence (Counts) 2.6x10 3 Fluorescence (Counts) 6.14x10 4 488nm Ar+ excitaton 7.4x10 4 2.2x10 3 Figure 3.16 An illustration of the typical procedure for NSOM imaging (a) Under conventional bright field microscopy the NSOM tip was brought to the area of interest on a 605nm QD labeled SKBR3 cell. (b) A fluorescence NSOM image of the selected area indicated by the white box in image a, obtained using 200nm diameter tip. (c) Higher resolution NSOM image zoomed into the selected area indicated by the white box in image b. (d) With the same tip stopped at the indicated position on image c, the fluorescent spectrum measured using excitation from the NSOM tip (488nm single mode Ar+ laser coupled). The broad peak in the spectrum corresponds to emission from multiple QDs, the two narrow peaks at 615nm and 625nm are the residual Ar+ laser line. 146 image of particular area of interest was obtained with further zoom-in with NSOM (figure 3.16(c) only NSOM image shown). Then if necessary, the NSOM tip can be moved to a position of interest on the image, the detection is switched to a spectrometer and a fluorescent spectrum was taken (figure 3.16(d)). The measured spectrum here corresponds to the fluorescence from the 605nm CdSe/ZnS QD labels. Note that in the QD labeled cancer cells, the NSOM imaging does not show isolated single QDs, indicating a high density of Her2/neu receptors on the cell surface. There is, on average one Her2/neu receptor in 20nmx20nm area on a typical SKBR3 cell (assuming 1x10 6 Her/neu per SKBR3, and 20 μmx20 μm cell size). Even assuming each Her/neu is labeled by one QD (neglecting the amplification by primary antibody), then within the excited area of a typical NSOM tip of 100-200nm diameter, there are tens of QDs being excited. The formed NSOM image is a reflection of the local QD density and hence Her2/neu receptor density. . In figure 3.17(a, b) is shown simultaneous AFM and fluorescent NSOM images for QD labeled SKBR3 cells, imaged using a tip of 100nm diameter aperture. The resolution of the NSOM image (panel b), as indicated by the smallest features on it, is ~150nm (see Xsec1 and Xsec2), which is 2.6 times better than the best far-field epifluorescence image at the same wavelength (400nm, see §3.2.1). Note that the fluorescence NSOM image of SKBR3 cell is marked by bright areas of ~500nm diameter. This indicates Her2/neu receptors on the SKBR3 cell surface are not distributed evenly but instead localized in clusters. The AFM and NSOM images of QD labeled MDA cells are shown in figure 3.17 (c) and (d), 147 1 μm 1 μm SKBR3 MDA AFM -- Morphology NSOM – Optical Response 1 μm SKBR3 0 200 400 600 800 1000 (nm) FWHM 150nm 0 200 400 600 800 1000 (nm) FWHM 150nm X-Sec2 1 μm MDA 189 1690 Height (nm) Height (nm) 60.5 263 0 4.2x10 3 Fluorescence (Counts) Fluorescence (Counts) 1.6x10 4 0 (a) (c) (b) (d) X-Sec1 1 μm 1 μm 1 μm 1 μm SKBR3 MDA AFM -- Morphology NSOM – Optical Response 1 μm SKBR3 1 μm SKBR3 0 200 400 600 800 1000 (nm) FWHM 150nm 0 200 400 600 800 1000 (nm) FWHM 150nm 0 200 400 600 800 1000 (nm) FWHM 150nm 0 200 400 600 800 1000 (nm) FWHM 150nm X-Sec2 1 μm MDA 1 μm MDA 189 1690 Height (nm) Height (nm) 60.5 263 0 4.2x10 3 Fluorescence (Counts) Fluorescence (Counts) 1.6x10 4 0 (a) (c) (b) (d) X-Sec1 Figure 3.17 Simultaneously obtained AFM and fluorescent NSOM images of QD (CdSe/ZnS 605nm) labeled SKBR3 cells (panel a, b) and MDA cells (panel c, d). Image obtained with tip of 100nm diameter aperture. Insets (XSec1) and (XSec2) show two cross-sections on the NSOM image (panel b) to illustrate the smallest features on the image. The resolution of the image as demonstrated by the FWHM of such feature is <150nm. 148 respectively. The fluorescence NSOM image (panel d) again shows areas marked by bright areas ~500nm in diameter, indicating the same clustered Her2/neu receptor distribution as on SKBR3 cells. The only difference is that the number of such clusters on MDA is much smaller than on SKBR3 cell, which is understandable given the total lower number of Her2/neu receptor on MDA (2x10 4 /cell) than on SKBR3 (1x10 6 /cell). The number of QDs in each cluster is difficult to quantify due to reasons stated in §3.3.3(b), one of these being the uncertain degree of QD yield reduction due to the nearby metallic coating on the tip. However, given the comparison to fluorescent signal of a single QD obtained in our single QD imaging experiments (§3.3.3(b)), the best estimated number of QDs in each cluster is on the order of 10 3 . Recall §3.4.1, the mutation in the Her2/neu receptors (with respect to normal Her2) are dimerized on cell surface. Our observation demonstrates that beyond the dimerization, the Her2/Neu receptors exist in clusters of ~500nm diameter on both SKBR3 and MDA surface. The same driving force for such clustering may be similar to that for dimerization which ultimately has to be related to the structural mutation of Her2/neu. Further study on the distribution of Her2 receptor on normal cells as the control experiment is important to check whether the clustering exists in normal cells or not, and whether the clustering is a reflection of the metastasis of the cell. However such experiments on normal cells could not be undertaken as there are strict rules governing used of normal human cells and such could not be made available to us. 149 By observing the corresponding AFM and NSOM images in figure 3.17 no obvious similarity is noticed between the cell morphology and the Her2/neu receptor distribution as reflected in the QD fluorescence. To quantify the relation of the two, we calculated the correlation coefficient between the simultaneously obtained AFM height and fluorescence intensity in the NSOM image: ( ) ( ) () () ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ − ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ − − − = ∑∑ ∑∑ ∑ ∑ mn mn mn mn mn mn mn F F H H F F H H CorrCoef 2 2 (3.X) where mn H and mn F are respectively the height and intensity of pixel (m, n) on the AFM and NSOM images. H and F are the average height and intensity. The calculated correlation coefficient is a number in [-1,1] and characterizes the degree to which the two variables (height and fluorescence intensity) at each pixel are related. In extreme cases, if the height and fluorescence has a linear relationship with positive (or negative) slope the correlation coefficient is 1 (or -1). If wherever the morphology is high, the fluorescent intensity is also high then the correlation coefficient is close to 1, and vice verse. For a set of 6 AFM and NSOM images obtained on two SKBR3 cell, the calculated correlation coefficients are {0.339, -0.405,-0.120, 0.163, 0.256, 0.602}. Therefore with the limited data so far, we can conclude no correlation between the AFM and NSOM image, hence no correlation between morphology and the Her2/neu receptor distribution. 150 §3.4.4 Future direction — NIR QDs as fluorescent labels So far in the experiments the QDs used for labeling are the commercially available CdSe/ZnS (605nm emission) nanocrystals in the visible wavelength regime. As stated in chapter 1, in comparison, labeling with near infrared (NIR) QDs has many advantages and is our focus of research. In the context of optical detection of disease, the most important advantage of NIR labeling is the lower autofluorescence background from the cell itself. Shown in figure 3.18 upper (lower) panel is the fluorescence spectrum of a SKBR3 cell with no labeling (with 605nm QD labeling). Excitation is with 488nm Ar + laser (far-field, through back port of IX71) that illuminates the whole cell. Fluorescence spectra in the 500-1000nm and 1000-1500nm regime are obtained, respectively, with CCD and InGaAs PD array detectors and are glued together. Obviously, the autofluorescence in the 900-1500nm is indeed significantly lower than in the visible range. With the 605nm QD labeling (figure 3.18 lower panel), the QD fluorescence is riding on a fluorescent background of comparable intensity. The background level is especially important in the spectroscopic detection of early stage disease, in which case one intends to distinguish fluorescence signal from the smallest number of QD labels against the background. The autofluorescence of the cell then will be the ultimate limitation on the sensitivity. Therefore it is highly desirable to move into the 900-1500nm NIR regime, using for example the InAs/ZnSe core/shell nanocrystals discussed in chapter 2. Indeed, the 151 biocompatible surface functionalization of the InAs/ZnSe nanocrystal and its use in cell labeling should be the next step of investigation. SKBR3 + QD Excitation 488nm SKBR3 Excitation 488nm Labeled with 605nm QD SKBR3 + QD Excitation 488nm SKBR3 Excitation 488nm Labeled with 605nm QD Figure 3.18 (Upper) Fluorescence spectrum of a SKBR3 cell with no labeling. (Lower) Fluorescence spectrum of a QD labeled SKBR3 cell. The peak at 605nm corresponds to the emission from 605nm emission CdSe/ZnS QDs. Both spectra were taken at room temperature with 488nm Ar + laser excitation (far-field, through back port of IX71). The 500-1000nm spectral range was measured with a Si CCD detector, and the 1000-1500nm spectral range was measured with an InGaAs array detector. Note the low fluorescence background from the SKBR3 cell in the NIR region (900-1500nm). 152 Chapter 3 References: [3.1] M. Bruchez Jr. M. Moronne, P. Gin, S. Weiss, A. P. Alivisatos, “Semiconductor Nanocrystals as Fluorescent Biological Labels”, Science 281, 2013- 2015 (1998). [3.2] W. C. W. Chan and S. Nie, “Quantum Dot Bioconjugates for Ultrasensitive Nonisotopic Detection”, Science, 281, 2016-2018 (1998). [3.3] A. P. Alivisatos, W. W. Gu, C. Larabell “Quantum dots as cellular probes”, Ann. Rev. Biomed. Eng. 7, 55-76 (2005) . [3.4] I. L. Medintz, A. R.Clapp, H. Mattoussi, E. R. Goldman, B. Fiber, J. M. Mauro, “Self-assembled nanoscale biosensors based on quantum dot FRET donors”, Nature Materials, 2, 630-638 (2003). [3.5] J. H. Kim, D. Morikis, M. Ozkan, “Adaptation of inorganic quantum dots for stable molecular beacons”, Sensor and Actuat. B – Chem. 102, 315-319 (2004). [3.6] K. Sokolov, M. Follen, R. Richards-Kortum, “Optical Spectroscopy for Detection of Neoplasia”, Curr. Opin. Chem. Biol., 6, 651-658 (2002). [3.7] R. R. Alfano, G. C. Tang, A. Pradhan, W. Lam, D. S. Choy, and E. Opher, “Fluorescence spectra from cancerous and normal breast and lung tissues”, IEEE J. Quantum Electronics, 23, 1806-1811 (1987). [3.8] X. H. Gao, Y. Y. Cui, R. M. Levenson, L. W. K. Chung, S. M. Nie, “In vivo cancer targeting and imaging with semiconductor quantum dots”, Nature Biotechnol., 22, 969-976 (2004). [3.9] L. Josephson, M. F. Kircher, U. Mahmood, Y. Tang, R. Weissleder, “Near- infrared fluorescent nanoparticles as combined MR/optical imaging probes”, Bioconjugate Chemistry, 13, 554-560 (2002). [3.10] U. Joos, T. Biskup, O. Ernst, I. Westphal, C. Gherasim, R. Schmidt, K. Edinger, G. Pilarczyk, C. Duschl, “Investigation of cell adhesion to structured surfaces using total internal reflection fluorescence and confocal laser scanning microscopy”, Eur. J. Cell Biol. 85, 225 –228 (2006). [3.11] T. Funatsu, Y. Harada, M.Tokunaga, K. Saito, T. Yanagida, “Imaging of single fluorescent molecules and individual ATP turnovers by single myosin molecules in aqueous-solution”, Nature, 374, 555-559 (1995). 153 [3.12] K. Betzig, J. K. Trautman, T. D. Harris, J. S. Weiner, R. L. Kostelak, “Breaking the diffraction barrier: Optical Microscopy on a nanometric scale.” Science, 251, 1468-1470 (1991). [3.13] J. J. Greffet, R. Carminatic. “Image formation in neear-field optics”, Prog. Surf. Sci. 56, 133-237 (1997). [3.14] E.H. Synge, "A suggested method for extending the microscopic resolution into the ultramicroscopic region" Phil. Mag. 6, 356 (1928). E.H. Synge, "An application of piezoelectricity to microscopy", Phil. Mag., 13, 297 (1932). [3.15] E. Abbe, “Beitrage zur theorie des mikroskops und der mikroskopischen wahrnehmung,” Archiv fur Mikroskopische Anatomie 9, 413-463 (1873). [3.16] Near-field nano/atom optics and technology, Ed. M. Ohtsu, Springer 1998. [3.17] D. Courjon, Near-field microscopy and near-field optics, Imperial College Press, London (2003). [3.18] E. Betzig, P. L. Finn, J. S. Weiner, “Combined shear force and near-field scanning optical microscopy”, Appl. Phys. Lett. 60, 2484-2486 (1992). [3.19] O. Zhong, D. Inniss, K. Kjoller, V. B. Elings, ”Fractured polymer silica fiber surface studied by tapping mode atomic-force microscopy”, Surf. Sci. 290, L688- L692 (1993). [3.20] J. P. Cleveland, B. Anczykowski, A. E. Schmid, V. B. Elings, “Energy dissipation in tapping-mode atomic force microscopy”, Appl. Phys. Lett. 72, 2613- 2615 (1998). [3.21] K. Lieberman, N. Ben-Ami, A. Lewis, “A fully integrated near-field optical, far-field optical, and normal force scanned probe microscope”, Rev. Sci. Instrum. 67 3567-3572 (1996). [3.22] T. R. Ramachandran, “Scanning probe microscopy studies of the highly strained epitaxy of InAs on GaAs(001) and Scanning Probe Based Imaging and Manipulation of Nanoscale Three-Dimensional Objects”, Ph. D. Dissertation, University of Southern California (1998). [3.23] Atomic Force Microscopy in Cell Biology (Methods in Cell Biology, Volume 68), Eds. B. P. Jena, J. K. H. Horber, Academic Press (2002). [3.24] M. Born, E. Wolf, “Principle of Optics”, Pergamon Press, London (1980). [3.25] J. D. Jackson, Classical Electrodynamics, Wiley (1998). 154 [3.26] H. A. Bethe, “Theory of diffraction by small holes”, Phys. Rev. 66, 163-182 (1944). [3.27] J. C. Bouwkamp, “Diffraction Theory”, Rep. Prog. Phys. 17, 35-100 (1954). [3.28] Y. Leviatan, “Study of near zone fields of a small aperture”, J. Appl. Phys. 60, 1577-1583 (1986). [3.29] U. Durig, D. W. Pohl, F. Rohner, “Near-field optical scanning microscopy“, J. Appl. Phys, 59, 3318-3327 (1986). [3.30] E. Betzig, A. Harootunian, A. Lewis, and M. Isaacson, “Near-field diffraction by a slit: implications for superresolution microscopy”, Appl. Opt. 25, 1890-1900 (1986). [3.31] N. A. McDonald, "Electric and Magnetic Coupling through Small Apertures in Shield Walls of Any Thickness," IEEE Trans. Microwave Theory Tech. MTT-20, 689 (1972). [3.32] Y. Ebenstein, T. Mokari, U. Banin, “Fluorescence quantum yield of CdSe/ZnS nanocrystals investigated by correlated atomic force and single-particle fluorescence microscopy”, Appl. Phys. Lett. 80, 4033-4035 (2002). [3.33] R. R. Chance, A. Prock, R. Silbey, “Molecular Fluorescence and Energy Transfer Near Interfaces” Vol. 37, p.1 in Advance in Chemical Physics, Eds. S. A. Rice and I. Prigogine, Wiley-Interscience, New York (1978). [3.34] R. R. Chance, A. Prock, R. Silbey, “Lifetime of an Excited Molecule Near a Metal Mirror: Energy Transfer in the Eu3+/Silver System” J. Chem. Phys. 60, 2184- 2185 (1974). [3.35] I. Chung, K. T. Shimizu, M. G. Bawendi, “Room temperature measurement of the 3D orientation of single CdSe quantum dots using polarization microscopy“, P. Natl. Acad. Sci. USA 100, 405-408 (2003). [3.36] Al. L. Efros, M. Rosen, M. Kuno, M. Nirmal, D.J. Norris, M. Bawendi, “Band-edge exciton in quantum dots of semiconductors with a degenerate valence band: Dark and bright exciton states”, Phys. Rev. B 54, 4843-4856 (1996). [3.37] W. P. Ambrose, P. M. Goodwin, J. C. Martin, R. C. Kell, “Alternations of single molecule fluorescence lifetimes in near-field optical microscopy”. Science 265, 364-367 (1994). 155 [3.38] L. Alroy, Y. Yarden. “The ErbB signaling network in embryogenesis and oncogenesis: signal diversification through combinatorial ligand-receptor interactions.” FEBS (Federation of European Biochemical Societies) Lett. 410, 83- 86 (1997). [3.39] N. E. Hynes, D. R. Stern, “The biology of erbB2/neu/HER-2 and its role in cancer”, Biochimica et Biophysica Acta 1198, 165-184 (1994). [3.40] C. I. Bargmann, M. C. Hung, R. A. Weinberg, “Multiple independent activations of the neu oncogene by a point mutation altering the transmembrane domain of p185” Cell 45, 649-657 (1986). [3.41] H. Lodish, A. Berk, S .L. Zipursky, P. Matsudaira, D. Baltimore, J. Darnell, Chapter 24, Section 2 and 3 in Molecular Cell Biology, 4 th Edition, Freeman, New York (2001). [3.42] T. W. Gadella, T. M. Jovin, “Oligomerization of epidermal growth factor receptors on A431 cells studied by time-resolved fluorescence imaging microscopy. A stereochemical model for tyrosine kinase receptor activation” J. Cell Biol. 129, 1543-1558 (1995). [3.43] J. Bange, E. Zwick E, A. Ullrich, "Molecular targets for breast cancer therapy and prevention". Nature Medicine 7, 548–552 (2001). [3.44] A. R. Tan, S. M. Swain. "Ongoing adjuvant trials with trastuzumab in breast cancer". Seminars in Oncology 30 (5 Suppl 16), 54–64 (2002). [3.45] J. C. Sarup, R. M. Johnson, K. L. King, B. M. Fendly, M. T. Lipari, M. A. Napier, A. Ullrich, H. M. Shepard, “Characterization of an anti-p185HER2 monoclonal antibody that stimulates receptor function and inhibits tumor cell growth.” Growth Regul. 1, 72-82. (1991). [3.46] Immunochemical Staining Methods Handbook, Ed. T. Boenisch, DakoCytomation (2001). 156 Chapter 4. Solid (Semiconductor) Surface Modification and Adsorption of NCQDs In the previous two chapters, we have discussed the synthesis of the nanocrystal quantum dots (chapter 2) and their use as biological labels (chapter 3). In this chapter we turn to the third ingredient involved in our proposed application of a hybrid colloidal/epitaxial structure for biological detection or activation (figure 4.1) as preludized in chapter 1 – the construction of colloidal/epitaxial hybrid structures. This topic is divided into four sections. In §4.1, we discuss an intrinsic challenge that the epitaxial quantum nanostructures face as being a part of such integrated structures – the degradation of optical performance due to surface recombination when brought close to the substrate surface and subjected to hostile biological fluid environment. The initial effort to ease such challenge via sulfur based surface passivation will also be presented. Then in §4.2, we discuss the simplest construction of colloidal/epitaxial hybrid structure based on direct deposition of nanocrystals on unmodified substrates using a refined evaporative assembly technique – dip-coating. In §4.3 we discuss the generic strategy to chemically modify he semiconductor surface via covalently grafted self-assembled monolayer (SAM). We present the procedure and result for SAM functionalization as well as the use of SAMs as bilinkers to direct chemisorption of nanocrystals on surface to create nanocrystal/SAM/substrate hybrid structures. 157 Figure 4.1 Schematic of biological sensing using integrated colloidal/epitaxial hybrid nanostructure. §4.1 Near-surface quantum nanostructure and GaAs surface passivation §4.1.1 Brief investigation of optical property of near-surface quantum structure As described in chapter 1, we are aiming at establishing communication between the NCQDs and the epitaxial quantum structures (such as quantum wells or self-assembled quantum dots) contained in the semiconductor substrates underneath. Therefore the epitaxial quantum structures have to be “near surface” to allow certain physical mechanism of interaction with the NCQDs. These interactions as will be detailed in chapter 6 typically have effective distance of a few nm or less. Naturally it is desirable to put the epitaxial nanostructure as close to the surface (hence to the NCQDs) as possible to allow maximum interaction. However, close to the surface is 158 a double-sided sword, since we fully expect the optical and electrical performance of quantum nanostructures to degrade when getting closer to the surface. This is so because “natural” semiconductor surface has localized surface states arising from dangling bonds, amorphousness and nonstoichiometry at the surface, or surface contamination (such as oxides). The energy of these surface states often resides in the semiconductor bandgap and gives rise to surface recombination of excited carriers [4.1]. This provides an enhanced nonradiative recombination channel for the excitations in the underlying quantum nanostructures. We expect the problem will be especially severe in our GaAs based material system which has a high surface state density on the order of 10 13 cm -2 eV -1 and surface recombination velocity on the order of 10 6 cm/sec [4.2]. Given the above considerations, it is important for us to understand the optical property of the near surface quantum structures and fold it into our design considerations for integrated nanostructures. Here as a model system we study the PL behavior of our molecular beam epitaxy (MBE) grown self-assembled 3- dimensional epitaxial island quantum dots (SAQDs) [4.3, 4.4] as a function of distance to the surface. The starting sample is our standard MBE grown 2.5ML InAs pyramid-shaped SAQDs on semi-insulating (001) GaAs buffer (16sec/ML, 500 C) and capped by 170ML (~48nm) GaAs using migration enhanced epitaxy (MEE) approach (16sec/ML, 350 C) [4.5]. This GaAs capping layer thickness is reduced to varying extant by a wet chemical etching process. The etchant used is a mixture of NH 4 ·OH:H 2 O 2 (30%):H 2 O (10:11:600 v/v) that has a GaAs etch rate of ~1nm/sec at 159 room temperature. In order to precisely determine the etch depth, before etching, the sample was partially masked by photoresist (PR). After etching, the PR was removed, and atomic force microscope (AFM) was used to measure the height difference between the etched area and the masked area, from which the remaining GaAs capping thickness was calculated. The root mean square (RMS) roughness of such etched surface is ~ 0.5nm. Four SAQD samples with GaAs capping layer thickness of 170ML (as grown), 105ML, 74ML and 40ML were prepared. The sample structure is schematically shown in figure 4.2. Given that the average height of the 2.5ML InAs SAQDs is ~6nm [4.5], the distance between the apex of the SAQD and the sample surface in these 4 samples is respectively, 42.0nm, 23.7nm, 14.9nm, 5.3nm. PL of the set of samples was measured using Ar + 514nm excitation at 78K. Excitation power density is low at 2-5W/cm 2 . The PL spectra are shown in figure 4.3(a). Figure 4.3(b)- (d) show the measured PL peak intensity, peak position and FWHM respectively. GaAs Substrate 2.5 ML InAs SAQD Cap Layer Thickness MEE GaAs Capping ~6nm Wetting Layer GaAs Substrate 2.5 ML InAs SAQD Cap Layer Thickness MEE GaAs Capping ~6nm Wetting Layer Figure 4.2 Schematic of pyramid InAs SAQD with GaAs capping. 160 Indeed, as expected, from this measurement we observe the loss of PL intensity in the near surface SAQDs (Figure 4.3(b)). When the capping layer thickness goes below ~100ML (28nm) the PL intensity of the SAQD starts to drop. At capping thickness 40ML (11nm) the PL intensity degrades significantly to ~20 times lower than for 170ML (48nm) capping. We note that similar loss of PL had been seen in the case of near surface GaAs/AlGaAs and InGaAs/GaAs quantum wells (QWs) by a few groups [4.6, 4.7]. In these studies, phenomenological models which assume surface recombination due to tunneling of confined carriers from the QW states to midgap surface state seemed to reasonably explain the drop in QW PL intensity and the decrease in the QW luminescence lifetime as function of the distance between the QW and the surface. We think the lowering of PL intensity from the near surface SAQD is likely due to a similar mechanism of surface-related recombination, although in our case, because of the 3D shape of the SAQD, the distance between SAQD to the surface is ill-defined. Interestingly, other than the decrease in intensity, significant red shifts and broadening of PL width are also observed as when the surface gets closer to the SAQDs ( figure 4.3(b)&(c) ). This observation provides experimental evidence for the prediction made by the stress simulation work from our group [4.8]: the hydrostatic and biaxial stress field distributions in the interior of the capping layer and the SAQD depend on the capping layer thickness. Consequently, the SAQD 3D confining potential and hence the electron energy states vary with the capping layer. With thinner capping layer, decreased hydrostatic compressive stress is expected in 161 1000 1100 1200 1300 1400 Wavelength (nm) (a) 1000 1100 1200 1300 1400 Wavelength (nm) 1000 1100 1200 1300 1400 Wavelength (nm) (a) (b) (b) (c) (c) (d) (d) Figure 4.3 (a) PL of 2.5ML InAs SAQD with 170ML, 105ML, 74ML, 40ML GaAs capping (at 78K). Corresponding PL Peak Intensity (b), Peak Position(c), FWHM (d). 162 the SAQDs, which lowers the conduction bandedge and raises the valence bandedge in the SAQD, and thus causes the PL to red shift. The broadening of the PL width reflects the fact that the larger SAQDs (the apexes of which are closer to the surface) experience a larger stress reduction and hence a larger red shift in PL. • Impact of Protein Adsorption Further deterioration of the emission from the near surface SAQDs is observed when exposed to biological fluids and contaminated with adsorbent from the solution. In figure 4.4 is shown the PL of as-prepared 2.5ML InAs SAQD with 40ML GaAs capping and after adsorption from solution of the protein ferritin. The ferritin protein adsorption is brought about by incubating the near surface SAQD sample in 1mg/ml cationized ferritin aqueous solution (from horse spleen; Sigma Corp.) for 5mins followed by thorough washing with 10ml of DI water. Sample is then blow dried with N 2 and loaded into the cryostat for PL measurements. The corresponding AFM measurements show one monolayer of ferritin was uniformly adsorbed on GaAs substrate. In figure 4.4, after ferritin adsorption the SAQD PL is seen to drop to half of the original, which is likely due to the surface contamination accompanying the ferritin adsorption. The SAQD PL peak position is also slightly red shifted by 7meV. The red shift observed may be the result of Stark effect due to the charge of the ferritin molecule and corresponding electrical field at the sample surface. 163 Figure 4.4 PL of as prepared 2.5ML InAs SAQD with 40ML GaAs capping (black) and after ferritin adsorption (red). §4.1.2 Sulfur based GaAs surface passivation One generally accepted strategy to tackle the degradation of optical property of near-surface quantum structures or bulk material due to surface states and attendant carrier recombination is “surface passivation”. In this process, certain chemical species are reacted with the surface to saturate the dangling bonds and to push the surface states out of the semiconductor’s bandgap. For the GaAs surfaces, sulfur based wet chemical passivation has been investigated in the literature and gives good results in removing surface oxide layer and saturating surface Ga and/or As bonds [4.9-11]. Specifically, we have studied GaAs surface passivation using a 164 (NH 4 ) 2 S based procedure adapted from [4.9] and observed improved PL response from bulk GaAs and near surface quantum dots as discussed below. Procedure for (NH 4 ) 2 S based sulfur passivation: GaAs substrate was first degreased sequentially in trichloroethylene (TCE), acetone, methanol and DI water for 5mins in each solvent and then dried by a N 2 jet. GaAs surface oxide was stripped by immersing the substrate in H 2 SO 4 :H 2 O (3:1 v/v) for two minutes. Then the substrate was transfer directly into 50 C 5% (NH 4 ) 2 S aqueous solution and was immersed for 10mins (while the solution is stirred). Then the substrate is taken out and blown dry under a flow of N 2 . Structurally, the passivation process does not noticeably increase the GaAs roughness. Shown in figure 4.5 is an AFM image of a MBE prepared GaAs sample after S-passivation. The rms roughness is 0.15nm. Cross-sectional TEM studies showed that the passivation above specified process etches away ~15ML of GaAs. Two samples, (1) an epi-ready semi-insulating GaAs substrate (as received) and (2) a 2.5ML InAs SAQD with 40ML GaAs capping were used to study the effect of passivation on the PL response. The corresponding results are shown in figure 4.6 (a) and (b). For both GaAs substrate and the InAs SAQDs, the room temperature PL enhancement was dramatic: 34 and 11 times, respective. Given this encouraging result, the above noted sulfur passivation procedure has been made a standard practice in constructing our colloidal/epitaxial heterostructures as used in the studies to be discussed in chapter 5. 165 Other than the quenching of the PL due to surface recombination, the important effect of the surface is the shift of the energy levels in the substrate due to the pinning of Fermi level by surface states, often referred to as surface band bending in bulk semiconductors. We expect this effect will play an important role in the excitation transfer between the nanocrystals and substrate in the hybrid structures, since it affects the relative energy level alignment between the two. There are no systematic data on this aspect in the literature for GaAs surface finished and treated by various chemical methods. For a thorough understanding of the excitation transfer phenomena to be discussed in chapter 5, the characterization of surface band bending for our samples is an important aspect of future studies. Figure 4.5 AFM image of Sulfur-passivated GaAs surface. (RMS roughness 0.15nm ) 166 Figure 4.6 (a) PL of untreated (black) and sulfur passivated (red) S-I GaAs substrate. After passivation, integrated PL intensity was enhanced 34 times. (b) PL of untreated (black) and sulfur passivated (red) near surface InAs SAQD sample (2.5ML InAs with 40ML GaAs capping). After passivation, integrated PL intensity from SAQD was enhanced 11 times. (a) (b) 167 §4.2 Adsorption of NCQDs on unmodified substrates via controlled evaporation: dip-coating technique In this section we discuss the simplest construction of colloidal/epitaxial hybrid structures through controlled assembly of nanocrystals from solution phase onto substrates without specific chemical functionalization. Although our limited aim here is to achieve controlled assembly of a monolayer high layer of nanocrystals uniformly over large area (> a few millimeters) on a semiconductor substrate for the subsequent optical study of the energy/charge transfer between the two (to be discussed in chapter 5), the controlled assembly of pre-synthesized nanometer size colloidal objects on a substrate is, in general an important paradigm to economically create complex nanostructured materials for many envisioned applications without invoking expensive nanolithography [4.12]. Whatever the method for assembly on solid surface may be, it always involves two conceptual elements: (1) the transport of the objects to the surface, and (2) the self-assembly of the objects due to their mutual interaction. The mechanism of transport and the force involved in the self-assembly classifies different methods and determines their applicability and limitation. For example if the transport of objects toward the surface has equal probability at all location on surface, then long- range 2D ordering of the objects is by definition not possible. Long-range 2D ordering has to rely on transport (at least initially) preferred towards specific direction, such as the edge of the substrate. The major methods reported in the literature for the ordered assembly of pre-synthesized colloidal objects on solid surface are: (a) Creating order templates using assembly of biological objects 168 (proteins, DNAs) on a solid surface, exploiting self-diffusion of objects and specific biochemical interaction between the objects (Biochemical assembly), see e. g. [4.13, 4.14]; (b) methods based on Lagmuir-Blodgett assembly of objects at liquid interface and subsequent transfer of the assembly onto a solid surface [4.15, 4.16]; (c) methods based on convective flow of objects in fluidic cells and flow induced interaction between objects, e.g. see [4.17, 4.18]; (d) methods based upon electrophoretic transport and flow induced interaction between objects. e.g. see [4.19-21]. (e) methods based on transport due to evaporation of solvent and the capillary interaction between objects (referred to hereafter as evaporative method), e.g. see [4.22-25]. Our focus has been on the evaporative method, given its simplicity and its potential to achieve long range ordering. Indeed amazingly simple, the drying of a droplet of colloidal solution can lead to ordered self-assembly of the colloidal particles due to the concentration of particles at the receding edge of the droplet and the mutual attractive capillary interaction. This phenomenon was accidentally discovered by Perrin in his classical work on determining the Avogadro number [4.26], in which monodispersed gomme-gutte ( μm size) particles formed ordered monolayer high array on glass slides after the suspension dried up. Lately, many authors have studied this method and applied it to form 2D assembly of nanocrystals on substrates [e.g. 4.24, 4.27]. In order to obtain large area uniformity, we have adapted a more controlled version of colloidal evaporative assembly as devised by Dimitrov and coworkers [4.23] – the dip-coating approach. 169 In dip coating, instead of letting the droplet of colloidal solution dry uncontrollably, a wettable substrate is dipped into a colloidal suspension and pulled out at a constant speed using an apparatus schematically show in figure 4.7. We constructed the apparatus using a motorized actuator (Encoder Mike, Oriel Instrument) driven linear translational stage. The speed of withdrawal can be varied between 0.11 μm/sec-110 μm/sec. If controlled appropriately, this dip-coating process results in a layer of 2D ordered array of particles to grow on the substrate near the substrate-suspension-air contact line. Figure 4.8 schematically depicts the meniscus of the suspension near the contact line where the leading edge of particle array grows. Below we first discuss how the array grows by incorporating particles in the suspension into it, and then return to the question of how the array initiates in the first place. Vial Colloial Suspension Clamp Substrate Linear Translational Stage Driven by Motorized Actuator Withdraw Vial Colloial Suspension Clamp Substrate Linear Translational Stage Driven by Motorized Actuator Withdraw Figure 4.7 Schematic of the apparatus for dip-coating. 170 Figure 4.8 Schematic depicts the meniscus of the suspension near the contact line where the leading edge of array grows by incorporating particles in the suspension. Figure adapted from [4.23]. §4.2.1 Growth of ordered colloidal particle array Colloidal particles from bulk suspension are transported towards the meniscus and the existing array due to the evaporation of the solvent from the meniscus and the induced convective flow. The detailed hydrodynamics of this transport is interesting subject for future research, as it involves a complex interplay between (a) pressures in the meniscus (capillary, gravitational, disjoining pressure); (b) solvent evaporation and convective flow into the meniscus; (c) particle 171 convective transport and diffusion (d) particle concentration dependent viscosity of the suspension, and (e) evaporation related temperature effect. However, qualitatively we can understand the transport process using the argument that the solvent evaporated from the meniscus will be exactly compensated by the flow into it from the bulk: ∫ ∞ = z e f dz z j z j z h ) ( ) ( ) ( (4.1) where ) (z h is the thickness of the meniscus, ) (z j f is the flux of solvent into the meniscus, ) (z j e is the flux of solvent evaporation, and the width of the substrate is set to be unity. If ) (z j f , as proportional to the evaporation rate of the solvent, is comparable to the diffusion of particles, the particles will become concentrated in the meniscus. Obviously a critical evaporation rate exists for this concentration to occur. This is particularly important for nanometer size objects, given their faster diffusion compared to micrometer and larger size objects. Defining the leading edge of assembled array to be z=0, the number of particles per unit time transported to the assembled array is: ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ ⋅ ⋅ = ∫ ∞ 0 ) ( ) 0 ( dz z j c n e β (4.2) where ) 0 ( c is the number of particles per unit volume of solvent at the edge of the assembled array (z=0), which is concentrated with respect to the bulk. The factor 1 / 0 < = < s p v v β is introduced to describe the ratio of the mean velocity of particles 172 p v to that of solvent molecules s v , since the interaction between the particle and substrate will cause a friction to slow down the particle. The particles transported to the leading edge of the existing array will incorporate in a close-packed fashion because of two forces: (1) the push by the convective flow due to solvent evaporation and (2) the inter-particle forces. The push by the convective flow is proportional to the flux of the solvent ) 0 ( f j at the leading edge of the array. Many inter-particle forces come into play. Other than the usual van der Waals and electrostatic, the capillary force between the particles in certain conditions can play the dominant role as the driving force for close packing of particles and their ordering. Consider the situation shown in Figure 4.9, where two near-by particles are partially immersed in a solvent. The difference in slope of the solvent surface results in an attraction force between the particles which is referred to as immersive capillary force. According to the developed theory [4.28] the lateral attraction force can be approximated as () ( ) L r F c c x / 1 sin 2 2 2 Ψ ≈ πσ , if ( ) 2 / 1 ) /( g L r c ρ σ Δ << << (4.3) where σ is the surface tension of the liquid, c r is the radius of the three phase contact line at the particle surface, c Ψ is the mean meniscus slope angle at the contact line, L is the center-to-center distance between particles, g is the gravitational acceleration, and ρ Δ is the difference of the solvent mass density in liquid and gas phase. For a thorough analysis of immersive capillary force between spherical and cylindrical objects on substrates or floating see [4.28-30]. 173 α α Ψ C 2r C 0 L R I 0 X Z INNER REGION OUTER REGION Substrate Colloidal Particle α α Ψ C 2r C 0 L R I 0 X Z INNER REGION OUTER REGION Substrate Colloidal Particle Figure 4.9 Schematic of two spheres partially immersed in a liquid layer on a horizontal solid substrate. The deformation of the liquid meniscus gives rise to inter- particle capillary attraction. Given its approximate L / 1 distance dependence, the capillary force is more long ranged than the van der Waals force between two spheres [4.31-32]. As a numerical example, the capillary force between a pair of 3nm radius nanocrystals partially immersed in toluene even at 100nm away is on the order of 10pN. The total capillary binding energy of such nanocrystals closely packed ( R L 2 = , R: Radius of nanocrystal) due to the meniscus between them is on the order of 100kT at room temperature. Due to this large attractive capillary force, the new coming particles have to be incorporated into the existing array close packed array, thus maintaining the order of the array. Moreover every particle transported to the existing array will likely incorporate irreversibly since the capillary binding energy is much larger than the thermal energy. The rate of growth of the array is then is simply A n v growth ⋅ = , where n is the number of particles transported to the assembled array per unit time; A is the area of 174 a particle, setting the width of substrate to be unity. Hence considering (4.2), in a dip- coating experiment, the coverage of the resulting particle array on surface is: wd e wd wd growth v A dz z j c v A n v v / ) ( ) 0 ( / / 0 ⋅ ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ ⋅ ⋅ = ⋅ = = ∫ ∞ β θ (4.4) where wd v is the speed at which the substrate is withdrawn. The coverage is proportional to the concentration of the solution and inversely proportional to the substrate withdraw speed. The growth of the 2D array on the surface due to particle transport and incorporation described above has been visualized using optical microscopy by Dimitrov and coworkers [4.23] for μm size latex particles dip-coated on glass substrate. Although real time monitoring the dip-coating process for nanocrystals is difficult, the end result is successful and conforms to the expectation. • Ordered Arrays of CdSe and InAs/ZnSe NCQDs: In our experiments CdSe and InAs/ZnSe nanocrystals (TOP or TOPO surfactant covered) were dip-coated onto substrates (GaAs, glass etc.) from their non-polar solution. Figure 4.10 is an illustrative high resolution AFM image of CdSe/TOPO nanocrystals (5nm diameter CdSe, 617nm emission peak, from Evident Technology) dip-coated on GaAs substrate (CdSe 0.25mg/ml in toluene, withdrawal speed 2 μm/sec, room temperature in the glovebox). The dip-coating experiments should be and are performed in an environment of low humidity to avoid water droplet condensation on the toluene solution surface which will change its surface stability. As expected, the CdSe forms monolayer high closely packed array on the 175 substrate. The imperfection in the order of the array is mainly due to the size distribution of the CdSe nanocrystals. In figure 4.11 are shown the large area AFM images of CdSe nanocrystals dip-coated on GaAs at fixed withdrawal speed (2 μm/sec) but with three different CdSe solution concentrations: 0.25mg/ml, 0.125mg/ml, and 0.05mg/ml. The corresponding coverage θ is ~100%, 50%, 30%, which indeed scales linearly with the nanocrystal concentration, as expected from Equ. (4.4). AFM was also used to image different areas on such dip-coated samples (CdSe or InAs/ZnSe on GaAs or Glass, sample area 5mm by 10mm) to test the uniformity of coverage. The nanocrystal coverage derived from 5 μm by 5 μm images shows typically less than 5% variation on different places on the sample. Figure 4.10 AFM image of ordered, one-monolayer-high, 5nm diameter CdSe/TOPO nanocrystal array deposited on GaAs substrate using a dip-coating technique. 176 (a) (a) (b) (b) (c) (c) Figure 4.11 AFM images of 5nm diameter CdSe nanocrystals dip-coated on GaAs. Substrate drawn speed fixed at 2 μm/sec. Panels (a), (b), (c) correspond to nanocrystal toluene solution concentration 0.25mg/ml, 0.125mg/ml, 0.05mg/ml, respectively. The resulting nanocrystal coverages are, respectively, ~100%, 50%, 30% and scale linearly with the solution concentrations. 177 §4.2.2 Initiation of ordered colloidal particle array Now returning to the question of how the ordered array is seeded in the first place before it grows via the mechanism discussed in §4.2.1 above. There is no quantitative model on this matter yet. The intuitive understanding proposed by Denkov et al [4.22] is as follows. At the very front of the wetting meniscus, when the meniscus thickness gets down to several hundred nanometers, the van der Waals force becomes important to change the shape for the meniscus. Due to the presence of the substrate, van der Waals force between solvent molecules mutually repulsive (the notion of disjoining pressure as referred to in the general literature) [4.32] which spreads the solvent to form a flat film (h has only slight dependence on z see figure 4.8). The solvent evaporation at this flat film region causes suspension influx which carries particles into the film. For small particles, Brownian motion maybe another factor to cause them to enter this flat film region. Detailed analysis by Dimitrov et al [4.23] indicates that the thickness of this flat film region (h 0 ) may play an important role in determining how the particle will be assembled initially. When the thickness of the film h 0 (see figure 4.12) is slightly smaller than the diameter of the colloidal particles d, the long-ranged lateral capillary force between partially immersed particles as discussed above (see figure 4.9) may be able to gather the particle to initiate ordered array (figure 4.12). If h 0 >d, lateral capillary force due to partially immersion does not exist, and irregular patches of the particles may be formed by short ranged (with respect to capillary force) van der Waals and electrostatic interaction. If h 0 <<d, calculation shows that particles will have a hard time to enter 178 the flat film region due to the capillary force that pushes them back to the bulk of suspension [4.33]. d>h 0 h 0 j e j p Capillary Force Parallel film region Initiation of ordered array (a) (b) (c) ~ d>h 0 h 0 j e j p Capillary Force Parallel film region Initiation of ordered array (a) (b) (c) ~ Figure 4.12 Schematic showing the initiation of the nanocrystal array on the surface. (a) the particles enter the flat film region due to solvent convective flow or Brownian motion (b) if the particles are of diameter on the order of but less than the film thickness, the partially immersed particles will experience lateral attractive capillary force (c) Initial ordered array formation due to capillary force. To summarize, the conditions necessary to initiate and subsequently grow ordered 2D array of nanoparticles on a substrate are: (1) the particles must be able to slide on the substrate before the film completely dries. Therefore particles adsorbing strongly on surface, such as through chemical bonding, should be avoided. (Note: typically the time of drying of the solvent is not the limiting factor itself. Before solvent drying, the nanoparticle can diffusion on the order of 1 μm. See estimation [4.24]). (2) The evaporation rate of the solvent should be above a critical value so that the convective flow can compete with the diffusion of the particles to allow them to be concentrated in the meniscus. (3) The substrate must be well wettable by the solution and a wetting solvent film relatively flat and of thickness on the order of but 179 less than the diameter of the particles needs to be formed to allow the initial formation of an ordered seed. Although the above three conditions are difficult to quantify, however general trends can be summarized by surveying experimental results. Toluene dispersed hydrocarbon terminated nanocrystals (CdSe or InAs on the order of 10nm diameter) can be dispersed on GaAs and glass substrates to form ordered structures, as our result shown in figure 4.10 demonstrate. Ordered assembly of water dispersed micro- diameter particles on glass was reported [4.23]. Our experimental efforts to dip-coat water dispersed protein particles (ferritin 12nm in diameter) on glass substrate only led to random distribution of ferritin on the substrate. However ordered protein assembly is possible on mercury surface from aqueous suspension [4.35, 4.36]. This author believes that the above noted successful and unsuccessful cases of dip-coating to form ordered array is mainly a reflection of condition 3 above: the particle diameter in relationship with the thickness of flat wetting film which is determined by the combination of solvent and surface property. §4.2.3 Dip-coating summary and future development To conclude, the dip-coating approach is a simple and flexible method to assemble uniform monolayer of particles on solid substrates, which we will use as the sample preparation technique to construct the simple nanocrystal/substrate hybrid structures for excitation transfer studies to be discussed in chapter 5. Dip-coating is unique in that the transport of particles is directional toward the air-liquid-solid contact line, instead of being random all over the substrate. The capillary force 180 involved in ordering is long ranged and has a 1/L distance dependence, extending >100nm for nanometer-sized objects. Therefore large single 2D crystalline domain can be formed. The orientation of the crystal is also possible to be controlled, given the directional transport of particles. For future developments relating to dip-coating, it will be an interesting to study the dip-coating assembly of particles on lithographically pre-patterned substrates. Specifically, substrates can be patterned with shallow depth similar to the diameter of the particles, and intuitively, as far as capillary force is concerned, the edge of the undulation should serve like the leading edge of the pre-assembled array on which new particles will incorporate. Therefore it seems possible to form single crystalline 2D array in every valley, and the arrays should be oriented by the edge of the valleys. Hopefully in this fashion with the help patterning, dip-coating can served as a workhorse to assemble nano-objects in the smallest length hierarchy (a few hundred nanometers) toward the formation of more complex nanostructures. §4.3 SAM surface modification and SAM mediated adsorption of small objects (nanocrystals, proteins) Following the discussion on adsorption of nanocrystals on unmodified substrate using dip-coating technique, in this section we discuss the role and use of surface modification mediated adsorption of nanoparticles. One generic strategy to modify the surface properties of a solid is to functionalize it with a chemisorbed a layer of designed organic (including bioorganic) molecules as schematically shown in figure 4.13 [4.37-39]. Such molecules typically consist of a hydrocarbon chain 181 (often alkyl chain) which has at the substrate end a functional group to bind to the surface atoms and appropriate functional group on the free end to serve the desired purpose. Indeed such a functional group may be as simple as a single atom or a molecule that enables attachment of a QD. The chemisorbed molecules on the surface often assemble into a closely packed monolayer referred to as a self- assembled monolayer (SAM). As far as the adsorbate on the SAM functionalized surface is concerned, their adsorption in the zeroth order is only governed by the properties of the surface determined by the free end functional group. The SAM is a powerful tool for surface modification because of its great design flexibility: thickness and electronic property of SAM can be tuned with the appropriate hydrocarbon chain length and structure; the functional group at the substrate and the free end can be chosen according to the substrate material and the surface binding to be achieved. In the following section we break the discussion into two parts: (1) the chemistry on the “substrate end” -- the procedure and characterization of SAM on substrates of interest; and (2) the chemistry on the “free end” -- the SAM mediated the adsorption of small objects (nanocrystals, proteins). §4.3.1 SAM functionalization for Semiconductors: The typical functional group for SAM grafting used in our work is summarized in figure 4.13: Silane group for Si/SiO 2 (or glass) [4.37, 4.40, 4.41], Thiol group for GaAs [4.42, 4.43]. The conditions and procedures to achieve high quality SAM adsorption has become a relatively mature subject with literature of the 182 last 20 years [4.44, 4.45], which we adapt for our purpose. It is worth noting that our main interest is to achieve uniform SAM layer on surfaces to mediate subsequent adsorption. Therefore the SAM functionalization is relatively easy, since it is not critical to pursue strict quality control in many respects, such as pin-hole defect density and “gauche” conformational defect density in the SAM layer, the control of which is absolutely essential for many envisioned electric applications of the SAM (e.g. as dielectric layer, etching resists etc.) [4.43, 4.46, 4.47]. Si R Si R SiO 2 /Si substrate Si R Cl Cl Cl S R S R GaAs substrate S R Si R S R H (CH 2 ) n (CH 2 ) n o o (a) (b) Substrate End Free End Hydrocarbon Chain Si R Si R Si R Si R SiO 2 /Si substrate Si R Cl Cl Cl S R S R S R S R GaAs substrate S R S R Si R Si R S R H (CH 2 ) n (CH 2 ) n o o (a) (b) Substrate End Free End Hydrocarbon Chain CH 3 Si Cl Cl Cl (CH 2 ) 17 CH 3 Si Cl Cl Cl (CH 2 ) 17 Si Cl Cl Cl (CH 2 ) 17 Si Br Cl Cl Cl (CH 2 ) 3 Si Br Cl Cl Cl Si Br Cl Cl Cl (CH 2 ) 3 H S CH 3 (CH 2 ) 17 H S CH 3 (CH 2 ) 17 S CH 3 (CH 2 ) 17 S H (CH 2 ) 6 S H S H (CH 2 ) 6 S H OTS BPTS ODT HDT Figure 4.13: Schematic of generic type of self-assembled monolayer” (a) SiCl 3 (CH 2 ) n R SAM on SiO 2 /Si. (b) HS(CH 2 ) n R SAM on GaAs. Here R represents the functional group on the free end of the molecule which could be: CH 3 -; HO-; HS-, NH 3 -, HO(CH 2 CH 2 O) n CH 2 - ……; (c) Structure of the four SAM molecules involved in our discussion: octadecyl-tricholorosilane (OTS); 3-bromopropyl- tricholoro-silane (BPTS); 1-octadecanethiol (ODT); 1, 6-Hexanedithiol (HDT) (c) 183 §4.3.1(a) Alkylsiloxane SAM functionalization of silicon The procedure for silicon SAM functionalization using alkylsiloxane was modified from [4.40]. Here as an example we describe the procedure for octadecyltrichlorosiloxane (ODS) SAM functionalization using octadecyl- tricholorosilane (OTS) molecule (see figure 4.12(c) for its structure). Other alkylsiloxane SAM can be grafted similarly with varied alkylsilane solution concentration and reaction time. The silicon wafer was first degreased sequentially in TCE, acetone, methanol and DI water for 5mins in each solvent and then was blown dry by N 2 . The silicon wafer was then immersed into freshly prepared hot piranha solution (absolute H 2 SO 4 and 30% aqueous H 2 O 2 70:30 v/v) for 45mins. This procedure yields a native SiO 2 layer on silicon wafer with hydroxyl (-OH) group terminated surface. Surface densities of –OH up to 5x10 14 cm -2 have been reported [4.40]. The wafer was thoroughly washed with DI water and immersed in DI water for storage. The wafer was taken out and blown dry in a jet of N 2 and then taken into a dry box purged by N 2 (1% relative humidity is more than sufficient). The pretreated substrate (-OH terminated) is very hydrophilic with water contact angle close to 0 . The glassware used is rinsed with acetone and dried thoroughly in an oven (120 C) for a few hours to remove the adsorbed water. Due to the susceptibility of the silicon-chloride bond to hydrolysis, it was necessary to restrict water present during SAM functionalization. However, exceptional care in control of the moisture 184 level used in handling organometallic reagent is certainly not necessary. The as- prepared native Si/SiO 2 substrate is not anhydrous anyway; in fact one opinion in the literature is that trace amount of water in the reaction solution is necessary to form highly-ordered SAM layer with low defect density [4.44]. The protocol we used is as follows. Inside a dry box, we prepared 0.5mM octadecyltrichlorosilane (OTS, Aldrich) solution in anhydrous toluene (Aldrich 99.8%). We immersed the pretreated silicon wafer in the OTS solution for 3 minutes at room temperature, then took it out and washed and immersed in anhydrous toluene. The prepared the ODS functionalized silicon was then taken out of the dry box. To remove residual physisorbed OTS, the sample was rinsed with ethanol (and then DI water in some cases) and blown dry with N 2 . After ODS SAM functionalization the surface property change is obvious. The CH 3 terminated silicon surface becomes very hydrophobic, the water contact angle observed by eye is certainly larger than 90 . An AFM (tapping mode, RTESP tip, DI multimode AFM) image of ODS functionalized SiO 2 /Si surface is shown in figure 4.14(a). The rms roughness of the functionalized surface is 0.16nm, the same as for bare starting SiO 2 /Si substrate (0.16nm rms roughness). This illustrates the uniformity of the SAM adsorption. Figure 4.14(b) shows an AFM image of 3- bromopropyl-tricholoro-siloxane (BPTS) SAM functionalized SiO 2 /Si surface. 185 (a) ODS/SiO 2/ Si (b) BPTS/SiO 2 /Si RMS roughness: 0.16nm RMS roughness: 0.24nm (a) ODS/SiO 2/ Si (b) BPTS/SiO 2 /Si (a) ODS/SiO 2/ Si (b) BPTS/SiO 2 /Si RMS roughness: 0.16nm RMS roughness: 0.24nm Figure 4.14 AFM image of (a) octadecyl-tricholorosiloxane (ODS) SAM functionalized Si/SiO 2 surface. (RMS roughness: 0.16nm) (b) 3-bromopropyl- tricholoro-siloxane (BPTS) SAM functionalized Si/SiO 2 surface. (RMS roughness: 0.24nm); §4.3.1(b) Alkylthiol SAM functionalization of GaAs (001) The procedure for GaAs (001) surface functionalization using alkylthiol SAM was modified from [4.48]. Here we describe the procedure for octadecylthiol (ODT) functionalization (see figure 4.13(c) for its structure). Other alkylthiol SAMs can be grafted similarly with varied alkylthiol solution concentration and reaction time. The GaAs (001) substrate was first degreased sequentially in (TCE), acetone, methanol and DI water for 5mins in each solvent and then was blown dry with N 2 . The degreased GaAs was etched in concentrated HCl (37%) to strip off its original contaminated native oxide. Etching was stopped by rinsing GaAs with DI water and thoroughly blown dry the GaAs in N 2 for a few minutes. This pretreated GaAs substrate is transferred into a dry box (1% humidity is more than sufficient.) 186 5mM ODT ethanol solution was prepared in the dry box by dissolving 0.143g ODT (Aldrich) in 100ml anhydrous ethonal, 3ml of 30% aqueous NH 4 OH was added into the solution. (The optimal concentration of oxide remover NH 4 OH was systematically studied in [4.48] ). The ODT solution is then purged with dry N 2 for a couple of hours to remove dissolved oxygen. Inside the dry box, the GaAs substrate is sealed in a 15ml air-tight vial filled with the ODT solution. The ODT functionalization was done for 8 hours at 50 C by heating the vial in a water bath. After finishing, the functionalized GaAs was taken out and rinsed in chlorobenzene, ethanol, and DI water to wash off physisorbed organic molecules and was dried in N 2 . The effect of ODT functionalization is obvious that the CH 3 terminated GaAs surface becomes very hydrophobic, the water contact angle observed by eye is certainly larger than 90 . An AFM (tapping mode, RTESP tip, DI multimode AFM) image of ODT functionalized GaAs surface is shown in figure 4.15(a).The functionalized surface remains flat (rms roughness 0.28nm, compared to 0.20nm before SAM func- (a) ODT/GaAs (b) HDT/GaAs RMS roughness: 0.28nm RMS roughness: 0.41nm (a) ODT/GaAs (b) HDT/GaAs (a) ODT/GaAs (b) HDT/GaAs RMS roughness: 0.28nm RMS roughness: 0.41nm Figure 4.15 AFM image of (a) 1-octadecanethiol (ODT) (b) 1, 6-hexanedithiol (HDT) SAM functionalized GaAs(001) surface. The RMS roughness for the two SAM functionalized surface are respectively: 0.28nm, 0.41nm. 187 tionalization), which illustrates the uniformity of the ODT SAM adsorption. Figure 4.14(b) shows an AFM image of 1,6-hexanedithiol (HDT) modified GaAs (001) which is discussed further in the next subsection. §4.3.2 SAM mediated nanocrystal adsorption Having discussed the chemistry at the substrate end of the SAM and the procedure for SAM grafting on Si/SiO 2 and GaAs surfaces, below we turn to the chemistry of the “free” end of the SAM and the SAM mediated adsorption. The ODS and ODT SAMs discussed above both have the methyl group (-CH 3 ) as the free end. Such SAM modifications result in a chemically inert and highly hydrophobic surface. Although the adsorption properties of objects can be tuned by the hydrophobicity/hydrophilicity of the surface, a more direct method to enhance the adsorption of desired objects is to functionalize the free end of the SAM with an appropriate chemical group that covalently bind to the objects to be adsorbed. In this subsection, we restrict the discussion to the covalent binding of nanocrystal quantum dots to the substrate via SAM bilinkers to construct hybrid colloidal/epitaxial structure. In next chapter (chapter 5), we generalize the concept of SAM to that made of biological molecules and discuss SAM directed adsorption of biological entities, such as proteins and live cells. The binding between thiol group (-SH) and the surface of common nanocrystals (CdSe, InAs, Au) is a well-known chemistry [4.49, 4.50]. Therefore, again as a prototypical example, we used 1,6-hexanedithiol (HDT, Aldrich, see figure 188 4.13(c) for its structure) SAM with the thiol at its free end to modify the GaAs (001) surface to bind to nanocrystals. The procedure for HDT functionalization is the same as for ODT discussed above except that the GaAs substrate is immersed in 0.65mM HDT isopropanol solution (10% v/v aqueous NH 4 OH added) for 4 hours. An AFM (tapping mode, RTESP tip, DI multimode AFM) image of the produced HDT/GaAs surface is shown in figure 4.15(b). The surface roughness (RMS 0.41nm) is notably increased with respect to that of bare GaAs (RMS 0.20nm). Blobby features ~10nm wide and 2.5nm high are noticeable. These features are likely a reflection of the polymerization of the HDT, in which thiols of different HDT molecules react to form disulfide bonds. Nanocrystal adsorption on such HDT modified GaAs surface was performed using CdSe/TOPO (5nm diameter, 617nm emission peak, Evident Technology) and InAs/TOP nanocrystals (4nm diameter, 1080nm emission peak, see chapter 2) synthesized by the author. The HDT/GaAs substrate (along with the control, bare GaAs) was immersed into 1.7mg/ml CdSe or 1.2mg/ml InAs toluene solution for 48 hours at room temperature. Then the samples are thoroughly washed in anhydrous toluene. Excessive toluene was wicked off with filter paper. Samples are then left to dry in glovebox. In figure 4.16 (left) are shown AFM (tapping mode, RTESP tip, DI multimode AFM) images of CdSe (a) and InAs (b) NCQDs adsorbed on HDT modified GaAs. From the cross-section analysis of these AFM images (figure 4.16 right), we notice that the height of the features on the AFM images match the diameter of the nanocrystals, which indicates that the adsorbed nanocrystal layer is 189 of one monolayer high. Note the distribution of these adsorbed nanocrystals is non- ordered on the surface in contrast to the close-packed distribution resulting from the dip-coating process (figure 4.10). We note the corresponding the AFM result for control sample (bare GaAs) shows no adsorbed nanocrystals after incubation in the same NC solution and toluene washing. The resistance of nanocrystals adsorbed on HDT modified GaAs to thorough toluene washing suggests that the nanocrystals were indeed chemically bound to the thiol group of the modified GaAs surface. Figure 4.16: (Left) AFM images of (a) CdSe (b) InAs NCQD adsorbed on HDT modified GaAs surface (Right) Cross-section analysis of the line as indicated in the corresponding AFM images. (b) InAs NC on Linked to HDT modified GaAs Line (a) CdSe NC on Linked to HDT modified GaAs Line 190 Hybrid NCQD/Semiconductor substrate Structures The above adsorption process resulted in a hybrid colloidal nanocrystal on crystalline semiconductor substrate integrated structure as schematically shown in figure 4.17. Here the colloidal QDs are linked to the semiconductor surface by SAM molecules. Such semiconductor substrates routinely arise in epitaxially buried semiconductor nanostructures, such as the so-called self-assembled quantum dots, quantum wires, and quantum wells, which are the backbone of advancing optoelectronic technology. Such structures are discussed in Chapter 5. Shown in figure 4.18 is the preliminary PL measurement from the CdSe linked to GaAs via HDT. The covalent binding of the nanocrystals did not entirely quench their PL. Similar nanocrystal/SAM/substrate hybrid structures synthesis and PL study was also reported by Marx et al [4.51]. For us, such hybrid structures have not been the main platform on which excitation transfer between the colloidal QD and epitaxial nanostructures has been studied in the past three years (see chapter 5). Nevertheless, these promotes future system to pursue for excitation transfer studies since one has the freedom of tuning the interaction between the colloidal QD and the epitaxial nanostructure by controlling the energy levels in the linker SAM molecules and their alignment with the states in the QD and epitaxial structure. Therefore vast richness in research along this line can be expected. 191 Substrate SAM Molecule Surfactant NCQD NCQD Substrate SAM Molecule Surfactant NCQD NCQD Figure 4.17 Schematic of NCQD/SAM/Substrate hybrid structure. Figure 4.18 PL of CdSe NC on Linked to HDT modified GaAs. Corresponding AFM image of the hybrid structure is shown in figure 4.16(a). 192 Chapter 4 References: [4.1] C. A. Colinge, J-P Colinge, Physics of Semiconductor Devices, Springer- Verlag, (2002). [4.2] D. E. Aspnes, “Recombination at semiconductor surfaces and interfaces”, Surf. Sci. 132, 406-421 (1983). [4.3] S. Guha, A. Madhukar, K. C. Rajkumar, ”Onset of incoherency and defect introduction in the initial stages of molecular-beam epitaxial-growth of highly strained InxGa1-xAs on GaAs (001)”, Appl. Phys. Lett. 57, 2110-2112 (1990). [4.4] A. Madhukar, Ch.2 in Nano-Optoelectronics, Ed. M. Grundmann, Springer- Verlag, Berlin, (2002) [4.5] E-T. Kim, Z. Chen, and A. Madhukar, “Selective manipulation of InAs quantum dot electronic states using a lateral potential confinement layer”, Appl. Phys. Lett. 81, 3473-3475 (2002). [4.6] V. Emiliani, B. Bonanni, C. Presilla, M. Capizzi, A.Frova, Y. L. Chang, I. H. Tan, J. L. Merz, M. Colocci, M. Gruioli, “Interaction mechanism of near-surface quantum well with oxidized and H-passivated AlGaAs surfaces”, J. Appl. Phys. 75, 5114-5122 (1994). [4.7] J. Dreybrodt, F. Daiminger, J. P. Reithmaier, and A. Forchel, “Dynamics of carrier-capture processes in Ga x In 1-x As/GaAs near-surface quantum wells”, Phys. Rev. B 51, 4657-4660 (1995). [4.8] M. A. Makeev, W. Yu, A. Madhukar, “Stress distributions and energetics in the laterally ordered systems of buried pyramidal Ge/Si (001) islands: An atomistic simulation study”, Phys. Rev. B 68 195301 (2003). [4.9] R. S. Besser, C. R. Helms, “Comparison of surface properties of sodium sulfide and ammonium sulfide passivation of GaAs”, J. Appl. Phys. 65, 4306-4310 (1989). [4.10] S. R. Lunt, G. N. Fiyba, P. G. Santangelo, N. S. Lewis, “Chemical studies of the passivation of GaAs surface recombination using surfides and thiols”, J. Appl. Phys. 70, 7449-7465 (1991). [4.11] H. Oigawa, J. F. Fan, Y. Nannichi, H. Hugahara, M. Oshima, “Universal passivation effect of (NH4)2Sx treatment of the surface of III-V compound semiconductors”, J. J. Appl. Phys. Part 2-Letters 30, L322-325 (1991). [4.12] M. Shimomura, T. Sawadaishi, “Bottom-up strategy of materials fabrication: a new trend in nanotechnology of soft materials”, Curr. Opin. Colloid. Interf. Sci. 6, 11-16 (2001). 193 [4.13] D. Pum, A. Neubauer, E. Gyorvary, M. Sara, and U. B. Sleytr, “S-layer proteins as basic building blocks in a biomolecular construction kit”, Nanotechnology 11, 100-107 (2000). [4.14] P. J. B. Koeck, H. K. Kagawa, M. J. Ellis, H. Hebert, and J. D. Trent, “Two- dimensional crystals of reconstituted beta-subunits of the chaperonin TF55 from Sulfolobus shibatae”, Biochimica et Biophysica Acta 1429, 40-44 (1998). [4.15] B. O. Dabbousi, C. B. Murray, M. F. Rubner, M. G. Bawendi, “Langmuir- Blodgett manipulation of size-selected CdSe nanocrystallites”, Chem. Mater. 6, 216- 219 (1994). [4.16] J. R. Heath, C. M. Knober, D. V. Leff, “Pressure/temperature phase diagrams and superlattices of organically functionalized metal nanocrystal monolayers: The influence of particle size, size distribution, and surface passivant”, J. Phys. Chem. B 101, 189-197 (1997). [4.17] S. H. Park, Y. N. Xia, “Assembly of mesoscale particles over large areas and its application in fabricating tunable optical filters”, Langmuir 15, 266-273 (1999). [4.18] S. H. Park, D. Qin, Y. Xia, “Crystallization of mesoscale particles over large areas”, Adv. Mater. 10, 1028 (1998). [4.19] M. Geirsig, P. Mulvaney, “Formation of ordered 2-dimensional gold colloidal lattices by electrophoretic deposition”, J. Phys. Chem. 97, 6334-6336 (1993). [4.20] M. Trau, D. A. Saville, I. A. Aksay, “Field-induced layering of colloidal crystals”, Science 272, 706-709 (1996). [4.21] S. R. Yeh, M. Seul, B. I. Shraiman, “Assembly of ordered colloidal aggregates by electric-field-induced fluid flow”, Nature 386, 57-59 (1997). [4.22] N. D. Denkov, O. D. Velev, P. A. Kralchevsky, I. B. Ivanov, H. Yoshimura, K. Nagayama, “Mechanism of formation of 2-Dimensional crystals from latex-particles on substrates”, Langmuir 8, 3183-3190 (1992). [4.23] A. S. Dimitrov, K. Nagayama, “Continuous convective assembling of fine particles into two-dimensional arrays on solid surfaces”, Langmuir, 12, 1303-1311 (1996). [4.24] C. Stowell, B. A. Korgel, “Self-assembled honeycomb networks of gold nanocrystals”, Nano Lett. 1, 595-600 (2001). [4.25] E. Rabani, D. R. Reichman, PL Geissler, L. E. Brus, “Drying-mediated self- assembly of nanoparticles”, Nature 426, 271-274 (2003). 194 [4.26] J. Perrin. Ann. Chim. Phys. 18, 1 (1909). [4.27] S. Narayanan, J. Wang and X. M. Lin, “Dynamical self-assembly of nanocrystal superlattices during colloidal droplet evaporation by in situ small angle x-ray scattering” Phys. Rev. Lett. 93, 135503 (2004). [4.28] P. A. Kralchevsky, V. N. Paunov, I. B. Ivanov, K. Nagayama, “Capillary meniscus interaction between colloidal particles attached to a liquid-fluid interface”, J. Colloid Interface Sci., 151, 79-94 (1992). [4.29] P. A. Kralchevsky, V. N. Paunov, N. D. Denkov, I. B. Ivanov, and K. Nagayama, “Energetical and force approach to the capillary interactions between particles attached to a liquid-fluid interface”, J. Colloid Interface Sci., 155, 420-437 (1993). [4.29] P. A. Kralchevsky, V. N. Paunov, N. D. Denkov, I. B. Ivanov, and K. Nagayama, “Energetical and force approach to the capillary interactions between particles attached to a liquid-fluid interface”, J. Colloid Interface Sci., 155, 420-437 (1993). [4.30] V. N. Paunov, P. A. Kralchevsky, N. D. Denkov, K. Nagayama, “Lateral capillary forces between floating submillimeter particles”, J. Colloid Interface Sci., 157, 100-112 (1993). [4.31] The non-retarded Van der Waals force between two identical spherical objects has the functional form 2 D R F vdw ∝ (when D<<R) and 7 6 D R F vdw ∝ (when D>>R), where R is the object radius, D is the surface distance between the two spherical objects, See Ref 4.32 for detail. [4.32] J. Israelachvili, Intermolecular & surface forces, Academic Press (1992). [4.33] P. A. Kralchevsky, K. Nagayama “Capillary forces between colloidal particles”, Langmuir 10, 23-36 (1994). [4.34] For example, the toluene evaporation rate is measured to be 170nm/sec under room temperature. So even disregarding compensation from bulk and the slowing of evaporation due to interaction with surface and particles, 6nm thick toluene taken 35ms to dry, during which time 6nm diameter particle can diffuse 1.5um, assume no friction between particle and surface. [4.35] H. Yoshimura, S. Endo, M. Matsumoto, K. Nagayama, Y. Kagawa, “Hexagnal Structure of two-dimensional crystals of the alpha-3-beta-3 complex of thermophilic ATP synthase”, J. Biochem., 106, 958-960 (1989). 195 [4.36] H. Yoshimura, S. Endo, M. Matsumoto, K. Nagayama, “2-dimensional crystallization of proteins on mercury” Ultramicroscopy, 32, 265-274 (1990). [4.37] J. Sagiv, “Organized monolayers by adsorption. 1. Formation and structure of oleophobic mixed monolayers on solid-surfaces.”, J. Am. Chem. Soc. 102, 92-98 (1980). [4.38] C. D. Bain, E. B. Troughton, Y. T. Tao, J. Evall, G. M. Whitesides, R. G. Nuzzo, “Formation of monolayer films by the spontaneous assembly of organic thiol from solution onto gold”, J. Am. Chem. Soc. 111, 321-335 (1989). [4.39] A. Ulman, An Introduction to Ultrathin Organic Films: From Langmuir-- Blodgett to Self—Assembly, Academic Press, San Diego (1991). [4.40] S. R. Wasserman, Y. T. Tao, G. M. Whitesides, “Structure and reactivity of alkylsiloxane monolayers formed by reation of alkyltrichlorosilanes on silicon substrates”, Langmuir 5, 1074-1087 (1989). [4.41] S. R. Wasserman, G. M. Whitsides, I. M. Tidswell, B. M. Ocko, P. S. Pershan, J. D. Axe, “The structure of self-assembled monolayers of alkylsiloxanes on silicon – A comparison of results from elliposmetry and low angle X-ray reflectivity”, J. Am. Chem. Soc. 111, 5852-5861 (1989). [4.42] O. S. Nakagawa, S. Ashok, C. W. Sheen, J. Martensson, D. L. Allara, ”GaAs interfaces with octadecyl thiol self-assembled monolayer – structural and electrical- properites”, J. J. Appl. Phys. Part 1 30, 3759-3762 (1991). [4.43] M. J. Lercel, R. C. Tiberio, P. F. Chapman, H. G. Graighead, C. W. Sheen, A. N. Parikh, D. L. Allara, “Self-assembled monolayer electron beam resists on GaAs and SiO 2 ” J. Vac. Sci. Tech. B 11 2823-2828 (1993). [4.44] J. B. Brzoska, I. B. Azuoz, F. Rondelez, “Silanization of solid substrates: A step toward reproducibility”, Langmuir 10, 4367-4373 (1994). [4.45] T. Vallant, H. Brunner, U. Mayer, H. Hoffmann, T. Leitner, R. Resch, G. Friedbacher, “Formation of self-assembled octadecylsiloxane monolayer on mica and silicon surfaces studied by atomic force microscopy and infrared spectroscopy“, J. Phys. Chem. B. 102, 7190-7197 (1998). [4.46] C. Boulas, J. V . Davidovits, F. Rondelez, D. Vuillaume, Suppression of Charge Carrier Tunneling through Organic Self-Assembled Monolayers”, Phys. Rev. Lett. 76, 4797-4800 (1996). [4.47] J. L. Wilbur, G. M. Whitesides, “Self-assembly and self-assembled monolayers in micro- and nanofabrication” Chapter 8 in Nanotechnology, Ed. G. Timp, Springer-Verlag, New York (1999). 196 [4.48] T. Baum, S. Ye, K. Uosaki, “Formation of Self-Assembled Monolayers of Alkanethiols on GaAs Surface with in Situ Surface Activation by Ammonium Hydroxide”, Langmuir, 15, 8577-8579, (1999). [4.49] V. L. Colvin, A. N. Goldstein, A. P. Alivisatos, “ Semiconductor nanocrystals covalently bound to metal surfaces with self-assembled monolayers”, J. Am. Chem. Soc. 114, 5221-5230 (1992). [4.50] R. Resch, C. Baur, A. Bugacov, B. E. Koel, P. M. Echternach, A. Madhukar, N. Montoya, A. A. G. Requicha, P. Will, “Linking and manipulation of gold multinanoparticle structures using dithiols and scanning force microscopy,” J. Phys. Chem. B 103, 3647-3650 (1999). [4.51] E. Marx, D. S. Ginger, K. Walzer, K. Stokbro, N. C. Greenham, “Self- assembled monolayer of CdSe nanocrystals on doped GaAs substurate“, Nano Lett. 2, 911-914 (2002). 197 Chapter 5. Bioconjugated SAM mediated cell adhesion on solid surfaces via specific ligand-receptor interaction Following last chapter’s discussion on semiconductor surface modification using hydrocarbon based SAM and SAM mediated nanocrystals adsorption, in this chapter we expand the application of SAM based surface modification to control the adsorption of biological entities. We first discuss non-specific and specific protein adsorption on semiconductor surface mediated by conventional hydrocarbon bases SAM induced. Then we introduce a more complicated version of SAMs with peptide ligands as their free ends which we dubbed Bioconjugated-SAM (BSAM), and discuss BSAM mediated cell adhesion via specific ligand-receptor interaction. As discussed in chapter 1, the ability to tune (enhance or reject) the adhesion of specific type of targeted cells using appropriate peptide ligand based SAM modified solid surface is an important aspect of for biological detection (figure 5.1). In addition, the tuning of cell adhesion behavior via surface modification is also demanded by another vast field of biomedical engineering -- surface coating of prosthetic implants to create a harmonious biotic-abiotic interface. In the context of the designing coating for cortical (brain) prosthetic device, we investigated peptide ligand based BSAM modification of ceramic (modified glass) substrates and its application to mediate hippocampal neuronal cell adhesion, as is reported in this chapter. However, we anticipate this experience on neuronal cell adhesion can also be applied to the problem of biological detection, in which appropriate surface modification will be designed to enhance diseased cells and repel normal cells. 198 Figure 5.1 A scheme for biological sensing involving (a) selective adhesion of target cells on surface functionalized by peptide ligand based SAM, and (b) subsequent QD labeling for detection. §5.1 SAM mediated protein adsorption The ODS and ODT SAMs discussed above both have the methyl group (-CH 3 ) as the free end. Such SAM modifications result in a chemically inert and highly hydrophobic surface. Historically, a vast literature has arisen on the SAMs based modification of the surface hydrophobicity/hydrophilicity and its effect on subsequent adsorption of proteins [5.1, 5.2]. A hydrophobic surface resulted from modification using the methyl terminated SAM, in general, is favorable for enhanced protein adsorption [5.1]. A simple illustration of protein adsorption on hydrophobic SAM modified surface is shown in figure 5.2, in which ferritin protein is adsorbed on ODS modified Si/SiO 2 surface. The procedure for ODS modification of Si/SiO 2 surface was described in chapter 4. (An illustrative AFM image of ODS modified 199 Si/SiO 2 surface was shown in figure 4.14.) The ferritin adsorption is performed by incubating ODS modified Si/SiO 2 substrate in 1mg/ml ferritin solution (from horse spleen, cationized with DPMA, in 0.015M NaCl, Simga) for 5 minutes. Then the substrate is rinsed thoroughly with DI water. The resulted sample (ODS adsorbed on Si/SiO 2 /ODS surface) was imaged using tapping mode AFM in DI water (NPS tip, DI multimode AFM). The round shaped features in the AFM image (figure 5.2) correspond to individual ferritin proteins (12nm in diameter). The high coverage (close to one monolayer) of the adsorbed ferritin shows high affinity between the ferritin and the hydrophobic ODS SAM modified surface. Figure 5.2 AFM image of ferritin protein adsorbed on ODS modified Si/SiO 2 surface. Other than the non-specific protein adsorption on hydrophobic SAM modified surface, we are interested in a more direct way to enhance the adsorption – via specific chemical binding between the SAM free end and the protein. As a 200 prototypical example, we chose 3-bromopropyl-tricholoro-silane (BPTS, see figure 4.13(c) for its structure) with Bromine as its free end to functionalize SiO 2 /Si surface. The procedure for BPTS functionalization is the same as for ODS except substrate is immersed in 1mM BPTS (Aldrich) toluene solution for 10mins. AFM of produced BPTS/SiO 2 /Si surface is shown in figure 5.3 which is of reasonable smoothness (RMS roughness 0.24nm, compared to 0.16nm roughness of the starting SiO2/Si substrate). The bromine end of the BPTS, via thiol-ether reaction [5.3], is designed to link to the cysteine amino acid exposed on the outer surface of a genetically engineered chaperonin protein (HSP60 from hyperthermophilic acidophilic archeon Sulfolobus Shibatae, 60kDa, ~17nm diameter, ~17 height) provided by J. Trent group at NASA-AMES [5.4, 5.5]. BPTS/SiO 2 /Si RMS roughness: 0.24nm BPTS/SiO 2 /Si RMS roughness: 0.24nm Figure 5.3 AFM image of 3-bromopropyl-tricholoro-siloxane (BPTS) SAM functionalized Si/SiO 2 surface. (RMS roughness: 0.24nm); 201 Chaperonin adsorption was performed as follows. BPTS modified Si(001)/SiO 2 substrate and the control (bare Si/SiO 2 substrate) was incubated in 0.075mg/ml chaperonin aqueous solution (buffer made of 25mM HEPES + 4mM ATP + 10mM MgCl 2 + 25mM NaN 3 ) for 45 minutes followed by thorough washing with the buffer. Then the resulting samples were imaged in tapping mode AFM (NPS tip, DI multimode AFM) in the aqueous buffer. Shown in figure 5.4 are the corresponding AFM images of chaperonin incubated on bare Si/SiO 2 (a) and BPTS modified Si (b). No adsorbed chaperonin is observed on the bare Si control. On the BPTS modified Si substrate, dramatically enhanced adsorption of chaperonin is observed. The resistance of adsorbed chaperonin to thorough washing indicates that such adsorption involves chemical bonding between the chaperonin and the BPTS modified Si substrate. (a) (b) (a) (b) Figure 5.4: AFM images show mutant Chaperonin adsorption on (a) bare Si surface (b) BPTS modified Si. Inset: higher resolution AFM image shows that disc-shaped ~30nm diameter features could be few-mers formed by aggregation of 3 or 4 donut- shaped chaperonins. 202 §5.2 Bioconjugated SAM (BSAM) mediated Cell adhesion. SAM as a general strategy for surface modification is not limited either to simple hydrocarbon based organic molecules, or to modifying the adsorption of small abiotic objects such as QDs and proteins as discussed in chapter 4 and the previous section. The generalized concept of SAM can also be applied to control the adsorption of complex biotic objects such as a living cell by creating the appropriate moiety at the free end of a SAM (adsorbed on the solid substrate of interest) to bind it to a cell. The trick is again to choose the appropriate free end of the SAM. Since the free end groups define the surface the cell senses and responds to, it allows the interaction between the cell and the substrate to be controlled. In the case of cell adhesion, the free end is more complicated than a simple chemical group, typically a small peptide or a protein. In our proposed scheme of biological detection using hybrid colloidal / epitaxial structures, such surface modification can obviously play an essential role -- to enhance adhesion of diseased target cells on specifically surface functionalized regions of substrate and keep the cell-substrate in close distance for detection. Further, such surface modification in general impacts biomedical engineering in the emerging field of implantable prosthetic devices such as those used to restore brain or eye function via electrical stimulation [5.6-5.8], In these applications an appropriate biotic-abiotic interface is essential to allow optimal function of the implant over extended period of time and not to damage the surrounding tissue or cells. 203 One type of prosthesis of interest to us is the cortical prosthesis that replaces brain function lost due to damage to the CA3 (CA= comu ammonis) subregion of the hippocampus of the brain [5.6, 5.9]. Such implants involve biomimetic electronics chip to mimic the function of the CA3 which receive inputs from and provide outputs to the undamaged regions adjacent to the CA3 through deep penetrating multi-electrode arrays placed according to the local cortical architecture. The multi- electrode array is schematically shown in figure 5.5 consisting the metal electrode and nonconducting ceramic substrate region. For the most effective neuron stimulation, it is thought desirable to preferentially adhere neurons in close contact to the metallic electrodes and to adhere the astrocytes (which are the supporting neuronal cells) on the ceramic substrate regions between where the neurons are adhered [5.6]. Motivated by the requirement of the two different yet connected applications noted above (biological detection using hybrid structure and implantable prosthesis), we explored the surface modification for cell adhesion aiming at (1) close cell/surface proximity (2) spatial-selectivity, and (3) cell-type-selectivity. One good way to achieve these aims is for the surface modification to mimic the biochemical environment which the specific cell type will recognize as natural – this means the extracellular matrix (ECM). The way the cells bind to and respond to the ECM is via the interaction between cell membrane receptors (mainly integrins) and the corresponding ligands in the ECM. With the recent progress, many short ligand peptides from the ECM proteins or cell membrane proteins have been identified as 204 the minimum sequence needed to serve as the ligand to bind to the particular receptors on different types of cells [5.10-5.12]. One well known peptide ligand is the RGD (Arg-Gly-Asp) from ECM protein fibronectin [5.11]. Following these discoveries, one recently proposed biomimetic strategy for surface modification is to coat the surface with peptide ligand for receptors of specific cell type [5.13-5.16]. Note this strategy is in contrast to the traditional coating for non-specific cell adhesion, mostly through electrostatic interactions, that is obtained through coating the abiotic surface with proteins such as the very often employed poly-D-lysine (PDL) and ECM proteins such as laminin [5.17]. This biomimetic strategy based on ligand-receptor specific interaction is obviously advantageous to meet the aim of intimate cell-surface contact and spatially-selective adhesion of selective cell type. Figure 5.5 Schematic of conducting electrodes on nonconducting substrates employed in prosthesis for electrical stimulation of cells and tissue. 205 We undertook systematic studies to examine the effectiveness of this ligand- receptor based surface modification to achieve enhanced selective neuronal cell adhesion in the context of cortical prosthesis [5.16]. The experience can be applied to the problem of biological detection, if the specific receptors on diseased cells and the corresponding ligands are known. To do so, a generalized version of SAM which we dub as Bioconjugated SAM (BSAM) was used for substrate surface modification. The BSAM involves a hydrocarbon based conventional SAM that is grafted to the solid surface, but on the free end is conjugated to the peptide ligands targeting the receptors of cells, neurons and astrocytes in this specific application. The BSAM functionalized surface and cell adhesion is schematically shown in figure 5.6. In our work the BSAM surface conjugation was broken into two steps: (i) Covalent adsorption of an appropriate hydrocarbon based SAM with exposed functional groups (carboxyl or amine) that would allow for conjugation of the ligand peptide. (ii) Covalent attachment of peptide ligand onto the SAM, thus creating a BSAM with peptide ligand as the free end for subsequent cell adhesion. Specifically we have employed the peptide IKVAV (see figure 5.7(b) for its structure) which has been identified as one of the neuron specific peptide ligands within the extracellular protein laminin [5.10]; likewise, we have employed the astrocyte receptor specific peptide ligand KHIFSDDSSE (see figure 5.7(b) for its structure) [5.12]. Below we present the a study of BSAM functionalization of 206 glass(SiO 2 ) substrates and surface characterization, followed by the results of rat hippocampal neuron adhesion studies on such well characterized surfaces. Figure 5.6. (a) Schematic of cell binding to the substrate via specific membrane receptor interaction with corresponding peptide ligands coated on the substrate. Note the undulation in the substrate surface is to depict a lateral scale (~200nm) of typical variation in the extra-cellular matrix topology and is on a length scale more than an order of magnitude smaller than the size of the cell (b) Magnified view of a laterally nanoscale (< 20nm) region to reveal significance of the surface vertical position (called surface roughness) in relation to the employed bilinker length. 207 Step 1: SAM attachment Glass OH OH OH Amino-phase glass OO O NH Si Si Si NH NH Amino-phase glass OO O Glass OO O OO O 2 22 O O + APTS Si OEt NH 2 OEt EtO Step 2: CAM attachment Amino phase glass Amino phase glass K K K COO - - Si NH 3 + Si OO O CO K CO K CO OO O K Amino phase glass Amino phase glass Glass Peptide (KHIFSDDSSE) O O Si + Si NH Si NH O O Si NH Glass COO - COO - NH 3 + NH 3 + Amino phase glass Amino phase glass I I I COO - - Si NH 3 + Si OO O CO I CO I CO OO O I Amino phase glass Amino phase glass Glass Peptide (IKVAV) O O Si + Si NH Si NH O O Si NH Glass COO - COO - NH 3 + NH 3 + (a) Step 1: SAM attachment Glass OH OH OH Amino-phase glass OO O NH Si Si Si NH NH Amino-phase glass OO O Glass OO O OO O 2 22 O O + APTS Si OEt NH 2 OEt EtO Step 1: SAM attachment Glass OH OH OH Amino-phase glass OO O NH Si Si Si NH NH Amino-phase glass OO O Glass OO O OO O 2 22 O O + APTS Si OEt NH 2 NH 2 OEt EtO Step 2: CAM attachment Amino phase glass Amino phase glass K K K COO - - Si NH 3 + Si OO O CO K CO K CO OO O K Amino phase glass Amino phase glass Glass Peptide (KHIFSDDSSE) O O Si + Si NH Si NH O O Si NH Glass COO - COO - NH 3 + NH 3 + Amino phase glass Amino phase glass K K K COO - - Si NH 3 + Si OO O CO K CO K CO OO O K Amino phase glass Amino phase glass Glass Peptide (KHIFSDDSSE) O O Si + Si NH Si NH O O Si NH Glass COO - COO - NH 3 + NH 3 + Amino phase glass Amino phase glass I I I COO - - Si NH 3 + Si OO O CO I CO I CO OO O I Amino phase glass Amino phase glass Glass Peptide (IKVAV) O O Si + Si NH Si NH O O Si NH Glass COO - COO - NH 3 + NH 3 + Amino phase glass Amino phase glass I I I COO - - Si NH 3 + Si OO O CO I CO I CO OO O I Amino phase glass Amino phase glass Glass Peptide (IKVAV) O O Si + Si NH Si NH O O Si NH Glass COO - COO - NH 3 + NH 3 + (a) NH 2 CH C CH 2 NH O CH 2 H 2 C H 2 C NH 2 CH C CH 2 NH O N HN CH C CH NH O H 3 C CH 2 CH 3 CH C CH 2 NH O CH C H 2 C NH O HO CH C CH 2 HN O COH O CH C H 2 C HN O C HO O CH C CH 2 HN O HO CH C CH 2 NH O HO CH C CH 2 OH O H 2 C C OH O NH 2 CH C CH HN O CH 3 H 2 C CH 3 CH C CH 2 H N O H 2 C CH 2 H 2 C H 2 N CH C CH NH O H 3 CCH 3 CH C H 3 C NH O CH C CH OH O H 3 C CH 3 KHIFSDDSSE -- specific to astrocytes K I V A V K H I F S D D S S E IKVAV – specific to neurons (b) Free End Free End Bind to APTS SAM Bind to APTS SAM NH 2 CH C CH 2 NH O CH 2 H 2 C H 2 C NH 2 CH C CH 2 NH O N HN CH C CH NH O H 3 C CH 2 CH 3 CH C CH 2 NH O CH C H 2 C NH O HO CH C CH 2 HN O COH O CH C H 2 C HN O C HO O CH C CH 2 HN O HO CH C CH 2 NH O HO CH C CH 2 OH O H 2 C C OH O NH 2 CH C CH HN O CH 3 H 2 C CH 3 CH C CH 2 H N O H 2 C CH 2 H 2 C H 2 N CH C CH NH O H 3 CCH 3 CH C H 3 C NH O CH C CH OH O H 3 C CH 3 KHIFSDDSSE -- specific to astrocytes K I V A V K H I F S D D S S E IKVAV – specific to neurons (b) Free End Free End Bind to APTS SAM Bind to APTS SAM Figure 5.7: (a) Schematic illustrating the chemistry of the peptide IKVAV and KHIFSDDSSE immobilization onto glass substrates. (b) Schematic showing amino acids in IKV A V and KHIFSDDSSE peptide. 208 BSAM functionalization The chemical steps involved in BSAM functionalization of glass substrate are shown in figure 5.7. In the first step, APTS (amino propyl triethoxysilane) SAM is chemisorbed onto glass substrate (coverslip, Corning) using a procedure similar to ODS SAM discussed in §4.3. The glass coverslip (Corning) was treated with a mixture of sulphuric and nitric acids, followed by adsorption of APTS which binds through the ethoxysilane group with the hydroxyl groups of the surface. This results in amine terminated surface. Step 2 complete the BSAM via the covalent attachment of the peptide ligand to the APTS SAM through the standard EDC (1-ethyl-3-(3- dimethylaminopropyl) carbodiimide) mediated formation of amide bonds. Peptide ligand attaches through their carboxyl end (glutamic acid (E) for KHIFSDDSSE and valine (V) for IKVAV) to the amine end of the SAM. The protocol for this step is adapted from [5.18]. Such surfaces, as well as negative (APTS only) and positive (poly-d-lysine coated) control samples were utilized in the cell adhesion/rejection studies described later. As noted above, the binding of the free end peptide ligand to the cell receptors is not much dependent upon the particular chemistry involved in binding of the BSAM at the solid surface end. This allows us to examine separately issues relating to the peptide ligand-receptor binding and the SAM binding to the particular solid surface (glass, alumina, metallic). For the bare glass surface, and the modified surfaces resulting from step 1 (SAM) and step 2 (BSAM), tapping mode atomic 209 force microscopy was used to ascertain the degree of surface roughness on the lateral size scale of the cells themselves (~10microns). Large area (several millimeters) uniformity of the adsorbed SAM and BSAM were examined via optical microscopy utilizing dyes to label the APTS SAMs and the peptide ligand. We emphasize that for surface functionalization at the level of a single BSAM thick layer (i.e. a BSAM monolayer) to be effective in direct binding with the cell receptor, the starting surface of the prosthetic materials must be flat on a height scale less than the size of the molecule. Hence the atomic scale characterization of the roughness of these surfaces is of significance in providing confidence in unambiguously interpreting the observed behavior in subsequent cell adhesion experiments, namely, is the specific adhesion, if observed, truly due to the direct peptide ligand-cell receptor binding. Finally, the BSAM modified and well characterized surfaces were exposed to cell cultures (E16 Hippocampal Neurons, culture protocol attached in appendix D). Unlike most cell culture studies reported in the literature, the samples studied here are based upon long (a week or longer) exposure time. This enables us to examine the adhesion characteristics of fully grown neuronal cells at long durations. We emphasize that even though in the neural prosthesis the neurons and astrocytes are to be attached preferentially and selectively on the metallic electrodes and adjacent ceramic/polymeric regions respectively, current experiments examined neuron adhesion only on glass surfaces modified with IKVAV and KHIFSDDSSE, along with control surfaces such as blank, APTS coated, and PDL coated. 210 Results: Bare, SAM, and Peptide Ligand Functionalized Surfaces Figure 5.8 shows illustrative large area optical microscope and small area AFM images, respectively, of the glass surface before (panels a and b) and after (panels c and d) modification with APTS. To enable optical microscopy, the amine reactive dye FITC (fluorescein isothiocyanate) was used along with the APTS. The uniformity of the surface coverage over macroscopic areas ~1 mm 2 and over nano- scale areas ~ 100 nm 2 is evident from these images. The AFM determined surface roughness of 0.28nm and 0.56nm for the bare and APTS covered glass indicate these to be fairy smooth. The APTS modified surfaces were subsequently treated with TAMRA (tetramethyl rhodamine) dye-labeled peptide ligand solutions. The resulting surfaces were examined again using fluorescence and AFM and typical results are shown in figure 5.8, panels (e) and (f). The peptide ligand coverage is also seen to be fairly uniform over macroscopic and nano-scale areas. Note that the peptide ligand modified surface has acquired significant surface roughness. Results: Cell Adhesion E16 Hippocampal neurons were cultured on such peptide immobilized glass substrates, alongside with negative and positive control surfaces to examine the nature of cell adhesion. Cultures were carried out for a period of one week. Figure 5.9 shows illustrative fluorescence microscopy images of neurons (with their nucleus labeled with propidium iodide fluorescent dye) adhered to APTS, PDL, IKVAV, and KHIFSDDSSE treated glass. Figure 5.10 shows the corresponding neuron nucleus surface coverage histogram for the differently functionalized glass surfaces. As 211 Figure 5.8: Optical fluorescence (left column) and AFM (right column) images of plain glass (top); APTS grafted glass treated with FITC dye (middle); and APTS grafted glass reacted with TAMRA labeled peptides ligands (bottom). 212 expected, an enhanced coverage of neurons on IKVAV functionalized surface is found compared to KHIFSDDSSE. However, PDL which represents nonspecific electrostatic binding, not the specific ligand-receptor binding, shows even higher coverage. Figure 5.9: Fluorescence microscopy images of neurons cultured on glass substrate functionalized as marked. Neuron nuclei were stained with propidium iodide dye for fluorescent imaging. 213 Figure 5.10: Neuron nuclei surface coverage histogram for glass substrate functionalized with APTS, PDL, IKVAV, and KHIFSDDSSE. The observation of higher neuron coverage on PDL covered glass compared to the IKVAV neuron adhesion molecule covered glass indicates that the peptide ligand density distribution and receptor-ligand binding are not optimal and further studies into the underlying factors that affect these are needed. Indeed, two such considerations are (1) the topology and roughness of the surface presented to the cell; and (2) the conformation of the peptide ligand. Flat starting substrate topology of the type used in most prior and these studies do not mimic the curvature of the extra- cellular matrix that most cells experience in their natural environment, by best estimates ~10nm amplitude undulation on the length scale (lateral) of ~100nm (as symbolically depicted in figure 5.6(a)). Distinct from this curvature, and riding on top of it, is the surface roughness, i.e. surface height fluctuations on lateral length 214 scales of <1nm (figure 5.6(b)). The relative magnitude of the starting solid surface height fluctuation (i.e. peak-to-valley ratio) in relation to the length of the conjugating BSAM (~1nm), as depicted in Fig. 5.6(b), can reduce the effective density of exposed peptide ligand accessible to the cell receptors. The second consideration is the affinity of the receptor-ligand binding which could be significantly reduced due to changes in the conformation of the peptide ligand as presented to the cell surface receptor. Indeed, a marked increase in surface roughness is seen after the peptide ligand adsorption (figure 5.8(f)) and suggests the need for closer examination of the underlying reasons [5.19]. In applications to biological detection, the function of surface modification is only to enhance the selective adsorption of targeted cells. The initial studies on the BSAM surface modification based on the specific binding between peptide ligand and cell receptor seems to be of promise for what is necessary for this application. In contrast, surface modification for prosthetic implants carries the additional task to maintain the long term vitality of the adhered cells. Although ligand-receptor binding is an approach being investigated in the context of on prosthetic surfaces modification for some time, the long term consequences of the strong mechanical disturbance due to ligand-receptor binding on the biological functioning of the adhered cells has not been sufficiently examined. It is however well known that the cell’s topological organization and intra- and extra-cellular biochemical processes are sensitive to its mechanical environment as well [5.20]. Weak cues of mechanical stress variations of the type inherent in the intimate binding of the cell through its 215 trans-membrane protein surface receptors can and are likely to cause changes in cell behavior over time. Recent studies involving control of the density and spatial distribution of ligand [5.21] reveal that below a certain ligand spacing (~60nm), aggregation of the receptors in the MC3T3 osteoblast cells used in the experiments involved gave rise to focal adhesion that activated viniculin and actin mediated impact on the cell. This should be expected and is important since the dynamics of the receptors in the membrane is impeded by their anchoring, thus leading not only to clustering and focal adhesion, but also potentially activating unidentified biochemical signaling cascades. In fact in the case of our current study and most of the past studies of cell adhesion to peptide ligand functionalized surfaces, the high level of surface roughness and simultaneously a high packing density of the peptide ligand, as noted above, may has an unintended benefit. That is the live cell has the ability to self-adjust its topology and the projections of its trans-membrane protein receptors participating in binding with the effectively lowered exposed ligand density presented by a rough surface. 216 Chapter 5 References: [5.1], K. L. Prime, G. M. Whiteside, “Self-assembled organic monolayers – model system for studying adsorption of proteins at surface”, Science 252, 1164-1167 (1991). [5.2] J. Buijs, D. W. Britt, V. Hlady, “Human Growth Hormone Adsorption Kinetics and Conformation on Self-Assembled Monolayers”, Langmuir, 14, 335-341 (1998). [5.3] P. S. Stayton, J. M. Olinger, S. T. Wollman, P. W. Bohn, S. G. Sligar, ”Engineering proteins for electrooptical biomaterials”, Chap. 19 in Molecular and Biomolecular Electronics, Ed. R. R. Birge, American Chemical Society, Washington DC (1994). [5.4] R. A. McMillan, C. D. Paavola, J. Howard, S. L. Chan, N. J. Zaluzec, J. D. Trent, “Ordered nanoparticle arrays formed on engineered chaperonin protein templates”, Nature Materials, 1, 247-252 (2002). [5.5] P. J. B. Koeck, H. K. Kagawa, M. J. Ellis, H. Hebert, J. D. Trent,” Two- dimensional crystals of reconstituted beta-subunits of the chaperonin TF55 from Sulfolobus shibatae”, Biochimica et Biophysica Acta-Protein Structure and Molecular Enzymology, 1429, 40-44 (1998). [5.6] Toward Replacement Parts for the Brain – Implantable Biomimetic Electronics as Neural Prostheses, Eds. T. W. Berger, L. Glanzman, MIT Press, Cambridge, MA (2005). [5.7] V. S. Polikov, P. A. Tresco, W. M. Reichert, “Response of brain tissue to chronically implanted neural electrodes” J. Neurosci. Meth. 148, 1-18 (2005). [5.8] J. W. Weiland, W. Liu, M. S. Humayun, “Retinal prosthesis”, Annu. Rev. Biomed. Eng. 7, 361-401 (2005). [5.9] T. W. Berger, G. Chauvet, R. J. Sclabassi, “A biologically-based model of functional-properties of the hippocampus”, Neural Networks, 7, 1031-1064 (1994). [5.10] K. Tashiro, G. C. Sephel, B. Weeks, M. Sasaki, G. R. Martin, H. K. Kleinman, Y . Yamada, “A synthetic peptide containing the IKV A V sequence from the A-chain of laminin mediates cell attachment, migration, and neurite outgrowth”, J. Biol. Chem. 264, 16174-16182 (1989). [5.11] E. Ruoslahti, “RGD and other recognition sequences or integrins” Annu. Rev. Cell Dev. Biol., 12, 697–715 (1996). 217 [5.12] Y. Rao, X. F. Wu, J. Gariepy, U. Rutishauser, C. H. Siu, “Identification of a peptide sequence involved in homophilic binding in the neural cell-adhesiion molecule NCAM”, J. Cell Biol., 118, 937-949 (1992). [5.13] J. P. Ranieri, R. Bellamkonda, E. J. Bekos, T. G. Vargo, J. A. Gardella, P. Aebischer, “Neuronal cell attachment to fluorinated ethylene propylene films with covalently immobilized laminin oligopeptides YIGSR and IKVAV.” J. Biomed. Mater. Res. 29, 779–785 (1995). [5.14] S. Saneinejad, M. S. Shoichet, “Patterned glass surfaces direct cell adhesion and process outgrowth of primary neurons of the central nervous system” J. Biomed. Mater. Res. 42, 13-19 (1998). [5.15] K. C. Olbrich, T. T. Andersen, F. A. Blumenstock, R. Bizios, “Surfaces modified with covalently-immobilized adhesive peptides affect fibroblast population motility” Biomaterials 17, 759-764 (1996). [5.16] S. Lu, A. Bansal, W. Soussou, T. W. Berger, A. Madhukar, “Receptor-Ligand- Based Specific Cell Adhesion on Solid Surfaces: Hippocampal Neuronal Cells on Bilinker Functionalized Glass“, Nano Lett. 6, 1977-1981 (2006). [5.17] Y. Zhong, X. Yu, R. Gilbert, R. V. Bellamkonda, “Stabilizing electrode-host interfaces: A tissue engineering approach”, J. Rehabil. Res. Dev. 38, 627-632 (2001). [5.18] L. Kam, W. Shain, J. N. Turner, R. Bizios, “Selective adhesion of astrocytes to surfaces modified with immobilized peptides”, Biomaterials 23, 511-515 (2002). [5.19] The marked roughness of the peptide surface is likely due to polymerization of the peptides in solution prior to adsorption on the SAM covered surface. (Mark Thompson, Private Communication) [5.20] V. Vogel, M. Sheetz, “Local force and geometry sensing regulate cell functions” Nat. Rev. Molec. Cell Biol. 7, 265-275 (2006). [5.21] M. Arnold, E. A. Cavalcanti-Adam, R. Glass, J. Blummel, W. Eck, M. Kantlehner, H. Kessler, J. P. Spatz, “Activation of integrin function by nanopatterned adhesive interfaces” Chem. Phys. Chem., 5, 383-388 (2004). 218 Chapter 6. Hybrid Colloidal/Epitaxial Nanostructures – A Study of Excitation Transfer In chapter 4, we discussed the synthesis of various hybrid colloidal/epitaxial nanostructure based systems and, in the process, discussed the following three common bases encountered: (a) the passivation of GaAs substrate; (b) nanocrystal deposition on substrates via dip-coating; and (c) nanocrystals linked to SAM functionalized substrates. Additionally, in the chapters 3 and 5, respectively, we discussed our studies of the selective binding of biological cells with nanocrystals and substrates via specific ligand-receptor interaction. Hence, towards our roadmap of using integrated colloidal/epitaxial structure platforms for biological applications as recalled here in figure 6.1, each aspect of the synthesis of such systems has been Figure 6.1 Schematic of sensing biological agents using integrated colloidal/epitaxial hybrid nanostructures chips. 219 investigated and reported in chapters 1 through 5 with different degrees of depth. In this chapter we now focus on our studies of the optical properties of the most basic of the colloidal/epitaxial hybrid structures and their unique optical properties. This study has revealed a non-conventional excitation transfer between nanocrystal quantum dots and the semiconductor substrate, an effect that might serve as a communication link between the two. The chapter is organized as follows: first we briefly review the current knowledge of photoluminescence from nanocrystals as it will be needed for the discussion of excitation transfer with the substrate; second we describe the PL instrumentation, focused mainly on the time-resolved PL system established by this author as part of this dissertation; then we move to the topic of excitation transfer between nanocrystals and substrates. Our study is performed on the system of InAs/ZnSe nanocrystals synthesized by the author (chapter 2) and physisorbed on passivated GaAs substrates via the dip-coating approach. Systematic experimental studies aimed at understanding the nature of the excitation transfer in this system are presented as guided by general theoretical understanding. Indeed, to our knowledge, these studies provide, for the first time, information on as fundamental an aspect of nanocrystal-solid substrate hybrid structures as their relative energy alignment, i.e. the equivalent of the so called band-edge discontinuity (or band-offset) in heterostructures. The band-offset, and their tailoring, are at the core of the semiconductor heterostructures science and electronic and optoelectronic technology. 220 Its equivalent for the NCQD substrate systems is thus the key to exploiting this new class of hybrid quantum structures (see §6.3). §6.1 Photoluminescence of Nanocrystal Quantum Dots. §6.1.1 Overview of Photoluminescence Process The size-tunable photoluminescence (PL) is the most important optical property of low dimensional quantum nanostructures, and also maybe the best known finite size effect of any kind. Different PL spectroscopic methods probe different aspects of the electronic structure and the optical processes in these quantum structures, which serve as the main workhorse for optical characterization. As illustrated in figure 6.2, PL in bulk semiconductors or low-dimensional quantum nanostructures can be approximated by a three step process: photogeneration, relaxation and recombination of carriers. Owing to the continuous density of electron states in bulk semiconductor and quantum wells, these three steps involve: (1) continuous absorption of photons above the material fundamental bandgap, thus generating electrons and holes in high excited states. Owing to the rapid electron- electron interaction (femtosecond), such electrons and holes have a “distribution” “temperature” much higher than the lattice temperature and hence are referred to as “hot” carriers; (2) given the continuous density of states (DOS) in bulk semiconductor and QWs, the hot electrons and holes quickly (sub-picosecond) relax to the states near the bottom of the conduction band and the top of the valence band by interaction with the lattice (mainly through the Fröhlich coupling with the 221 LO-phonons), thus establishing a certain local distribution, usually a few degrees higher than the lattice temperature; (3) from there, the electron and hole pair (coulomb bound, called exciton) recombines at much slower rate compared to relaxation (on the order of 1ns). Radiative recombination generates photoluminescence that reflects the distribution of the carriers in the near bandedge states. Excited electron Excited hole hv PL CB Edge or LUMO VB Edge or HOMO hv Abs (1) Photon Generation (2) Carrier Relaxation (3) Radiative Recombination (2) Carrier Relaxation Excited electron Excited hole hv PL CB Edge or LUMO VB Edge or HOMO hv Abs (1) Photon Generation (2) Carrier Relaxation (3) Radiative Recombination (2) Carrier Relaxation Figure 6.2 Schematic showing the 3-step model of the PL process: (1) photon generation, (2) carrier relaxation, (3) radiative recombination. 222 The PL process in quantum dots approximates to the same three general steps as above; however, owing primarily to the discrete nature of the electron density of states arising from the 3D confinement, the details involved are quite unique. (For a detailed review of the photoluminescence processes in NCQDs see [6.1]). Below we capture some of the important features as needed for the discussion of excitation transfer between nanocrystals and substrates. §6.1.1(a) Inhomogeneous Broadening Dominates the PL Width Unlike the continuous density of states (DOS) in bulk semiconductors or QWs, the δ-function DOS in quantum dots combined with small finite number of electrons precludes the notion of carrier distribution from applying to a single QD. The width of the HOMO-LUMO transition line is thus expected to be of Lorentian shape: 2 2 0 2 0 ) ( ) ( Γ + − Γ = ω ω ω A A (6.1) Here 0 ω is the resonant frequency of the HOMO-LUMO transition, 0 A is an arbitrary constant, and Γis the width of the Lorentian PL line. The linewidth Γis inversely proportional to the dephasing time of the HOMO-LUMO exciton. Common dephasing mechanisms include radiative/non-radiative recombination of exciton and the exciton-phonon coupling. If recombination is the only dephasing mechanism, then 1 2 1 T = Γ , where 1 T is the recombination lifetime of the exciton (as measured in time-resolved PL experiment). 223 Indeed narrow PL linewidths as small as 100 μeV (limited by instrument resolution) are observed in low temperature (10K) single QD PL experiments [6.2]. Such narrow line width is close to what is expected from exciton dephasing due to recombination alone, thus providing evidence for nearly δ-function energy of electronic states in QDs. However most PL measurements of NCQDs, given the small absorption cross-section of the QDs (on the order of 1nm 2 ), are performed on an ensemble of large number of QDs. The PL of an ensemble of NCQDs is the superposition of the PL from individual NCQDs. The width of PL from such an ensemble of QDs is broadened due to the QD size variation and the corresponding spread of HOMO- LUMO gap (referred to as inhomogeneous broadening). The QDs size distributions resulting from the best synthesis (chapter 2) still have a significant size variation (~10%) and an inhomogeneous PL broadening ~100meV (for both CdSe and InAs nanocrystals). §6.1.1(b) Relaxation in QDs. Due to the δ-function DOS in QDs, their carrier relaxation mechanisms are quite different from those in bulk semiconductors and QWs which are dominated by the electron-phonon coupling. The phonon mediated relaxation of carriers in NCQDs is expected to slow dramatically since now the phonon energy has to match exactly the energy difference between discrete states, a less probable situation. This phenomenon is well-known as the “phonon bottleneck” [6.3, 6.4, 6.5]. In QDs, due 224 to the small electron effective mass, the energy levels of electron have large spacing. In figure 6.3 are shown the energy levels of 5.2nm InAs nanocrystals calculated using p k r r ⋅ method by the author as presented in chapter 2. The electron 1Pe to 1Se level has a large separation of ~500meV which is far larger than the LO-phonon energy (~29meV for InAs). Hence the relaxation of electron is expected to be hindered, since now the electron-phonon interaction can only occur via multiple phonon process, which is exponentially slower as a function of the number of phonons involved. InAs Core 5.2nm dia. Surfactant Surfactant 1Se 1Pe 1VB 2VB Eg(InAs)=0.418 (LT) …… …… 1eV ΔE(1Pe-1Se)~500meV InAs Core 5.2nm dia. Surfactant Surfactant 1Se 1Pe 1VB 2VB Eg(InAs)=0.418 (LT) …… …… 1eV InAs Core 5.2nm dia. Surfactant Surfactant 1Se 1Pe 1VB 2VB Eg(InAs)=0.418 (LT) …… …… 1eV ΔE(1Pe-1Se)~500meV Figure 6.3 Calculated electronic structure of InAs nanocrystal (diameter 5.2nm). Details of calculation are provided in chapter 2. 225 Experimentally the electron relaxation has been thoroughly studied in the CdSe nanocrystal system using ultrafast spectroscopy techniques such as transient absorption (TA) pump-probe technique [6.6, 6.7, 6.8]. However, such measurements conclude the existence of subpicosecond relaxation of the electrons in contradiction to the slowdown expected from the “phonon bottleneck”. As shown by TA data (figure 6.4), in 8nm diameter CdSe the 1Pe to 1Se relaxation time is ~500 fs in spite of an energy separation of about 8 LO-phonons between these two states. Therefore such a relaxation can not be phonon mediated. It has been proposed that the fast electron relaxation the QDs is due to the Auger-type e-h scattering [6.9] in which the Figure 6.4 Transient absorption (TA) spectrum of CdSe/TOPO NCQD showing sub- ps electron relaxation to its 1S (LUMO) state. Data taken from [6.6 Klimov 2000] 226 electron transfers its excessive energy to the hole via the e-h coulomb interaction and relaxes to the LUMO level. The hole subsequently relaxes through its dense spectrum of states (see below) to the HOMO level. This model of electron and hole relaxation is shown schematically in figure 6.5. Indeed in strongly confined QDs, the electron-hole coulomb interaction is enhanced due to the large overlap of their wavefunctions and can provide an intrinsic fast relaxation pathway for the electrons. A key experimental prove for the above proposed mechanism is the electron relaxation time as a function of nanocrystal surface passivation. TA measurements have been performed on CdSe nanocrystals with ZnS shell which were expected to have no hole traps and on CdSe nanocrystals capped with pyridine surfactant which were expected to have strong hole traps. [6.6, 6.8] (Note: pyridine surfactant stabilizes hole on its conjugated ring which reduces the hole coulomb interaction with the electron.) The two CdSe samples (ZnS coated and pyridine capped), respectively, show electron 1Pe to 1Se relaxation time of 0.3ps and 3ps [6.6, 6.8], which is consistent with the proposed role of hole in electron relaxation: with the hole trapped by the pyridine, the electron relaxation is slowed due to lack of Auger relaxation pathway. 227 Auger Energy Transfer Hole relaxation in dense energy levels through consecutive phonon emission (a) Electron Relaxation (b) Hole Relaxation Auger Energy Transfer Hole relaxation in dense energy levels through consecutive phonon emission (a) Electron Relaxation (b) Hole Relaxation Figure 6.5 Schematic showing the mechanism of carrier energy relaxation in nanocrystal quantum dots: (a) electron relaxation via Auger energy transfer to hole (b) hole relaxation via consecutive phonon emission. Compared to the electrons levels, the hole levels in the NCQDs have much smaller interlevel spacing due to the large effective mass of the holes. As shown in figure 6.3., the hole levels in 5.2nm diameter InAs nanocrystals (calculated using p k r v ⋅ in chapter 2) are separated by around or less than 100meV. These levels split into fine structures due to the effect of NCQD nonspherical shape [6.10, 6.11] and e- h exchange interaction [6.12]. The coupling to lattice vibrations (phonon coupling) also causes broadening of these levels [6.13]. These effects all smear out hole levels and lead to small interlevel separation. The hole relaxation dynamics has been examined again in the CdSe NCQD system via the second-harmonic-generation (SHG) up-conversion based TRPL technique of sub-ps resolution. (The TA 228 technique for probing electron relaxation dynamics noted above does not work effectively for holes due to the difficulty in saturating the hole levels due to high spectral density.) Figure 6.6 shows an example of the SHG TRPL spectra of 3.6nm diameter CdSe NCQDs detected at different energies (Data taken from [6.14]). This measurements indicate that the hole energy descends gradually from the high excited states to the HOMO level (figure 6.4(b)). The hole energy relaxation rate derived is ~1.5eV/ps (except that the last stage of relaxation from 2VB to 1VB is slowed to ~0.2eV/ps due to large hole level separation [see Ref 6.14]). This relaxation rate in NCQDs is comparable to that of hole in bulk CdSe (~1.4eV/ps) [6.15]. This suggests pump pulse autocorrelation PL from HOMO- LUMO Transition pump pulse autocorrelation PL from HOMO- LUMO Transition Figure 6.6 Dynamics of ‘‘hot’’ PL detected at different spectral energies in CdSe NCQD (R=1.8nm, HOMO-LUMO PL emission 2.16eV). This measurement shows that the hole energy relax gradually from the high excited states to the HOMO level through a ladder of intermediate states, as depicted in figure 6.5(b). Data Taken from [6.14 Xu 2002]. 229 that the model for hole relaxation in NCQD is similar to that in bulk – via a cascade of single phonon scattering event through a ladder of quasi-continuous intermediate hole levels as depicted in figure 6.5(b), and not necessarily hindered by the phonon bottleneck. This hole relaxation model is expected given the small hole level separation as argued above. §6.1.1(c) PL Intermittence and “dark” charged QDs The distinct feature of the PL of an individual QD (especially the NCQDs) not present in bulk or QW PL, is its intermittence: the PL intensity of a QD is found to fluctuate in an “on-off” fashion on a time scale of millisecond to minutes [6.16]. This PL intermittence is colloquially referred to as QD “blinking”. This phenomenon has been well studied in many single QD PL experiments (mainly on CdSe NCQDs) [6.17, 6.18, 6.19]. As an illustrative example, in figure 6.7 is shown a time trace of PL intensity of a CdSe NCQD (data adapted from ref 6.20). Given the essential physics involved (to be discussed below), PL intermittence is also expected in our InAs based NCQDs. The PL intermittence is correlated to another important single QD PL phenomenon -- spectral diffusion (i.e. a discrete or continuous shift of QD emission wavelength as a function of time). For CdSe QDs the spectral diffusion can be as large as 50meV [6.2]. It is observed by Bawendi et al. that the discrete shift of emission wavelength coincides with the on/off transition events in CdSe QD PL [6.19]. 230 Figure 6.7 Time trace of photoluminescence from a single ~29 Å radius CdSe QD. (Taken from 6.20 Kuno 2000). The initial model [6.16] for single QD PL intermittence is built upon the theory of Auger-ionization induced nanocrystal photodarkening [6.21] and is later refined by Bawendi et al [6.19] considering the connection between PL intermittence and spectral diffusion. Bawendi’s model captures a process depicted in figure 6.8, as described below. A NCQD before photoexcitation is typically charge neutral, which we refer to as in “resting” state (figure 6.8(a1)). Such a QD can be turn into excited state (creation of exciton in QD) after absorbing an excitation photon (figure 6.8(a2)). The exciton in the excited QD normally recombines to generate photoluminescence and return the NCQD to the “resting” state. 231 hv Abs e - h + h + e - h + e - e - e - e - e - e - h + h + h + h + h + h + Charged NCQD with lone carrier are “Dark”. e - (a1) Resting QD e - h + h + e - (a2) Excited (a3) “Dark” (a4) Resting Auger Energy Transfer Lone hole in a positively charged NCQD (b) Auger energy transfer from the exciton to the lone carrier in charge NCQD quenches photoluminscence hv Abs e - h + h + e - h + e - e - e - e - e - e - h + h + h + h + h + h + Charged NCQD with lone carrier are “Dark”. e - (a1) Resting QD e - h + h + e - (a2) Excited (a3) “Dark” (a4) Resting Auger Energy Transfer Lone hole in a positively charged NCQD Auger Energy Transfer Lone hole in a positively charged NCQD (b) Auger energy transfer from the exciton to the lone carrier in charge NCQD quenches photoluminscence Figure 6.8 (a) Schematic showing the capture and release of lone charge in a NCQD that cause the NCQD photoluminescence to be turned “off” and “on”. Schematic adapted from [6.23 Shimizu 2004] (b) Schematic showing the Auger energy transfer from an exciton to a lone carrier in the NCQD that quenches the photoluminescence of a charged NCQD. However, in an excited QD, other then exciton recombine, there is also a finite probability for one of the carriers to be trapped to the QD surface or in the surrounding media (mechanism to be discussed below), which leaves behind a charged QD with a lone carrier, as schematically shown in figure 6.8(a3). Such QDs with lone carrier upon further light absorption do not generate photoluminescence, and are referred to as in “dark” state (figure 6.8(a3)). The reason for the quenching of 232 the PL is that a QD with lone charge carrier upon photon absorption will be turned into a 3 particle system involving one exciton and a lone carrier. The energy transfer from the exciton to the lone charge carrier (illustrated in figure 6.8(b)) in the same QD is predicted to be ~100ps [6.22], much faster than the typical radiative recombination (10ns-1 μs) of the exciton. The photoluminescence from a “dark” QD figure 6.8(a3) will be restored when (a) the alone charge in the “dark” QD also becomes trapped at the QD surface or in the surrounding media; or (b) the “dark” QD captures another opposite charge from its surrounding to recombine with the lone charge carrier. Hence the “dark” QD will be neutralized and restored into “resting” state again (figure 6.8(a4)), which upon further photoexcitation is capable of PL emission. At low excitation power densities, the time a QD spends in excited state is little. The quantum dot is switched between “resting” and “dark” state, upon capture (release) of charge from (to) the trap on NCQD surface or surrounding media. In its “resting” state, the QD generate PL emission upon photoexcitation, while in its “dark” state, the QD do not. Because in typical single QD PL measurement the integration time (ms-sec) smears out single photon emission events, in the PL intensity trace as function of time (figure 6.7), one observe the single QD PL being switched back and forth between continuous periods of “on” and “off”, which physically corresponds to the QD switching between neutral “resting” and charged “dark” states. Hence in the general literature of NCQD, the “resting” and “dark” state are, respectively, referred to as “on” and “off” state. The time duration in which a 233 NCQD is in the “dark” (“off”) state is decided by the rate at which the lone carrier is released from a QD. The time duration of the “resting” or “on” state on the other hand is decided by the rate of photoexcitation, and the probability of single carrier trapping after each photoexcitation events. The capture (release) of charge from (to) the trap on NCQD surface or surrounding media, as described above, not only switching the QD between the normal “resting”(”on”) state and the charged “dark”(“off”) state, it also changes the local electrical field experienced by the NCQD (as schematically shown in figure 6.8(a1) and (a4), the resting state after capturing and releasing the lone carrier do not necessarily has the same charge distribution in its surrounding. This change in local electrical field distribution cause emission wavelength of QD to shift due to Zeeman effect. Therefore the PL intermittence is correlated with the spectral shift [6.19]. Systematic studies of the distribution of single QD PL “off” and “on” time duration as a function of temperature and excitation power density have been reported by many groups for CdSe nanocrystals [6.20]. The “off” time duration follows a power law probability distribution: α − ⋅ = t A t P ) ( , where ) (t P is the probability of the “off” time being equal to time t, A is a constant, and α is the power law coefficient, α~1.5 for CdSe NCQDs [6.20]. Surprisingly, the “off” time is nearly independent of temperature or excitation power density, which suggests that the process (charge transfer from/to surface or surrounding media) that turns the NCQD into “on” state is a tunneling process (not phonon assisted). 234 The “on” time distribution follows a similar power law as the “off” time at lower excitation powers and temperature. The probability of long “on” time deviates (decreases) from the power law at higher excitation power and temperature. Therefore the process that turns the NCQD into the “off” state can have two separate mechanisms: (1) temperature-independent tunneling process; and (2) temperature and excitation power dependent photo-ionization. Bawendi et al proposed a phenomelogical “random walk” model that explained the “off’-“on” time probability distribution and the independence of the “off” time from temperature and power density [6.23]. Briefly, this model assumes that the relevant energy level in the NCQD and a hypothetical trap state at the QD surface or in the surrounding are fluctuating in energy in a 1D random walk fashion; when the two energy states are in resonance, charge transfer occurs due to temperature independent tunneling process which switches the QD between the “on” and “off” states. It is worth to hint here the close relation between the PL intermittence and the main subject of this chapter – the excitation transfer between NCQDs and substrates. Both are modification of the PL behavior of NCQDs due to charge transfer to its environment (surface trap, media, or a nearby substrate). The PL intermittence and spectral diffusion of QD is not observed in bulk semiconductor or QWs, but is very similar to what observed in fluorescence from molecular species [6.24, 6.25]. Indeed both NCQDs and dye fluorophores have large surface-to-volume ratio and typically reside in a non-solid-state surrounding, such as in solution or air. Therefore their 235 optical properties are susceptible to local environment change (charge and electric field, etc). However, it is exactly this susceptibility that needs to be utilized cleverly to enable the QD to serve as the desired detector. §6.1.2 Photoluminescence Spectroscopy Methods In usual PL measurements of NCQDs the excitation energy is fixed and the detection energy is scanned. Typically the excitation is set well above the HOMO- LUMO transition of the smallest QDs, so that PL emission occurs from the entire NCQD ensemble. Such PL is referred to as full luminescence spectrum. As explained above in §6.1.1, the full luminescence spectrum of QD ensembles is inhomogeneously broadened and gives the distribution of the NCQD HOMO-LUMO gap. With increasing excitation power density, it is also possible to exhibit higher excited states in PL spectrum by creating multi-exciton excitation in QDs and saturating the HOMO-LUMO levels. The threshold for multi-exciton event can be derived from the absorption cross-section σ , and exciton recombination lifetime τ : τ σ ⋅ hv I th ~ (6.2) where th I is the threshold excitation power density and hv is the excitation photon energy. For a typical InAs NCQD, σ ~1nm 2 , τ ~1ns, hv ~1.6eV, the threshold excitation power density th I ~2.4x10 4 W/cm 2 . Beyond this threshold, the PL process becomes lot more complicated involving state-saturation and exciton-exciton 236 interaction. The multi-exciton PL process is an important field for the application of the NCQDs for lasers and is beyond the scope of this dissertation. All of our experimental work is performed at excitation powers (1-100W/cm 2 ) much smaller than the threshold. Thus, in which no multi-exciton excitation is involved and the PL intensity scales linearly with the excitation power density. Photoluminescence Excitation Spectroscopy: The PL measurement also can be performed by fixing the detection at certain energy and scanning the excitation energy. Such a spectrum of PL intensity as a function of excitation energy is called the PL excitation (PLE) spectrum. The PLE is similar to absorption spectrum in reflection at different energy. The PLE spectrum differs from the absorption, however, as it also depends on the choice of detection energy, the relaxation process of the carrier, and the competition between radiative recombination and non-radiative recombination. PLE is an important tool for the study of nanocrystal/substrate hybrid structure because of the following reasons: (1) It is applicable when the material is too thin for absorption measurement or the substrate is not transparent. Specifically in our study, it probes the optical process in a submonolayer of nanocrystals deposited on the substrate. (2) The PLE spectrum reflects the loss of carriers during relaxation process, which can be important during excitation transfer between nanocrystals and substrates. 237 (3) In an ensemble of nanocrystals of a large size distribution and inhomogeneously broadened PL, by setting a narrow detection window, PLE provides size-selectivity. Because the excitons in nanocrystals always relax to their HOMO-LUMO states before emission, PLE probes the optical process in a selectable subset of nanocrystals with a narrow size distribution that emit only in the detection window. Connected to the size-selectivity of the PLE, PL spectrum can also be measured with excitation set at a special energy (called selective PL) to provide size selectivity and enhance certain PL features. A commonly applied selective PL technique for nanocrystals is fluorescence line-narrowing (FLN) spectroscopy [6.26]. In FLN the excitation is set at the low-energy end of the first absorption peak of nanocrystal ensemble. Thus only the largest nanocrystals are excited, and a significantly narrowed PL spectrum can be achieved. FLN can reveal fine electronic structures in the NCQDs otherwise smeared out in the full luminescence spectrum [6.10, 6.27]. Time-Resolved Photoluminescence: Other than the time-integrated PL technique discussed above, it is useful to probe to the evolution of PL emission as a function of time delay after excitation. Such measurement is called time-resolved photoluminescence (TRPL). The typical TRPL technique based on the direct detection of PL emission has a resolution less than 10ps. It therefore can probe only the recombination of carriers from HOMO- 238 LUMO states. Recently, the emergences of the up-conversion PL technique and ultrahigh resolution streak cameras have enabled ~200fs resolution. Thus the TRPL technique can also be used to study the carrier relaxation dynamics [6.14]. In spite of all the powerfulness of the PL-based different techniques discussed above, PL has two basic limitations: (1) the sample has to generate enough emission to be detected. (This used to severely limit the application of PL in the early works on nanocrystals when nanocrystals had very low quantum efficiency due to structural defects and surface traps); (2) the PL techniques by themselves can not determine absolute position of energy levels with respect to vacuum. Only separation between different levels can be probed. §6.2 Photoluminescence instrumentation Following the introduction of photoluminescence process and different PL techniques in §6.1, in this section we introduce the particular PL instrumentation used in the studies undertaken for this dissertation. Two PL/PLE setups were used for this dissertation work: (a) a pre-existing time-integrated PL/PLE setup (b) a new PL setup that includes time-integrated PL/PLE and time-resolved PL capability, established as a part of this dissertation work. Below we discuss the instrumentation of these two setups, with the focus on the time-resolved PL instrumentation as it was the first implementation of TRPL in our group. 239 §6.2.1 Pre-existing Time-Integrated PL/PLE Setup: In the pre-existing time-integrated PL/PLE setup, a CW Ti:Sapphire tunable laser (spectra physics 3900) pumped by a 10W Ar + laser (Coherent 310 ML-Vis) is used for excitation, the photoluminescence is dispersed with a 1m double monochromator (SPEX 1404), and detected using LN2 cooled Ge detector (North Coast). The instrumentation detail of this old PL/PLE setup has been discussed in the dissertations of many former group members [e.g. Ref 6.28, 6.29] and hence is not repeated here. However it is important for this author to note the following two observations regarding its usage for characterization of the nanocrystals on semiconductor substrates: (1) The Ar + laser (Innova 310) in the setup is a multi-line gas laser. Although the non-lasing emission from Ar + at different wavelengths (other than the few main lasing wavelengths 476nm, 488nm, 496nm, 514nm) used to be neglected in the PL measurement, since they are overwhelmed by the strong PL from the self-assemble quantum dots (SAQDs), such emission from the laser do interfere with the weak PL signal from NCQDs. Therefore a 514nm laserline filter (Newport) is used to pick out 514nm Ar + line for NCQD PL excitation. (2) Since the PL characterization of the InAs nanocrystal quantum dots on GaAs substrates often requires the spectrum to cover a wide wavelength range from 800nm to 1500nm, the calibration of the spectral response of the detection system (monochromator and Ge detector) is essential to generate reliable PL spectra. For our desired accuracy, the calibration is done by using the PL setup to measure the 240 spectrum of a light source of known spectral irradiance, such as a tungsten lamp, which can be approximated by a blackbody of a color temperature specified by the manufacturer. §6.2.2 The New PL/PLE/TRPL Setup: Overview In the following part of §6.2 we focus on the instrumentation of the new PL/PLE/TRPL setup built to extend our optical characterization capability. The setup, as schematically shown in figure 6.9, integrates PL, PLE and TRPL capabilities. In figure 6.9, all parts for TRPL are shown in red, parts for time integrated PL/PLE are shown in blue, parts shared by TRPL and PL/PLE are shown in green. Below we discuss respectively the PL/PLE and TRPL instrumentation of this setup. The TRPL being the first one built in our group is presented with more details. §6.2.2(a) Time integrated PL/PLE in the new setup For time-integrated PL/PLE measurements, an Ar + laser (Innova 310 ML-Vis) or the Ar + laser pumped Ti:S laser (Spectra-Physics 3900) laser is used as the excitation source. The Ti:S laser has a tunable range of 700-850nm or 840-1010nm using two interchangeable mirror sets dubbed, respectively, “blue set” and “red set”. However change of the mirror set is a time consuming process that requires recalibration of the output wavelength. The laser beam is chopped typically at 80Hz, attenuated by the ND filter (if necessary) and focused on to the sample mounted inside a continuous flow cryostat (Janis ST-B, 5.5K-360K) by a f=200mm lens (L1). 241 Coherent Mira 900D Mode-Locked Ti:S Laser M Delay ORTEC 425A or ORTEC 416A 1GHz x100 Pre-Amp Ortec 9306 Pico-Timing Discriminator Ortec 9307 Innova 310 Ar+ laser, w/ LD optics, 8W Time-to- Amplitude Converter (Ortec 457) MCA Ortec trump 8k Start NIN Stop NIM M L2: 3” Lens f=79.9mm ND Coherent pulse swich IntraCavity Dumper M 2”M flipping HV Power Supply C4840 L1: 1” Lens f=200mm ND InGaAs PD Acton ID-441-C Pico-Timing Discriminator Ortec 9307 Janis cryostat sample Signal from Si P.D. out HV Laser System Pulse Processing Electronics Detection Optics ND Beam Sampler Inverting Transformer Phillips460 Si PD assembly Hamamatsu S5973 15dB Attenuator 665-15-1 Acton sp300i Spectrograph SP-300i InGaAs PD Array 1/4m subtractive Monochrometer CVI DK242 PMT housing C4878 MCP-PMT Hama- matsu R3809U-59 PC & Control Program COM Counter Oriel 76915 Winspec cw Ti:S Laser (spectra-physics 3900) Flipping M Flipping M Step Motor Controller (oriel) Lock-in SR830 Chopper GPIB Beam Sampler M M M Electro meter Keithley 617 Power Meter Camera Ctrl (PI ST133) Time-Integrated Electronics Coherent Mira 900D Mode-Locked Ti:S Laser M Delay ORTEC 425A or ORTEC 416A 1GHz x100 Pre-Amp Ortec 9306 Pico-Timing Discriminator Ortec 9307 Innova 310 Ar+ laser, w/ LD optics, 8W Time-to- Amplitude Converter (Ortec 457) MCA Ortec trump 8k Start NIN Stop NIM M L2: 3” Lens f=79.9mm ND Coherent pulse swich IntraCavity Dumper M 2”M flipping HV Power Supply C4840 L1: 1” Lens f=200mm ND InGaAs PD Acton ID-441-C Pico-Timing Discriminator Ortec 9307 Janis cryostat sample Signal from Si P.D. out HV Laser System Pulse Processing Electronics Detection Optics ND Beam Sampler Inverting Transformer Phillips460 Si PD assembly Hamamatsu S5973 15dB Attenuator 665-15-1 Acton sp300i Spectrograph SP-300i InGaAs PD Array 1/4m subtractive Monochrometer CVI DK242 PMT housing C4878 MCP-PMT Hama- matsu R3809U-59 PC & Control Program COM Counter Oriel 76915 Winspec cw Ti:S Laser (spectra-physics 3900) Flipping M Flipping M Step Motor Controller (oriel) Lock-in SR830 Chopper GPIB Beam Sampler M M M Electro meter Keithley 617 Power Meter Camera Ctrl (PI ST133) Time-Integrated Electronics Figure 6.9: Schematic of the new PL/PLE/TRPL setup. (This figure does not represent the actual layout of the optical components) 242 Given the divergence of the two lasers (Ar + 0.7mrad, Ti:S 1.0mrad), the expected diameters of the focused spots sizes are 133 μm and 200 μm, respectively, for the Ar + and Ti:S lasers. The photoluminescence from the sample is focused by a 3” f=79.9mm lens (L2) into the entrance slit of a 0.3m single stage imaging spectrograph (Acton SP300i). The ND filters are inserted in front of the entrance slit as necessary. Three gratings of 300g/mm, 600g/mm, 1200g/mm blazed, respectively, at 1.0 μm, 1.0 μm and 0.75 μm are installed on Acton SP300i. The spectrograph has a resolution of ~2.5nm per mm slit width (at 500nm using the 1200g/mm grating). Spectrally dispersed light is steered to either of the two exit ports of the spectrograph which are fitted with LN2 cooled InGaAs photodiode array (1x512 linear array, Princeton Instruments Inc.) and an Peltier cooled InGaAs photodiode (Acton Research Inc.), respectively. The InGaAs array detector allows high throughput time-integrated PL measurements. The InGaAs array has spectral response from 0.8-1.6 μm. It is controlled by a PI ST-133 camera controller which is in turn connected to a PC with PI spectra interfacing card and Winspec software (Princeton Instrument). General consideration and caution of using an imaging spectrograph need to be followed. Especially, adjustment is needed to ensure the dispersed light is focused sharply on the InGaAs array detector to avoid blurring of spectrum. The InGaAs array has a relatively high dark current (~15,000e/sec/pixel) which, as a background, needs to be subtracted from the measured spectrum. 243 The PLE measurement is implemented similar to the pre-existing PLE setup [6.28], except that an InGaAs PD instead of the Ge detector is used. The PL signal is detected at a fixed wavelength using the InGaAs PD. The output from the PD is measured by a lock-in amplifier (Stanford Research 830) using the chopper signal as the reference. The tunable excitation for the PLE is provided by the Ti:S laser controlled by a stepper mike controller (Oriel). Part of the Ti:S laser beam is split towards a power meter (connected to a Keithley multimeter) that measures the Ti:S power as a function of wavelength for the normalization of the PLE spectra. A custom program (first established by Dr. R. Heitz, and developed by Dr. I. Mukhametzhanov and this author) is used to control the relevant instruments and generate the PLE spectra. §6.2.2(b) TRPL in the new setup: overview and design consideration Given our interest in the NIR application of NCQDs and its integration with the NIR quantum optoelectronic nanostructures, the TRPL is designed for NIR wavelength regime (700-1000nm excitation, 400-1200nm detection) and 10ps- 1000ns time dynamic range for the measurement of such structures. Different implementations of TRPL fall into two categories: direct methods which use high speed optoelectronic devices to directly measure the time interval between the excitation photon and the luminescence photon, such as utilizing streak camera or time-correlated single photon counting (TC/SPC); or indirect methods which convert the time difference into some other physical property that can be handled with low 244 speed detection, such as second-harmonic generation (SHG) based up-conversion PL technique and modulation spectroscopy based TRPL. An introduction to the above noted time-resolved optical spectroscopy techniques can be found in Ref 6.30 and Ref 6.31. In our setup, TRPL is implemented using a direct detection method -- TC/SPC, which is chosen due to two considerations imposed by our anticipated study on hybrid epitaxial/colloid nanostructures. First, the epitaxial QW and SAQDs (especially near surface quantum structure) typically have a PL decay time ~10- 1000ps, and the colloidal nanocrystals typically have a PL decay time ~10-1000ns. Therefore the TRPL to cover the time range for both of these two types of nanostructures needs a dynamic range of 5 orders of magnitude, which is impossible to achieve using methods other than TC/SPC. Second, TC/SPC is most suited to measure the anticipated weak PL signal from monolayer coverage of NCQDs and near surface quantum structures. In TC/SPC every photon of PL emission can be utilized in the measurement (subject only to the quantum efficiency of the detector cathode), with no additional conversion steps. As shown in schematic figure 6.9, a cavity-dumped mode-locked fs/ps Ti:S laser (Coherent Mira 700-980nm) pumped by an Ar + laser (Coherent Innova 310 ML-Visible) is used to generate fs/ps pulsed laser excitation. The 4% intensity of the laser pulse is split by a beam sampler (Newport) onto a homemade fast photodiode based trigger (>1GHz) to turn the timing of the excitation light pulses into electric pulses. The rest of the pulsed laser intensity is steered by mirrors (protected Silver 245 coating), attenuated by ND filters (if necessary) and focused using a lens (L1, f=200mm) onto the sample mounted inside a continuous flow cryostat (Janis ST-B, 5.5K-360K). The excitation spot on the sample is ~300 μm, as calculated using the divergence of the laser beam (1.5mrad) and the focal length of L1 (200mm). The photoluminescence generated from the sample is collected by a 3” lens (L2, f=79.9mm). For time-resolved PL, the photoluminescence is steered by a flipping mirror into the entrance slit of a subtractive monochromator (CVI DK242). Motorized filter wheel (CVI) is placed before the entrance of the monochromator to attenuate the PL intensity, if necessary. Spectrally dispersed PL photon exits the monochromator, hits the fast-timing Microchannel Plate Photomultiplier Tube (MCP-PMT), and generates electric pulses corresponding to the PL timing. Pulse processing electronics utilizing constant-fraction differential discrimination (CFD) and time-to-pulse-amplitude conversion (TAC) technique are used to measure and record the time difference between the electric pulses from the trigger (corresponding to the excitation timing) and those from MCP-PMT (corresponding to PL timing) to generate the TRPL spectrum. The pulse processing electronics has a time-resolution ~40ps (before data processing) which is the limitation of the time resolution of whole TRPL setup. As this is the first implementation of TRPL in our group, below we provide the details on its component instrumentation. 246 §6.2.3 TRPL instrumentation: Optics Laser System for Excitation: The light source for the TRPL setup is a cavity dumped fs/ps Ti:S laser consisting of a Coherent Mira 900D fs/ps laser with Mira PulseSwitch cavity dumper (Mira900D and PulseSwitch together are hereafter referred to as Mira), pumped by Ar + laser (Coherent Innova 310 ML-Vis). Mira 900D is a passive mode-locked Ti:S laser, capable of generating fs (200fs) or ps (<3ps) pulsed light at repetition rate 76MHz and maximum average power >1.5W (depending on wavelength and pumping power). The Mira900D exploits the Kerr lens effect of the Ti:S crystal which narrows the beam diameter of fs/ps high-energy pulsed lasing mode. A slit is strategically placed in the laser cavity to select the narrowed pulsed lasing mode and suppress the cw mode. The detailed mechanism of mode-locking laser is beyond the scope of this dissertation can be found in [6.32]. The repetition rate of the Mira900D pulses by itself is 76MHz which is too fast for our desired TC/SPC application. Therefore we modify the Mira900 laser with the Mira PulseSwitch (a cavity dumper) which employs an acoustooptic Brag cell to select desired pulses out of the pulse train of Mira900D for experiment use. Note that unlike standard extracavity dumper (such as Coherent Mira pulse picker), the PulseSwitch is an intracavity dumper that constitutes a part of the Mira laser cavity and requires a major reconfiguration of the Mira900D. However, PulseSwitch offers about 10 times higher peak pulse power compared to extracavity dumper. For the PulseSwitch in use, we have observed an inconvenience that the lateral position of 247 the slit assembly in the PulseSwitch has severe drift over time (typically takes over an hour to stabilize), which works against achieving stable mode-locked operation. Lubrication of the contact between the slit assembly and the PulseSwitch case seems helpful to lessen the drift. When operating in the 900-980nm range (especially ~950nm), the Mira must be purged with dry N 2 , since in this wavelength regime there is strong absorption by the water vapor in air. When the laser pulse from the Mira output (200fs or 3ps for fs/ps operation) propagate through the optical system, the pulse width are broadened as a result of transmission through lens or windows or reflecting off mirrors that changes the relative amplitude or phase of different wavelength components in the laser pulse. Such pulse broadening is an important consideration in femto-second spectroscopy but is insignificant given our intended time resolution. Estimated from the number of mirrors (protected silver coating) and the thickness of glass (BK7) lens or window used in the setup, the pulse broadening is < 2ps. A coherent Innova 310 Ar + laser (multiline-visible) is used to pump the Mira. For Mira to achieve stable modelocking (pulsed operation), the ability to tightly focus the pumping laser in the Ti:S crystal to achieve the highest power density is the single most important factor. Therefore a low divergence output coupler must be used for the Innova 310. The Ar + laser also must be forced into TEM 00 transverse mode by reducing its intracavity aperture at the cost of a reduced output power. A 5 to 8W of Ar + laser output in TEM 00 is necessary for stable modelocking of the Mira, depending on the desired output wavelength and repetition rate. 248 Subtractive Monochromator: When a light pulse is diffracted by a grating in a monochromator, besides spectral dispersion, the light pulse will be broadened (temporal dispersion) due to the variation in the path length among the rays that hit different parts of the grating. Therefore, for pico-second resolution TRPL measurements, subtractive monochromator is needed. A subtractive monochromator combines two single monochromators whose gratings rotate in opposite directions so that their temporal dispersions cancel each other. In our TRPL setup a 1/4m subtractive monochromator (CVI DK242) is used. Two pairs of gratings are installed in the monochromator (600g/mm, blazing at 500nm, and 600g/mm blazing at 1100nm) to cover, respectively, visible and NIR regimes. The resolution of the subtractive DK242 is 3.2nm at 1mm slit width. The temporal dispersion of the monochromator is <500fs. Trigger: A trigger based on a fast Si photodiode (Hamamatsu 5973, 1GHz) was built to record the timing of the excitation pulse. The circuit is shown in figure 6.10. In building this circuit, the photodiode leads and the stripped conductors of the coaxial cable should be kept as short as possible to avoid distortion of the radio frequency signal. A chip capacitor should be used as the bypass capacitor to avoid lead. MCP-PMT: To achieve the desired 10ps time resolution to measure the PL rise and decay time of the epitaxial nanostructures, a MCP-PMT (Hamamatsu R3809-U, 25ps T.T.S., 400-1200nm) is used as the detector. In MCP-PMT, unlike conventional PMT, the secondary electron multiplier is a micro-channel plate -- an 249 array of millions of inner metal coated glass capillaries (~10 μm inner diameter, <1mm long) which lead to very small temporal spread of electrons passing through and amplified by the MCP. The temporal spread of electrons is referred to as the transit time spread (T.T.S) of the MCP. For the R3809-U in our use, the T.T.S is ~25ps. The MCP-PMT is installed in a water assisted thermo-electrically (TE) cooled housing (-30 C, Hamamatsu C4878). To be effective in the desired NIR regime, the chosen MCP-PMT uses an Ag-O-Cs (S1) cathode. The S1 cathode has a very low quantum efficiency (~0.2% at 800nm, 0.001% at 1100nm) and high dark counts (~5000c/s at -30 C). High dark counts make the device vulnerable to mishandling. Dark counts of the S1 cathode itself at room temperature will damage the MCP if high voltage (HV) is applied! Pulsed light Photodiode Hamamatsu 5973 A K Resistor 10k Ω Chip Capacitor 0.1 μF Battery 9V - + 50 Ω Coaxial Cable Cable Outer Conductor BNC plug 50 Ω terminated BNC Receptor Pulsed light Photodiode Hamamatsu 5973 A K Resistor 10k Ω Chip Capacitor 0.1 μF Battery 9V - + 50 Ω Coaxial Cable Cable Outer Conductor BNC plug 50 Ω terminated BNC Receptor Figure 6.10 Trigger circuit based on Hamamatsu 5973 fast photodiode 250 §6.2.4 TRPL instrumentation: Time correlated single photon counting (TC/SPC) electronics The electric pulses from the MCP-PMT designating the luminescence timing and those from Si diode trigger designating the excitation pulse timing are fed into the TC/SPC electronics to generate the TRPL spectrum. The block diagram of our TC/SPC electronics using the inverted photon counting mode (as detailed below) is shown in figure 6.11. Precise Timing of Pulse – Constant Fraction Discriminator (CFD): The output pulses from MCP-PMT are amplified by a 1GHz preamplifier (40dB, Ortec 9036) and attenuated by a (15dB) attenuator so that they are in the appropriate amplitude range to be fed into the timing-discriminator (Ortec 9037). The anode output from a MCP-PMT has a narrow width (~250ps) but its amplitude varies over large range (~2 orders of magnitude) due to the fluctuations in the electron multiplication. Therefore the conventional discriminator based on a fixed threshold level cannot precisely measure the pulse arrival time. Instead, the discriminator needs to time the arrival of the pulse using a dynamic level proportional to the pulse amplitude. This is done using a constant fraction discriminator (CFD). In CFD, the input pulse is split into two pulses of equal amplitude. The first pulse is inverted and delayed for a preset period of time. The second pulse is attenuated and added to the first one, generating a bipolar pulse with zero-crossing that occurs at the same fraction of the input pulse independent of its amplitude. Based on the zero-crossing, CFD sends a NIM-standard logic pulse representing the precise timing of the MCP- 251 PMT output to the time-to-pulse-amplitude converter (TAC) to serve as the “start”. The uncertainty in the pulse timing determined by the CFD, often referred to as “time walk”, for pulses of 1:10 amplitude range is less than 25ps for the CFD used (Ortec 9037) in use. Similarly the output pulse from the Si PD trigger is inverted by an inversing transformer (Philips Scientific 460) and then processed by another CFD to generate another NIM pulse. The NIM pulse is delayed by a delay unit (Ortec 425A or 416A) and is sent to TAC as the “stop”. Delay ORTEC 425A or ORTEC 416A 1GHz Pre-Amp Ortec 9306 Time-to-Amplitude Converter (Ortec 457) Start NIN Stop NIM Pico-Timing Discriminator Ortec 9307 Inverting Transformer Phillips460 15dB Attenuator 665-15-1 MCP-PMT Hamamatsu R3908-U Monochromator DK242 sample Pico-Timing Discriminator Ortec 9307 Fast Si-Diode based Trigger MCA Ortec trump 8k PC & Control Program Counter Oriel 76915 TTL Mira 900D + Pulse Switch Pumped by Ar + Delay ORTEC 425A or ORTEC 416A 1GHz Pre-Amp Ortec 9306 Time-to-Amplitude Converter (Ortec 457) Start NIN Stop NIM Pico-Timing Discriminator Ortec 9307 Inverting Transformer Phillips460 15dB Attenuator 665-15-1 MCP-PMT Hamamatsu R3908-U Monochromator DK242 sample Pico-Timing Discriminator Ortec 9307 Fast Si-Diode based Trigger MCA Ortec trump 8k PC & Control Program Counter Oriel 76915 TTL Mira 900D + Pulse Switch Pumped by Ar + Figure 6.11 The block diagram of the TC/SPC electronics using the inverted photon counting mode. 252 Time-to-Amplitude Conversion: The TAC (Ortec 457) measures the time interval between the “start” and “stop” NIM pulses, and generates an analogous output whose voltage is proportional to the time interval, namely the time difference between the excitation photon and PL photon (to be precise the first PL photon counted by the MCP-PMT, see below). The TAC output is then fed into a multi- channel analyzer (MCA, Ortec Trump 8K). Upon a large number of such measurements the MCA accumulates a histogram of the TAC output voltages that present the probability distribution of photon emission in various time windows after the excitation, namely the TRPL spectrum. Photon Counting Rate and Inverted Timing: In practice the TAC measures first accepted “start” pulse and the immediate next “stop” pulse and ignores any additional “start” or “stop”. Therefore, if the PMT detects multiple photons after each excitation light pulse, the TAC will automatically take the timing of the first detected photon and bias the TRPL spectrum toward shorter time. To avoid this systematic error, one should attenuate the excitation pulse intensity or the luminescence to ensure the probability of detecting more than one photon after each excitation pulse is negligible. In practice we restrict the PMT counting rate to <1% of the laser repetition rate. This restriction necessitates a practice which at first glance seems abnormal -- the so called delayed coincidence scheme or inverted photon counting mode. Normally one expects the excitation pulse to be the “start” and the luminescence photon detecting by the PMT to be the “stop”. However this will lead to the majority of “start” having no corresponding “stop”, which will generate 253 excessive dead time in the TAC and inefficient use of MCA. Therefore instead we invert the “start” and “stop”: use the luminescence photon detection timing as the “start”, while the excitation pulse timing is delayed and used as the “stop”. In this way each “start” always corresponds to a “stop”. Of course, the delay of the “stop” needs to be long enough to account for the time of light propagation, all instrument delay, and the desired time range of TRPL measurement. In summary, the timing configuration of the inverted photon counting mode is shown in figure 6.12. Laser output Trigger output PL intensity Averaged PMT output Delayed Trigger output Single Photon Detection Event NIM Start NIM Stop Controlled Delay Δt Time-to-Amplitude Converter Output t A Δ ∝ No Photon Detected Multichannel Analyzer Oupt MCA Counts PL Intensity Channel Number ( Δt) Accumulate Laser output Trigger output PL intensity Averaged PMT output Delayed Trigger output Single Photon Detection Event NIM Start NIM Stop Controlled Delay Δt Time-to-Amplitude Converter Output t A Δ ∝ No Photon Detected Multichannel Analyzer Oupt MCA Counts PL Intensity Channel Number ( Δt) Accumulate Figure 6.12 The timing configuration of TC/SPC in the inverted photon counting mode as being implemented in the electronics shown in figure 6.11. 254 Since the TC/SPC electronic works at as high as 1GHz frequency, it goes without saying that all cautions in constructing a UHF instrumentation need to be observed, such as the proper choice and termination of cables, and the attenuation and distortion of signal as a function of cable length. §6.2.5 Time Resolution and Instrument Response Function Under the assumption that the TRPL measurement is a linear time-invariant system, the measured TRPL curve is a convolution of instrument decided impulse function and undistorted time dependent photoluminescence response from the sample: ∫ ′ ′ − ′ = t t d t t g t h t f 0 ) ( ) ( ) ( (6.3) where ) (t f is the measured TRPL curve, ) (t g is the undistorted TRPL response from the sample, ) (t h is the instrument impulse response (response from a δ-function input). In the case of TRPL, and ) (t h is a measure of the detection distorted intensity-time profile of the excitation pulse, which is referred to as instrument response function (IRF). The IRF is decided by the accumulated effect of uncertainties in timing from all instruments (for numbers see description in §6.2.3 and §6.2.4): the width of the excitation laser pulse, the temporal dispersion during pulsed light propagation (including in the monochromator), the transit time spread of the MCP-PMT and the various uncertainties in the TC/SPC electronics. Among these, the two dominant 255 ones are the T.T.S of the MCP-PMT (~25ps) and the “time walk” of CFD (~25ps), from which we expect the FWHM of IRF to be ~40ps. Operationally the IRF is measured by focusing attenuated excitation laser on the copper sample holder and measuring the scattered laser light time distribution using exactly the same procedure and instrument configuration in which the TRPL measurements will be perform. Figure 6.13 is an example of a measured IRF. The FWHM is 27ps, better than expectation. The rise time (10%-90%) of IRF is ~11ps and the decay time (10%-90%) is ~38ps. The FWHM of IRF at zeroth order decides the time-resolution of the TRPL setup. However, since in Equ. (6.3) we have the Figure 6.13 Measured instrument response function (IRF) of the TRPL setup. The FWHM, rise time and decay time of the IRF are, respectively, 27, 11 and 38ps. 256 privilege to measure the IRF ) (t h as well as the convoluted TRPL curve ) (t f under the same condition, reversing Equ. (6.3), we can deconvolute to reveal the undistorted TRPL response from the sample, ) (t g . If the data is of reasonable signal- to-noise ratio (SNR), often time-resolution as small as 1/4 of the FWHM of the IRF can be achieved. Below we briefly discuss the data processing and deconvolution in §6.2.6. §6.2.6 TRPL data processing and deconvolution In equ. (6.3), ∫ ′ ′ − ′ = t t d t t g t h t f 0 ) ( ) ( ) ( , (6.3) knowing the convoluted TRPL data ) (t f and the IRF ) (t h , how to recover the TRPL response ) (t g is a typical problem of deconvolution in data or imaging processing. Many methods, from the simple Fourier transform based deconvolution [6.33] to sophisticated algorithms based on maximum entropy analysis [6.34, 6.35] or fast global fitting algorithm [6.36], are available to deal with such problems of different levels of demands. For TRPL data deconvolution, the simple Fourier transform based algorithms implemented in most common data analysis software (such as Microcal Origin) is not suitable, since this class of algorithms takes no account of the noise or give too many constraining parameters to fiddle with that leads to somewhat arbitrary results. 257 For TRPL data deconvolution, since most of the time other than knowing ) (t f and ) (t h , we can also guess the functional form of the undistorted TRPL response ) (t g (typically some form of exponential function), a robust but relatively slow method to deal with such deconvolution problem is the least square iterative reconvolution method [6.37, 6.38]. Briefly, it is performed as follows: If the undistorted TRPL response ) (t g has a known functional form ) ... ; ( 1 n a a t G in which n a a a ... , 2 1 are the parameters to be determined, we can guess a set of fitting parameters, and reconvolute ) ... , ; ( 2 1 n a a a t G with the IRF ) (t h which leads to a fitting function: ∫ ′ ′ − ′ = t n n t d a a a t t G t h a a a t y 0 2 1 2 1 ) ... , ; ( ) ( ) ... , ; ( (6.4) Then the fitting is done by adjusting n a a a ... , 2 1 to minimize the weighted residue between the fitting function and the measured TRPL curve ) ( i t f , ( } { i t , i=1..m are the m th time points at which the PL intensities are measured): { } ∑ − = i i n i i t f a a a t y 2 2 1 2 2 )] ( ) ... , , ( )[ / 1 ( σ χ (6.5) where 2 χ is the residue, and i σ is the expected error at i th data point. Given the dominance of the shot noise due to Poisson statistics of the photon-counting (see discussion in §3.3.3(b)), i σ is taken as the square root of the measured PL intensity at each data point. The minimization of 2 χ is performed using Levenberg-Marquardt 258 nonlinear optimization [6.32] to obtain the best set of fitting parameters n a a a ... , 2 1 and the deconvoluted TRPL response from the sample ) ... , ; ( 2 1 n a a a t G . Following the least square iterative reconvolution method above, a custom program was developed to deconvolute our TRPL data. An example of its application is shown in figure 6.14. The black dots represent the measured TRPL response from a sample of MBE grown 2.0ML InAs SAQD with GaAs capping (RG970221-01-H). The dotted line is the corresponding IRF (FWHM 89ps). (Note here the IRF has a larger FWHM compared to that shown in figure 6.13 because the TRPL setup measures a larger time range.) The TRPL response of the SAQDs is modeled using a three level system involving electron relaxation and HOMO-LUMO exciton recombination [6.40] which leads to the following fitting function: ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎝ ⎛ − − × ⎥ ⎦ ⎤ ⎢ ⎣ ⎡ ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎝ ⎛ − − − × + = Decay Rise t t t t t t A C y 0 0 exp exp 1 (6.6) in which Rise t and Decay t are the rise and decay times of the PL, respectively reflecting the relaxation of electron to the LUMO level and the radiative recombination of the HOMO-LUMO exciton [6.40]. The iterative reconvolution method leads to the deconvoluted TRPL response (green, figure 6.14) of rise and decay time 87ps and 705ps, respectively. The red curve in figure 6.14 is the reconvoluted fitting to the measured data. As a concluding remark on data fitting, we note the least square iterative reconvolution is a robust method for data fitting, especially when distortion exists in 259 Measured TRPL Spectrum Reconvoluted Fitting IRF FWHM 89ps Deconvoluted Rise Time t Rise =87ps Decay Time: t Decay =705ps ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎝ ⎛ − − × ⎥ ⎦ ⎤ ⎢ ⎣ ⎡ ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎝ ⎛ − − − × + = Decay Rise t t t t t t A C y 0 0 exp exp 1 Fitting function: Measured TRPL Spectrum Reconvoluted Fitting IRF FWHM 89ps Deconvoluted Rise Time t Rise =87ps Decay Time: t Decay =705ps ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎝ ⎛ − − × ⎥ ⎦ ⎤ ⎢ ⎣ ⎡ ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎝ ⎛ − − − × + = Decay Rise t t t t t t A C y 0 0 exp exp 1 Fitting function: Figure 6.14 Example of TRPL data deconvolution. Black dots show the measured TRPL response from a 2.0ML InAs SAQD sample (RG970221-01-H) at 6K. Dotted line shows the IRF of the measurement (89ps FWHM). Green and red curves show, respectively, the deconvoluted TRPL response and the reconvoluted fitting to the measured data. The deconvoluted PL response of the 2.0ML InAs SAQD has a rise time of 87ps and a decay time of 705ps. the original data or data is of low SNR [6.38]. The shortcomings, however, are that this method requires knowledge of the functional form of the measured spectrum and requires relatively large computational power. For applications in which a large amount of data deconvolution is required, such as in fluorescence lifetime imaging (FLIM), a fast algorithm is definitely required (such as global fitting algorithm [6.36]). 260 §6.2.7 TRPL Future Upgrade The current the TRPL setup uses a fs/ps mode-locked laser for excitation, MCP-PMT and TC/SPC electronics for detection which limits the time resolution of setup to be ~10ps even after IRF deconvolution. Therefore the current TRPL setup is mostly suitable only to monitor HOMO-LUMO transition dynamics, but not the carrier relaxation process. One of the directions for TRPL future upgrade is to improve its resolution to subpicosecond to study the carrier relaxation dynamics important for the excitation transfer between nanocrystals and substrates. This can be achieved using either the up-conversion based PL technique or exploring the latest development in streak cameras. Indeed with the latest ultrahigh resolution streak camera (e.g. Hamamatsu C6860, C6138), a resolution up to a few hundred fs can be achieved. The shortcoming though is the very limited maximum time range (~1ns). §6.3 Excitation transfer in NCQD/Semiconductor Hybrid structures After having discussed the background knowledge of photoluminescence process in semiconductor nanocrystal quantum dots and the PL instrumentation, we now turn to the actual study of the excitation transfer involved in hybrid structures made of nanocrystals and crystalline semiconductor substrates. Towards this end, given the importance of the GaAs/InAs based III-V epitaxial structures in optoelectronics, as well as the importance of 1.1-1.5 μm wavelength regime provided by this material system to the future of biological detection (see the details argued in chapter 3), the canonical nanocrystal system to be examined is the InAs/ZnSe based core-shell quantum dots in direct contact with a GaAs crystalline semiconductor 261 substrate (figure 6.15) and the potential of control on energy/charge exchange between the two. In this section we thus first briefly review the general problem of excitation transfer between the two bodies and then discuss in detail our findings for the GaAs(substrate)/(InAs/ZnSe) NCQD hybrid system. A non-conventional example of “hot” excitation transfer from the excited states of the InAs/ZnSe NCQDs into the GaAs substrate is revealed in these studies. (a) (a) ~ ~ Vacuum HOMO Substrate CB VB ? ? LUMO E g (substrate) NCQD e V Δ h V Δ ~ ~ Χ sub Χ NCQD (b) ~ ~ Vacuum HOMO Substrate CB VB ? ? LUMO E g (substrate) NCQD e V Δ h V Δ ~ ~ Χ sub Χ NCQD (b) Figure 6.15: (a) Schematic of InAs/ZnSe core/shell nanocrystal quantum dots in direct contact with a GaAs substrate. (b) Shows the unknown nature of the relative alignment ( e V Δ and h V Δ ) of the electron and hole energies of the NCQDs and the substrate band edges. A common vacuum level and the simple electron-affinity (or photothreshold) rule from semiconductor heterojunctions is used to draw figure 6.16(b). Χ sub and Χ NCQD stand for, respectively, the electron-affinity of the substrate and the NCQD. 262 §6.3.1 Overview of the excitation transfer process. From a historical prospective, two vast classes of past studies are closely related to and beneficial to the understanding of the excitation transfer in the colloidal/epitaxial hybrid structures examined for the first time as part of this dissertation. These two classes are: (1) Energy/charge transfer between adsorbed molecules and solid substrates with varying types and degrees of coupling between the two [6.41, 6.42]. The reported studies conceptually most relevant to the combination of nanocrystal QDs on crystalline semiconductor substrates of focus here are (a) organic dyes on crystalline semiconductor with an intervening inert atom layer, such as pyrene/Xe/Si(111) [6.43], (b) donor/acceptor molecules on metals or semiconductors such as ruthenium hexamine/functionalized alkanethiol SAMs/Au(111) [6.44] or acceptor on GaAs surface quantum well [6.45], and (c) nanocrystals on polycrystalline metals, such as CdS/organic monolayers/Au [6.46]. These investigations laid down the frame work of some transfer mechanisms that we may encounter in the colloidal/epitaxial hybrid structures. (2) Energy/charge transfer processes between nanocrystals and the solvent environment [6.47], between nanocrystals and surface adsorbed molecules [6.48, 6.6] or between nanocrystals and a surrounding conducting or semiconducting polymeric matrix [6.49, 6.50]. Such energy/charge transfer processes are extensively studied in the context of nanocrystal based inorganic/organic optoelectronic devices such as light-emitting diodes [6.51] or photovoltaic devices [6.52]. In all such studies, the 263 “band-offsets” have been treated as parameters with unknown values that are approximated as following the electron affinity rule. We note that, as some of the studies have shown, charge transfer between nanocrystals and adsorbed molecules can happen on very fast time scale of ~200fs [6.48, 6.6]. If a comparable transfer can be achieved between the adsorbed nanocrystals and the underlying substrate, certainly an efficient communication link between the two can be achieved. Regardless of the specific system involved in the above noted studies in which transfer occurs, quantum mechanically, the energy/charge transfer process always involves the coupling between an initial state and a final state (of the entire system in consideration) by an interaction Hamiltonian. The rate of transfer, to a good approximation, is decided by Fermi’s golden rule: ( ) i f i f tr E E H w − Φ Φ = δ π 2 int 2 h (6.7) where tr w is the transition probability, f Φ and i Φ are the wavefunctions of the final and initial state, int H is the interaction Hamiltonian, f E and i E are the energies of the initial and the final state. Different types of initial and final states, as well as interactions between them classify the transfer mechanism. Specifically for the energy/charge transfer involving nanocrystals in which electronic states are localized due to the confining potential, the dominant transfer mechanisms are: 264 (1) dipole-dipole coupling mediated Förster resonant energy transfer (FRET), and, (2) tunneling process. Below we briefly summarize the characteristics of the two transfer mechanisms. §6.3.1(a) Förster Resonant Energy Transfer (FRET): FRET involves a process in which an exciton excited in one object (the donor) recombines and creates another exciton of the same energy in a “nearby” object (the acceptor) without emission or absorption of a real photon. The transfer is the strongest if the excitonic transitions in both the donor and the acceptor are an electric dipole transition, in which case the interaction is mediated by the relatively long range (compared to quadrapole, octapole) dipole-dipole interaction. Therefore specifically for FRET, the initial state is 0 A E D Φ Φ (donor in excited state E D Φ , acceptor in ground stats 0 A Φ ), and final state is E A D Φ Φ 0 (donor in ground state 0 A Φ , acceptor in excited state E D Φ ), the dipole-dipole interaction Hamiltonian dd H is: [ ] [ ] {} 2 3 / ) ( ) ( 3 ) ( ) ( 1 AD AD AD AD dd R R A p R D p A p D p R H r r r r r r ⋅ ⋅ − ⋅ = ε (6.8) where ) (D p r and ) (A p r are, respectively, the electric dipoles of the donor and the acceptor, AD R r is a vector pointing from the donor to the acceptor. ε is the dielectric constant of the environment. 265 Following the above general formalism, the FRET transfer rates in different cases can be derived, such as between two localized donor and acceptor (e.g. between two dye molecules [6.53, 6.54]), from a localized donor to a delocalized acceptor (e.g. from a dye molecule to a semiconductor substrate [6.55]), or from a delocalized donor to a localized acceptor (e.g. from a quantum well to a dye molecule [6.56]). Without going into the derivation we note here the well known FRET rate between localized sensitizer and acceptor (such as between dye molecules or QDs) [6.53]: 6 0 1 ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ = r R w D FRET τ (6.9) where FRET w is the FRET rate; D τ is the sensitizer radiative recombination life time in absence of the acceptor; r is distance between the sensitizer and the acceptor; and 0 R is the FRET critical radius defined as: v d v v v F N n QE R S S A ~ ~ ) ~ ( ) ~ ( 128 ) 10 ln( 9000 0 4 4 5 2 6 0 ∫ ∞ ⋅ ⋅ ⋅ = ε π κ (6.10) where 0 R is in the unit of cm, QE is the sensitizer luminescence quantum efficiency in the absence of the acceptor, A N is the Avogadro’s number, and n is the refractive index of the environment. In the integrand, ) ~ (v F S is the normalized shape-function of luminescence spectrum of the sensitizer, and ) ~ (v S ε is the molar absorption coefficient of the acceptor in unit of (M -1 cm -1 ), and v ~ is the wavenumber in cm -1 . Finally κ is an orientational factor which equals to: 266 ( ) ( ) ˆ ) ( ˆ ˆ ) ( ˆ 3 ) ( ˆ ) ( ˆ AS AS R A R S A S ⋅ ⋅ − ⋅ = μ μ μ μ κ (6.11) where ) ( ˆ S μ and ) ( ˆ A μ are the unit vectors of the sensitizer and acceptor transition moments, and AS R ˆ is the unit vector between sensitizer and acceptor. The averaged κ value for sensitizers and acceptors of randomly oriented transition moments is 2/3 [6.53]. The typical FRET critical radius 0 R lies in the range of 1-10nm. The maximum FRET rate between two molecular dyes or two quantum dots is on the order of 10 8 -10 9 sec -1 (a calculation of FRET rate between nanocrystals of typical parameters is presented in §6.3.3). Therefore, given the typical rate of the processes involved in the photoluminescence in QDs (§6.1), FRET can typically only compete with the radiative recombination process but is too slow to compete with carrier (electron and hole) relaxation to their individual ground states. The FRET rate has no explicit temperature dependence; implicit temperature effect is through the temperature dependence of the QE and the shape of the emission/absorption spectra of the donor and the acceptor. §6.3.1(b) Tunneling: Other than the resonant energy transfer discussed above, tunneling is also an important mechanism for excitation transfer. The tunneling can be in the form of single carrier (electron or hole) transfer or in the form of exciton transfer. The tunneling process involves the coupling of the wavefunctions of the initial state and 267 the final state penetrating through a classically forbidden potential (barrier). The simplest single particle tunneling rate depends on the overlap of the wavefunction of the initial and final states: ( ) i f i f tr E E V w − Φ Φ = δ π 2 2 h (6.12) where V is the potential. The tunneling model has been applied to describe charge transport in photoexcited close packed 3D CdSe nanocrystals solid under an electrical field [6.57]. We note that single particle tunneling has to occur between two states resonant (equal) in energy and the tunneling rate has no explicit temperature dependence. Off-resonance tunneling has to be assisted by yet another particle (such as a phonon) in order to conserve energy, which is therefore slower than the on- resonance tunneling. In the charge transfer into or out of nanocrystal quantum dots involving tunneling, given the same tunneling barrier (same shell layer and surfactant thickness) for different core size NCQDs, the main factor that decides the transfer efficiency is the alignment of the relevant energy levels inside and outside the NCQDs. One such example is seen in the NCQD PL intermittence dynamics as discussed in §6.1.2(c). Bawendi et al used a “random walk” model which neatly explained the power law “off’ and “on” time probability distribution for NCQD PL intermittence [6.23]. This model assumes that the energy level in the QD and a hypothetical trap state at the QD surface, or in the surrounding, are fluctuating in energy in a 1D 268 random walk fashion; when the two energy states are in resonance, charge transfer occurs because of tunneling, which switches the PL of the NCQD “on” and “off”. Although the single particle tunneling is useful in describing charge transfer involving NCQDs, we note that such a description implies adiabatic approximation. This approximation is shown to be often not appropriate for the electron transfer process in molecular systems [6.58], between adsorbed molecule and metal or semiconductor surface [6.59], or in charge transfer involving nanocrystals [6.47]. Indeed the charge transfer rate in/out of nanocrystals can be quite comparable to the lattice vibration frequency, and thus the adiabatic approximation does not necessarily hold. The theoretical framework describing non-adiabatic charge transfer has been developed in the context of charge transfer in molecules [6.60] and also has been applied to deal with charge transfer involving nanocrystals [6.47]. The details of such theory can be found in [6.61] and is beyond the scope of this dissertation. However, this author notes that, for the description of charge transfer between nanocrystals and substrates, the non-adiabatic charge transfer theory can be an important perspective to be exploited in future research. §6.3.2 Sample Preparation and Optical Measurements in Excitation Transfer Studies §6.3.2(a) Sample Preparation The main system involved in our studies of excitation transfer between nanocrystals and substrates is the InAs/ZnSe core shell nanocrystals deposited on 269 sulfur-passivated GaAs (001) surface as schematically shown in figure 6.15. The InAs/ZnSe core/shell NCQDs were prepared using procedures described in detail in §2.3. The most probable InAs core diameter is ~5 nm and the range is from ~3nm to ~ 7nm. The ZnSe shell thickness is ~1.8ML. The room temperature PL of these InAs/ZnSe nanocrystals in toluene is shown in figure 6.16 which has the dominant peak at 1.02eV (1220nm). The GaAs substrates used are MBE grown semi-insulating (p-type background doping ~1x10 15 /cm 3 ) GaAs buffer layers supplied by my colleagues Dr. Eui-tae Kim, Ms. Yi Zhang, and Mr. Tetsuya Asano. The as-grown GaAs substrates are sulfur-passivated using the protocol describe in §4.1. Then the GaAs substrate is glued to a piece of InAs substrate using a thin layer of epoxy. The function of the InAs substrate is to prevent the excitation laser in the PL measurement from reaching the copper sample holder and causing a luminescence background which interferes with the emission from the InAs/ZnSe nanocrystals. The NCQDs were deposited onto the passivated GaAs surface using the dip-coating method introduced in §4.2. Typically InAs/ZnSe toluene solution (~0.5mg/ml) and substrate withdrawal speed 2 μm/sec was used for dip-coating, which resulted in a one monolayer high closely-packed layer of InAs/ZnSe nanocrystals on GaAs substrates. An illustrative AFM (tapping mode, RTESP tip, DI multimode AFM) image of InAs/ZnSe nanocrystals on GaAs is shown in figure 6.17. (Note the close packing of InAs/ZnSe nanocrystals.) The coverage and uniformity of NCs on the surface were measured from such AFM images. Typically, NCQD coverage 270 measured from randomly selected 5µmx5µm areas on a given sample showed variation less than 5%. PL of InAs/ZnSe in Toluene PL of InAs/ZnSe in Toluene Figure 6.16 Room temperature photoluminescence of InAs/ZnSe nanocrystals in toluene solution. Figure 6.17: Tapping mode AFM image of InAs/ZnSe NCQDs deposited on sulfur passivated GaAs (001) via dip-coating. 271 Along with InAs/ZnSe nanocrystals on GaAs samples described above, reference samples consisting of InAs/ZnSe nanocrystals dip-coated on glass coverslip (same protocol as above) are also always included in the PL measurements by placing the GaAs based sample and glass based reference sample on the sample holder. Since the bandgap of glass (SiO 2 ), ~8eV, is much larger than the InAs/ZnSe nanocrystals, we do not expect excitation transfer to occur between the two. After preparation, both InAs/ZnSe on GaAs samples and reference samples (InAs/ZnSe on glass) were transferred, without exposure to air, into a cryostat and pumped to ~2x10 -7 torr vacuum for the photoluminescence studies. Experiments showed that the samples can stay in the cryostat for a couple of weeks without degradation in the NCQD PL behavior. §6.3.2(b) Optical Measurements: The pre-existing PL setup described in §6.2.1 was used for the time- integrated PL and PLE spectroscopy measurements. The TRPL measurements were performed using the new PL/PLE/TRPL setup described in §6.2.2-§6.2.7. Unless otherwise stated, the excitation power densities used in the time-integrated PL/PLE measurement are 100W/cm 2 . At this low excitation power density, the exciton occupation in NCQDs is estimated to be less than 0.01 per NCQD. In addition, all PL/PLE data shown have their intensities normalized to the NCQD coverage of individual samples obtained from AFM measurements performed after the PL/PLE measurements. 272 §6.3.3 Inter-NCQD energy transfer Given the closely-packed distribution of InAs/ZnSe NCQDs on GaAs or glass substrate, other than the coupling between the NCQDs and the substrates, a significant coupling between the nearby NCQDs is also expected. The subject of inter-NCQD energy transfer or electron transport in 3-dimensionally-closely-packed CdSe nanocrystal solids has been examined [6.62, 6.57, 6.63]. It was shown that in the CdSe nanocrystal solids, without an externally applied electric field, the dominant coupling between nearby NCQDs is via the dipole-dipole interaction mediated resonant energy transfer (see §6.3.1(a)). Here we present the first evidence of inter-NCQD energy transfer in the near infrared InAs/ZnSe NCQD assembly which, different from the 3D CdSe NCQD solid noted above, is of sub one monolayer coverage on the substrate. The sample for this study is InAs/ZnSe nanocrystals dip-coated on glass cover-slip schematically show in figure 6.18. The sample preparation is described in Figure 6.18 Schematic showing sample structure employed for inter-dot transfer study. 273 §6.3.2(a). Glass substrate is chosen to avoid any ambiguity that might otherwise arise from communication with the substrate surface. Given the large ~8eV bandgap of glass, no excitation transfer between it and the adsorbed nanocrystals is expected. The room temperature PL spectrum of the InAs/ZnSe NCQDs on glass is shown in figure 6.19(a). In figure 6.19(b) is shown the room temperature PL decay detected at 1.02eV/1220nm (the PL peak) and a higher energy 1.127eV/1100nm which correspond to emission from, respectively, larger (~5nm diameter) and smaller (~4nm diameter) nanocrystals. The faster decay time for the smaller NCQDs (1.82ns) compared to for the larger ones (3.58ns) indicates excitation transfer from smaller to the larger dots. Further clearer evidence for inter-NCQD FRET was obtained from the temperature dependence of the decay time measured at the above two detection energies and shown in figure 6.20. Note that the decay times of the larger and the smaller NCQDs in the ensemble have dramatically different temperature dependence. The decay time of the larger dots decreases by an order of magnitude when the temperature increases from 79K to 297K. By contrast, that of the smaller dots remains nearly independent of the temperature (1.82ns at 297K and 3.56ns at 79K). The decrease in the decay time with increasing temperature is expected in nanocrystals given two contributing factors: (1) the enhancement in non-radiative recombination rate, and (2) the impact of the fine structure of their electronic states. Indeed due to the non-spherical shape or the crystal field in the nanocrystals, the bandedge exciton splits into a fine structure in which the exciton state of lowest 274 1.13eV (1100nm) (a) 1.13eV (1100nm) (a) Detection: 1.01eV(1220nm) Detection: 1.13eV(1100nm) t=1.82ns t=3.58ns 297K (b) Detection: 1.01eV(1220nm) Detection: 1.13eV(1100nm) t=1.82ns t=3.58ns 297K Detection: 1.01eV(1220nm) Detection: 1.13eV(1100nm) t=1.82ns t=3.58ns 297K (b) Figure 6.19 (a) Room temperature PL of InAs/ZnSe nanocrystals dip-coated on glass substrate. (b) Time resolved PL spectra detected at two energies 1.02 eV(black) and 1.13eV(red), corresponding to emission, respectively, from larger (~5nm diameter) and smaller (~4nm diameter) InAs/ZnSe nanocrystals. The single exponential fitted luminescence lifetimes are 3.58ns and 1.82ns, respectively. 275 Det.1.13eV(1100nm) Det.1.02eV(1220nm) Estimated for 1100nm emission NCQDs, as detailed in the text. Det.1.13eV(1100nm) Det.1.02eV(1220nm) Estimated for 1100nm emission NCQDs, as detailed in the text. Figure 6.20 Measured temperature dependence of the luminescence lifetime detected at 1.02eV (black square) and 1.13eV (red square). Black Crosses show the luminescence lifetime of the 1.13eV emission nanocrystals estimated using the calculated FRET time constant and the assumption that non-FRET related recombination time of the smaller (1.13eV emission) nanocrystals is the same as the lager (1.02eV emission) nanocrystals. (See text for details). This estimation fits reasonably with the observed luminescence lifetime of the 1.13eV emission nanocrystals (red square). energy is forbidden from dipole-transition (“dark” exciton) [6.10, 6.11]. With increasing temperature, a larger fraction of excitons are thermally activated from the lowest energy “dark” state to the higher dipole-transition-allowed “bright” states, therefore a decrease in luminescence lifetime is expected. Such theoretical expectation was experimentally confirmed in TRPL of CdSe nanocrystals [6.64 45]. 276 Clearly the decay time of the smaller nanocrystals, if they were isolated, should follow the same trend of temperature dependence as that of the larger ones. The near constant decay time with varying temperature observed here in the smaller nanocrystals thus indicates that their decay time, instead of reflecting their intra-dot recombination, is limited by the resonant energy transfer to the nearby larger NCQDs. Such resonant energy transfer has no temperature dependence. Using Equ (6.9)-(6.11) and the following assumed parameters we estimate the FRET time constant between a smaller (1.13eV/1100nm emission) nanocrystal and a neighbouring average size nanocrystal (1.02eV/1220nm emission): (1) the smaller NCQDs have a δ-function emission at 1.13eV/1100nm; (2) the molar absorption coefficient of average size nanocrystals at 1100nm is 2.5x10 5 M -1 cm -1 (optical density 1.0 at 1mg/ml concentration see [6.65 67]). (3) the nanocrystals are randomly orientated thus 3 / 2 = κ ; (4) the dielectric constant of the environment is 1.25 (the average dielectric constant of air and glass); (5) the QE of the nanocrystals is the same as in solution: ~20% at room temperature. (6) the center-to-center distance of the nearby QDs is 8nm (accounting for InAs core diameter 5nm, ZnSe shell thickness 2ML, and the surfactant length 1nm). Inserting the above parameters into Equ (6.10) leads to a critical FRET radius: R FRET =10.2nm. The FRET time constant using Equ (6.9) is QE tot Rad FRET τ τ τ ⋅ = ⋅ = 23 . 0 23 . 0 , where Rad τ and tot τ are, respectively, the radiative recombination rate and the total recombination rate (radiative and nonradiative) of the nanocrystals without FRET. 277 Assuming at room temperature, without FRET, the smaller (1100nm emission) nanocrystals have a QE=20% and the total recombination time is 3.58ns, the same as the larger ones, we obtain an estimated FRET time constant ~4ns. Assuming that the non-FRET related recombination times of the smaller (1.13eV emission) nanocrystals are the same as the lager (1.02eV emission) nanocrystals at different temperatures, the non-FRET recombination time combined with the FRET time constant (4ns) gives an estimate of the decay time for the smaller (1.13eV/1100nm emission) nanocrystals (figure 6.20, crosses), which fits reasonably well with the measured decay time temperature dependence (figure 6.20) red squares). We note that the FRET coupling is quite significant in closely-packed InAs/ZnSe nanocrystals (as estimated above, 4 times faster than the radiative decay rate). This coupling smears out the PL spectrum of the nanocrystal ensemble, which presents an unavoidable difficulty in the optical studies of excitation transfer in the nanocrystal/substrate structures to be discussed. Compared to the conventional CdSe nanocrystals, the FRET coupling between the InAs nanocrystals is of longer range because of the longer wavelength involved (see Equ. 6.10). Therefore, in this respect, using InAs based NIR nanocrystals is certainly advantageous in applications such as FRET based molecular-ruler, or FRET based biological detection [6.66, 6.67]. Other than the energy transfer between InAs NCQDs, from a historical prospective, our group was the first one to discover the energy transfer between 278 vertically-paired, epitaxially grown, InAs self-assembled quantum dots (though the transfer is via a different mechanism--phonon assisted tunneling) [6.68 41]. One of the interesting future directions is thus to study FRET in a colloidal/epitaxial hybrid structures between the adsorbed NCQDs and near-surface self-assembled quantum dots in the underlying substrate, as schematically shown in figure 6.21. Given the typical FRET critical radius for InAs QDs ~10nm estimated above, the FRET coupling between the two types of quantum dot seems to be a viable way to establish Surface Undulation Driven by Strain Field of the SAQDs FRET Surface Undulation Driven by Strain Field of the SAQDs FRET Figure 6.21 Schematic showing a future direction of colloidal/epitaxial hybrid structure made of vertically aligned near surface SAQDs and adsorbed NCQDs. FRET coupling between the NCQDs and SAQDs can be exploited to establish communication link between the two. 279 communication between the two. The challenge, however, is how to vertically align the NCQDs and the SAQDs as shown in the schematic of figure 6.21. We note that the surface undulation of the capping layer driven by the strain field of the SAQDs demonstrated by our group [6.69] may be exploited to realize such vertical alignment between SAQDs and adsorbed NCQDs. §6.3.4 Excitation Transfer between NCQDs and Substrate §6.3.4(a) Observation of excitation transfer Having examined inter-NCQD energy transfer between monolayer thickness NCQDs adsorbed on a solid surface, we next turn to excitation transfer between the nanocrystals and the underlying crystalline semiconductor substrates. This is a field that has been barely investigated. However such transfer provides communication link between nanocrystals and substrates, and hence is at the center of our proposed application of colloidal/epitaxial hybrid structures for chip-based biological detection (figure 6.1). Recently, a study of the room temperature energy transfer in the UV- visible range between CdSe/ZnS core-shell NCQDs and InGaN quantum well buried below in a GaN substrate at extremely high (~100 μJ/cm 2 , 200fs duration pulse excitation, equivalent to ~500MW/cm 2 ) pumping powers has appeared [6.70, 6.71]. As authors claimed, efficient FRET occurs from the quantum well electron-hole pairs (not exciton) to the NCQDs at the very high excitation density (~500MW/cm 2 ) employed (carrier density ~10 13 /cm 2 in the quantum well), thus providing a means of indirectly injecting electron-hole pairs in nanocrystal quantum dots which otherwise, 280 owing to the presence of the insulating organic capping layer, is not possible through electrical injection. Below we discuss our investigation of excitation transfer from adsorbed InAs/ZnSe NCQDs to GaAs substrates (sample schematically shown in figure 6.15) at the opposite and i.e. extremely low excitation power densities (100W/cm 2 ), and via a very different mechanism than FRET. The simplest initial observation of the presence of such excitation transfer we found in the PL behavior of the adsorbed InAs/ZnSe NCQDs. Figure 6.22 summarizes the room temperature, low excitation power (100W/cm 2 ) PL spectra of InAs/ZnSe NCQDs deposited on passivated GaAs (red lines) and on glass (black lines), for excitation at 1.38eV/900nm (panel a) and at 1.46eV/850nm (panel b). At room temperature, the former is below and the latter above the GaAs band gap (1.42eV/872nm). Comparison of these spectra shows a richness of information which is fairly revealing of the nature of the components in the sample and the communication phenomena between these components. First, the PL responses from NCQDs on glass (figure 6.22, black lines) for excitation at either energies are, as expected, the same. They are also essentially the same as that from NCQDs in solution with a peak at 1.02eV/1220nm (figure 6.16). This suggests that the ZnSe shell and the TOP/TOPO surfactants of the InAs/ZnSe core-shell NCQDs are protecting the optical response quite well even away from the solution environment. Second, for excitation energy below the GaAs bandgap energy, the PL intensity of these NCQDs on GaAs is not too far below that on glass. This, once again, suggests the dominant role of the ZnSe shell and the surfactants in protecting the InAs core 281 Figure 6.22: Room temperature PL of NCQDs on different substrates for excitation at (a) below (1.38eV) and (b) above (1.476eV) GaAs band gap (1.42eV). In both panels, NCQDs on glass (black); NCQDs passivated GaAs (red). from being too adversely affected by the protocol that we have developed for the NCQD adsorption process. Third, and of greatest significance, for the NCQDs on GaAs substrate, the NCQD PL peak intensity at ~1.02eV/1220nm is seen to drop by a factor of ~5 as the excitation goes from below to above GaAs band gap. The occurrence of the NCQD PL intensity drop only on GaAs substrate at higher excitation energy (near GaAs bandgap) indicates there is onset of charge or energy transfer between the NCQDs and the GaAs substrate that significantly reduces the PL from the NCQDs. On On On On 282 §6.3.4(b) Onset of PL Drop and Electronic Structure of InAs/ZnSe on GaAs Following this encouraging discovery, we used systematic PLE studies to probe the “onset” excitation energy of this dramatic drop in the PL of NCQDs on GaAs. In figure 6.23 are shown the room temperature PLE spectra for detection at 1.02eV/1220nm, the peak of the NCQD PL distribution. Note that the PLE of NCQDs on glass shows only small variation in the excitation wavelength region of 1.24eV(1000nm)-1.46eV(850nm), manifesting the near constancy of the NCQD absorption coefficient in this regime. By contrast, the PLE intensity of the NCQDs on passivated GaAs samples is seen to exhibit a nearly step-like drop at ~1.36eV as the excitation energy changes from below to above the GaAs band gap. Hereafter we refer to the excitation energy at which the PL intensity of NCQD on the GaAs substrate starts to drop as the “onset” excitation energy of the PL drop, or simply “onset” energy. The nature of the “onset” excitation energy as measured by the PLE needs to be revealed by its relation to the relative positioning of the energy levels in the NCQDs and the substrate. As noted at the opening of this chapter and in conjugation with figure 6.15, this relative positioning, by extension of the terminology from the vast field of heterojunctions, we will call “band-offset” or “band edge discontinuity”. While for several bulk semiconductor heterojunctions these discontinuities have been well established, for hybrid NCQD and semiconductor substrates, with or without an intervening surfactant, these have never been examined before. Our studies presented here constitute the first set of data whose interpretation, for the first time, reveals the qualitative nature of the band-offset between any NCQD and a substrate. 283 On Glass Eg(GaAs) Det.@1.02eV E(1S e -2VB) On GaAs 1P e -1VB E(1S e -1VB) Onset On Glass Eg(GaAs) Det.@1.02eV E(1S e -2VB) On GaAs 1P e -1VB E(1S e -1VB) Onset Figure 6.23: Room temperature PLE of NCQDs deposited on: Glass (black), passivated GaAs (red), The detection is at 1.02eV (NCQD PL peak). Energies of the most relevant optical transitions in the NCQDs are marked on the PLE data for reference and ease of discussion in the text. Shown in figure 6.24 is the energy diagram of the InAs/ZnSe NCQD on GaAs substrate. The substrate band edges to the NCQD HOMO and LUMO band offsets chosen in drawing this figure, in fact, already accounts for a self-consistent analysis and interpretation of the various observations, as discussed in the following. The NCQD energy levels correspond to that of ~5.2nm core diameter at room temperature (RT) whose HOMO-LUMO emission is at 1.02eV/1220nm (the detection energy in the PLE). The NCQD energy levels are adapted from figure 2.12 calculated using 8 band p k r r ⋅ method but with the following two modifications: 284 (a) the InAs bandgap difference between 0K and RT is accounted for, and (b) the Coulomb energy between the electron and the hole are equally distributed to the electron and hole states. Surfactant ZnSe ZnSe InAs Core 5.2nm dia ~ ~ Vac 1Se LUMO 1Pe 1VB HOMO 2VB Eg(InAs)=0.36eV(RT) 1.02eV 0.18eV 0.44eV 1.22eV …… …… GaAs CB VB 1.42eV (RT) 1eV ?? ?? Surface States 1.13eV Surfactant ZnSe ZnSe InAs Core 5.2nm dia ~ ~ Vac 1Se LUMO 1Pe 1VB HOMO 2VB Eg(InAs)=0.36eV(RT) 1.02eV 0.18eV 0.44eV 1.22eV …… …… GaAs CB VB 1.42eV (RT) 1eV ?? ?? Surface States 1.13eV Figure 6.24 Energy level diagram of InAs/ZnSe NCQDs on GaAs substrate drawn for a relative positioning of the substrate band edges and the NCQD HOMO-LUMO energies consistent with the findings (see text for details). The question marks indicates the unknown nature of the hypothetical concept of defining a confinement potential for the entire system and in equivalent to the unknown nature of e V Δ and h V Δ defined in figure 6.15(b). 285 The validation of the original 8 band p k r r ⋅ calculation is obtained by comparison to the published results of optical transitions in the InAs nanocrystals (see figure 2.13). For the InAs/ZnSe NCQDs themselves, the band discontinuity between the InAs core and the epitaxial ZnSe shell is taken from [6.72]. The RT GaAs bandedge as well as surface band tailing is also shown in figure 6.24. The relative position of the GaAs bandedges and the energy levels in the InAs NCQDs however is uncertain due to (a) the unknown GaAs band bending related to the sulfur passivated surface and, (b) the unknown nature of the interface between the GaAs and the InAs/ZnSe NCQDs, such as the steady state charge transfer. Guided by the energy diagram (figure 6.24) we can examine the PLE data in figure 6.23 more quantitatively. The PLE peak of NCQDs on glass seen at ~1.42eV is the 1VB-1Pe transition (calculated value 1.46eV) in the NCQDs. With respect to the 1VB-1Pe transition energy, the “onset” of the PL drop of NCQDs on GaAs is seen to occur at ~1.36eV, which is ~60meV below. But with respect to the (2VB-1Se) transition (~1.20eV, as marked in figure 6.23) the “onset” excitation at ~1.36eV is 160meV in excess. Thus the photon absorption and optical transition involved in the NCQDs at the “onset” excitation energy has to be between the LUMO level (1Se) and the high excited hole states above 2VB where the hole states are quasi- continuous (see §6.1.2(b)). In addition, with respect to the GaAs bandgap at room temperature (1.42eV) the “onset” is also ~60meV below. Thus, when the PLE of the NCQDs on GaAs starts to drop, the bulk (i.e. band-to-band) transitions in GaAs are not yet photoexcited. 286 §6.3.4(c) Hole transfer from InAs/ZnSe NCQD to GaAs substrate. Given the above discussed relationship of the “onset” of PLE drop with respect to the optical transitions in the InAs/ZnSe NCQDs and the GaAs substrate, we realize that the straightforward explanation of the NCQD PL drop is the excitation (hole) transfer from the NCQDs to the substrate, moreover. Strikingly however, the excitation transfer must occur when the exciton consists of an electron at the LUMO level and the hole at excited states above 2VB. This is the first observation of such an excitation transfer from NCQDs involving “hot” carriers. It is remarkable because electron/hole relaxation itself in the NCQDs occurs on a 1 picosecond time scale (see discussion in §6.1.1(b)). This type of “hot carrier” transfer from NCQDs to the underlying substrate can thus open a new regime of studies aimed at transfer of weak sensing signals generated by NCQDs that have been appropriately surface-functionalized to act as sensors of chemical, biochemical, and/or biological agents. Further evidence for the fact that that excitation transfer occurs at high excited states is obtained by examining time-resolved PL behavior. Shown in figure 6.25 are the measured HOMO-LUMO PL decays (of NCQDs on GaAs) at two excitation energies 1.37eV(900nm) and 1.55eV(800nm) that are, respectively, below and above the GaAs bandgap (1.42eV at RT). We note that the detection for TRPL is set at 1.13eV/1100nm, off the PL peak, because at 1.02eV/1220nm the quantum efficiency of MCP-PMT detection is too low to allow a measurement. (When PL intensity (photon counts) is low, the TRPL curve shows a certain degree of 287 oscillation. This is an instrument artifact related to the dark count from the MCP- PMT). At 1.55eV/800nm excitation for which transfer occurs, the HOMO-LUMO transition decay time remains essentially the same as at 1.37eV/900nm excitation for which transfer has essentially not started (refer to PLE data in figure 6.23), although the absolute intensity at 1.55eV/800nm excitation is ~1/5 of that at 1.37eV/900nm excitation. Hence it is clear that the transfer has already happened before the carriers relax to HOMO-LUMO level. This further confirms the “hot” nature of the transfer. Figure 6.25: Room temperature TRPL measurement of NCQDs on passivated GaAs for excitation at 1.55eV (800nm black) and 1.37eV (900nm red) respectively above and below the GaAs bandgap. Detection is at 1.12eV (1100nm). 288 It is obvious that such “hot” transfer cannot be due to the dipole-dipole interaction mediated Förster resonant energy transfer (FRET §6.3.1(a)) of the whole exciton (electron and hole together). FRET occurs on the time scale of ~1-10ns, too slow to compete with the carrier relaxation. Reduction of the PL due to electron transfer is also not likely since the electrons are always at the LUMO level. The PL reduction due to electron transfer is not likely to has an onset occurs at excitation far above the NCQD HOMO-LUMO gap. Therefore the most probable mechanism for PL reduction at the observed excitations higher than the “onset” energy is still the hole transfer to the substrate, as depicted in figure 6.26. In this picture, after the excitation energy goes above the “onset” energy, a hole is created at the excited state (above ~2VB state) in the NCQD. During the hole relaxation to its ground state, when the energy of the hole is in resonance to some external “trap” states present due to the GaAs substrate, the hole tunnels into such trap states and the PL from the NCQD is dramatically reduced. However if the excitation energy is smaller than the “onset” energy, the hole will not be in resonance to the GaAs substrate induced states during its relaxation and thus the PL of the NCQDs will not be affected. The carrier transfer “onset” excitation energy is determined by the energy separation between the LUMO level of the NCQDs and the ‘trap” states outside NCQDs induced by the passivated GaAs substrate, potentially within its own bandgap as interfacial or surface disorder- induced band tailing states. 289 \ Surfactant ZnSe ZnSe InAs Core 5.2nm dia ~ ~ Vac 1Se LUMO 1Pe 1VB HOMO 2VB 1.02eV 0.18eV 0.44eV 1.22eV …… …… GaAs CB VB 1.42eV (RT) 1eV ?? ?? Surface States 1.13eV Hole Transfer “Onset” excitation energy Lone electron leads to “Dark” QD Surfactant Surfactant ZnSe ZnSe InAs Core 5.2nm dia ~ ~ Vac 1Se LUMO 1Pe 1VB HOMO 2VB 1.02eV 0.18eV 0.44eV 1.22eV …… …… GaAs CB VB 1.42eV (RT) 1eV ?? ?? Surface States 1.13eV Hole Transfer “Onset” excitation energy Lone electron leads to “Dark” QD Surfactant Figure 6.26 Schematic depicting quenching of NCQD PL due to hole transfer to the substrate. Note that after the hole transfer, the NCQD remains “dark” due to the presence of the lone electron. As discussed in §6.1.2(b), because of the small spacing of the hole levels, as well as the smearing of these levels due to fine structure splitting and phonon coupling, the hole levels in NCQDs are a quasi-continuum. The hole relaxation in a 290 NCQD occurs through a cascade of single phonon scattering events through the ladder of intermediate hole levels. So since the drop of NCQD PL intensity above the “onset” is step like and the drop stops ~30meV after onset, the hole trap states due to the GaAs substrate must be narrowly distributed. This fact, and that the “onset” is just ~60meV below the GaAs bandgap, together suggest that the hole trap states outside the NCQDs is not the intrinsic GaAs valence band, but rather narrowly distributed (energywise) states present at the interface of the sulfur passivated GaAs surface and the nanocrystals. It is not clear if these states are induced by the NCQD adsorption. However, the presence of these states is definitely related to the GaAs surface, because no PL drop is observed from the reference sample (NCQD on the glass). Therefore in the remainder of this section, we refer to these states (that act as acceptors for NCQD holes) as the GaAs/NCQD interfacial states. §6.3.4(d) Connection to QD PL Intermittence and Estimation of Transfer Rate It is interesting to note the similarity between the NCQD PL intermittence as discussed in §6.1.2(c) and the NCQD PL reduction on GaAs substrate at excitation energy above “onset” due to hole transfer. Both of these phenomena are related to charge transfer between NCQD and traps in the surrounding environment. As the model for the PL intermittence (see §6.1.1(c)) suggests, the charge transfer from a NCQD to the surrounding is via tunneling of carriers when the NCQD state and trap states are in resonance. In case of InAs/ZnSe NCQDs on GaAs, the immediate environment of the NCQDs is the GaAs surface, the energetics of the “traps” are thus by definition connected to the GaAs band structure and surface states. Hence not 291 surprisingly, we observe charge (hole) transfer only at above “onset” excitation energy, when resonance between states in the NCQDs and GaAs surface states is available. Similar to what happens in PL intermittence, we expect that the hole transfer from a NCQD to the substrate will also lead to a negatively charged “dark”-state NCQD whose PL with continued photon excitation will be quenched due to Auger energy transfer from the photon generated exciton to the lone electron in the NCQD (see §6.1.1(c)). The NCQD in “dark” will be turned into “resting” state again after the alone electron is released (takes ~millisecond to minutes). This “dark” time duration of the NCQD will be the main consequence of the hole transfer that leads to the PL intensity reduction. In the ensemble of the NCQDs adsorbed on the substrate, a fraction of NCQDs are in the “resting” state ( st QD N Re ), another in the excited state ( . Exc QD N ), and the remainder in the “dark” state ( Dark QD N ). These fractions are decided by the following factors: (1) rate of photo excitation ( . exc r ); (2) the average time during which the hole dwells in the NCQD excited states which are in resonance of the GaAs/NCQD interfacial states . ~ exc t ; (3) the transfer (tunneling) rate of the hole to the interfacial states trans R ; and (3) the average period of time for which a NCQD remains charged and “dark” after the hole transfer ( Dark τ ~ ). If we neglect the effect of “natural” PL intermittence of the NCQDs unrelated to the GaAs substrate, the relationship between the above quantities can be written as: 292 ⎪ ⎩ ⎪ ⎨ ⎧ = + + = × × × × 1 ~ ~ . Re . Re Dark QD Exc QD st QD Dark QD Dark trans exc exc st QD N N N N R t r N τ ) 13 . 6 ( ) 13 . 6 ( b a We note that at low excitation power density the fraction of QDs in excited states is small 1 . << Exc QD N (6.14) Further we assume that relaxation to its ground state or transfer to GaAs/NCQD interface states are the only two destinations of an excited hole. Then the average dwell time of the hole in excited state in resonance to NCQD/GaAs interfacial states is: trans rel trans rel exc t t t t t + × = . . . ~ (6.15) where . rel t is the time for hole to relax out of the excited states in resonance with the GaAs/NCQD interface states, and . trans t is the average time for the hole to tunnel into the interface states trans trans R t / 1 . = (6.16) Assuming that the entering of NCQD into “dark” state due to hole transfer is the only reason for the NCQD PL drop, the ratio of the PL intensity at excitations above “onset” energy above I (transfer occurs) to the PL intensity at excitation below “onset” below I (no transfer) is: . Re st QD below above N I I = (6.17) 293 Combining (6.14) to (6.17) above, we obtain: above below trans rel rel Dark exc I I t t t r = + × × + . ~ 1 τ (6.18) Relating to our experimental situation, the photoexcitation rate is ~10 7 /sec (at 100W/cm 2 excitation, and NCQD absorption cross-section ~1nm 2 ). The distribution of “dark”(“off”) time for an InAs NCQD and under the influence of the GaAs substrate has not been studied, but presumably it follows the same power-law, α − ⋅ = t A t P ) ( , as observed in CdSe NCQDs [6.20]. Here, for estimation purposes, we thus take the average “dark” time ( Dark τ ~ ) to be 0.1 sec which is the typically average measured CdSe “dark” (“off”) time in single QD PL experiments [6.20]. The ratio of the PL intensity below to above “onset” excitation is ~5 as shown in the PLE data (figure 6.22). Plugging the above estimate into (6.18) we obtain the ratio of the time of hole relaxation to time of hole transfer as, trans rel t t . = 4x10 -6 , namely the hole transfer time is ~10 5 longer than the relaxation time. The message from the above simple analysis is that even though the hole transfer rate may not be as fast as the relaxation process (namely, hole transfer probability after one photoexcitation event is much less than unity), nevertheless, since after the transfer of the hole the PL of the negatively charged NCQD is quenched for a fairly long time (compared to the rate of photoexcitation), the PL of the NCQDs can still be lowered significantly. The full analysis of the reduction of PL due to hole transfer will have to take into account the natural PL intermittence of 294 a NCQD unrelated to the GaAs substrate, and the effect of the photoexcited GaAs substrate on the NCQD “off” time distribution, such as carrier transfer from the substrate to the NCQD. Such an analysis is currently made difficult due to the lack of independent knowledge of the PL intermittence of the InAs/ZnSe NCQDs and the nature of interface between the NCQDs and the passivated GaAs. §6.3.4(e) PLE temperature dependence and detection energy dependence Temperature dependence of NCQD PLE: Further insight into the correspondence between the “onset” of NCQD PL intensity reduction and the relative positioning of the energy levels in the NCQDs and the GaAs substrate is gained from examination of the dependence of the PLE on the temperature and detection energy. In figure 6.27(a) are shown the PLE spectra at several temperatures between liquid helium and room temperature. These spectra are taken for detection at the NCQD peak, which is essentially independent of the temperature at 1.02eV/1220nm. The “onset” excitation energies in PLE as a function of temperature are plotted in figure 6.27(b) (red squares). Note that the “onset” of PLE tracks largely the change in the GaAs bandgap and is consistently ~45meV below the corresponding bandgap until ~250K. At room temperature the onset energy is ~55meV below the GaAs band gap. The inflexion points of the PLE intensity drop, obtained by Gaussian fitting to the 1st derivative of the PLE curves, are also plotted in figure 6.27(b) (blue squares). Note that these values are consistently ~30meV below the GaAs band edge up to room temperature. 295 1S e -1VB Det.@1.02eV Onset Inflexion Point (a) 1S e -1VB Det.@1.02eV Onset Inflexion Point 1S e -1VB Det.@1.02eV Onset Inflexion Point (a) (b) PLE Inflextion Point PLE “Onset” (b) PLE Inflextion Point PLE “Onset” Figure 6.27 (a) PLE spectra of NCQDs on passivated GaAs substrate (as normalized to that on glass) at various temperatures. Detection is at the NCQD PL peak 1.02eV (which is temperature insensitive). (b) Corresponding PLE drop “onset” energies (red squares) and the inflexion point (blue squares) as a function of temperature; Solid curve: GaAs bandgap; Dotted red curve: GaAs bandgap-46meV, the best fit for PLE drop “onset” energies; Dotted blue curve: GaAs bandgap -30meV, the best fit for the PLE inflexion points. 296 As argued above in §6.3.4(c), within the model of hole transfer from NCQDs to GaAs/NCQD interfacial states (figure 6.26), the “onset” excitation energy is determined by the distance between the LUMO levels of the NCQDs and the interfacial states. Therefore the tracking of the “onset” excitation energy to the GaAs bandgap indicates that the GaAs/NCQD interfacial states are in fact similar to the near valence bandedge GaAs surface states in their nature. Indeed, typical shallow semiconductor surface states move with the corresponding bandedge. Therefore at lower temperatures and larger GaAs bandgap, the “onset” excitation will have to increase accordingly for holes in the NCQDs to remain in resonance with the GaAs surface states. Detection energy dependence of NCQD PLE: Now we turn to the PLE dependence on the detection energy. In figure 6.28(a) are shown the room temperature PLE spectra for NCQDs on passivated GaAs, detected at 1.13eV/1100nm, 1.02eV/1220nm, and 0.94ev/1320nm which correspond to luminescence from narrow windows of increasing average size nanocrystals with the attendant decreasing ground state luminescence energies. In figure 6.28(b) are plotted the corresponding inflexion points of the PLE curves: note these are constant at ~30meV below the GaAs independent of the detection energy, i.e. the NCQD size. One contributing factor to the insensitivity of the PLE drop “onset” to the detection energy is the inter-dot lateral energy transfer as discussed in §6.3.3, which redistributes exciton population among all nanocrystals. However, if we neglect the 297 (a) (a) (b) PLE Inflexion Point (eV) (b) (b) PLE Inflexion Point (eV) Figure 6.28 (a) Room temperature PLE spectra of NCQDs on passivated GaAs (as normalized to that on glass) at various detection energies. (b) Excitation energies at the PLE inflexion points that mark the drop of each PLE curve are plotted as a function of detection energy. 298 effect of inter-dot energy transfer, within the model of hole transfer from NCQDs to the substrate, the insensitivity of the PLE drop “onset” to the detection energy indicates that the LUMO levels of NCQDs of different sizes are aligned with each other within an uncertainty of ~10meV and thus have essentially the same distance to the GaAs/NCQD interfacial states, as depicted in figure 6.29. The reason for the near alignment of the LUMO levels of the NCQDs will require further investigation. Vac 1eV InAs Core 5.2nm dia InAs Core 5.6nm dia InAs Core 4.6nm dia ZnSe ~ ~ Hole Transfer Surface States Detection 1.02eV 1.13eV “Onset” excitation energy 1Se LUMO 1Pe 1VB 2VB GaAs GaAs GaAs ZnSe ZnSe Surfactant CB VB 0.94eV 1Se LUMO 1Se LUMO VB CB VB CB Vac 1eV InAs Core 5.2nm dia InAs Core 5.6nm dia InAs Core 4.6nm dia ZnSe ~ ~ Hole Transfer Surface States Detection 1.02eV 1.13eV “Onset” excitation energy 1Se LUMO 1Pe 1VB 2VB GaAs GaAs GaAs ZnSe ZnSe Surfactant CB VB 0.94eV 1Se LUMO 1Se LUMO VB CB VB CB Figure 6.29 The calculated energy diagram of InAs/ZnSe nanocrystals of 4.6nm, 5.2nm and 5.6nm core diameter on GaAs substrate. Emission energies of these three sizes correspond to the detection energies in the PLE measurements shown in figure 6.28(a) (1.13eV, 1.02eV and 0.94eV). On this plot the LUMO (1Se) levels of the nanocrystals of different sizes are assumed to be aligned in order to explain the observation (figure 6.28) that the “onset” excitation energy for PL reduction is independent of the detection energy. 299 §6.3.4(f) Hole transfer from GaAs substrate to the InAs/ZnSe NCQDs. So far we have discussed hole transfer from NCQDs to the GaAs/NCQD interfacial states (similar to GaAs surface states in their nature). We note however that the hole transfer in the opposite direction, namely from the GaAs surface states to the NCQD as illustrated in figure 6.30, is also a possible pathway leading to the NCQD PL reduction above an “onset” excitation energy. Obviously, if the transfer is due to tunneling as discussed above, the matrix elements for the tunneling from the NCQD to the substrate or vice versa are identical. Therefore, subject only to initial state carrier occupancy, the two processes always happen in parallel. The ratio of transfer rates in one direction or the opposite is thus decided by the carrier occupancy of the relevant states in the NCQDs and on the substrate surface. Similar to the consequence of hole transfer from a NCQD to a substrate, the hole transfer from a substrate to a NCQD will lead to a positively charged “dark” NCQD whose PL will be quenched until the lone hole is released (see §6.1.2(c)). Thus the hole transfer from the substrate to the NCQD is also a candidate model to explain the observed PL intensity reduction at an excitation energy above a threshold, i.e. an “onset” behavior. As illustrated in figure 6.30, the “onset” excitation energy for the PL reduction within this model is the minimum photon energy needed to create holes in the GaAs surface (gap) states near the valence bandedge. Once a hole is created in the surface states, given the quasi-continuous hole states in the NCQD, a nearly resonant state in the NCQD is always available for the created hole to tunnel into. Therefore the insensitivity of the “onset” to the detection energy (figure 6.28) is 300 naturally explained without having to invoke the inter-dot energy transfer or the alignment of the LUMO level of the NCQDs. The tracking of the “onset” excitation energy to the GaAs bandgap is also expected since, the shallow surface states typically move with the bandedge. Vac Surfactant ZnSe ZnSe InAs Core 5.2nm dia ~ ~ 1Se LUMO 1Pe 1VB HOMO 2VB 1.02eV 0.18eV 0.44eV 1.22eV …… …… GaAs CB VB 1eV ?? ?? Surface States 1.13eV Hole Transfer “Onset” excitation energy Lone Hole leads to “Dark” QD Surfactant Vac Surfactant ZnSe ZnSe InAs Core 5.2nm dia ~ ~ 1Se LUMO 1Pe 1VB HOMO 2VB 1.02eV 0.18eV 0.44eV 1.22eV …… …… GaAs CB VB 1eV ?? ?? Surface States 1.13eV Hole Transfer “Onset” excitation energy Lone Hole leads to “Dark” QD Surfactant Figure 6.30 Schematic depicting the quenching of NCQD PL due to hole transfer from a substrate surface state to a NCQD. Such a hole transfer causes the NCQD to have a lone hole which quenches the PL of the NCQD. 301 The weakness of this model of hole transfer from the substrate to the NCQD however is that, as shown in figure 6.23, the “onset” of the PLE reduction is at ~1.36eV, ~55meV below the GaAs bandgap, and the PLE reduction is already completed at 1.40eV before the excitation energy reaches GaAs bandgap (1.42eV at RT). Given the typical small absorption cross-section of surface states (~1A 2 ), the surface state is not likely to be saturated with the low excitation power (100W/cm 2 ) at room temperature. Therefore when the excitation energy reaches GaAs bandgap, the excitation created in the GaAs bulk will feed the surface states, and significantly increase the hole occupancy in the surface states (compared to below GaAs bandgap excitation) and the probability of hole transfer into the NCQDs. Thus one would expect further reduction of NCQD PL intensity at excitation energy above the GaAs bandgap. However the lack of such PLE drop at excitation energy equal to GaAs bandgap shows that the hole transfer from GaAs substrate to the NCQD is unlikely to be the correct model to explain the reduction of NCQD luminescence at excitations above “onset” energy. §6.3.4(g) Conclusions To summarize this section, we have observed excitation transfer between the adsorbed InAs/ZnSe NCQDs and the GaAs substrate at very low excitation power density (~100W/cm 2 ). The excitation transfer is manifested in an abrupt reduction of the PL from the NCQDs when the excitation energy goes above a threshold, thus defining an “onset” energy which is found to be ~55meV below the GaAs bandgap. 302 Given the analysis presented in this section, we believe that the reduction of the NCQD PL is caused by the transfer of holes from high excited states in the NCQD into the GaAs/NCQD interfacial states before the hole relaxes to its HOMO level. The observed excitation transfer from high excited states of NCQDs is unique and the first observation of its kind. It may be exploited to establish a communication link between NCQDs and the underlying substrate, contributing towards our proposed application of the hybrid colloidal/epitaxial structures for biological sensing applications. We note the similarity between (a) the observed reduction of PL from adsorbed NCQDs at excitation energy above “onset” (presented in this section), and (b) the turning-“off” of NCQD PL during PL intermittence (as discussed in §6.1.2(c)). Both of these phenomena are related to charge transfer from a NCQD to the surrounding environment (including a substrate) and the quenching of the PL from the resulting NCQD with a lone carrier. Specifically, in the studied hybrid structures of InAs/ZnSe NCQDs on a GaAs substrate, the hole transfer from NCQDs into the substrate will leave behind “dark” NCQDs with lone electrons. We estimate that because of the relative long “dark” period subsequent to hole transfer (compared to the photoexcitation rate), even when the hole transfer rate is much slower than its intradot relaxation rate (namely, hole transfer probability build-up after every photoexcitation event is much less than unity), significant enough hole transfer can still occur and lead to a significant reduction in the NCQD PL intensity. 303 For further progress on this problem of excitation transfer between InAs/ZnSe NCQDs and a GaAs substrate, an interesting direction is toward the PL behavior of NCQDs sparsely dispersed on a substrate, so that the ensemble PL is not affected by the inter-dot transfer. The ultimate experiment on this line is of course the PL of a single InAs/ZnSe NCQD on GaAs and to measure how the NCQD PL intermittence is modified by the presence of the GaAs substrate. Does the “on” time of the NCQD (on GaAs) indeed decrease dramatically because of hole transfer, and if so, what is the “on” time distribution? Although today there is no available NIR detector of sufficient detectivity to perform such measurements, such experiments may be possible in the future owing to the developing optoelectronics technology in the NIR regime. In addition, the latest technologies have enabled TRPL at sub-picosecond resolution using ultrafast streak camera or the up-conversion technique. For our standard samples of submonolayer of NCQDs dip-coated on substrates, future experiments could be designed to directly measure the loss of hole population during relaxation due to transfer to the substrate. However, these ultrafast time-resolved experiments may be difficult due to the weak PL signal from submonolayer of NCQDs and the lack of good NIR detector. Besides finding out more details on the transfer mechanism in the current NCQD/substrate system, an important direction for future research work could be excitation transfer in the nanocrystal/SAM/substrate hybrid structures (figure 6.31) as discussed in chapter 4. With an appropriate self-assembled monolayer of desired 304 electronic structure between NCQDs and a substrate, it should be possible to tune (enhance or prohibit) the charge transfer process between the NCQDs and the substrate as different applications demand. Substrate SAM Molecule Surfactant NCQD NCQD Substrate SAM Molecule Surfactant NCQD NCQD Figure 6.31 Schematic of NCQD/SAM/Substrate hybrid structure. 305 Chapter 6 Reference: [6.1] Semiconductor and metal nanocrystals, Chap. 3-6, Ed. V. I. Klimov, Marcel Dekker, Inc (2004). [6.2] S. A. Empedocles, D. J. Norris, and M. G. Bawendi, “Photoluminescence Spectroscopy of Single CdSe Nanocrystallite Quantum Dots”, Phys. Rev. Lett. 77, 3873–3876 (1996). [6.3] U. Bockelmann, G. Bastard, “Phonon-scattering and energy relaxation in 2- dimensional, one dimensional and zero-dimensional electron gases”, Phys. Rev. B. 42, 8947-8951 (1990). [6.4] H. Benisty, C. M. Sotomayortorres, C. Weisbuch, “Intrinsic mechanism for the poor luminescence properties of quantum-box systems”, Phys. Rev. B 44, 10945- 10948 (1991). [6.5] R. Heitz, A. Kalburge, Q. Xie, M. Grundmann, P. Chen, A. Hoffmann, A. Madhukar, D. Bimberg, ‘Excited states and energy relaxation in stacked InAs/GaAs quantum dots”, Phys. Rev. B 57, 9050-9060 (1998). [6.6] V. I. Klimov, “Optical nonlinearities and ultrafast carrier dynamics in semiconductor nanocrystals”, J. Phys. Chem. B 104, 6112-6123 (2000). [6.7] V. I. Klimov and D. W. McBranch, “Femtosecond 1P-to-1S Electron Relaxation in Strongly Confined Semiconductor Nanocrystals”, Phys. Rev. Lett. 80 4028-4031 (1998). [6.8] P. Guyot-Sionnest, M. Shim, C. Matranga, M. Hines “Intraband relaxation in CdSe quantum dots”, Phys. Rev. B 60, R2181-R2184 (1999). [6.9] Al. Efros, V. A. Kharchenko, M. Rosen, Breaking the phonon bottleneck in nanometer quantum dos –role of Auger-like processes”, Solid State Comm. 93, 281- 284 (1995). [6.10] M. Nirmal, D. J. Norris, M. Kuno, M. G. Bawendi, Al. Efros, M. Rosen, “Observation of the dark exciton in CdSe quantum dot”, Phys. Rev. Lett. 75 3728- 3731 (1995). [6.11] Al. L. Efros and M. Rosen, M. Kuno, M. Nirmal, D. J. Norris, and M. Bawendi, “Band-edge exciton in quantum dots of semiconductors with a degenerate valence band: Dark and bright exciton states”, Phys. Rev. B 54, 4843-4956 (1996). 306 [6.12] U. Banin, J. C. Lee, A. A. Guzelian, A. V. Kadavanich, A. P. Alivisatos, “Exchange interaction in InAs nanocrystal quantum dots”, Superlattices and Microstructures 22, 559-567 (1997). [6.13] A. V. Uskov, A. P. Jauho, B. Tromborg, J. Mork, R. Lang, “Dephasing times in quantum dots due to elastic LO phonon-carrier collisions”, Phys. Rev. Lett. 85, 1516-1519 (2000). [6.14] S. Xu, A. A. Mikhailovsky, J. A. Hollingsworth, V. I. Klimov, “Hole intraband relaxation in strongly confined quantum dots: Revisiting the "phonon bottleneck" problem”, Phys. Rev. B 65, 045319 (2002). [6.15] S. S. Prabhu, A. S. Vengurlekar, S. K. Roy, J. Shah, “Nonequilibrium dynamics of hot carriers and hot phonons in CdSe and GaAs”, Phys. Rev. B. 51 14233-14246 (1995). [6.16] M. Nirmal, B. O. Dabbousi, M. G. Bawendi, J. J. Macklin, J. K. Trautman, T. D. Harris, L. E. Brus, “Fluorescence intermittency in single cadmium selenide nanocrystals”, Nature, 383, 802-804 (1996). [6.17] B. R. Fisher, H-J. Eisler, N. E. Stott, and M. G. Bawendi, “Emission Intensity Dependence and Single-Exponential Behavior in Single Colloidal Quantum Dot Fluorescence Lifetimes”, J. Phys. Chem. B, 108, 143-148 (2004). [6.18] U. Banin, M. Bruchez, A. P. Alivisatos, T. Ha, S. Weiss, D. S. Chemla, ”Evidence for a thermal contribution to emission intermittency in single CdSe/CdS core/shell nanocrystals”, J. Chem. Phys. 110, 1195-1201 (1999). [6.19] R. G. Neuhauser, K. T. Shimizu, W. K. Woo, S. A. Empedocles, M. G. Bawendi, “Correlation between fluorescence intermittency and spectral diffusion in single semiconductor quantum dots”, Phys. Rev. Lett. 85, 3301-3304 (2000). [6.20] M. Kuno, D. P. Fromm, H. F. Hamann, A. Gallagher, D. J. Nesbitt, “Nonexponential "blinking" kinetics of single CdSe quantum dots: A universal power law behavior”, J. Chem. Phys. 112, 3117-3120 (2000). [6.21] D. I. Chepic, Al. Efros, A. I. Ekimov, M. G. Vanov, V. A. Kharchenko, I. A. Kudriavtsev, T. V. Yazeva, “Auger ionization of semiconductor quantum drops in a glass matrix”, J. Lumin. 47, 113-127 (1990). [6.22] P. Roussignol, D. Ricard, J. Lukasik, C. Flytzanis, “New results on optical- phase conjugation in semiconductor-doped glasses”, J. Opt. Soc. Am. B, 4, 5-13 (1987). 307 [6.23] K. T. Shimizu and M. G. Bawendi, “Optical dynamics in single semiconductor quantum dots”, Chap. 6 in Semiconductor and metal nanocrystals, Ed. V. I. Klimov, Marcel Dekker, Inc (2004). [6.24] T. Bashe, “Fluorescence intensity fluctuations of single atoms, molecules and nanoparticles”, J. Lumin. 76, 263-269 (1998). [6.25] R. M. Dickson, A. B. Cubitt, R. Y. Tsien, W. E. Moerner, “On/off blinking and switching behaviour of single molecules of green fluorescent protein”, Nature 388, 355-358 (1997). [6.26] A. P. Alivisatos, T. D. Harris, P. J. Carrol, M. L. Steigerwald, L. E. Brus,"Electron-vibration coupling in semiconductor clusters studied by resonance Raman spectroscopy", J. Chem. Phys. 90, 3463-3468 (1989). [6.27] Al. Efros, M. Rosen, M. Kuno, M. Nirmal, D. J. Norris, M. Bawendi, ”Band- edge exciton in quantum dots of semiconductors with a degenerate valence band: Dark and bright exciton states”, Phys. Rev. B 54, 4843-4856 (1996). [6.28] I. Mukhametzhanov, “Growth control, structural characterization, and electronic structure of Stranski-Krastanow InAs/GaAs (001) quantum dots”, Ph.D. Disseration, University of Southern California (2002). [6.29] E. T. Kim, “Tuning InAs quantum dot electronic structure using (InAlGa)As capping layers and application to infrared photodetectors”, Ph.D. Dissertation, University of Southern California (2003). [6.30] V. Bruckner, Applications of Time-Resolved Optical Spectroscopy, Elsevier (1990). [6.31] J. Shah, Ultrafast Spectroscopy of Semiconductors and Semiconductor Nanostructures, Springer-Verlag (1999). [6.32] J. T. Verdeyen, Laser Electronics, Prentice Hall (1995). [6.33] W. H. Press, B. P. Flannery, S. A. Teukolsky, W. T. Vetterling, §13.1 in Numerical Recipes in C, 2nd Edition, Cambridge Press (1992). [6.34] F. J. Grunthaner, P. J. Grunthaner, R. P. Vasquez, B. F. Lewis, J. Maserjian, A. Madhukar, “Local atomic and electronic-structure of oxid-GaAs and SiO2-Si interfaces using high-resolution XPS.” J. Vac. Sci. Tech. 16, 1443-1453 (1979). And reference therein. [6.35] T. J. Ulrych, T. N. Bishop, ”Maximum entropy spectral analysis and autoregressive decomposition”, Rev. Geophys. 13, 183-200 (1975). 308 [6.36] S. Pelet, M. J. R. Previte, L. H. Laiho, P. T. C. So “A fast global fitting algorithm for fluorescence lifetime imaging microscopy based on image segmentation”, Biophys. J. 87, 2807-2817 (2004). [6.37] A. E. McKinnon, A. G. Szabo, D. R. Miller, “Deconvolution of photoluminescence data”, J. Phys. Chem. 81, 1564-1570 (1977). [6.38] D. V. O’Conner, W. R. Ware, J. C. Andre, “Deconvolution of fluorescence decay curves – Critical comparison of techniques”, J. Phys. Chem. 83, 1333-1343 (1979). [6.39] W. H. Press, B. P. Flannery, S. A. Teukolsky, W. T. Vetterling, §15.5 in Numerical Recipes in C, 2nd Edition, Cambridge Press (1992). [6.40] R. Heitz, H. Born, F. Guffarth, O. Stier, A. Schliwa, A. Hoffmann, D. Bimberg,”Existence of a phonon bottleneck for excitons in quantum dots”, Phys. Rev. B 64, 241305 (2001). [6.41] R. R. Chance, A. Prock, R. Silbey, “Molecular Fluorescence and Energy Transfer Near Interfaces” Vol. 37, p.1 in Advance in Chemical Physics, Eds. S. A. Rice and I. Prigogine, Wiley-Interscience, New York (1978). [6.42] T. T. T. Li, M. J. Weaver, “Intramolecular electron-transfer at metal-surfaces 4. Dependence of tunneling probability upon donor-acceptor spectration distance”, J. Am. Chem. Soc. 106, 6017-6108 (1984). [6.43] A. P. Alivisatos, M. F. Arndt, S. Efrima, D. H. Waldeck, and C. B. Harris, “Electronic-energy transfer at semiconductor interfaces 1. Energy-transfer from two-dimensional molecular films to Si (111)”, J. Chem. Phys. 86, 6540-6549 (1987). [6.44] C. E. D. Chidsey, and D. N. Loiacono, “Chemical functionality in self- assembled monolayers – structural and electrochemical properties” Langmuir, 6, 682-691 (1990). [6.45] S. J. Diol, E. Poles, Y. Rosenwaks, R. J. Dwayne Miller, “Electron-transfer dynamics at GaAs surface quantum well”, J. Phys. Chem. B 102, 6193-6201 (1998). [6.46] E. P. A. M. Bakkers, E. Reitsma, J. J. Kelly, and D. Vanmaekelbergh “Excited-State Dynamics in CdS Quantum Dots Adsorbed on a Metal Electrode”, J. Phys. Chem. B 103, 2781-2788 (1999). [6.47] L. Brus, “Model for carrier dynamics and photoluminescence quenching in wet and dry porous silicon thin films”, Phys. Rev. B 53, 4649-4656 (1996). 309 [6.48] S. Logunov, T. Green, S. Marguet, M. A. el-Sayed, “Interfacial carriers dynamics of CdS nanoparticles”, J. Phys. Chem. A 102, 5652-5658 (1998). [6.49] Y. Wang, N. Herron, “Photoconductivity of CdS nanocluster-doped polymers”, Chem. Phys. Lett. 200, 71-75 (1992). [6.50] J. X. Cheng, S. H. Wang, X. Y. Li, Y. J. Yan, S. H. Yang, C. L. Yang, J. N. Wang, W. K. Ge, “Fast interfacial charge separation in chemically hybridized CdS- PVK nanocomposites studied by photoluminescence and photoconductivity measurements”, Chem. Phys. Lett. 333, 375-380 (2001). [6.51] V. L. Colvin, M. C. Schlamp, A. P. Alivisatos, “Light-emitting-diodes made from Cadium Selenide nanocrystals and a semiconducting polyer”, Nature, 370, 354- 357 (1994). [6.52] N. C. Greenham, X. G. Peng, A. P. Alivisatos, “Charge separation and transport in conjugated-polymer/semiconductor-nanocrystal composites studied by photoluminescence quenching and photoconductivity”, Phys. Rev. B 54, 17628- 17637 (1996). [6.53] Th. Förster, “10th Spiers Memorial Lecture. Transfer mechanisms of electronic excitation”, Discuss. Faraday Soc. 27, 7 – 17 (1959). [6.54] D. L. Dexter, “A Theory of Sensitized Luminescence in Solids”, J. Chem. Phys. 21, 836-850 (1953). [6.55] M. Stavola, D. L. Dexter and R. S. Knox, “Electron-hole pair excitation in semiconductor via energy transfer from an external sensitizer”, Phys. Rev. B 31, 2277-2289 (1985). [6.56] D. Basko, G. C. La Rocca, F. Bassani, V. M. Agranovich, “Förster energy transfer from a semiconductor quantum well to an organic material overlayer”, Eur. Phys. J. B 8, 353-362 (1999). [6.57] C. A. Leatherdal, C. R. Kagan, N. Y. Morgan, S. A. Empedocles, M. A. Kastner, M. G. Bawendi, “Photoconductivity in CdSe quantum dot solids”, Phys. Rev. B 62, 2669-2680 (2000). [6.58] M. A. Ratner, A. Madhukar, “On the role of nuclear motions in electron and excitation transfer rates: Importance of transfer-integral dependence upon nuclear coordinate”, Chem. Phys. 30, 201-215 (1978). [6.59] S. Song, R. A. Clark, E. F. Bowden, M. J. Tarlov, “Characterization of Cytochrome-C Alkanethiolate structures prepared by self-assembly on gold”, J. Phys. Chem. 97 6564-6572 (1993). 310 [6.60] R. A. Marcus, N. Sutin, “Electron transfer in chemistry and biology”, Biochimica et Biophysica Acta, 811, 265-322 (1985). [6.61] Chap. 1 & 2 in Molecular Electronics: A 'Chemistry for the 21st Century' Monograph (Iupac Chemistry for the 21st Century Series), Eds. J. Jortner, M. Ratner, Blackwell Science (1997). [6.62] C. R. Kagan, C. B. Murray, and M. G. Bawendi. “Long-range resonance transfer of electronic excitations in close-packed CdSe quantum-dot solids”, Phys. Rev. B. 54, 8633-8643 (1996). [6.63] C. B. Murray, C. R. Kagan, M. G. Bawendi, “Synthesis and characterization of monodisperse nanocrystals and close-packed nanocrystal assemblies”, Ann. Rev. Mater. Sci. 30, 545-610 (2000). [6.64] O. Labeau, P. Tamarat, and B. Lounis “Temperature Dependence of the Luminescence Lifetime of Single CdSe=ZnS Quantum Dots”, Phys. Rev. Lett. 90, 257404 (2003). [6.65] J. Wickham, “Investigation of the Mechanism of Pressure-Induced Solid-Solid Structural Transformations in Semiconductor Nanocrystals”, Ph. D. Dissertation, University of California at Berkeley (2001). [6.66] I. L. Medintz, A. R.Clapp, H. Mattoussi, E. R. Goldman, B. Fiber, J. M. Mauro, “Self-assembled nanoscale biosensors based on quantum dot FRET donors”, Nature Materials, 2, 630-638 (2003). [6.67] J. H. Kim, D. Morikis, M. Ozkan, “Adaptation of inorganic quantum dots for stable molecular beacons”, Sensor and Actuat. B – Chem.l 102, 315-319 (2004) [6.68] R. Heitz, I. Mukhametzhanov, P. Chen, A. Madhukar, “Excitation transfer in self-organized asymmetric quantum dot pairs”, Phys. Rev. B 58, R10151-R10154 (1998). [6.69] Q. H. Xie, P. Chen, A. Madhukar, “InAs island-induced-strain driven adatom migration during GaAS overlayer growth”, Appl. Phys. Lett. 65, 2051-2053 (1994). [6.70] M. Achermann, M. A. Petruska, S. Kos, D. L. Smith, D. D. Koleske, and V. I. Klimov, “Energy-transfer pumping of semiconductor nanocrystals using an epitaxial quantum well”, Nature, 429, 642-646 (2004). [6.71] S. Kos, M. Achermann, V. I. Klimov, D. L. Smith, “Different regimes of Förster-type energy transfer between an epitaxial quantum well and a proximal monolayer of seminconductor nanocrystals”, Phys. Rev. B 71, 205309 (2005). 311 [6.72] H. Kroemer, Chap. 17 in Molecular Beam Epitaxy and Heterostructrues, Eds. L.L. Chang and K. Plog, NATO ASI Series, Martinus, Nijhoff, (1985). 312 Chapter 7. Conclusion and Outlook In this concluding chapter, we summarize the main results reported in the preceding chapters and provide some potential future research directions. §7.1 Nanocrystal synthesis and electronic structure. (Chapter 2 summary) In chapter 2 we discussed the nanocrystal synthesis and electronic structure. To-date, the most successful nanocrystal synthesis route for both II-VI and III-V nanocrystals is via thermolysis of organometallic precursor in polar Lewis base (trioctylphosphine or trioctylphosphine oxide) acting as both coordinating solvent and capping agent. Practically, in order to achieve narrow nanocrystal size distribution, a conventional two step growth strategy involving separated nanocrystal nucleation and growth phase is often employed. In InAs nanocrystal growth, we find the two step growth fails to keep the nanocrystal size distribution narrow beyond ~4nm diameter, due to the difficulty in suppressing nanocrystals nucleation during continued growth phase. This difficulty can be eased by solvent injection during growth to lower the precursor concentration or more significantly, by the special “punctuated” growth approach introduced by this author. In the “punctuated” approach, we separate the nucleation and continued growth stage into two different growth cycles, thus decoupling their contradictory requirements on the precursor concentration. Pre-formed InAs nanocrystal nuclei (~4nm diameter) are further grown with a significantly lowered precursor concentration. Punctuated growth approach thus effectively suppresses the nucleation during continued growth and 313 produces large diameter InAs nanocrystals (8.5nm diameter), ~2nm larger than possible with conventional growth approach. The as-prepared InAs nanocrystals covered by TOP surfactant have low quantum efficiency (QE) ~1% which can be significantly improved with ZnSe shell coating. The optimal thickness of the ZnSe shell is ~1.8ML (for book-keeping of shell thickness see §2.3.3). At this optimal shell thickness, the QE of the InAs/ZnSe nanocrystals increases ~20 times compared to InAs/TOPO nanocrystal. With further increased shell thickness, the nanocrystal evolves into dendrimers with shape similar to CdSe tetrapods with short arms. The shape evolution is likely to be accompanied by defect formation in the shell layer, as a decrease in the QE of nanocrystal is simultaneous observed. This is, to the author’s knowledge, the first observation of such nanocrystal dendrimerization during stressed shell overgrowth. The detailed mechanism will require further investigation. If the process can be understood and properly controlled, it may lead to a new methodology to create complex nanostructures. One further step on the synthesis is to convert the aqueous functionalization of the as-synthesized InAs/ZnSe nanocrystal to molecules compatible with solubility in water, and the subsequent attachment of appropriate biological ligand for near- infrared cell labeling application. Since the shell layer of InAs nanocrystal is limited to fairly thin thickness (~2ML ZnSe) compared to the shell thickness of standard CdSe/ZnS nanocrystals (~6ML), the optical property of InAs/ZnSe is expected to be more susceptible to surfactant exchange. Therefore phospholipids micelle 314 encapsulation based aqueous functionalization is likely to be a better strategy for InAs/ZnSe nanocrystals compared to surfactant exchange. In strongly confined nanocrystal quantum dots (nanocrystal radius smaller than exciton Bohr radius), the confinement energy of electrons and holes is far larger than their mutual Coulomb interaction which can be thus treated as a perturbation. The strong confinement resulted NCQDs properties, such as size-tunable optical transition and δ-function energy states can be qualitatively understood using a simple “particle-in-a-box” model as backed by the effective mass approximation. However such “particle-in-a-box” model has no quantitative value because it fails to account for the mixing of different bulk bands due to the confinement potential. To calculate the electronic structure of the nanocrystals made of narrow bandgap InAs, we employed an 8 band p k r r ⋅ method (Pidgeon-Brown model) that explicitly considers the coupling between the two lowest conduction bands and six highest valence bands. The calculated InAs energy levels (assuming infinite barrier height) fit reasonably well with experimentally measured optical transitions for large InAs nanocrystals (> 5nm diameter). The deviation of calculated results for smaller nanocrystals can be due to the assumed infinite barrier height or the fundamental limitation of the p k r r ⋅ method as a perturbative method restricted to the k~0 region. In the future, a method that can describe the whole Brillioun zone such as tight- binding should be tried to calculate the electronic structure of the small InAs nanocrystals. 315 §7.2 Quantum Dot based Cell labeling and Simultaneous AFM/NSOM imaging. (Chapter 3 Summary) An integrated far-field and near-field optical microscopy and spectroscopy setup was built as part of this dissertation work. Complementary to the conventional far-field epifluorescence microscopy which does not have sufficient depth resolution for cells, this setup also includes a unique normal force tapping mode atomic-force microscopy (AFM) based near-field scanning optical microscope (NSOM). The AFM based NSOM enables simultaneous morphological and surface sensitive fluorescence imaging of the cells which can be beneficial to bridge the traditional pathological knowledge of cell/tissue (i.e. morphology, density, shape) to the physiology, i.e. the biochemical processes and mechanisms underlying diseases. In our implementation of the NSOM, the near-field optical excitation is generated through the use of a tapered fiber tip with an aperture defined by metal coating. Qualitatively the problem of electromagnetic wave diffraction through a subwavelength aperture is solved using Born’s angular spectrum theory. We found the near-field components start to dominate the diffracted light when the diameter of the aperture is below half the wavelength. On practical experimental consideration, we note that there exists significant loss of light power throughput with reduced aperture diameter due to attenuation of light in the aperture defined by finite thickness metal coating. The available excitation power, and thus the fluorescence signal level, is the limiting factor of the NSOM resolution. The noise of the NSOM 316 image is dominated by shot-noise, which limits the signal-to-noise ratio of our NSOM image to ~7. Results of single QD NSOM imaging are presented. The observed fluorescence signal is ~50 times less than the expectation estimated using the measured excitation power and the assumption that QD quantum efficiency remains the same as in solution. This discrepancy is not yet solved. A contributing factor could be the quenching of QD quantum yield due to resonant energy transfer to the metallic coating on the tip. As calculated using the CPS theory, the QE of the QD drops an order of magnitude when the QD is 10nm away from the metallic coating. The author speculates such fluorescence quenching can be used beneficially to improve the imaging resolution. In a clinically relevant case study, the combined AFM/NSOM is used to image breast cancer cells SK-BR-3 and MDA-MB-231 with their cancer-inducing mutant surface receptor Her2/neu labeled by nanocrystal quantum dots. The procedure and considerations for quantum dot labeling are discussed. Using AFM based NSOM imaging (150nm optical resolution) we find the distribution of Her2/neu receptor on surface of both types of cells is in clusters of ~500nm diameter, consistent with the expected dimerization/oligomerization of the mutant Her2/neu. However, so far we have observed no correlation of cell morphology (height) vs. Her2/neu distribution. We measured the fluorescence spectra of unlabeled and QD labeled SK-BR-3 and MDA-MB-231. These cells have significantly reduced autofluorescence 317 background in the 900-1500nm near-infrared wavelength range compared to the visible range (500-900nm). Since the autofluorescene of the cell is the ultimate limitation on the sensitivity in the detection of fluorescently labeled biomarkers, it is highly desirable to move QD labeling to the 900-1500nm NIR regime using, for example, the InAs/ZnSe core/shell nanocrystals discussed in chapter 2. This can be an important direction of future research. §7.3 Semiconductor surface “passivation”/modification and nanocrystal adsorption on untreated/modified semiconductor surfaces. (Chapter 4 Summary) An intrinsic challenge that the epitaxial quantum nanostructures face as being a part of the hybrid colloidal/epitaxial hybrid structure is the degradation of optical performance due to surface recombination. As a model system we study the PL behavior of InAs SAQDs as a function of their GaAs capping layer thickness. The PL of the SAQDs degraded ~20 times when the capping decrease to 40ML. Red shift and broadening of PL width is also observed to accompany reduced capping thick, which can be explained by the variation of stress in the SAQD and the capping layer as a function of capping thickness, previously predicted by molecular dynamics based simulation. One way to reduce carrier surface recombination and improve the efficiency of near surface quantum structure is via appropriate passivation of the surface. We studied wet chemical sulfur “passivation” of GaAs surface, which successfully 318 improved the room temperature PL from near surface SAQDs or PL from GaAs substrate by better than one order of magnitude. One important aspect of surface effect (resulted from our specific surface finishing and treatment) not yet investigated is the shift of the energy levels in the substrate due to surface band bending. The characterization of surface band bending in our samples is an important aspect of future studies for a thorough understanding of the excitation transfer in the colloidal/epitaxial hybrid structure. Also in chapter 4, we discussed the simplest construction of colloidal/epitaxial hybrid structure which is based on deposition of nanocrystals on unmodified substrates using a refined evaporative assembly technique – dip-coating. Large area uniform, monolayer height, closely packed nanocrystal arrays have been deposited on substrate using dip-coating technique. We discussed the mechanism of dip-coating as unique technique for self-assembly of colloidal particles. In dip- coating the transport of particles is directional toward the air-liquid-solid contact line. The capillary force involved in ordering is long ranged and of 1/L distance dependence, extending >100nm for nanometer-sized objects. Thus dip-coating is potentially an economic method to construct large area ordered nanoparticle array with controllable orientation. For future development relating to dip-coating, it will be interesting to study the dip-coating assembly of nanoparticles on pre-patterned substrates. Finally in chapter 4 we discuss the generic strategy to chemically modify the semiconductor surface via covalently grafted self-assembled monolayer (SAM) and 319 SAM induced chemisorption of nanocrystals. Procedures for Si/SiO 2 and GaAs SAM functionalization are presented. These procedures lead to uniform adsorption of SAM layer. With thiol group (-SH) as its free end, the SAM molecules can be used as a bilinker to direct chemisorption of nanocrystals on surface to create colloidal/epitaxial hybrid structures. Uniform layer of InAs and CdSe nanocrystal has been linked to hexanedithiol modified GaAs surface and characterized by AFM. The nanocrystal/SAM/substrate structures can be a future platform to study the excitation transfer between nanocrystals and substrates. On this platform, one has the freedom of tuning the interaction between the colloidal QD and the epitaxial nanostructure by controlling energy levels in the linker SAM molecules and their alignment with the states in the QD and epitaxial structure. §7.4 Bioconjugated SAM mediated cell adhesion on solid surfaces via specific ligand-receptor interaction (Chapter 5 Summary) In chapter 5, we expand the application of SAM based solid surface modification to control the adsorption of biological entities such as proteins and living cells. Via appropriate choice of the chemical group on its free end, SAM can act as bilinker to covalently bind to protein molecules. As an example, 3- bromopropyl-tricholoro-silane (BPTS) was used to functionalize Si/SiO 2 surface. The bromine free end of the BPTS SAM via thiol-ether reaction links to the cysteine amino acid exposed on the outer surface of a genetically engineered chaperonin 320 protein ((HSP60), hence significantly enhance the adsorption of the chaperonin protein compared to unmodified surfaces, as characterized by AFM. SAM, as a general strategy for surface modification, can also be applied to control the adsorption of living cells on a solid substrate (referred to as cell adhesion in the general literature). One proposed method is to functionalize surface with SAM molecules that has a peptide ligand as their free ends, which we dub as bioconjugated SAM (BSAM). The peptide ligand is designed to bind to surface receptors that present on the type of targeting cells via specific ligand-receptor interaction to achieve selective cell adhesion. In our proposed scheme of biological using hybrid colloidal/epitaxial structures, such surface modification can play the essential role to enhance adhesion of diseased target cells and keep the cell-substrate in close distance for detection. Further, such surface modification in general impacts biomedical engineering in the emerging field of implantable prosthetic devices. In these applications an appropriate biotic-abiotic interface is essential to allow optimal function of the implant over extended period of time. In the context of cortical prosthesis, we undertook systematic studies to examine the effectiveness of surface modification using the peptide ligand based BSAM to achieve enhanced selective neuronal cell adhesion. Specifically we have employed two BSAMs whose free ends are, respectively, peptide IKVAV (neuron specific peptide ligands) and peptide KHIFSDDSSE (astrocyte specific peptide ligand) [Ref 4.49 Rao 1992] to modify glass (SiO 2 ) substrates. The procedure for BSAM functionalization was presented. Uniform BSAM surface coverage at both 321 micrometer and millimeter length scale is achieved as characterized using AFM and fluorescence microscopy, respectively. The BSAM functionalized glass substrate was exposed to hippocampal neuron (E16) cell culture. As expected, a higher coverage of neurons on IKVAV (neuron specific ligands) functionalized surface is found compared to KHIFSDDSSE (astrocyte specific ligand), demonstrating the effectiveness of the specific ligand- receptor interaction in achieving enhanced selective cell adhesion. However, compared to IKVAV modified surface, we observe even higher neuron coverage on glass surface modified with the poly-D-lysine (PDL) which represents nonspecific electrostatic binding, not the specific ligand-receptor binding. This observation indicates that the peptide ligand density distribution and receptor-ligand binding are not optimal and further studies into the underlying factors that affect these are needed. We speculate two potential factors are (1) the topology and roughness of the surface presented to the cell; and (2) the conformation of the peptide ligand upon surface adsorption. Indeed in our current study the roughness (i.e. surface height fluctuations on lateral length scales of <1nm) of the glass substrate is quite comparable to the length of the conjugating BSAM, which can reduce the effective density of exposed peptide ligand accessible to the cell receptors. The second consideration is the affinity of the receptor-ligand binding which could be significantly reduced due to changes in the conformation of the peptide ligand as presented to the cell surface receptor. Indeed, a marked increase in surface roughness 322 is seen after the peptide ligand adsorption and suggests the need for closer examination of the underlying reasons. We note that the high level of surface roughness and simultaneously a high packing density of the peptide ligand in our current study and most of the past studies of cell adhesion to peptide ligand functionalized surfaces, may has an unintended benefit. That is the live cell has the ability to self-adjust its topology and the projections of its trans-membrane protein receptors participating in binding with the effectively lowered exposed ligand density presented by a rough surface. Future experiments can be design to compare cell adhesion on substrate surface of controlled roughness and of atomic flatness which are functionalized with the same peptide ligand. This may reveal the impact of substrate surface roughness on cell adhesion. §7.5 Excitation transfer in nanocrystal quantum dot / crystalline semiconductor substrate hybrid structures (Chapter 6 Summary) In chapter 6, we focus on our studies of the optical properties of the most basic of the colloidal/epitaxial hybrid structures and their unique optical properties. Given the importance of the GaAs/InAs based III-V epitaxial structures in optoelectronics, as well as the importance of 1.1-1.5 μm wavelength regime provided by this material system to the future of biological detection, the investigation is carried out in a canonical system of InAs/ZnSe based core-shell NCQD in direct 323 contact with a surface passivated GaAs (001) substrate and the corresponding reference system of InAs/ZnSe NCQD on glass substrate. The NCQDs were adsorbed on substrates (GaAs or glass) via dip-coating technique, which results in a sub one-monolayer height of close-packed NCQDs on the substrate. In the ensemble of adsorbed NCQDs, we observed significant inter-dot excitation transfer via dipole-dipole interaction mediate Forster resonant energy transfer (FRET). Compared to the reported investigation of FRET in 3D packed CdSe nanocrystal solids, this is the first observation of FRET in NCQDs submonolayer coverage and in NIR InAs/ZnSe nanocrystals. The studies are carried out in the system of InAs/ZnSe NCQD on glass substrate to avoid confusion from excitation transfer between NCQD and substrates. In the ensemble of adsorbed NCQDs, the luminescence decay time of the smaller NCQD (1.13eV/1100nm emission) is shorter compared to that of larges ones (1.02eV/1220nm emission). More importantly, unlike the large nanocrystals whose luminescence decay time increases with decreasing temperature, the decay time of the smaller NCQD has very small temperature dependence, as they are limited by the rate of energy transfer to nearby larger NCQDs. The critical radius for FRET transfer and the inter-dot FRET rate is estimated to be, respectively, ~10nm and 4ns. Indeed this estimation reasonably explained the observed luminescence decay time of the smaller NCQDs and their temperature dependence. For future research, the longer FRET critical radius in NIR NCQDs (compared to visible NCQD) is advantageous for NIR NCQDs application in FRET based biological detection application. The FRET 324 mechanism may also be pursued to establish communication link between vertical aligned NCQDs and near surface epitaxial island SAQDs. Of more significance, compared to the inter-dot resonant energy transfer, we have observed a non-conventional excitation transfer between the adsorbed InAs/ZnSe NCQDs and the GaAs substrate at very low excitation power density (~100W/cm 2 ). The excitation transfer is manifested in an abrupt reduction of the PL from the NCQDs when the excitation energy goes above a threshold, thus defining an “onset” energy which is found to be ~55meV below the GaAs bandgap. Given the analysis presented in chapter 6, we believe that the reduction of the NCQD PL is caused by the transfer of holes from high excited states in the NCQD into surface state of GaAs substrate (possibly related or induced by NCQD adsorption) before the hole relaxes to its HOMO level. The observed excitation transfer from high excited states of NCQDs is unique and the first observation of its kind. It may be exploited to establish a communication link between NCQDs and the underlying substrate, contributing towards our proposed application of the hybrid colloidal/epitaxial structures for biological sensing applications. We note the similarity between (a) the observed reduction of PL from adsorbed NCQDs at excitation energy above “onset” (presented in this section), and (b) the turning-“off” of NCQD PL during PL intermittence. Both of these phenomena are related to charge transfer from a NCQD to the surrounding environment (including a substrate) and the quenching of the PL from the resulting NCQD with a lone carrier. Specifically, in the studied hybrid structures of 325 InAs/ZnSe NCQDs on a GaAs substrate, the hole transfer from NCQDs into the substrate will leave behind “dark” NCQDs with lone electrons. We estimate that because of the relative long “dark” period subsequent to hole transfer (compared to the photoexcitation rate), even when the hole transfer is much slower than its relaxation (namely, hole transfer probability after every photoexcitation event is much less than unity), the hole transfer can still lead to a significant reduction in the NCQD PL intensity. For further progress on this problem of excitation transfer between InAs/ZnSe NCQDs and a GaAs substrate, an interesting direction is toward the PL behavior of a single InAs/ZnSe NCQD on GaAs and to measure how the NCQD PL intermittence is modified by the presence of the GaAs substrate. Although today there is no available NIR detector of sufficient detectivity to perform such measurements, such experiments may be possible in the future owing to the developing optoelectronics technology in the NIR regime. In addition, the latest technologies have enabled TRPL at sub-picosecond resolution using ultrafast streak camera or the up-conversion technique. For our standard samples of submonolayer of NCQDs dip-coated on substrates, future experiments could be designed to directly measure the loss of hole population during relaxation due to transfer to the substrate. However, these ultrafast time-resolved experiments may be difficult due to the weak PL signal from submonolayer of NCQDs and the lack of good NIR detector. Besides finding out more details on the transfer mechanism in the current NCQD/substrate system, an important direction for future research work could be 326 excitation transfer in the nanocrystal/SAM/substrate hybrid structures as discussed in chapter 4. With an appropriate self-assembled monolayer of desired electronic structure between NCQDs and a substrate, it should be possible to tune (enhance or prohibit) the charge transfer process between the NCQDs and the substrate as different applications demand. §7.6 Outlook – Nanocrystal heterostructure for functional biological coupling As summarized in the previous sections of this concluding chapter, in this dissertation we have covered the following research grounds: (1) the synthesis of InAs based near-infrared emitting nanocrystal quantum dots (NCQDs), (2) NCQD cell labeling and imaging, (3) semiconductor surface modification/functionalization and the synthesis of nanocrystal/substrate hybrid structure (4) Bioconjugated SAM based surface functionalized for selective cell adhesion (5) excitation transfer in nanocrystal/crystalline semiconductor substrate hybrid structure. However, towards our roadmap of hybrid colloidal/epitaxial nanostructure for chip-based, ultrasensitive biological detection, there is still one essential component which is not yet discussed: the electronic coupling between nanocrystals and the biological entities. Unfortunately, discussion of such coupling is virtually non- existing in the field of NCQD based biological detection as it stands today. In the current scheme of NCQD based biological detection, NCQDs are always performing their only role as passive emitters. It is the biological ligand-receptor interaction that binds the ligand conjugated NCQDs to the corresponding targets (receptors) and thus 327 manifests the existence/distribution of the target. Such binding between the conjugated NCQDs and the targets is the only interaction between the two and the response of the NCQD is not sensitive to the environment. In this sense, NCQDs are inferior to the fluorescent molecular dyes which can often reflect the environmental change (PH, ions, electric potential, ligand binding, etc.) via variation in their fluorescence properties. In this author’s view, how to establish a meaningful functional coupling between nanocrystals and biological entities (cells, proteins) beyond the notion of such “attachment” is indeed a most significant problem for the application of nanocrystals in biological detection. Recently, moving toward this direction a few studies have reported the Foster resonant energy transfer (FRET) between a pair of NCQD and dye conjugated via biological molecules such as protein and DNA [7.1, 7.2]. However, even these cases still do not involve any electronic coupling between NCQDs and biomolecules themselves. The FRET coupling is also severely limited to absorbing/fluorescent species, and therefore is not a universal coupling mechanism. In analogy to the fact that a piece semiconductor alone does not make an electronic device, we believe that simple homogeneous nanocrystals alone will not be able to provide the desired functional coupling to the biological molecules. Instead, more complex nanocrystal heterostructures are necessary for this purpose. Thus, here, as an epilogue of this dissertation and an outlook for a major research field for the future, we briefly discuss our work on semiconductor nanocrystal heterostructures and their prospective for biological application. 328 We conceive that the simplest nanocrystal heterostructure for functional biological manipulation is a semiconductor/metal (S/M) junction (Schottky junction) as schematically shown in figure 7.1. The corresponding schematic band structure of a Schottky junction is shown in figure 7.2. Owing to the difference in the chemical potentials of the two components (metal and semiconductor), the energy bands of the semiconductor bend near the junction, which in turn creates a built-in electrical field in the semiconductor. Upon light excitation, photo-created electron-hole pairs in the semiconductor get separated by the built-in field, and lead to the photovoltaic (PV) effect. The photovoltage that can be achieved is decided by the type of metal and semiconductor in contact, and is typically tens to hundreds of mV. We expect that the electrical field due to the PV effect in a nanocrystal S/M junction can provide an effective electrical coupling to biological entities. Metal Semiconductor Metal Semiconductor Figure 7.1 Schematic of nanocrystal semiconductor/metal junction. 329 E vacuum E c E v Semiconductor Metal hv E f - - - - + ++ + W m Φ n Χ E vacuum E c E v Semiconductor Metal hv E f - - - - + ++ + W m Φ n Χ Figure 7.2 schematic band structure of a semiconductor/metal junction (Schottky junction) and its photovoltaic effect. Synthesis and characterization of Au/CdSe nanocrystal S/M junction Towards this end, as the first material system, we investigated the synthesis of Au/CdSe nanocrystal S/M junction. The Au/CdSe nanojunctions were synthesized using a two-step organometallic routine. First, using standard procedure, CdSe nanorods were formed by reacting Cd and Se precursors in coordinate solvent TOPO and alkyl-phosphonic acid. In the second step, Au precursors were added into the CdSe rod solution. Owing to the chemical difference of the two ends of the CdSe rods, Au particles can be grown on one end preferentially by appropriate control on 330 growth conditions. The detailed procedure of the synthesis is presented in Appendix- E. In figure 7.3 are shown two illustrative TEM images ( (a) low resolution, (b) high resolution) of an Au/CdSe S/M junction sample synthesized by this author. In both images the dark end of the rod corresponds to the Au tip which has higher electron density compared to the CdSe. As measured from TEM images, 71% of the CdSe rods have a single Au tip on one end of the rods, 23% of the CdSe rods have Au tips on both ends of the rods, and 6% of the CdSe rods have no Au tip. The average Au tip diameter is 5nm. Note that in the high resolution TEM image (figure 7.3(b)) evenly spaced CdSe lattice planes can be observed, demonstrating that the CdSe integrity is not compromised by Au growth. We have observed the photovoltaic effect in the Au/CdSe junction in solution environment utilizing the transient electrical birefringence (TEB) measurements. CdSe rods possess optical birefringence due to their anisotropy. In the TEB method, while an external electric field is switched on and off, we measure the rise ( Δn r ) and fall ( Δn f ) of the birefringence of the CdSe rods or Au/CdSe junction (in solution). The rising edge of the birefringence of the rods reflects the rotation of Au/CdSe junction to preferentially align its dipole with the external field. The falling edge reflects the randomization of the rod orientation once the external field is removed, whose slope gives rotational diffusion constant of the rods. 331 20nm (a) 20nm (a) 5nm (b) 5nm (b) Figure 7.3 Illustrative TEM images ((a) low resolution, (b) high resolution) of as- synthesized Au/CdSe nanocrystal metal/semiconductor heterojunctions. (Courtesy of Bryce Sadtler and Steven Hughes, University of California, Berkeley) 332 Experimental setup for the measurement is as described in reference [7.3], except that a 532nm Nd:YAG laser is added for the photoexcitation of Au/CdSe junctions. Au/CdSe sample of large aspect ratio (CdSe rods ~30nm long and 3nm in diameter, 3nm diameter Au tip) were used to achieve sufficient TEB signal. Figure 7.4 shows the rising (a) and falling (b) edge of the transient birefringence for three cases: CdSe rod before Au tip growth (black), CdSe/Au heterojunction with light excitation off (blue), and CdSe/Au with excitation on (Red). Excitation was via cw 532nm laser at 1W/cm 2 . Heating by the excitation laser is considered and ruled out. Compared to CdSe rod, the fall and rise of CdSe/Au birefringence are significantly slower (~20 times). Moreover, a significant difference in the TEB response of CdSe/Au samples was observed when excitation is switched on and off, which suggests that the photoexcitation causes charge redistribution in individual Au/CdSe S/M junction so that their rotational response under an external electric field is affected, confirming to the expected PV effect in a S/M junction. Nanocrystal S/M junction for noninvasive electrical stimulation of cells Upon light illumination, the photovoltaic effect of the nanocrystal S/M junction offers, in general, an electrode-free method of non-invasive, local and site- specific electrical excitation of biological entities. This would be a valuable tool for the study of electric field related cellular processes. An example of such envisioned application is the usage of nanocrystal S/M junction for neuron stimulation. As schematically shown in figure 7.5(a), we envision the nanocrystal S/M junction to 333 (a) (a) (b) (b) Figure 7.4 Panel (a) (panel (b)) shows the rise (fall) of birefringence upon application (removal) of external electrical field of the following samples: CdSe rod (black), Au/CdSe NCHJ with light excitation off (blue) and light excitation on (red). 334 Resting K + Channel h + e - hv K + ~140mM Na + ~12mM Cl - ~4mM …… K + ~4mM Na + ~150mM Cl - ~120mM …… Extracelluar 0mV Cytosolic -60mV 1e-h / hv C M =~1 μF/cm 2 R K =60k Ω·cm 2 E K =60mV R serial R parallel Na + Cl - Na + Cl - K + A - K + A - S M (a) (b) Nanocrystal S/M junction Transmembrane Embedded Cell Plasma Membrane Resting K + Channel h + e - hv K + ~140mM Na + ~12mM Cl - ~4mM …… K + ~4mM Na + ~150mM Cl - ~120mM …… Extracelluar 0mV Cytosolic -60mV 1e-h / hv C M =~1 μF/cm 2 R K =60k Ω·cm 2 E K =60mV R serial R parallel Na + Cl - Na + Cl - K + A - K + A - S M (a) (b) Nanocrystal S/M junction Transmembrane Embedded Cell Plasma Membrane Figure 7.5: (a) Schematic of nanocrystal S/M junction embedded neuron membrane and (b) corresponding equivalent circuit diagram. be transmembrane embedded in a neuron to achieve the maximum efficiency of stimulation. The transmembrane embedment can likely be achieved via appropriate amphiphilic surface modification of the nanocrystal S/M junctions which makes them compatible with the nature of the bilipid cell membrane. 335 For a zeroth order estimation of the magnitude of electrical stimulation a nanocrystal S/M junction can create, we approximate the transmembrane S/M junction embedded neuron by the equivalent circuit diagram shown in figure 7.5(b). The equivalent circuit of the bilipid / distributed protein containing plasma membrane is depicted in purple. The membrane is typically 5-10nm in average thickness and functions as a capacitor (capacitance C M ~1μF/cm 2 ) [7.4]. The ion channels on the membrane (mainly resting K + channel) lead to a typical resistivity of R k =60k Ω·cm 2 [7.4]. The ionic gradient across the membrane, created by the ATP driven ion-transporters, leads to an electromotive potential typically ~60mV directing from inside of the membrane to the outside [7.4]. (This means that at steady state the potential inside the cell is ~60mV below outside, referred to as resting potential of the cell.) An Embedded nanocrystal S/M junction (metal inside the cell) acts essentially as a photo cell, which is equivalent to a current source with a resistor and a capacitor in parallel (figure 7.5(b) depicted in blue). The current is determined by the rate of photon absorption. The parallel resistor presents the internal resistance within the S/M junction. The connection between the S/M junction and the membrane is mediated via the motion of ions in the solution environment, which is represented by a series resistor between the S/M junction and the membrane. Assuming the object is to stimulate the neuron sufficiently for it to fire an action potential, upon light excitation of the embedded nanocrystal S/M junction, the photovoltage of the S/M junction needs to reduce the potential across the neuron cell membrane by ~10mV [7.4] (called membrane depolarization) in order to statistically 336 cause the opening of efficient number of voltage-sensitive Na + ion channels which, in turn, cause the action potential to be generated and travel down the axon. To estimate the needed density of nanocrystal S/M junctions and the light flux necessary to create the ~10mV membrane depolarization, we perform the following semiquantitative estimation. Consider that the S/M junction consists of 5nm diameter gold tip grown on 10nm diameter CdSe nanorods, and are embedded in the membrane at a density of 10 10 /cm 2 . At such a density the S/M junctions occupy only a 1% of the membrane area and their capacitance is much smaller than that of the membrane. Such an S/M junction will have ~10nm 2 absorption cross-section for light at visible wavelength. Assume every photon absorbed by the S/M junction creates one e-h separation across the plasma membrane and neglect the recombination of photon generated e-h within the FAN, (R parallel >>R K +R Serial ). Given the capacitance and resistance of the membrane, in order to generate one action potential, (namely to charge the membrane 10mV) every second, the embedded nanocrystal S/M junctions need to create ~10 -8 C/cm 2 charge displacement across the membrane per second. This in turn requires only 0.02mW/cm 2 light (visible wavelength) power density impinging on the cell membrane. The above analysis indicates that the nanocrystal S/M junctions can indeed offer an effective method of electrical stimulation of neurons using optical excitation of fairly low power density. Further improvement of the stimulation efficiency may be achieved by selective binding of the PV nanocrystal S/M junction to the voltage sensitive ion channels which initiate neuron firing. We note that this non-invasive 337 electrical simulation using embedded S/M junctions may offer a variety of opportunities ranging from probing fundamental intracellular electric processes to endowing/restoring light sensitivity to retinal neurons (such as the retinal ganglion cells) and serving as nanoscale devices for visual prosthesis. 338 Chapter 7 References: [7.1] I. L. Medintz, A. R.Clapp, H. Mattoussi, E. R. Goldman, B. Fiber, J. M. Mauro, “Self-assembled nanoscale biosensors based on quantum dot FRET donors”, Nature Materials, 2, 630-638 (2003). [7.2] J. H. Kim, D. Morikis, M. Ozkan, “Adaptation of inorganic quantum dots for stable molecular beacons”, Sensor and Actuat. B – Chem. 102, 315-319 (2004). [7.3] L. S. Li, A. P. Alivisatos, “Origin and scaling of the permanent dipole moment in CdSe nanorods” Phys. Rev. Lett. 90, 097402 (2003). [7.4] B. Hille, Ionic channels of excitable membranes, Sinauer Associates Inc. (1992). 339 Bibliography E. Abbe, “Beitrage zur theorie des mikroskops und der mikroskopischen wahrnehmung,” Archiv fur Mikroskopische Anatomie 9, 413-463 (1873). M. Achermann, M. A. Petruska, S. Kos, D. L. Smith, D. D. Koleske, and V. I. Klimov, “Energy-transfer pumping of semiconductor nanocrystals using an epitaxial quantum well”, Nature, 429, 642-646 (2004). J. D. Aiken III, R. G. Finke, “A review of modern transition-metal nanoclusters: their synthesis, characterization, and applications in catalysis”, J. Mol. Catal. A – Chem. 145, 1–44 (1999). R. R. Alfano, G. C. Tang, A. Pradhan, W. Lam, D. S. Choy, and E. Opher, “Fluorescence spectra from cancerous and normal breast and lung tissues”, IEEE J. Quantum Electronics, 23, 1806-1811 (1987). A. P. Alivisatos, M. F. Arndt, S. Efrima, D. H. Waldeck, and C. B. Harris, “Electronic-energy transfer at semiconductor interfaces 1. Energy-transfer from two- dimensional molecular films to Si (111)”, J. Chem. Phys. 86, 6540-6549 (1987). A. P. Alivisatos, T. D. Harris, P. J. Carrol, M. L. Steigerwald, L. E. Brus,"Electron- vibration coupling in semiconductor clusters studied by resonance Raman spectroscopy", J. Chem. Phys. 90, 3463-3468 (1989). A. P. Alivisatos, W. W. Gu, C. Larabell, “Quantum dots as cellular probes”, Ann. Rev. of Biomed. Eng. 7, 55-76 (2005). L. Alroy, Y. Yarden. “The ErbB signaling network in embryogenesis and oncogenesis: signal diversification through combinatorial ligand-receptor interactions.” FEBS (Federation of European Biochemical Societies) Lett. 410, 83- 86 (1997). W. P. Ambrose, P. M. Goodwin, J. C. Martin, R. C. Kell, “Alternations of single molecule fluorescence lifetimes in near-field optical microscopy”. Science 265, 364- 367 (1994). M. Arnold, E. A. Cavalcanti-Adam, R. Glass, J. Blummel, W. Eck, M. Kantlehner, H. Kessler, J. P. Spatz, “Activation of integrin function by nanopatterned adhesive interfaces” Chem. Phys. Chem., 5, 383-388 (2004). 340 D. E. Aspnes, “Recombination at semiconductor surfaces and interfaces”, Surf. Sci. 132, 406-421 (1983). C. D. Bain, E. B. Troughton, Y. T. Tao, J. Evall, G. M. Whitesides, R. G. Nuzzo, “Formation of monolayer films by the spontaneous assembly of organic thiol from solution onto gold”, J. Am. Chem. Soc. 111, 321-335 (1989). E. P. A. M. Bakkers, E. Reitsma, J. J. Kelly, and D. Vanmaekelbergh “Excited-State Dynamics in CdS Quantum Dots Adsorbed on a Metal Electrode”, J. Phys. Chem. B 103, 2781-2788 (1999). J. Bange, E. Zwick E, A. Ullrich, "Molecular targets for breast cancer therapy and prevention". Nature Medicine 7, 548–552 (2001). U. Banin, Private Communication. U. Banin, J. C. Lee, A. A. Guzelian, A. V. Kadavanich, A. P. Alivisatos, “Exchange interaction in InAs nanocrystal quantum dots”, Superlattices and Microstructures 22, 559-567 (1997). U. Banin, C. J. Lee, A. A. Guzelian, A. V. Kadavanich, A. P. Alivisatos, W. Jaskolski, G. W. Bryant, Al. L. Efros, M. Rosen, “Size-dependent electronic level structure of InAs nanocrystal quantum dots: Test of multiband effective mass theory”, J. Phys. Chem., 109, 2306-2309 (1998). U. Banin, Y. W. Cao, D. Katz, O. Millo, “Identification of atomic-like electronic states in indium arsenide nanocrystal quantum dots”, Nature, 400, 542-544 (1999). U. Banin, M. Bruchez, A. P. Alivisatos, T. Ha, S. Weiss, D. S. Chemla, ”Evidence for a thermal contribution to emission intermittency in single CdSe/CdS core/shell nanocrystals”, J. Chem. Phys. 110, 1195-1201 (1999). C. I. Bargmann, M. C. Hung, R. A. Weinberg, “Multiple independent activations of the neu oncogene by a point mutation altering the transmembrane domain of p185” Cell 45, 649-657 (1986). T. Bashe, “Fluorescence intensity fluctuations of single atoms, molecules and nanoparticles”, J. Lumin. 76, 263-269 (1998). D. Basko, G. C. La Rocca, F. Bassani, V. M. Agranovich, “Förster energy transfer from a semiconductor quantum well to an organic material overlayer”, Eur. Phys. J. B 8, 353-362 (1999). G. Bastard, “Super-lattice band-structure in the envelop-function approximation”, Phys. Rev. B. 24, 5693-5697 (1981). 341 G. Bastard, J. A. Brum, “Electronic states in semiconductor heterostructures”, IEEE J. Quantum Electron. 22, 1625-1644 (1986). T. Baum, S. Ye, K. Uosaki, “Formation of Self-Assembled Monolayers of Alkanethiols on GaAs Surface with in Situ Surface Activation by Ammonium Hydroxide”, Langmuir, 15, 8577-8579, (1999). M. G. Bawendi, A. R. Kortan, M. L. Steigerwald, L. E. Brus,“X-ray structural characterization of larger CdSe semiconductor clusters.“ J. Chem. Phys. 91, 7282- 7290 (1989). H. Benisty, C. M. Sotomayortorres, C. Weisbuch, “Intrinsic mechanism for the poor luminescence properties of quantum-box systems”, Phys. Rev. B 44, 10945-10948 (1991). T. W. Berger, G. Chauvet, R. J. Sclabassi, “A biologically-based model of functional-properties of the hippocampus”, Neural Networks, 7, 1031-1064 (1994). Toward Replacement Parts for the Brain – Implantable Biomimetic Electronics as Neural Prostheses, Eds. T. W. Berger, L. Glanzman, MIT Press, Cambridge, MA (2005). R. S. Besser, C. R. Helms, “Comparison of surface properties of sodium sulfide and ammonium sulfide passivation of GaAs”, J. Appl. Phys. 65, 4306-4310 (1989). H. A. Bethe, “Theory of diffraction by small holes”, Phys. Rev. 66, 163-182 (1944). E. Betzig, A. Harootunian, A. Lewis, and M. Isaacson, “Near-field diffraction by a slit: implications for superresolution microscopy”, Appl. Opt. 25, 1890-1900 (1986). K. Betzig, J. K. Trautman, T. D. Harris, J. S. Weiner, R. L. Kostelak, “Breaking the diffraction barrier: Optical Microscopy on a nanometric scale.” Science, 251, 1468- 1470 (1991). E. Betzig, P. L. Finn, J. S. Weiner, “Combined shear force and near-field scanning optical microscopy”, Appl. Phys. Lett. 60, 2484-2486 (1992). G. L. Bfr, G. E. Pikus, Sov. Phys. Solid State 3, 2221 (1962). D. Bimberg, M. Grundmann, F. Heinrichsdorff, N. N. Ledentsov, c, V. M. Ustinov, A. E. Zhukov, A. R. Kovsh, M. V. Maximov, Y. M. Shernyakov, B. V. Volovik, A. F. Tsatsul’nikov, P. S. Kop’ev and Zh. I. Alferov, “Quantum dot lasers: breakthrough in optoelectronics”, Thin Solid Films 367, 235-249 (2000). 342 U. Bockelmann, G. Bastard, “Phonon-scattering and energy relaxation in 2- dimensional, one dimensional and zero-dimensional electron gases”, Phys. Rev. B. 42, 8947-8951 (1990). Immunochemical Staining Methods Handbook, Ed. T. Boenisch, DakoCytomation (2001). M. Born, E. Wolf, “Principle of Optics”, Pergamon Press, London (1980). C. Boulas, J. V. Davidovits, F. Rondelez, D. Vuillaume, Suppression of Charge Carrier Tunneling through Organic Self-Assembled Monolayers”, Phys. Rev. Lett. 76, 4797-4800 (1996). J. C. Bouwkamp, “Diffraction Theory”, Rep. Prog. Phys. 17, 35-100 (1954). J. B. Brzoska, I. B. Azuoz, F. Rondelez, “Silanization of solid substrates: A step toward reproducibility”, Langmuir 10, 4367-4373 (1994). M. Bruchez Jr. M. Moronne, P. Gin, S. Weiss, A. P. Alivisatos, “Semiconductor Nanocrystals as Fluorescent Biological Labels”, Science 281, 2013-2015 (1998). V. Bruckner, Applications of Time-Resolved Optical Spectroscopy, Elsevier (1990). L. Brus, “A simple model for the ionization potential, electron affinity, and aqueous redox potentials of small semiconductor crystallites”, J. Chem. Phys. 79, 5566-5571 (1983). L. Brus, “Model for carrier dynamics and photoluminescence quenching in wet and dry porous silicon thin films”, Phys. Rev. B 53, 4649-4656 (1996). J. Buijs, D. W. Britt, V. Hlady, “Human Growth Hormone Adsorption Kinetics and Conformation on Self-Assembled Monolayers”, Langmuir, 14, 335-341 (1998). Y. Cao, U. Banin, “Synthesis and characterization of InAs/InP and InAs/CdSe core/shell nanocrystals.” Angew. Chem. Int. Ed. 38, 3692-3694 (1999). Y. Cao and U. Banin, “Growth and Properites of Semiconductor Core/Shell Nanocrystals with InAs Cores”, J. Am. Chem. Soc. 122, 9692-9702 (2000). W. C. W. Chan and S. Nie, “Quantum Dot Bioconjugates for Ultrasensitive Nonisotopic Detection”, Science, 281, 2016-2018 (1998). R. R. Chance, A. Prock, R. Silbey, “Molecular Fluorescence and Energy Transfer Near Interfaces” Vol. 37, p.1 in Advance in Chemical Physics, Eds. S. A. Rice and I. Prigogine, Wiley-Interscience, New York (1978). 343 R. R. Chance, A. Prock, R. Silbey, “Lifetime of an Excited Molecule Near a Metal Mirror: Energy Transfer in the Eu3+/Silver System” J. Chem. Phys. 60, 2184-2185 (1974). Y. C. Chang, G. D. Sanders, D. Z. –Y. Ting, in Exciton in Confined System, Eds. R. Del Sole, A. D’Andrea, and A. Lapiccirella, Spring-Verlog, Berlin, (1988). L. Chen, “Optical Properties of InGaAs/GaAs quantum well structures: Application to light modulators and switches”, Ph. D. Disseratation, University of Southern California (1993). J. X. Cheng, S. H. Wang, X. Y. Li, Y. J. Yan, S. H. Yang, C. L. Yang, J. N. Wang, W. K. Ge, “Fast interfacial charge separation in chemically hybridized CdS-PVK nanocomposites studied by photoluminescence and photoconductivity measurements”, Chem. Phys. Lett. 333, 375-380 (2001). D. I. Chepic, Al. Efros, A. I. Ekimov, M. G. Vanov, V. A. Kharchenko, I. A. Kudriavtsev, T. V. Yazeva, “Auger ionization of semiconductor quantum drops in a glass matrix”, J. Lumin. 47, 113-127 (1990). C. E. D. Chidsey, and D. N. Loiacono, “Chemical functionality in self-assembled monolayers – structural and electrochemical properties” Langmuir, 6, 682-691 (1990). I. Chung, K. T. Shimizu, M. G. Bawendi, “Room temperature measurement of the 3D orientation of single CdSe quantum dots using polarization microscopy“, P. Natl. Acad. Sci. USA 100, 405-408 (2003). J. P. Cleveland, B. Anczykowski, A. E. Schmid, V. B. Elings, “Energy dissipation in tapping-mode atomic force microscopy”, Appl. Phys. Lett. 72, 2613-2615 (1998). S. Coe, W. K. Woo, M. Bawendi, V. Bulovic, “ Electroluminescence from single monolayers of nanocrystals in molecular organic devices”, Nature, 420, 800-803 (2002). C. A. Colinge, J-P Colinge, Physics of Semiconductor Devices, Springer-Verlag, (2002). V. L. Colvin, A. N. Goldstein, A. P. Alivisatos, “ Semiconductor nanocrystals covalently bound to metal surfaces with self-assembled monolayers”, J. Am. Chem. Soc. 114, 5221-5230 (1992). V. L. Colvin, M. C. Schlamp, A. P. Alivisatos, “Light-emitting-diodes made from Cadium Selenide nanocrystals and a semiconducting polyer”, Nature, 370, 354-357 (1994). 344 D. Courjon, Near-field microscopy and near-field optics, Imperial College Press, London (2003). Y. Cui, Q. Q. Wei, H. K. Park. C. M. Lieber, Nanowire nanosensors for highly sensitive and selective detection of biological and chemical species, Science, 293, 1289-1292 (2001). B. O. Dabbousi, C. B. Murray, M. F. Rubner, M. G. Bawendi, “Langmuir-Blodgett manipulation of size-selected CdSe nanocrystallites”, Chem. Mater. 6, 216-219 (1994). B. O. Dabbousi, J. Rodriguez-Viejo, F. V. Mikulec, J. R. Heine, H. Mattoussi, R. Ober, K. F. Jensen, M. G. Bawendi, “(CdSe)ZnS core-shell quantum dots: Synthesis and characterization of a size series of highly luminescent nanocrystallites” J. Phys. Chem. B 101, 9463-9475 (1997). N. D. Denkov, O. D. Velev, P. A. Kralchevsky, I. B. Ivanov, H. Yoshimura, K. Nagayama, “Mechanism of formation of 2-Dimensional crystals from latex-particles on substrates”, Langmuir 8, 3183-3190 (1992). D. L. Dexter, “A Theory of Sensitized Luminescence in Solids”, J. Chem. Phys. 21, 836-850 (1953). R. M. Dickson, A. B. Cubitt, R. Y. Tsien, W. E. Moerner, “On/off blinking and switching behaviour of single molecules of green fluorescent protein”, Nature 388, 355-358 (1997). A. S. Dimitrov, K. Nagayama, “Continuous convective assembling of fine particles into two-dimensional arrays on solid surfaces”, Langmuir, 12, 1303-1311 (1996). S. J. Diol, E. Poles, Y. Rosenwaks, R. J. Dwayne Miller, “Electron-transfer dynamics at GaAs surface quantum well”, J. Phys. Chem. B 102, 6193-6201 (1998). J. Dreybrodt, F. Daiminger, J. P. Reithmaier, and A. Forchel, “Dynamics of carrier- capture processes in Ga x In 1-x As/GaAs near-surface quantum wells”, Phys. Rev. B 51, 4657-4660 (1995). U. Durig, D. W. Pohl, F. Rohner, “Near-field optical scanning microscopy“, J. Appl. Phys, 59, 3318-3327 (1986). Y. Ebenstein, T. Mokari, U. Banin, “Fluorescence quantum yield of CdSe/ZnS nanocrystals investigated by correlated atomic force and single-particle fluorescence microscopy”, Appl. Phys. Lett. 80, 4033-4035 (2002). 345 Al. Efros, V. A. Kharchenko, M. Rosen, Breaking the phonon bottleneck in nanometer quantum dos –role of Auger-like processes”, Solid State Comm. 93, 281- 284 (1995). Al. L. Efros, M. Rosen, M. Kuno, M. Nirmal, D.J. Norris, M. Bawendi, “Band-edge exciton in quantum dots of semiconductors with a degenerate valence band: Dark and bright exciton states”, Phys. Rev. B 54, 4843-4856 (1996). Al. L. Efros, M. Rosen, “Quantum size level structure of narrow-gap semiconductor nanocrystals: effect of band coupling”, Phys. Rev. B 58, 7120-7135 (1998). A. I. Ekimov, Al. L. Efros, A. A. Onushchenko, “Quantum size effects in semiconductor microcrystals”, Solid State Commun. 56, 921-924 (1985). A. I. Ekimov, F. Hache, M. C. Schanne-Klein, D. Ricard, C. Flytzanis, I. A. Kudryavtsev, T. V. Yazeva, Al. Efros, “Absorption and intensity-dependent photoluminescence measurements on CdSe quantum dots: assignment of the first electronic transitions”, J. Opt. Soc. Am. B, 10, 100-107 (1993). V. Emiliani, B. Bonanni, C. Presilla, M. Capizzi, A.Frova, Y. L. Chang, I. H. Tan, J. L. Merz, M. Colocci, M. Gruioli, “Interaction mechanism of near-surface quantum well with oxidized and H-passivated AlGaAs surfaces”, J. Appl. Phys. 75, 5114- 5122 (1994). S. A. Empedocles, D. J. Norris, and M. G. Bawendi, “Photoluminescence Spectroscopy of Single CdSe Nanocrystallite Quantum Dots”, Phys. Rev. Lett. 77, 3873–3876 (1996). R. J. Errington, Advanced practical inorganic and metalorganic chemistry, Chapman & Hall, London (1997). Z. M. Fang, K. Y. Ma, D. H. Jaw, R. M. Cohen, G. B. Stringfellow, “Photoluminescence of InSb, InAs, and InAsSb Grown by organometallic vapor- phase epitaxy.” J. Appl. Phys. 67, 11 (1990) 7034-7039. M. Faraday, “The Bakerian Lecture: Experimental Relation of Gold (and other Metals) to Light”, Philosophical Transactions of the Royal Society of London, 147, 145-181 (1857). B. R. Fisher, H-J. Eisler, N. E. Stott, and M. G. Bawendi, “Emission Intensity Dependence and Single-Exponential Behavior in Single Colloidal Quantum Dot Fluorescence Lifetimes”, J. Phys. Chem. B, 108, 143-148 (2004). Th. Förster, “10th Spiers Memorial Lecture. Transfer mechanisms of electronic excitation”, Discuss. Faraday Soc. 27, 7 – 17 (1959). 346 T. Funatsu, Y. Harada, M.Tokunaga, K. Saito, T. Yanagida, “Imaging of single fluorescent molecules and individual ATP turnovers by single myosin molecules in aqueous-solution”, Nature, 374, 555-559 (1995). T. W. Gadella, T. M. Jovin, “Oligomerization of epidermal growth factor receptors on A431 cells studied by time-resolved fluorescence imaging microscopy. A stereochemical model for tyrosine kinase receptor activation” J. Cell Biol. 129, 1543-1558 (1995). G. Galilei, Dialogues Concerning Two New Sciences, translated by H. Crew and A. de Salvio, Prometheus Books, 1991. X. H. Gao, Y. Y. Cui, R. M. Levenson, L. W. K. Chung, S. M. Nie, “In vivo cancer targeting and imaging with semiconductor quantum dots”, Nature Biotechnol. 22, 969-976 (2004). M. Geirsig, P. Mulvaney, “Formation of ordered 2-dimensional gold colloidal lattices by electrophoretic deposition”, J. Phys. Chem. 97, 6334-6336 (1993). B. L. Gel’mont, M. I. Dyakonov, “Acceptor levels in diamond type semiconductors ” Sov. Phys. Semicond. 5, 1905 (1972). D. S. Ginger and N. C. Greenham, "Electrical Properties of Semicondcutor Nanocrystals", Chap. 7 in Semiconductor and Metal Nanocrystals, Ed.V. Klimov, Marcel Dekker, Inc. (2004). N. C. Greenham, X. G. Peng, A. P. Alivisatos, “Charge separation and transport in conjugated-polymer/semiconductor-nanocrystal composites studied by photoluminescence quenching and photoconductivity”, Phys. Rev. B 54, 17628- 17637 (1996). J. J. Greffet, R. Carminatic. “Image formation in neear-field optics”, Prog. Surf. Sci. 56, 133-237 (1997). Nano-Optoelectronics, Ed. M. Grundmann, Springer-Verlag, berlin (2002). F. J. Grunthaner, P. J. Grunthaner, R. P. Vasquez, B. F. Lewis, J. Maserjian, A. Madhukar, “Local atomic and electronic-structure of oxid-GaAs and SiO2-Si interfaces using high-resolution XPS.” J. Vac. Sci. Tech. 16, 1443-1453 (1979). And reference therein. S. Guha, A. Madhukar, K. C. Rajkumar, ”Onset of incoherency and defect introduction in the initial-stages of molecular-beam epitaxial-growth of highly strained InxGa1-xAs on GaAs (100)”, Appl. Phys. Lett. 57, 2110-2112 (1990). 347 I. Gur, N. A. Fromer, M. L. Geier, A. P. Alivisatos, “Air-stable all-inorganic nanocrystal solar cells processed from solution”, Science 310, 462-465 (2005). P. Guyot-Sionnest, M. Shim, C. Matranga, M. Hines “Intraband relaxation in CdSe quantum dots”, Phys. Rev. B 60, R2181-R2184 (1999). A. A. Guzelian, U. Banin, A. V. Kadavanich, X. Peng, and A. P. Alivisatos, “Colloidal chemcial synthesis and characterization of InAs nanocrystals quantum dot”, App. Phys. Lett. 69, 1432-1434 (1996). J. R. Heath, C. M. Knober, D. V. Leff, “Pressure/temperature phase diagrams and superlattices of organically functionalized metal nanocrystal monolayers: The influence of particle size, size distribution, and surface passivant”, J. Phys. Chem. B 101, 189-197 (1997). J. Heath, J. J. Shiang, “Covalency in semiconductor quantum dots”, Chem. Soc. Rev. 27, 65-71 (1998). R. Heitz, I. Mukhametzhanov, P. Chen, A. Madhukar, “Excitation transfer in self- organized asymmetric quantum dot pairs”, Phys. Rev. B 58, R10151-R10154 (1998). R. Heitz, A. Kalburge, Q. Xie, M. Grundmann, P. Chen, A. Hoffmann, A. Madhukar, D. Bimberg, ‘Excited states and energy relaxation in stacked InAs/GaAs quantum dots”, Phys. Rev. B 57, 9050-9060 (1998). R. Heitz, H. Born, F. Guffarth, O. Stier, A. Schliwa, A. Hoffmann, D. Bimberg,”Existence of a phonon bottleneck for excitons in quantum dots”, Phys. Rev. B 64, 241305 (2001). B. Hille, Ionic channels of excitable membranes, Sinauer Associates Inc. (1992). M. A. Hines, P. Guyot-Sionnest, “Synthesis and Characterization of Strongly Luminescing ZnS-Capped CdSe Nanocrystals”, J. Phys. Chem. 100, 468-471 (1996). N. E. Hynes, D. R. Stern, “The biology of erbB2/neu/HER-2 and its role in cancer”, Biochimica et Biophysica Acta 1198, 165-184 (1994). J. Israelachvili, Intermolecular & surface forces, Academic Press (1992). J. D. Jackson, Classical Electrodynamics, Wiley (1998). Atomic Force Microscopy in Cell Biology (Methods in Cell Biology, Volume 68), Eds. B. P. Jena, J. K. H. Horber, Academic Press (2002). 348 U. Joos, T. Biskup, O. Ernst, I. Westphal, C. Gherasim, R. Schmidt, K. Edinger, G. Pilarczyk, C. Duschl, “Investigation of cell adhesion to structured surfaces using total internal reflection fluorescence and confocal laser scanning microscopy”, Eur. J. Cell Biol. 85, 225–228 (2006). Chap. 1 & 2 in Molecular Electronics: A 'Chemistry for the 21st Century' Monograph (Iupac Chemistry for the 21st Century Series), Eds. J. Jortner, M. Ratner, Blackwell Science (1997). L. Josephson, M. F. Kircher, U. Mahmood, Y. Tang, R. Weissleder, “Near-infrared fluorescent nanoparticles as combined MR/optical imaging probes”, Bioconjugate Chemistry, 13, 554-560 (2002). C. R. Kagan, C. B. Murray, and M. G. Bawendi. “Long-range resonance transfer of electronic excitations in close-packed CdSe quantum-dot solids”, Phys. Rev. B. 54, 8633-8643 (1996). L. Kam, W. Shain, J. N. Turner, R. Bizios, “Selective adhesion of astrocytes to surfaces modified with immobilized peptides”, Biomaterials 23, 511-515 (2002). E. O. Kane “Band structure of Indium Antimonide“, J. Phys. Chem. Solids, 1, 249 (1957). J. E. B Katari, V. L. Colvin, A. P. Alivisatos, “X-ray photoelectron-spectroscopy of CdSe nanocrystals with application to studies of the nanocrystal surface”, J. Phys. Chem. 98, 4109-4117 (1994). E. T. Kim, Z. Chen, and A. Madhukar, “Selective manipulation of InAs quantum dot electronic states using a lateral potential confinement layer”, Appl. Phys. Lett. 81, 3473-3475 (2002). E. T. Kim, A. Madhukar, Z. M. Ye, and J. C. Campbell, “High detectivity InAs quantum dot infrared photodetectors”, App. Phys. Lett. 84, 3277-3279 (2004). E. T. Kim, “Tuning InAs quantum dot electronic structure using (InAlGa)As capping layers and application to infrared photodetectors”, Ph.D. Dissertation, University of Southern California (2003). J. H. Kim, D. Morikis, M. Ozkan, “Adaptation of inorganic quantum dots for stable molecular beacons”, Sensors and Actuat. B – Chem. 102, 315-319 (2004). V. I. Klimov and D. W. McBranch, “Femtosecond 1P-to-1S Electron Relaxation in Strongly Confined Semiconductor Nanocrystals”, Phys. Rev. Lett. 80 4028-4031 (1998). C. Kittle, Introduction to Solid State Physics, Wiley, New York, (1995). 349 V. I. Klimov, “Optical nonlinearities and ultrafast carrier dynamics in semiconductor nanocrystals”, J. Phys. Chem. B 104, 6112-6123 (2000). V. I. Klimov, A. A. Mikhailovsky, S. Xu, J. A. Hollingsworth, C. A. Leatherdale, H. J. Eisler, and M. G. Bawendi, “Optical gain and stimulated emission in nanocrystal quantum dots”, Science, 290, 314-317 (2000). Semiconductor and metal nanocrystals, Chap. 3-6, Ed. V. I. Klimov, Marcel Dekker, Inc (2004). P. J. B. Koeck, H. K. Kagawa, M. J. Ellis, H. Hebert, J. D. Trent,” Two-dimensional crystals of reconstituted beta-subunits of the chaperonin TF55 from Sulfolobus shibatae”, Biochimica et Biophysica Acta-Protein Structure and Molecular Enzymology, 1429, 40-44 (1998). S. Kos, M. Achermann, V. I. Klimov, D. L. Smith, “Different regimes of Förster- type energy transfer between an epitaxial quantum well and a proximal monolayer of seminconductor nanocrystals”, Phys. Rev. B 71, 205309 (2005). P. A. Kralchevsky, V. N. Paunov, I. B. Ivanov, K. Nagayama, “Capillary meniscus interaction between colloidal particles attached to a liquid-fluid interface”, J. Colloid Interface Sci., 151, 79-94 (1992). P. A. Kralchevsky, V. N. Paunov, N. D. Denkov, I. B. Ivanov, and K. Nagayama, “Energetical and force approach to the capillary interactions between particles attached to a liquid-fluid interface”, J. Colloid Interface Sci., 155, 420-437 (1993). P. A. Kralchevsky, K. Nagayama “Capillary forces between colloidal particles”, Langmuir 10, 23-36 (1994). H. Kroemer, Chap. 17 in Molecular Beam Epitaxy and Heterostructrues, Eds. L.L. Chang and K. Plog, NATO ASI Series, Martinus, Nijhoff, (1985). M. Kuno, D. P. Fromm, H. F. Hamann, A. Gallagher, D. J. Nesbitt, “Nonexponential "blinking" kinetics of single CdSe quantum dots: A universal power law behavior”, J. Chem. Phys. 112, 3117-3120 (2000). O. Labeau, P. Tamarat, and B. Lounis “Temperature Dependence of the Luminescence Lifetime of Single CdSe=ZnS Quantum Dots”, Phys. Rev. Lett. 90, 257404 (2003). V. K. LaMer, R. H. Dinegar, “Theory, Production and Mechanism of Formation of Monodispersed Hydrosols”, J. Am. Chem. Soc. 72, 4847-4854 (1950). 350 P. Lawaetz, “Valence band parameters in cubic semiconductors”, Phys. Rev. B, 4, 3460-3467 (1971). C. A. Leatherdal, C. R. Kagan, N. Y. Morgan, S. A. Empedocles, M. A. Kastner, M. G. Bawendi, “Photoconductivity in CdSe quantum dot solids”, Phys. Rev. B 62, 2669-2680 (2000). M. J. Lercel, R. C. Tiberio, P. F. Chapman, H. G. Graighead, C. W. Sheen, A. N. Parikh, D. L. Allara, “Self-assembled monolayer electron beam resists on GaAs and SiO 2 ” J. Vac. Sci. Tech. B 11 2823-2828 (1993). Y. Leviatan, “Study of near zone fields of a small aperture”, J. Appl. Phys. 60, 1577- 1583 (1986). Handbook Series on Semiconductor Parameters, Vol. 1, M. Levinshtein, S. Rumyantsev and M. Shur, ed., World Scientific, London, (1996). L. S. Li, A. P. Alivisatos, “Origin and scaling of the permanent dipole moment in CdSe nanorods” Phys. Rev. Lett. 90, 097402 (2003). T. T. T. Li, M. J. Weaver, “Intramolecular electron-transfer at metal-surfaces 4. Dependence of tunneling probability upon donor-acceptor spectration distance”, J. Am. Chem. Soc. 106, 6017-6108 (1984). K. Lieberman, N. Ben-Ami, A. Lewis, “A fully integrated near-field optical, far-field optical, and normal force scanned probe microscope”, Rev. Sci. Instrum. 67 3567- 3572 (1996). I. M. Lifshitz V. V. Slyozov, “Kinetics of precipitation from super saturated solid solutions”, J. of Phys. and Chem. of Solids, 19, 35-50 (1961). L. C. Liu, S. H. Bisbud, “Quantum dot size distribution analysis and precipitation stages in semiconductor doped glass”, J. Appl. Phys. 68, 28-32 (1990). H. Lodish, A. Berk, S .L. Zipursky, P. Matsudaira, D. Baltimore, J. Darnell, Chapter 24, Section 2 and 3 in Molecular Cell Biology, 4 th Edition, Freeman, New York (2001). S. Logunov, T. Green, S. Marguet, M. A. el-Sayed, “Interfacial carriers dynamics of CdS nanoparticles”, J. Phys. Chem. A 102, 5652-5658 (1998). S. Lu, A. Bansal, W. Soussou, T. W. Berger, A. Madhukar, “Receptor-Ligand-Based Specific Cell Adhesion on Solid Surfaces: Hippocampal Neuronal Cells on Bilinker Functionalized Glass“, Nano Lett. 6, 1977-1981 (2006). 351 S. R. Lunt, G. N. Fiyba, P. G. Santangelo, N. S. Lewis, “Chemical studies of the passivation of GaAs surface recombination using surfides and thiols”, J. Appl. Phys. 70, 7449-7465 (1991). J. M. Luttinger, W. Kohn, “Motion of Electrons and Holes in Perturbed Periodic Fields”, Phys. Rev. 97, 869 (1955). J. M. Luttinger, “Quantum Theory of Cyclotron Resonance in Semiconductors: General Theory”, Phys. Rev. 102, 1030 (1956). A. Madhukar, Ch.2 in Nano-Optoelectronics, Ed. M. Grundmann, Springer-Verlag, Berlin, (2002). A. Madhukar, “Surface Modification Engineered Assembly of Novel Quantum Dot Architectures For Advanced Electronics“, Proposal to DoD, (2000), DARPA/AFOSR Grant. No. F49620-01-1-0479. A. Madhukar, S. Lu, A. Konkar, Y. Zhang, M. Ho, S. M. Hughes, A. P. Alivisatos, “Integrated semiconductor nanocrystal and epitaxical nanostructure systems: Structural and optical behavior”, Nano Lett. 5, 479-482 (2005). M. A. Makeev, W. Yu, A. Madhukar, “Stress distributions and energetics in the laterally ordered systems of buried pyramidal Ge/Si (001) islands: An atomistic simulation study”, Phys. Rev. B 68 195301 (2003). L. Manna, E. C. Scher, A. P. Alivisatos, ”Synthesis of soluble and processable rod-, arrow-, teardrop-, and tetrapod-shaped CdSe nanocrystals”, J. Am. Chem, Soc. 122, 12700-12706 (2000). R. A. Marcus, N. Sutin, “Electron transfer in chemistry and biology”, Biochimica et Biophysica Acta, 811, 265-322 (1985). E. Marx, D. S. Ginger, K. Walzer, K. Stokbro, N. C. Greenham, “Self-assembled monolayer of CdSe nanocrystals on doped GaAs substurate“, Nano Lett. 2, 911-914 (2002). N. A. McDonald, "Electric and Magnetic Coupling through Small Apertures in Shield Walls of Any Thickness," IEEE Trans. Microwave Theory Tech. MTT-20, 689 (1972). A. E. McKinnon, A. G. Szabo, D. R. Miller, “Deconvolution of photoluminescence data”, J. Phys. Chem. 81, 1564-1570 (1977). 352 R. A. McMillan, C. D. Paavola, J. Howard, S. L. Chan, N. J. Zaluzec, J. D. Trent, “Ordered nanoparticle arrays formed on engineered chaperonin protein templates”, Nature Materials, 1, 247-252 (2002). I. L. Medintz, A. R.Clapp, H. Mattoussi, E. R. Goldman, B. Fiber, J. M. Mauro, “Self-assembled nanoscale biosensors based on quantum dot FRET donors”, Nature Materials, 2, 630-638 (2003). M. Meyer, C. Wallberg, K. Kurihara, J. H. Fendler, “Photosensitized charge separation and hydrogen production in reversed micelle entrapped platinized colloidal cadium sulfide. J. Chem. Soc. Commun. 88, 969-977 (1984). O. I. Micic and A. J. Nozik, “Synthesis and characterization of binary and ternary III-V quantum dots. J. Lumin. 70, 95-107 (1996). O. I. Micic, B. B Smith, A. J. Nozik, “Core-shell quantum dots of lattice mathced ZnCdSe 2 shells on InP cores: experiments and theory”, J. Phys. Chem. B, 104, 12149-12156 (2000). T. Mokari, E. Rothenberg, I. Popov, R. Costi, U. Banin, “Selective growth of metal tips onto semiconductor quantum rods and tetrapods”, Science 304, 1787-1790 (2004). I. Mukhametzhanov, “Growth control, structural characterization, and electronic structure of Stranski-Krastanow InAs/GaAs (001) quantum dots”, Ph.D. Disseration, University of Southern California (2002). C. B. Murray, D. J. Norris, M. G. Bawendi, “Synthesis and characterization of nearly monodisperse CdE (E=S, Se, Te) semiconductor nanocrystallites.” J. Am. Chem. Soc. 115, 8706-8715 (1993). C. B. Murray, C. R. Kagan, M. G. Bawendi, “Synthesis and characterization of monodisperse nanocrystals and close-packed nanocrystal assemblies”, Ann. Rev. Mater. Sci. 30, 545-610 (2000). S. V. Nair, L. M. Ramaniah, K. C. Rustagi, “Electron-states in a quantum dot in an effective-bond-orbital model”, Phys. Rev. B 45, 5969-5979 (1992). O. S. Nakagawa, S. Ashok, C. W. Sheen, J. Martensson, D. L. Allara, ”GaAs interfaces with octadecyl thiol self-assembled monolayer – structural and electrical- properites”, J. J. Appl. Phys. Part 1 30, 3759-3762 (1991). S. Narayanan, J. Wang and X. M. Lin, “Dynamical self-assembly of nanocrystal superlattices during colloidal droplet evaporation by in situ small angle x-ray scattering” Phys. Rev. Lett. 93, 135503 (2004). 353 R. G. Neuhauser, K. T. Shimizu, W. K. Woo, S. A. Empedocles, M. G. Bawendi, “Correlation between fluorescence intermittency and spectral diffusion in single semiconductor quantum dots”, Phys. Rev. Lett. 85, 3301-3304 (2000). M. Nirmal, D. J. Norris, M. Kuno, M. G. Bawendi, Al. Efros, M. Rosen, “Observation of the dark exciton in CdSe quantum dot”, Phys. Rev. Lett. 75 3728- 3731 (1995). M. Nirmal, B. O. Dabbousi, M. G. Bawendi, J. J. Macklin, J. K. Trautman, T. D. Harris, L. E. Brus, “Fluorescence intermittency in single cadmium selenide nanocrystals”, Nature, 383, 802-804 (1996). D. J. Norris, “Electronic structure in semiconductor nanocrystals”, Chap. 2 in Semiconductor and Metal Nanocrystals, Ed. V. I. Klimov, Marcel Dekker Inc., New York (2004). D. V. O’Conner, W. R. Ware, J. C. Andre, “Deconvolution of fluorescence decay curves – Critical comparison of techniques”, J. Phys. Chem. 83, 1333-1343 (1979). Near-field nano/atom optics and technology, Ed. M. Ohtsu, Springer 1998. H. Oigawa, J. F. Fan, Y. Nannichi, H. Hugahara, M. Oshima, “Universal passivation effect of (NH4)2Sx treatment of the surface of III-V compound semiconductors”, J. J. Appl. Phys. Part 2-Letters 30, L322-325 (1991). K. C. Olbrich, T. T. Andersen, F. A. Blumenstock, R. Bizios, “Surfaces modified with covalently-immobilized adhesive peptides affect fibroblast population motility” Biomaterials 17, 759-764 (1996). M. A. Olshavsky, A. N. Goldstein, A. P. Alivisatos, ”Organometallic synthesis of GaAs crystallites exhibiting quantum confinement”, J. Am. Chem. Soc. 112, 9438- 9439 (1990). S. H. Park, D. Qin, Y. Xia, “Crystallization of mesoscale particles over large areas”, Adv. Mater. 10, 1028 (1998). S. H. Park, Y. N. Xia, “Assembly of mesoscale particles over large areas and its application in fabricating tunable optical filters”, Langmuir 15, 266-273 (1999). V. N. Paunov, P. A. Kralchevsky, N. D. Denkov, K. Nagayama, “Lateral capillary forces between floating submillimeter particles”, J. Colloid Interface Sci., 157, 100- 112 (1993). S. Pelet, M. J. R. Previte, L. H. Laiho, P. T. C. So “A fast global fitting algorithm for fluorescence lifetime imaging microscopy based on image segmentation”, Biophys. J. 87, 2807-2817 (2004). 354 X. Peng, J. Wickham, A. P. Alivisatos, “Kinetics of II-VI and III-V Colloidal Semiconductor Nanocrystal growth: “Focusing” of Size Distribution”, J. Am. Chem. Soc. 120, 5343 (1998). X. Peng, M. C. Schlamp, A. V. Kadavanich, A. P. Alivisatos, “ Epitaxial growth of highly luminescent CdSe/CdS core/shell nanocrystals with photostability and electronic accessibility”, J. Am. Chem. Soc. 119, 7019-7029 (1997). J. Perez-Conde, A. K. Bhattacharjee, “Electronic structure of CdTe nanocrystals: a tight-binding study”, Solid Stat. Comm. 110, 259-264 (1999). J. Perrin. Ann. Chim. Phys. 18, 1 (1909). C. R. Pidgeon, R. N. Brown, “Interband magneto-absorption and Faraday rotation in InSb”, Phys. Rev. 146, 575-& (1966). V. S. Polikov, P. A. Tresco, W. M. Reichert, “Response of brain tissue to chronically implanted neural electrodes” J. Neurosci. Meth. 148, 1-18 (2005). S. S. Prabhu, A. S. Vengurlekar, S. K. Roy, J. Shah, “Nonequilibrium dynamics of hot carriers and hot phonons in CdSe and GaAs”, Phys. Rev. B. 51 14233-14246 (1995). W. H. Press, B. P. Flannery, S. A. Teukolsky, W. T. Vetterling, §13.1 in Numerical Recipes in C, 2nd Edition, Cambridge Press (1992). W. H. Press, B. P. Flannery, S. A. Teukolsky, W. T. Vetterling, §15.5 in Numerical Recipes in C, 2nd Edition, Cambridge Press (1992). K. L. Prime, G. M. Whiteside, “Self-assembled organic monolayers – model system for studying adsorption of proteins at surface”, Science 252, 1164-1167 (1991) multinanoparticle structures using dithiols and scanning force microscopy”, J. Phys. Chem. B, 103 3647-3650 (1999). D. Pum, A. Neubauer, E. Gyorvary, M. Sara, and U. B. Sleytr, “S-layer proteins as basic building blocks in a biomolecular construction kit”, Nanotechnology 11, 100- 107 (2000). E. Rabani, D. R. Reichman, PL Geissler, L. E. Brus, “Drying-mediated self- assembly of nanoparticles”, Nature 426, 271-274 (2003). T. R. Ramachandran, “Scanning probe microscopy studies of the highly strained epitaxy of InAs on GaAs(001) and Scanning Probe Based Imaging and Manipulation of Nanoscale Three-Dimensional Objects”, Ph. D. Dissertation, University of Southern California (1998). 355 J. P. Ranieri, R. Bellamkonda, E. J. Bekos, T. G. Vargo, J. A. Gardella, P. Aebischer, “Neuronal cell attachment to fluorinated ethylene propylene films with covalently immobilized laminin oligopeptides YIGSR and IKVAV.” J. Biomed. Mater. Res. 29, 779–785 (1995). Y. Rao, X. F. Wu, J. Gariepy, U. Rutishauser, C. H. Siu, “Identification of a peptide sequence involved in homophilic binding in the neural cell-adhesiion molecule NCAM”, J. Cell Biol., 118, 937-949 (1992). M. A. Ratner, A. Madhukar, “On the role of nuclear motions in electron and excitation transfer rates: Importance of transfer-integral dependence upon nuclear coordinate”, Chem. Phys. 30, 201-215 (1978). H. Reiss, “The Growth of Uniform Colloidal Dispersions”, J. Chem. Phys. 19, 482- 487 (1951). R. Resch, C. Baur, A. Bugacov, B. E. Koel, P. M. Echternach, A. Madhukar, N. Montoya, A. A. G. Requicha, P. Will, “Linking and manipulation of gold multinanoparticle structures using dithiols and scanning force microscopy,” J. Phys. Chem. B 103, 3647-3650 (1999). A. V. Rodina, A. Y. Alekseev, A. L. Efros, M. Rosen, B. K. Meyer, “General boundary conditions for the envelope function in the multiband k·p model”, Phys. Rev. B 65, 125302 (2002). R. Rossetti, L. Brus, “Electron-hole recombination emission as a probe of surface chemistry in aqueous CdS colloids”, J. Phys. Chem. 86, 4470-4472 (1982). E. Ruoslahti, “RGD and other recognition sequences or integrins” Annu. Rev. Cell Dev. Biol., 12, 697–715 (1996). P. Roussignol, D. Ricard, J. Lukasik, C. Flytzanis, “New results on optical-phase conjugation in semiconductor-doped glasses”, J. Opt. Soc. Am. B, 4, 5-13 (1987). J. Sagiv, “Organized monolayers by adsorption. 1. Formation and structure of oleophobic mixed monolayers on solid-surfaces.”, J. Am. Chem. Soc. 102, 92-98 (1980). S. Saneinejad, M. S. Shoichet, “Patterned glass surfaces direct cell adhesion and process outgrowth of primary neurons of the central nervous system” J. Biomed. Mater. Res. 42, 13-19 (1998). 356 J. C. Sarup, R. M. Johnson, K. L. King, B. M. Fendly, M. T. Lipari, M. A. Napier, A. Ullrich, H. M. Shepard, “Characterization of an anti-p185HER2 monoclonal antibody that stimulates receptor function and inhibits tumor cell growth.” Growth Regul. 1, 72-82. (1991). P. C. Sercel, K. J. Vahala, “Analytical formalism for determining quantum-wire and quantum-dot band structure in the multiband envelop-function approximation”, Phys. Rev. B 42, 3690-3710 (1990). J. Shah, Ultrafast Spectroscopy of Semiconductors and Semiconductor Nanostructures, Springer-Verlag (1999). V. I. Sheka, D. I. Sheka, “Local states in a semiconductor with a narrow forbidden band”, Sov. Phys. JETP 24, 975 (1967). K. T. Shimizu and M. G. Bawendi, “Optical dynamics in single semiconductor quantum dots”, Chap. 6 in Semiconductor and metal nanocrystals, Ed. V. I. Klimov, Marcel Dekker, Inc (2004). M. Shimomura, T. Sawadaishi, “Bottom-up strategy of materials fabrication: a new trend in nanotechnology of soft materials”, Curr. Opin. Colloid. Interf. Sci. 6, 11-16 (2001). K. Sokolov, M. Follen, R. Richards-Kortum, “Optical Spectroscopy for Detection of Neoplasia”, Curr. Opin. Chem. Biol., 6, 651-658 (2002). D. H. Son, S. M. Hughes, Y. D. Yin, A. P. Alivisatos, “Cation exchange reactions-in ionic nanocrystals” Science 306, 1009-1012 (2004). S. Song, R. A. Clark, E. F. Bowden, M. J. Tarlov, “Characterization of Cytochrome- C Alkanethiolate structures prepared by self-assembly on gold”, J. Phys. Chem. 97 6564-6572 (1993). M. Stavola, D. L. Dexter and R. S. Knox, “Electron-hole pair excitation in semiconductor via energy transfer from an external sensitizer”, Phys. Rev. B 31, 2277-2289 (1985). P. S. Stayton, J. M. Olinger, S. T. Wollman, P. W. Bohn, S. G. Sligar, ”Engineering proteins for electrooptical biomaterials”, Chap. 19 in Molecular and Biomolecular Electronics, Ed. R. R. Birge, American Chemical Society, Washington DC (1994). M. L. Steigerwald, A. P. Alivisatos, J. M. Gibson, T. D. Harris, R. Kortan, A. J. Muller, A. M. Thayer, T.M. Duncan, D. C. Douglass, L. E. Brus, “Surface derivatization and isolation of semiconductor cluster molecules.” J. Am. Chem. Soc. 110, 3046-3050 (1988). 357 O. Stier, M. Grundmann, D. Bimberg “Electronic and optical properties of strained quantum dots modeled by 8-band k·p theory”, Phys. Rev. B 59, 5688-5701 (1999). C. Stowell, B. A. Korgel, “Self-assembled honeycomb networks of gold nanocrystals”, Nano Lett. 1, 595-600 (2001). T. Sugimoto, “Preparation of Monodisperse Colloidal Particles”, Adv. Colloidal Interfac. Sci, 28, 65-108, (1987). Y. P. Sun, K. F. Fu, Y. Lin, W. J. Huang, “Functionalized carbon nanotubes: Properties and applications”, Accounts Chem. Res. 35, 1096-1104 (2002). E.H. Synge, "A suggested method for extending the microscopic resolution into the ultramicroscopic region" Phil. Mag. 6, 356 (1928). E.H. Synge, "An application of piezoelectricity to microscopy", Phil. Mag., 13, 297 (1932). D. V. Talapin, A. L. Rogach, M. Haase, H. Weller, “Evolution of an Ensemble of Nanoparticles in a Colloidal Solution: Theoretical Study”, J. Phys. Chem. B 105, 12278-12285 (2001). A. R. Tan, S. M. Swain. "Ongoing adjuvant trials with trastuzumab in breast cancer". Seminars in Oncology 30 (5 Suppl 16), 54–64 (2002). K. Tashiro, G. C. Sephel, B. Weeks, M. Sasaki, G. R. Martin, H. K. Kleinman, Y. Yamada, “A synthetic peptide containing the IKVAV sequence from the A-chain of laminin mediates cell attachment, migration, and neurite outgrowth”, J. Biol. Chem. 264, 16174-16182 (1989). S. H. Tolbert, A. P. Alivisatos, “High-pressure structural transformation in semiconductor nanocrystals”, Ann. Rev. Phys. Chem. 46, 595-625 (1995). M. Trau, D. A. Saville, I. A. Aksay, “Field-induced layering of colloidal crystals”, Science 272, 706-709 (1996). H. Uchida, C. J. Curtis, A. J. Nozik, “GaAs nanocrystals prepared in quinoline.” J. Phys. Chem. 95, 5382-5384 (1991). A. Ulman, An Introduction to Ultrathin Organic Films: From Langmuir--Blodgett to Self—Assembly, Academic Press, San Diego (1991). T. J. Ulrych, T. N. Bishop, ”Maximum entropy spectral analysis and autoregressive decomposition”, Rev. Geophys. 13, 183-200 (1975). 358 A. V. Uskov, A. P. Jauho, B. Tromborg, J. Mork, R. Lang, “Dephasing times in quantum dots due to elastic LO phonon-carrier collisions”, Phys. Rev. Lett. 85, 1516-1519 (2000). K. J. Vahala, P. C. Sercel, “Application of a total-angular-momentum bassis to quantum-dot bandstructure”, Phys. Rev. Lett. 65: 239-242 (1990). T. Vallant, H. Brunner, U. Mayer, H. Hoffmann, T. Leitner, R. Resch, G. Friedbacher, “Formation of self-assembled octadecylsiloxane monolayer on mica and silicon surfaces studied by atomic force microscopy and infrared spectroscopy“, J. Phys. Chem. B. 102, 7190-7197 (1998). W. G. J. H. M. van Sark, “Enhancement of solar cell performance by employing planar spectral converters”, Appl. Phys. Lett. 87, 151117 (2005). J. T. Verdeyen, Laser Electronics, Prentice Hall (1995). V. Vogel, M. Sheetz, “Local force and geometry sensing regulate cell functions” Nat. Rev. Molec. Cell Biol. 7, 265-275 (2006). H. H. vonGrunberg, “Energy levels of CdSe quantum dots: Wurtzite versus zinc- blende structure”, Phys. Rev. B. 55, 2293-2302 (1997). Y. Wang, N. Herron, “Photoconductivity of CdS nanocluster-doped polymers”, Chem. Phys. Lett. 200, 71-75 (1992). L. W. Wang, A. Zunger, “ High-energy excitonic transitions in CdSe quantum dots”, J. Phys. Chem. B, 102, 6449-6454 (1998). S. R. Wasserman, Y. T. Tao, G. M. Whitesides, “Structure and reactivity of alkylsiloxane monolayers formed by reation of alkyltrichlorosilanes on silicon substrates”, Langmuir 5, 1074-1087 (1989). S. R. Wasserman, G. M. Whitsides, I. M. Tidswell, B. M. Ocko, P. S. Pershan, J. D. Axe, “The structure of self-assembled monolayers of alkylsiloxanes on silicon – A comparison of results from elliposmetry and low angle X-ray reflectivity”, J. Am. Chem. Soc. 111, 5852-5861 (1989). J. W. Weiland, W. Liu, M. S. Humayun, “Retinal prosthesis”, Annu. Rev. Biomed. Eng. 7, 361-401 (2005). R. L. Wells, C. G. Pitt, A. T. McPhail, A. P. Purdy, S. Shafieezad, and R. B. Hallock, “Use of Tri(trimethylsilyl)arsine To Prepare Gllium Arsenide and Indium Arsenide”, Chem. Mater., 1, 4-6 (1989). 359 J. Wickham, “Investigation of the Mechanism of Pressure-Induced Solid-Solid Structural Transformations in Semiconductor Nanocrystals”, Ph. D. Dissertation, University of California at Berkeley (2001). J. L. Wilbur, G. M. Whitesides, “Self-assembly and self-assembled monolayers in micro- and nanofabrication” Chapter 8 in Nanotechnology, Ed. G. Timp, Springer- Verlag, New York (1999). A. J. Williamson, A. Zunger, “InAs quantum dots: Predicted electronic structure of free-standing versus GaAs-embedded structures”, Phys. Rev. B 59, 15819-15824 (1999). G. H. Wu, R. H. Datar, K. M. Hansen, T. Thundat, R. J. Cote, A. Majumdar, “Bioassay of prostate-specific antigen (PSA) using microcantilevers”, Nature Biotechnol. 19, 856-860 (2001). J. B. Xia, “Electronic structure of zero-dimensional quantum wells”, Phys. Rev. B 40, 8500-8507 (1989). Q. H. Xie, P. Chen, A. Madhukar, “InAs island-induced-strain driven adatom migration during GaAS overlayer growth”, Appl. Phys. Lett. 65, 2051-2053 (1994). Q. H. Xie, A. Madhukar, P. Chen. N. P. Kobayashi, “Vertically self-organized InAs quantum box island on GaAs (001)”, Phys. Rev. Lett. 75, 2542-2545 (1995). S. Xu, A. A. Mikhailovsky, J. A. Hollingsworth, V. I. Klimov, “Hole intraband relaxation in strongly confined quantum dots: Revisiting the "phonon bottleneck" problem”, Phys. Rev. B 65, 045319 (2002). S. R. Yeh, M. Seul, B. I. Shraiman, “Assembly of ordered colloidal aggregates by electric-field-induced fluid flow”, Nature 386, 57-59 (1997). H. Yoshimura, S. Endo, M. Matsumoto, K. Nagayama, Y. Kagawa, “Hexagnal Structure of two-dimensional crystals of the alpha-3-beta-3 complex of thermophilic ATP synthase”, J. Biochem., 106, 958-960 (1989). H. Yoshimura, S. Endo, M. Matsumoto, K. Nagayama, “2-dimensional crystallization of proteins on mercury” Ultramicroscopy, 32, 265-274 (1990). O. Zhong, D. Inniss, K. Kjoller, V. B. Elings, ”Fractured polymer silica fiber surface studied by tapping mode atomic-force microscopy”, Surf. Sci. 290, L688-L692 (1993). 360 Y. Zhong, X. Yu, R. Gilbert, R. V. Bellamkonda, “Stabilizing electrode-host interfaces: A tissue engineering approach”, J. Rehabil. Res. Dev. 38, 627-632 (2001). Y. H. Zhu, X. W. Zhang, J. B. Xia, “Electronic States in InAs quantum spheres and ellipsoids”, Phys. Rev. B 73, 165326 (2006). 361 Appendix A: Synthesis of (TMS) 3 As This procedure is adapted from reference [J. Wickham, Ph. D. Disseration, University of California at Berkeley (2001), pp 44-49] and modified with this author’s experience. Material: Arsenic metal (power 99.999+% Aldrich); Sodium Potassium Alloy (NaK, 22%/78%, Aldrich); Trimethylsilyl chlorine (TMS-Cl, Aldrich 99% redistilled); ethylene glycol dimethylether (DME, anhydrous 99.5%, Aldrich ); Calcium Hydride (CaH, powder, 90-95%, Aldrich). The synthesis of (TMS) 3 As is a four day procedures as described below. Be aware that the component materials involved are highly volatile, toxic and pyrophoric. Extreme care needs to be taken in their handling. Refer to the MSDS of NaK before proceeding. NaK, TMS-Cl and (TMS) 3 As are highly oxygen sensitive and moisture sensitive. Hence all possible measures for air-sensitive techniques should be employed. Preferably the glovebox should be regenerated before the synthesis to ensure the minimum oxygen and moisture level. Glassware were rinsed with acetone and baked in 120 C oven before usage. Day 1. Reduction of As metal to As 3- using NaK, Preparation for TMS-Cl distillation. Inside glovebox, measure 125ml of DME into a 250ML 3-neck round bottom flask. Weigh 2.5g of NaK and added into DME over a few minutes while stirring. Slowly add 1.9g of Arsenic powder at rate of ~0.1g/min. If the temperature of the DME flask rises, the addition needs to slow down. Arsenic 362 should be dissolved and the solution turns black and quite viscous. Attach a vacuum adaptor with stopcock to the main neck of the flask; attach a vacuum adaptor with stopcock and a glass thermal couple adaptor to the other two necks of the flask. Close the stopcocks. Properly quench and dump all the wastes, including glass pipette contaminated with Arsenic and NaK. (To quench NaK, first add lots of toluene, then quench with methanol. Toluene has a high flash point and does not explode easily). Take the flask containing DME, Arsenic and NaK out of glovebox. Attach a condenser to the flask via the adaptor on the main neck. Attach the condenser to the Schlenk line and a bubbler for standard refluxing. Pump/purge the condenser three times before open the stopcock connecting the flask and the condenser. Circulate cooling water in the condenser. Make sure that the solution is stirring. Slowly heat the solution to 85-88 C, until the solution is refluxing. Leave the solution for refluxing overnight. Make sure the setup is air-tight and the stirrer will not break the glassware (specially the thermal couple adaptor). In preparation for the distillation of TMS-Cl on day 2, assemble a standard Ar protected distillation apparatus. (Note: insert a vacuum adaptor with Teflon stopcock between the collection flask and the rest of the distillation apparatus.) Add two spoons of CaH and 25ml of TMS-Cl into a 50ml round bottom flask with a sidearm valve. Connect the flask to the distillation apparatus under positive Ar pressure. Stir the TMS-Cl overnight under Ar to remove any water present. 363 Day 2: Distill TMS-Cl, Add distilled TMS-Cl to Ar 3- solution to form (TMS) 3 As. (Byproducts are NaCl and KCl precipitates). Heat the flask containing TMS-Cl under Ar to distill TMS-Cl. TMS-Cl comes out at ~57 C. Discard the first 5ml, collect the rest in the collection flask. After finishing, close the stopcock between the collection flask and the rest of the distillation apparatus. Disconnect the collection flask (with the vacuum adaptor) from the distillation apparatus and cap it with a rubber septum. Reduce the temperature of the Ar 3- solution to 50 C. Pump/purge an addition funnel (50ml) capped with a rubber septum. Attach the addition funnel to the vacuum adaptor on the side neck of the Ar 3- containing flask. Close the stopcock of the funnel. Flush a glass syringe (>15ml capacity) with Ar. Withdraw 12ml of distilled TMS-Cl and inject into the addition funnel. Open the stopcock of the vacuum adaptor between addition funnel and the 3-neck flask. Slowly open the addition funnel stopcock, so that the TMS-Cl is added dropwise over 30mins to the 3-neck flask containing As 3- to form (TMS) 3 As. The addition of the TMS- Cl should be slow enough to keep the reaction solution temperature between 45 C to 55 C. After ~5ml of TMS-Cl is added, the reaction solution turns into green with lots of precipitation (NaCl and KCl salt). Adjust to keep stirring gently. The reaction solution changes to orange to red to brown as the TMS-Cl addition progress. Final solution color is brownish. After addition is completed, close the vacuum adaptor stopcock, remove the addition funnel. Raise the reaction solution temperature back to 85-88 C and leave stirring overnight. 364 Day 3. Separate DME/(TMS) 3 As solution from the KCl and NaCl salt precipitation via vacuum filtration in the glovebox. In order to move the 3-neck flask containing (TMS) 3 As into the glovebox for filtration, the solution must be degassed. Otherwise when pumping the flask in the antechamber the imbalance in pressures could cause an explosion, if the flask is not sealed extremely tightly. A freeze-pump-thaw procedure is used for degassing as described below. Cooled the 3-neck flask containing the mixture of DME/(TMS)3As and NaCl and KCl precipitation (hereafter referred to mixture) to room temperature. While flowing Ar, immerse the 3-neck into a dewar filled with LN2, until DME solution is completely frozen. Install dewars on both traps of the Schlenk line. Fill the two dewars COMPLETELY with LN2. Pull vacuum on the mixture, remove the LN2 dewar under the 3-neck flask and let the mixture start to thaw and warm up. It is very important to keep the dewars under the traps filled with LN2 during the whole process. As mixture warms up, some DME will be pulled into the traps by vacuum due to its high volatility. If only the bottom of trap is in contact with LN2, the DME will collect at the bottom, clog the trap and cut off the vacuum. However, if the entire surface of the trap is kept cold, the DME will freeze over the entire surface area, leaving a passageway for the vacuum. As the mixture warms up, bubbles of gas (dissolved Ar) should come from the liquid. As soon as no more gas bubbles are visible and the outside of the flask is dry, close the stopcock of the vacuum adaptor on the main neck of the 3-neck flask so that the flask remains under vacuum. Transfer the flask into the glovebox. 365 Note when thawing the mixture, at around -10 C (depending on vacuum strength), the solution will start to bubble violently. This means that the DME is boiling. Do not continue to put vacuum on the warm flask for a long time, otherwise the evaporated DME will clog the trap. If the vacuum level during the thaw process becomes suspiciously high, check if the traps are clogged. If clogged clean up the trap and do the freeze-pump-thaw again. Do not run the risk of the 3-neck flask explode in the antechamber of the glovebox. Author recommends that before transferring the 3-neck flask into the glovebox, put it in a secondary container and wrap up with plastic bag, so that the contamination will be confined, even in case the flask explodes or breaks during the transfer. Connect the vacuum line to the glovebox. Vacuum filter the mixture of DME/(TMS) 3 As and NaCl and KCl precipitation through a 350ml fritted glass Buchner funnel. The filtered solution (DME/(TMS) 3 As) should be bright yellow. Flush the funnel with extra DME to get all the (TMS) 3 As out. Try to keep the oxygen level in the glovebox as low as possible. If there is excess oxygen present, the filtrate will turn from yellow to red over time. (At 4ppm O2 level, it takes only a few minutes for solution to turn red.) Pour the filtrate through a funnel into a 250ml round bottom flask with sidearm valve and put in a stir bar. (Note pipette the filtrate into the flask is not a good choice, because it takes too long to complete.) Attach a vacuum adaptor to the 250ml flask, close the stopcock. Leave the flask in the glovebox for tomorrow’s usage. Put all Arsenic contaminated glassware into a secondary container and wrap with plastic bag. Purge the glovebox with Ar to clean up arsenic contamination of the glovebox atmosphere before take the glassware out of the 366 glovebox. Otherwise the glovebox atmosphere comes out the antechamber will contaminate lab air. Dispose the filtered solids and wash all glassware (including the traps). Note these are arsenic contaminated so handle with extreme care in the fumehood. All the glasswares and leftover DME solution contaminated with arsenic on exposure to air will turn into brownish and smell terrible. Day 4. Separate (TMS) 3 As from DME solution via vacuum distillation In day 4, the (TMS)3As is separated from DME solution via vacuum distillation. To order to establish a good vacuum during the distillation, freeze- pump-thaw needs to be applied again before the distillation to degas the solution. During the vacuum sequence of freeze-pump-thaw, most of the DME is removed. The remaining DME is distilled off. Setup the apparatus for standard vacuum distillation. Connect to distillation apparatus two 25ml flasks (via vacuum adaptor with Teflon stopcock) to collect (TMS) 3 As and one 250ml flask to collect DME. It is useful to use a long neck 250ml flask to facilitate LN2 cooling. Do not insert vacuum adaptor between this 250ml flask and the distillation apparatus. Take the flask containing DME/(TMS) 3 As (prepared on day 3) out of the glovebox and connect it to the distillation apparatus under positive Ar pressure. Perform freeze-pump-thaw degassing of the DME/(TMS) 3 As solution as described for day 3. The only difference is that most of the excess DME should be collected in the large 250ml collection flask. In order to do so, the 250ml collection flask must be submerged in LN2 (as completely as possible). As described for day 3, both traps also have to be completely submerged in LN2 to 367 prevent them from being clogged. Most of the DME should be transferred to the collection flask by the time the freeze-pump-thaw is finished and the DME/(TMS) 3 As solution returns to room temperature. Start to distill (TMS) 3 As by slowly heating the flask containing the remaining DME/(TMS) 3 As under vacuum. The DME will come off first. Be sure to keep the DME collection flask submerged in LN2 all the time, otherwise the traps will clog. The (TMS) 3 As should begin to come off around 45 C and 50 C depending on the vacuum strength. Collect (TMS) 3 As in the 25ml collection flask. The (TMS) 3 As should be colorless liquid. After all (TMS) 3 As is collected, close the stopcock on the vacuum adaptor to the (TMS) 3 As collection flask (leave the (TMS) 3 As under vacuum). Detached the collection flask from the distillation apparatus and pump it into the glovebox. The prepared (TMS) 3 As should be stored in an amber vial in the refrigerator, as it is light sensitive. The above reaction should produce ~3 grams of (TMS) 3 As product. If more product is required, one can double the amount of this synthesis reaction. As on day 3, clean up all arsenic contaminated glassware with caution in the fumehood. 368 Appendix B: Angular part of the envelop wavefunction for electron and hole states in spherical quantum dots. Adapted from reference [2.55]: V. I. Sheka, D. I. Sheka, “Local states in a semiconductor with a narrow forbidden band”, Sov. Phys. JETP 24, 975 (1967) For even states of total angular momentum F and z projection Z F : ⎟ ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎜ ⎝ ⎛ ⋅ − ⋅ + = Ω + − − − 2 1 , 2 1 2 1 , 2 1 2 / ) ( 2 / ) ( Fz F Z Fz F Z C Y F F F Y F F F , ⎟ ⎟ ⎟ ⎟ ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎜ ⎜ ⎜ ⎜ ⎝ ⎛ ⋅ + + + + − ⋅ + + − ⋅ + − + ⋅ + − + − + = Ω + + + + − + − + 2 3 , 2 1 2 1 , 2 1 2 1 , 2 1 2 3 , 2 1 1 1 ) 2 )( 1 )( ( 3 ) 1 ( ) 3 ( ) 1 ( ) 3 ( ) 2 )( 1 )( ( 3 Fz F Z Fz F Z Fz F Z Fz F Z h Y Fz F Fz F F F i Y F F Fz F Y F F Fz F i Y Fz F Fz F F F N ⎟ ⎟ ⎟ ⎟ ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎜ ⎜ ⎜ ⎜ ⎝ ⎛ ⋅ − − − − − − ⋅ − − − + ⋅ + − + − ⋅ − + − + + − = Ω + − + − − − − − 2 3 , 2 3 2 1 , 2 3 2 1 , 2 3 2 3 , 2 3 2 2 ) 2 )( 1 )( ( ) 1 )( )( ( 3 ) )( 1 )( ( 3 ) 2 )( 1 )( ( Fz F Z Fz F z z Z Fz F z Z z Fz F Z h Y Fz F Fz F F F i Y F F F F F F Y F F F F F F i Y Fz F Fz F F F N ⎟ ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎜ ⎝ ⎛ ⋅ + + + − ⋅ + + − = Ω + + − + 2 1 , 2 1 2 1 , 2 1 ) 1 ( 2 / ) 1 ( ) 1 ( 2 / ) 1 ( Fz F Z Fz F Z S Y F F F i Y F F F For odd states of total angular momentum F and z projection Z F : ⎟ ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎜ ⎝ ⎛ ⋅ + + + ⋅ + + − − = Ω + + − + 2 1 , 2 1 2 1 , 2 1 ) 1 ( 2 / ) 1 ( ) 1 ( 2 / ) 1 ( Fz F Z Fz F Z C Y F F F Y F F F 369 ⎟ ⎟ ⎟ ⎟ ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎜ ⎜ ⎜ ⎜ ⎝ ⎛ ⋅ − − − + + − ⋅ − + + ⋅ + + − − ⋅ − + + − + = Ω + − + − − − − − 2 3 , 2 1 2 1 , 2 1 2 1 , 2 1 2 3 , 2 1 1 1 ) 1 )( )( 1 ( 3 ) ( ) 1 3 ( ) ( ) 1 3 ( ) 1 )( 1 )( ( 3 Fz F Z Fz F Z Fz F Z Fz F Z h Y Fz F Fz F F F i Y F F Fz F Y F F Fz F i Y Fz F Fz F F F N ⎟ ⎟ ⎟ ⎟ ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎜ ⎜ ⎜ ⎜ ⎝ ⎛ ⋅ + + + + + + ⋅ + + + + + − ⋅ + − + − + + − ⋅ + − + − + − − = Ω + + + + − + − + 2 3 , 2 3 2 1 , 2 3 2 1 , 2 3 2 3 , 2 3 2 2 ) 3 )( 2 )( 1 ( ) 2 )( 1 )( 1 ( 3 ) 1 )( 2 )( 1 ( 3 ) 3 )( 2 )( 1 ( Fz F Z Fz F Z Z Z Fz F Z Z z Fz F Z h Y F F Fz F Fz F i Y F F F F F F Y F F F F F F i Y Fz F Fz F F F N ⎟ ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎜ ⎝ ⎛ ⋅ − ⋅ + = Ω + − − − 2 1 , 2 1 2 1 , 2 1 2 / ) ( 2 / ) ( Fz F Z Fz F Z S Y F F F i Y F F F 370 Appendix C: Breast Cancer Cell Line MDA-MB231 and SK-BR-3 Cell Culture Protocol §C.1 MDA-MB231 and SK-BR-3 Cell Culture Protocol recommended by America Type Tissue Culture Collection (ATCC). (The following protocols are reproduced from ATCC literature.) I. MDA-MB-231 subculture protocol (taken from America Type Tissue Culture Collection) Propagation: 1. ATCC complete growth medium: Leibovitz's L-15 medium with 2 mM L- glutamine, 90%; fetal bovine serum, 10% 2. Temperature: 37.0 C 3. Atmosphere: air, 100% Subculturing Protocol: 1. Remove and discard culture medium. 2. Briefly rinse the cell layer with 0.25% (w/v) Trypsin-0.53mM EDTA solution to remove all traces of serum that contains trypsin inhibitor. 3. Add 2.0 to 3.0 ml of Trypsin-EDTA solution to flask and observe cells under an inverted microscope until cell layer is dispersed (usually within 5 to 15 minutes). Note: To avoid clumping do not agitate the cells by hitting or shaking the flask while waiting for the cells to detach. Cells that are difficult to detach may be placed at 37°C to facilitate dispersal. 4. Add 6.0 to 8.0 ml of complete growth medium and aspirate cells by gently pipetting. 5. Add appropriate aliquots of the cell suspension to new culture vessels. 6. Incubate cultures at 37°C without CO 2 . Subcultivation ratio: A subcultivation ratio of 1:2 to 1:4 is recommended Medium renewal: 2 to 3 times per week Preservation: 1. Freeze medium: Complete growth medium supplemented with 5% (v/v) DMSO 2. Storage temperature: liquid nitrogen vapor phase II. SK-BR-3 subculture protocol (taken from America Type Tissue Culture Collection) Propagation: 1. ATCC complete growth medium: McCoy's 5a medium (modified) with 1.5 mM L-glutamine adjusted to contain 2.2 g/L sodium bicarbonate, 90%; fetal bovine serum, 10% 2. Temperature: 37.0C 3. Atmosphere: air, 95%; carbon dioxide (CO 2 ), 5% Subculturing Protocol: 1. Remove and discard culture medium. 2. Briefly rinse the cell layer with 0.25% (w/v) Trypsin, 0.53 mM EDTA solution to remove all traces of serum which contains trypsin inhibitor. 371 3. Add 2.0 to 3.0 ml of Trypsin-EDTA solution to flask and observe cells under an inverted microscope until cell layer is dispersed (usually within 5 to 15 minutes). Note: To avoid clumping do not agitate the cells by hitting or shaking the flask while waiting for the cells to detach. Cells that are difficult to detach may be placed at 37°C to facilitate dispersal. 4. Add 6.0 to 8.0 ml of complete growth medium and aspirate cells by gently pipetting. 5. Add appropriate aliquots of the cell suspension to new culture vessels. 6. Incubate cultures at 37°C. Subcultivation ratio: A subcultivation ratio of 1:2 is recommended Medium renewal: 2 to 3 times per week Preservation: 1. Freeze medium: Complete growth medium, 95%; DMSO, 5% 2. Storage temperature: liquid nitrogen vapor temperature §C.2 Detailed procedure used for MDA-MB231 and SK-BR-3 cell culture in this dissertation work. (courtesy of Henry Lin, Department of Pathology, University of Southern California). 1. Autoclave the glass coverslips (25mm diameter, Corning) on which MDA- MB231 and SK-BR-3 cells are to be cultured. 2. General rule: Before Start, sterilizing the Bio-cabinet with 70% alcohol. Use 70% alcohol often to sterilize everything used for culture (reagent bottle, hand glove, etc.). 3. General rule: all reagents, glass substrates, culture flasks, and culture wells used need to be preheated to 37 C in circulation water bath. 4. Observe MDA-MB231 and SK-BR-3 in original culture flask under optical microscope. SK-BR-3 cells are circularly shaped and clamped together. MDA-MB231 cells are elongated and are separated from each other. 5. The following procedures apply to both SK-BR-3 and MDA-MB231 unless otherwise stated. 372 6. Remove the culture medium in the original culture flasks containing SK-BR- 3 and MDA-MB231 cells. 7. Wash the original culture flasks with 2.5ml PbS buffer (for flask of area 7.5cm 2 ). (The purpose of this step is to wash off the remaining Trypsin inhibitor in the flasks.) 8. Suck away PbS from the corner of the original culture flasks. 9. Add 0.5ml 0.25% (w/v) Trypsin-0.53mM EDTA solution into the original culture flask (for flask of area 7.5cm 2 ). Put flask into incubator to keep the temperature at 37 C which is optimal for Trypsin activity. 10. Get a fresh culture flask (7.5cm 2 area) for continued culture. 11. Prepare Serum + Media solution For SK-BR-3: McCoy's 5a medium (modified) with 1.5 mM L-glutamine adjusted to contain 2.2 g/L sodium bicarbonate, 90%; fetal bovine serum, 10%. For MDA-MB231: Leibovitz's L-15 medium with 2 mM L-glutamine, 90%; fetal bovine serum, 10%. 12. Add ~5ml Serum + Medium solution into the fresh culture flask for continued culture. 13. Heat up the substrates (glass coverslips), culture wells and culture flasks to 37 C. 14. Gently shake the original culture flask with the cells, so that the cells detach from the wall of the original culture flask (observe under optical microscope). IMPORTANT, the interval between step 9 (adding Trypsin) and this step should be precisely controlled, if the interval is too long, the Trypsin can also 373 digest cells. In our operation this interval is controlled to be 2 minutes for MDA-MB231 and 5 minutes for SK-BR-3. 15. Split the cell solution from the original culture flask, ½ solution for continued culture (add into the fresh flask for continued culture which contains media and serum). Label the flask (cell type, date, media used, transfer times, operator). Put the flask for continued culture into the incubator (37 C, 5% CO 2 ). 16. Add 2ml serum+medium solution into each culture well. Put the glass coverslips (autoclaved beforehand) into the wells. 17. Split the rest ½ of the cell solution from the original culture flask equally into 6 culture wells. (Or for more precise control of the density of the cells on glass substrate, cell counter should be used.) 18. Put the culture wells into incubator (37 C, 5% CO 2 ). 19. Renew the media 2-3 times per week. 374 Appendix D: Rat Hippocampal Neurons Culture Protocols (Courtesy of Anubhuti Bansal) Preparation: CoverSlips Cleaning Coverslips: a. Regular Cleaning Place coverslips in ceramic racks. Soak racks in 1 N HCl overnight (HCl can be replaced with a zero-residue detergent) Rinse racks in ddH2O several times and overnight Store coverslips in 70% EtOH Before use, rinse coverslips in sterile ddH 2 O b. Alternative Rapid Cleaning: Put coverslips in petri dishes Rinse 1x with sterile ddH 2 O Dry and UV for 1 hour Rinse 1x with sterile ddH 2 O Dry, flip, and UV for 1 hour. Coating Coverslips Varies depending on coating material (see substrates protocol) Washing coated Coverslips Rinse coated coverslips with PBS several times (4 times at 1 hour interval, twice each time, and once overnight) 375 Rinse one more time the next day, and cover with a thin volume of NBM (+/- GLU) Preparing NBM solutions Desired Volumes NBM –GLU 200 ml NBM +GLU 100 ml Total 300 ml NBM -GLU NBM 293.1 ml B-27 (50x) 6 ml (final conc. 1x or 2%) Glutamine (200mM) 0.75 ml (final conc. 0.5 mM) Pen-Strep (10,000u/ml) 0.15 ml (final conc. 5u/ml) Filter 200 ml NBM +GLU Take remaining 100 ml NBM-GLU and add L-Glutamate (1mg/ml) 0.37 ml (final conc. 25uM) Filter these 100 ml Place in incubator at 37 °C and 10% CO 2 overnight (bottles cracked open) Dissection preparations: Clean incubator, UV sterilize metal and plastic trays, fill water tray with sterile water and antimitotic and antifungal, and set to 37 °C and 10% CO 2 . Flame pipet tips to make ends narrower, rank 1-4 (based on how long it takes bulb to reload after squeezing air out.) Autoclave: - Pasteur Pipets, short and long and flamed longs 376 - Surgical tools (sciscors, tweezers, fine forceps…) - Petri dishes, large, medium, and small, and ice container - Folded and foil-wrapped Kimwipes - Pipet tips if needed - ddH 2 O (in separate cycle) Filter sterilize 100 ml Hanks and PBS Have 2 liters of 70% Ethanol Place NBM-Glu, NBM+Glu ,0.25% Trypsin and 10% FBS in incubator Dissection: 1. Add 5-10 ml water to 500g ground dry ice in ice bucket 2. Put rat in bucket and close lid for 2 mins 3. Check reflexes and put rat in 70% ethanol for 10 mins 4. Pin rat limbs to dissection board 5. Cut skin over the stomach and detach the skin to uncover abdomen 6. Pin skin to sides, avoid overstretching and compressing the body 7. Lift and cut lower abdominal muscle being very careful not to puncture organs 8. Remove pups while cutting the attachments, be careful not to drop them. 9. Place pups in large sterile petri dish containing ice cold sterile PBS 10. Remove pups from embryonic sacs and decapitate 11. Place heads in PBS containing medium size petri dish 12. Take heads to dry petri dish, and stabilize with straight forceps at 45 ° through the eye orbits 377 13. Break the skull bone with curved forceps and pull back the skin to expose the brain 14. Scoop the brain out with the curved forceps and place in ice cold Hanks’. 15. Dry the brain on sterile Kimwipes, and put on petri dish under the microscope 16. Stabilize the brain from the cerebellum, and pull cortices to the side 17. Suck out the hippocampi with long Pasteur pipette and put into tube with 5ml cold Hanks’. Dissociation: 1. Rinse hippocampi with 5 ml Hanks 3 times, while fishing any loose membranes out 2. Add 500 ul Trypsin and incubate for 5 min @ 37 °C 3. Rinse 1 time with Hanks 4. Replace Hanks with 10% FBS and wait 3 mins 5. Rinse 3 times with Hanks 6. Using Pipette Size 0, pipette tissue up and down until is goes smoothly through the pipette. 7. Repeat #6 with pipettes # 1,2, and 3. 8. Put a drop in a petri dish and check for clump 9. If there are clumps Repeat #6 with Pipette #4 10. Check for clumps 11. Let the big clumps settle for 30+ seconds 378 12. Pass the floating cells (leave the clumps that settled) through Cell strainer into 50 ml tube 13. Transfer small amounts of the dissociated cells into the appropriate volume of NBM+Glu 14. Mix well, and take 10 μl of this mix to each side of the cell counter 15. Adjust cell count to desired value 16. Plate cells (3 mls into small petri dishes, 5 mls into medium, 12 mls into large) 17. Put dishes in incubator 18. Feed with NBM-GLU 3-4 days later, by removing half the volume of NBM+GLU and replacing with equal amount of NBM-GLU. 379 Appendix E: CdSe Rods and Au/CdSe nanocrystal metal/semiconductor junction synthesis. §E.1 CdSe rod synthesis. (Procedure courtesy of Steven Hughes, University of California, Berkeley) Materials Cadmium Oxide (CdO, Strem 99.99%) Trioctylphosphine oxide (TOPO, 99% Acros) Trioctylphosphine (TOP, 97% Strem) Hexylphosphonic Acid (HPA, Polycarbon Industry Inc. ) Octadecyl phosphonic acid (ODPA, Polycarbon Industry Inc. ) Selenium (Selenium shots 99.99%, Strem) Toluene (anhydrous, 99.8%, Aldrich) Methanol (anhydrous, 99.8%, Aldrich) Procedure Selenium/TOP (10% Selenium in mass) was prepared by dissolving Selenium in TOP. Inside a glovebox, Selenium shots need to be stirred overnight at room temperature in TOP to dissolve completely. Insert 200mg CdO, 3g TOPO, 0.16g HPA, 0.85g ODPA into a 25ml 3 neck flask. Connect the flask to the Schlenk line via a condenser (do not pass cooling water through the condenser). Purge the flask 3 times with Ar gas. Stir the mixture. Raise the temperature of the mixture in the flask to 120 C and degas the mixture under vacuum for 20mins. Switch back to Ar gas, continued to heat the mixture to 300 C. Reflux the mixture at 300 C for 10mins to get clear 380 solution. Bring the temperature of the solution back to 120 C, degas for 2hours under vacuum. Switch the flask back to Ar gas, and raise the temperature of the solution to 330 C. Rapidly inject 0.62g Selenium/TOP (10% Selenium in mass) at 330 C into the 3 neck flask. Temperature of the solution in the flask dropped to 315 C after injection. Set the temperature of the solution to 300 C. Stop the reaction 50mins after Selenium/TOP injection (stop heating and cool the flask to room temperature by blowing it with N 2 ). The produced CdSe rods were dissolved in toluene, precipitated by methanol and then redissolved in toluene for storage. Result As characterized by TEM, the obtained CdSe rods have an average diameter 9.5nm and an average length 28.5nm. The molar mass of the rods is 7x10 6 g. Assuming 80% Selium yield, the total amount of CdSe rods produced by the above procedure is 120mg (17nM). §E.2. Au/CdSe nanocrystal metal-semiconductor junction synthesis. Materials Gold Chloride (AuCl 3 , Aldrich 99.99%) Didodecyldimethyl ammonium bromide (DDAB, Aldrich 98%) Hexyldecylamine (HDA, Aldrich 98%) CdSe rods (synthesized as described in §E.1) Toluene (anhydrous, 99.8%, Aldrich) Methanol (anhydrous, 99.8%, Aldrich) 381 Procedure The Au/CdSe nanocrystal semiconductor/metal junction was formed by growing Au tips on one end of the CdSe rods. The procedure is detailed below. Two stock solutions were prepared by mixing the following chemicals, respectively, Stock solution 1: 1.66mg AuCl 3 + 8.33mg DDAB + 15mg HDA + 1g toluene Stock solution 2: 3.33mg AuCl 3 + 16.66mg DDAB + 5mg HDA + 48mg TOPO + 2g toluene Take 20mg of CdSe rods (synthesized as described in §E.1), dilute to 11mg/ml with toluene and store in an air free vial. Inside the glovebox, while stirring, inject 0.66ml Stock Solution 1 into the vial at a rate of 0.1ml/min. Then inject 1ml Stock Solution 2 into the vial at a rate of 0.1ml/min. Precipitate the produced An/CdSe nanocrystal semiconductor/metal junctions with methanol, and then redissolved in anhydrous toluene for storage. Result As measured from TEM images of obtained sample of Au/CdSe, 71% of the CdSe rods have a single Au tip on one end of the rods, 23% of the CdSe rods have Au tips on both ends of the rods, and 6% of the CdSe rods have no Au tip. The average Au tip diameter is 5nm. For illustrative TEM images of the Au/CdSe nanocrystal semiconductor/metal junctions see figure 7.3.

Abstract (if available)

Abstract Integration of epitaxial and colloidal semiconductor nanostructures into hybrid structures can potentially open unprecedented functionalities that combine the strengths of the epitaxial nanostructures as manifest in optoelectronics with the versatility of the colloidal nanocrystal and their application in solution environment. We envision that such hybrid structures can be a promising platform for chip-based, high-sensitivity detection of early disease and biohazard. This dissertation is devoted to several different research grounds that contribute to this envisioned application of colloidal/epitaxial hybrid structure.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Expanding the library of surface ligands for semiconductor nanocrystal synthesis and photovoltaic applications

PDF

Semiconductor inks for solution processed electronic thin‐films

PDF

Growth control and design principles of self-assembled quantum dot multiple layer structures for photodetector applications

PDF

Quantifying the surface chemistry of semiconductor nanocrystals spanning covalent to ionic materials

PDF

Non-radiative energy transfer for photovoltaic solar energy conversion: lead sulfide quantum dots on silicon nanopillars

PDF

Investigations of Mie resonance-mediated all dielectric functional metastructures as component-less on-chip classical and quantum optical circuits

Asset Metadata

Creator Lu, Siyuan (author)

Core Title Some studies of nanocrystal quantum dots on chemically functionalized substrates (semiconductors) for novel biological sensing

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Physics

Publication Date 12/01/2006

Defense Date 10/25/2006

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag biological sensing,cell adhesion,excitation transfer,nanostructure,OAI-PMH Harvest,quantum dot,surface functionalization

Language English

Advisor Madhukar, Anupam (committee chair), Bergmann, Gerd (committee member), Bozler, Hans M. (committee member), Hellwarth, Robert W. (committee member), Humayun, Mark S. (committee member)

Creator Email siyuanlu@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-m209

Unique identifier UC1455982

Identifier etd-Lu-20061201 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-167848 (legacy record id),usctheses-m209 (legacy record id)

Legacy Identifier etd-Lu-20061201.pdf

Dmrecord 167848

Document Type Dissertation

Rights Lu, Siyuan

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Repository Name Libraries, University of Southern California

Repository Location Los Angeles, California

Repository Email cisadmin@lib.usc.edu