Citation
The Optical and Photoconductive Response in Germanium Quantum Dots and Indium Tin Oxide Composite Thin Film Structures

Material Information

Title:
The Optical and Photoconductive Response in Germanium Quantum Dots and Indium Tin Oxide Composite Thin Film Structures
Copyright Date:
2008

Subjects

Subjects / Keywords:
Charge carriers ( jstor )
Conductivity ( jstor )
Electrons ( jstor )
Excitons ( jstor )
Germanium ( jstor )
Indium ( jstor )
Quantum dots ( jstor )
Tin oxides ( jstor )
X ray film ( jstor )
X ray spectrum ( jstor )

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright the author. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Embargo Date:
5/4/2004
Resource Identifier:
83991723 ( OCLC )

Downloads

This item is only available as the following downloads:


Full Text

PAGE 1

THE OPTICAL AND PHOTOCONDUCTIVE RESPONSE IN GERMANIUM QUANTUM DOTS AND INDIUM TIN OXI DE COMPOSITE THIN FILM STRUCTURES By TRACIE J. BUKOWSKI A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2002

PAGE 2

The path throughout my years of education would never have been such a joy or success without the constant love and support from my parents Tom and Vicki Bukowski. They taught me some of the most important lessons in life, without which, I would not be where I am today.

PAGE 3

ACKNOWLEDGMENTS Through any great adventure, especially one as consuming and life altering as obtaining a Ph.D., there are always special people that lend a hand along the way. First, and foremost, I would like to thank my advisor and mentor, Dr. Joseph H. Simmons. From the first time I spoke with him, he inspired me to see the joy and wonder that the pursuit of science can bring. I feel very fortunate to have met and worked with both him and his wife, Cate. A special thank you goes to another University of Florida professor, Dr. Paul Holloway. At a time when I was very uncertain which direction my path led, he was generous enough to essentially accept me into his group and offer support and guidance through the final stages of the research. I would also like to recognize the rest of my committee members – Dr. Jack Mecholsky, Dr. Cammy Abernathy, Dr. David Norton and Dr. David Tanner. I was fortunate enough to spend the last two years of my graduate career working at Sandia National Laboratories in New Mexico. For this opportunity, I am indebted to Kelly Simmons-Potter, Gerry Hays and B.G. Potter. Working with them allowed me to gain incredible experience, as I was able to interact with many professional people and scientists to build a network of friends and colleagues. In fact, I owe many of them a note of thanks for helping me collect critical data. For TEM sample preparation and imaging I would like to express my gratitude to Dr. Paul Kotula and Michael Rye. For the use and instruction with the laser sources I thank Dr. Joe Thomes and Dr. Darrell iii

PAGE 4

Armstrong. Other people that deserve thanks are Professor Pierre Lucas and Phil Anderson with the University of Arizona who provided me with XRD data, as well as Valentin Craciun from the University of Florida. Finally, I would like to share my appreciation for the two souls closest to my heart – my (soon-to-be) husband, Chip Berniard, and my cat, Luna. Chip’s amazing support, love, faith and patience are immeasurable. They made the entire journey one filled with happiness, and for that I will always be grateful. iv

PAGE 5

TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iii LIST OF TABLES............................................................................................................vii LIST OF FIGURES.........................................................................................................viii ABSTRACT.....................................................................................................................xiii CHAPTERS 1 MOTIVATION................................................................................................................1 Solar Cell Technology Today.........................................................................................2 The p-n Homojunction.............................................................................................2 Loss Mechanisms in Solar Cells..............................................................................7 Applying the Quantum Dot to Solar Cell Technology...................................................9 2 BACKGROUND...........................................................................................................12 The Quantum Confinement Effect................................................................................12 Low-dimensional Structures..................................................................................12 Quantum Dot Energy States...................................................................................14 The Exciton...................................................................................................................15 Exciton Bohr Diameter.................................................................................................16 Confinement Regimes...................................................................................................17 Photoconductivity.........................................................................................................20 3 THIN FILM FABRICATION........................................................................................26 Materials Selection........................................................................................................26 Indium Tin Oxide...................................................................................................26 Germanium.............................................................................................................33 Sapphire.................................................................................................................37 RF Magnetron Sputtering.............................................................................................38 Quantum Dot Formation...............................................................................................43 4 MICROSTRUCTURE...................................................................................................46 Transmission Electron Microscopy..............................................................................46 v

PAGE 6

Theory of Operation...............................................................................................46 Sample Preparation................................................................................................48 Quantum Dot Growth with Temperature...............................................................49 Quantum Dot Size Analysis...................................................................................53 Dot Size Variation Across Film.............................................................................56 X-ray Spectrum Imaging...............................................................................................57 Theory of Technique..............................................................................................57 2-D Compositional Mapping Data.........................................................................58 Raman Spectroscopy.....................................................................................................65 Theory of Technique..............................................................................................65 Raman Data............................................................................................................66 X-ray Diffraction..........................................................................................................70 Theory of Technique..............................................................................................70 X-ray Diffraction Data...........................................................................................71 Materials Interaction Concerns.....................................................................................76 Porosity and Germanium Ripening...............................................................................76 Discussion.....................................................................................................................84 5 OPTICAL PROPERTIES..............................................................................................90 Reflection, Transmission and Absorption.....................................................................90 Absorbance Data...........................................................................................................93 Discussion.....................................................................................................................98 6 ELECTRICAL PHOTORESPONSE...........................................................................101 Four-Point Probe Technique.......................................................................................101 Photoconductivity Data...............................................................................................105 Discussion...................................................................................................................112 7 CONCLUSIONS..........................................................................................................117 APPENDIX A RAMAN PEAK FITS.................................................................................................122 B BAND GAP DETERMINATION..............................................................................127 LIST OF REFERENCES.................................................................................................130 BIOGRAPHICAL SKETCH...........................................................................................137 vi

PAGE 7

LIST OF TABLES Table page 2-1 Exciton Bohr diameters and band gap energies for various semiconductors.............17 2-2 Electron and hole masses for various semiconductors................................................20 3-1 Dissociation energy per oxygen atom for various molecules.....................................34 3-2 Basic materials properties of Germanium...................................................................37 3-3 Basic materials properties of sapphire........................................................................38 3-4 RF magnetron sputtering conditions for both Ge and ITO..........................................43 4-1 Average quantum dot sizes for various 15/5% Ge multilayer films........................54 4-2 Characteristic XRD peaks for cubic germanium.........................................................74 4-3 Comparison of characteristic XRD peaks for Ge, In 2 O 3 and In 2 Ge 2 O 7 .......................84 6-1 Photoconductivity data for films taken at 800 nm....................................................106 6-2 Photoconductivity data for films taken at 300 nm....................................................110 vii

PAGE 8

LIST OF FIGURES Figure page 1-1 A p-n junction showing (a) variation of carrier concentrations, n and p, with position; (b) a shift in energy levels to create uniform chemical potential, , across junction....................................................................................................................4 1-2 Solar spectral irradiance for air mass zero....................................................................4 1-3 Graphical method to determine conversion efficiency from a plot of the number of photons in the solar spectrum versus photon energy...............................................6 1-4 Ideal solar cell efficiency versus energy, plotted for conditions of 1 sun (C=1) and 1000 sun (C=1000) concentrations..........................................................................7 1-5 Graphical method to determine conversion efficiency of 3 band gaps from a plot of # photons in solar spectrum versus photon energy.....................................................7 1-6 Excitation mechanisms that lead to conversion efficiency losses due to (a) energy less than E g , (b) thermal relaxation of energy in excess of E g , (c) radiative recombination, and (d) Auger recombination..........................................................8 1-7 Absorbance versus wavelength for a Ge film and for various Ge quantum dots 150 , 46 , 12 , and 4 in diameter.......................................................................10 2-1 Low-dimensional structures 3D-0D...........................................................................13 2-2 Density of states versus energy for a bulk material (3D), quantum well (2D), quantum wire (1D), and quantum dot (0D)...........................................................14 2-3 Schematic band diagrams for a free exciton, confined, coupled exciton, and a confined, decoupled exciton..................................................................................18 2-4 Rise and decay curves of photoconductivity for n-type silicon with (a) shallow traps and (b) deep traps...................................................................................................24 3-1 Transmittance and sheet resistance for various transparent conducting oxides; Note: fabrication method identified in parentheses.........................................................27 3-2 Free electron density of states as a function of % Sn doping in In 2 O 3 .......................29 3-3 Typical absorbance, A, transmittance, T, and reflectance, R, curves for an ITO film.31 viii

PAGE 9

3-4 Relative degree of bandgap energy shifts due to quantum confinement for several direct and indirect semiconductors........................................................................35 3-5 Energy band diagrams for germanium, Ge, and cadmium telluride, CdTe................36 3-6 Schematic representation of the dual gun rf magnetron sputtering system................39 3-7 Absorbance versus wavelength for ITO films deposited at different rf powers.........40 3-8 Sheet Resistance and Resistivity of ITO films as a function of rf power...................42 3-9 Deposition rate of the ITO gun as a function of rf power..........................................42 4-1 Signals generated when an electron beam interacts with a specimen........................47 4-2 The principle of forming a scanned TEM image........................................................48 4-3 Indium Tin Oxide film annealed to 800C for 1 hour................................................50 4-4 STEM images of 15/5% germanium multilayer composite films as a function of temperature (a) as-deposited, (b) 400C 1 hour, (c) 600C 1 hour, (d) 650C 6 minutes, (e) 660C 30 minutes, and (f) 700C 12 minutes....................................51 4-5 STEM images of composite films containing quantum dots of germanium and indicating dot size (a) 650C 6 minutes a = 10nm, (b) 660C 30 minutes a = 16 nm, and (c) 700C 12 minutes a = 14.5 nm...........................................................55 4-6 STEM image of a 15/5% composite film annealed to 700C for 6 minutes indicating three distinct growth regions – A (bottom region), B (middle region) and C (top region)..................................................................................................56 4-7 X-ray Spectral Imaging data for a 15/5% film heated to 400C 1 hour...................60 4-8 X-ray Spectral Imaging data for a 15/5% film heated to 600C 1 hour...................61 4-9 X-ray Spectral Imaging data for a 15/5% film heated to 650C 6 minutes.............62 4-10 X-ray Spectral Imaging data for a 15/5% film heated to 660C 30 minutes.........63 4-11 X-ray Spectral Imaging data for a 15/5% film heated to 700C 12 minutes.........64 4-12 Block diagram of the Raman spectroscopic experiment..........................................67 4-13 Raman spectra for sapphire, bulk germanium and indium tin oxide films annealed to 600C 1 hour and 1000C 30 minutes................................................................68 4-14 A more detailed look at the region around the Ge Raman peak from figure 4-13...68 ix

PAGE 10

4-15 Raman data for bulk Ge film heated to 600C 1 hour, 250 Ge film layered between ITO heated to 600C 1 hour, 125 Ge multilayer composite film heated to 600C 1 hour, and 15 Ge multilayer composite film heated to 700C 30 minutes...................................................................................................................69 4-16 Raman data for bulk Ge film heated to 600C 1 hour, 15/5% multilayer heated to 400C 1 hour, 15/5% multilayer film heated to 660C 30 minutes, and 15/5% multilayer film heated to 700C 12 minutes..........................................................70 4-17 XRD spectrum for an ITO film heated to 600C 1 hour..........................................73 4-18 XRD spectrum for a Ge film heated to 600C 1 hour..............................................73 4-19 XRD spectrum for Ge film heated to 600C 1 hour, focusing on the (111) reflection................................................................................................................74 4-20 XRD spectrum for a 15/5% multilayer film heated to 650C 6 minutes, focusing on the (111) reflection for Ge................................................................................75 4-21 XRD spectrum for a 15/5% Ge multilayer film heated to 660C 30 minutes, focusing on the (111) reflection for Ge..................................................................75 4-22 GIXD spectra of ITO as-deposited, ITO heated to 600C 1 hour, and a 15/5% Ge multilayer film heated to 660C 30 minutes..........................................................76 4-23 X-ray Spectral Imaging data for a 15/5% Ge multilayer film heated to 700C 12 minutes...................................................................................................................79 4-24 X-ray Spectral Imaging data for a 15/5% Ge multilayer film heated to 700C 30 minutes...................................................................................................................80 4-25 X-ray Spectral Imaging data for a 15/5% Ge multilayer film heated to 1000C 6 minutes...................................................................................................................81 4-26 X-ray Spectral Imaging data for a 15/5% Ge multilayer film heated to 1000C 12 minutes...................................................................................................................82 4-27 Magnified images of the films surfaces using an optical microscope for an ITO film heated to 800C 1 hour, a 15/5% multilayer film heated to 400C 1hour, a 15/5% multilayer film heated to 700C 12 minutes (with corresponding STEM image), a 15/5% mulitlayer film heated to 1000C 30 minutes, and a 15/5% mulitlayer film heated to 850C 30 minutes..........................................................83 5-1 Photographic images of pure indium tin oxide films (a) in the as-deposited state and (b) annealed to 1000C for 1 hour.........................................................................94 5-2 Absorbance data for indium tin oxide films in the as-deposited state, annealed to 400C for 1 hour, and annealed to 600C for 1 hour.............................................94 x

PAGE 11

5-3 Plot of sqrt( h) versus energy, h, for indium tin oxide films in the as-deposited state, annealed to 400C 1hour, and annealed 600C 1 hour.................................95 5-4 Photographic image of a 0.5 m pure germanium film annealed to 600C 1 hour....95 5-5 A data for a pure germanium film annealed to 600C for 1 hour...............................96 5-6 Plot of the square root of ( h) versus energy, h, for a pure Ge film annealed to 600C for 1 hour....................................................................................................96 5-7 Photographic images of 15/5% composite multilayer films in (a) the as-deposited state and (b) annealed to 660C for 30 minutes......................................................97 5-8 A data for 15/5% composite multilayer films at various temperatures...................98 6-1 Schematic of the four-point probe experimental set-up...........................................102 6-2 Graphical method for determining the correction factor, CF, for the calculation of sheet resistance from four-point probe data.........................................................103 6-3 Photographic image of the four-point probe head contacting the sample................104 6-4 Absorbance data for various ITO and Ge/ITO multilayer films..............................105 6-5 Conductivity versus time for the photoresponse of an ITO film heated to 600C for 1 hour. (On and Off indicate when the film was exposed to the light and when it was removed).......................................................................................................107 6-6 Conductivity versus time for the photoresponse of an ITO film heated to 800C for 1 hour......................................................................................................................108 6-7 Conductivity versus time for the photoresponse of an 15/5% multilayer film heated to 650C for 6 minutes.........................................................................................108 6-8 Conductivity versus time for the photoresponse of an 15/5% multilayer film heated to 660C for 30 minutes.......................................................................................109 6-9 Conductivity versus time for the photoresponse of an 15/5% multilayer film heated to 700C for 30 minutes.......................................................................................109 6-10 A normalized plot of the measured voltage response versus time for an ITO film heated to 600C and a multilayer film heated to 660C......................................110 6-11 Conductivity versus time indicating the decay behavior of a multilayer film heated to 650C for 6 minutes.........................................................................................111 6-12 Conductivity versus time indicating the decay behavior of a multilayer film heated to 660C for 30 minutes.......................................................................................111 xi

PAGE 12

6-13 Conductivity versus time indicating the decay behavior of a multilayer film heated to 700C for 30 minutes.......................................................................................112 6-14 Band diagram for Ge/ITO after contact..................................................................116 xii

PAGE 13

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy THE OPTICAL AND PHOTOCONDUCTIVE RESPONSE IN GERMANIUM QUANTUM DOTS AND INDIUM TIN OXIDE COMPOSITE THIN FILM STRUCTURES By Tracie J. Bukowski May 2002 Chair: Joseph H. Simmons Cochair: Paul H. Holloway Department: Materials Science and Engineering Films consisting of ~35 alternating layers of germanium (15/layer) and indium tin oxide (280/layer) were deposited onto sapphire substrates using the rf magnetron sputtering deposition technique, producing a total film thickness of 0.5 – 1.0 m. The deposited multilayer structures were heat-treated to temperatures of 600-1000C for times of 6-60 minutes in an argon atmosphere, creating germanium (Ge) quantum dots 9-25 nm in size surrounded by indium tin oxide (ITO). Beyond temperatures of 700C, porosity and loss of germanium are evident. The quantum dot films were characterized using transmission electron microscopy (TEM). From the TEM images, the microstructure of the films and the size of the quantum dots were obtained. Raman and X-ray diffraction data revealed that the films heated above 600C consist of cubic, polycrystalline Ge and In 2 O 3 . The Raman data showed a shift in the crystalline Ge peak toward the high frequency side by 12 cm -1 as the confinement of the Ge was increased. This shift is opposite to the trend expected for quantum confinement, likely due to stresses in the film. xiii

PAGE 14

The effect that varying the dot size has on the optical properties of these films is obvious from their absorbance data. A clear blue-shift in the band edge energy, 2.7-3.7 eV shifted from bulk germanium, results as the quantum dot size decreases. The search for enhanced photoconductivity due to the presence of the Ge nanocrystals in the ITO was established using a Ti:sapphire laser tuned to 300 nm and a four-point probe conductivity measurement system. The photoconductive response measured for the pure ITO films showed a very small increase in conductivity of 0.1 0.06%. However, a composite multilayer film annealed to 660C for 30 minutes, containing an average germanium dot size of 16 nm, produced a photoconductive response of 3.9 0.05%. While the overall increase in conductivity for the quantum dot film is still small, its value is a clear enhancement over the pure ITO films containing no quantum dots. This indicates that the germanium, while being in a quantum confined state, was able to inject photogenerated carriers into the conducting matrix. The result was an increase in the free carrier density of the films containing the germanium quantum dots, which is ultimately responsible for the enhanced photoresponse measured. xiv

PAGE 15

CHAPTER 1 MOTIVATION Nanoscale structures, particularly quantum dots, have been of enormous interest given their unique potential. The thrust in device-based research has been towards smaller and smaller dimensions. It is in this realm that the quantum dot lives and its inimitable properties, due to its small size, have opened up many new directions of research. Quantum dots have revealed interesting phenomena such as quantum interference effects, tunneling effects and band gap variations. 1 These behaviors give nanoscale structures an important place in advancing technology. Among the areas being advanced are information and telecommunications, electronics and electro-optics, including the quantum dot laser, chemical and gas sensors, and nonlinear optics. 2-20 One of the more dramatic properties of the quantum dot is its ability to vary the band gap of a material – something once thought to be a definitive materials property. The ability of the quantum dot to tailor a band gap lies in its small dimensions, which give rise to a quantization of its energy states. An exciting question concerning quantum dot physics is – “Can a quantum dot be spatially and energetically confined and simultaneously be able to conduct away its photoexcited carriers for use in an external load?” This question is intriguing not just at the basic science level, but also at the application level, specifically where solar cell technology is concerned. In order to understand the significant impact that quantum dots can have on solar cells, the fundamental aspects of solar cell technology must first be described, the details of which 1

PAGE 16

2 are outlined. Then, the motivation behind applying semiconductor quantum dot structures to the problems that plague solar cells is presented. Solar Cell Technology Today Ever since the inception of the solar cell in 1954 by Chapin, Fuller, and Pearson, much research has been conducted in search of improved efficiency and performance. 21 In addition to studying many different semiconductor materials and structures, including single-crystal, polycrystal and amorphous, researchers have become very clever in device configuration in an attempt to yield high conversion efficiency. 22-24 For example, a major source of loss in the conventional silicon, Si, solar cell is due to reflection. To help prevent such a loss, techniques such as the textured cell and anti-reflection coatings have been devised. Although much research has been performed for the purpose of improving device efficiencies, the performance of solar cells is still far from thermodynamic efficiencies. The p-n Homojunction In spite of all the solar cell research conducted over the years, the silicon p-n homojunction still remains the conventional technology in use for space and terrestrial applications. The p-n homojunction is created when one side of the material is p-type and the other n-type, forming an abrupt junction. 25 On the p-side the majority carriers are holes, while electrons dominate on the n-side. When such a junction is created, the electrons from the n-side diffuse toward the p-side, while the holes diffuse from p to n-side. These electrons and holes recombine in the region around the junction, creating the so-called depletion layer, an area defined by few charge carriers. Figure 1-1(a) is a graphical representation of this variation in charge carriers, including the depletion region, surrounding the junction. Another important result of the electron-hole

PAGE 17

3 interaction around the junction is a shift in energy levels. The now positively charged ionized donors in the n-region of the depletion layer leave the n-region positively charged. Similarly, the negatively charged ionized acceptors in the p-region leave it negatively charged. Since there must be charge neutrality across the junction, the electron energy levels on the n-side are lowered and those on the p-side are raised. The criterion that causes this is the requirement that the chemical potential must be independent of position. Another useful way to think of this is that the Fermi levels must be equal on both sides of the junction. The shifting of energy levels to satisfy this condition creates a built-in potential, also known as the contact potential, o . Figure 1-1(b) helps illustrate this concept. Once the p-n junction is created it can be utilized as a solar cell device. Any carriers excited within the depletion layer will be driven across the junction (direction dependent on the type of carrier) where they can be collected. However, the Si p-n homojunction solar cell has its disadvantages. Figure 1-2 shows the solar spectrum irradiance as a function of wavelength for various air mass conditions. 26 Note that air mass zero (AM0) represents the solar spectrum from space, i.e. outside the Earth’s atmosphere. It can be seen that visible light makes up a significant part of the spectrum. It is this range of energies we are most concerned with for the transformation of light rays to electronic carriers within the solar cell device.

PAGE 18

4 Figure 1-1 A p-n junction showing (a) variation of carrier concentrations, n and p, with position; (b) a shift in energy levels to create uniform chemical potential, , across junction. Figure 1-2 Solar spectral irradiance for air mass zero. Perhaps the single most important issue in solar cell performance is ideal conversion efficiency. The ideal conversion efficiency is defined as the ratio of the maximum power output to the input power. So, the more incident light that is absorbed and turned into carriers, the better the conversion efficiency. However, basic materials

PAGE 19

5 properties dictate the absorption characteristics of a material. The conventional p-n junction solar cell has only a single band gap, E g , as shown in Figure 1-1. Once the cell is exposed to the solar spectrum, any photon with energy less than E g will not be absorbed by the device and thus contributes nothing to the cell output. This is known as spectral mismatch, in which light composed of a spectrum of energies has only one band gap for absorption. The device is a low pass filter, absorbing light of greater energy than the band gap. However, the excess energy is lost as heat and only the energy equal to E g will contribute. Figure 1-3 illustrates graphically the conversion efficiency of a single band gap device (E g = 1.35 eV). The ideal conversion efficiency is the ratio of the maximum power output to the incident power. This is obtained graphically by first finding the short-circuit current density, J L , for a given band gap energy, then finding its corresponding E m , the energy per photon delivered to the load at the maximum power point. The maximum conversion efficiency is then just the ratio of the area under the rectangle to the area under the curve. In this case the maximum efficiency is only 31%, meaning that only 31% of the total incident power from the sun is converted into output power through the device. Photons with a lower energy than E g are lost through transparency and those with high energy only contribute E g to the cell potential. It would seem that the band gap value of the semiconductor is of great importance in determining the conversion efficiency. Note from Figure 1-4 that if the band gap is decreased to capture more photons the efficiency does not necessarily go up. There are many semiconductors that reach near that maximum efficiency. In fact the maximum is so broad that the conversion efficiency is not significantly dependent on E g , so long as it is in the range of approximately 1-2 eV. One way to increase the efficiency is to alter the

PAGE 20

6 cell to incorporate several band gaps. Figure 1-5 shows the effect that multiple band gaps can have on the maximum conversion efficiency. Simply increasing the number of band gaps from one to three enhances the efficiency by about 20%, as each region with a different band gap captures more energy and provides additional photon conversion. Multiple band gaps have a dramatic effect on increasing the ideal conversion efficiency of solar cells. This is true because the cell is now able to absorb at several wavelengths, which helps to prevent conversion losses. Figure 1-3 Graphical method to determine conversion efficiency from a plot of the number of photons in the solar spectrum versus photon energy.

PAGE 21

7 Figure 1-4 Ideal solar cell efficiency versus energy, plotted for conditions of 1 sun (C=1) and 1000 sun (C=1000) concentrations. Figure 1-5 Graphical method to determine conversion efficiency of 3 band gaps from a plot of # photons in solar spectrum versus photon energy. Loss Mechanisms in Solar Cells It is interesting to note that the calculated maximum thermodynamic efficiency of solar energy conversion is 93%. 27 And yet, as was reported previously, a single band gap device is only capable of 31% ideal conversion efficiency. In addition to the efficiency loss discussed above, there are many other losses that can occur during the process of

PAGE 22

8 converting light into energy. Figure 1-6 illustrates some of the different losses that play a role in reducing solar cell efficiency. While a direct transfer of energy equal to the band gap is the ideal process to occur, many different things can happen. There can be losses due to (a) insufficient energy, (b) too much energy or thermal relaxation, (c) radiative recombination, and (d) non-radiative recombination. In the radiative case, the energy that went into creating an electron in the conduction band is given up in the form of light as the electron in the conduction band and the hole in the valence band recombine. The non-radiative case also involves an electron and hole recombination. However, instead of releasing the energy as light, it is transferred to another carrier. Such an example, given in (d), is known as Auger recombination. These additional problems can cause a decrease of 33-50% from the efficiency of the solar cell. These latter problems can be mitigated to some extent and much work has been done to produce high quality crystals, etc., explaining why some cell devices vary between 9 – 23% efficiency. However, the true limit of the cell is still the spectral mismatching. Figure 1-6 Excitation mechanisms that lead to conversion efficiency losses due to (a) energy less than E g , (b) thermal relaxation of energy in excess of E g , (c) radiative recombination, and (d) Auger recombination.

PAGE 23

9 Applying the Quantum Dot to Solar Cell Technology Clearly, the great leap forward in solar cell technology that will dramatically improve conversion efficiency is to design cells that incorporate more than one band gap into a single device. This requires a device with several different materials or a single material whose band gap is tailorable. By choosing to include several different materials into the device, one could help solve the conversion efficiency problem, but in the process create many more problems. One such problem is known as lattice mismatching. When two different materials are forced to bond together, stresses can develop between the two layers if the unit cell spacing of the materials is not equal. The stress is a result of the forced stretching and compression of bonds at the interface. This lattice mismatching creates a non-equilibrium state for the atoms involved, making them more susceptible to change. They act as defects so that electrons and holes created within the cell are often drawn to these lattice mismatched sites and trapped. This effect serves to reduce the overall efficiency of the solar cell. Another potential negative effect on conversion efficiency due to multiple materials is the issue of materials interaction. When a material is brought together with another material, there is always a likelihood that an alloy can form at the interface, where essentially a mixing process occurs. However, as is often the case, this resulting material may not possess the kind of optical or electrical properties needed. As a consequence, either the incident light reflected or scattered, or carrier trap sites are created. Whichever may be the case, both result in large carrier losses and poor conversion efficiency. Quantum dots offer an interesting option since they can be fabricated in different sizes and can exhibit a range in band gap energies. Since their band gap is tailorable, based on size, different layers could act together to reduce spectral mismatch and increase

PAGE 24

10 the conversion efficiency, as seen in Figure 1-5. However, quantum dots are usually formed in a matrix of different composition, and so there may be additional concerns of lattice mismatch and materials interaction. In addition, since a process of quantum confinement alters the band gap, there may also be barriers to the injection of these carriers into the matrix, limiting carrier transport in the device. If these problems can be resolved, quantum dots have the potential to improve the present state of solar cell technology because these unique structures have the ability to change their band gap. Figure 1-7 Absorbance versus wavelength for a Ge film and for various Ge quantum dots 150 , 46 , 12 , and 4 in diameter. The phenomenon of band edge shift due to quantum confinement has been well established experimentally. Much effort has been put forth in studying semiconductor quantum dots grown within an insulating glass matrix; silicon or germanium, Ge, nanocrystals in silica, SiO 2 , being a common system. 28-32 Bukowski et al. published data on germanium quantum dots in a SiO 2 matrix. 33 Their absorbance data illustrate the dramatic shift in band gap that is possible with varying the dot size (see Figure 1-7). As

PAGE 25

11 can be seen, a blue-shift in energy results as one goes from a thin film of Ge to quantum dots of Ge 4 in diameter. In fact the shift in band gap energy spans several eVs. The data prove that not only does quantum confinement of the Ge quantum dots produce a dramatic blue-shift in band gap energy, but that the size of the dots produced can be tailored to vary E g over a wide range of desirable values. However, if the special qualities of quantum dots are to be utilized in a photoconductive or photovoltaic device, such as a solar cell, photoexcited carriers must diffuse into the matrix material. This is only possible if the matrix material is not only transparent, but also electrically conducting. This becomes the basis for the current research – “What happens when the insulating SiO 2 matrix is replaced with a transparent conducting oxide matrix?” The implications of such a question are important and numerous. The following chapters will address some of the aspects involved in defining, studying and answering such a question. The goal of this research was to show experimentally that the presence of semiconductor quantum dots inside the transparent conducting matrix produces an enhanced electrical photoresponse.

PAGE 26

CHAPTER 2 BACKGROUND The unique properties of the quantum dot can be explained using simple models of the quantum confinement effect. The details of this effect are outlined and include a discussion on the density of states for these systems. Then, the importance of the exciton in quantum dot energy transitions is explained, and the concept of the exciton Bohr diameter as an important, material dependent property in quantum confinement is described. A discussion will follow on the different confinement regimes possible, based on the size of the quantum dot, will be outlined. Because the goal of the present research is aimed at applying the quantum dot to photoconductive and photovoltaic applications, the pertinent elements of the theory of photoconductivity are presented. The Quantum Confinement Effect Low-dimensional Structures The quantum dot belongs to a group known as low-dimensional structures (2D-0D). The dimensionality refers to the number of directions in which the carriers of the material act as free carriers. For example, a 3D structure is one in which the electronic carriers are free in all three directions. This is the case for a bulk material. Now imagine that a completely free bulk material is spatially confined in one direction. What results is a thin film, whose carriers may now be confined in one of the three directions, creating the 2D structure, also known as a quantum well. Taking this spatial confinement process to the next step and further confining the thin film, one is left with a 1D system, or the 12

PAGE 27

13 quantum wire. Finally, when a material is spatially confined in all three directions, a 0D structure is formed, i.e., the quantum dot. This progression is illustrated in Figure 2-1 Figure 2-1 Low-dimensional structures 3D-0D. Quantum confinement effects play a role in the properties of a material as long as the distance in the confined direction is on the order of the carrier de Broglie wavelength, , which is expressed as )3(kTmheff (2-1) where h and k are constants and T is the temperature. Here the mass, m eff , the effective mass of the carrier in the confined crystal. The effective mass is often much smaller than the true mass. This reduced mass results in size quantization effects at a thickness much larger than expected,10-100 times the lattice constant. 34 Of course, the dimensions below which confinement effects are seen are completely dependent on the material involved. The significance of this spatial confinement of the carriers can be seen in a plot of the density of states for each low-dimensional structure. Depicted in Figure 2-2 is the change in the density of states that occurs as a material evolves from a completely unconfined state to the three-dimensionally confined state. Interesting changes occur in

PAGE 28

14 the allowed carrier energies as a result of the confinement process. In the bulk, the carriers can exist in continuous bands. This is clearly not the case for the 0D structure where the material has been spatially confined in all three directions forming a small dot of material. Instead of a continuous band of allowable energies, the carriers within a 0D structure are restricted to a specific set of completely quantized energy states. This quantization of allowed energies and the dot-like shape of the confined material are what give the quantum dot its name. Figure 2-2 Density of states versus energy for a bulk material (3D), quantum well (2D), quantum wire (1D), and quantum dot (0D). Quantum Dot Energy States It is clear that spatially confining a material to very small dimensions results in a change in the density of states. The exact effect in the energy states of the system can be obtained by solving for the eigenenergies of the Schroedinger wave equation. In this way, an direct relationship between the size of the confined system and the resulting change in its energy state can approximated using simple confinement theory. 35-37 As stated before, the quantum dot is confined in all three directions and so experiences a potential barrier change in each direction. In order to solve for the energy of this system, the Schroedinger equation must be solved for in three dimensions. The resulting expression for the energy of the confined system is

PAGE 29

15 22222mLnEnn (2-2) where m is the carrier mass, n is the n zeroes of the spherical Bessel function of order 1, and L is the confinement dimension. 38 The resulting energy state is inversely proportional to the square of L. In effect, the band gap of a material can be shifted toward higher energies by confining the carriers. The carrier mass plays a role in determining the characteristic dimensions necessary for creating size quantization effects. This idea is presented in detail in the following section, where the concept of the exciton is presented and its importance in mapping out the different confinement regimes explained. As will be seen, the correct form of the energy equations in a quantum dot changes according to its size regime. The Exciton It has been shown that in the case of a fully confined quantum dot there are no free carrier contributions to the density of states. If there are no free carriers, then what means of excitation and conduction are possible? This is where the idea of the exciton comes in. The concept of an exciton did not surface until there was experimental evidence in a semiconductor absorption spectrum that then could not be explained. 39 The study of semiconductor materials has been enormous. Their basic band structure makes them very interesting to study and they have revolutionized the state of technology today. The feature that makes semiconductor materials so useful is the distribution of allowed electron quantum states. This is the range of energies within which no electron states exist; i.e., an electron cannot occupy an energy state that lies within this forbidden region. In the simple case, the electrons in the valence band must be given energy greater than or equal to the band gap in order for the electrons to conduct. It was assumed in the early

PAGE 30

16 studies of semiconductors that the conduction band electron and the valence band hole had no interaction. However, the absorption edge spectra, whose explanations were based on such an assumption, could not explain many details observed in optical experiments. Hence, the idea of an exciton was proposed. In this case, the energy provided to the electron is not enough to produce a free electron and a free hole. This means that an electron in the valence band cannot be excited into the conduction band. Although technically an electron without sufficient energy is not allowed to occupy an energy above its equilibrium state in the valence band, sometimes it is possible to form a stable structure by exciting the electron into an orbit about its potential minimum. In this way, a stable state is formed below the conduction band as the electron orbits the hole. Thus, the electron is bound to a hole, creating what is called an exciton. In the case of spatially confined systems, such as the quantum dot, excitons are the carriers that create the resulting quantized energy states. Exciton Bohr Diameter The structure of an electron bound to a hole is very reminiscent of the hydrogen atom. In the hydrogen atom case, a single electron is bound to a single proton by a Coulombic attractive force. The behavior of the exciton is very similar. In fact, as the electron orbits the hole, a set of hydrogen-like states is created. This is the reason that the Schroedinger wave equation is so easily adaptable to solving for the energy states in a quantum dot. In Bohr’s model of the electron’s orbit around the nucleus, he derived expressions for the radii and energies of these circular orbits. This same idea is applied to the quantum dot, not just regarding the energy states, but also regarding the carrier orbit dimension. Just as an electron orbiting a nucleus has a characteristic dimension, called the Bohr radius, so too does the electron orbiting a hole in a quantum dot exciton.

PAGE 31

17 This characteristic dimension is called the exciton Bohr diameter, a x , and is essentially a measure of the diameter (or radius) of the exciton. The exciton Bohr diameter is a critical parameter that provides a basis upon which to judge the criteria for size confinement in materials. For example, the exciton Bohr diameter for a CdTe quantum dot is 150 . This means, in general, that in order for size confinement effects to be observable, the size of the CdTe nanocrystal needs to be roughly 150 or smaller. It will be seen in the next section that there are more detailed requirements on size depending on the confinement regime, but the exciton Bohr diameter is a useful property. Table 2-1 lists several different semiconductors and their corresponding exciton Bohr diameters. 38 Notice that it is a material dependent property since the criteria for confinement change according to the nature of the material involved. Table 2-1 Exciton Bohr diameters and band gap energies for various semiconductors. Semiconductor Exciton Bohr Diameter Band gap Energy CuCl 13 3.4 eV ZnSe 84 2.58 eV CdS 56 2.53 eV CdSe 106 1.74 eV CdTe 150 1.50 eV GaAs 280 1.43 eV Si 37(longitudinal) 90(transverse) 1.11 eV Ge 50(longitudinal) 200(transverse) 0.67 eV PbS 400 0.41 eV Confinement Regimes The exciton Bohr diameter is a useful parameter for evaluating the conditions for creating quantum confinement effects in a specific material. In general, confinement effects must be taken into account as the material dimension is reduced to a size

PAGE 32

18 approaching the exciton Bohr diameter. However, the effects due to confinement do not just act as a switch and immediately turn on at a certain size constraint. Instead, there are different regimes of confinement, strong and weak, that give different resulting energy state equations. 39 These strong and weak states are determined by the degree of coupling between the electron and hole in the exciton. Figure 2-3 helps to illustrate the process that occurs as a free exciton is confined to a coupled state, and further confined to the decoupled state. Note that when the electron and hole are no longer coupled, this is the strong confinement case. Figure 2-3 Schematic band diagrams for a free exciton, confined, coupled exciton, and a confined, decoupled exciton. When an exciton is created in a semiconductor quantum dot and the size of the crystal is roughly 3-10 times the exciton Bohr radius of the material, then the exciton is said to be in a weakly confined state. This means that the Coulomb interaction energy is on the order of the electron and hole sublevel separations, and so must be considered. The electron and hole are then treated as a correlated pair whose center of mass motion is

PAGE 33

19 quantized by the confinement potential. It is possible to solve for the energy states in a weakly confined quantum dot using the parabolic approximation of the Schroedinger equation with the boundary condition that the potential at the radius of the crystal, a o , is zero; (a o ) = 0. A solution can be found by separating variables into center of mass coordinates. 38 The resulting expression for the energy quantization in 0D structures is 2*222onlnlaME (2-3) where M * = effective exciton mass (M * =m e * + m h * ), and nl are the n zeroes of the spherical Bessel function of order l. This equation gives the shift in energy that occurs when a bulk solid exciton is confined in three dimensions in the range of 3-10 times the exciton Bohr radius. When the size of the quantum crystal, a o , approaches the size of the exciton, a x , the solution to the energy states, presented in the previous section, begins to break down. In this situation, the Coulomb effects are small compared to the quantization of the kinetic energy of the exciton. When these conditions are present, the exciton is said to be strongly confined. 38 For example, if the quantum dot radius is below that of its exciton Bohr radius, then the electron and hole wave functions are uncorrelated due to strong confinement. The approach to solving for the energy states in such a system is the same as that presented in the weak confinement case, except that, in the case of strong confinement, the electron and hole have independent Bessel wave functions. The resulting expression for the energy shift is: 2*222onlnlaE (2-4) where now the effective reduced mass, * is defined by

PAGE 34

20 ***111hemm (2-5) It is useful to note the differences between the energy expressions for the strong and weak confinement cases, and the 3D particle in a box treatment. The relationship between energy and crystal size is the same; a blue-shift in energy still results as the size of the crystal decreases. The differences are the presence of the Bessel function term, nl , and the changes in reduced carrier mass. In regards to the carrier mass, in a bulk semiconductor the electrons dominate and so the electron mass is used. However, in an exciton, both the electron and hole contribute to the carrier mass, resulting in a reduced mass that is lower than the effective mass of the electron. This is especially true for excitons strongly confined since the reduced mass is smallest and hence the energy shift is more pronounced. Of course the relative shifts in energy are very dependent on the material involved since the electron and hole masses can vary greatly with the material. Table 2-2 lists the electron and hole masses for several different semiconductors. 38 Table 2-2 Electron and hole masses for various semiconductors. Semiconductor m e * m h * GaN 0.2 0.8 CdS 0.2 0.9 CdSe 0.13 0.8 CdTe 0.11 0.35 GaAs 0.07 0.5 Si 0.98(longitudinal) 0.19(transverse) 0.52 Ge 1.58(longitudinal) 0.08(transverse) 0.3 PbS 0.1 0.1 Photoconductivity Simply stated, photoconductivity is a phenomenon where extra carriers are generated within a semiconductor or insulator when it is illuminated by photons of the

PAGE 35

21 appropriate energy. 40 The presence of the photogenerated carriers results in a lowering of the resistance of the material, i.e. the electrical conductivity is enhanced. This effect, however, is only measurable under the influence of an electric field, whether it is inherent in the material or externally applied. Hence, photoconductivity is a two-step process whereby excess carriers are first generated optically, and then transported across the material by an electric field. The energy, or wavelength ( p ), of light necessary for creating the photoexcited carriers is determined by the band gap, E g , of the material. So, for all wavelengths of light equal to or less than p , the energy is absorbed to produce electron–hole pairs. While testing for enhanced conductivity in the presence of light seems a very straightforward task, there are many complexities involved in trying to understand the dynamics of how the carriers are generated and where they ultimately go. When a semiconductor thin film is illuminated, generating an electron-hole pair, both the electron and the hole can contribute to the resulting conductivity. In all of the treatments of the issue of conductivity there are two important terms that help define the conductive behavior of a material. These terms are the density of free carriers, n, and the mobility of the carriers, . Of the two, the density of carriers often has a more dramatic effect on conductivity, as it is more strongly a function of energy. When both the electron and hole contributions to the overall conductivity are considered, the final expression is )(pnpnq (2-6) where q is the electronic charge of the carrier, n and p are the number of electrons and holes, and n and p are the corresponding electron and hole mobilities, respectively. Clearly, by increasing the number of carriers, both electrons and holes, and their corresponding mobilities, the conductivity of the material can be increased.

PAGE 36

22 However, the process of excitation and enhanced conductivity can be complicated by various mechanisms that interfere with the successful transport of any excited carriers. When photoexcitation produces a free hole and free electron, both carriers will contribute to the increased conductivity as long as they are unhindered and pass out of the material at the appropriate contact end, while being replaced at the other contact end to maintain charge neutrality. But the carriers may meet a different fate while being driven through the material. For a material with defects, the carriers can be captured by these local imperfections, or the electron and hole can simply recombine. Often, recombination occurs between a trapped carrier and a free carrier of opposite type, or also by two closely trapped carriers. The ultimate effect is that the carrier density is decreased, which reduces the enhanced photoconductive response. Such carrier trapping sites can have a pronounced effect on the dynamics of carrier transport, specifically in photoexcited carrier decay time. Imagine that a photoconducting material is illuminated, exciting carriers at a rate of g o carriers per second-cm 3 . The conductivity increases and reaches a steady state when the excitation source is turned off. Without the excitation source providing energy for continued production of carriers, those in excited states will begin to decay to their ground states. The rate at which this occurs can offer an insight into the number of trapping sites in a material. Consider the case where the number of electrons in traps, n t , is much smaller than the density of free electrons, n. This is known as trap saturation and is common at high excitation levels or high temperature. In such a situation, the current will decay exponentially, with a decay time equal to the electron lifetime o . Knowing o makes it possible to calculate the microscopic mobility of the majority carriers, o . 40

PAGE 37

23 oooqg (2-7) However, if instead the number of free electrons is much smaller than the number of trapping sites, then the decay can be significantly longer due to slow release of carriers from their trap sites. While the relative intensity of the light source can affect the decay rates, the type of traps present also has a pronounced effect in the decay behavior. In n-type silicon (n-Si), studies have been done to evaluate the decay time in photoconductivity experiments. 41 Figure 2-4 (a) shows a typical rise and decay curve for n-Si, illustrating the effect that trapping sites have on the resulting measured behavior. 41 In this case, shallow traps, as indicated by the short decay times, dominate the decay behavior. The light source is turned on at point A, creating electrons and holes which increase the conductivity to point B where they reach equilibrium. Between B and C, hole trapping occurs, causing the overall conductivity rate to decrease somewhat. The light is then turned off at point C and a decay is observed corresponding to the recombination of excess electrons and holes between C and D. Finally, the shallow traps are emptied resulting in the slower decay rate seen between points D and E. A similar trend is seen for n-Si containing deep level traps. Figure 2-4 (b) presents the photoconductivity curve, illustrating the typical long-time decay behavior associated with deep hole traps. When the light source is removed, the expected drop in conductivity is seen between D and E. This initial fast decay corresponds to recombination and the emptying of shallow traps. The deep traps are represented in the region from E to F and can take considerably longer to decay. Such a curve can provide quantitative information, allowing the calculation of the density of deep traps from the difference in conductivity between D and F. 41

PAGE 38

24 Figure 2-4 Rise and decay curves of photoconductivity for n-type silicon with (a) shallow traps and (b) deep traps. Clearly, the rate at which the measured photoconductive response of a material decays when the light is turned off is dependent on the type of process taking place. The fastest decay rate process, alluded to in the above example, is the direct recombination of free electrons and holes. The next fastest decay process is the emptying of shallow traps. The slowest decay rate processes are due to the emptying of deep traps and the retrapping of carriers. Photoconductivity in low-dimensional structures possesses the same issues as in bulk materials. The primary difference being that there is a potential barrier to carrier transport due to the presence of confinement. This barrier height can be different for the electrons and holes of the excited exciton, and, thus, each can contribute in different ways. It is important in the material selection process that the band diagrams be

PAGE 39

25 considered for insight into the potential electronic behavior of the confined carriers. Too high a potential barrier can make transporting the carriers to the electrodes difficult. Too low a barrier and confinement will not be present. While the dynamics involved in photoconductivity can be very complex, by measuring the conductivity in both dark and illuminated conditions any resulting gain can be measurable. It is expected that the presence of the germanium dots will effectively increase the carrier density available for transport, resulting in an increased photoresponse. Furthermore, the rate at which the excited carriers decay back to their ground state can offer an interesting look into the density of trapping sites.

PAGE 40

CHAPTER 3 THIN FILM FABRICATION In order to investigate the properties of the proposed quantum dot thin films, they must first be fabricated. This chapter presents considerations involved in making the final materials selections for the quantum dot material, the matrix material and the substrate upon which the films are deposited. Also included are the details of the rf magnetron sputtering technique used for depositing the films. Finally, the post-deposition processing steps necessary to form the Ge quantum dots are outlined. Materials Selection Clearly, the matrix material needs to be transparent over the wavelength spanning the solar spectrum, and electrically conductive. The former is necessitated by the design of the solar cell in which only the quantum dot material is photovoltaic. The latter is needed to transport the excited carriers from the quantum dots to the electrodes. A class of doped and non-stoichiometric materials has the unique characteristic of being simultaneously transparent and conducting. They are called transparent conducting oxides, and have been used to serve many different purposes such as transparent electrodes. Indium Tin Oxide It has been indicated that intrinsic, stoichiometric materials are not capable of achieving high optical transparency and high electrical conductivity. 42 The way in which a material is essentially forced into such a dual role is through doping and introducing 26

PAGE 41

27 non-stoichiometry. This can be easily achieved in oxide materials such as zinc oxide, indium oxide and tin doped indium oxide (ITO). 43-44 Work done by Chopra et al. has produced a review of transparent conductors, including ITO. 42 Figure 1-3 shows graphically how various oxides rate for transmission and resistance properties. 42 ITO is compared with other transparent conducting oxides such as zinc oxide (ZO), tin oxide (TO) and antimony tin oxide (ATO). Clearly, ITO films possess one of the highest percent transmission, %T, and are among the lowest in measured sheet resistance, R s . ITO has been one of the most widely used and successful transparent conductors, and is well known for its high transparency in the visible region of the spectrum (>80%), strong reflectance in the infrared region and high radiation resistance. 45-46 In order to understand where ITO obtains its superior properties, we must first discuss the mechanisms behind its conductivity and optical transparency. Figure 3-1 Transmittance and sheet resistance for various transparent conducting oxides; Note: fabrication method identified in parentheses.

PAGE 42

28 In the undoped state, indium oxide (IO) films are generally polycrystalline with a cubic bixbyte structure, a lattice constant, a, of 10.118 , and a typical grain size of about 100 . IO is considered a semiconductor with a direct band gap of 3.5 eV and an indirect gap of 2.5 eV. It is an n-type conductor, meaning that the primary carriers are electrons produced from oxygen vacancies in the film. Doping with tin (Sn), to create indium tin oxide (ITO), enhances its conductivity. ITO films have the same structure as IO but have a small change in the lattice constant, 10.118
PAGE 43

29 between the Sn atoms that occur as they begin to come in close contact. 47 As a result, there is an optimum doping level in Sn:In 2 O 3 that generates the lowest film resistivity, which occurs around 10% Sn. Figure 3-2 Free electron density of states as a function of % Sn doping in In 2 O 3 . The change in the free electron density in ITO films, due to the incorporation of tin, also has a pronounced effect on its optical behavior. Figure 3-3 shows typical absorbance, A, transmittance, T, and reflectance, R, curves for an ITO film. 22 Absorbance is the fraction of light absorbed by the material relative to the incident light. Transmittance is that fraction of light that passes through the material, and reflectance is that which is reflected at the surface. Note that the oscillations seen at the short wavelengths are Fabry-Perot interference oscillations resulting from the measurement of a thin film. ITO is characterized by a high optical transparency in the visible and near-IR regions, but is limited due to reflection losses at the surface (1-2%) and absorption in the film (1-2%) due to free carrier absorption. This effect of free carrier absorption becomes pronounced in the near-IR as evidenced by the sharp onset of reflection. This behavior

PAGE 44

30 can be modeled using classical Drude theory, which describes the plasma oscillation of the free electrons through the equation, 21242cMpmNec (3-1) where M is the dielectric constant of the material, m c is the effective mass for electrons in the conduction band, and N is the free electron density. The plasma frequency, p , is by definition the wavelength at which the real part of the complex dielectric constant, r , is equal to zero. The observed absorption peak that occurs with the rise in reflection is due to plasma resonance. For > p , r is negative, and as a result, the plasma is reflective, coinciding with the R onset in the spectrum. It is important to note that the free electron density affects where the wavelength of the plasma resonance occurs. As the free electron density increases, the plasma wavelength shifts towards shorter wavelengths, i.e. higher energies. Hence, the free carrier density in ITO can greatly alter the transparency region in these films. In addition to affecting the near-IR reflection, the density of free carriers also influences the UV absorption edge in ITO. Similar to the plasma oscillation, an increase in the free electron density results in a blue-shift in the intrinsic absorption edge. This effect is known as the Moss-Burstein shift and is due to a filling of states in the conduction band. As the number of free electrons is increased, the conduction band begins to fill at the bottom, essentially forcing the bottom of the conduction band to higher energies. Assuming parabolic band edges, this shifting in the band edge energy can be described by 3238*2NmhEvcBM (3-2)

PAGE 45

31 where N is the free electron density and m vc * is the conduction band effective mass, which also considers valence band curvature effects. In the discussion of the absorbance data in a later chapter, it will be important to consider all potential reasons for a shift in the band edge of the quantum dot films. This is especially true since a blue-shift in the absorption edge is also an obvious response in a film due to quantum confinement effects, as previously discussed above in Chapter 2. Figure 3-3 Typical absorbance, A, transmittance, T, and reflectance, R, curves for an ITO film. Finally, there are certain changes in the resistivity and optical behavior of ITO as a result of thermal annealing. Because applying a heat treatment to the nanocomposite films is an important step in developing the quantum dot, as discussed in section 3.3, it is necessary to understand the influence that temperature can have on the properties of

PAGE 46

32 ITO. 48-51 The two most prominent reactions that occur in ITO films during thermal anneals are grain growth and oxygen diffusion into the films, both of which can have pronounced effects on the resulting electrical and optical properties. ITO, in its as-deposited state, has a pale yellow appearance. At temperatures around 100-200C, the films become clear, losing their color. This happens because of a rapid grain growth process that occurs at these temperatures, removing any absorbing defects. As this process takes place, the conductivity is also greatly increased as the rapid grain growth spurs an increase in the oxygen vacancy concentration. 48 Because oxygen vacancies are so vital to the film’s conductivity, it is important to note that the atmospheric partial pressure of oxygen inside the furnace can profoundly affect the resistivity. Often, an inert gas or a reducing gas environment is passed through the furnace to displace any atmospheric oxygen that might be present. Assuming that the oxygen partial pressure during heat treatment is sufficiently low, then the effects of grain growth dominate the changes that happen in resistivity and absorption. Initially, the decrease in resistivity is rapid as many new carriers are created due to an increase in oxygen vacancies. But above around 200C, the resistivity tends to flatten out. At this point grain growth slows down, and any further decrease in resistance is due to mobility contributions as a result of any grain size increases. In general, the point to make is that, in the proper atmosphere, an increase in temperature results in an increase in carrier density, carrier mobility, and, hence, an increase in conductivity and transparency. Given all of the data published on ITO films and the effects that various parameters, such as deposition method, dopant concentrations, and temperature, have on

PAGE 47

33 the resulting electrical and optical properties, it is important to understand these effects. 42,47-62 In the case of this research, ITO was chosen because of its superior conductivity and high transparency region. All of the properties pertaining to the deposited ITO films in this study will be described in future chapters. Germanium For the quantum dot structures, the important aspect in materials selection is that the material be easily made into nanocrystals and should possess large energy shifts with crystal size. It is also necessary that the quantum dot not interact strongly with the matrix material, for example, a material that forms an oxide coating around the dot would prevent carrier transport. In addition, the band gap energies, as a function of dot size, must coincide with the majority of the energy in the solar spectrum in order to have the best conversion efficiency. Considering all of the important criteria, germanium was chosen for the quantum dot material. Germanium is a popular material for synthesizing quantum dot structures using many different deposition techniques, indicating that ease of fabrication is achievable. 63-66 Based on electronic structure and the band gap energy of the bulk material, both Si and Ge make interesting candidates. Both exhibit a large shift in energy, E, with size as was shown earlier. However, their oxidation energy is very different. The formation of an oxide at the interface of the quantum dot and the ITO matrix material would greatly affect the electrical properties of the films since intrinsic oxides are known dielectrics. In such a situation, even if carriers were created within the quantum dot, they would be prohibited from being transported across the junction and, hence, no enhanced photoconductivity would be measured. Table 3-1 compares the

PAGE 48

34 oxygen dissociation energies for various oxides. 67 When comparing germanium dioxide, GeO 2 , and silicon dioxide, SiO 2 , with the ITO components (indium oxide, In 2 O 3 and tin oxide, SnO 2 ), it is clear that Si will take oxygen from both, while Ge is less likely to do so. In 2 O 3 is a more stable oxide than GeO 2 , and SnO 2 is only slightly less stable than GeO 2 . This table indicates that Ge is less likely to oxidize in ITO than Si. However, some care is necessary since the GeO 2 dissociation energy is greater than that for SnO 2 . As mentioned previously, it is also critical that the quantum dot be capable of varying its band gap, with decreasing size, over a relatively large range. Recall from Figure 1-7 that the creation of Ge dots is possible over a wide size range, and that this results in a dramatic shift in band gap energy. The energy shift spans a region from around 1.5 eV to 5 eV, making it perfectly suitable for solar spectrum irradiation. Table 3-1 Dissociation energy per oxygen atom for various molecules. Molecule Dissociation Energy (kcal/mol) SeO 2 27.5 TeO 2 38.8 Bi 2 O 3 46.0 CdO 60.9 SnO 2 69.4 GeO 2 96.3 TiO 2 109.0 In 2 O 3 111.3 CeO 2 122.5 B 2 O 3 150.9 SiO 2 153.9 Al 2 O 3 195.9

PAGE 49

35 0123456CdSCdSeCdTeSiGeSemiconductorEnergy Shift (eV) Figure 3-4 Relative degree of bandgap energy shifts due to quantum confinement for several direct and indirect semiconductors. It is of interest to discuss where this dramatic energy shift evolves from in germanium, and why the effect is not so pronounced in other semiconductors such cadmium telluride. Figure 3-4 compares the reported energy shifts for several semiconductors, illustrating the fact that confinement effects in Ge lead to more dramatic energy changes. Comparing the shifts seen in Ge and CdTe, it is seen that over similar size regions, CdTe shifts its optical band edge by approximately 0.5 eV, while Ge far exceeds this with a shift of 5 eV. When evaluating the reasons for the large difference, the first obvious thing to think about is the differences in carrier masses for each material. Recall from equation 2-2 that the energy states of a quantum dot are inversely proportional to the effective carrier mass. The smaller the mass, the larger the blue-shift in energy. Germanium has a carrier mass of 1.64m o for the electron and 0.04m o for the

PAGE 50

36 hole, while those for cadmium telluride are 0.1m o and 0.4m o respectively. 26,68 The resulting effective carrier mass for Ge is 0.04m o and that for Si is 0.08m o , indicating the Ge clearly has a smaller effective carrier mass. However, this difference can not account for the large difference in band edge seen experimentally. The answer lies in a comparison of the band diagrams for the two materials, shown if Figure 3-5. Figure 3-5 Energy band diagrams for germanium, Ge, and cadmium telluride, CdTe. When a material is confined to a finite size, k is no longer considered a good quantum number, as the once nearly continuous energy bands in the bulk become discrete combinations of the bulk bands. 69 The result is a shift in energy levels that derive from the highly curved regions in the band diagrams of the bulk. So the degree of curvature in the bands holds the key to the resulting extent of energy shift of the confined material. The bands in CdTe which give rise to the lowest energy transitions are relatively flat,

PAGE 51

37 implying that the size of the quantum structure will have only a small effect on the optical properties. In Ge, however, the bands that contribute to energy shifts are highly curved, resulting in more dramatic changes in the optical band gap with size confinement. Based on all of the evidence discussed, germanium is an excellent choice for the quantum dot material. Not only are they easily fabricated, but they also have the ability to absorb over a wide range of energies. Several of the important materials properties associated with bulk germanium are listed in Table 3-2. 70-71 Table 3-2 Basic materials properties of Germanium. Lattice Constant () Density (g/cm 3 ) Dielectric Constant Energy Gap (eV) Melting Point (C) Refractive Index Intrinsic Resistivity (cm) 5.66 5.32 16.0 0.67 937 4.0 47 Sapphire Because temperature effects play a large role in the resulting film structures, it is of primary importance that the substrate be non-reactive and possess a thermal expansion coefficient, , similar to that of the ITO film. It was seen that if a glass silica substrate was used, the post-annealed films were fraught with cracks due to the mismatch in . Also, since the two most important properties for evaluating the usefulness of germanium quantum dots in indium tin oxide were absorption and photoconductivity, it was necessary that the substrate upon which the films were deposited be transparent in the spectral region of interest. For these reasons, random-axis oriented sapphire was chosen as the ideal substrate material for this application. Sapphire is a synthetic crystal form of aluminum oxide, Al 2 O 3 , with a hexagonal crystal structure. The transmission of sapphire is excellent, limited primarily by surface

PAGE 52

38 reflections. In the wavelength regions extending from visible through infrared, sapphire transmits well over 80% of the incident photons. In addition, sapphire is chemically inert, and is extremely hard, ranking 9 on the Moh’s hardness scale. 72 Finally, with respect to its temperature dependent properties, sapphire has a coefficient of thermal expansion of 7.7x10-6/C (compared to 7.2x10-6/C of ITO) and a softening point of 1800C. 72 Table 3-3 provides a list of some of the fundamental materials properties of sapphire. Table 3-3 Basic materials properties of sapphire. Density (g/cm 3 ) Softening Point (C) Refractive Index Moh Hardness Coefficient of Linear Expansion (/C) 3.98 1800 1.8 9 7.7x10 -6 RF Magnetron Sputtering Given the practical and theoretical importance of quantum dot structures, much research has been done, and continues to be done, towards optimizing the conditions for their fabrication. To this end, many techniques have been used in creating Ge quantum dots, including ion implantation, inorganic solution phase synthesis, dc magnetron co-sputtering and rf co-sputtering. 63-65,73 It was discovered within our laboratories that a dual gun, rf magnetron sequential sputtering technique was very successful in creating Ge quantum dots within a SiO 2 matrix. 33 This served as a basic model for the fabrication of Ge within ITO. The method consists of alternately exposing a substrate to two separate rf magnetron sputtering guns (see Figure 3-6). Each gun is fitted with a different target, one containing the ITO glass and one containing the Ge semiconductor material. By varying the amount of time that the substrate is held over each target and the deposition rate at

PAGE 53

39 each gun, the relative volume fraction of semiconductor present in the sample can be varied. The result is a multilayer, thin film made up of repeating layers of Ge and ITO. Figure 3-6 Schematic representation of the dual gun rf magnetron sputtering system. To create these multilayer films, a substrate is secured onto a sample platter mounted 5 inches above the targets, all enclosed within a stainless steel box chamber. The chamber is evacuated by a turbomolecular pump, which is backed by a roughing pump. The base and operating pressures, 2x10 -6 T and 3 mT respectively, are observed using Penning and thermocouple vacuum gauges. 74 The relative percentage and thickness of each material deposited onto the substrate are completely tailorable by changing the initial computer program input. This automated program would make the necessary calculations based on the user input, and use this to control the rotation of the sample platter by means of a stepper motor. Once the optimum deposition conditions for rf power and chamber vacuum pressure was experimentally determined, a set of investigative samples were prepared. In order to produce the highest quality film possible, precautions were taken to ensure cleanliness and purity. The substrates were carefully set in a beaker of methanol and placed in an ultra-sonic bath for 30 minutes and then dried with bursts of N 2 . During

PAGE 54

40 deposition, it is important to note that the substrate was not heated and no gas, other than Ar, was introduced into the chamber. The targets consisted of 2 inch round disks mounted onto their respective guns using silver paste. The ITO target was made up of In 2 O 3 -SnO 2 (90/10 wt%) of 99.99% purity, and the Ge target was of 99.999% purity. Figure 3-7 Absorbance versus wavelength for ITO films deposited at different rf powers. The ideal deposition parameters for sputtering Ge in the current system were previously determined. 33 Successful films of Ge quantum dots in SiO 2 were obtained under an operating pressure of 3mT and a rf power of 40W, corresponding to a power density of 1.97 W/cm 2 . What was not known, however, was the optimum rf power for the ITO target. Because high transparency and conductivity of the ITO were the most desirable properties, a series of ITO films were prepared at various rf powers to determine the best condition. Figures 3-7 and 3-8 show absorbance, A, and resistivity, , for as-deposited ITO films produced at rf powers of 20-100W, all fabricated under the same deposition conditions. As can be seen in the absorbance plot, the transparency of

PAGE 55

41 the films seem to be nearly identical regardless of the rf power chosen, at least in the wavelength range of interest. In contrast, the resistivity of the ITO films is quite dependent on the rf power. In fact, the higher the power the higher the resistivity. One interesting thing to point out, however, is the deposition rate of the ITO with rf power. Figure 3-9 shows the deposition rate of the ITO gun with applied rf power. The deposition rate was calculated by sputtering an ITO film for a pre-determined amount of time. The resulting film was measured for thickness using a profilometer and averaged over several measurements. With a film thickness and deposition time, a deposition rate was calculated. As is expected, the rate at which the ITO gun deposits the ITO film onto the substrate increases with applied rf power. A high deposition rate is desirable not only for time’s sake, but also to produce high purity films. This is obtained with high rf powers. But we have seen that high rf powers also produce a more electrically resistive film, which is undesirable. To balance these two competing terms, a higher rf power was chosen that also gave an acceptable measured film resistivity. Thus, an rf power of 80 W was determined to be the optimum setting for deposition of ITO in this rf magnetron sputtering system. Table 3-4 outlines the deposition parameters used for the fabrication of the Ge/ITO multilayer films.

PAGE 56

42 Figure 3-8 Sheet Resistance and Resistivity of ITO films as a function of rf power. Figure 3-9 Deposition rate of the ITO gun as a function of rf power.

PAGE 57

43 Table 3-4 RF magnetron sputtering conditions for both Ge and ITO. Base Pressure, P base (Torr) Operating Pressure, P op (Torr) Rf Power, Ge (Watts) Ge deposition rate (/s) Rf Power, ITO (Watts) ITO deposition rate (/s) 2*10 -6 3*10 -3 40 0.8 80 2.5 With the best set of thin film fabrication conditions determined, a thin film consisting of germanium and indium tin oxide alternating layers was created. However, remember that this simply gets us to the multilayer film stage. In order to create the Ge quantum dots from the stacked layers, post-deposition processing is needed. The details involved in transforming sandwiched Ge layers into isolated quantum dots are discussed in the following section. Quantum Dot Formation Once the Ge and ITO multilayer films have been deposited, they still need to be annealed in order to obtain Ge quantum dots from the layered structure. The application of heat provides the energy necessary for the as-deposited amorphous germanium layers to diffuse and form critical size crystalline Ge clusters for nucleation and growth. The Ge forms and grows isolated clusters to reduce its surface area to volume ratio and, thereby, lower its energy state. 75 Fortunately, the nucleation rate is high enough to produce many separate nuclei of germanium, allowing the formation of many dots per layer. Of course, anneal time and temperature play an important role in determining the size of the resulting crystals. In addition to simple nucleation and growth taking place during anneals, there is also the issue of diffusion between the dot material and the matrix. The details of this potential materials interaction will be discussed in detail in Chapter 4. For now, the focus is on the furnace set-up and ambient conditions seen by the films during the post-deposition anneals.

PAGE 58

44 The heat treatments were performed using a fused silica tube programmable furnace. The films were placed, film side up, in a fused silica boat and positioned in the center of the hot zone of the furnace. The temperature was monitored using two type-K thermocouples; one inside the tube and one just outside, to ensure precise temperature readings. The atmosphere was purged with ultra-high purity Argon (Ar) gas, and was allowed to purge the tube for a minimum of one hour prior to the start of the furnace run. The system was created to minimize the likelihood that outside air was present during anneal; both ends of the tube were capped with silica end-caps that permitted the Ar to flow in one end and out the other through a bubbler. In this way, the ambient air was freed of oxygen, preventing oxidation reactions to occur with the film. The variables commonly associated with heat treatments are ramp rate, temperature and time. In order to avoid thermal shock from occurring in the film, the ramp should be kept at a reasonable level, typically 10C/min. Because the material to be grown into crystallites is germanium, the appropriate crystallization temperature must be kept in mind when choosing the anneal temperature. In this case, the crystallization temperature is 630C, which means that this should be the target temperature. However, in order to understand the dynamics of crystallization and growth of these structures, a series of temperatures were chosen of 600-1000C. Similarly, several times were used to determine its effect on the resulting film structure, in the range of 6 minutes to 1 hour. Another important parameter for film fabrication is the relative thickness of each of the germanium and indium tin oxide layers. This is controlled by the volume percent of germanium and its layer thickness. Because the goal is to produce small, isolated clusters of germanium from a layer, it is necessary that the initial layer thickness be

PAGE 59

45 sufficiently thin. If the germanium layer is too thick or there is too much of it in the overall film, then quantum dots will not be favored to form during anneals. Also, the volume percent of germanium was kept small in order to promote quantum dot growth and to limit interactions between the dots. For these reasons, the base conditions for creating these quantum dot composite structures were a total germanium content of 5vol% and a layer thickness of 15 . These parameters caused the ITO volume percent to be 95% and its thickness to be 280 . Previous studies of rf magnetron sputtered Ge, Si and CdTe in SiO 2 has indicated that these fabrication conditions are ideal. 76 Using these initial choices and basic information, a sample matrix was produced. The idea was to be able to make correlations between the variables of temperature, time, Ge vol% and Ge layer thickness in order to determine their impact on the resulting film microstructure and behavior. In the following chapters, insights will be made into how the processing conditions for the films resulted in specific film morphology, and optical and electrical behaviors.

PAGE 60

CHAPTER 4 MICROSTRUCTURE In order to have a clear understanding of the microstructure or overall physical make-up of the multilayered quantum dot thin films several techniques were used. While it is expected, from previous studies, that from the as-deposited amorphous germanium layers isolated quantum dots will form, it is important to be able to measure their size and determine the structure. 76 From micrographs taken with the transmission electron microscope, TEM, a qualitative evaluation was made of the films as well as a quantitative measurement of the quantum dot size. Further information regarding the compositional constituents present and their relative positions within the film were revealed through automated x-ray spectral imaging (SI) analysis. Details about the crystallinity and structure of the germanium and indium tin oxide were investigated by Raman spectroscopy and x-ray diffraction methods. Transmission Electron Microscopy Theory of Operation The TEM uses electrons instead of photons, but operates in a similar way to an optical microscope, with condensing and objective lenses. The resolution of a TEM image can be as high as 1.5 , but the sample must be made very thin, down to approximately 100 nm. Figure 4-1 illustrates the secondary signals created when an ionizing electron is directed onto a sample. 77 The electrons that are transmitted through the foil or thinned sample are used to create an image in the TEM. Different TEM 46

PAGE 61

47 configurations have been developed, including the high resolution TEM (HRTEM) and the scanning TEM (STEM). 77 Figure 4-1 Signals generated when an electron beam interacts with a specimen. The images created in the current work were obtained using a JEOL 5900LV instrument. To take advantage of the x-ray spectrum imaging capabilities of this particular microscope, a detailed discussion of which is presented in a following section, the TEM was operated in scanning mode, i.e., STEM. Figure 4-2 illustrates the principles of operating in STEM mode. 77 Scan coils are used to scan the beam over the specimen, while simultaneously scanning the CRT, where the image is formed by a serial process, rather than a parallel process in conventional TEM. All of the STEM images produced for the multilayer quantum dot films in the current study were done with annular dark-field (ADF) imaging. Instead of using a bright field (BF) detector that collects the direct beam electrons, an annular detector is used to collect all of the scattered electrons.

PAGE 62

48 Figure 4-2 The principle of forming a scanned TEM image. Sample Preparation To generate an image using the TEM the specimen must allow the electrons to pass through. Because of this, the TEM sample must be made sufficiently thin to allow electron transparency. A novel technique known as a focused ion beam or FIBS was used to generate cross-sectional samples. The FIBS is essentially a scanning electron microscope (SEM) with an ion beam attachment. With the sample tilted relative to the ion beam source, a very small section can be cut away in cross-section and smoothed to the correct thickness. By using the SEM imaging capabilities, the operator can visually monitor its progress, which is important since the dimensions of the resulting TEM sample are much smaller than can be distinguished with the naked eye. The final sample is cut away from the bulk film where it falls into the trench created in its absence and is taken out of the SEM. Then, using equipment that was first developed for invitro fertilization, the TEM sample is plucked from its trench and placed on a copper TEM grid for analysis.

PAGE 63

49 Quantum Dot Growth with Temperature As indicated previously in Chapter 3, germanium quantum dots are formed through post-deposition anneals of the germanium/indium tin oxide alternating thin film structures. To understand the dynamics of such a process, it is important to monitor the development of the quantum dots as a function of anneal temperature. STEM images offer a unique and useful way to observe the changes in the microstructure of the films as the once continuous germanium layers transform into isolated quantum dots of various sizes. As will be discussed in a further section, such imaging also provides necessary insight into potential materials interactions that may occur during the heat treatments. By developing a series of images as a function of temperature and using them in conjunction with other characterization techniques, the growth and crystallization processes that occur in the films can be understood. All of the films presented in this section consist of germanium (Ge) and indium tin oxide (ITO) alternating layers. The overall Ge content is 5% by volume, corresponding to a nominal germanium layer thickness of 15 and ITO layer thickness of 285 . The first layer deposited onto the sapphire substrate is always ITO, and the top layer is also always ITO. The range of temperatures used for comparison are the as-deposited state, or heat-treated at 400C, 600C, 660C, or 700C, for times of 30-60 minutes as specified. All anneals were done in an argon gas atmosphere (see Chapter 3 for details). Figures 4-3 and 4-4 present the annular dark field STEM images obtained.

PAGE 64

50 Figure 4-3 Indium Tin Oxide film annealed to 800C for 1 hour. For a baseline indicator, a film containing only ITO is presented in Figure 4-3. It is a 5200 thick film that has been annealed to 800C for one hour. Notice that it is a continuous film whose ITO crystalline grain structure is columnar, as expected. 52 Also worthy of note is the structure present near the substrate interface. Keep in mind that the topmost, bright layer is simply a gold/palladium (Au/Pd) metallization layer deposited as a means to prevent charging effects during FIBS preparation. Another important baseline indicator is the as-deposited state of the multilayer films. Figure 4-4(a) shows what such a film looks like before any heat treatment has been performed. The structure consists of alternating layers of the Ge and ITO materials. It is a continuous multilayer structure free of voids or quantum dots.

PAGE 65

51 Figure 4-4 STEM images of 15/5% germanium multilayer composite films as a function of temperature (a) as-deposited, (b) 400C 1 hour, (c) 600C 1 hour, (d) 650C 6 minutes, (e) 660C 30 minutes, and (f) 700C 12 minutes.

PAGE 66

52 Of particular interest is to observe the changes, if any, that take place when the composite Ge and ITO films are annealed. Figure 4-4 illustrates the evolution that takes place upon increased temperature anneals. Note that the appropriate scale for each micrograph is indicated in the figure. As discussed, Figure 4-4(a) shows the as-deposited state for the multilayers. Figure 4-4(b) is a 1m thick, 15/5% Ge composite multilayer film that has been heat-treated to 400C for one hour. After being exposed to 400C, the alternating layered structure is still intact. The only significant change has occurred within the ITO layers, where clearly they have crystallized. There is no indication of quantum dot formation within the Ge layers. However, as seen in Figure 4-4(c), a similar film heat-treated to 600C for one hour visibly demonstrates that something is beginning to happen in the Ge layers, indicating the start of quantum dot formation. Clearly, as evidenced in Figure 4-4(d), once the film has been annealed to 650C there is an obvious break-up in the Ge layer as nanocrystals are nucleating and growing. Once the film is annealed to 660C for 30 minutes, as in a 5000 multilayer film shown in Figure 4-4(e), the formation of the Ge quantum dots is complete, as obvious dot-like structure can be seen within the once continuous Ge layers. This point is further made in Figure 4-4(f) for a 1 m film heat-treated to 700C for 12 minutes. (Please note that in these TEM samples, the data is collected over a thickness of approximately 100 nm, meaning that there may be other quantum dots superimposed behind the dots seen in the images, or there may still be some continuous layer structure to the Ge dots). As is apparent from the results, the formation of germanium quantum dot structures within an indium tin oxide matrix is possible for the temperature region shown. In later discussions it will be established that by annealing beyond the range of

PAGE 67

53 temperatures presented other dynamics take over, limiting the processing boundaries for quantum dot formation in this system. But first it is of interest to analyze the sizes of the quantum dots obtained in the present films and see how they vary with temperature. In addition, by studying the variation of size over the film thickness, some insights into the growth process can be recognized. Quantum Dot Size Analysis Many techniques have been used to determine the average size of quantum dots, including indirect measurements from Raman and x-ray diffraction data. 78-79 However, the most direct way is by physically measuring the dots from the STEM micrographs. Software, for example Adobe Photoshop as was used in this case, makes measuring and averaging the dot size very straightforward. In addition, by acquiring quantum dot size information in this way, one can also distinguish different size ranges over the thickness of the film, if any exist, which ultimately leads to more insightful data. Figure 4-5 provides larger images of some of the previously illustrated cross-sectional samples in order to make the quantum dot sizes more apparent. By enlarging an image and using the measurement tool in the software, an average crystal size can be obtained. In Figure 4-5 (a)-(c), the sizes of the nanocrystals for the particular region shown are evident. In the film annealed to 650C for 6 minutes, dots are just beginning to become isolated, averaging 10 nm in size. However, note that many of them are elongated. The shape of the nanocrystal has pronounced effects on the resulting quantized energy spectrum, which will be discussed in more detail in Chapter 5 when the optical absorbance of the films is analyzed. In Figure 4-5 (b) and (c), whose films were annealed to 660C for 30 minutes and 700C for 12 minutes respectively, the

PAGE 68

54 nanocrystals have become more spherical. The average quantum dot size for the 660C treated film is 16nm and for the 700C treated film the average size is 12.6nm. It is interesting to note that, due to the small temperature range over which dot formation is possible, the variation in the controllable average Ge dot size is limited. Table 4-1 illustrates this point by presenting the calculated average quantum dot sizes for the sample set. For anneal temperatures ranging from 400C, where no dot formation is apparent, to 700C, where isolated quantum dots are evident, the range in sizes is only 9 nm – 22nm. The smallest dot size was produced in the film heated to 600C for 1 hour, however they were within a somewhat continuous thin Ge layer. The largest quantum dots were found in a film heated to 700C for 30 minutes. Table 4-1 Average quantum dot sizes for various 15/5% Ge multilayer films. Film Thickness (m) Anneal Temperature (C) Anneal Time Average Quantum Dot Size 1.0 600 1 hour 9 nm (within layer) 0.5 660 30 minutes 11-21 nm 0.5 700 6 minutes 12 nm 1.0 700 12 minutes 12.6-25nm 1.0 700 30 minutes 10-22 nm 1.0 1000 6 minutes 15 nm

PAGE 69

55 Figure 4-5 STEM images of composite films containing quantum dots of germanium and indicating dot size (a) 650C 6 minutes a = 10nm, (b) 660C 30 minutes a = 16 nm, and (c) 700C 12 minutes a = 14.5 nm.

PAGE 70

56 Dot Size Variation Across Film While the analysis of the average quantum dot size from TEM micrographs may seem straightforward, the task is complicated by the fact that the size varies over the film thickness. As seen in the STEM image of Figure 4-6, the formation of quantum dots of Ge from the layered structure produces three distinct regions (A-C) of quantum dot sizes. The example shown is of a 15/5% Ge multilayer 5,500 thick film that has been annealed to 700C for 6 minutes. Such a short anneal time allows a glimpse into the beginning stages of quantum dot formation throughout the film. Figure 4-6 STEM image of a 15/5% composite film annealed to 700C for 6 minutes indicating three distinct growth regions – A (bottom region), B (middle region) and C (top region). Region A, consisting of the bottom 1/3rd of the film is characterized by a loss of the multilayered structure in exchange for the presence of pores. In region B, the middle 1/3rd of the film, has pronounced quantum dot formation with no porosity. Region C, the top 1/3rd, seems to have maintained the continuous multilayered structure with neither

PAGE 71

57 pores nor quantum dots present. The significance of this variation of quantum dot formation as a function of depth in the film will be dealt with in detail in the materials interaction section where the dynamics of the germanium and indium tin oxide diffusion processes will be discussed. But for now, it is important to recognize that there exist different regions within a single film, where the size of the quantum dot can vary. This must be kept in mind when using average crystal size information. X-ray Spectrum Imaging Theory of Technique By combining imaging and spectrometry a very powerful analysis tool is created. This is why x-ray spectrum imaging (SI) has become such a useful technique. 80 Simply having a micrograph of the structure contained within the sample will not enlighten the researcher on its compositional make up. Tools do exist that complement TEM images by giving compositional information, such as electron energy loss (EELS) and energy dispersive x-ray (EDX) spectroscopies, but these techniques only allow spot analysis and are inherently dependent on the interpretations of the user. 77 For example, in order to obtain insightful information about the chemical components and regions in a sample, the operator must first be able to identify where the chemically distinct regions exist. In addition, to get a comprehensive view of the area of interest it would take considerable time to accomplish. Over the last decade, much progress has been made in resolving these issues and recently, due to the advancement in computer power and speed, x-ray spectrum imaging systems have become an invaluable tool. A spectrum image is essentially a 2-dimensional chemical map of a material. From a 2-D STEM image, an array of points is created and a complete x-ray spectrum is taken from each pixel in the array using EDX.

PAGE 72

58 An immense amount of chemical data is created in this way, and in order to extract the relevant information and display it in a comprehensible manner, it is run through spectral imaging analysis software. The software was created and written at Sandia National Laboratories in New Mexico and is beginning to emerge as a commercially available technique through Thermo Noran. The spectral images present for the quantum dot films in the current research were obtained using this software. Note that for a 200nm square array, the resolution is 4 nm/pixel. 2-D Compositional Mapping Data It has been established, through the STEM images presented in a previous section, that quantum dot formation within the germanium layers in the composite films does occur at some critical post-deposition anneal temperature. But only relying on the contrast present within the micrograph to infer details about the chemical composition can lead to false conclusions. For example, in annular dark field images, porosity can easily be mistaken for quantum dots. It is therefore important to use SI mapping to help positively distinguish between different regions of interest within the films. Further, it is important to determine what chemical phases are present, especially when potential materials interactions are a concern. In this way, x-ray spectral imaging was utilized to help establish what chemical signatures are present in the micrographs and where the boundaries of the different signatures are located. SI data sets were obtained on the same sample series presented in the STEM images shown in the previous section. Figures 4-7 – 4-11 present the STEM images with the corresponding SI compositional maps, illustrating the evolution of the films with temperature. The point in evaluating the Ge/ITO multilayer films as a function of anneal

PAGE 73

59 temperature is to determine what changes are taking place at the different temperatures and to gain insights into the chemical nature of the resulting dot-like structures. Figure 4-7 illustrates the chemical signatures detected for a specific region in the 15/5% multilayer film heated to 400C for one hour . To be clear, the red square shown in the STEM image is the specific area from which the 2-D map was created, typically a 200 nm square section. The software outputs the x-ray spectra detected with a corresponding map. In each case, a separate color-coded map is produced for each chemical signature detected. The pixels colored red indicate the regions where the specific signature is strongest, while blue indicates little or no signal from that signature. Figure 4-8 shows the SI data collected for the 600C one hour film, where the beginnings of germanium quantum dot formation are evident as the germanium diffuses through the layers and clumps. At 660C for 6 minutes the film, as illustrated in Figure 4-9, has begun to form isolated germanium quantum dots surrounded by indium tin oxide. Finally, as seen in Figures 4-10 and 4-11, the germanium dots grow under anneal conditions of 660C for 30 minutes and 700C for 12 minutes, respectively. Further insight into these findings will be presented in the discussion section at the end of the chapter.

PAGE 74

60 Figure 4-7 X-ray Spectral Imaging data for a 15/5% film heated to 400C 1 hour.

PAGE 75

61 Figure 4-8 X-ray Spectral Imaging data for a 15/5% film heated to 600C 1 hour.

PAGE 76

62 Figure 4-9 X-ray Spectral Imaging data for a 15/5% film heated to 650C 6 minutes.

PAGE 77

63 Figure 4-10 X-ray Spectral Imaging data for a 15/5% film heated to 660C 30 minutes.

PAGE 78

64 Figure 4-11 X-ray Spectral Imaging data for a 15/5% film heated to 700C 12 minutes.

PAGE 79

65 Raman Spectroscopy Theory of Technique Raman spectroscopy takes advantage of the Raman effect, so named after the man who discovered it in 1928. 81 Simply stated, the Raman effect is the inelastic scattering of light by matter. When a photon of visible light too low in energy to excite an electronic transition interacts with a molecule, it can be scattered in one of three ways. It can be elastically scattered, retaining its incident energy (also known as Rayleigh scattering), or it can be inelastically scattered by either giving energy to the molecule (Stokes scattering) or taking energy away from the molecule (anti-Stokes scattering). 81-82 Raman scattering occurs when light incident on a material interacts with the rotational and vibrational energies in the molecule thus, altering the wavelength of the incident light. Since the change in the incident energy is characteristic of the type of molecule present, the scattered radiation from a sample can be collected and used to help characterize length, strength and type of molecular bonds inherent in a sample. 83 The Raman technique is also a good tool for monitoring the crystallinity of a material. For example, in going from an amorphous to a fully crystalline sample of the same material, the Raman peak can be seen to evolve from a broad peak to a strong, narrow crystalline peak and combinations of the two, depending on the degree of crystallinity. This is specifically useful for germanium films, where a broad amorphous peak centered at 280 cm -1 can shift toward a strong, narrow peak centered at 300 cm -1 upon crystallization. 84 However, quantum confined materials can also cause a once strong crystalline peak to broaden. 85 The presence of small spherical particles of a material introduces another phenomenon observed in Raman spectra. Small particles, such as quantum dots,

PAGE 80

66 can actually shift the Raman line towards the low frequency end, i.e. towards the laser line. This has been seen experimentally in various semiconductor quantum dot materials, including silicon and germanium. 33,85-86 The fact that nanoparticles have little material to interact with means that the vibrational strength of the molecules with the incident radiation is diminished somewhat. As a result, less energy is given to the molecule during excitation producing in a small shift in the bulk crystalline Raman line towards the low frequency end. So, Raman spectroscopy can be used as a tool for establishing the presence of quantum dots within a film. Other factors, however, can also affect the shifting of the Raman lines, which can compensate any shift that may be present due to quantum confinement of particles. In particular, stresses that develop at the interface of the particles and the matrix can force the Raman line toward higher frequencies. 87 What is clear is that Raman spectra can differentiate between crystalline and amorphous structures. The effect of size is less clear. Raman Data Raman scattering spectra were obtained using a 0.6 m spectrometer with a liquid nitrogen cooled charge coupled detector (CCD) and a holographic notch filter. An Ar ion laser tuned at 514.5 nm excited the specimens. The CCD detector has a groove density of 1200 g/mm, corresponding to a pixel resolution of 0.6 cm -1 . The integration time for acquiring a spectrum varied between 1 and 30 seconds using software by SpectraMax. A block diagram of the basic Raman spectroscopic set-up is depicted in Figure 4-12.

PAGE 81

67 Figure 4-12 Block diagram of the Raman spectroscopic experiment. In order to determine which peaks are of special interest, a baseline must be created to identify known peaks from the substrate, etc. Figure 4-13 illustrates where the Raman peaks are located for sapphire, bulk crystalline germanium and indium tin oxide films annealed to 600C for one hour and 1000C for 30 minutes. The important points to note here is that there is no difference in the Raman signal between the sapphire and ITO films analyzed, suggesting that there is no observable ITO signal. It should also be seen that the Ge peak is located at 288 cm -1 for a pure germanium film annealed to 600C for one hour. A closer view of this region is provided in Figure 4-14. It is this peak that will be searched for in the composite multilayer films. Modeling of the Raman peaks was performed using Origin software to determine the experimental peak centers. The complete set of fits produced is detailed in Appendix A.

PAGE 82

68 Figure 4-13 Raman spectra for sapphire, bulk germanium and indium tin oxide films annealed to 600C 1 hour and 1000C 30 minutes. Figure 4-14 A more detailed look at the region around the Ge Raman peak from figure 4-13.

PAGE 83

69 Figure 4-15 helps illustrate the effect that quantum confinement has on the peak strength and position. For comparison is a 250 Ge layer deposited between two layers of ITO with a strong peak centered at 297 cm -1 , a 125/5% multilayer with a peak at 294 cm -1 , and a 15/5% multilayer film possessing a peak at 293 0.6 cm -1 . All anneal conditions for each film is indicated in the figure. Clearly, both quantum confinement effects and effects due to stresses are playing a role in the resulting Raman behavior in these films. Figure 4-15 Raman data for bulk Ge film heated to 600C 1 hour, 250 Ge film layered between ITO heated to 600C 1 hour, 125 Ge multilayer composite film heated to 600C 1 hour, and 15 Ge multilayer composite film heated to 700C 30 minutes. Finally, the changes in the Ge Raman peak for multilayer films heat-treated at different temperatures are presented in Figure 4-16. At 400C the Raman peak is dominated by the broad, amorphous behavior with combined peak positions at 264 cm -1 and 281 cm -1 . Clearly, a dramatic change has taken place once the film has been

PAGE 84

70 annealed to 660C, as the Ge peak shrinks, broadens and shifts. At 700C, the film possesses a sharper peak near 299 cm -1 . Figure 4-16 Raman data for bulk Ge film heated to 600C 1 hour, 15/5% multilayer heated to 400C 1 hour, 15/5% multilayer film heated to 660C 30 minutes, and 15/5% multilayer film heated to 700C 12 minutes. X-ray Diffraction Theory of Technique It was seen that the Raman spectra of the multilayer, quantum dot films helped to provide insight into the growth and crystallization of the structures as a function of anneal temperature. But one of the most widely used techniques for evaluating the crystallinity and structure of films and powders is x-ray diffraction, and serves as a useful counterpart to the Raman data. X-ray diffraction is essentially a method used for evaluating the lattice spacings, also known as d-spacings, present in a sample. 25 The technique is analogous to an optical diffraction experiment whereby the line spacings of the grating can be determined by the resulting diffraction of light. In x-ray diffraction, the incident

PAGE 85

71 energy is x-rays. Because of its small wavelength, relative to visible light, x-rays can interact with the atomic planes in the internal structure of materials, causing the x-rays to diffract at certain angles of incidence, and create a unique set of diffracted or reflected x-ray peaks. 25 The technique is based on Bragg’s law, which states that these reflected x-rays from a set of lattice planes will only occur at specific angles. These angles can be predicted using the mathematical expression of Bragg’s law, stated as 2d sin = n, where n is an integer, is the angle of incidence for the x-rays, is the wavelength of the diffracted x-rays, and d is the spacing of the planes in the sample. The x-ray diffraction data presented were produced using a Scintag XDS 2000 PTS diffractometer and a grazing angle x-ray diffractometer (GIXD). 88 GIXD has an advantage in evaluating thin films because it effectively increases the thickness of the film that is able to interact with the x-ray source. Without taking advantage of the grazing incidence, the x-ray path length through the film is too short, which can greatly reduce the intensity of the diffracted signal. The GIXD method overcomes this by causing the film to seem thicker than it actually is. This method was employed for the current films in an attempt to sample more of the germanium in the films. Because the overall germanium content was generally only 5%, it was hoped that GIXD would effectively increase the Ge signal seen by the x-rays. X-ray Diffraction Data The goal in obtaining x-ray diffraction (XRD) data for the composite multilayer samples was to isolate the Ge signal in order to say explicitly that crystalline Ge was present. Before this can be done, the XRD peaks corresponding to the ITO matrix material must first be identified. Figure 4-17 provides the XRD data for an ITO film

PAGE 86

72 annealed to 600C for one hour. The peaks present in this sample are consistent with cubic In 2 O 3 . It is also important to establish where the strongest Ge peak occurs in XRD data. Figure 4-18 provides the peaks for a Ge film heat-treated to 600C for one hour. The peaks obtained correspond to cubic Ge as shown in Table 4-2. Figure 4-19 is a slow step scan for the same Ge film in the region of the strongest peak. As seen, this peak occurs at 2 = 27.5. It is this peak that will be of particular interest in the composite films. However, this can be difficult to obtain for a small particle of material. Similar to the effect seen in Raman, the XRD peaks for a crystallite is also reduced and broadened. This can make it very difficult to discern among the background x-rays, especially for samples containing only 5vol% of the material of interest. Figures 4-20 and 4-21 provide XRD data in the region of the (111) Ge peak for 15/5% Ge composite films annealed to 650C for 6 minutes and 660C for 30 minutes, respectively. The (111) Ge peak is quite clear for the 650C film, whose peak center is located at 27.55. The same peak is more difficult to discern for the 660C sample, but a broad peak is suggested near 27.58. Because it is difficult to definitively say that, in fact, germanium is present in the 660C composite sample, it is instructive to look at the (100) In 2 O 3 peak near 30.5. Figure 4-22 compares this peak for pure ITO films and the 660C sample. What is obvious is the shift the peak toward a higher 2-theta value. This implies that the lattice spacing is somewhat smaller, which may be attributed to the presence of the Ge in the sample.

PAGE 87

73 Figure 4-17 XRD spectrum for an ITO film heated to 600C 1 hour. Figure 4-18 XRD spectrum for a Ge film heated to 600C 1 hour.

PAGE 88

74 Table 4-2 Characteristic XRD peaks for cubic germanium Germanium (cubic) a=5.66 2 Intensity h k l 27.2840 100 1 1 1 45.3061 57 2 2 0 53.6830 39 3 1 1 66.0174 7 4 0 0 72.8046 10 3 3 1 83.6872 17 4 2 2 90.0599 7 5 1 1 100.7616 3 4 4 0 107.3335 11 5 3 1 118.8705 6 6 2 0 126.4521 4 5 3 3 141.2272 2 4 4 4 152.9328 8 5 5 1 Figure 4-19 XRD spectrum for Ge film heated to 600C 1 hour, focusing on the (111) reflection.

PAGE 89

75 Figure 4-20 XRD spectrum for a 15/5% multilayer film heated to 650C 6 minutes, focusing on the (111) reflection for Ge. Figure 4-21 XRD spectrum for a 15/5% Ge multilayer film heated to 660C 30 minutes, focusing on the (111) reflection for Ge.

PAGE 90

76 Figure 4-22 GIXD spectra of ITO as-deposited, ITO heated to 600C 1 hour, and a 15/5% Ge multilayer film heated to 660C 30 minutes. Materials Interaction Concerns As with any composite materials system there is always the potential for the components to interact with each other. Often, this can lead to undesirable consequences, but ultimately can aid in the understanding of the behavior of the particular materials system under study. Such materials interactions are of particular concern and interest when exposed to high temperatures. It will be established that there exists a narrow range of fabrication conditions that permit the growth of germanium quantum dots within the indium tin oxide. These positive growth conditions will be described and the interesting interactions that occur when they are exceeded will be explained. Porosity and Germanium Ripening The first indication that something strange was happening in the germanium indium tin oxide materials system was apparent in the STEM and SI data collected in the films that were post-annealed to temperatures at and above 700C. It should be noted

PAGE 91

77 that, for the application desired for the quantum dot thin films, porosity could be a problem. The pores have the potential of acting as surface traps, causing any carriers created by the quantum dots to become trapped and recombine. Thus, it is important to understand the fabrication limits for the system and how it can be avoided. Figure 4-23 provides evidence of the first steps in the Ge diffusion and agglomerate formation that takes place within the films at elevated temperatures. This 15/5% Ge composite film, annealed at 700C for 12 minutes, shows an intact quantum dot structure in the center of the film. However, near the substrate interface the Ge has begun to diffuse and create large clumps. In this area signs of pore formation are apparent. When a similar film is now heated to 700C for 30 minutes, as shown in Figure 4-24, the quantum dot structure is still present in the two topmost regions. Notice, however, in the lower region near the substrate, that the once large particles of Ge are now empty pores. This phenomenon of Ge agglomeration leading to pore formation is even more pronounced in films annealed at 1000C. Figure 4-25 gives the SI data for a 15/5% multilayer film heated to 1000C for 6 minutes. The top layers demonstrate the characteristic Ge agglomerates, while the bottom layers have evolved beyond this stage to the pore stage. When a similar film is annealed at 1000C for 12 minutes, as shown in Figure 4-26, gross pore formation is dominant throughout the film. It is important to note that the relative chemical signature intensities, while they can not be used to determine strict stoichiometry, can provide some information about the reactions occurring. However, as a note of caution, it is important to understand that the presence of any In/Sn or O peak in a predominantly Ge signature can indicate one of two

PAGE 92

78 things. First it can imply that some reaction is taking place between the Ge and ITO into some combined phase of the elements. Secondly, it could simply be an artifact of the STEM process since it samples a thickness about 100 nm. What is clear is that some type of reaction is taking place that causes pores to form in the multilayer films at elevated temperatures. One of two things seems likely to be the answer to what is occurring in the films. It can either be due to an overall reduction in volume due to changing into a new phase, or it can be forming some gaseous species that escapes leaving the pores behind. To provide some insight, microscopic images of the surface of the annealed films were taken. Figure 4-27 illustrates the condition of the surface of the films after the heat treatments. It shows an ITO film heated to 800C for one hour compared with a composite film heated to 400C for one hour. Nothing very unusual seems to be happening, but then no porosity is present in these films. Also included in Figure 4-27 are surface images of multilayer films heated to 700C, 850C and 1000C, including the corresponding STEM image of the 700C sample. As seen in the figure, a composite film annealed to 700C for 12 minutes, where porosity is present, has clear evidence surface bubbling. In the cross-sectional image, it is obvious that gaseous products found a way to escape from the bottom of the film by creating an opening that extends to the top, leaving the bubble appearance on the film surface. These bubbles can leave some interesting patterns on the surface as seen in the surface images of composite films heated to 1000C and 850C. While it may seem obvious to test for the presence for an In 2 Ge 2 O 7 (IGO) phase, this is not easily accomplished. Table 4-3 compares the strongest XRD peaks for pure Ge, pure In 2 O 3 and In 2 Ge 2 O 7 . The differences between the x-ray signatures for the IGO

PAGE 93

79 are small, as the peak spectrum is almost identical to the combined spectra of Ge and ITO. Other researchers have attempted to make this distinction using XRD, but have concluded that it is not feasible. 89-90 Little data is reported in the literature for IGO. There have been studies performed where Ge doping of In 2 O 3 is done to enhance the conductivity of the films, but no indication of XRD or Raman characteristics have been published to date. 91-95 Figure 4-23 X-ray Spectral Imaging data for a 15/5% Ge multilayer film heated to 700C 12 minutes.

PAGE 94

80 Figure 4-24 X-ray Spectral Imaging data for a 15/5% Ge multilayer film heated to 700C 30 minutes.

PAGE 95

81 Figure 4-25 X-ray Spectral Imaging data for a 15/5% Ge multilayer film heated to 1000C 6 minutes.

PAGE 96

82 Figure 4-26 X-ray Spectral Imaging data for a 15/5% Ge multilayer film heated to 1000C 12 minutes.

PAGE 97

83 Figure 4-27 Magnified images of the films surfaces using an optical microscope for an ITO film heated to 800C 1 hour, a 15/5% multilayer film heated to 400C 1hour, a 15/5% multilayer film heated to 700C 12 minutes (with corresponding STEM image), a 15/5% mulitlayer film heated to 1000C 30 minutes, and a 15/5% mulitlayer film heated to 850C 30 minutes.

PAGE 98

84 Table 4-3 Comparison of characteristic XRD peaks for Ge, In 2 O 3 and In 2 Ge 2 O 7 Ge 2 Int. h k l In 2 O 3 2 Int. h k l In 2 Ge 2 O 7 2 Int. h k l 27.2804 100 1 1 1 30.5808 100 2 2 2 27.275 85 1 1 1 45.3061 57 2 2 0 33.1027 2 3 2 1 27.463 100 0 2 1 53.6830 39 3 1 1 35.4666 30 4 0 0 29.624 100 -2 0 1 45.6921 10 4 3 1 33.541 90 1 3 0 51.0388 35 4 4 0 34.295 90 -2 2 0 54.3719 2 6 0 0 46.633 80 -2 2 2 53.445 90 1 3 2 Discussion Figures 4-3 and 4-4 show the dramatic changes that take place when one goes from a simple ITO film to composite multilayer structures annealed at different temperatures. TEM images provide a detailed look into the microstructure of thin films, allowing insightful inferences about the growth process that takes place. It is very clear that up to 600C, no significant changes occur. However, at 600C germanium begins to diffuse within its layer, i.e. intralayer diffusion, to create the beginnings of quantum dot formation. Upon increased temperature and time, such quantum dot structures grow and become discrete and isolated within the ITO matrix. However, as seen in Figure 4-6, such quantum dot growth is not consistent throughout the thickness of the film. Three separate regions are created that seem to possess very different diffusion kinetics. In the example of the 700C 6 minute film, the short anneal time essentially froze the film in the preliminary growth process allowing a look into the different regions as they begin to form. The top layer maintains the intact multilayer structure that is inherent to the as-deposited state. The middle layer has quantum dot structures present as the Ge layers begin to diffuse and form dots of Ge. The bottom layer, that closest to the substrate, shows a loss of the multilayer structure

PAGE 99

85 with large pores forming. The formation of different regions within the film could be seen to some extent in all post-annealed films, becoming more pronounced in the films with a smaller overall thickness. The formation of different growth regions within a sample is often contributed to the presence of a temperature gradient. Normally for thin films such as these, such temperature gradients that may be present during anneals are neglected or non-existent due to the thin nature of the films. The configuration of the furnace and the properties of the substrate material could facilitate temperature gradient effects in thicker films, since the sapphire substrate is highly thermally conductive and Ar gas is passed over the top of the films. However, the thickness of the films in the current samples are too thin to have the temperature gradients necessary to explain the growth regions. Its more likely that regions develop due to the existence of pockets of trapped gas incorporated during the sputtering deposition process. These pockets of trapped gas are found near the substrate interface, apparent in the annealed pure ITO film shown in Figure 4-3. These features could facilitate the reactions and formation of pores that occur at elevated temperatures near the substrate in the annealed films and could explain the different growth regions observed. The presence of these regions, while unexpected and potentially devastating from an application point of view, offers a unique look into the growth processes for these quantum dot structures. The x-ray automated spectral imaging data shown in figures 4-7 – 4-11, helps explain some of the growth processes taking place. At 400C the multilayer structure is completely intact, with strong ITO and Ge signals consistent with the image. There is no evidence of any quantum dot formation or any materials interactions taking place. When heated to 600C for one hour the film shows definite signs of Ge diffusion both within the

PAGE 100

86 Ge layer as well as some diffusion between adjacent Ge layers. These diffusion processes lead to the formation of small dot-like structures within the Ge layers. Again, the chemical signatures detected indicate ITO and Ge where they would be expected. Note, however, that the Ge signature is showing increased signals of O and In/Sn. This could mean that there is some reaction occurring between the Ge and the ITO. But it could also be due to the fact that the x-ray data from the TEM sample is taken over a thickness. This means that it could be detecting some of the ITO matrix material that is present behind the Ge layer. This could be the case when the Ge dots become isolated or if the cross-sectional sample is not exactly normal to the plane of the detector. When a film is annealed to 650C for 6 minutes, such a short time allows some inter-diffusion of the Ge, but prohibits the intralayer-diffusion between the Ge layers. What results are dot-like structures of Ge beginning to separate out from the once continuous Ge layers. The chemical signature is strongly Ge, with little to no contributions indicated from the ITO. Again, if any materials interactions are to take place, such short time anneals seem to have prevented it. This may not be the case in the film heated to 660C for 30 minutes. This film shows pronounced dot formation within the Ge layers with an average size of 16 nm, but seemingly larger at the bottom of the film. The SI data indicates intralayer and interlayer diffusion of Ge and suggests that the Ge diffuses along the ITO grain boundaries. There is also an increased presence of In/Sn and O in the Ge chemical signature. This combined with the knowledge that much diffusion is present, leads one to think that there is a reaction occurring between the ITO and the Ge. However, the Ge signature is still predominantly Ge, indicating that it may just be surface reactions. Finally, at 700C for 12 minutes, the film contains many isolated Ge quantum dots

PAGE 101

87 surrounded by ITO. This is supported by the corresponding SI data. All of the SI data as a function of temperature concludes that, in fact, Ge quantum dots are able to form within the ITO up to 700C. This is further substantiated by the supporting Raman data of the annealed multilayer films. Figure 4-13 establishes that there is no obvious ITO Raman peak, but illustrates clearly that the Ge peak for a pure Ge film annealed to 600C for one hour occurs around 288 cm -1 in the spectrum. The Ge Raman peak has a tail at the low frequency end indicating that there may still be an amorphous Ge component to the film. Also clear from this figure is that there are no significant peaks in this region for the ITO or sapphire materials. This means that a peak found near 300 cm -1 can be attributed to the presence of Ge. Figure 4-15 shows the effect that confinement of a material has on its Raman peak center and intensity. It has been well established both experimentally and theoretically that as the average size of the quantum dot decreases, the Raman peak inhomogeneously broadens and softens in intensity, while the peak center shifts to higher frequencies. 85 This phenomenon of broadening and softening can be seen in the figure as one goes from a Ge film down to a ~15 nm quantum dot structure. However, the Ge peak position does not shift in the direction consistent with quantum confinement effects. Instead the Raman line experiences a shift in the opposite direction. The reason for this is not that confinement effects do not play a role in these films. Rather, the stresses that develop between the Ge crystallites and the surrounding ITO matrix force the Raman line to shift toward the opposite direction, compensating any effect that may be present due to confinement. In this case it is likely the result of stress arising from a surface free energy

PAGE 102

88 difference between the semiconductor and the oxide matrix, as would occur in lattice mismatch. Both the Ge and the ITO are crystalline at the temperatures tested. While they are both cubic materials, they have vastly different lattice constants, 5.66 for Ge and 10.118 for ITO, so lattice mismatching is likely a contributing factor to the internal stresses in the film. These quantum confinement and stress effects are also evident in the Raman data illustrated in figure 4-16. While the Ge Raman lines are broadened and softened for the composite films annealed to 660C for 30 minutes and 700C for 12 minutes, the peak is still present. The important point here is that there does exist some Ge in the multilayer films, indicating that crystalline Ge quantum dots are present to some degree. This point is further supported by the XRD data. Figure 4-21 shows the XRD scan for the 660C 30 minute film. While the peak, centered near 2-theta = 27.5, is weak and broadened, likely due to confinement effects and the low percentage of Ge in the films, it does exist. This, together with the STEM images, SI data and Raman, confirms the idea that crystalline Ge is still present in the quantum dot structures of the annealed composite films. This is true for all of the films annealed up to temperatures of 700C. For films heated at 700C for longer times or above 700C, the microstructure and chemical make-up of the films begins to change. It is beyond 700C that we leave the workable processing range for quantum dot formation in this materials system. Figures 4-23 – 4-26 illustrate the process of pore formation within the films. It seems that, indeed, there is a materials interaction occurring between the Ge and the ITO, which is facilitated at high temperatures. The first step in the growth process is the desired growth of Ge quantum dot structures. However, over time and at elevated temperatures, the Ge

PAGE 103

89 quantum dots begin to diffuse and ripen, creating large particles of Ge. It is during this process that the Ge and the ITO can react. The formation of bubble structures on the surface implies the presence of a volatile species. X-ray data are not conclusive, but suggest that it is an In 2 Ge 2 O 7 phase. At high enough temperatures, something interesting happens that reduces the Ge content in the films and leaves behind pore structures. It was reported by Sarkisov et al. in 1971 that the In 2 Ge 2 O 7 , upon an increase in temperature, decomposes and a gaseous phase of O 2 and GeO evolves. 96 The presence of a gaseous material escaping the composite multilayer films is demonstrated in Figure 4-27, where bubble formations are evident in the films heated at and above 700C. This fact greatly limits the workable processing range for Ge quantum dot formation in this materials system. For a novel materials system it is especially important to completely characterize the physical nature of the films created. With this knowledge, provided by STEM, SI, Raman and XRD data, insightful conclusions can be made regarding the optical and electrical behaviors observed.

PAGE 104

CHAPTER 5 OPTICAL PROPERTIES At a most fundamental level, the optical properties of a material are defined by the ways in which it interacts with light. As the electromagnetic radiation of light travels through a material it is altered by the presence of charge within that material. These interactions effectively change the wave velocity and intensity of the light in measurable ways. Such interactions are what define many of the things we see around us, including the many colors of the rainbow and the beautiful metallic shine of gold. Measurements of the optical response of a material, coupled with the basic knowledge of the interaction, can provide an important look into its basic structure. Reflection, Transmission and Absorption When light is incident on a material, it can either be reflected from the surface, absorbed by the material, scattered at the surface or from imperfections within the film, or simply transmitted through the material. All of these effects added together equals the total incident light upon the sample, through the equation R + A + S + T = 1, where R is the reflectance, A is the absorbance, S is the scattered component, and T is the transmittance. 88 Each of the reflectance, absorbance and transmittance effects will be considered in turn in the following paragraphs. Consider the case of light passing through a medium of index n o (note: n o = 1 for a material surrounded by air) and interacting with a transparent, i.e. non-absorbing, material with a refractive index of n 1 . By solving the Maxwell equations, with the appropriate boundary conditions, the amplitudes of the reflected and transmitted waves 90

PAGE 105

91 can be determined. The results are known as the Fresnal coefficients for reflection, r, and transmission, t, at the interface of the two media. 97 Assuming that the incident light is normal to the surface, then the reflection and transmission coefficients are given by r = (n o -n 1 )/(n 1 +n o ) and t = 2n o /(n 1 +n o ). The fraction of reflected light from the top surface of the material, R, assuming normal incidence, is given by R = r 2 = ((n 1 -n o )/(n 1 +n o )) 2 (5-1) and the fraction of transmitted, T, light is given by T = t 2 (n 1 /n o ) 2 = 4n 1 n o /(n 1 +n o ) 2 . (5-2) If, however, the film is absorbing, then the complex index of refraction (N = n +ik) must be used, giving the fraction of reflected light, R as R = ((n o -n 1 ) 2 + k 1 2 )/(n o +n 1 ) 2 + k 1 2 )), (5-3) where k is the index of absorption, also known as the extinction coefficient and n is the real part of the refractive index. The values of n and k are linked but their values relative to each other vary depending on the type of material. For example, in highly absorbing materials, such as metals, n is usually much smaller than k. But for dielectric films, which are highly non-absorbing, k is almost negligible with respect to n. 88 It is important to note that the equations explained above are only valid for the case where only one interface is present. The case of a film deposited onto a substrate complicates the equations, and multilayer films demand an even more rigorous approach, but all are attainable. It can be especially important to evaluate such interactions at interfaces when dealing with thin films, where Fabry-Perot interference effects can dominate in the non-absorbing and weakly absorbing regions of the spectra. Several researchers have handled these situations in the literature. 98-102

PAGE 106

92 One important parameter seems to be missing from the above expression for the light transmitted through a specimen. Obviously, the thickness of the film, x, interacting with the incident light will have an effect on the degree to which the light is able to transmit. A thicker sample will allow less light to pass, as a thin one will allow much of the light to pass. Another way in which the transmission coefficient is defined is as the ratio of the transmitted to incident power. The overall transmission is then stated as T = ((1-R) 2 exp(-x))/(1-R 2 exp(-2x)), (5-4) where R is the reflectivity, x is the sample thickness, and is the absorption coefficient. 103 Sometimes it can be assumed that the reflectivity of the film is so small as to be neglected, as is often the case in films with a top layer of ITO. Using this and the relationship between A, the optical absorbance, and T, one can solve for by the expression = 2.303x10 4 A/x, where x is in cm and is in units of cm -1 . 104 Now it is possible to take the experimental absorbance data collected and transform it into a by knowing the thickness of the film tested. In this way, the data is normalize by the thickness and makes comparison of absorption edges possible. For indirect gap semiconductors like Ge and ITO, the absorption coefficient is proportional to the energy, h, squared. So plotting the square root (sqrt) of ( h) versus h provides a straightforward measurement of the optical band gap of the material. 38 Another method for determining the band gap is to simply take the derivative of the A data. This provides a measurement of the inflection point in the absorbance edge, allowing consistent comparisons of E g between films. Measuring the absorption spectrum of semiconductor materials offers a simple way to investigate their band structure. In this way, a direct measurement of the band gap

PAGE 107

93 energy of the particular film can be determined. Comparison of the calculated E g from the A data will allow some insight into the ways that quantum confinement effects play a role in altering the band structures of the quantum dot composite films. Absorbance Data Recall from Figure 1-7 the expected effect that quantum confining the Ge will have on the absorbance spectra. However, because the matrix material is ITO instead of silica, the degree to which the band edges are shifted is not known. In order to understand and interpret any shifts present in the data, the band edges for the pure ITO and Ge materials must first be established. It is of interest to see images of the actual ITO samples in both the as-deposited state and in the annealed state. Figure 5-1(a) and (b) shows how the ITO appears both before heat treatment (a) and after being annealed to 1000C for 1 hour (b). Figure 5-2 illustrates this effect in the absorbance edges in ITO films. From the as-deposited state to an anneal temperature of 600C, the measured band gap of these ITO films ranges from 4.45 eV to 5.18 eV. It is seen from the absorption plot in Figure 5-3, that the as-deposited ITO film has the lowest band gap, while the film annealed to 400C has the highest measured band gap. It is clear that for temperatures up to 600C an increase in E g is seen. Figures 5-4 and 5-5 show the photographic image and measured A data for a pure Ge film heated to 600C for one hour. It possesses the dark gray color expected for a Ge film. Also, the absorbance spectrum has the gradual sloping off of the absorbance tail that is characteristic of an indirect-gap semiconductor such as germanium. However, the presence of the Fabry-Perot fringes makes determination of the direct – indirect transition difficult. By plotting the absorption coefficient (see Figure 5-6), as shown previously for ITO, the indirect band edge energy was determined to be 0.8 eV.

PAGE 108

94 Figure 5-1 Photographic images of pure indium tin oxide films (a) in the as-deposited state and (b) annealed to 1000C for 1 hour. Figure 5-2 Absorbance data for indium tin oxide films in the as-deposited state, annealed to 400C for 1 hour, and annealed to 600C for 1 hour.

PAGE 109

95 Figure 5-3 Plot of sqrt( h) versus energy, h, for indium tin oxide films in the as-deposited state, annealed to 400C 1hour, and annealed 600C 1 hour. Figure 5-4 Photographic image of a 0.5 m pure germanium film annealed to 600C 1 hour.

PAGE 110

96 Figure 5-5 A data for a pure germanium film annealed to 600C for 1 hour. Figure 5-6 Plot of the square root of ( h) versus energy, h, for a pure Ge film annealed to 600C for 1 hour.

PAGE 111

97 With the band edges determined for the pure ITO and Ge films, the focus can now be turned towards the annealed quantum dot films. First, Figure 5-7 illustrates visually what happens to the 15/5% multilayer films as they go from the as-deposited state (a) to an annealed state at 660C for 30 minutes (b). An obvious change has taken place, but for details one must look at the resulting absorbance spectra. Figure 5-8 presents a plot of the absorbance data for several films. This absorbance data corresponds to the same sample set provided in the STEM images of Figure 4-4. The absorbance data for the ITO film heated to 600C for one hour is provided as a reference point. What is obvious is the shifting of the edges that occurs in the quantum dot films. A clear blue-shift in energy takes place due to the presence of the Ge quantum dot structures within the ITO film up to anneal temperatures of 660C. However, at 700C, the absorbance edge red-shifts, likely due to an overall growth of the Ge dots. Figure 5-7 Photographic images of 15/5% composite multilayer films in (a) the as-deposited state and (b) annealed to 660C for 30 minutes.

PAGE 112

98 Figure 5-8 A data for 15/5% composite multilayer films at various temperatures. Discussion Figure 5-1 presents the photographic images of the rf sputtered ITO films. As-deposited ITO has the characteristic yellow hue, which becomes transparent upon heat treatment. The trend of the indium tin oxide band edge shifting as a function of temperature is shown in figure 5-2. The optical and electrical nature of ITO is especially susceptible to the deposition and post-processing conditions. Clearly, temperature does have an effect on the band structure of the ITO films. Further, it is important to discern between the edges of an ITO film and that of a shifted quantum dot film. From these data, a range of ITO band edge energies is obtained. For the temperature range from the as-deposited state to 600C, E g shifts from 4.46 eV to 5.18 eV, respectively. This provides the low-wavelength end limit for the multilayer films. It should be noted that all of the measured band edges for the rf sputtered ITO films are in the range of the published data for ITO. 42 Also, they exhibit very good transparency, especially in the visible wavelength range.

PAGE 113

99 Similarly, the pure Ge film band edge provides the long wavelength end limit. Figure 5-5 shows the absorbance data for a Ge film heated to 600C for one hour. Again, plotting sqrt ( h) versus h (Figure 5-6), the indirect band gap can be obtained. By extrapolating the slope of the line to intersect with the x-axis, the indirect gap appears near 0.8 eV. This compares reasonably well with published data for partially amorphous Ge thin films. 105 With this information a comparison can be made for the progression of a film from the pure Ge state to the multilayer composite films annealed at various temperatures. Figure 5-8 illustrates the evolution that takes place when the composite films transform from continuous layered Ge multilayer films to one containing quantum dot structures. A pronounced shifting in the band edge results from the presence of the germanium quantum dots within the ITO. In going from a 15/5% Ge mulitlayer film annealed to 400C to a similar film annealed to 600C, a blue-shift occurs from 3.52 eV to 4.27 eV (see Appendix B for the method of obtaining the band gap values). For insight into this transformation, recall the STEM images of Figure 4-4. Clearly the change that facilitates this band edge shift is the evolution of continuous Ge layers to the beginnings of Ge dots within the layers. Upon further increase in the anneal temperature, the STEM images show that isolated quantum dots appear in the 660C 30 minute composite film. This is reflected in the absorbance data as the edge is shifted even farther toward shorter wavelengths, with a corresponding E g of 4.5 eV. However, for the film heated to 700C for 12 minutes, the absorbance edge is red-shifted relative to the 660C film to an E g of 3.91 eV. This implies that the quantum confinement effect is not as strong because of a growth in the size of the quantum dots. This seems to agree with

PAGE 114

100 the computed ranges of the quantum dot sizes reported previously. Also, this agrees with what would be expected intuitively. An increase in temperature allows the diffusion of the germanium to occur more quickly, providing the conditions for Oswald ripening of the dots. This ripening of the Ge was apparent from the spectral imaging data. It should be noted that stress effects can also produce a shift in the band gap energy of a material. However, stress can not account for the dramatic band gap shifts observed in these films. We have seen from the data provided in previous chapters that the germanium and indium tin oxide system does have the potential to create quantum dot structures within a narrow post-deposition anneal range. The presence of quantum dots implies that quantum confinement effects dictate the film behavior, and the clear blue-shift in the absorbance data shown provides this proof.

PAGE 115

CHAPTER 6 ELECTRICAL PHOTORESPONSE With the composition and structure of the germanium and indium tin oxide composite films measured and the presence of the quantum confinement effect established, the focus is now turned towards their electrical behavior. Specifically, the photoconductive response of the films containing Ge quantum dots will be compared with films made from pure ITO in order to determine if an enhanced response due to the quantum dots is present. Further, the decay behavior of the photoconductivity curves will be evaluated to offer some basic insights into the nature of the electronic processes involved. Four-Point Probe Technique The most common method for measuring the resistivity of a material is the four-point probe technique. The four-point probe system contains four thin, collinearly placed silicon carbide probe tips, spaced an equal distance, s, apart for the purpose of contacting with the material being tested. Each tip is supported by springs to minimize sample damage during probing. However, the measurement can be very sensitive to the pressure applied to the film, so the probe head was lowered onto the sample a set distance specified by the micrometer reading on the lowering device to maintain the same pressure on all films tested. A current, I, is made to flow between the two outermost probes, while the resulting voltage drop, V, is measured between the inner two probes, as shown in Figure 6-1. For a very thin sheet of material, the calculated sheet resistance, R s , is obtained from the expression R s = CF(V/I), where CF is the correction factor. The 101

PAGE 116

102 correction factor is necessary to take into account any edge effects that may be present due to geometrical considerations. For example, a thin wafer with a thickness of w, length of a, and width of d, the appropriate CF can be determined graphically from Figure 6-2. 26 For semi-infinite samples, the correction factor is 4.54. 106 As the geometry of the sample being tested approaches the four-point probe spacing distance CF becomes smaller, dependent on the ratio of d/s as shown in the figure. From the calculated sheet resistance, the corresponding resistivity (), i.e. conductivity (), can be found using the equation = 1/ = R s *w, most often given in the units of -cm. Figure 6-1 Schematic of the four-point probe experimental set-up.

PAGE 117

103 Figure 6-2 Graphical method for determining the correction factor, CF, for the calculation of sheet resistance from four-point probe data. In order to test for a change in conductivity due to photoinduced processes, the four-point probe head was mounted onto an optics table with the appropriate light source directed to the area being probed. An example of the system described is shown in the photographic image of Figure 6-3. With light incident on the film, the resulting voltage drop measured between the probe tips is monitored as a function of time, before illumination, during illumination and after the light source is removed. In this way, a measure of the response of the film can be established and compared with other films. In this case, the voltage drop measured is the voltage developed between the inner probe tips for a constant current running between the outer probes. Hence, a drop in the voltage measured under illumination conditions is indicative of an increase in the conductivity of the film. The current source and voltage measurement device was a Keithley 236 Source-Measure Unit, which had the capability of simultaneously sourcing the current

PAGE 118

104 and measuring for low voltage samples with a sensitivity of 10 V. Note that for the measurements taken there was a very small noise component of 0.1 mV. Figure 6-3 Photographic image of the four-point probe head contacting the sample. To prove that any change in conductivity measured is a result of the photoexcitation of carriers from the germanium quantum dots two different wavelengths of light were used. It is important that the first wavelength corresponds to an area in the absorption spectrum where little or no absorption is present. In this way, a baseline measurement can be made where no change in conductivity is expected for either the ITO or multilayer films. For this, an 800 nm Ti:sapphire ring laser was used at a power of 100-300 mW. Then, a wavelength source was chosen within the band edges of the films to ensure that absorption was taking place to produce the necessary excess carriers needed to create an enhanced photoresponse. For this an Ar+ laser fitted with UV optics and tuned to 300 nm was used at a power of 100 mW. It is important to note that the four-point probe measurements were very sensitive to the amount of pressure applied to the film. The measurements could vary for the same film depending on the imparted pressure. To avoid a pressure dependence in the photoconductivity data, all tests performed were done using consistent, constant pressure. Although the readings were

PAGE 119

105 pressure dependent, the relative changes in the photoresponses were reproducible for each film tested. Photoconductivity Data Figure 6-4 is a plot of the optical absorbance spectra for the films studied in the photoconductivity measurements. The point to notice is where the band edges lie relative to the two light wavelengths used. The films being compared include two ITO films annealed to 600C and 800C for one hour, and 15/5% multilayer films that each have been annealed to 650C for 6 minutes, 660C for 30 minutes, and 700C for 30 minutes. In all of the films no strong absorption effects are observed at the 800 nm wavelength, while the 300 nm wavelength is within the absorption band edges of all the films tested. Figure 6-4 Absorbance data for various ITO and Ge/ITO multilayer films. Table 6-1 presents the photoconductivity data measured for the films at 800 nm. Listed are the voltage readings both before, V b , and after, V a , illumination, the corresponding change in resistivity and the percent change in conductivity, , due to the

PAGE 120

106 incident light. As expected, there is almost no change 0.1mV in the electronic response to the 800 nm wavelength light in both the ITO films and the multilayer films. Table 6-1 Photoconductivity data for films taken at 800 nm. Film Film thickness (m) Applied current (mA) V b (V) V a (V) (-cm) b (-cm) -1 % ITO 600C 0.5 10 .3141 .3138 6.75E-6 148.0 .100.03 ITO 800C 0.5 5 .2750 .2748 9.00E-6 84.6 .070.04 ML 650C 0.5 1 .1087 .1084 6.45E-5 42.8 .280.09 ML 660C 0.5 1 .2415 .2411 8.60E-5 19.3 .160.04 ML 700C 1.0 10 .4084 .4081 1.29E-5 56.9 .070.02 However, as seen in Figures 6-5 – 6-9 the photoconductive response of the films to the 300 nm light shows some dramatic changes. Figures 6-5 and 6-6 illustrate the photoconductive response of ITO films, heated to 600C and 800C respectively, as a function of time. These were continuously monitored readings that indicate the change in the electronic behavior of the films before, during and after illumination. Clearly there is some photoconductive nature to the ITO films, but the response is relatively small, similar to that seen at 800 nm, with a maximum increased conductivity of 0.1% for ITO heated to 600C and 0.2% for ITO heated to 800C. However, Figures 6-7 – 6-9 tell a dramatically different story for the annealed multilayer composite films containing germanium quantum dots. Figure 6-7 is the conductivity data as a function of time for a 15/5% multilayer film annealed to 650C for 6 minutes. Figure 6-8 is data for a similar film heated to 660C for 30 minutes and Figure 6-9 is data for a 700C 30 minute annealed film. Clearly, there is an enhanced response to the incident light from the film containing the germanium quantum dots. Table 6-2 summarizes the maximum changes in conductivity under exposure for each film presented. Included are the voltage readings

PAGE 121

107 both before, V b , and after, V a , light exposure, the measured conductivity before exposure, b , and the resulting percent increase in conductivity, %, due to exposure. To illustrate the point that the photoresponse of the multilayer films are much enhanced over the ITO films, Figure 6-10 presents a normalized plot of the voltage change for the ITO film heated to 600C and the multilayer film heated to 660C. Figure 6-5 Conductivity versus time for the photoresponse of an ITO film heated to 600C for 1 hour. (On and Off indicate when the film was exposed to the light and when it was removed)

PAGE 122

108 Figure 6-6 Conductivity versus time for the photoresponse of an ITO film heated to 800C for 1 hour. Figure 6-7 Conductivity versus time for the photoresponse of an 15/5% multilayer film heated to 650C for 6 minutes.

PAGE 123

109 Figure 6-8 Conductivity versus time for the photoresponse of an 15/5% multilayer film heated to 660C for 30 minutes. Figure 6-9 Conductivity versus time for the photoresponse of an 15/5% multilayer film heated to 700C for 30 minutes.

PAGE 124

110 Table 6-2 Photoconductivity data for films taken at 300 nm. Film Film thickness (m) Applied current (mA) V b (V) V a (V) b (-cm) -1 % ITO 600C 0.5 100 .1479 .1481 3144.7 -0.1.06 ITO 800C 0.5 100 .2667 .2662 1744.0 0.2.04 ML 650C 0.5 100 .5578 .5518 834.8 1.1.02 ML 660C 0.5 10 .2150 .2067 216.4 3.9.05 ML 700C 1.0 100 .4170 .4135 557.7 0.8.02 Figure 6-10 A normalized plot of the measured voltage response versus time for an ITO film heated to 600C and a multilayer film heated to 660C. To help understand the nature of the photoconductivity the decay curves for each of the multilayer films is investigated. Recall from Chapter 3 the discussion of the example of n-type silicon photoconductive decay. Parallels can be drawn between the published behavior of silicon and the response measured for the germanium/indium tin oxide films. Figures 6-11, 6-12 and 6-13 show a closer view of the decay behavior in the photoconductive response of the 650C, 660C and 700C films, respectively.

PAGE 125

111 Figure 6-11 Conductivity versus time indicating the decay behavior of a multilayer film heated to 650C for 6 minutes. Figure 6-12 Conductivity versus time indicating the decay behavior of a multilayer film heated to 660C for 30 minutes

PAGE 126

112 Figure 6-13 Conductivity versus time indicating the decay behavior of a multilayer film heated to 700C for 30 minutes. Discussion As seen in Figure 6-4 there is little absorption taking place at 800 nm in all of the films compared. This fact is mirrored in the photoresponse of both the ITO films and the multilayer films. In both there was little or no change in the measured voltage under exposure to the light. Remember that the expectation is that for the films containing the germanium quantum dots, any excited carriers created within the dots will be given to the ITO matrix to enhance the electrical response. Clearly at 800 nm the carriers within the germanium will not be excited since it is far below the absorption band edge for these structures. At 300 nm light the situation is very different as there should be sufficient energy to excite carriers within the germanium. For the films containing just indium tin oxide (Figures 6-5 and 6-6), there is some photoconductive response, but the response is minimal, causing a small increase in conductivity of 0.2% for the 800C ITO film, and a decrease in the conductivity by 0.1% for the 600C ITO film. However, the presence of

PAGE 127

113 the germanium within the ITO has an enhanced effect on the photoresponse of the multilayer films. This fact is reflected in the data presented in Figures 6-7 to 6-13. Upon exposure to the UV light source the films exhibit an increased electrical response. It is important to note that this effect is reproducible for each film. Figure 6-7 presents the conductivity versus time response for the multilayer film heated to 650C for 6 minutes. Recall from the previous STEM images and XRD data for this film that it consists of small semi-crystalline germanium quantum dots within the continuous germanium layers. Also of importance is the fact that no porosity is present, which has a positive effect on the resulting conductivity. The measured conductivity of the film at time = 0 is 834.14 (-cm) -1 . After 20 seconds of illumination to the 300 nm light the conductivity increased to 842.91 (-cm) -1 , resulting in an increased conductivity change of = 8.77 (-cm) -1 , corresponding to a 1.1% increase. While fundamentally this is still a small change, it is an unmistakable increase over the response from the ITO films. This increased response was also seen in the 660C multilayer film from figure 6-8. It had a base conductivity at time = 0 of 216.33 (-cm) -1 , which increased to 225.02 (-cm) -1 after exposure to light, increasing the conductivity by = 8.69 (-cm) -1 , corresponding to a 3.9% increase. This is a further increase in the response seen in the 650C film discussed above. Recall from STEM data that this particular film has isolated crystalline germanium quantum dots. However, a small amount of porosity was present near the surface of the film. This likely had the effect of lowering the overall conductivity of the film by creating trapping sites for free carriers that would otherwise contribute to the response. By eliminating porosity, but keeping the isolated dot

PAGE 128

114 structure, an even more pronounced increase in the photoresponse would likely be seen. In addition, it can be seen from the photoconductivity data that the increased conductivity did not reach saturation. Had the photoconductive response been allowed to reach saturation, the measured increase in conductivity due to exposure would have been higher than reported. It is also important to note that the attenuation distance for 300 nm light through the 660C multilayer film is 50 . Since the film is only 0.5 thick, the light is able to penetrate the entire thickness of the film, providing the possibility that all Ge present in the film can photogenerate carriers. The effect of the presence of pores on the electrical response is more obvious in the multilayer film heated to 700C for 30 minutes (Figure 6-9). Here pore formation is significant, while there is still some existence of germanium dots. The resulting photoresponse is, hence, smaller than the 650C and 660C films, with a base conductivity of 557.69 (-cm) -1 , increasing to 562.41(-cm) -1 after exposure, for a conductivity change of = 4.72 (-cm) -1 , corresponding to an increase of 0.8%. While it is a smaller increase than that seen for the other multilayer films, it is still an enhanced effect over the pure ITO film response. For each measurement, the UV source was incident on the film being tested for 20 seconds, after which it was shuttered. The continuing change in the electrical response of the films was monitored during the decay time after exposure. These results are presented in Figures 6-11 to 6-13. In all of the films, heated to 650C, 660C and 700C, the behavior of the decay is similar. At point A the light source is turned on, resulting in the formation of electrons and holes, which increase the conductivity from point A to B until an equilibrium is reached. However, from point B to C the rate of increase in

PAGE 129

115 conductivity begins to decrease in spite of the fact that the light is still incident on the sample. The reasons for this behavior is purely speculative, however, it is likely due to a saturation in the number of available carriers and/or due to hole trapping as was seen in the example of n-silicon discussed in Chapter 3. This phenomenon of the trapping of the holes seen in the multilayer films can be explained by studying the band diagram of the germanium – indium tin oxide system seen in Figure 6-14. When in contact it seems that the electrons will flow freely, while the holes have a barrier that could cause trapping. This helps illustrate that, according to the band diagram for this system, simultaneous quantum confinement and carrier transfer may be possible. The large difference in the valence bands of the ITO and Ge could be one source of the trapping effect seen in the decay response of these films. Note that this hole trapping response was not seen in the ITO film data. It is important to also note that the relative positions of the Ge and ITO bands will vary somewhat as the band gap of the Ge is shifted with size. This shifting of the band gap for Ge begins in the size range of 3-10 times the exciton Bohr diameter, which is 200 in the transverse direction. So transnational confinement begins for Ge dot sizes ranging from 2000 down to 600 for this weak confinement regime. Upon further reduction in the size of the quantum dot, the electron and hole bands begin to shift even more. In the case of the 660C annealed multilayer sample, the valence band shifts down by 3.8 eV, while the conduction band shifts up by only 0.71 eV, resulting in an increased band gap of 4.5 eV measured by the absorbance data. Although the valence band of the confined Ge shifts closer to that of the ITO, the requirement that the Fermi energies be equal at the contact interface will still produce a relatively large difference in valence band energies as seen in the diagram.

PAGE 130

116 Vacuum Level Ec Ev EgGe = .67eV o= .14eV Ev EgITO = 4.5eV Ge = 4eV ITO/ITO = 4.2eV n-type ITO intrinsic Ge Figure 6-14 Band diagram for Ge/ITO after contact. The decay behavior in the multilayer films after the light exposure is terminated indicates that a somewhat complex mix of decay processes is taking place. At first there is a relatively fast decay between points C and D, which can correspond to either the recombination of the excess free electrons and holes or the decay of shallow traps or a combination of both. However, the majority of the decay is a slow exponential, reflecting the presence of many deep level traps. The exact nature of these traps is not possible to decipher from these curves, but the fact that the materials system of Ge and ITO involves reactions and pore formation makes deep level traps likely to occur. These trapping states, both shallow and deep, not only make the response time extremely slow, but also suppresses the photoconductive effect in these films. However, the presence of the germanium quantum dots within the ITO matrix clearly has an enhanced photoresponse over the ITO films alone. This fact indicates that quantum dots may have a positive role to play in the photoconductive and photovoltaic devices of the future.

PAGE 131

CHAPTER 7 CONCLUSIONS Low-dimensional structures, specifically quantum dots, have proved to be interesting candidate materials for many device applications, including quantum memory and lasers. The purpose of the current study was to investigate the potential of quantum dot thin film structures for use in photoconductive and photovoltaic applications, specifically to show that simultaneous quantum confinement and carrier transfer is possible. In order for this to be established, it must be seen that the presence of quantum dots within the film leads to an enhanced photoresponse over similar films lacking quantum dots. In this study rf magnetron sputtering was used to fabricate multilayered films consisting of germanium, Ge, and indium tin oxide, ITO. The films studied were made up of alternating layers of the materials with an overall content of 5% germanium. Each alternating layer of Ge had a thickness of 15 , and each ITO layer had a thickness of 280 . Under post-deposition anneals in the temperature range of 600-1000C, the continuous germanium layers are transformed into isolated quantum dots surrounded by the transparent, conducting ITO matrix. It is postulated that the germanium quantum dots, when exposed to light of a suitable wavelength, will create excess carriers. In the presence of an applied electric field, the extra carriers would be transported through the conducting matrix to the appropriate contact, producing an enhanced electrical response. By correlating the microstructural, optical and electrical data from the films, an understanding of the behavior of the germanium quantum, dot multilayer films can be obtained. It was found that, in fact, germanium quantum dots did form within the 117

PAGE 132

118 multilayer structures for anneal temperatures in the range of 600-700C for times of 6 minutes to 1 hour. Through micrographs taken with scanning transmission electron microscopy (STEM) and the corresponding x-ray automated spectral imaging data, it was found that small germanium dots 9 nm in size nucleated from the continuous layer around 600C. An increase in temperature and time caused the germanium to diffuse within the layer and between layers along the ITO grain boundaries to form larger, isolated dots. Within the temperature range of 600-700C, the achievable range of dot sizes was 9 – 25 nm. However, the germanium–indium tin oxide materials system proved reactive at temperatures of 700-1000C. In this processing range it was clear that the germanium began to form large particles that ultimately reacted with the surrounding ITO matrix. Although the x-ray data was inconclusive, it seems likely that the reaction forms In 2 Ge 2 O 7 (IGO). IGO is volatile at temperatures near 700C, forming a gaseous phase of O 2 and GeO through a decomposition process of In 2 Ge 2 O 7 In 2 O 3 + O 2 + 2GeO, consequently leaving behind large pore structures. To further corroborate the presence of germanium quantum dots within the ITO, Raman and x-ray diffraction, XRD, studies were done. For ITO films there was no obvious Raman peak. The only peaks present in the pure ITO films tested were contributed by the sapphire substrate. Pure germanium films were grown to examine the corresponding Raman peak, which is centered around 280 and 300 cm-1 for the amorphous and crystalline components, respectively. The theoretical prediction for a shift in the Raman line due to quantum confinement effects states that the center of the peak moves toward smaller wavenumbers. However, it was observed that for the multilayer films, were the Ge Raman peak was present, that the line shifted in the

PAGE 133

119 opposite direction. Stress effects can cause such a shifting in the Raman peak, and it was concluded that this was the likely cause behind the observed shift in the films, which overshadowed any effects due to confinement of the germanium. Another trend seen in Raman data due to quantum confinement effects is the softening and broadening of the Raman line with increasing confinement. This effect was observed in the Ge/ITO multilayer films. Similar softening and broadening effects due to quantum confinement also manifest themselves in x-ray diffraction data. This was seen in the multilayer films, where the germanium peak corresponding to (111) plane in a cubic germanium crystal, were broader and less intense in the films containing germanium quantum dots. However, it must be considered that the low intensity of the peaks could be due to the small volume fraction of germanium in the films. While the STEM micrographs, Raman and XRD data help to support the idea that quantum confinement is playing a role in these multilayer films, the strongest and most obvious evidence comes from the shifting of the band edges observable in UV-Vis absorbance data. When a material is confined to a small dimension, approaching the material’s exciton Bohr diameter, then interesting changes occur in the electronic transitions. As the degree of confinement increases, i.e. as the diameter of the material becomes smaller, the band edge is forced to higher energies. This phenomenon was established for the multilayer films containing germanium quantum dots. The absorbance data for films annealed 600-700C shows a dramatic blue-shift in band gap energy with decreasing dot size, and a corresponding red-shift in energy when the dots grew in size. The shift in the measured absorption edge was consistent with the acquired dot sizes for the films. The most pronounced shift occurred in the film annealed to 660C for 30

PAGE 134

120 minutes, where isolated germanium quantum dots grew to an average size of 16 nm in diameter. Here, the band gap corresponded to 3.27 eV; an energy significantly higher than the bulk germanium band gap of 0.67 eV. Such a dramatic shifting in the absorption edge, coupled with the data collected from STEM, Raman and XRD, leave little doubt that quantum confinement effects, due to the incorporation of germanium quantum dots, are present in these films. With germanium dots successfully fabricated into the indium tin oxide matrix the corresponding electrical photoresponse of the films was measured. From a device perspective, by varying the size of the quantum dots many different wavelengths of light can be absorbed and converted into electron–hole pairs. This allows high conversion efficiency of the solar radiation necessary for solar cell technology. The premise for the current research was to prove that a quantum confined material surrounded by a transparent conducting oxide matrix will, under exposure to light with an energy above its band gap, create these excess carriers and, in the presence of an electric field, transport them to the matrix. This would result in an increased electrical response. To prove that the ITO films containing the Ge dots were, in fact, behaving in this fashion, their photoconductive behavior was compared to pure ITO films. It was seen that at 800 nm, a wavelength below the band gaps of all the films, no significant response in the measured conductivity was present for either the pure or quantum dot films. However, when the response was measured at a wavelength above the band edges of the films, 300 nm, an unmistakable enhancement in the photoresponse was observed for the films containing the germanium dots. Pure ITO films tested showed a response very similar to their response in the below-band gap test, producing an increase in the conductivity under

PAGE 135

121 illumination of roughly 0.1%. However, in the quantum dot containing films the enhanced photoconductive response was as high as 3.9%. While the increases seen in the quantum dot films seem small, it is still a clear indication of the potential that such composite materials can have for creating novel photoconductive and photovoltaic devices. The photoconductive decay curves measured for these films indicate that significant deep level traps are present. By better understanding the nature of these trapping centers and by optimizing the fabrication technique to limit the materials interactions, composite quantum structures such as these could help advance the current state of solar cell technology. The most important contribution of this work is that it experimentally demonstrates that quantum dot structures can simultaneously exhibit a shift in band gap energy resulting from quantum confinement and an increase in photoconductivity resulting from the injection into the conductive matrix of photoexcited carriers created in the quantum dot. This combination of effects promises a better means of harnessing solar energy for use in devices of the future.

PAGE 136

APPENDIX A RAMAN PEAK FITS All Raman peaks were modeled using the fitting routines in Origin software. The peaks were fit to Gaussians. 122

PAGE 137

123

PAGE 138

124

PAGE 139

125

PAGE 140

126

PAGE 141

APPENDIX B BAND GAP DETERMINATION The band gap values provided were determined by taking the first derivative of the absorbance spectrum for each film. This was done using Origin, where the spike in the results were given a wavelength value and then converted into an energy value. 127

PAGE 142

128

PAGE 143

129

PAGE 144

LIST OF REFERENCES 1. C. Weisbuch, H. Benisty and R. Houdre, Journal of Luminescence 85 (2000) p. 271. 2. El-Hang Lee and Kyoungwan Park, Materials Science and Engineering B74 (2000) p. 1. 3. John H. Reina, Luis Ouiroga and Neil F. Johnson, Physical Review A 62 (2000) p. 102305. 4. T. Lundstrom, W. Schoenfeld, H. Lee and P.M. Petroff, Science Magazine 286 (1999) p. 2312. 5. H.C. Liu, Physica E 8 (2000) p. 170. 6. D. Bimberg, M. Grundman, F. Heinrichsdorff, N.N. Ledentsov, V.M. Ustinov, A.E. Zhukov, A. R. Kovsh, M.V. Maximov, Y.M. Shernyakov, B.V. Volovik, A.F. Tsatsul’mikov, P.S. Kop’ev and Zh.I. Alferov, Thin Solid Films 367 (2000) p. 235. 7. M. Grundmann, F. Heinricksdorff, C. Ribbat, M.-H. Mao and D. Bimberg, Applied Physics B 69 (1999) p. 413. 8. Mitsuru Sugawara, Journal of Crystal Growth 189/190 (1998) p. 585. 9. J. H. Marsh, D. Bhattacharyya, A. Saber Helmy and E.A. Avrutin, Physica E 8 (2000) p. 154. 10. Marius Grundmann, Physica E 5 (2000) p. 167. 11. M. K. Zundel, K. Eberl, N.Y. Jin-Phillipp, F. Phillipp, T. Riedl, E. Fehrenbacher and A. Hangleiter, Journal of Crystal Growth 201/202 (1999) p. 1121. 12. Simon Fafard, Physica E 8 (2000) p. 107. 13. K. Hinzer, C.Ni. Allen, J. Lapointe, D. Picard, Z.R. Wasilewski and S. Fafard, Journal of Vacuum Science and Technology A 18(2) (2000) p. 578. 14. Zhihong Jin, Huan-Jun Zhou, Zhang-Li Jin, Robert F. Savinell and Chung-Chiun Liu, Sensors and Actuators B 52 (1998) p. 188. 130

PAGE 145

131 15. G. Blaser, Th. Ruhl, C. Diehl, M. Ulrich and D. Kohl, Physica A 266 (1999) p. 218. 16. E. S. Rodrigues, E.V. Anda, P. Orellana and H.N. Nazareno, Solid State Communications 111 (1999) p. 217. 17. L. Yue and Y. He, Journal Materials Science Letters 15 (1996) p. 263. 18. Eiichi Hanamura, Physical Review B 37(3) (1988) p. 37. 19. Joseph H. Simmons, Barrett G. Potter, Jr. and O. Romulo Ochoa, CRC Handbook of Laser Science and Technology (CRC Press, Inc., Boca Raton, 1995) p. 250. 20. S. Vijayalakshmi, F. Shen and H. Grebel, Applied Physics Letters 71(23) (1997) p. 3332. 21. D.M. Chapin, C.S. Fuller and G.L. Pearson, Journal of Applied Physics 25 (1954) p. 676. 22. Martin A. Green, MRS Bulletin 17, No.10 (1993) p. 26. 23. Allen M. Barnett, Robert B. Hall and James A. Rand, MRS Bulletin 17, No.10 (1993) p. 33. 24. Y. Hamakawa, W.Ma and H. Okamoto, MRS Bulletin 17, No.10 (1993) p. 38. 25. J.R. Hook and H.E. Hall, Solid State Physics (John Wiley & Sons 1991). 26. S.M. Sze, Physics of Semiconductor Devices (John Wiley & Sons 1981). 27. C.H. Henry, Journal of Applied Physics 51(8) (1980) p. 4494. 28. Monique Pauthe, Estela Bernstein, Jean Dumas, Lucien Saviot, Annie Pradel and Michel Ribes, J. Mater. Chem. 9 (1999) p. 187. 29. Masaki Yamamoto, Ryo Hayashi, Keiji Tsunetomo, Kenji Kohno and Yukio Osaka, Jpn. J. Appl. Phys. 30(1) (1991) p. 136. 30. M. Zacharias, J. Blasing, J. Christen, P. Veit, B. Dietrich and D. Bimberg, Superlattices and Microstructures 18(2) (1995) p. 139. 31. Jinzhong Cui, Guanghua Chen and Xinwang Zhang, Phys. Stat. Sol.(b) 189 (1995) p. K9. 32. T. Yamaguchi, T. Ohmi and M. Aoyama, SPIE 2364 (1994) p. 351.

PAGE 146

132 33. Tracie J. Bukowski, T.M. Neidt, Romulo Ochoa and Joseph H. Simmons, Journal of Non-Crystalline Solids 274 (2000) p. 87. 34. N.V. Joshi, Photoconductivity: Art, Science, and Technology (Marcel Dekker, New York 1990). 35. S. Nishikawa, Thin Solid Films 135 (1986) p. 219. 36. Alex Zunger, MRS Bulletin Feb (1998) p. 35. 37. M. Boero, J.M. Rorison, G. Duggan and J.C. Inkson, Surface Science 377-379 (1997) p. 371. 38. Joseph H. Simmons and Kelly S. Potter, Optical Materials (Academic Press, San Diego 2000). 39. Jacques I. Pankove, Optical Processes in Semiconductors (Prentice-Hall, Inc. 1971). 40. J.Mort and D. M. Pai, Photoconductivity and Related Phenomena (Elsevier Scientific Publishing Co., New York 1976). 41. Richard H. Bube, Photoconductivity of Solids (John Wiley & Sons, Inc., New York 1971). 42. K.C. Chopra, S. Major and D.K. Pandya, Thin Solid Films 102 (1983) p. 1. 43. Roy G. Gordon, MRS Bulletin 25(8) (2000) p. 52. 44. David S. Ginley and Clark Bright, MRS Bulletin 25(8) (2000) p. 15. 45. S. Nishikawa, Thin Solid Films 135 (1986) p. 219. 46. D.V. Morgan, Thin Solid Films 135 (1995) p. 283. 47. H. Kstlin, R. Jost and W. Lems, Physical State of Solids (a) 29 (1975) p. 87. 48. D.V. Morgan, Thin Solid Films 312 (1998) p. 268. 49. K. B. Sundaram and Jila Blanchard, Engineering the New Century Conference Proceedings IEEE SOUTHEASTCON (1997) p. 230. 50. H. Morikawa and M. Fujita, Thin Solid Films 359 (2000) p. 61.

PAGE 147

133 51. F. Zhu, C.H. A. Huan, Keran Zhang and A.T.S. Wee, Thin Solid Films 359 (2000) p. 244. 52. Wen-Fa Wu and Bi-Shiou Chiou, Thin Solid Films 298 (1997) p. 221. 53. M. Bender, W. Seelig, C. Daube, H. Frankenberger, B. Ocker and J. Stollenwerk, Thin Solid Films 326 (1998) p. 72. 54. E. Terzini, G. Nobile, S. Loreti, C. Minarini, T. Polichetti and P. Thilakan, Japenese Journal of Applied Physics 38 (1999) p. 3448. 55. Wen-Fa Wu, Bi-Shiou Chiou and Shu-Ta Hsieh, Semiconductor Science and Technology 9 (1994) p. 1242. 56. Li-Jian Meng and M. P. dos Santos, Journal of Vacuum Science Technology A 18(4) (2000) p. 1668. 57. E. Terzini, G. Nobile, T. Polichetti and P. Thilakan, IEEE 26 th PVSC Conference (1997) p. 667. 58. Wen-Fa Wu and Bi-Shiou Chiou, Semiconductor Science and Technology 11 (1996) p. 196. 59. Yuzo Shigesato, Itaru Yasui and David C. Paine, Journal of Materials March (1995) p. 47. 60. Mamoru Mizuhashi, Thin Solid Films 76 (1981) p. 97. 61. Yutaka Ohhata, Fujitoshi Shinoki and Sadafumi Yoshida, Thin Solid Films 59 (1979) p. 255. 62. F.T.J. Smith and S.L. Lyu, Journal of the Electrochemical Society: Solid-State Science and Technology (1981) p. 2388. 63. Soumyendu Guha, Mark Will and Lloyd Chase, Nuclear Instruments and Methods in Physics Research B 147 (1999) p. 367. 64. Minoru Fujii, Shinji Hayashi and Keiichi Yamamoto, Japanese Journal of Applied Physics 30, No.4 (1991) p. 687. 65. James R. Heath, J.J. Shiang and A.P. Alivisatos, Journal of Chemical Physics 101(2) (1994) p. 1607. 66. M. Zacharias, J. Blasing, M. Lohmann and J. Christen, Thin Solid Films 278 (1996) p. 32.

PAGE 148

134 67. David R. Lide, ed., CRC Handbook of Chemistry and Physics (Academic Press, Inc., Boca Raton 1999). 68. B.G. Potter, Dissertation entitled Electronic Band Structure and Optical Characteristics of Quantum-Size CdTe Crystals in Glass Films, University of Florida (1991). 69. G.D. Sanders, C.J. Stanton and Y.C. Chang, Physical Review B 48, No.15 (1993) p. 11067. 70. W. E. Beadle ed., Quick Reference Manual For Silicon Integrated Circuit Technology (John Wiley & Sons, Inc., New York 1974). 71. Edward D. Palik, ed., CRC Handbook of Optical Constants of Solids (Academic Press, Inc., Orlando 1985). 72. Melles Griot Company, optical products and support April 22, 2002. Available. < www.mellesgriot.com > 73. M. Zacharias, R. Weigand, B. Dietrich, F. Stolze, J. Blasing, P. Veit, T. Drusedau and J. Christen, Journal of Applied Physics 81(5) (1997) p. 2384. 74. John F. O’Hanlon, A User’s Guide to Vacuum Technology (John Wiley & Sons, Inc., New York 1989). 75. Philip Moriarty, Rep. Prog. Phys. 64 (2001) p. 297. 76. B.G. Potter and J. H. Simmons, Journal of Applied Physics 68(3) (1990) p. 1218. 77. David B. Williams and C. Barry Carter, Transmission Electron Microscopy (Plenum Press, New York 1996). 78. M. Gotic, M. Ivanda, S. Popovic, S. Music, A. Sekulic, A. Turkovic and K. Furic, Journal of Raman Spectroscopy, 28 (1997) p. 555. 79. David C. Paine, Christine Caragianis and Yuzo Shigesato, Applied Physics Letters 60(23) (1992) p. 2886. 80. Paul G. Kotula, Michael R. Keenan and Joseph R. Michael, Journal of Microscopy and Microanalysis in press (2002). 81. C. R. Brundle, C. A. Evans Jr., S. Wilson, Encyclopedia of Materials Characterization (Butterworth-Heinemann, Boston 1992).

PAGE 149

135 82. D.J. Gardiner, P.R. Graves, H.J. Bowley, Practical Raman Spectroscopy (Springer-Verlag, New York 1987). 83. J. Twardowski, and P. Anzenbacher, Raman and IR Spectroscopy in Biology and Biochemistry (Ellis Horwood Limited, New York 1994). 84. C. E. Bottani, C. Martini, P. Milani, M. Manfredini, A. Stella, P. Tognini, P. Cheyssac and R. Kofman, Applied Physics Letters 69(16) (1996) p. 2409. 85. A Stella, P. Tognini, C.E. Bottani, P. Milani, P. Cheyssac and R. Kofman, Thin Solid Films 318 (1998) p. 100. 86. M. Fujii, S. Hayashi and K. Yamamoto, Applied Physics Letters 57(25) (1990) p. 2692. 87. Atkinson and S. C. Jain, Journal of Raman Spectroscopy 30 (1999) p. 885. 88. M. Ohring, The Materials Science of Thin Films (Academic Press, Boston 1992). 89. A. Hakam, L. Ouboumalk, M. Banouq, Y. Saidi, M. Taibi and J. Aride, Ann. Chim. Sci. Mat. 25 (2000) p. 387. 90. E. Baba Ali, G. Sorensen, J. Ouerfelli, J.C. Bernede and H.E. Maliki, Applied Surface Science 152 (1999) p. 1. 91. Pfeifer, R. Wartchow and M. Binnewies, Z. Kristallogr. NCS 216 (2001) p. 191. 92. M. Mizuno and T. Miyamoto, Japanese Journal of Applied Physics 39 (2000) p. 1849. 93. G. Campet, S.D. Han, S.J. Wen, M.C.R. Shastry, B. Chaminade, E. Marquestaut and J. Portier, Materials Science and Engineering B22 (1994) p. 274. 94. G. Campet, S.D. Han, S.J. Wen, J.P. Manaud, J. Portier, Y. Xu and J. Salardenne, Materials Sci. and Engineering B19 (1993) p. 285. 95. T. Maruyama and T. Tago, Applied Physics Letters 64(11) (1994) p. 1395. 96. Sarkisov, Neorg. Mater. 7(2) (1971) p. 341. 97. Eugene Hecht, Optics (Addison-Wesley Publishing Co., Reading 1974). 98. R. Swanepoel, Journal of Physics E: Sci. Instrum. 16 (1983) p. 1214. 99. R. T. Phillips, Journal of Physics D: Appl. Phys. 16 (1983) p. 489.

PAGE 150

136 100. Kelly Simmons-Potter and J. H. Simmons, Journal of the Optical Society of America B 13(2) (1996) p. 268. 101. W.Nazarewicz, P. Rolland, E. da Silva and M. Balkanski, Applied Optics 1(3) (1962) p. 369. 102. R. Bowling Barnes and M. Czerny, Physical Review 38 (1931) p. 338. 103. F. Abeles, ed., Optical Properties of Solids (North Holland Publishing Co., Amsterdam 1972). 104. J. R. Meyer-Arendt, Introduction to Classical and Modern Optics (Prentice Hall, Englewood Cliffs 1989). 105. J. Tauc, R. Grigorovici and A. Vancu, Phys. Status Solidi 15 (1966) p. 627. 106. Bridge Technology, a company specializing in resistivity measurements. April 22, 2002. Available.

PAGE 151

BIOGRAPHICAL SKETCH Tracie J. Bukowski received her Bachelor of Science in materials science and engineering from the University of Arizona in 1996. While there, she worked under the direction of Professor B.J.J. Zelinski and Professor D. Uhlmann doing sol-gel processing of ceramics. After a semester studying at the University of Warwick in Coventry, England, she returned to Tucson and finished her degree while employed at Donnelly, a local research company, conducting research involving piezoelectric thin films and colloidal chemistry. Upon meeting Dr. Joseph Simmons, she agreed to work with him on his on-going quantum dot project at the University of Florida. She moved to Gainesville, Florida, and began her graduate career with him in 1997. She spent the last two years of her graduate research working with Kelly Simmons-Potter at Sandia National Laboratories in Albuquerque, New Mexico, while completing her degree. 137