You are on page 1of 226

Introduction to Nanoscience and Nanotechnology

:
A Workbook

M. Kuno
August 22, 2005

Copyright © 2005 By Masaru Kenneth Kuno, University of Notre Dame, All Rights Reserved

From: http://nd.edu/~mkuno/Class_downloads/Chem647_nano_text.pdf

2

Contents
1 Introduction

1

2 Structure

11

3 Length scales

37

4 Excitons

45

5 Quantum mechanics review

57

6 Confinement

75

7 Nondegenerate perturbation theory

107

8 Density of states

121

9 More density of states

129

10 Even more density of states

139

11 Joint density of states

151

12 Absorption

165

13 Interband transitions

173

14 Emission

195

15 Bands

213

16 K · P Approximation
(Pronounced k dot p)

235
i

ii

CONTENTS

17 Tunneling

261

18 The WKB approximation

279

19 Synthesis

307

20 Tools

325

21 Applications

333

Acknowledgments

355

. . . . .8 2.2 6. . . . . . . . . . . . . . . . . . . . . . . Particle in a finite box: solutions 76 78 79 80 81 iii . . . .4 2. . . . . . . .12 2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5 Particle in an infinite box . . . . Hexagonal unit cell . . . . . . . . . . . . .3 1. .15 14 3D Bravais lattices . . . . . . . . . . . . . . . . . . . . . . . . . . . . BCC unit cell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .9 2. . Atom sharing . . . . .1 2.3 2. . . . . . . . .5 2. . . . .1 6. . . . . . . . . . . .2 1. . . . . . . . . . . . . . . 46 46 6. . . . . . . . . . . . . . . . . . . . . . .1 4. . . . . . . . .2 Exciton types . . . . . . . . . . . . . Simple cubic . . . . . Half a harmonic oscillator . . . . . . . . . . . . . . . . . . . . . .5 CdSe quantum dot . . . . . . . . . . . Miller index examples . . . . . . . . . .14 2. . . . . Various 1D potentials . . . . . . . . . . Quantum confinement . . . . . Particle in a finite box . . 12 14 15 16 16 17 18 19 20 21 22 23 24 28 30 4. . . . . . .4 1. CdSe . . . . . . . . . . Wurtzite unit cell . . Size dependent absorption Artificial solid . . . . . .13 2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . FCC unit cell . . CsCl structure unit cell . .10 2.7 2. . . . Zincblende or ZnS structure unit cell NaCl structure unit cell . . . . . . . . . . . . . . . Dimensionality . . . . and emission of . . . . . . . . . . . . . . . . . . . . .11 2. . .3 6.6 2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2 2. . . . . . . . . . . . . . . . . . . Atoms per unit cell . . .List of Figures 1. . . . . . . . . . . . . . . Diamond structure unit cell . . .1 1. . . . . . . . .4 6. . . . . . . . . . . . . . . . . . . FCC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Exciton absorption . . . . . . . . . . . 4 6 7 8 9 2. . . . . . . . . . . . . . . . . . . . . . . . . More Miller index examples . . . . . . . .

. . . . . . . . . .5 15. . . . . . . . . . . . . Summary. . . . . . . . . . . . . 140 142 144 147 11. . . joint density of states . . . . . . . . . .3 15. . . . . . . . . . . . Radiative decay of excited state . . . . . . . .4 3D 2D 1D 0D 9. . . . . . . . . . . . . . . . . . . . . . . 1D joint density of states . . . . . . . .3 10. . . . . . . . . . . . . 8. . . . . .6 6. . . . . . . . . . . . . . . . 153 155 158 161 162 14. . . . . .1 15. . . .3 14. . . . . . . . . . . . . . . . . .3 8. . . .1 11. . . . . Derived emission spectrum: Einstein A and B Pulsed experiment and lifetime . . . . . .4 2D 2D 1D 1D . . . . . . . . .4 11. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2 11. . . . . . . . 196 209 209 211 211 Kronig-Penney rectangular potential . . . . . . . . . . 134 3D Fermi level . . 261 17. coefficients . . . .4 15. . . . . . . . . . . . . . . . . . . . . . . .4 14. . . .5 Einstein A and B coefficients . . . . . . . .2 15.2 14. . . . . .1 8. . . Kronig-Penney delta function potential .7 density density density density density density density density of of of of of of of of states states states states CB VB CB VB . .3 11. . . . . . .2 3D density of CB and VB states . . .8 From metals to insulators . . delta functions continued: Mathcad solutions . General Kronig Penney model continued: Mathcad solutions Kronig Penney model revisited: Mathcad solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2 10. . . . . . . . . . . . . . .8 Particle in a finite well: Mathcad solutions . . . . . . . . . . . states states states states . . . . . . . . . . . . . . . . . . . . . . . . . Kronig Penney model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213 224 227 228 229 230 15.iv LIST OF FIGURES 6. . . . . . . .6 15.1 14. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262 . .5 Vertical transitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2 8. . . . . . . . . . . . . . 85 87 92 123 125 126 127 231 232 17. .1 Potential step . . . . .2 Potential step (ε > V ) . . . Multiple pathway decay of excited state . . . . .1 10. . . . Particle in a sphere . . . . . . . . . . . Kronig Penney model. . General Kronig Penney model: Mathcad solutions . . . . . Particle in an infinite circle . . . . . . . . . 2D joint density of states . . . . . . 3D joint density of states . . . . . . . . . . delta functions: Mathcad solutions . .1 9.7 6. . . . . . . . . . . . . . . . . 15. . . 137 10. . . .

. Potential barrier . . step . . . . . . . . .3 17. . . . . . . . . . . .4 21. . . . . . . . . . . . . . . . Linear barrier . . . . . . . . . . . . .1 18. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 326 327 328 329 330 331 21. . . . . . . . . . . . . . . . . barrier . . . . . . Semiconductor junction . . . . . . . . . . . . . . . . . .5 17. . . . . . . . . . .9 Nanowire device . . . . . . . . . . .5 18. . . . . . . . . . . . . . Scanning tunneling microscopy . . . . . . . . . . . 308 309 311 317 319 20. Nanowire sensor . . . . . . . . . . . . Potential barrier (ε < V ) . . . . . .6 17. . . .LIST OF FIGURES v 17. . . . .7 Potential step (ε < V ) . . . . . . . . . . . . . . . . . . . . . . Orthodox model of single electron tunneling . . . . . .7 21. . . . . . . . . . . . . . . . . . . . .6 Transmission electron microscopy Secondary electron microscopy . . . .6 18. . . . . .1 21. . . . . . . . . . . . . . . Coulomb Staircase . . LaMer model: size distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Quantum dot/dye photobleaching . . . . . . . . . . . . .4 19. . . . . . . . . . . . . . . . . . . . . Solar spectrum and QD absorption/emission spectra Quantum dot LED schematic . Potential barrier (ε > V ) . . . .3 18. . . Microcontact printing . . . . .3 19. . . . . . . . . . . . . .3 21. . . . . . .2 20. . . . . . . . . LaMer model . .2 19. . . . . . . . . .8 21. . . . . . . . . Density of states for lasing . . . . . . .4 17. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281 286 290 298 300 301 303 304 19.5 Cartoon of a MBE apparatus . . . . . . . .3 20.8 Arbitrary potential Arbitrary potential Arbitrary potential Field emission . . . . . . . . . . . . . . . . Atomic force microscopy . . . . . . . . . . . Parabolic barrier . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .7 18. . . . . . . . . drop . . . . . . . .1 20. . .6 21. . . . . .4 20.5 20. . . . Shottky barrier . 334 335 337 338 340 342 344 346 350 . . . . . . . .2 18. . . . . . . . . .2 21. Parabolic barrier . . . . . Dip pen nanolithography . . . . .5 21. . . . Quantum dot/dye absorption/emission spectra . . . . . . . . . . . 263 269 269 275 278 18.1 19. .4 18. . . . . . . . . Cartoon of a MOCVD apparatus Colloidal synthesis apparatus .

vi LIST OF FIGURES .

. . . . Group III-V semiconductors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Group II-VI semiconductors .4 2. . . . . . . . . . . .3 2. . . . . . . . . . . . . . . . .5 Common metals . Group IV-VI semiconductors vii . .List of Tables 2. Group IV semiconductors . . . . . . . . . . . . .1 2. . 25 25 26 26 27 . . . .2 2. .

viii LIST OF TABLES .

Preface This set of lecture notes about nanoscience and nanotechnology was initially written over the spring and summer of 2003. the goal is to provide the student with a foundation by which they can critically analyze. I agreed to teach an introductory class on nanoscience and nanotechnology for incoming graduate students and upper level undergraduates at the University of Notre Dame. these lecture notes can be converted into an introductory text so that others may benefit as well. Masaru Kenneth Kuno Notre Dame. contribute to the emerging nano field. 2003 ix . The motivation for this was to describe to the student the physics behind each concept or assumption commonly encountered in the nano literature rather than just providing a qualitative overview of developements in the field. IN August 24. let alone any textbook. In this manner. After my initial appointment as an assistant professor in chemistry. I undertook it upon myself to compile a series of lecture notes that would explain to the student some of the underlying concepts behind “nano”. it quickly became apparent to me that there were few resources available for teaching such a class. So while waiting for equipment to arrive. However after accepting this task. It is also my hope that one day. I have also tried to illustrate and motivate these concepts with instances in the current literature where such concepts are applied or have been assumed. and possibly in the future.

x LIST OF TABLES .

Revision History • Version 1.0 complete 8/03 • Corrected for typos and such as well as added the Kane chapter 6/04 • Added Absorption section 7/05 • Added Exciton section 7/05 • Added Quantum mechanics review section 7/05 • Added Non-degenerate perturbation theory section 7/05 • Added Time dependent perturbation theory section 7/05 xi .

xii LIST OF TABLES .

a current GE commercial for washers and dryers features the storyline of: geeky nanotechnologist bumps into a supermodel at the laundromat resulting in love at first sight. the Willem Dafoe character (the Green Goblin) is a famous (and wildly wealthy) nanotechnologist whose papers the Tobey McGuire character (Spiderman) has read and followed (see the scene outside of Columbia university). in the recent blockbuster. This scene involves a retinal eye transplant aided by so called “nano reconstructors”. without providing a definite answer to this question. A scene in the DC metro shows him reading a newspaper with the headline “nanotechnology breakthrough”. Nano has even entered popular culture. It has attracted the attention of researchers from all walks of life. Likewise. nano is a popular (emerging) area of science and technology today. A prime example of this is the new National Nanotechnology Initiative (NNI) created by former President Bill Clinton. For example. implicitly. Further impetus for this movement comes from the temendous increase in public and private funding for nano over the last ten years. from physics to chemistry to biology and engineering. private sector contributions have jumped dramatically as evidenced by the plethora of small startup firms lining the tech corridors of the east and west. in the movie “Minority Report” Tom Cruise’s character undergoes eye surgery to avoid biometric fingerprinting. It’s used as a buzzword in contemporary books. told to believe that their mix of 1 . Spiderman. movies and television commercials. In television.Chapter 1 Introduction Preliminaries What is “nano”? Well. The NNI increases funding in national nanoscience and nanotechnology research by hundreds of millions of dollars yearly. We’re therefore. In addition.

questions such as how the optical and electrical properties of a given material evolve from those of individual atoms or molecules to those of the parent bulk. preventing heart attacks. the word nano describes physical lengthscales that are on the order of a billionth of a meter long. any such developments will ultimately rely on the scientific research of today. The Matrix and Prey. mend bones when broken. INTRODUCTION brains and beauty embody GE’s new washer/dryer line. Concepts being discussed include developing tiny nano robots that will “live” inside us and repair our blood vessels when damaged. nanotechnology will provide a solution to all of mankind’s problems whether it be hunger in developing countries or pollution in developed ones. Now. chemistry. which is. concurrent with his visions of a utopian future is a darker side. This mesoscopic size regime has previously been unexplored and beckons the researcher with images of a scientific wild wild west with opportunites abound for those willing to pack their wagons and head into the scientific and technological hinterland. laying down the foundation for tomorrow’s nanoscience and nanotechnology. the New York Times bestseller “Prey” by Michael Crighton features nanotechnology run amok with spawns of tiny nano bots escaping from the laboratory and hunting down people for food. make us smarter and even give us immortality. In today’s scientific realm. macroscale. Nanoscale materials therefore lie in a physical size regime between bulk. Such beings could then sow the seeds to mankind’s own destruction in the spirit of recent movies and books such as The Terminator. However. (Sounds like the “Andromeda Strain” except recast with nano as opposed to an alien virus. yet unanswered. on a daily basis. devoted to exploring his ideas. Drexler therefore envisions a vast industrial revolution of unprecendented size and scale. cure us when we are sick. biology and engineering asks basic.). At the same time. whether such predictions and visions of the future will ever become reality remains to be seen. The mantle of nano has also been adopted by various scientific visionaries. called the Foresight Institute. nanoscale physics. In books. These nano robots will also serve as tiny factories. manufacturing anything and everything from food to antibiotics to energy.2 CHAPTER 1. materials (the realm of condensed matter physics) and molecular compounds (the realm of traditional chemistry). Perhaps the most prominent is Eric Drexler who has founded an institute. involving themes where such nano robots escape from the laboratory and evolve into sentient beings completely out of mankind’s control. In turn. Other questions that nanoscience asks include: • How does one make a nanometer sized object? . They will also kill cancer. In this respect.

In addition. personal computing.3 • How do you make many (identical) nanometer sized objects? • How do the optical and electrical properties of this nanoscale object change with size? • How does its optical and electrical properties change with its “dimensionality”? • How do charges behave in nanoscale objects? • How does charge transport occur in these materials? • Do these nanoscale materials posess new and previously undiscovered properties? • Are they useful? The transition to nanoscience begins at this last point when we ask how these nanoscale materials might be exploited to improve our lives. So whether it be for household lighting or consumer electronics. from amorphous (glass-like) to polycrystalline (multiple domains) to crystalline. Solids come in a number of forms. such as GE. have also started their own in house nano programs to revolutionalize consumer lighting. More established companies. HP. the composition of solids is discussed to introduce common crystal structures found in nanomaterials. data storage and so forth. a nano solution exists and there is very likely a company or person pursuing this vision of a nano future. materials are so interesting and potentially useful. hence the emphasis on crystal structure. Overview The idea behind these lecture notes is as follows: First in Chapter 2. Lucent and IBM. the structure . Venture capital firms and mainstream industry have therefore taken up this challenge with many small startups trying to apply nanoscale materials in products ranging from better sunscreen lotions to fluorescent labels for biological imaging applications to next generation transistors that will one day store the entire content of the Library of Congress on the head of a pin. nanoscale. So what is nano? This series of lecture notes tries to answer this question by explaining the physical concepts behind why such small. Much of nanoscience and nanotechnology focuses on nanometer sized crystalline solids.

the concept of crystal structure and the periodic potential due to the ordered arrangement of atoms is central to the concept of electronic bands. Finally. These natural length scales can either be referred to by their deBroglie wavelength or by the exciton Bohr radius. Developments in this area. the increase in surface area is important to applications where the surface to volume ratio plays a critical role such as in catalysis as well as in photovoltaics.1: Transmission electron micrograph of individual CdSe quantum dots Chapter 3 introduces the concept of length scales to put into perspective the actual physical lengths relevant to nano. Figure 1. Thus.4 CHAPTER 1. INTRODUCTION section also illustrates the increase in surface to volume ratio for nanomaterials over bulk. The surface is therefore potentially important in dictating a material’s optical and electrical properties when nanometer sized. which we will discuss later on. using nanostructures. Although being nanometer sized is often considered the essence of “nano”. the relevant physical length scales are actually relative to the natural electron or hole length scales in the parent bulk material. Next the concept of quantum confinement is introduced in Chapter 6 through the simple quantum mechanical analogy of a particle in a 1 di- . have led to increasingly efficient solar cells such as the Gratzel cell. in direct contrast to macroscopic solids where such numbers are typically much smaller. while a given nanometer sized object of one material may qualify for nano. Furthermore. a similar sized object of another material may not. This is because in nanometer sized systems up to 50% of the atoms lie at the surface of a nanostructure.

and 10. Likewise in the case of a quantum wire. confinement occurs along all three physical dimensions. However. there are electronic states associated with these extra two degrees of freedom. Analogies comparing the particle in a one dimensional box to a quantum well. insulators. The density of states argument is subsequently applied to both the valence band and conduction band of a material. 2 dimensional and 3 dimensional box. and semi-metals all have bands due to the periodic potential experienced by the electron in a crystal. Finally. The concept of quantum yields and lifetimes. and 0D (quantum dot) systems we turn to the concept of photoluminescence. This is one reason why quantum dots or nanocrystals are often called artificial atoms.y. This topic is important because metals.9. Quantum confinement is most commonly associated with nano in the sense that bulk materials generally exhibit continuous absorption and electronic spectra. and semi-metals can all be distinguished through . Bands are introduced in Chapter 15. the emission spectrum of a bulk 3D material is calculated using the derived Einstein A and B coefficients.5 mensional. Putting together both valence and conduction band density of states introduces the concept of the joint density of states (JDOS) in Chapter 11 which. this periodic potential occurs due to the ordered and repeated arrangement of atoms in a crystal. 1D (quantum wire). upon reaching a physical length scale equivalent to or less than either the exciton Bohr radius or deBroglie wavelength both the optical and electronic spectra become discrete and more atomic-like.and z such that the optical and electrical spectra become truly atomic-like. After describing the absorption spectra of 3D (bulk). in turn. systems will also emit light of certain frequencies. x. the Einstein A and B coefficients and their relationships are introduced and derived in Chapter 14. In the extreme case of quantum dots. is related to the absorption coefficient of a material. the particle in a two dimensional box to a quantum wire and the particle in a three dimensional box to a quantum dot provide only half the solution. semiconductors. To describe this process. 2D (quantum well). with two degrees of confinement. which describe the efficiency and timescale of the emission. To gain better insight into these additional states we introduce the concept of density of states (DOS) in Chapters 8. Generally speaking. Furthermore. in addition to absorbing light. completes this section. metals. there exists one degree of freedom. As mentioned earlier in the section on structure. So solving the particle in a two dimensional box problem models the electronic states along the two confined directions but does not address states associated with this remaining degree of freedom. If one considers that in a quantum well only one dimension is confined and that two others are “free”. semiconductors.

larger dots absorb and emit red light.6 CHAPTER 1. INTRODUCTION Figure 1.2: Photograph of the size dependent emission spectra of both HgS (top) and CdSe (bottom) quantum dots. . Small quantum dots absorb and emit blue/green light.

These bands and associated gaps are formed from the delocalization of carriers in this new periodic potential defined by the ordered arrangement of quantum dots. photodetectors. or quantum wells. An exploration of the band gap concept ultimately touches on the effects quantum confinement has on the optical and electrical properties of a material. or wires.7 Figure 1. much like the arrangement of atoms in a crystal. in turn. will determine their usefulness in applications such as solar energy conversion. or dots. or lasing applications. the occupation of these bands by electrons. For example. The band gap is extremely important for optoelectronic applications of semiconductors. 2 and 3 dimensions. Imagine artificial elements or even artificial metals. This forbidden range of energies (a no man’s land for electrons) is referred to as the band gap. and semiconductors. the band gap will generally determine what colors of light a given semiconductor material will absorb or emit and. semimetals. in the case of semiconductors and insulators there is a range of energies that cannot be populated by carriers separating the valence band from the conduction band. Talk about designer materials. At the same time. quantum wires. quantum wire and quantum dot. Analogous to a quantum well. which leads to the realization of a size dependent band gap. No wonder visionaries such as Drexler envision . Metals have “full” conduction bands while semiconductors and insulators have “empty” conduction bands. can ultimately lead to new “artificial” solids with artificial bands and corresponding band gaps.3: Cartoon of confinement along 1. Introducing the concept of bands is also important for another reason since researchers have envisioned that ordered arrays of quantum wells.

4: Size dependent absorption and emission spectra of colloidal CdSe quantum dots.8 CHAPTER 1. . INTRODUCTION Figure 1.

Figure 1. stacks of closely spaced wells have been grown leading to actual systems containing these minibands. metals. In the case of quantum dots and wires. Transport properties of these systems take center stage in the realm of devices where one desires to apply quantum dots. and quantum wires within opto-electronic devices such as single electron transistors. but they also have electrical properties as well. and semi-metals absorb and emit light.9 so much potential in nano. These properties are greatly affected by quantum confinement and the discreteness of states just as with the aforementioned optical properties. The concept of creating artificial solids with tailor made bands from artificial atoms has been tantalizing to many. The white space in between dots is the organic ligands passivating the surface of the particle. insulators. In the case of quantum wells. we introduce the concept of tunneling in Chapter 17 to motivate carrier trans- . Each dark spot is an individual quantum dot.5: TEM micrograph of an array of colloidal CdSe quantum dots ordered into an artificial crystal. Not only do semiconductors. To this end. minibands have not been realized yet but this is not for lack of trying. quantum wells.

quantum wires. In this fashion. Special emphasis is placed on the Coulomb blockade and Coulomb staircase problem. INTRODUCTION port in nanometer-sized materials. we turn to three topics that begin the transition from nanoscience to nanotechnology. the classical LaMer and Dinegar growth model is derived and explained. The WKB approximation is subsequently introduced in Chapter 18 to provide an approximate solution to Schrodinger’s equation in situations where the potential or potential barrier for the carrier varies slowly. which is the basis of potential single electron transistors to be used in next generation electronics. more general. and quantum wells using examples from the current literature. more uniform. This becomes important when one considers that the discreteness of states in confined systems may mean that there are substantial “barriers” for carrier transport along certain physical directions of the material. After providing a gross overview of optical. secondary electron microscopy (SEM). Special emphasis is placed on better understanding colloidal growth models currently being used by chemists to make better. Newer techniques now coming into prominence are also discussed at the end of the section. fashion. electrical and transport properties of nanostructures. . To this end. quantum dots and nanorods. Relations to the behavior of an ensemble size distribution are then discussed completing the section. Chapter 21 discusses applications of quantum dots.10 CHAPTER 1. as well as in another emerging field called molecular electronics. Chapter 20 describes some of the classical techniques used to characterize nanostructures such as transmission electron microscopy (TEM). Finally. A common result for the form of the tunneling probablity through an arbitrary barrier is derived. the concepts behind dip-pen nanolithography and microcontact printing are illustrated. Once created. one can repeat the same tunneling calculations as in the previous section in a faster. metal-organic chemical vapor deposition (MOCVD) and colloidal sysnthesis are described. atomic force microscopy (AFM) and scanning tunneling microscopy (STM). Chapter 19 describes current methods for making nanoscale materials. Tunneling is a quantum mechanical effect where carriers can have non-zero probability of being located in energetically fobidden regions of a system. A brief overview of each technique is given. This expression is commonly seen and assumed in much of the nano literature especially in scanning tunneling microscopy. In particular. tools are needed to study as well as manipulate nanoscale objects. Techniques such as molecular beam epitaxy (MBE).

it is also very common to see a basis consisting of multiple atoms. is to illustrate the crystal structure of common materials often encountered in the nano literature. However. from a material’s electronic spectra to its density and even to its powder x-ray diffraction pattern. potentially causing some confusion when you first look at the crystal structures of these materials. Since nano typically concerns itself with crystalline metal nanoparticles and semiconductor nanocrystals. crystal structure comes into play in many aspects of research. which is the case when one deals with binary or even ternary semiconductors. amorphous (no long range order. Solids generally appear in three forms. and wells. The goal. Individual atoms (or groups of atoms) are hung off of the lattice. glass-like). Atoms in a crystal are generally pictured as being arranged on an imaginary lattice. the basis consists of only a single atom and each atom is located directly over a lattice point. wires. These individual (or groups of) atoms are referred to as the “basis” of the lattice. Here the basis atoms do not necessarily sit at the same position as a lattice point. polycrystalline (multiple domains) or crystalline (a single extended domain with long range order). In the simplest case. The interested reader may consult a number of excellent introductory references such as Kittel’s introduction to solid state physics. There are 14 three dimensional Bravais lattices shown in Figure 2. however. The endless repetition of basis atom(s) on a lattice makes up the crystal.1. much like Christmas ornaments. In this respect. 11 . having a basic picture of how the elements arrange themselves in these nanocrystalline systems is important.Chapter 2 Structure Crystal structure of common materials This section is not meant to be comprehensive.

Rather. It is often useful to know the number of atoms in a unit cell. STRUCTURE These are also referred to as conventional unit cells (i.1: 14 3-dimensional Bravais lattices. From Ibach and Luth.12 CHAPTER 2. used in everyday life) as opposed to the primitive unit cell of which only the simple cubic (aka “primitive”) lattice qualifies. one can find even simpler repeating units by looking harder. A general counting scheme for the number of atoms per unit cell follows. Figure 2. • atoms entirely inside the unit cell: worth 1 • corner atoms: worth 1 8 .e. That is. most of these unit cells are not the simplest repeating units of an extended lattice. these conventional cells happen to be easy to visualize and interpret and hence are the ones most commonly used.

Ag. A simple way to differentiate the two is the atomic packing order. Note that an alternative name exists for the FCC unit cell: cubic close packed (CCP). Typical elements that crystallize in the BCC structure include: Fe. By counting atoms in this fashion. It contains 2 atoms per unit cell (8 on the corners and 1 inside). Au. The HCP structure has a conventional unit cell shown in Figure 2.2 illustrates the positions of an atom in the corner. V.13 • face atoms: worth • edge atoms: worth 1 2 1 4 Figure 2. 4 . 4 . represent different atom planes. 0. Ni. The weighting scheme follows because each of these unit cells is surrounded by other unit cells.3. Likewise. The diamond structure differs from its FCC and BCC counterparts because it has a multi atom basis. using the above counting scheme. hence all atoms with the exception of interior ones are shared. W and Mo. the diamond structure has a FCC lattice with a two atom basis. This is illustrated in Figure 2. Pt. Both unit cells are shown in Figures 2. and C etc. 0) and the second atom is at 4 . Any atom in the interior of a unit cell is exclusive to that cell. Another conventional unit cell that is often encountered is called the “diamond” structure. . Cr. they strongly resemble the Bravais lattices or conventional unit cells seen in the previous diagram. one can determine the number of atoms per unit cell. The letters A. the cubic lattices are important. edge and interior positions. which follows ABCABC in the case of FCC and ABABA in the case of HCP in the ¡001¿ and ¡0001¿ directions respectively. B. and Al. More complete tables can be found in Kittel. The first atom is at  1 1 1 (0. The number of atoms per unit cell in the FCC case is 4 (8 corner atoms and 6 face atoms). it does not immediately resemble any of the 14 Bravais lattices. Nb.6. It is adopted by elements that . In fact. Both FCC and BCC structures have a single atom basis. Therefore. Single element crystals In the case of metals. thus. which should be remembered when reading the literature. the number of atoms per BCC unit cell. is 2 (1 interior atom and 8 corner atoms). The corner atoms are shared by 8 unit cells while the face atoms are shared by 2 cells and finally the edge atoms by four. . Pd.5. Analogous to the FCC lattice is the hexagonal close packed (HCP) structure. face. Ta. with particular emphasis on the face centered cubic (FCC) and body centered cubic (BCC) structures.15 and 2. Typical elements that crystallize in the FCC structure include: Cu.

STRUCTURE Figure 2.2: Number of atoms per unit cell. Counting scheme. .14 CHAPTER 2.

3: Cartoon showing sharing of atoms by multiple unit cells.15 Figure 2. .

4: FCC unit cell Figure 2. STRUCTURE Figure 2.5: BCC unit cell .16 CHAPTER 2.

4. 6 face atoms). Summary. Ge and Sn. 0.17 Figure 2. As a side note.7). diamond • lattice:FCC • basis atom 1: (0. Si. such as III-V and II-V semiconductors. 41 . 41 ) displacement.6: Hexagonal unit cell. offset from each other by a ( 14 . 4  Compound crystals In the case of binary compounds. things get a little more complicated. One doesn’t have the benefit of conventional unit cells that resemble any of the 14 standard Bravais lattices. 0) • basis atom 2: 1 1 1 4. Some common elements that crystallize in the diamond structure include: C. have a tendency to form strong covalent bonds. . The number of atoms per unit cell in this case is 8 (8 corner atoms. resulting in tetrahedral bonding arrangements (Figure 2. 4 interior atoms. one way to visualize the diamond unit cell is to picture two interpenetrating FCC lattices.

41 . “NaCl” and “CsCl” lattices. 41 ) in Figure 2. The only real difference is that now we have two elements making up the atom basis of the unit cell.7: Diamond structure unit cell Instead these conventional unit cells often have names such as the “NaCl” structure or the “ZnS” structure and so forth. Using the above counting scheme we find that there are 8 atoms per unit cell. . The two atom. STRUCTURE Figure 2. Common crystal lattices for semiconductors include the “ZnS”. structure can be visualized as two interpenetrating FCC lattices offset by ( 14 . and 4 atoms of element 2.18 CHAPTER 2.8. This is because. 0. 41 . 0) and element2 14 . 41 . also called zinc blende (ZB) or sphalerite. This is further subdivided into 4 atoms of element 1. The underlying lattice is FCC. we no longer have a single atom basis. It is identical to the diamond structure we saw in the case of single element crystals. but rather a basis consisting of multiple atoms as well as a basis made up of different elements. unlike simple FCC or BCC metals. two  element basis has positions of element1 (0. The ZnS. You will notice in the figure that the 4 atoms of one element are completely inside the unit cell and that the atoms of the other element are arranged as 8 corner and 6 face atoms.

This is broken up into 4 atoms from element 1 and 4 atoms from element 2. PbSe.8: Zincblende or ZnS structure unit cell. and PbTe. 0) • basis. 0) . 4. element 1 (0. One can see in the figure that for element 1 there are 8 corner atoms and 6 face atoms. element 2 1 1 1 4. 0. 21 . The two element. 0. 0.19 Summary. 0) in Figure 2. The underlying lattice is FCC. element 1 (0. ZnS • lattice: FCC • basis.9. Examples of materials that crystallize in this structure include PbS. It has 8 atoms per unit cell. 4  Figure 2. The NaCl structure (also called rocksalt structure) can be visualized as 2 interpenetrating FCC lattices offset by ( 21 .  two atom basis has locations of element1 (0. 21 . Summary. 0. NaCl • lattice: FCC • basis. 0) and element2 12 . For element 2 there are 12 edge atoms and 1 interior atom.

CsCl and CuZn. Examples of materials that crystallize in this structure include. The underlying lattice is simple cubic and the twoatom. 0. 2.10 where one can see that there are two elements present with one of them being the center atom. 1 from each element. Summary. 0. element 2 1 1 1 2. . CsCl • lattice: SC • basis. 2 atoms from element 1 and 2 atoms from element 2. The CsCl has two atoms per unit cell. 0) 2 1 1 and element 2 2 . 0) and element 2 at 12 . 2  Figure 2. two element basis is element 1 at (0. 21 . Examples of materials that crystallize in this structure include CdSe and GaN. The unit cell is shown in Figure 2. It has a multi atom basis. 2. The atoms from the other element take up corner positions in the unit cell. element 2 1 1 1 2.11 and contains 4 atoms per unit cell. 3 . 21 . 0) • basis. The underlying lattice is hexagonal. two atom basis is element 1 (0.9: NaCl structure unit cell. 0. 2 . STRUCTURE • basis. 2  The wurtzite crystal structure is the compound material version of the single element HCP structure. element 1 (0. The  two element.20 CHAPTER 2. It is shown in Figure 2. The CsCl structure is the compound material version of the single element BCC unit cell.

21 Figure 2. 0. These numbers are referred to as Miller indices. and z directions. element 1 (0. 3. For the case of cubic lattices the lattice constant.e. divide out any common integer) • Express the plane through these integers in parentheses as (abc) .10: CsCl structure unit cell. y. • Take the desired plane and see where it intersects each x. (i. z axis in multiples of the lattice constant. • Next take the reciprocal of each intersection point and reduce the three values to their lowest integer values. Summary. element 2 2 1 1 3. 2  Miller indices Sometimes you will see the orientation of a crystal plane described by (001) and so forth. 0) • basis. y. wurtzite • lattice: hexagonal • basis. is the same in all x. a. They are generated using some simple rules described below.

STRUCTURE Figure 2. • Should the plane not intersect an axis. for example (a¯bc).12 and 2. say the y axis. just put a bar over the number.13. just write a 0. say the z axis.11: Wurtzite unit cell. For example (ab0) • If the intercept is in the negative side of an axis.22 CHAPTER 2. . Miller notation • (hkl) =crystal plane • {hkl} =equivalent planes • [hkl] =crystal direction • < hkl >=equivalent directions Examples are illustrated in Figures 2.

12: Examples of using Miller indices.23 Figure 2. .

24 CHAPTER 2.13: More examples of using Miller indices. STRUCTURE Figure 2. .

2: Group IV semiconductors I II Cu Ag Au Zn Cd Hg III B Al Ga In Tl IV C Si Ge Sn Pb V N P As Sb Bi VI S Se Te Po • Si=diamond structure • Ge=diamond structure III-V • GaN=ZB [cubic] (alternatively called ZnS structure) .25 Quick tables Short tables of common metals and semiconductors are provided below with their standard crystal structure. Common Metals Table 2.1: Common metals I II Cu Ag Au Zn Cd Hg III B Al Ga In Tl IV C Si Ge Sn Pb V N P As Sb Bi VI S Se Te Po • Ag=FCC [cubic] (alternatively called cubic closest packed) • Au=FCC [cubic] (alternatively called cubic closest packed) Common Semiconductors Group IV Table 2.

26 CHAPTER 2.4: Group II-VI semiconductors I II Cu Ag Au Zn Cd Hg III B Al Ga In Tl IV C Si Ge Sn Pb V N P As Sb Bi VI S Se Te Po • ZnS=ZB [cubic] • ZnSe=ZB [cubic] • CdS=ZB [cubic] • CdSe=wurtzite [hexagonal] • CdTe=ZB [cubic] IV-VI • PbS=NaCl structure (alternatively called rocksalt) • PbSe=NaCl structure (alternatively called rocksalt) • PbTe=NaCl structure (alternatively called rocksalt) .3: Group III-V semiconductors I II Cu Ag Au Zn Cd Hg III B Al Ga In Tl IV C Si Ge Sn Pb V N P As Sb Bi VI S Se Te Po • GaAs=ZB [cubic] (alternatively called ZnS structure) • InP=ZB [cubic] (alternatively called ZnS structure) • InAs=ZB [cubic] (alternatively called ZnS structure) II-VI Table 2. STRUCTURE Table 2.

1) .27 Table 2. Recall that each corner sphere is shared by 8 other neighboring cubes. Vsphere = where 4 3 πr 3 r=a 2 4  a 3 π 3 2 3 πa 6 Vsphere = = Also there is only 1 sphere in the cube. Assume that the spheres touch. Next the volume of the cube is Vcube = a3 so that the ratio of volumes is Ratio = = πa3 6 a3 π 6 The desired simple cubic packing fraction is Ratiosc = 0. what fraction of space is filled by atoms in the primitive cubic and fcc lattices.524 (2.5: Group IV-VI semiconductors I II Cu Ag Au Zn Cd Hg III B Al Ga In Tl IV C Si Ge Sn Pb V N P As Sb Bi VI S Se Te Po Packing fraction Supposing atoms to be rigid spheres. Simple cubic Determine the packing fraction of the simple cubic lattice.

28 CHAPTER 2.14: simple cubic. STRUCTURE Figure 2. .

In the fcc lattice there are 8 corner spheres and 6 face spheres.29 FCC Determine the packing fraction of the fcc lattice c2 = a2 + a2 = 2a2 √ 2a c = = 4r This results in √ 2 r= a 4 The volume of a single sphere is Vsphere = = = 4 3 πr 3 √ !3 2 4 π a3 3 4 √ 2 3 πa 24 Next we calculate how many spheres are in the cube. The total volume of the spheres in the cube is therefore √ ! 2π Vspheretotal = 4Vsphere = 4 24 √ 2 3 πa = 6 The volume of the cube is simply Vcube = a3 . The calculation is  • Corner: 8 81 = 1  • Face: 6 12 = 3 The total number of spheres in the cube is therefore 4.

STRUCTURE The desired ratio of volumes is then Ratio = √ 2πa3 6 a3 √ 2 = π 6 = 0. Smaller sizes translate to larger surface to volume ratios.741 (2.741 The packing fraction of the fcc lattice is Ratiof cc = 0.15: FCC.2) Figure 2. .30 CHAPTER 2. Surface to volume ratio Here we show that the surface to volume ratio of a number of systems scales as an inverse power law.

Volume V = a3 Surface to volume ratio R= S V = = 6a2 a3 6 a Sphere Surface area S = 4πa2 Volume of sphere V = 4 3 πa 3 Surface to volume ratio R= S V = = Cylinder Surface area of cylinder S = 2πal Volume of cylinder V = πa2 l 4πa2 4 3 3 πa 3 a .31 Cube Surface area S = 6a2 where a is the radius.

consider a collection of 1nm radius spheres that will have the same total volume as the original sphere. However in other cases. In one case illustrated below it can lead to more efficient catalysts. how many little spheres do I need to make up the same volume. Next.32 CHAPTER 2. there is often an incentive to make nanostructured materils in the hopes that the increase in surface to volume ratio will lead to more catalytic activity. Start with a 1 cm radius sphere. Vlittle = = 4 π(10−7 )3 3 4 π10−21 3 . To illustrate consider the following example. How does it compare? The surface of the large sphere is Slarge = 4πa2 = 4πcm2 The volume of the large sphere is Vlarge = = 4 3 πa 3 4 πcm3 3 Next. such as in the optical microscopy of nanostructures it can be a bad thing because of defects at the surface which quench the emission of the sample. Why. This is both good and bad. But the bottom line is that as the size of the system decreases. Calculate the surface area. the surface to volume ratio follows an inverse power law. surface to volume ratio? Since surfaces are useful for things such as catalysis. What is the total surface area of these little spheres. the fraction of atoms on the surface will increase.3) In all cases. STRUCTURE Surface to volume ratio R= S V = = 2πal πa2 l 2 a (2.

Do the same for As atoms.4 nm diameter Pt nanoparticle using the bulk density of Pt. Exercises 1.33 The ratio of the large to small sphere volumes is Vlarge Vlittle = 4 3π 4 −21 3 π10 21 = 10 Next the surface area of an individual little sphere is Slittle = 4π(10−7 )2 = 4π10−14 cm2 Multiply this by 1021 to get the total surface area. Calculate the number of silicon atoms in a cubic centimeter.65 ˚ A. How many total atoms exist on this device.5g/cm3 . Calculate the total number of atoms in the device. 3. 2. 4. Calculate the number of Ga atoms per cubic centimeter in a GaAs crystal. Vlittle. before we had a total surface area of 4πcm2 . The lattice constant of GaAs is 5. Consider a FCC unit cell with a lattice constant of a = 0. Consider ρ = 21.tot = 4π10−14 1021 = 4π107 cm2 107 So finally. 5. The lattice constant of Si is 5. Now we have times more surface area with the little spheres. Calculate the number of atoms in a 1. Calculate the number of atoms in a 1. Its dimensions are 100 x 100 x 20 microns.391 nm. 6.4 nm diameter Pt nanoparticle using the total number of unit cells present. Consider an actual silicon device on a wafer with physical dimensions of 5 x 5 x 1 microns.43 ˚ A. Consider a slightly larger GaAs laser on the same wafer. .

D. Cobalt is usually found with a hexagonal crystal structure. 3. Calculate the number of atoms in a 2 nm diamter -cobalt nanocrystal through a density argument. Rinehart and Winston. CdSe has a hexagonal unit cell (unit cell volume ' 112˚ A3 ). Calculate the number of surface atoms and percentage of surface atoms in a 2 nm diameter -cobalt particle. Calculate the number of atoms in a 2 nm diameter nanocrystal through a unit cell argument. Draw the surface of a Ag crystal cut along the (111) and (100) plane.097˚ A. The unit cell contains 20 atoms. Berlin 1990. Kittel Wiley. 8. “Solid-state physics” H. Ibach and H. The density is ρ = 8. Mermin Holt. W.635g/cm3 . “Solid state physics” N. One can calculate this through a unit cell approach but use whatever approach you like. 9. “Introduction to solid state physics” C. 13. New York. For the same CdSe dots considered. How many atoms of each element are there? 12. 11. 6 nm diameter particle. STRUCTURE 7. New York 1986. 2. . Ashcroft and N. calculate the fraction of surface atoms in each case and plot this on a graph.635g/cm3 . The lattice constants are a = 4. Estimate the number of surface atoms and percentage of surface atoms in a 1.3˚ A and c = 7˚ A.34 CHAPTER 2. New York 1976. It was recently found to crystallize with a simple cubic structure. 3. Calculate the total number of atoms in a CdSe quantum dot for a: 1. New York.4 nm diameter Pt nanoparticle. References 1. 2. 10. now called -cobalt. 5. Luth Springer Verlag. The lattice constant of -cobalt is a = 6. Use ρ = 8. 4.

“A solution-phase chemical approach to a new crystal structure of cobalt” D.35 Relevant literature 1. P. Bawendi Angew. Int. 38. . Chem. Dinega and M. 1788 (1999). Ed. G.

36 CHAPTER 2. STRUCTURE .

But as Uzi Landman once said. deBroglie comes along later and says “both”. optical and electrical properties of matter all become size and shape dependent. physics and biology. less filling anyone? So to make a long story short. This was his Ph.D thesis for which he got the Nobel prize in 1929. 37 . This is the opposite extreme-calling traditional chemistry. but for now just picture the classic Bohr model of electrons circling a heavy nucleus in stable orbits.Chapter 3 Length scales What are the relevant length scales for nano? Well. Was light a wave or was it a stream of particles? On one side you had Huygens who thought it was a wave. this is pretty high on the chutzpa scale. One useful perspective on a definition for the appropriate lengths scales for nano is a regime where the chemical. This could mean delaing with stuff on the hundreds of nm scale. We will describe excitons (electron hole pairs) in more detail shortly. Now others will tell you that nano is for extremely small things like molecules and such. I guess it depends on who you talk to. Then on the other side you had Newton who thought it was a stream of particles. physical. Tastes great. But isn’t this just traditional chemistry or even biology? After all chemists. nano to put a new spin on things. For semiconducting materials this is given by the bulk exciton Bohr radius. DeBroglie wavelength So what is the deBroglie wavelength? Well this arose within the context of a classic debate about the true nature of light back in the old days. On one hand some people call nano anything smaller than stuff on the micro level. biologists and physicists have been dealing with small things for a pretty long time even before the development of instruments such as transmission electron micrographs.

85×10−12 F/m (permittivity).11 × 10 kg (mass of a free electron) and q = 1. We will mostly deal with electrons from here on out. h ¯ = 1. h is Plank’s constant (6. The wave-like properties of matter become important when you deal with small things so although macroscopic objects have wave-like properties they are better described by Newton’s laws.2) where 0 = 8. Therefore according to deBroglie associated with each object is a wavelength. in fact.38 CHAPTER 3. both statements essentially say the same thing. me = 9. an atom.054×10−34 J ·s (Planck’s −31 constant over 2π). an electron. a car. a plane. λ= h p (3. Textbook Bohr radius Here is the textbook equation for the Bohr radius of an electron a0 = 4π0 ¯ h2 mq 2 (3. In this section we show that the two are related and that. a baseball. LENGTH SCALES But deBroglie’s claim actually goes beyond light because what it really says is that matter (whether it be you.3) .602 × 10−19 C (charge).62×10−34 J · s) and p = mv is the momentum with m as the mass. me. If you plug all the numbers in and do the math you come up with the result a0 = 5. a quantum dot or a nanowire–all of these things have both wavelike and particle-like properties). At other times one sees the statement that a nanomaterial is quantum confined because its size is smaller than the corresponding exciton Bohr radius. The reason we do this is that often in the literature one sees a statement that a nanomaterial is in the “quantum confinement” regime because its size is smaller than the corresponding deBroglie wavelength of an electron or hole. We ask if these are the same statement.1) where λ is the deBroglie wavelength.28 × 10−11 meters = 0.528 Angstroms (3. DeBroglie wavelength and exciton Bohr radius Here we derive the relationship between the deBroglie wavelength and the exciton Bohr radius.

Derivation Basically we need to equate the centrifugal force of a carrier with the Coulomb attractive (inward) force.39 This is the standard Bohr radius one sees all the time.3) q2 n2 ¯h2 = mr 4π0 Rearrange this to get r= 4π0 n2 ¯h2 mq 2 If n = 1 (the lowest orbit) this gives us the Bohr radius a0 = 4π0 ¯h2 mq 2 . that an integer number of wavelengths must “fit” into the circumference of the classic Bohr orbit. Starting with the above equation we rearrange it to get λ= h h 2πr = = n p mv Solve for v to get (express v in terms of r) v= nh n¯h = 2πmr mr Replace this into the main equation (3. Namely. The deBroglie relation comes in by relating the wavelength λ = hp where h is Planck’s constant and p is the momentum of the particle (p = mv).4) Here we make use of the relation 2πr = nλ (3.5) where n is an integer. q2 mv2 = r 4π0 r2 (3.

40 CHAPTER 3. So. Any mobile charges or dipoles in the system will move or re-orient to oppose the field. using the values below. The deBroglie wavelength or exciton Bohr radius are therefore natural length scales by which to compare the physical size of a nanomaterial. . then one must account for possible extra charges or dipoles that can move to oppose the electric field that one establishes between the original two oppositely charged species. This list is not meant to be comprehensive and the interested reader should consult the Landolt Bornstein tables for more complete values. in effect. Furthermore. In general objects with dimensions smaller than these natural length scales will exhibit quantum confinement effects.4 basically gives the relation between the deBroglie wavelength and the exciton Bohr radius.6) This is because if you are not in vacuum which you aren’t in a solid.7) where me and mh are the effective masses of electron and hole in the material. Examples Here are some values for some common systems where I’ve taken values of the relative dielectric constant. (Instead of  replace it with 0). The net effect is that the original electric field as well as Coulomb potential are diminished by the response of the material. for the case of an exciton (electron hole pair) in a semiconductor just replace the mass of the electron with the effective mass of the exciton. LENGTH SCALES which is the standard textbook equation we showed earlier. a0 = 4π0 ¯ h2 mq 2 (3. The factor by which it is diminished is the relative dielectric constant. trying to cancel it out. Note that equation 3. our initial statements about confinement dealing with either the exciton Bohr radius or deBroglie wavelength are essentially one and the same. 1 mef f = 1 me + 1 mh (3. This will be discussed in more detail in subsequent chapters. liquid or gas. electron and hole effective masses from the literature. At this point we note that if the electron or carrier is not in vacuum the equation should be modified to take into account the dielectric constant of the medium. in a straightforward fashion. One can derive the exciton Bohr radius of these systems. denoted commonly as .

07m0 mh = 0.45m0  = 9.13m0 mh = 0.2m0 mh = 0.5 • InP: me = 0.067m0 0.6 • CdSe: me = 0.45m0 leading to the effective mass mef f = 0.058m0 .067m0 mh = 0.4 Worked example Case: (GaAs) 1 mef f = = 1 1 + me mh 1 1 + 0.4m0  = 14.7m0  = 8.02m0 mh = 0.4m0  = 14 • CdS: me = 0.45m0  = 12.4 • InAs: me = 0.41 • GaAs: me = 0.

4)(1.058m0 q 2 4π(8.054 × 10−34 )2 = (0.3 nm ab = Exciton Bohr radius for GaAs.85 × 10−12 )(12.97 nm ab = Exciton Bohr radius for CdSe. Now finally let’s put things into perspective.602 × 10−19 )2 = 4. Thus the radius of the electron orbit is about 20 unit cells away from the counterpart hole. We just found that the bulk exciton Bohr radius was aB = 11.3nm.058)(9.13m0 0. You can see that in semiconductors the electron can move very far away from its counterpart due to the screening effect of the material.054 × 10−34 )2 = (0.602 × 10−19 )2 = 11.4)(1.42 CHAPTER 3. So in the end you can see that because of the variable relative dielectric constant and effective masses of carriers in a semiconductor there will be various exciton Bohr radii. Consider the case of GaAs.11 × 10−31 )(1. Case: (CdSe) 1 mef f = = 1 1 + me mh 1 1 + 0. .85 × 10−12 )(9.1)(9.45m0 leading to the effective mass mef f = 0.65˚ A.058m0 q 2 4π(8.1m0 4π0 ¯h2 0. The lattice constant of GaAs which crystallizes in the zinc blende form is a = 5.11 × 10−31 )(1. LENGTH SCALES 4π0 ¯h2 0.

What is the deBroglie wavelength of C60 moving at 220 m/s? Read the corresponding article if you are interested. Nature 401. What is the wavelength of a 3 eV photon? 4. What is the wavelength of a 1 eV photon? 2. What is the deBroglie wavelength of a baseball (0. Arndt et. Calculate the exciton Bohr radius for the following semiconductors.15 kg) moving at 50 m/s? 6. What is the wavelength of a 2 eV photon? 3.43 Exercises 1. If needed use values for what is called the heavy hole. al. II-VI compounds CdS CdSe CdTe III-V compounds InP InAs IV-VI compounds PbS PbSe PbTe 8. Consult a good resource such as Landolt Bornstein. Explain what the size of the exciton Bohr radius means for achieving quantum confinment. 7. What systems are easiest for achieving confinement. 680 (1999). . What is your deBroglie wavelength (what’s your weight in kg?) when moving at 10 m/s? 5.

LENGTH SCALES .44 CHAPTER 3.

Frenkel excitons are commonly seen in organic semiconductors. the exciton energy Eexc = Eg − Ebind where Eg is the bulk band gap of the semiconductor (analogous to the HOMO-LUMO gap in molecular systems) and Ebind is the exciton binding energy that we will derive shortly.Chapter 4 Excitons Excitons are bound electron-hole pairs caused by the absorption of a photon in a semiconductor. Corresponding binding energies are on the order of 100meV . Corresponding binding energies are on the order of 10meV . Mott-Wannier excitons have weak electron-hole interactions caused by a small Coulomb attraction and hence the carriers are relatively far apart. There are two types of excitons. The excitons most commonly dealt with in semiconductors is the MottWanner exciton due to the large relative dielectric constant of most bulk semiconductors. Frenkel excitons have strong Coulomb interactions between the electron and hole. Excitons can be seen in the absorption spectrum of bulk semiconductors. They generally appear just below the band edge of the semiconductor. By contrast. This is because the energy of the exciton is lower than the band edge transition by its binding energy. Mott-Wannier excitons and Frenkel excitons. Namely. we have an electron in the conduction band of the semiconductor and a hole in the valence band of the semiconductor. distinguished by how close the electron and hole are. Excitons are potentially mobile and are charge neutral. The carriers are close togther. 45 . Specifically.

1: Exciton types Figure 4.2: Exciton absorption .46 CHAPTER 4. EXCITONS Figure 4.

. ao .47 Exciton Bohr radius and binding energies From the previous chapter we know that the Bohr radius of the electron-hole pair is aB = 4πo¯h2 µq 2 where µ is the exciton reduced mass (i used m last time. It isn’t readily apparent which  to use in the above expressions since there is the static dielectric constant. The total energy of the exciton relative to its ionization limit is the difference between its kinetic energy and its Coulomb (attractive) potential energy. Next we want the energy of the exciton. sorry for the notation change). This is where ao is everyone’s favorite textbook Bohr radius (ao = 0. use the dielectric constant at the same frequency corresponding to the energy required to create the exciton. 1 q2 tot = mv2 − 2 4πo r .528A a more convenient way to go about calculating the exciton Bohr radius of something. Recall that ao = 4πo ¯h2 mo q 2 Therefore the bulk exciton Bohr radius can be re-written as aB = 4πo ¯h2 mo mo q 2 µ where  is the relative dielectric constant of the semiconductor and µ is the reduced mass of the exciton. Generally speaking. Note that we have omitted something pretty important.1) ˚). In real life the relative dielectric constant. the high frequency dielectric constant ∞ and anything inbetween. is frequency dependent. This gives aB =  mo µ  ao (4. Usually folks like to write the exciton Bohr radius in terms of the standard textbook Bohr radius for the electron.

2) Again. So we now have o tot = −R  µ 2  mo  1 n2 Next define the exciton Rydberg as the product RX = R desired textbook expression tot = − RnX2  µ 2 mo  to get the (4. EXCITONS Since we previously saw in deriving the Bohr radius that mv2 r = mv2 = q2 4πor2 q2 4πor so that    q2 q2 − 4πor 4πo r   2 1 q = − 2 4πo r tot = 1 2  But recall that we just found that in a bound exciton r = aB = 4πo ¯h2 n2 µq 2 Note that many times you will see the exciton Bohr radius written as aX . Same thing. To express this like everyone else    µ mo q 4 1 1 tot = − 2 2 2 2 (4πo) ¯h  mo n2 Now recognize that   the first term in the expression is our textbook Rydberg mo q 4 1 R = 2 (4π )2 ¯h2 which refers to an energy of 13. Then we get our desired energy tot = − 12  µq 4 (4πo )2 ¯ h2  1 n2 (4.3) . you will often see this expressed in many texts in terms of the Rydberg and the exciton Rydberg (RX ).6 eV.48 CHAPTER 4.

602 × 10−19 )4 8(8. If so then divide this by 1.185 × 10−18 The units are Joules.602 × 10−19 J/eV to get R = 13.6eV So we did things right.62 × 10−34 )2 all in SI units. Evaluate this to get R = 2.49 Check me You can check that indeed the Rydberg is correctly defined.11 × 10−31 )(1. In terms of units to prove that the last number was really in units of joules we have kgc4 = = = = F2 2 2 j s m2   kgm2 s2 jc4 j2F 2 c4 jF 2 c4 V 2 jc2 sinceF = c4 F 2j2 c V c2 V 2 j 2 j = j sincej = cv = = j . Start with our above definition and simplify it a little to get mo q 4 82o h2 R= Numerically we have R= (9.85 × 10−12 )2 (6.

Finally we want the binding energy of the exciton. EXCITONS Ok.5) Exciton Schrodinger equation Excitons can be described as quasi particles and hence there is a quantum mechanical description for their energies and wavefunctions. the units work out. We begin with the Hamiltonian for two independent particles (an electron and a hole) coupled by a common Coulomb term and then re-express the equation into one for the bound exciton. If   µ 1 tot = −R 2  mo n2 then bind = ∞ − n=1 giving bind = −R  µ 2  mo  1 1 − 2 2 n∞ n1  For the lowest orbit this gives bind = R In general bind =  µ 2 m mq 4 2(4πo )2 ¯ h2 o  1 n2 (4. This is the difference in energy between the electron-hole pair in a given orbit (say n = 1) and being at an infinite separation (n = ∞). Great.50 CHAPTER 4.4)  (4. In one dimension we have   h2 d2 ¯ ¯h2 d2 e2 − Ψe Ψh = Ψe Ψh − − 2me dx2e 2mh dx2h 4πo (re − rh ) More generally however we have   ¯h2 2 e2 h2 2 ¯ Ψe Ψh = Ψe Ψh ∇ − ∇ − − 2me e 2mh h 4πo (re − rh ) To begin converting this into an exciton picture let R = = me re mh rh + me + mh me + mh mh rh me re + M M .

It can be shown that where µ = me mh (me +mh ) re = R + µr me (4. First expression re = R + = = = = = = µr me me re mh rh + + M M mh rh me re + + M M me re mh rh + + M M mh re me re + M M me + mh re M re me mh (re − rh ) M me mh (re − rh ) M mh re mh rh − M M “Proof”.6) where M = me + mh is the total mass of the exciton.7) be the relative electron hole coordinate. Second expression rh = R − = = = = = µr mh me re mh rh + − M M me re mh rh + − M M mh rh me rh + M M  me + mh rh M rh me mh (re − rh ) M mh me re me rh + M M . Next let r = re − rh (4.51 R= me re M + mh rh M (4.9) is the reduced mass. “Proof”. This is the center of mass coordinate.8) rh = R − µr mh (4.

11) . First expression ∇e = ∂ ∂re = = = ∂ ∂r ∂ ∂R + ∂r ∂re ∂R ∂re ∂ ∂  me  (1) + ∂r ∂R M m  ∂ ∂ e + ∂r M ∂R yielding ∇e = ∇r + me M ∇R (4.52 CHAPTER 4. EXCITONS By the chain rule we can now re-express ∇e and ∇h as follows since both re and rh are seen to depend on both r and R.10) Second expression ∇h = ∂ ∂h ∂ ∂r ∂ ∂R + ∂r ∂rh ∂R ∂rh ∂ ∂  mh  = (−1) + ∂r ∂R M ∂ mh ∂ = − + ∂r M ∂R = yielding ∇h = −∇r + mh M ∇R Put everything together now Back to our original expression   ¯2 2 h ¯2 2 h e2 − Ψtot = Ψtot ∇ − ∇ − 2me e 2mh h 4πo(re − rh ) (4.

53 where we will flip this Hamiltonian to something in terms of r and R only.    me h2  ¯ me ∇R ∇r + ∇R − ∇r + 2me M M    2  h ¯ mh mh ∇R −∇r + ∇R − −∇r + 2mh M M  e2 Ψtot = Ψtot − 4πo r  ¯2 h − 2me    m 2 me me e 2 + ∇r ∇R + ∇R ∇r + ∇R M M M    m 2 h2 ¯ mh mh h − ∇2R ∇r ∇R − ∇R ∇r + ∇2r − 2mh M M M  2 e Ψtot = Ψtot − 4πo r ∇2r    ¯2 h ¯h2  me 2 2 ¯ 2  mh 2 2 h h2 ¯ 2 − ∇r + − ∇R − ∇R − 2me 2mh 2me M 2mh M h2  me  ¯ h2  me  ¯ ∇r ∇R − ∇R ∇r − 2me M 2me M ¯h2  mh  h2  mh  ¯ ∇r ∇R + ∇R ∇r + 2mh M 2mh M  e2 − Ψtot = Ψtot 4πo r   ¯2 h − 2  1 1 + me mh  ∇2r ¯2 h − 2M  me ∇2R mh ∇2R + M M  h2 ¯ ¯2 h ¯2 h ¯2 h ∇r ∇R − ∇R ∇r + ∇r ∇R + ∇R ∇r 2M  2M 2M 2M e2 − Ψtot = Ψtot 4πo r −   ¯ 2 (me + mh ) 2 h e2 ¯2 2 h Ψtot = Ψtot − ∇r − ∇ − 2me mh 2M R 4πo r   ¯2 2 h e2 ¯2 2 h Ψ(r)Ψ(R) = Ψ(r)Ψ(R) ∇ − − ∇r − 2µ 2M R 4πo r .

GaAs has a zinc blende crystal structure with a lattice constant of a = 0. . Low temperature experiments are commonly conducted at 4K and 77K. Provide the following: (a) the bulk exciton Bohr radius of this material. Any suggestions as to why? 2.2mo . h 2 h ¯ − 2µ ∇2r − h e2 4πo r i Ψ(r) = rel Ψ(r) (4. divide by Ψ(r)Ψ(R) to get 1 ¯ 1 ¯h2 2 h2 2 0 ∇r Ψ(r) −  − ∇ Ψ(R) − cm = 0 Ψ(r) 2µ Ψ(R) 2M R This gives you two separate equations.12) i h2 ¯ ∇2R Ψ(R) = cm Ψ(R) − 2M (4. 10K and 4K. 300K. Provide a value for the thermal energy. (b) the number of unit cells contained within the lowest exciton orbit n = 1. a dielectric constant of  = 12. Solve for the eigenvalues of the relative expression to get the energies of the bound electron-hole pair. 77K. EXCITONS Since this is a separable problem.8.067mo and mh = 0. (c) The highest temperature at which stable excitons are observable. let  = cm + rel giving  ¯2 2 h h2 2 ¯ e2 − ∇r − ∇ − − cm − rel Ψ(r)Ψ(R) = 0 2µ 2M R 4πo r     e2 ¯h2 2 h2 ¯ − rel + − ∇R − cm Ψ(r)Ψ(R) = 0 − ∇2r − 2µ 4πo r 2M  −Ψ(R) 0 where  = − ¯h2 2 ¯2 2 h 0 ∇r Ψ(r) −  Ψ(r)Ψ(R) − Ψ(r) ∇ Ψ(R) − cm Ψ(r)Ψ(R) = 0 2µ 2M R e2 4πo r − rel .56nm. So you get the two separate equations for the exciton relative motion and the center of mass motion. in units of meV at these temperatures.13) Exercises 1. Next. One which depends only on r(left one) and one which depends only on R (right one). and electron/hole effective masses of me = 0. kT.54 CHAPTER 4. Both are solved independently.

55

Confinement regimes
At the heart of the binding energy is the Coulomb attraction between the
oppositely charged electron and hole. Recall that this Coulomb term is proportional to 1r . Now it will be shown that the confinement energy associated
with either the electron or hole is proportional to r12 . Its clear that the latter
term will grow faster than the former when the size of a material becomes
small. So in bulk materials, the Coulomb term and associated exciton is
important. By contrast, in nanomaterials the exciton begins to experience
competition from the “confinement” of individual electrons and holes.
There are three size (confinement) regimes that one must consider in
real life. They are referred to as the strong, intermediate and weak confinement regimes. More about confinement will be illustrated in the following
chapters. However, for now note that life isn’t all that simple.
• Strong confinement: a < aB,e , aB,h
• Intermediate confinement: aB,h < a < aB,e
• Weak confinement: a > aB,e , aB,h
Weak confinement
In this case a > aB,e , aB,h . As a consequence the binding energy of the
exciton is larger than the individual confinement energies of the electron
and hole. This is definitely the case with bulk materials and with “large”
nano scale materials. The energy of optical transitions here is basically the
band gap energy minus the exciton binding energy.
Intermediate confinement
In this case we have the physical radius of the material smaller than one
carrier’s individual Bohr radius but larger than the individual Bohr radius
of the other carrier. Because the effective mass of the electron is smaller than
the effective mass of the hole, one often sees the criteria, aB,h < a < aB,e .
We have
aB,e =
aB,h =

4πo¯h2
me q 2
4πo¯h2
mh q 2

56

CHAPTER 4. EXCITONS

Strong confinement
This is commonly seen in small nanomaterials. Here the physical size of
the system is smaller than both individual Bohr radii of the electron and
hole, a < aB,e , aB,h . In this regime the optical properties of the material are
dominated by quantum confinement effects of the electron and hole. This
will be seen in the following Chapters.

Exercises
1. For InP which has the following parameters, me = 0.077mo , mh =
0.2mo and  = 12.4, what are the individual Bohr radii of the electron
and hole. Provide physical size ranges for this material to be in the
strong, intermediate and weak confinement regimes.
2. For CdSe which has the following parameters, me = 0.13mo , mh =
0.45mo and  = 9.4, what are the individual Bohr radii of the electron
and hole. Provide physical size ranges for this material to be in the
strong, intermediate and weak confinement regimes.

Chapter 5

Quantum mechanics review
Wavefunctions and such
Given the deBroglie wave-particle duality it turns out that we can mathematically express a particle like a wave using a “wavefunction” (usually
denoted Ψ). This wavefunction replaces the classical concept of a trajectory
and contains all the dynamical information about a system that you can
know. Usually much of the work we will do here will be to find out what
this wavefunction looks like given certain constraints on the system (called
boundary conditions).
There is a probabilistic interpretation of the wavefunction called the
Born interpretation. In this respect
|Ψ|2 = Ψ∗ Ψ is considered as a probability density

|Ψ|2 dx = Ψ∗ Ψdx is considered as a probability

Through these quantities one can determine the probability that the particle
is somewhere. Note that from a physical perspective only |Ψ|2 has some
physical significance. Ψ can be real or imaginary or negative but |Ψ|2 will
be real and positive.
One consequence of the probabilistic interpretation of the wavefunction
is that the wavefunction must be normalized.
Z
|Ψ|2 dx = 1
This is because the probability of finding the particle somewhere must be
unity. So generally you will see that in front of Ψ(x, t) will be a constant
N which ensures normalization is met. Physical particles therefore have
57

58

CHAPTER 5. QUANTUM MECHANICS REVIEW

normalizable wavefunctions, non-physical ones have wavefunctions that are
not normalizable.
There are some important mathematical properties (or mathematical
constraints) of the wavefunction described as follows
• Ψ must be “well behaved” in general
• Ψ must be finite (it does not blow up or show singularities. This constraint of course is related to being able to normalize the wavefunction.
Can’t do it if it goes nuts.)
• Ψ must be single valued. Obviously it doesn’t make sense to have
multiple probabilities at a given position.
• Ψ must be continuous.
• Ψ must have a first derivative (related to previous statement)
• Ψ must have a second derivative (related to previous, previous statement)
• Ψ must be integrable and hence normalizable (basically a consequence
of 1 and 2)
Other properties to remember include
R
• normalization |Ψ|2 dx = 1
R
• orthogonality Ψ1 Ψ2 dx = 0. Thus these two wavefunctions Ψ1 and
Ψ2 have no mutual spatial overlap.

Observables
All dynamical information about the particle or system is contained in the
wavefunction. This includes observables such as
• position
• momentum
• angular momentum
• energy

For example. Some examples of operators include • x (multiply by x) • ∂ ∂x (take the partial derivative with respect to x) • ∂ ∂t (take the partial derivative with respect to time) You will sometimes see operators denoted by a little hat on top of it. read as multiply by x ∂ • momentum operator. Two fundamental operators are the position operator x ˆ and the momentum operator pˆ. An operator will act on a wavefunction to transform it into another function. observables are represented by mathematical ”operators”. In quantum mechanics.59 So while the wavefunction itself is not a measurable quantity. x ˆ = x. just like the particle is represented by a mathematical wavefunction. these observables are in principle. For example • x ˆ = x operator (multiply by x) • yˆ = y operator (multiply by y) • and so forth Correspondence principle In quantum mechanics all physical observables have corresponding operators. for the kinetic energy operator (used to extract out the energy of the particle through its wavefunction) T = = = p2 2m    1 ∂ ∂ −i¯h −i¯h 2m ∂x ∂x   2 1 ∂ −¯h2 2 2m ∂x = − ¯ 2 ∂2 h 2m ∂x2 . • position operator. pˆ = −i¯h∇ = −i¯h ∂x (read as take the derivative with respect to x and multiply by −i¯h) All other operators can be constructured from combinations of these two fundamental operators.

4) (5. ∂x ∂y ∂z In addition ∆ = ∇2 (French) So to summarize you can see that other operators can be built up from the fundamental operators x ˆ and pˆ. .60 CHAPTER 5. y.1) For the potential energy V (x) we have V (x) = V (x) (5. y. QUANTUM MECHANICS REVIEW This leads to 2 2 h ∂ ¯ T = − 2m ∂x2 (5. y. z) 2 h ¯ ∇2 H = − 2m (5. z) 2m 2m 2m    p2y 1 ∂ ∂ p2 −i¯h −i¯h + + z + V (x. H H = T +V ¯h2 ∂ 2 + V (x) = − 2m ∂x2 Therefore 2 2 h ∂ ¯ H = − 2m + V (x) ∂x2 (5. z) 2m ∂x ∂x 2m 2m = − ¯ 2 ∂2 h ¯h2 ∂ 2 ¯h2 ∂ 2 − − + V (x. y. z) 2m ∂x2 2m ∂y 2 2m ∂z 2 This results in 2 h ¯ H = − 2m  ∂2 ∂x2 + ∂2 ∂y 2 + ∂2 ∂z 2  + V (x. For the total energy in 1D we have the operator H = T + V where H is called the Hamiltonian operator. .2) Read this as multiply by V (x).3) This is the total energy operator called the Hamiltonian operator or Hamiltonian for short.5) where ∇2 is called the Laplacian or “Del-squared” and   ∂ ∂ ∂ ∇= . For the total energy in 3D H = = p2y p2 p2x + + z + V (x.

61 Eigenvalues and eigenfunctions Now a special class of function which when operated on yields itself again (may be multiplied by a constant or something) is called an eigenfunction (or eigenvector) of the operator. ∂ ∂x operator and the constant Expectation values The average of mean value of an observable A is by definition < A >= R ˆ Ψ∗ AΨdx This is called “sandwiching”. In the special case that Ψ is an eigenfunction of the operator A ˆ = aΨ AΨ Therefore <A> = Z ˆ Ψ∗ AΨdx Z Ψ∗ aΨdx Z = a Ψ∗ Ψdx = = a since by normalization the integral is 1 Therefore < A >= a in the special case that Ψ is an eigenfunction of A.6) . (5. For example. you could have ∂ αx (e ) = αeαx ∂x Here the function eαx is an eigenfunction of the α is an eigenvalue of the operator. The constant that falls out is called an eigenvalue of the operator.

62 CHAPTER 5. In this notation • A wavefunction Ψ is denoted by |Ψ > and is called a ket • The complex conjugate Ψ∗ is denoted by < Ψ| and is called a bra Bras and kets follow the rules of linear algebra • < a|b >=< b|a >∗ • |aA >= a|A > if a is a constant (a scalar) • < aA| = a∗ < A| Likewise • < A|bB + cC >= b < A|B > +c < A|C > Integrals are represented Z Ψ∗1 Ψ1 dx Z Ψ∗2 Ψ1 dx Z ˆ 1 dx Ψ∗1 AΨ Z ˆ 1 dx Ψ∗2 AΨ in the following fashion = < 1|1 >= 1 normalization = < 2|1 >= 0 orthogonality ˆ > Sandwiching = < 1|A|1 ˆ > Sandwiching = < 2|A|1 and so forth. they are important restrictions on operators that represent observable quantities. Operator math Just as wavefunctions have to obey certain constraints. QUANTUM MECHANICS REVIEW Dirac Bra-Ket notation (a quick aside) Dirac suggested a shorthand notation for writing and dealing with wavefunctions. operators and such. This new notation turns out to be easier and faster than the default calculus approach. This is called “linearity” and “hermiticity” .

eigenfunctions of a Hermitian operator associated with different eigenvalues are all orthogonal. Furthermore. “Proof”: ˆ >∗ = < k|A|j ˆ > < j|A|k k < j|k >∗ = j < k|j > where < k|j >=< j|k >∗ (k − j) < k|j > = 0 Since k 6= j (they are different eigenvalues) < k|j >= 0 . If its going to be observable in real life the number has to be real (not complex).63 Linearity: An operator Aˆ is linear for any constants α and β and wavefuncions |f > and |g > if ˆ > +β A|g ˆ > Aˆ (α|f > +β|g >) = αA|f So the operator acting on the whole superposition wavefunction Ψtot = α|f > +β|g > is the same as the operator acting individually on the parts. Because Aˆ is Hermitian ˆ >∗ = < k|A|k ˆ > < k|A|k k∗ = k Therefore k is real valued. “Proof”: |k > is an eigenfunction of Aˆ with eigenvalue k. Hermiticity: An operator Aˆ is Hermitian if its integrals obey the following relationship for all valid |f > and |g > Z Z ∗ ˆ ˆ f dx Ψf AΨg dx = Ψ∗g AΨ ˆ >∗ = < g|A|f ˆ > < f |A|g This restriction ensures that eigenvalues of Aˆ are all real-valued (as opposed to complex).

9) The first expression reflects operation on the right.8) ˆ >=< Aˆ† φ|Ψ > < φ|A|Ψ (5. 3) ˆB) ˆ †=B ˆ † Aˆ† (A Proof: Let ˆ |φ > = (AˆB)|Ψ > ˆ ˆ = A|BΨ > (5. 2) ˆ >=< φ|AΨ ˆ > < φ|A|Ψ (5.10) .64 CHAPTER 5. The second expression reflects operating on the left. More games with the adjoint 1) ˆ > = < φ|AΨ ˆ > < φ|A|Ψ ˆ = < AΨ|φ >∗ = < Ψ|Aˆ† |φ >∗ Therefore ˆ >=< Ψ|Aˆ† |φ >∗ < φ|A|Ψ (5. QUANTUM MECHANICS REVIEW More on operators ˆ >= |AΨ ˆ >= |Ψ0 > A|Ψ so that 0 0 |Ψ > ←→ < Ψ | ˆ > ←→ < AΨ| ˆ |AΨ ˆ > ←→ < Ψ|Aˆ† A|Ψ where Aˆ† is called the adjoint of Aˆ operator.7) of course if A = A† then A is Hermitian.

This will be introduced in the next section. the corresponding operators must “commute” Definition of “commute” ˆ taken in succession on any Ψ The action of the two operators say Aˆ and B is identical to the action of the operators taken in reverse order. the two operators. this Ψ does not have to have a characteristic value of any other observable. ˆ A] ˆ (A (5. .13) or equivalently that the “commutator” of the two operators equals zero.65 Take adjoint now ˆ BΨ ˆ > → < BΨ|Aˆ† A| ˆ > → < Ψ|B ˆ † Aˆ† AˆB|Ψ ˆ = AˆB ˆ or Cˆ † = B ˆ † Aˆ† resulting in Call C ˆ > → < Ψ|C ˆ† C|Ψ This ends up with ˆ †=B ˆ † Aˆ† (AˆB) (5. For a Ψ to have a characteristic value of 2 observables simultaneously. ˆ=B ˆ Aˆ AˆB (5.12) Commutators A wavefunction Ψ with a characteristic well-defined value of some observable quantity is an eigenfunction of the corresponding operator.11) ˆ are Hermitian (Aˆ = Aˆ† and B ˆ=B ˆ †) 4) If Aˆ and B ˆB) ˆ †=B ˆ † Aˆ† = B ˆ Aˆ = AˆB ˆ + [B. However. ˆ A] ˆ (A ˆ A] ˆ is called the “commutator” of This results in the relationship where [B. ˆB) ˆ † = AˆB ˆ + [B.

ˆ A])|β ˆ B| > ˆ A] ˆ =0 since [B.66 CHAPTER 5. QUANTUM MECHANICS REVIEW Definition of “commutator” ˆ B] ˆ = AˆB ˆ−B ˆ Aˆ [A. A special case is ˆ B] ˆ = AˆB ˆ −B ˆ Aˆ = 0 [A. ˆ A] ˆ = 0). ˆ > = AˆB|β ˆ >) = β(A|β ˆ > = β|Aβ . To illustrate: ˆ Aβ ˆ > = B ˆ A|β ˆ >= (A ˆB ˆ + [B. tion of both Aˆ and B Alternative derivation ˆ Claim that if |β > is an eigenfunction of B ˆ >= β|β > B|β ˆ commute (i. which occurs when Aˆ and B Now if such a Ψ exists then it is an eigenfunction of both operators. [B.e. For example ˆ B|β ˆ >) = Aβ|β ˆ ˆ > A( >= β A|β ˆ=B ˆ Aˆ then But now if AˆB ˆ > = B ˆ A|β ˆ > AˆB|β ˆ > = B( ˆ A|β ˆ >) β A|β ˆ >) is an which means that (A|β ˆ call it |β 0 > eigenfunction of B. |Aβ ˆ > is also an eigenfunction Then if Aˆ and B ˆ of B with eigenvalue β. ˆ if they commute. ˆ commute. ˆ 0> = B|β 0 = β|β > ˆ Hence |β > is an eigenfuncThis implies that |β > is an eigenfunction of A.

pˆ]|Ψ >= i¯h|Ψ > (5. otherwise it is easy to get confused.16) or alternatively that This value is non-zero showing that x ˆ and pˆ do not commute.67 One therefore sees that ˆ Aβ ˆ >= β|Aβ ˆ > B| (5. And as advertised earlier this will lead to the Uncertainty principle ∆x∆p ≥ ¯ h 2 (5. pˆ]|Ψ > = (ˆ xpˆ − pˆx ˆ)|Ψ > = x ˆpˆ|Ψ > −ˆ px ˆ|Ψ > d|Ψ > d = −i¯ hx ˆ + i¯h (x|Ψ >) dx dx   d|Ψ > d|Ψ > +x + |Ψ > = i¯ h −x dx dx = i¯ h|Ψ > Resulting in [ˆ x. given 2 commuting observables Aˆ and B sible to construct an orthonormal basis with eigenstates common to both Aˆ ˆ and B.) [ˆ x. Proof: (note that when working with operators its easier to have an imaginary test function to work on. if A and B do not commute ([A.15) [x. x ˆ and pˆ do not commute. Additional commutator math Before we go on we will want to summarize some important commutator relations .17) ˆ it is always posFurthermore. ˆ ˆ ˆ ˆ Conversely. This is the origin of the uncertainty principle.14) ˆ Again this implies that |β > is an eigenfunction of A. you cannot specify the eigenvalues of both simultaneously (only one or the other). B] 6= 0). p] = i¯h (5. To illustrate this locally however. which we will discuss later in more detail.

ˆ B] ˆ + [A. x ˆ]. ˆ C ˆ + D] ˆ = [A. Examples Evaluate [ˆ p. ˆ C] ˆ • [A. ˆ C] ˆ + [A. ˆ B] ˆ Cˆ + B[ ˆ A. ˆ C] ˆ = [A. ˆ C] ˆB ˆ • [AˆB.18) Evaluate [ˆ p2 . h i ˆ [B. ˆ Cˆ B] ˆ • A. Remember to add some dummy wavefunction Ψ to help . ˆ C] ˆ • [A. ˆ C] ˆ + [B. ˆ C] ˆ + [A. ˆ B ˆ + C] ˆ = [A. x ˆx ˆ]Ψ = pˆx ˆx ˆ−x ˆ2 pˆΨ d where pˆ = −i¯h dx d 2 dΨ = −i¯h (ˆ x Ψ) − x ˆ2 (−i¯h) dx dx   dΨ dΨ = −i¯h x ˆ2 + 2ˆ xΨ + i¯hx ˆ2 dx dx   dΨ dΨ − 2ˆ xΨ + x ˆ2 = i¯h −ˆ x2 dx dx = −2i¯hx ˆΨ Therefore we conclude that [ˆ p. ˆ B ˆ C] ˆ = [A. ˆ C] ˆ = A[ ˆ B. QUANTUM MECHANICS REVIEW ˆ A] ˆ =0 • [A. ˆ B] ˆ = −[B. ˆ B] ˆ • [A. Remember to add some dummy wavefunction Ψ to help out.68 CHAPTER 5. x ˆ2 ] = −2i¯hx ˆ (5. ˆ bB] ˆ = b[A. x ˆ2 ]. ˆ D] ˆ • [Aˆ + B. ˆ A] ˆ • [A. [ˆ p. ˆ D] ˆ + [B. ˆ B ˆ C] ˆ − [A.

t) = − ∂t 2m where V is some generic potential. t) = HΨ(r. Alternatively   ¯h2 2 ∂ ∇ + V Ψ(r. t) i¯ h ∂t where H is the Hamiltonian operator.20) . x ˆ] = −2i¯hpˆ (5. (5.69 out.19) The time independent Schr¨ odinger equation The general form of the Schrodinger equation which we will need to solve many times is ∂ Ψ(r. [ˆ p2 . t) i¯ h Ψ(r. x ˆ]Ψ = (ˆ p2 x ˆ−x ˆpˆ2 )Ψ = pˆ2 x ˆΨ − x ˆpˆ2 Ψ   d2 d d ˆ 2 (−i¯h)2 = −i¯ h −i¯h (xΨ) − x dx dx dx   d dΨ d2 Ψ = −¯ h2 x +Ψ +h ¯ 2x 2 dx dx dx   2 d2 Ψ d Ψ dΨ dψ +h ¯ 2x 2 + = −¯ h2 x 2 + dx dx dx dx   2 2 dΨ d Ψ d Ψ +h ¯ 2x 2 = −¯ h2 x 2 + 2 dx dx dx 2 dΨ d2 Ψ d Ψ h2 +h ¯ 2x 2 = −¯ h2 x 2 − 2¯ dx dx dx dΨ = −2¯ h2 dx   d = −2¯ h2 Ψ dx  2  2¯ h d = Ψ (−i¯h) i¯ h dx = −2i¯ hpˆΨ Resulting in [ˆ p2 .

. QUANTUM MECHANICS REVIEW This equation derived by Schrodinger describes the wavefunction of the particle and hence is a pretty important. The general form of the Schrodinger equation has time dependence as you see above. t) Note that the left side has time dependence while the right side only has a spatial dependence. t) = Ψ(r)f (t) = Ψf shorthand expression We now have ∂ ¯h2 ∇2 (Ψf ) = − Ψf + V Ψf ∂t 2m ¯h2 f 2 ∂f =− ∇ Ψ + V Ψf i¯hΨ ∂t 2m   2 2 1 ¯h ∇ i¯ hΨ ∂f = Ψ+VΨ − f ∂t Ψ 2m i¯ h Note that the left hand side is independent of position and the right hand side is independent of time. To distinguish these two equations the former is often referred to as the time-dependent Schrodinger equation. However if can be shown that there is a time independent form of the equation also called the Schrodinger equation. It is only dependent on position V → V (r). t) = ∂t  ¯ 2 ∇2 h − +V 2m  Ψ(r. To obtain the time-independent version of the Schrodinger equation assume that V does not depend on time. Look for solutions of the form Ψ(r. Note that one typically deals with the time-independent version for the first semester of quantum and then get into the time-dependent version in the second semsester. Both sides must equal a constant (call it ) in order to be consistent.70 CHAPTER 5. It turns out that this constant is the system eigenenergy which is why we are calling it  already. We have ∂ i¯ h Ψ(r.

So what does this mean? Well.22) . it means that one has plane waves. it Ψ = Ψ(r)e− h¯ Recall that additional constraints on Ψ(r) include • Ψ(r) is finite (5.21) The form of the function is exponential but it has an imaginary exponent. Summary So putting everything together we have the full form of the wavefunction (both the time-dependent and time-independent parts). Its our favorite eigenvalue and eigenvector problem.71 Left hand side i¯ h ∂f f ∂t 1 ∂f f ∂t 1 ∂f f ∂t lnf =   i¯ h −i = h ¯ it = − h ¯ note that there is a constant of integration = but we ignore it here This results in f =e −it h ¯ (5. Right hand side  2 2  h ∇ Ψ ¯ 1 − +VΨ =  Ψ 2m HΨ = Ψ This expression is something that we are familiar with. Note that eikx = coskx + isinkx.

t) = HΨ(r. Here you will note that typically the first semester of quantum mechanics is spent solving the time independent Schrodinger equation for a number of different situation and boundary conditions. We will solve this equation a bunch of times for a number of model systems. t) ∂t where we just found that Ψ(r. These problems are referred to as model problems and will be described in the next sections. Later on we will go back and deal more directly with the time dependent Schrodinger equation. QUANTUM MECHANICS REVIEW • Ψ(r) is continuous 0 • Ψ (r) is continuous Furthermore. HΨ(r)e −et h ¯ HΨ(r)e −it h ¯ t HΨ(r)e−i h¯ HΨ(r) leaving HΨ(r) = Ψ(r) (5.72 CHAPTER 5. What do we mean What exactly did we mean above. boundary conditions on the wavefunction will induce quantization on the energies  as well as the mathematical form of the wavefunction. Some relevant applications of the time independent Schrodinger equation include .23) which is the time independent Schrodinger equation or our familiar eigenvalue and eigenvector problem. t) = Ψ(r)e− it h ¯ Replace this into the Schrodinger equation i −it ∂ h Ψ(r)e h¯ = i¯h ∂t h i ∂ − it i¯hΨ(r) e h¯ =   ∂t it i i¯ h − Ψ(r)e− h¯ = ¯h   i Ψ(r) = i¯h − ¯h and evaluate. Well let’s go back to the full timedependent Schrodinger equation i¯h ∂ Ψ(r.

73 • particle in a box (absorption) • harmonic oscillator (vibrations) • rigid rotor (rotational spectra) .

74 CHAPTER 5. QUANTUM MECHANICS REVIEW .

a 2D box and a 3D box. a cylindrical wire and a sphere (particle in a spherical box) to illustrate the discreteness as well as progression of states in wells. wires and dots. wires and dots are often described using the analogy to a particle in a 1D box. wire or dot will also play a role in dictating the ordering and spacing of states. This is because when the actual physical length scale of the system is smaller than the exciton Bohr radius or corresponding deBroglie wavelength (as we saw in the previous section). The appearance of discrete states is one of the fundamental signatures of nanomaterials. In turn. Since solving the Schrodinger equation of a carrier to find its eigenvalues and eigenfunctions involves using boundary conditions one can also immediately predict that the actual shape of a quantum well. the energies of the carrier along that dimension of the material are no longer continuous as in the case where there is no confinement. In this chapter we solve the simple analytical problems of a particle in a 1 dimensional rectangular box. 75 . The same applies to quantum wells as well as more exotic shapes of nanostructures. either or both the electron and hole experience confinment. A nanowire will have a similar but different progression of states than a quantum dot (or nanocrystal).Chapter 6 Confinement Quantum wells.

1: Cartoon of a one dimensional infinite barrier potential 1 Dimension of confinement Particle in a 1D infinite box The potential is   ∞ ifx <= 0 V (x) = 0 if0 < x < a   ∞ ifx >= a The boundary conditions are Ψ(0) = 0 Ψ(a) = 0 The Shrodinger equation to solve is 2 2 h d Ψ ¯ − 2m dx2 + V Ψ = εΨ (6.76 CHAPTER 6.2) .1) Rearrange to yield in the box region where V = 0 d2 Ψ + k2Ψ = 0 dx2 (6. CONFINEMENT Figure 6.

77 where k = q 2mε h2 ¯ General solutions are of the form Ψ = Aeikx + Be−ikx (6. q This is readily integrated to give N = 2a and the complete wavefunction as q Ψ(x) = 2a sin(kx) (6. ka = nπ. to give ε= n2 h2 8ma2 (6. Now by normalizing the wavefunction Ψ∗ Ψ = 1 one basically gets the equation Z a sin2 (kx)dx = 1 N2 0 The integral can be evaluated by recalling that sin2 (kx) = 12 (1 − cos(2kx)).5) . This leads to r 2mε a = nπ ¯h2 2 The energy can be solved for. Apply the boundary conditions now to simplify Ψ(0) = A + B → B = −A Ψ(a) = Aeikx − Ae−ikx = 0 This latter equation reduces to 2iA eikx − e−ikx =0 2i 2iAsin(ka) = 0 ikx −ikx ikx +e−ikx and cos(kx) = e where we employ the identities sin(kx) = e −e 2i For this to be true and non-trivial (A = 0).3) Note that the first term describes a right going wave and the second describes a let going wave. Also if you need it eikx = cos(kx) + isin(kx) and e−ikx = cos(kx) − isin(kx). .4) where n is an integer.

That is as the energies get larger. Estimate the first few energies (n = 1. Assume knowledge of the harmonic oscillator wavefunctions and energies. In all cases assume an infinite box model. Use Mathcad. 4.2. Assume the mass is 0.082m0 ). Mathematica or your favorite mathematical modeling program and numerically determine the first 10 energies of the particle in an infinite box (previous problem). 2. 2. Consider the following potential in Figure 6.5m0 ) and “light” hole (mlh = 0. Using your knowledge of the energy level spacing for a particle in an infinite box as well as the energy level spacing in a harmonic oscillator predict whether the energy levels will converge or diverge for the following two potentials.2: Half a harmonic oscillator .067m0 . Draw the corresponding wavefunctions as well. Figure 6. Find the energy levels and eigenfunctions. 3) for an electron in GaAs quantum wells of width 10 nm and 4 nm. 3. Matlab.78 CHAPTER 6. Repeat the same calculation for a “heavy” hole (mhh = . do the spacings between states increase or decrease. CONFINEMENT Exercises 1. (look them up if you need to) This problem does not require any extensive work.

. The potential is The solutions are   V V (x) = 0   V ifx <= 0 if0 < x < a ifx >= a Ψ1 (x) = Aeβx + Be−βx Ψ2 (x) = Ceikx + De−ikx Ψ3 (x) = F eβx + Ge−βx q q 2m(V −ε) 2mε where β = and k = . From quantum mechnanics we know that the solution in the box region where the potential is zero will be wavelike.79 Figure 6.3: Various 1D potentials Particle in a 1D finite box This problem is a little more complicated. In addition F = 0. In the barrier region we also know that the solutions will be exponentially decaying. By finiteness of the wavefunction h2 ¯ h2 ¯ B = 0. Must be well behaved. The wavefunction must not blow up in the barrier region.

C.G).4: Cartoon of a one dimensional finite barrier potential This leaves us with Ψ1 (x) = Aeβx ikx Ψ2 (x) = Ce (6.7) (6.8) Apply the boundary conditions and matching conditions as follows Ψ1 (0) = Ψ2 (0) → A = C + D 0 0 Ψ1 (0) = Ψ2 (0) → Aβ = ikC − ikD Ψ2 (a) = Ψ3 (a) → Ceika + De−ika = Ge−βa 0 0 Ψ2 (a) = Ψ3 (a) → ikCeika − ikDe−ika = −βGe−βa This leads to a system of four equations and four unknowns (A.6) −ikx + De −βx Ψ3 (x) = Ge (6.D.80 CHAPTER 6. Arranged in matrix form it looks like   A 1 −1 −1 0   β −ik ik 0  C  =0  −ika  0 eika e −e−βa  D G 0 ikeika −ike−ika βe−βa  Here either we get the trivial solution where A=C=D=G=0 or that the . CONFINEMENT Figure 6.

.5: Sketch showing the behavior of the wavefunctions in the finite box model determinant of the large matrix is zero.81 Figure 6.

.

.

.

1 −1 −1 0 .

.

.

β −ik ik 0 .

.

.

=0 .

0 eika e−ika −e−βa .

.

.

.

0 ikeika −ike−ika βe−βa .

9) Simplify this determinant. (6. One possible path is shown. This can be done a number of ways. −ik(row 3) + (row 4) → (row4) The determinant becomes .

.

.

1 .

−1 −1 0 .

.

.

0 (β − ik) (β + ik) .

0 .

.

=0 ika −ika −βa .

0 .

e e −e .

.

−ika −βa .

0 .

0 −2ike (β + ik)e Followed by − 1 (row 2) → (row 2) β − ik e−ika (row 3) → (row 3) .

CONFINEMENT yielding .82 CHAPTER 6.

.

.

1 −1 .

−1 0 .

.

.

0 −1 − (β+ik) .

0 .

.

(β−ik) .

.

=0 −2ika −a(β+ik) .

.

0 1 e −e .

0 0 −2ike−ika (β + ik)e−βa

Additional steps

(row 2) + (row 3) → (row 3)

−(row 2) → (row 2)

giving

.

1 −1 −1 .

(β+ik) .

0 1 .

(β−ik) .

.

0 0 e−2ika − (β+ik) (β−ik) .

.

0 0 −2ike−ika .

.

.

.

.

.

=0 −a(β+ik) .

−e .

−βa .

(β + ik)e 0 0 Finding this determinant basically means finding the sub 2x2 determinant .

.

.

−2ika (β+ik) −a(β+ik) .

e − −e .

.

(β−ik) .

.

=0 .

−2ike−ika (β + ik)e−βa .

This is the same as   (β + ik) (β + ik)e−βa − 2ike−ika−a(β+ik) = 0 e−2ika − (β − ik) (β + ik)e−2ika−βa − (β + ik)2 −βa e − 2ike−2ika−βa = 0 (β − ik) Drop the e−βa term to give (β + ik)e−2ika − (β + ik)2 − 2ike−2ika = 0 (β − ik) (β + ik − 2ik)e−2ika = (β + ik)2 (β − ik) (β − ik)2e−2ika = (β + ik)2 .

Exercises 1. Also assume.83 Multiply both sides by eika (β − ik)2e−ika = (β + ik)2 eika (β + ik)2eika − (β − ik)2e−ika = 0 Continue simplifying (β 2 + 2iβk − k 2)eika − (β 2 − 2iβk − k 2 )e−ika = 0 (β 2 − k 2 )(eika − e−ika ) + 2iβk(eika + e−ika ) = 0 2i(β 2 − k 2 )sin(ka) + 4iβkcos(ka) = 0 (β 2 − k 2 )sin(ka) + 2βkcos(ka) = 0 (β 2 − k 2 )tank(ka) + 2βk = 0 Giving our final expression tan(ka) = 2βk (k2 −β 2 ) (6. now rather than an infinite box.067mo .e. Calculate the energy . Increase the barrier to 1 eV. find the roots).10) q q 2m(V −ε) 2mε and k = . Consider the same GaAs quantum wells in the previous exercise (well widths of 4 nm and 10 nm).11) Solve this numerically to get all allowed values of the energy (i. we have a finite box with a 0. Assume the effective mass of the electron is 0. If we replace this into the above where β = h2 ¯ h2 ¯ equation we get the equation tan q  2mε a h2 ¯ = 2 √ ε(V −ε) 2ε−V (6. Calculate the energy of electron states trapped in the box.3 eV barrier.

. let zo = 1 2meF h ¯2 1 3 =  ¯2 h 2meF 1 3 . We then get d2 Ψ 2m − z¯Ψ = − 2 z¯2 ¯h  ¯2 h 2meF  23 Ψ . First multiply through by 2m   2meF z 2m d2 Ψ − Ψ = − 2 Ψ 2 dz 2 ¯h ¯h   2meF 2m d2 Ψ − zΨ = − 2 Ψ dz 2 ¯h2 ¯ h Now let z¯ = z zo .84  CHAPTER 6. Matlab. The Schrodinger equation is   ¯h2 d2 + V (z) Ψ(z) = Ψ(z) − 2m dz 2 The relevant boundary condition for the wavefunction is that it must go to zero at the origin Ψ(0) = 0. We get   2m 1 d2 Ψ 2meF zo z¯Ψ = − 2 Ψ − 2 2 2 zo d¯ z ¯h ¯h   2 2m d Ψ 2meF zo3 z¯Ψ = − 2 zo2 Ψ − 2 2 d¯ z ¯h ¯ h Next. beginning at the origin. . The left side of the potential is infinite. Mathematica etc. Particle in a 1D triangular well Here the potential is V (z) = eF z where e is the fundamental unit of charge and F is the magnitude of the electric field giving rise to such a potential. . CONFINEMENT of electron states trapped in the box (also known as bound states). This can be done using your favorite mathematical modeling software such as Mathcad. To simplify things and converge on a name equation that has already h2 ¯ been solved we do the following.

85 Figure 6.6: Mathcad sheet showing numerical solutions to the particle in a finite box problem .

.86 Let ¯ =  CHAPTER 6.14) which are the desired energies of the particle in a 1D triangular potential. Once this is done we are left with d2 Ψ = sΨ d¯ z2 (6. We are then left with the generic solution for the wavefunctions Ψ = N Ai (s) (6. Let α be the (negative) zeros of the Airy function Ai . To find the energies of the particle. The boundary condition is therefore Ψ(−¯ ) = 0. we apply the one boundary condition that we have Ψ(z = 0) = 0. CONFINEMENT  o leading to  2  32 2m¯ o d2 Ψ ¯h Ψ − z¯Ψ = − 2 z¯2 2meF ¯ h  1  2 1 3 2m 3 o ¯Ψ = − 2 eF ¯h  1 3 2m = − o ¯Ψ (eF ¯h)2 Let o = 1 2m (eF h ¯ )2 1 3 =  (eF ¯ h)2 2m 1 3 giving d2 Ψ − z¯Ψ = −¯ Ψ z¯2 d2 Ψ = (¯ z − ¯)Ψ z¯2 Let s = z¯ − ¯ . Since s = z¯ − ¯  and z¯ = 0 if z = 0 we get s = −¯ . In general Bi will diverge as z → ∞. Therefore ¯ = |α| and recall ¯ = o . Since we can’t have the wavefunction diverge we drop the latter Airy function. Solving for  we get  = αo or (eF ¯h)2 =α 2m   13 (6.12) This is Airy’s equation and has general solution that are linear combinations of Airy functions (Ai and Bi ).13) where N is a normalization constant.

15) .7: Cartoon of the infinite circular potential In the region where the potential is zero − ¯2 2 h ∇ Ψ = εΨ 2m where ∇2 . is   1 ∂ ∂ 1 ∂2 2 ∇ = r + 2 2 r ∂r ∂r r ∂θ (6. θ coordinates.87 2 Dimensions of confinement Paticle in an infinite circular box The Schrodinger equation here is 2 h ¯ − 2m ∇2 Ψ + V Ψ = εΨ The potential is V (x) = ( 0 ifr < a ∞ ifr >= a Figure 6. the Laplacian in r.

00 y =− y 00 0 r2 x rx + + (kr)2 x x ! (6. Assume a form of the wavefunction that is Ψ = x(r)y(θ). dz 2    ∂Ψ 1 ∂2Ψ 1 ∂ r + 2 2 − εΨ = 0 r ∂r ∂r r ∂θ   ∂Ψ 1 ∂ 2 Ψ 2mε 1 ∂ r + 2 2 + 2 Ψ=0 r ∂r ∂r r ∂θ ¯h   2 1 ∂ ∂Ψ 1 ∂ Ψ r + 2 2 + k2Ψ = 0 r ∂r ∂r r ∂θ  Multiply through by r2 to get   ∂Ψ ∂2Ψ ∂ =0 r + (kr)2Ψ + r ∂r ∂r ∂θ2 Note that you have part of this expression depends only on r and the other exclusively on θ. Let’s evaluate the first two terms in the main equation above. CONFINEMENT Note that the Laplacian in cartesian coordinates is ∇2 = Insert this into the main equation to get ¯2 h − 2m d2 dx2 + d2 dy 2 + d2 .16) . As shorthand just denote Ψ by xy. • ∂Ψ ∂r = yx 0  0 • r ∂Ψ ∂r = yrx  00 0 r ∂Ψ ∂r = yrx + yx  0 ∂ 2 00 • r ∂r r ∂Ψ ∂r = yr x + yrx • ∂ ∂r Replace the last item into the above main equation to get 00 0 00 yr2 x + yrx + (kr)2xy + xy = 0 Divide through by xy to get 00 0 00 rx y r2 x + + (kr)2 + =0 x x y Solve for y first.88 CHAPTER 6.

to make this look like something you look up in a book. let z = kr.17) or 00 2 y +m y =0 leading to a general solution of the form y(θ) = Aeimθ + Be−imθ (6.21) which is a linear combination of Jm (z) called Bessel functions of the first kind and Ym (z) called Bessel functions of the second kind.89 let 00 0 r2 x rx + + (kr)2 x x 2 m = ! which you notice is independent of θ and so to y appears as a constant. Alternatively. Equation 13 becomes 0 00 rx r2 x + + (kr)2 − m2 = 0 x x   00 0 x x m2 2 + + k − 2 =0 x rx r leading to 00 x + 0 x r  + x k2 − m2 r2  =0 (6. Note that Bessel functions of the first kind are well behaved at the origin but Bessel functions . Replacing this into the above expression gives 00 y = −m2 y (6.20) Solutions to this equation (for integer m) take the general form x(r) = AJm (z) + BYm (z) (6. This results in Bessel’s equation 2 2 2 z 2 ddzx2 + z dx dz + x(z − m ) = 0 (6.19) which is the normal Bessel equation.18) 00 Now given an explicit form for yy (equation 15) replace this back into equation 14. This is because ultimately what we want is x(r).

third etc. Calculate the energy of the first 3 optical transitions by adding these particle in a circular box energies to the bulk room temperature band gap of InP. Mathematica etc.38mo . Chem. .3˚ A. Since k = eigenenergies that we desire are ε= h2 α2 ¯ 2ma2 2mε h2 ¯ and z = kr. Assume the diameter of the corral is 71.90 CHAPTER 6. Phys. Use your favorite mathematical modeling program such as Mathcad. m is the mass of the particle and α are the roots (first.4mo . visible. B 106. Wang. Assume the effective mass of the electron is me = 0. M. second. x(r) = AJm (z) (6. not the order of the Bessel function.078mo and mh = 0. How does it compare to the paper? 3.) of the Bessel function. Calculate the first 3 electron energy levels of this corral. C. Ignore the length of the wire for simplicity.. . . Gudiksen. IR? How does it compare to the paper? 2.22) Finally. . 15 and 20 nm (see the Lieber paper below). UV. What colors do you expect these wires to emit. Lieber J. Matlab. Consider a quantum corral as described below in the Eigler paper.. CONFINEMENT of the second kind will diverge. Consider an InP nanowire of diameter 10. Jm (ka)q= 0 Therefore ka = α which is the root of the Bessel function. draw the first three wavefunctions of the system.23) where a is the radius of the circle. 4036 (2002). Relevant reading • “Size dependent photoluminescence from single indium phosphide nanowires” M. Therefore. from our other boundary conditions. Assume me = 0. to obtain physically relevant solutions drop the Bessel functions of the second kind and only consider Jm (z). J. Sorry for the notation glitch) Exercises 1. S. (Please note that the m in the denominator is the mass of the particle. For the same quantum corral above. the (6.

1. Nanda. 262. 83. Lutz. E. 3 Dimensions of confinement Particle in an infinite spherical box This is a more complicated problem.25) r2 sin2 (θ) ∂φ2 If replaced into the above equation. E. C.91 • “Confinement of electrons to quantum corrals on a metal surface” M. Eigler Science. C. • “Quantum corrals” M. F. 98 (1995). Fissan Nano Lett. The Schrodinger equation is 2 h ¯ ∇2 Ψ + V Ψ = εΨ − 2m The potential is V (x) = ( 0 ifr < a ∞ ifr >= a In the region inside the sphere where V = 0. Crommie. Lutz.24) where 1 ∂ ∇ = 2 r ∂r 2  ∂ r ∂r 2  1 ∂ + 2 r sin(θ) ∂θ  ∂ sin(θ) ∂θ  + 1 ∂2 (6. M. Two approaches to a solution are illustrated with one leading to what are know as spherical Bessel function and the other to a solution involving regular Bessel functions of half integer order. J. P. H. P. K. M. Heller Physica D. • “Energy levels in embedded semiconductor nanoparticles and nanowires” K. 218 (1993). Crommie. F. this reduces to − ¯2 2 h ∇ Ψ = εΨ 2m (6. D. D. Kruis. 605 (2001). Eigler. multiply by 2mr2 on both sides to simplify giving −¯ h2 r2 ∇2 Ψ = 2mr2 εΨ . F.

CONFINEMENT Figure 6.92 CHAPTER 6.8: Cartoon of the infinite spherical potential .

Basically     ∂ 1 1 ∂ ∂2 2 2 ˆ L = −¯ h sin(θ) − sin(θ) ∂θ ∂θ sin2 (θ) ∂φ2 (6.93 Expanded out this looks like       1 ∂Ψ 1 1 ∂ ∂ ∂2Ψ 2 2 2 ∂Ψ −¯h r r + 2 sin(θ) + 2 2 r2 ∂r ∂r r sin(θ) ∂θ ∂θ r sin (θ) ∂φ2 = 2mr2 εΨ −¯ h 2  ∂ ∂r      ∂ ∂2Ψ 1 ∂Ψ 1 2 ∂Ψ r + sin(θ) + ∂r sin(θ) ∂θ ∂θ sin2 (θ) ∂φ2 = 2mr2 εΨ Rearrange to give   ∂ 2 ∂Ψ r − 2mr2 εΨ − h ¯ ∂r ∂r     ∂ 1 1 ∂ ∂2 2 − ¯ h sin(θ) − Ψ=0 sin(θ) ∂θ ∂θ sin2 (θ) ∂φ2 2 where recall that the third term looks familiar.27) . recall that L ¯ 2 l(l + 1) (angular momentum eigenvalues) leading to   2 ∂ 2 ∂Ψ −¯ h r − 2mr2 εΨ + h ¯ 2 l(l + 1)Ψ = 0 ∂r ∂r Simplify this −¯ h Let k 2 = 2mε h2 ¯ 2    2 ∂Ψ r − Ψ 2mr2 ε − ¯h2 l(l + 1) = 0 ∂r     2mε 2 ∂ 2 ∂Ψ r +Ψ r − l(l + 1) = 0 ∂r ∂r ¯h2 ∂ ∂r giving    ∂ 2 ∂Ψ r + Ψ k 2 r2 − l(l + 1) = 0 ∂r ∂r (6. Our main equation then becomes   2 ∂Ψ 2 ∂ ˆ2Ψ = 0 r − 2mr2 εΨ + L −¯ h ∂r ∂r ˆ2Ψ = h Furthermore.26) (Angular momentum operator).

The other will lead to solutions involving regular or normal Bessel functions of half integer order. φ) The we can evaluate the following items using x and y as shorthand for x(r) and y(θ. The first will lead to solutions involving so called “spherical” Bessel functions. 00 0 r2 x + 2rx + x(k 2r2 − l(l + 1)) = 0 2 dz 2 Let z = kr (therefore dr = dz k and dr = k2 . Bessel functions are proportional to each other so ultimately the solutions are the same.28) This is the general spherical Bessel equation whose solutions take the form: x(r) = Ajl (z) + Byl (z) (6.29) . these will be useful in a moment) and replace into the above expression 00 0 r2 x + 2rx + x(z 2 − l(l + 1)) = 0 d2 x dx r2 2 + 2r + x(z 2 − l(l + 1)) = 0 dr dr dx d2 x + x(z 2 − l(l + 1)) = 0 k 2 r2 2 + 2rk dz dz or 2 2 z 2 ddzx2 + 2z dx dz + x(z − l(l + 1)) = 0 (6.94 CHAPTER 6. Solution with spherical Bessel functions Let Ψ = x(r)y(θ. normal. CONFINEMENT At this point there are two ways to proceed from this equation. Spherical and half integer order. φ) • ∂Ψ ∂r = yx 0 2 • r2 ∂Ψ ∂r = yr x • ∂ ∂r 0  0 2 00 r2 ∂Ψ ∂r = y(r x + 2rx ) Replace the last item into equation 23 giving 00 0 y(r2 x + 2rx ) + xy(k 2r2 − l(l + 1)) = 0 Divide out y and continue simplifying.

one must drop the spherical Bessel functions of the second kind leaving x(r) = Ajl (z) (6. Furthermore. So to get a physical solution. Note that these spherical Bessel functions are denoted by little jn (z) as opposed to big Jn (z) which are normal Bessel functions (same with spherical Bessel functions of the second kind).95 where jl (z) are spherical Bessel functions of the first kind and yl (z) are spherical Bessel functions of the second kind (also known as spherical Neumann functions written as nl (z)). Note that jl (z) is finite and well behaved at the origin while yl (z) diverges. a subclass of normal Bessel functions as shown at the end of this chapter. φ) = . Solution with half integer (normal) Bessel functions Alternatively assume that the wavefunction has the form ψ = R(r)y(θ. φ) Replace this in equation 24 and start simplifying. spherical bessel functions are related to half integer Bessel functions. Examples of the spherical Bessel function solutions are • j0 (z) = sin(z) z • j1 (z) = sin(z) z2 − cos(z) z cos(z) • j2 (z) = 3 sin(z) z3 − 3 z2 − sin(z) z and so forth. pg 437.30) The equation and its solutions can be looked up in a text like “Handbook of Mathematical Functions” Abramowitz and Stegun. ∂ ∂r   2 ∂Ψ r + Ψ(k 2 r2 − l(l + 1)) = 0 ∂r We need the following bits of information  y 0 1 • ∂Ψ ∂r = yx − r2 + r x • r2 ∂Ψ ∂r = −xy + ryx • ∂ ∂r 0  0 00 0 00 r2 ∂Ψ ∂r = −x y + x ry + x y = x ry x(r) r y(θ.

96 CHAPTER 6. CONFINEMENT Replace the last item into our main equation 00 xy 2 2 x ry + (k r − l(l + 1)) = 0 r Get rid of y x 2 2 (k r − l(l + 1)) = 0 r   l(l + 1) 00 2 x +x k − =0 r2  √ Now let z = kr and x = zφ(z) dr = dz giving k   l(l + 1) d2 x 2 +x k − =0 dr2 r2   2 √ k 2l(l + 1) 2d x 2 k + zφ k − =0 dz 2 z2 00 rx + Use the following bits of information √ 0 1 zφ + φ 21 z − 2 • dx dz =    3 √ 00 1 1 0 2 0 • ddzx2 = zφ + φ 21 z − 2 + 12 φ − 12 z − 2 + z − 2 φ • d2 x dz 2 = √ 1 3 zφ + φ z − 2 − 41 φz − 2 00 0 especially the last term to get     √ φ −3 00 0 −1 k 2l(l + 1) 2 √ 2 2 2 k =0 zφ + φ z − z + zφ k − 4 z2 Drop k 2 and continue simplifying   √ 00 1 φ 3 √ l(l + 1) 0 =0 zφ + φ z − 2 − z − 2 + zφ 1 − 4 z2   00 0 φ l(l + 1) zφ + φ − + zφ 1 − =0 4z z2   l(l + 1) φ 00 0 =0 z 2 φ + zφ − + z 2 φ 1 − 4 z2   0 1 2 00 2 z φ + zφ + φ − + z − l(l + 1) = 0 4    1 0 2 00 2 + l(l + 1) =0 z φ + zφ + φ z − 4 .

R(r).34) Summary So in the end.36) 2 . since x(r) = zφ(z) and R(r) = x(r) r the general radial solution. Furthermore. regardless of how you came to the solution. is φ(z) = AJn (z) + BJn (z)  √  0 0 z A Jn (z) + B Yn (z) x(r) = R(r) = or more explicitly x(r) = √1 z  0  0 A Jn (z) + B Yn (z)  √  z AJ(l+ 1 ) (z) + BY(l+ 1 ) (z) 2 2 (6. to find the eignevalues we need to find the roots.35) J(l+ 1 ) (kr) = 0 half integer Bessel (6. of either jl (kr) = 0 spherical Bessel (6.33) We can look up the equation and its solution in a book like Abramowitz and Stegun but since Yn diverges near the origin so it should be dropped immediately.32) (6. This leaves R(r) = 0 A √ (z) J z (l+ 21 ) (6. In this case where we have halfinteger orders n then general solutions for φ are linear combinations of half √ integer Bessel functions. α.97 Note that the term in the inner parenthesis is equal to l + fies the full expression to   ! 1 2 0 2 00 2 z φ + zφ + φ z − l + =0 2 Let n = l + 1 2  00 0 z 2 φ + zφ + φ(z 2 − n2 ) = 0  1 2 2 This simpli- (6.31) This is the normal Bessel function equation.

Note that for each value of l there are an infinite number of roots.37) Finally. Calculate the eigenenergies of a free electron (mass mo ) in a 5 nm diameter sphere for l = 0. CONFINEMENT So on finding the root one also has α = kr and from this the desired value of k. 3 using the lowest root of the Bessel function. Summarize all these energies in ascending order of energy in one table. 2. Exercises 1. The eigenvalues are then r ka = α 2mε a = α ¯h2 ¯h2 α2 ε = 2ma2 leading to ε= ¯ 2k2 h 2m ε= h2 α2 ¯ 2ma2 (6. 1. 3. 1. the relation between spherical Bessel functions and half integer Bessel functions are: r π (6. Its your choice on what’s most convenient. . 3 using the third lowest root of the Bessel function.39) Y 1 (z) 2z (l+ 2 ) So either approach to solving the problem is valid.98 CHAPTER 6. Calculate the eigenenergies of a free electron (mass mo ) in a 5 nm diameter sphere for l = 0. 3 using the second lowest root of the Bessel function. 4. 2.38) J 1 (z) jl (z) = 2z (l+ 2 ) r π yl = nl (z) = (6. 1. 2. 2. Calculate the eigenenergies of a free electron (mass mo ) in a 5 nm diameter sphere for l = 0.

Mattoussi. A. 517 . Now Bashir (Dabbousi) has been good enough to plot the radial wavefunction of both the electron and hole in this system. J. Mikulec. Kim. Often enough there is a Coulomb attraction between the electron and hole in a confined system. As a consequence the energies one calculates via particle in a box type expression are often corrected for this Coulomb attraction. Use Mathcad. J. 125. Hint: think of how these energies all scale with r or a. Basically ask yourself how all these energies scale with r or a. Reproduce the first (CdSe/matrix) and second (CdSe/ZnS) graphs. 9463. Schooss. H. al. Mathematica or your favorite mathematical modeling program to draw the radial wavefunctions corresponding to the lowest three energies in your table. an infinite circular potential (wire) and a particle in an infinite spherical box (dot). Soc. 9. Dabbousi. Calculate the first 7 energy levels of a free electron in a 5 nm length three dimensional box. 8. Phys. Read the following papers with particular emphasis on the first one. V. 11466. R. O. (2) “Quantum-dot quantum well CdS/HgS/CdS: Theory and experiment” D. J. M. Heine. You too can do this. (All sides of the cube have length of 5 nm). Eisler. 6. Fisher. K. Weller Phys. Don’t forget to normalize the wavefunction. R. Am. B 1994. 101. Chem. Bawendi J. G. F. Matlab. Any similarities? Now take a look a Yu et.520 (01 Aug 2003) Letters and comment on it. Rev. Core shell semiconductor particles are pretty popular these days since they were found to improve the quantum yield of many semiconductor systems. B 1997. Rodriguez-Viejo. Jensen. Compare these results to the particle in a sphere case. See figure 14. 2003. Eychmuller. Chem. 49. A. B. Qualitatively compare the size dependence of the energies for a particle in a 1D box (well). Nature Materials 2. M. Ober. 7. .99 5.. 17072. (1) (CdSe)ZnS core-shell quantum dots: Synthesis and characterizatio of a size series of highly luminescent nanocrystallites” B. H. F. Think back to the basic expression for the Coulomb energy and qualitatively explain why this term is important or not important. (3)“Type-II quantum dots: CdTe/CdSe (core/shell) and CdSe/ZnTe (core/shell) heterostructures” S. Bawendi J. Mews. H.

θ. We get   2  ∂ ∂ ∂2 2 ∂ r r + 2 +r Ψ = −κ2 r2 Ψ ∂r ∂r ∂θ ∂z 2   ∂Ψ ∂2Ψ ∂ ∂2Ψ r + r 2 2 + κ2 r 2 Ψ + r ∂r ∂r ∂z ∂θ2  Now assume a solution of the form Ψ(r. Replacing this into the Schrodinger equation we get     h2 1 ∂ ¯ ∂2 ∂ 1 ∂2 − r + 2 2 + 2 Ψ = Ψ 2m r ∂r ∂r r ∂θ ∂z giving Multiply both sides by − 2m h2 ¯  Let k 2 = 2m h2 ¯ 1 ∂ r ∂r    ∂ 1 ∂2 ∂2 2m r + 2 2 + 2 Ψ=− 2 Ψ ∂r r ∂θ ∂z ¯h giving  1 ∂ r ∂r  ∂ r ∂r   1 ∂2 2 2 + 2 2 + ∂ ∂z Ψ = −κ2 Ψ r ∂θ Multiply by r2 to get rid of the middle 1 r2 term. CONFINEMENT Particle in a cylinder Here in cylindrical coordinates   1 ∂ ∂ 1 ∂2 ∂2 2 ∇ = r + 2 2+ 2 r ∂r ∂r r ∂θ ∂z Solve the Schrodinger equation HΨ = Ψ where H=− ¯2 2 h ∇ +V 2m where V = 0 inside the cylinder. Elsewhere the potential is infinite. z) = A(r.40) . θ)B(z) ≡ AB (6. The effective mass of the electron and the hole are used.100 CHAPTER 6.

Let’s call it m2 . θ or z. Begin with − 1 ∂2B B ∂z 2 d2 B dz 2 = m2 = −m2 B leading to d2 B + m2 B = 0 dz 2 (6. θ) (although offset by κ2 ).43) A general solution to this differential equation is B = Ao eimz + Bo e−imz (6. The RHS is a function of z.44) .42) − B ∂z 2 Let’s solve the easy equation first. If these expressions are equivalent then the must both equal a constant with no dependence of r. We then have   ∂ ∂AB ∂ 2 AB ∂ 2 AB 2 r r + r2 + (κr) AB + = 0 ∂r ∂r ∂z 2 ∂θ2   ∂2A ∂A ∂2B ∂ r + Ar2 2 + (kr)2AB + B 2 = 0 Br ∂r ∂r ∂z ∂θ Divide through by AB to get   r ∂ ∂A r2 ∂ 2 B 1 ∂2A 2 r + + (kr) + =0 A ∂r ∂r B ∂z 2 A ∂θ2 Divide through by r2 to get   ∂A 1 ∂2B 1 ∂2A 1 ∂ 2 r + + κ + =0 rA ∂r ∂r B ∂z 2 r2 A ∂θ2 Consolidate terms to get 1 ∂A 1 ∂2A 1 ∂2B 2 + 2 + κ = − rA ∂r r A ∂θ2 B ∂z 2 The LHS is a function of (r.41) rA ∂r ∂r r A ∂θ2 1 ∂2B = m2 (6. Therefore we get   1 ∂ ∂A 1 ∂2A r + 2 + κ2 = m2 (6.101 through the method of separation of variables.

CONFINEMENT where Ao and Bo are constants. This is then the longitudinal part of the wavefunction.47) therefore we either get the trivial solution (Ao = 0) or we get sinmL = 0. The latter solution requires that mL = nπ nπ m = L where n = 0.48) where No is a normalization constant. Applying the first boundary condition we get B(0) = Ao + Bo = 0 therefore Bo = −Ao Applying the second boundary condition we get B(L) = Ao eimL − Ao e−imL = 0 or 2iAo   eimL − e−imL = 0 2i 2iAo sinmL = 0 (6. 2. 3 . Begin with   1 d dA 1 d2 A r + 2 + κ2 = m2 rA dr dr r A dθ2 . The boundary conditions here require that B(0) = 0 (6.46) where L is the length of the rod or the wire. Now we go and find the harder solutions.102 CHAPTER 6. . . Putting everything together we get B(z) = No sin( nπ z) L (6. 1.45) B(L) = 0 (6.

50) C ∂r ∂r 1 ∂2D = p2 (6. θ) ≡ C(r)D(θ) again by the separation of variables approach. Begin with − 1 d2 D D dθ2 d2 D dθ2 = p2 = −Dp2 .103 Multiply through by r2 to get   r d dA 1 d2 A r + + (kr)2 = (mr)2 A dr dr A dθ2 Now multiply through by A    dA d2 A d r + 2 + (κr)2 − (mr)2 A = 0 r dr dr dθ Assume A(r. This gives us the following two equations     r ∂ ∂C r + (κr)2 − (mr)2 = p2 (6.49) Note that the left side depends only on r and the right side depends only on θ.51) − D ∂θ2 Let’s start by solving the easy equation. A(r. Both sides therefore equal a constant. θ) = C(r)D(θ) ≡ CD giving   ∂CD ∂ 2 CD  r + + (κr)2 − (mr)2 CD = 0 2 ∂r ∂θ    ∂C ∂2D  ∂ r + C 2 + (κr)2 − (mr)2 CD = 0 Dr ∂r ∂r ∂θ ∂ r ∂r  Divide through by CD    r ∂ ∂C 1 ∂2D  2 2 + (κr) − (mr) =0 r + C ∂r ∂r D ∂θ2 Consolidate common terms     ∂C 1 ∂2D r ∂ r + (κr)2 − (mr)2 = − C ∂r ∂r D ∂θ2 (6. We will call it p2 .

104 CHAPTER 6. 1. 3. . . Now we go and solve the harder equation. . CONFINEMENT giving d2 D + p2 D = 0 dθ2 (6.52) The general solutions to this differential equation are of the form D = Co eipθ + Do e−ipθ (6.53) where Co and Do are constants. 2. By continuity we also know that D(θ) = D(θ + 2π) giving D(0) = Co + Do D(2π) = Co eip(2π) + Do e−ip(2π) by inspection p is an integer and p = 0. Begin with     r d dC r + (κr)2 − (mr)2 = p2 C dr dr Multiply through by C     dC d r + (κr)2 − (mr)2 c r dr dr  2    d C dC r r 2 + + (κr)2 − (mr)2 C dr dr 2  d C dC  r2 2 + r + (κr)2 − (mr)2 C dr dr 2C   d dC r2 2 + r + (κr)2 − (mr) − p2 C dr dr = p2 C = p2 C = p2 C = 0 Divide through by r2   1 dC p2 d2 C 2 2 + + (κ − m ) − 2 C = 0 dr2 r dr r Let a2 = (κ2 − m2 )   d2 C 1 dC p2 2 + + a − 2 C=0 dr2 r dr r .

z) = n Jp (ar) Coipθ + Do e−ipθ sin( z) L (6. This leaves us with C(ξ) = AJp (ξ) 2m . 2.55) This is the wavefunction of this part of the problem. Now we summarize everything here C(r) = AJp (ar) D(θ) = Co eipθ + Do e−ipθ nπ B(z) = No sin( z) L where p = 0. C = AJp (ξ) + BYp (ξ) √ where A. . . 2. Putting it all together we have the total wavefunction as the product of the individual terms Ψ = C(r)D(θ)B(z) giving h  i nπ Ψ(r.54) This is Bessel’s equation whose solutions are Bessel functions.105 Multiply through by r2 again r2  dC dC  +r + (ar)2 − p2 C = 0 dr dr Let ξ = ar or dξ = adr or dr = a2 r2 dξ a . and n = 0. The desired energies come from the roots of the Bessel function Jp (ar) = 0 (6. . θ. . 3. This gives dC dC + ar + [ξ 2 − p2 ]C = 0 dξ dξ or ξ2 dC dC +ξ + [ξ 2 − p2 ]C = 0 dξ dξ (6. B are constants and ξ = ar where a = κ2 − m2 and κ2 = Because Yn (ξ) → ∞ as r → ∞ we have B = 0. . 1. . 3. . 1.56) This is our desired wavefunction.57) . h2 ¯ (6.

106 CHAPTER 6.58) 2m R L . α will be the root of the Bessel function and α = aao where a = κ2 − m2 . ao is the radius of the cylinder. CONFINEMENT where r = ao is the radius of the√cylinder. and α are the roots of the Bessel function. Then we have p α = κ2 − m2 ao r 2m − m2 ao = ¯h2 r 2m α = 2 ao ¯h 2 2m α 2 − m = a2o ¯h2 2m α2 + m2 = a2o ¯h2 giving ¯2 h = 2m  α2 + m2 a2o  where m = nπ L . If ao = R the the final desired energies of the particle in a cylinder are   ¯h2  α 2  nπ 2 = + (6.

. . . can be included if desired.  = (0) + λ(1) + λ2 (2) + . In the end we will usually set λ = 1 but the method is valid for any fraction of λ. . . Our wavefunctions and energies are Ψ = Ψ(0) + λΨ(1) + λ2 Ψ(2) + . • We introduce a parameter λ to keep track of the order or alternatively.Chapter 7 Nondegenerate perturbation theory ˆ is close to an approximate The basic idea here is that the true Hamiltonian H (0) ˆ Hamiltonian H whose eigenfunctions and eigenvalues are known. Usually though you will stop at H the first or second correction to either the energies or wavefunctions. ˆ (1) have a λ in front of it. . The ˆ (1) is called a perturbation. • So 1st order corrections from H ˆ (2) will have λ2 in front of it • 2nd order corrections from H • and so on in this fashion • Also λ does not represent any real physical quantity. the order of refinement of the solutions. H ˆ (3) . Our corrected Hamiltonian is therefore ˆ =H ˆ (0) + λH ˆ (1) + λ2 H ˆ (2) + . It is simply a mathematical technique for keeping track of the order of approximation. 107 . . . Higher order corrections such as difference H ˆ (2) . H ˆ (0) is our unperturbed Hamiltonian whose eigenvalues and eigenwhere H vectors we know.

In addition. drop λ3 and higher terms). We will separately evaluate the left and right hand side of these expressions.e. Left hand side H (0) Ψ(0) + λH (0) Ψ(1) + λ2 H (0) Ψ(2) + λ3 H (1) Ψ(2) + λ4 H (2) Ψ(2) + λH (1) Ψ(0) + λ2 H (1) Ψ(1) + λ3 H (2) Ψ(1) + λ2 H (2) Ψ(0) When we consider terms only up to second order we get H (0) Ψ(0) + λ[H (0) Ψ(1) + H (1) Ψ(0) ] + λ2 [H (0) Ψ(2) + H (1) Ψ(1) + H (2) Ψ(0) ] Right hand side (0) Ψ(0) + λ(0) Ψ(1) + λ2 (0) Ψ(2) + λ3 (1) Ψ(2) + λ4 (2) Ψ(2) + λ(1) Ψ(0) + λ2 (1) Ψ(1) + λ3 (2) Ψ(1) + λ2 (2) Ψ(0) When we consider terms only up to second order we get (0)Ψ(0) + λ[(0) Ψ(1) + (1) Ψ(0) ] + λ2 [(0) Ψ(2) + (1)Ψ(1) + (2) Ψ(0) ] Now because the solution must be true for all values of λ one can equate the coefficients of power of λ giving H (0) Ψ(0) = (0) Ψ(0) H (0) Ψ(1) + H (1) Ψ(0) = (0)Ψ(1) + (1) Ψ(0) H (0) Ψ(2) + H (1) Ψ(1) + H (2) Ψ(0) = (0) Ψ(2) + (1)Ψ(1) + (2) Ψ(0) . NONDEGENERATE PERTURBATION THEORY The results of perturbation theory are derived by taking the exact Schrodinger equation HΨ = Ψ and substituting in our above expressions for Ψ and . we will consider terms only up to second order (i.108 CHAPTER 7. Next we group all terms with the same power of λ. [H (0) + λH (1) + λ2 H (2) ][Ψ(0) + λΨ(1) + λ2 Ψ(2) ] = [(0) + λ(1) + λ2 Ψ(2) ][Ψ(0) + λΨ(1) + λ2 Ψ(2) ] Ok things get complicated here.

< Ψ(0) |H (0) − (0) |Ψ(1) > = 0 < Ψ(0) |H (0) − (0) |Ψ(2) > = 0 this is because H (0) is Hermitian H (0) = H †(0) . Alternatively.1) (H (0) − (0) )Ψ(1) + (H (1) − (1) )Ψ(0) = 0 (7. We get < Ψ(0) |H (1) |Ψ(0) > −(1) < Ψ(0) |Ψ(0) > = 0 < Ψ(0) |H (1) |Ψ(1) > −(1) < Ψ(0) |Ψ(1) > + < Ψ(0) |H (2) |Ψ(0) > −(2) < Ψ(0) |Ψ(0) > = 0 reducing to (0) <Ψ |H (1) |Ψ(0) > −(1) = 0 < Ψ(0) |H (1) |Ψ(1) > + < Ψ(0) |H (2) |Ψ(0) > −(2) = 0 finally simplifying to < Ψ(0) |H (1) |Ψ(0) > = (1) < Ψ(0) |H (1) |Ψ(1) > + < Ψ(0) |H (2) |Ψ(0) > = (2) . This leaves < Ψ(0) |H (1) − (1)|Ψ(0) > = 0 1st order < Ψ(0) |H (1) − (1)|Ψ(1) > + < Ψ(0) |H (2) − (2)|Ψ(0) > = 0 2nd order Next let’s break each expression up and drop any zero terms. It provides our zero-order solutions.2) (H (0) − (0) )Ψ(2) + (H (1) − (1) )Ψ(1) + (H (2) − (2) )Ψ(0) = 0 (7. Now multiply all expression by Ψ∗(0) and integrate.109 Let’s rearrange these expressions to get (H (0) − (0) )Ψ(0) = 0 (7. you could also invoke the orthogonality of < Ψ(0) | and |Ψ(1) > after operating on the bra. < Ψ(0) |H (0) − (0)|Ψ(0) > = 0 < Ψ(0) |H (0) − (0)|Ψ(1) > + < Ψ(0) |H (1) − (1)|Ψ(0) > = 0 < Ψ(0) |H (0) − (0)|Ψ(2) > + < Ψ(0) |H (1) − (1)|Ψ(1) > + < Ψ(0) |H (2) − (2)|Ψ(0) > = 0 Now in the case of the first order and second order expressions the first term in each case is zero.3) The first of the three equations is our unperturbed case. Nothing special here. Also let’s switch to bra-ket notation so I don’t have to write out the integrals. So you can operate with H (0) on the bra and see that each expression will die.

Usually you stop at first order corrections.  = (0) + λ(1) + λ2 (2) + λ3 (3) + . NONDEGENERATE PERTURBATION THEORY Therefore for the energies we have the following first and second order corrections to the zero order solutions. .4) (2) =< Ψ(0) |H (1) |Ψ(1) > + < Ψ(0) |H (2) |Ψ(2) > (7.110 CHAPTER 7. Left hand side H (0) Ψ(0) + λH (0) Ψ(1) + λ2 H (0) Ψ(2) + λ3 H (0) Ψ(3) + λ4 H (1) Ψ(3) + λ5 H (2) Ψ(3) + λ6 H (3) Ψ(3) + λH (1) Ψ(0) + λ2 H (1) Ψ(1) + λ3 H (1) Ψ(2) + λ4 H (2) Ψ(2) + λ5 H (3) Ψ(2) + λ2 H (2) Ψ(0) + λ3 H (2) Ψ(1) + λ4 H (3) Ψ(1) + λ3 H (3) Ψ(0) We want the third order corrections here. (1) =< Ψ(0) |H (1) |Ψ(0) > (7. . . We will revisit this later after finding out what |Ψ(1) > looks like. . Example 1. . The same applies to higher order corrections and means that this can get pretty tedious. Plugging this into the Schrodinger equation we get [H (0) + λH (1) + λ2 H (2) + λ3 H (3) ][Ψ(0) + λΨ(1) + λ2 Ψ(2) + λ3 Ψ(3) ] = [(0) + λ(1) + λ2 (2) + λ3 (3) ][Ψ(0) + λΨ(1) + λ2 Ψ(2) + λ3 Ψ(3) ] Again. So right now our second order expression is not necessarily in a useful form. since this gets a little involved we will consider the left and right hand side of this equation separately. expression for 3 Please calculate the expression for the third order correction 3 H = H (0) + λH (1) + λ2 H (2) + λ3 H (3) + . . Ψ = Ψ(0) + λΨ(1) + λ2 Ψ(2) + λ3 Ψ(3) + .5) Note that to get the second order energy correction you already need to know what |Ψ(1) > is. .

Switching to the Dirac bra-ket notation we get (H (0) − (0) )|Ψ(3) > +(H (1) − (1) )|Ψ(2) > +(H (2) − (2))|Ψ(1) > +(H (3) − (3) )|Ψ(0) >= 0 < Ψ(0) |H (0) − (0) |Ψ(3) > + < Ψ(0) |H (1) − (1)|Ψ(2) > (3) + < Ψ(0) |H (2) − (2) |Ψ(1) > + < Ψ(0) |H (3)− |Ψ(0) >= 0 the first term drops out leaving < Ψ(0) |H (1) − (1) |Ψ(2) > + < Ψ(0) |H (2) − (2)|Ψ(1) > + < Ψ(0) |H (3) − (3) |Ψ(0) >= 0 becoming < Ψ(0) |H (1) |Ψ(2) > −(1) < Ψ(0) |Ψ(2) > + < Ψ(0) |H (2) |Ψ(1) > −(2) < Ψ(0) |Ψ(1) > + < Ψ(0) |H (3) |Ψ(0) > −(3) < Ψ(0) |Ψ(0) >= 0 by orthonormality a bunch of terms drop out giving < Ψ(0) |H (1) |Ψ(2) > + < Ψ(0) |H (2) |Ψ(1) > + < Ψ(0) |H (3) |Ψ(0) >= (3) Therefore our desired 3rd order correction is (3) =< Ψ(0) |H (1) |Ψ(2) > + < Ψ(0) |H (2) |Ψ(1) > + < Ψ(0) |H (3) |Ψ(0) > (7. This leaves H (0) Ψ(3) + H (1) Ψ(2) + H (2) Ψ(1) + H (3) Ψ(0) = (0) Ψ(3) + (1)Ψ(2) + (2) Ψ(1) + (3)Ψ(0) Multiply both sides by Ψ∗(0) and integrate.6) .111 Right hand side (0)Ψ(0) + λ(0) Ψ(1) + λ2 (0) Ψ(2) + λ3 (0)Ψ(3) + λ4 (1)Ψ(3) + λ5 (2) Ψ(3) + λ6 (3) Ψ(3) + λ(1) Ψ(0) + λ2 (1) Ψ(1) + λ3 (1)Ψ(2) + λ4 (2)Ψ(2) + λ5 (3) Ψ(2) + λ2 (2) Ψ(0) + λ3 (2)Ψ(1) + λ4 (3)Ψ(1) + λ3 (3) Ψ(0) We only want the 3rd order corrections here.

du = dx. We get   Z a V x 1 V a − 2 sin2bx|0 − sin2bxdx = 2 a 2b 0 2b   V V x 1 1 a a = − 2 sin2bx|0 + cos2bx|0 2 a 2b 2b 2b the first term is zero  V V 1 = − 2 cos2bx|a0 2 a 4b2 the term in brackets is also zero . Particle in a slanted box We have a particle in a 1D box with a slanted bottom. NONDEGENERATE PERTURBATION THEORY Example 2. Let u = x. The zeroth order wavefunction q (0) = n2 h2 . The perturbation term is H (1) = Va x where V is some constant. v = 1 2b sin2bx. is Ψ(0) = 2a sin nπx a and the zeroth orther energies are  8ma2 What is the 1st order correction to (0) ? From nondegenerate perturbation theory the first order correction is (1) = < ψ(0) |H (1) |Ψ(0) > ! !  r Z a r nπx nπx V 2 2 dx = sin x sin a a a a a 0  Z a 2 V nπx = dx xsin2 a a a 0 nπ let b = Z a a 2V = xsin2 bxdx a2 0 1 where sin2 bx = (1 − cos2bx) 2   Z 2V 1 a x(1 − cos2bx)dx = a2 2 0 Z a  Z a V = xdx − xcos2bxdx a2 0 0   Z a V x2 a | − xcos2bxdx = a2 2 0 0   Z a V a2 = − xcos2bxdx a2 2 0 Solve the integral by parts. dv = cos2bxdx.112 CHAPTER 7.

The final corrected energies are then = n2 h2 8ma2 + V 2 (7. We have 1  = (n + )hν 2 1 Ψ = Nn Hn (α 2 x)e− αx2 2 q and the normalization constant is Nn = where α = kµ h2 ¯ The first order energy correction is √1 2n n! α π 1 4 (1) =< Ψ(0) |H (1) |Ψ(0) > where H (1) = 61 γx3 .113 therefore (1) = V 2 (7. An anharmonic oscillator Consider an anharmonic oscillator whose potential is U (x) = 12 kx2 + 16 γx3 . .7) This is our desired 1st order correction. Example 3. Now by symmetry their product is overall odd and over the range of the integral it evaluates to zero. We have     Z αx2 αx2 1 1 γ ∞ − − 3 = Nn Hn (α 2 x)e 2 (x ) Nn Hn (α 2 x)e 2 dx 6 −∞ Z 1 1 γ 2 ∞ 2 [Hn (α 2 x)][x3][Hn (α 2 x)]e−αx dx Nn = 6 −∞ 1 Now if n = 0 H0 (α 2 x) = 1 leaving Z ∞ γ 2 x3 e−αx dx = Nn2 6 −∞ The first term in the integral is odd. Therefore (1) = 0 (7.8) where n = 1. Note that this problem does not require a lot of math. The second is even. Calculate the 1st order correction to the ground state energy.9) . 3. 2. . .

So to first order the overall energies are  = (n + 12 )hν (7. Another anharmonic oscillator Calculate the 1st order correction to the ground state energy of an anharmonic oscillator whose potential energy is 1 2 1 3 1 kx + γx + bx4 2 6 24 1 1 3 γx + bx4 H (1) = 6 24 1  = (n + )hν 2 U (x) = 1 Ψ = Nn Hn (α 2 x)e− αx2 2 q kµ where α = and the normalization constant is N = h2 ¯ 1  N0 = απ 4 if n = 0. NONDEGENERATE PERTURBATION THEORY This is our desired first order correction. By symmetry it drops out leaving.10) Example 4. Z 1 1 2 b 2 ∞ N0 (H0 (α 2 x))(x4)(H0 (α 2 x))e−αx dx = 24 −∞ 1 where H0 (α 2 x) = 1 Z b 2 ∞ 4 −αx2 x e dx = N 24 0 −∞ this integral does not dissapear by symmetry and must be evaluated Z bN02 ∞ 4 −αx2 = x e dx 12 −∞ looking this up in a table of integrals where R∞ 0 2 x2ne−αx dx = 1·3·5·(2n−1) p π α 2n+1 αn . The first order energy correction is √1 2n n! α π 1 4 and (1) = < Ψ(0) |H (1) |Ψ(0) >    Z ∞ 2 2 1 1 1 4 1 3 − αx − αx 2 2 2 2 γx + bx N0 H0 (α x)e N0 H0 (α x)e dx = 6 24 −∞ We have already done the 1st term.114 CHAPTER 7.

We then have (1) Ψi = X j cij |j > and where cij are weights representing how much of each unperturbed wavefunction is present..11) Wavefunctions Now that we have the energy corrections.115 Our integral evaluates to 3 8α2 pπ The expression becomes  r bN02 π 3 12 8α2 α r bN02 π 32α2 α r α 2 where N0 = giving π b 32α2 kµ where α2 = 2 giving ¯h b¯ h2 32kµ = = = = α. A convenient way to express Ψ(1) is as a linear expansion in unperturbed (0) wavefunctions which we will denote as Ψj or |j >.. Note that this is also called the Raleigh Ritz method. The overall energy with the first order correction is therefore 1 b¯h2  = (n + )hν + 2 32kµ = hν 2 + b¯ h2 32kµ (7. We know that we have corrections like (0) Ψtot = Ψi (1) + λΨi (2) + λ2 Ψi +. what about the wavefunctions? Note that you will usually go to second order in the energies but only 1st order in the wavefunctions. (0) We have Ψi which is our unperturbed wavefunction. .

(0) k X k (0) cik+ < k|H (1)|i > = i X k Consolidate our desired cik terms to get X k (0) (0) cik (k − i ) = − < k|H (1) |i > cik . X j (0) j (0) (0) cij j |j > +H (1) |i > = i X j cij |j > +H (1) |i > = (0) i X j X j (1) cij |j > +i |i > (1) cij |j > +i |i > Next to start finding our desired coefficients cij multiply both sides by < k| which is a member of the unperturbed basis. NONDEGENERATE PERTURBATION THEORY If we go back to our expression where we derived the 1st order corrections in energy we have (0) (1) (1) (0) (H (0) − i )Ψi + (H (1) − i )Ψi (1) H (0) Ψi (1) Now recall that Ψi H (0) X j X j = P j cij |j = 0 (0) + H (1) Ψi (0) (1) (1) (0) = i Ψi + i Ψi > and replace into the above expression (0) cij |j > +H (1) |i > = i cij H (0) |j > +H (1) |i > = X j (0) i X j (1) cij |j > +i |i > (1) cij |j > +i |i > Recall that |j > are eigenfunctions of H (0) since we expanded in the unperturbed basis. Note that k 6= i.116 CHAPTER 7. (0) j X j (0) j X j (0) cij < k|j > + < k|H (1) |i > = i (0) cij < k|j > + < k|H (1) |i > = i X (1) cij < k|j > +i < k|i > j X cij < k|j > j Now due to the orthogonality between |k > and |j > the only terms in the above expression that survive are those where k = j.

Now that we have an explicit form we can substitute it in. Also recall that Ψ(1) = so you can see that the corrected wavefunction to first order is |Ψ >= |i > + <k|H (1) |i> k ((0) −(0) ) |k i k P P j cij |j (7.13) > The correction is <k|H (1) |i> k ((0) −(0) ) |k i k Ψ(1) = P > (7. Recall that (0) (1) (2) =< Ψi |H (1) |Ψi (0) (0) > + < Ψi |H (2) |Ψi > (1) where we didn’t know |Ψi > before.15) This is a more user friendly 2nd order energy correction.12) (0) (i −k ) These are our desired correction coefficients. > . We get (0) (1) (2) = < Ψi |H (1) |Ψi > X < k|H (1) |i >< i|H (1) |k > = (0) (0) (i − k ) k The final expression is (2) = P k <k|H (1) |i><i|H (1) |k> (0) (0) (i −k ) (7. Also note that i 6= k.117 Therefore cik = − < k|H (1)|i > (0) (0) (k − i ) cik = <k|H (1) |i> (0) (7. thus H (2) = 0.14) Now that we have the 1st order correction to the wavefunction. To make life even easier assume that the total perturbation goes only to first order. we can go back and express the 2nd order correction in energy in a more useful form. This is called non-degenerate perturbation theory.

NONDEGENERATE PERTURBATION THEORY Summary (2) = P k <k|H (1) |i><i|H (1) |k> (1) =< i|H (1) |i > (7. Starting with (3) i (0) (2) =< Ψi |H (1) |Ψi (0) (1) (0) > + < Ψi |H (2) |Ψi (0) (2) > and with (0) (1) (H − i )Ψi (1) (1) (H (1) − i )Ψi (0) (H (0) − i )Ψi + (0) > + < Ψi |H (3) |Ψi + (H (1) (1) (0) = 0 (2) (0) = 0 − i )Ψi + (H (2) − i )Ψi Using the second and third equations we will try and get rid of the first term in the first equation.16) |<k|H (1) |i>|2 (7. multiply the second expression by Ψ∗(2) and integrate.118 CHAPTER 7. Multiply the third expression by Ψ∗(1) and integrate. What about 3rd order corrections? Going back for the 3rd order non-degenerate energy correction (3).18) > So you see that we will typically go to second order corrections in the energies and only to 1st order in the wavefunctions.17) = (0) (0) (i −k ) |Ψ >= |i > + P P k (0) (0) (i −k ) <k|H (1) |i> k ((0) −(0) ) |k i k (7. Second expression From the second expression we get the following (0) (1) ∗(2) Ψ∗(2)(H (0) − i )Ψi + Ψi < (2) Ψi |H (0) (0) (1) − i |Ψi >+< (1) (0) (H (1) − i )Ψi (2) Ψi |H (1) (1) (0) − i |Ψi = 0 > = 0 giving (2) (1) (0) < Ψi |H (1) − i |Ψi (2) (0) (1) >= − < Ψi |H (0) − i |Ψi > . To do this. Here is (2) (1) an alternative expression where we don’t need to know |Ψi > only |Ψi >.

(1) < Ψi |H (1) − i |Ψi < and i (0) (2) Ψi |H (1) |Ψi > = − < Ψi |H (0) − i |Ψi (0) (2) > > = −< (1) Ψi |H (0) (0) (2) − i |Ψi > (0) (1) therefore (3) i (1) (0) (2) = − < Ψi |H (0) − i |Ψi > + < Ψi |H (2) |Ψi (0) (0) > + < Ψi |H (3) |Ψi > Third expression Now from the third expression we get (0) (1) (0) (2) < Ψi |H (0) − i |Ψi −< (2) (1) (1) (2) (0) Ψ∗(1) (H (0) − i )Ψi + Ψ∗(1)(H (1) − i )Ψi + Ψ∗(1) (H (2) − i )Ψi (1) Ψi |H (0) − (1) (1) (1) > + < Ψi |H (1) − i |Ψi (0) (2) i |Ψi >=< (1) Ψi |H (1) (1) (1) − i |Ψi (1) (2) (0) > + < Ψi |H (2) − i |Ψi >+< (1) Ψi |H (2) = 0 > = 0 or (2) (0) − i |Ψi > Finally using this last expression we can replace it into the result from the second expression above to get (3) i (1) (0) (2) = − < Ψi |H (0) − i |Ψi = + (0) (1) > + < Ψi |H (2) |Ψi (0) (0) > + < Ψi |H (3) |Ψi (1) (1) (1) (1) (2) (0) [< Ψi |H (1) − i |Ψi > + < Ψi |H (2) − i |Ψi (0) (1) (0) (0) < Ψi |H (2) |Ψi > + < Ψi |H (3) |Ψi > > >] The only term that simplifies is the second one in brackets. Now we get for the 3rd order correction (3) i (1) (1) (1) =< Ψi |H (1) − i |Ψi (1) (0) > + < Ψi |H (2) |Ψi (0) (1) It is apparent that all we need to know now is |Ψi Using (1) |Ψi > = X < k|H (1) |i > (0) k (1) < Ψi | = (0) (i − k ) X < i|H (1) |j > (0) j (0) (i − j ) (1) > + < Ψi |H (2) |Ψi > |k > < j| (0) (0) > + < Ψi |H (3) |Ψi > . It reduces to (1) (0) (1) (0) < Ψi |H (2) |Ψi > since |Ψi > and |Ψi > are orthogonal.119 (1) (0) where since H (1) and H (0) and i (0) (1) (2) are all Hermitian.

i i (0) (i −j )(i −k ) =< i|H (1) |i > so that <i|H (1) |j><j|H (1) |k><k|H (1) |i> (0) (0) (0) (0) (i −j )(i −k ) − < i|H (1) |i > P k6=i <i|H (1) |k><k|H (1) |i> (0) (0) (i −k ) (7.19) Alternatively. NONDEGENERATE PERTURBATION THEORY Now assume (to make life simple) that the Hamiltonian only has a first order perturbation H (2) = 0 and H (3) = 0.120 CHAPTER 7. This leaves behind (3) i (1) (1) (1) = < Ψi |H (1) − i |Ψi (1) (1) = < Ψi |H (1) |Ψi (1) (1) Ψi |H (1) |Ψi > (1) (1) (1) > − < Ψi |i |Ψi (1) −i > (1) (1) Ψi |Ψi = < > < >   # " X < i|H (1) |j > X < k|H (1)|i > (1) (1) (1)  H (1) =  |k > − i < Ψi |Ψi > (0) (0) < j| ( −  ) j = k (0) (i kj6=i = i k X < i|H (1) |j >< j|H (1) |k >< k|H (1) |i > (0) (0) − j )(i (0) − k ) (1) − i (1) (1) < Ψi |Ψi > X < i|H (1) |j >< j|H (1) |k >< k|H (1) |i > (0) (0) (0) (0) (i − j )(i − k ) "  # (1) X < k|H (1) |i > X < i|H |j > (1)  < j| |k > − i (0) (0) (0) (0) ( −  ) (i − k ) j k i j kj6=i = X < i|H (1) |j >< j|H (1) |k >< k|H (1) |i > (0) (1) − i (0) (0) (0) (i − j )(i − k ) kj6=i X < i|H (1) |j >< j|k >< k|H (1) |i > (0) (0) (0) (0) (i − j )(i − k ) kj6=i Now since we have < j|k > the only terms that are non-zero are those where j=k X < i|H (1) |j >< j|H (1) |k >< k|H (1) |i > X < i|H (1) |k >< k|H (1)|i > (1) = −  i (0) (0) (0) (0) (0) (0) (i − j )(i − k ) (i − k )2 kj6=i k6=i Therefore P (3) i = kj6=i <i|H (1) |j><j|H (1) |k><k|H (1) |i> (0) (1) (3) = P kj6=i (0) (0) (1) P <i|H (1) |k><k|H (1) |i> (0) (0) k6=i (i −k )2 − i (7.20) .

So when put together with the confined energies from the last section.Chapter 8 Density of states The idea here is that while the confinement of an axis gives rise to discrete energies (as we saw in the previous section on confinement). As a consequence one way to get an idea of what the energies look like is to perform the following density of states argument. these density of states calculations provide a more thorough description of what the electronic structure of 3D. there are additional states along the unconfined axes (degrees of freedom). 1D and 0D materials look like. named after Max Born and Theodore Von 121 . 3 Dimensions (bulk) Consider the volume in “k” space 4 Vk = πk 3 3 (k 2 = kx2 + ky2 + kz2 ) where for a particle in this sphere kx = ky = kz = 2π Lx 2π Ly 2π Lz Note that the 2π arises from the constraints of a periodic boundary condition (Born-Von Karman conditions. Here it is not practical to try and calculate the energies of each state since they are actually continuous bands or energies. 2D. in cases such as the quantum well and quantum wire.

1) dε dε 3π dε q (units of 1 over length) for a free electron or alterwhere recall k = 2mε h2 ¯ 3   2 . N 0 = 2N = 2 4 3 3 πk Lx Ly Lz 8π 3 k3 Lx Ly Lz total number of states in sphere 3π 2 = Consider the density 0 ρ= k3 N = 2 number of states/unit volume Lx Ly Lz 3π Now consider the energy density defined as  3 k d dρ 1 dk 3 0 3π 2 ρ = = = 2 (8.2) This is the “density of states” in 3 dimensions. (8.3) . Note the square root dependence on energy. .z = 2πn Lx. The volume of a given mode is then kx ky kz The number of modes (N ) in the sphere are then 4 πk 3 Vk = 3 3 Lx Ly Lz kx ky kz 8π N= Say the particle is an electron and we consider the spin degeneracy (up and down).y.y. Note also that the energies of the system are ε= where k = q ¯ 2k2 h 2m kx2 + ky2 + kz2 and kx.122 CHAPTER 8. then we multiply N by 2. 3 . The expression becomes natively k 3 = 2mε 2 h ¯ 1 ρ = 2 3π  2m ¯h2 3 dε 2 dε 1 ρ = 2 3π  2m ¯h2 3 3 1 ε2 2 0 0 0 ρ3D = 1 2π 2  2 2 2m h2 ¯ 3 3 √ 2 ε (8. 1. 2.z . . . DENSITY OF STATES Karman) as opposed to the more general nπ where n = 0.

2. It is expressed as εtot = ¯ 2 kz2 ¯h2 k 2 h + = εn + εx.1 eV. just alter the expression for k below equation 5.y 2m 2m where k 2 = kx2 + ky2 and kz = nπ Lz Consider now an area in k space Ak = πk 2 (8. Hint. Express the 3D density of states in this situation. Assume the electron resides in a non-zero potential V . The total energy of this system is a sum of the energy along the quantized direction + the energy along the other 2 “free” directions. 2 Dimensions (well) Here we have 1 dimension that is quantized. Express your answer in terms of eV and cm3 .1: Density of states in a 3D material (bulk) Exercises 1.123 Figure 8.4) . Let’s assume its the z direction.1. Calculate the 3D density of states for free electrons with energy 0.

124 CHAPTER 8. Exercises 1.5) where Θ is the Heavyside function. DENSITY OF STATES where for the particle kx = ky = 2π Lx 2π Ly The area of a given mode is then kx ky with the total number of modes (N ) in the area being N = k 2 Lx Ly πk 2 L L = x y 4π 2 4π Again if the particle is an electron and we consider spin. Consider only the lowest subband.1 eV. Calculate the 2D density of states for free electrons with energy 0. multiply by 2 to get 0 N = 2N = k 2 Lx Ly total number of states in area 2π Consider now a density 0 ρ= where recall that k = ρ= q 2mε . . h2 ¯ N k2 = Lx Ly 2π Replacing this above gives mε number of states per unit area ¯h2 π Now consider the energy density 0 ρ = dρ m = 2 dε ¯ π h This is the energy density of the subband for a given kz (or εn ). For each sucterm and hence another subband. cessive kz there will be an additional ¯hm 2 π Therefore the density of states is written 0 ρ2D = m ¯2π h P n Θ(ε − εn ) (8. Express your answer in eV and cm3 units.

n are integers.125 Figure 8.6) where k = kx = L2πx Furthermore along the confined directions. kz = nπ Lz . ky = mπ and m. Consider a length 2k. The number of modes Ly along this length is N= 2k 2k kLx = = number of state along the line 2π kx π Lx Now if we consider an electron again. The total energy of the system can be written as εtot = h2 ky2 ¯h2 k 2 ¯ 2 kz2 ¯ h + + = εn + εm + εx 2m 2m 2m (8. ensure to take into account the spin degeneracy 0 N = 2N = 2kLx π Now a density is 0 2k 2 N = = ρ= Lx π π r 2mε number of states per unit length h2 ¯ .2: Density of states in a 2D material (quantum well) 1 Dimension (wire) Consider now the situation where there are 2 dimensions confined and only 1 degree of freedom (say the x direction).

126

CHAPTER 8. DENSITY OF STATES

Consider the energy density
0

ρ

r
r


2 2m d ε
2 2m 1 − 1
=
=
ε 2

π ¯h2 dε
π ¯h2 2
r
1 2m 1

π ¯h2 ε

=
=

(8.7)

This is the energy density for a given n, m value or (εn , εm combination).
The complete expression taking into account all m, n combinations is
0

ρ1D =

1
π

q

2m
h2
¯

P

n,m

√ 1
ε−εn,m Θ(ε − εn,m )

(8.8)

where again, Θ is the Heavyside function. Notice the inverse square root
dependence of the density of states with energy.

Figure 8.3: Density of states in a 1D material (quantum wire)

Exercises
1. Calculate the 1D density of states for free electrons with energy 0.1 eV
above εn,m . Consider only the lowest subband. Express your result in
units of eV and cm3 .

127

0 Dimension (dot)
Here since all three dimensions are confined the density of states is basically
a series of delta functions. The total energy of the system is
εtot =

h2 ky2 ¯h2 kz2
¯ 2 kx2 ¯
h
+
+
= εm + εn + εo
2m
2m
2m

where m, n, o are integers and kx =
states is


Lx ,

ky =


Ly ,

kz =

(8.9)

Lz

The density of

0

ρ0D = δ (ε − εm,n,o )

(8.10)

Figure 8.4: Density of states in a 0D material (quantum dot)

Exercises
1. Compare and contrast the density of states in the 0D case to the
previous 3D, 2D and 1D cases.

128

CHAPTER 8. DENSITY OF STATES

Chapter 9

More density of states
Density of states in the conduction band
To do this calculation we need to know the probablility that an electron will
occupy a given state of energy ε. This probability, Pe (ε), is refered to as the
Fermi Dirac distribution. In addition, we need to know the density of states
0
(ρ ) which we calculated in the last section. Recall that this density of states
0
has units of number per unit volume per unit energy. Therefore ρ dε is the
number of states per unit volume. The number of occupied states at a given
energy per unit volume (or alternatively the concentration of electrons at a
given energy) is therefore
0

ne (ε) = Pe (ε)ρ (ε)dε
where the Fermi Dirac distribution is
Pe (ε) =

1
1+e

ε−εF
kT

(9.1)

Here εF is the Fermi energy (also called the chemical potential sometimes).
The total concentration of electrons in the conduction band is therefore
the integral over all available energies
Z ∞
0
Pe (ε)ρ (ε)dε
(9.2)
ntot =
εC

where εC is the energy where the conduction band starts. Consider the
density of states we just derived for a 3D (bulk) material in the previous
0
section, ρ3D . 
3 

1
0
2me 2 √
ε
ρ3D = 2

¯h2
129

130

CHAPTER 9. MORE DENSITY OF STATES

Now rather than have the origin of the energy start at 0, start it where the
conduction band begins (εC ). This expression can then be recast as 
3 

1
0
2me 2 √
ρ3D = 2
ε − εC
(9.3)

¯h2
Replace this and the Fermi Dirac distribution into the above expression for
the total concentration of electrons in the conduction band to get  

3
Z ∞
2me 2 √
1
1
ntot =
ε − εC dε
ε−εF
2
¯h2
εC 1 + e kT 2π  

3 Z

2me 2 ∞
1
1
ε − εC dε
=
ε−εF
2
2

¯h
εC 1 + e kT
Z ∞

1
ε − εC dε
= A
ε−εF
εC 1 + e kT
Z ∞

1
ε − εC dε
= A
ε−εC +εC −εF
εC 1 + e
kT
Z ∞

1
= A
ε − εC dε
εC −εF
ε−εC
εC 1 + e kT e kT
(9.4)
Here, let η =

ε−εC
kT

and µ =

−εC +εF
leading
kT
Z ∞

to

1 √
ηdη

0

ntot = A

eη−µ

εC  

3
0
2
e
where A = 2π1 2 2m
and A can be found from the above change of
2
h
¯
variables. Note at this point that someone has already solved this integral
and one can just look it up. The integral is called the Fermi integral or
Fermi Dirac integral. It is defined as follows and its solutions labeled F 1 (η)
2
can be looked up.
F 1 (η) =
2

R∞
0


η
1+eη−µ dη

However, to stay instructive let’s just consider the case where ε − εF >>
kT . In this case the exponential in the denominator of the Fermi Dirac
distribution will dominate and the expression basically becomes
1
1+e

ε−εF
kT

'

1
e

ε−εF
kT

= e−

ε−εF
kT

131
which has the Boltzman distribution form. Our expression for ntot becomes
Z ∞
ε−εF √
e− kT ε − εC dε
ntot = A
(9.5)
εC
Z ∞
[(ε−εC )+(εC −εF )] √
kT
ε − εC dε
= A
e−
ε0

C
Now change variables and let x = ε−ε
kT such that ε = εC + xkT and dε =
kT dx. Note also that the limits of integration will change accordingly. This
leads to
Z ∞
1
1
e− kT [(ε−εC )+(εC −εF )] (xkT ) 2 kT dx
ntot = A

0

3
2

= A(kT ) e

εC −εF
kT

Z

1

e−x x 2 dx

0

The last integral is a common function called the Gamma function. It is
defined as follows
Z ∞
Γ(n) =
e−x xn−1dx
0

and its values
can be looked up in most reference books. In our case we 

have Γ 23 . Therefore ntot can be expressed as 

εC −εF
3
3

nC ≡ ntot = A(kT ) 2 e kT Γ
2
To be consistent with other texts, we use nC in lieu of ntot at this point
3
and also define NC = A(kT ) 2 Γ 32 . This leads to the common textbook
expression
nC = NC e−

εC −εF
kT

(9.6)

This is the expression for the effective density of states of the conduction
band.

Density of states in the valence band
The way this is calculated mirrors the approach for the conduction band
with a few minor changes. Again, we need the probability of finding a hole
at a given energy Ph (ε). Likewise we need the density of states calculated

132

CHAPTER 9. MORE DENSITY OF STATES

in the previous section. We will calculate the number of holes at a given
energy per unit volume (or concentration of holes at a given energy)
0

nh (ε) = Ph (ε)ρ (ε)dε
where we use the relationship
Pe (ε) + Ph (ε) = 1
or
Ph (ε) = 1 − Pe (ε)
This lead to
Ph (ε) = 1 −

1
1+e

(9.7)

ε−εF
kT

The density of states is
0

ρ3D

1
= 2
2π 

2mh
¯h2 

3
2


εV − ε

(9.8)

where εV is the energy where the valence band starts. Consequently the total
concentration of holes in the valence band is the integral over all energies.
ptot =

Z

=

Z

εV

0

Ph (ε)ρ (ε)dε

−∞
εV

1−

−∞

1
1+e

ε−εF
kT

For notational convenience let B =

ptot = B

Z

εV
−∞

1−

!

1
2π 2  

2mh
h2
¯ 

3

1
2π 2

1
1+e

ε−εF
kT

!

2

2mh
¯h2 

3
2


εV − ε dε

. This leads to

εV − ε dε

Since, generally speaking ε < εF
ptot = B

Z

εV
−∞

1−

1
ε −ε
− FkT

1+e

!


εV − ε dε

We get the following expression upon recognizing this   εF −ε 3 3 − ptot = B(kT ) 2 e kT Γ 2 To be consistent with common notation and other texts ptot can be expressed as pV ≡ ptot = Nv e− εF −εV kT (9. 1 1+x 1 ε −ε − FkT 1+e ' 1 − x + x2 − x3 + x4 + . make a change of variables by letting x = εVkT−ε as well as ε = εV − kT x and dε = −kT dx. Therefore 1− 1 ε −ε − FkT 1+e !   εF −ε εF −ε = e− kT ' 1 − 1 − e− kT Replacing this in the ptot expression gives Z εV εF −ε √ εV − ε dε ptot = B e− kT −∞ As with the conduction band case earlier.. The last expression gives the effective density of states for the valence band. .9)  3 where Nv = B(kT ) 2 Γ 23 . ' 1 − e− εF −ε kT + . .133 Approximate the stuff in parenthesis through the binomial expansion keeping only the first 2 terms (where |x| < 1). .. Note that the limits of integration change accordingly resulting in Z 0 √ 1 ptot = B e− kT [(εF −εV )+(εV −ε)] kT x(−kT )dx Z∞∞ εV −ε εF −εV 3 1 = B e− kT e− kT (kT ) 2 x 2 dx 0 Z ∞ εF −ε 3 1 − = B(kT ) 2 e kT e−x x 2 dx 0 Recall that the last integral above is the Gamma function.

. their density of states and their population at the conduction and valence band edges.134 CHAPTER 9.1: Sketch of the electron and hole distribution functions. MORE DENSITY OF STATES Figure 9.

135 Summary Fermi level of an intrinsic semiconductor If the bulk semiconductor is intrinsic. there has been no doping of the material and hence no extra electrons or holes anywhere. In this situation nC = pV where previously we found that − nC = NC e pV = NV e− εC −εF kT εF −εV kT   3 and NC = A(kT ) Γ 2   3 3 and NV = B(kT ) 2 Γ 2 3 2 This leads to NC e− εC −εF kT = NV e −εF −εV kT or Ae− εC −εF kT = Be− εF −εV kT .

the Fermi level is at the midway point between valence and conduction bands. 77K. 300K and 600K.08mo and . To a good approximation however.10) One can therefore see that at T = 0 the Fermi energy of an intrinsic semiconductor is at the halfway point between the top of the valence band and the bottom of the conduction band. Note also that since generally speaking mh > me the temperature dependent term is positive and slowly increases the Fermi level position with increasing temperature. Exercises • Calculate the intrinsic Fermi level of silicon at 0K.136 CHAPTER 9. MORE DENSITY OF STATES where A = 1 2π 2  2me h2 ¯ 3 2 and B = 3 εC −εF kT me2 e− e− εC −εF kT εF −εV kT 3 1 2π 2  2mh h2 ¯ 3 3 2 = mh2 e− and ultimately reduces to εF −εV kT 3 = mh2 3  2 1 m h (−ε +ε +ε −ε ) e kT C F F V = me 3  2εF εC +εV mh 2 − e kT e kT = me  3 2εF mh 2 εC +εV e kT e kT = me 3   2εF   εC +εV  mh 2 ln e kT + ln e kT = ln me 3  2εF mh 2 εC + εV + = ln kT me kT 3  mh 2 + εC + εV 2εF = kT ln me  3 mh 2 εC + εC kT ln εF = + 2 me 2 e me2 This yields the final expression εF = εC +εC 2 + 43 kT ln  mh me 3 2 (9. Note that Eg = Ec + Ev and assume me = 1. 10K.

137 Figure 9. .55mo . mh = 0. Leave the answer in terms of Eg or if you desire look up the actual value of Eg and express the final answers in units of eV.2: Sketch showing the Fermi level position at 0 degrees halfway between the conduction band and valence band positions.

138 CHAPTER 9. MORE DENSITY OF STATES .

In this section we repeat the same calculations for materials with lower dimensions. we could choose the first band. For example. Density of states in the conduction band: 2D We start with the Fermi Dirac distribution for electrons and also the density 0 of states (ρ2D ) that we derived in an earlier section. Recall that 0 ρ2D = me X Θ(ε − εn ) h2 π n ¯ Consider only one of the subbands. Rather than use ntot as before let’s just stick to nc 139 .Chapter 10 Even more density of states In the previous section we ran through the calculation for the conduction and valence band density of states for a 3D (bulk) material. In particular we do this for 2D and 1D materials that are representative of quantum wells and quantum wires respectively. In this case the density of states simplifies to 0 ρ2D = me ¯h2 π Now recall from the previous section that the number of states at a given energy per unit volume (or the concentration of electrons at a given energy) is 0 ne (ε) = Pe (ε)ρ (ε)dε The total concentration of electrons in this first subband is the integral over all available energies.

EVEN MORE DENSITY OF STATES Figure 10.1: Conduction band density of states for a 2D material. from the start nc = 1 where Pe (ε) = 1+e Z ∞ 0 Pe (ε)ρ (ε)dε (10. The nth band is shaded to distinguish it from the other subbands.1) εC is the Fermi Dirac distribution. Putting everything ε−εF kT together we get nC = Z ∞ 1 1+e εC = me ¯h2 π Z ∞ εC ε−εF kT ! 1 1+e ε−εF kT me dε ¯h2 π dε Since the band really begins at εn as opposed to εC like in the bulk the integral changes from Z ∞ Z ∞ → εC εn leading to me nC = 2 ¯h π Z ∞ εn dε 1+e ε−εF kT .140 CHAPTER 10.

141 if now ε − εF >> kT me nC = 2 h π ¯ Z ∞ − e ε−εF kT εn .

me kT − ε−εF .

.

∞ dε = − 2 e kT .

as in the conduction band case.3) −∞ !  Z εV mh 1 = 1− dε ε−εF ¯h2 π −∞ 1 + e kT ! Z 1 mh εV dε 1− = ε−εF h2 π −∞ ¯ 1 + e kT 0 . This is denoted Ph (ε) and is evaluated from Pe (ε) + Ph (ε) = 1 Ph (ε) = 1 − Pe (ε) where Pe (ε) is the Fermi Dirac distribution leading to Ph (ε) = 1 − 1 1+e ε−εF kT The number of states at a given energy per unit volume (concentration at a given energy) is 0 nh (ε) = Ph (ε)ρ (ε)dε If. ¯h π εn This leads to the final expression for the carrier density of the nth subband. The total concentration of holes in this first subband is the integral over all energies. which in our example we assumed was the lowest one. Hence this is the carrier density at the conduction band edge of a 2D material.2) Density of states in the valence band:2D As with the conduction band case we need the probability of occupying a given state in the valence band. We get Z εV 0 pV = Ph (ε)ρ (ε)dε (10. nC = me kT − e h2 π ¯ εn −εF kT (10. we consider only the nth subband ρ = ¯hm2hπ and for simplicity the first.

. keeping only the first two terms. Recall that 1 = 1 − x + x2 − x3 + . 1+x ε −ε 1 − FkT +.. .142 CHAPTER 10. εF −ε = 1 − e 1 + e− kT such that mh pV = 2 h π ¯ Z εn −∞ − e εF −ε kT . The nth band is shaded to distinguish it from the other subbands. Since the nth (in our case the first) subband begins at εn rather than εV as in the bulk. the limits of the integral change Z εV Z εn → −∞ −∞ The expression becomes mh pV = 2 ¯h π Z εn −∞ 1− 1 1+e ε−εF kT ! dε Since for the valence band ε < εF ! Z 1 mh εn 1− dε pV = 2 εF −ε ¯h π −∞ 1 + e− kT Apply the binomial expansion of the term in parenthesis to simplify. .2: Valence band density of states for a 2D material. EVEN MORE DENSITY OF STATES Figure 10.

mh kT − εF −ε .

.

εn dε = e kT .

¯h2 π −∞ .

To a first approximation however the Fermi level is midway between the valence band beginning and conduction band beginning.143 This reduces to our desired final expression mh kT − e h2 π ¯ pV = εF −εn kT (10. The above expression reduces   1 mh e kT (−εn1 +εF +εF −εn2) = me   2ε 1 mh − kT (εn1 +εn2 )+ kTF e = me   εn1 +εn2 2εF mh e− kT e kT = me   2εF εn1 +εn2 mh e kT = e kT me    εn1 +εn2   2εF  mh kT + ln e kT ln e = ln me   εn1 + εn2 mh 2εF + = ln kT me kT   mh 2εF = kT ln + εn1 + εn2 me resulting in the final expression εF = εn1 +εn2 2 + kT 2 ln  mh me  (10. If we assume an intrinsic semiconductor with no doping such that there are no additional electrons or holes present nC = pV then mh kT εF −εn2 me kT − εn1 −εF kT = 2 e− kT e 2 h π ¯ ¯h π Note that n1 and n2 have been used to distinguish the start of the conduction band subband and the valence band subband respectively.4) Fermi level position:2D The procedure for finding the Fermi level position is the same as in the 3D case. since mh > me the second term is positive and grows slowly with increasing temperature.5) As before. .

n ¯h m. The number of states at a given energy per unit volume (or the concentration at a given energy) is 0 ne (ε) = Pe (ε)ρ (ε)dε The total concentration in the first subband is the integral over all possible energies Z ∞ 0 nc = Pe (ε)ρ (ε)dε εC where as before the Fermi Dirac distribution is 1 Pe (ε) = ε−εF 1 + e kT . EVEN MORE DENSITY OF STATES Density of states in the conduction band:1D We start with the Fermi Dirac distribution for electrons and the density of 0 states (ρ1D ) that we derived earlier. n)th subbband beginning at energy εm. We consider only the (m.144 CHAPTER 10.n)th band is shaded to distinguish it from the other subbands.n ) ρ1D = √ 2 π ε − εm. For convenience we could say this is the first subband.3: Conduction band density of states for a 1D material.n where Θ is the Heavyside function. Recall that r 1 1 2me X 0 Θ(ε − εm. Figure 10.n . The (m.

n 1 dε ε − εm.n εm.n −εF kT e h2 ¯ e− ε−εF kT εm.145 On introducing this.n ∞ e− ε−εm.n +εm.n If ε − εF >> kT 1 1+e ε−εF kT → e− ε−εF kT leading to nC = = = = ∞ 1 π r 2me h2 ¯ 1 π r Z 2me h2 ¯ 1 π r Z 2me h2 ¯ 1 π r Z 2me − εm.n ε−εm.n −εF kT εm.n ε−εF kT ! 1 dε √ ε − εm. Also make corresponding changes to the limits of integration resulting in √ r Z kT 2me − εm. the total expression becomes nC Z = ∞ ε−εF kT 1 π r 1 2me dε 2 √ε − ε ¯h m.n −εF kT 1 dε √ ε − εm.n such that ε = εm.n −εF ∞ −x − 1 kT nC = e x 2 dx e π h2 ¯ 0 .n e− m.n kT e− √ εm.n + kT x and Make a change of variables and let x = ε−ε kT dε = kT dx.n 1+e Z 2me ∞ 1 1 dε ε−εF √ 2 ε − εm.n kT 1 dε √ ε − εm.n h ¯ εC 1 + e kT εC = 1 π 1 r Since the band actually begins at εm.n Therefore 1 nC = π r 2me h2 ¯ Z ∞ 1 1+e εm.n ∞ e− 1 dε √ ε − εm.n the limits of integration become Z ∞ εC → Z ∞ εm.n Z ∞ εm.n ε−εm.

n − ε ¯h2 The number of states at a given energy per unit volume (or concentration at a given energy) is 0 nh (ε) = Ph (ε)ρ (ε)dε . As before we need the probability of occupying a given state in the valence band. (and for convenience that could be the first one) the expression simplifies to r 1 0 1 2mh ρ1D = √ π εm.n − ε where Θ is the Heavyside function and the band explicitly starts at εm.n −εF kT Γ e nC = 2 π 2 ¯h The final expression for the carrier concentration of a 1D material at the conduction band edge is nC = 1 π q 2me kT − e h2 ¯ εm. a few of the terms change slightly in appearance.146 CHAPTER 10.n −εF kT Γ 1 2  (10. n)th band. The occupation probability is calculated from Pe (ε) + Ph (ε) = 1 Ph (ε) = 1 − Pe (ε) Here Pe (ε) is the Fermi Dirac distribution leading to Ph (ε) = 1 − 1 1+e ε−εF kT The density of states is r 1 2mh X 1 0 ρ1D = Θ(εm.n m. However. EVEN MORE DENSITY OF STATES Recall that the last integral is the Gamma function √ r   1 kT 2me − εm. We also need the density of states calculated in the previous section. Now if we consider only the (m.n causing its presence in the square root denominator.n − ε) √ 2 π ε ¯h m.6) Density of states in the valence band:1D The way this is calculated is very similar to that for the conduction band.

The (m.n − ε dε 1 dε √ εm.n rather than at εV as in the bulk.n (10.n pV = π h2 −∞ ¯ Furthermore.n − ε ¯h −∞ 1 + e kT ! r Z 1 1 1 2mh εV 1− = dε √ ε−εF 2 π εm.147 Figure 10.4: Valence band density of states for a 1D material.7) pV = −∞ ! r Z εV 1 2mh 1 1 = 1− dε √ ε−εF 2 π εm.n − ε h ¯ −∞ 1 + e kT Since the band begins at εm. The total concentration in the subband is the integral over all energies Z εV 0 Ph (ε)ρ (ε)dε (10.n)th band is shaded to distinguish it from the other subbands.n pV = π h2 −∞ ¯ 1− 1− −∞ 1 1+e ε−εF kT ! 1 ε −ε − FkT 1+e √ ! 1 εm.8) → −∞ The expression becomes r Z 1 2mh εm.n − ε . since ε < εF r Z 1 2mh εm. the limits of integration change Z εV Z εm.

n −ε kT εm.9) −ε To simplify.n kT Γ 1 2  (10.n kT e ¯h2 εF −ε kT 1 dε √ εm. Let x = m.n − ε (10.n − ε dε 1 dε εm. .n − kT x kT and dε = −kT dx.n −∞ εm.n − ε √ √ dε 1 εm.n +εm.10) .n −ε kT −∞ ε 1 εm. pV = = r Z 1 2mh − εF −εm.n e− e− √ εm. .n −∞ Z εm. EVEN MORE DENSITY OF STATES Using the binomial expansion and keeping only the first two terms 1 1+x 1 ε −ε − FkT 1+e = 1 − x + x2 − x3 + .n − ε εF −εm. Note also the corresponding changes in the limits of integration.n −∞ e− e− e− 2mh − εF −εm. = 1 − e− εF −ε kT the term in parenthesis becomes 1− 1 ε −ε − FkT 1+e ! = e− εF −ε kT yielding pV = 1 π = 1 π = = 1 π 1 π r r r r 2mh ¯h2 2mh ¯h2 2mh ¯h2 Z Z εm.n kT e−x x− 2 dx 2 e π ¯h 0 The resulting integral is the now familiar Gamma function.n 0 −x 1 kT e √ √ (−kT )dx e 2 π ¯h kT x ∞ √ r Z ∞ 1 kT 2mh − εF −εm.n −ε kT εF −εm.148 CHAPTER 10.n or ε = εm.n kT Z εm. This lead to our final expression for the total concentration of holes in the valence band pV = 1 π q 2mhkT − e h2 ¯ εF −εm. make a change of variables.

11) Since mh > me the second term is positive and as a consequence the Fermi energy has a slight temperature dependence.n2 are used to refer to the energies where the conduction and valence bands begin repsectively.n2 1 2me kT kT kT e e = Γ Γ 2 2 π 2 π 2 h ¯ ¯h Here note that εm1.n1 +εm2.n2 ) kT = εm1.n2 2 + kT 4 ln  mh me  (10.n2 mh 2 kT + ln e ln me 1  mh 2 εm1.n1 +εF +εF −εm2. to a first approximation. This expression reduces as follows e− εm1.n2 kT e me  ε 1   m1.149 Fermi level position:1D This evaluation goes the same was as for the 2D or 3D material.n1 + εm2.n1 and εm2. If we are dealing with intrinsic semiconductors with no additional doping nC = pV From what we have just evaluated r r   ε   ε −ε 1 2mh kT 1 − m1. it can be taken as the midpoint between the conduction band and valence band energies (the zero temperature limit). However. as before.n2 √ kT mh e− r mh me r mh me r mh εm1.n1 +εm2.n1 +εm2.n1 +εm2.n2 ln + me kT   kT mh + εm1.n1 + εm2.n1 −εF 1 − F m2. .n1 −εF √ kT me e− = (−εm1.n2 kT = e− 2εF e kT 2εF e kT =  2εF  ln e kT = 2εF kT = 2εF = εF −εm2.n2 ln 2 me This leads to our final expression εF = εm1.

EVEN MORE DENSITY OF STATES .150 CHAPTER 10.

Chapter 11 Joint density of states In the previous sections we have calculated the “density of states” in both the conduction band and valence band separately. Interband transitions occur between both bands giving rise to optical transitions of the semiconductor. recall kx ky kz where kx = ky = kz = 2π Lx 2π Ly 2π Lz The number of modes or states in the given sphere is then N = 4 πk 3 Vk = 3 3 Lx Ly Lz kx ky kz 8π 151 . Consider a spherical volume of 4 Vk = πk 3 3 The volume of a given mode was. 3D bulk We derive the density of states again as done previously. As a consequence it is instructive to calculate the “joint” density of states which is proportional to the absorption coefficient of the material.

3) we obtain dk dε k 2 dk 2π 2 dε 0 ρj (ε) = ρ1 (k) = (11.  3 1 2m 2 √ ε (previously) 2π 2 ¯h2 k2 (this time) π2 Starting with the energy density 0 ρ2 = k2 dρ = 2 dk π (11.1) divide by 2 to go back to only 1 spin orientation since in an optical transition spin flips are generally forbidden. let’s solve for dρ dk . the density 0 ρ= k3 N = 2 number of states/unit volume Lx Ly Lz 3π Now consider the energy density 0 ρ = dρ dρ alternatively dε dk Note that previously we solved for dρ dε dρ dk = = dρ dε . multiply this by 2 to account for spin N 0 4 πk 3 = 2N = 2 3 3 Lx Ly Lz 8π k3 = Lx Ly Lz total number of states in sphere 3π 2 Consider as before. 0 k2 ρ ρ1 = 2 = 2 2 2π 0 (11. This time however.152 CHAPTER 11. JOINT DENSITY OF STATES For an electron.2) This expression applies to either conduction band or valence band. Applying the following equivalence 0 ρj (ε)dε = ρ1 (k)dk (11.4) .

where ka is the k value in the valence band and kb is the value in the conduction band). transitions in k are vertical such that the intitial k value in the valence band is the same k value as in the conduction band (ka = kb = k.153 where ρj is the desired joint density of states. Figure 11. Now from the conservation of momentum.1: Cartoon showing momentum conserving vertical transition between the valence band and conduction band in k space The energy of the initial state in the valence band is εa = εv − ¯ 2 ka2 h 2mh Likewise the energy of the final state in the conduction band is εb = εc + ¯ 2 kb2 h 2me .

5) where εg = εc − εv .7) 2π 2 ¯h2 mh where for notational simplicity we have used the reduced mass µ = mmee+m h Now to continue towards our final expression.6)       k 1 k2 me mh me mh = 2 2 ρj (ε) = 2π 2 ¯h2 k me + mh me + mh 2π ¯h kµ = (11. explicitly describe what k is ε = εg + ↓ k2 = ¯ 2k2 h 2µ 2µ(ε − εg ) ¯h2 or p 2µ(ε − εg ) k= ¯h Replacing this into our main expression for ρj gives µ k ρj (ε) = 2π 2 ¯h2 r  2µ(ε − εg ) µ = 2 2 2π ¯h ¯h2 (11. JOINT DENSITY OF STATES The energy of the transition is ε = εb − εa     ¯ 2k2 h ¯h2 k 2 − εv − = εc + 2me 2mh ¯h2 k 2 ¯h2 k 2 + = (εc − εv ) + 2me 2mh  2 2  ¯h k 1 1 = εg + + 2 me mh (11.8) . Now from above me + me dε =h ¯ 2k dk me mh leading to the desired expression dk 1 = 2 dε ¯ k h Since ρj (ε) = k2 2π 2 dk dε  me mh me + mh  (11.154 CHAPTER 11.

155 which after some simplification gives ρj (ε) = 1 4π 2  2µ h2 ¯ 3 √ 2 ε − εg (11.2: Joint density of states for a 3D material (bulk) 2D well As we did before. the number of modes in the area is N= Ak πk 2 k2 = L L = Ly Lz y z ky kz 4π 2 4π .9) Figure 11. consider an area in k-space of Ak = πk 2 where the area occupied by a given mode or state is ky kz (we implicitly assume that kx represents the confined direction) 2π Ly 2π kz = Lz ky = Together.

0 k ρ ρ1 = 2 = 2 2π 0 (11.10) with the energy density given by 0 ρ = dρ dρ or alternatively dε dk Note that previously we solved for dρ dε dρ dk = = dρ dε .156 CHAPTER 11. This time. JOINT DENSITY OF STATES Multiply by 2 to account for spin 0 N = 2N = k2 Ly Lz 2π Now consider the density 0 N k2 ρ= = Ly Lz 2π (11. the conservation of momentum means that transitions in k-space are “vertical”. spin flip optical transitions are forbidden.11) Now applying the following equivalence 0 ρj (ε)dε = ρ2 (k)dk (11.13) where ρj is the desired joint density of states. As before in the 3D case. let’s consider dρ dk . m (previously) π¯h2 k (this time) π Starting with the energy density 0 ρ2 = dρ k = dk π divide by 2 to get rid of the spin since formally speaking. . however.12) one obtains dk dε k dk 2π dε 0 ρj (ε) = ρ1 (k) = (11.

not to be confused with the bulk band gap) mh This leads to = εn1 − εn2 and µ is the reduced mass µ = mmee+m h dε ¯h2 k = dk µ or µ dk = 2 dε ¯ k h (11.16) . the initial k value in the valence band is the same as the final k value in the conduction band (ka = kb = k) where ka (kb ) is the valence (conduction) band k values.157 That is.15) such that when replaced into our main expression the desired expression for the joint density of states is ρj (ε) = k 2π  µ h2 k ¯  1D wire Consider the length in k-space Lk = 2k = µ 2¯ h2 π (11.14) where εg (effective band gap. The energy of the initial state in the valence band is εa = εn2 − ¯ 2k2 h 2mh Likewise the energy of the final state in the conduction band is εb = εn1 + ¯ 2 k2 h 2me The transition energy is ε = εb − εa ¯ 2k2 h = εn1 − εn2 + 2 = εg + ¯ 2k2 h 2µ  1 1 + me mh  (11.

158 CHAPTER 11.3: Joint density of states for a 2D material (quantum well) The length occupied by a given mode or state is kx where kx = 2π Lx The number of states in the given length is N= Lk 2k kLx = Lx = kx 2π π Multiply this by 2 to account for spin kLx total number of states π 0 N =2 Consider the density 0 ρ= 2k N = number of states per unit length Lx π Then the energy density is 0 ρ = Previously we solved for dρ dε dρ dk dρ dρ or alternatively dε dk dρ dε = = r 1 2m 1 √ (previously) π ¯h2 ε 2 (this time) π . JOINT DENSITY OF STATES Figure 11.

19) . the conservation of momentum means that transitions in k-space are vertical so that ka = kb = k.18) where ρj (ε) is the desired joint density of states. 3D and 2D. As in the other two cases. Here ka is the k value in the valence band and kb is the k value in the conduction band. 0 ρ 1 ρ1 = 2 = 2 π 0 Now apply the following equivalence 0 ρj (ε)dε = ρ1 (k)dk leading to 0 ρj (ε) = ρ1 (k) = dk dε 1 dk π dε (11.n2 − ¯ 2k2 h 2mh Likewise the energy of the final state in the conduction band is εa = εm1.n2 + 2 = εg + ¯ 2 k2 h 2µ  1 1 + me mh  (11.n1 + ¯ 2k2 h 2me The transition energy is ε = εb − εa ¯ 2k2 h = εm1.159 Starting with the energy density 0 ρ2 = dρ 2 = dk π (11. The energy of the initial state in the valence band is εa = εm2.17) divide by 2 to consider only one spin orientation since spin flip transitions are generally forbidden.n1 − εm2.

This leads to dε ¯h2 k = dk µ or µ dk = 2 dε ¯ k h Since ρj (ε) = 1 dk π dε (11.21) or more directly analogous to last time = 1 2π q 2µ √ 1 h2 −g ¯ (11. Since 2 2 ε = εg + ¯h2µk we get 2µ(ε − εg ) ¯h2 k2 = ↓ r k = 1 k ↓ 2µ (ε − εg ) ¯h2 ¯h √ √ 2µ ε − εg = This leads to the final expression for the joint density of states ρj (ε) = 1 π q µ √ 1 2¯ h2 ε−εg (11. . they are probably smeared out by Coulomb effects.n1 − εm2. JOINT DENSITY OF STATES where εg = εm1.20) the joint density becomes ρj (ε) = µ 1 π¯h2 k Now to continue towards our final expression we express k fully.160 CHAPTER 11.n2 and the reduced mass µ = me mh me +mh .22) Note that the singluarities are referred to as Van Hove singularities. In the case of real quantum wires.

2D and 1D materials we can summarize their implications for the absorption spectra of these materials.4: Joint density of states for a 1D material (quantum wire) Summary Now that we have the explicit joint density of states for 3D.161 Figure 11. The calculation to achieve this involves Fermi’s golden rule and will not be discussed at this point. To summarize √ • 3D: ρj ∝ ε − εg • 2D: ρj ∝ constant depending on the reduced mass • 1D: ρj ∝ √ 1 ε−εg • 0D: ρj ∝ δ(ε − εn ) . It can be shown that the absorption coefficient and joint density of states are proportional to each other.

2D. 1D. and 0D materials .5: Summary of the joint density of states for 3D. JOINT DENSITY OF STATES Figure 11.162 CHAPTER 11.

163 ov .

JOINT DENSITY OF STATES .164 CHAPTER 11.

Picture next a bulk solid which can be divided into very small slices or “slabs” of thickness ∆x. each slab absorbs some light and transmits the rest. Recall now that x = n∆x giving x I = (1 − f ) ∆x Io 165 . re-arrange this into the following I = (1 − f )n Io This is the fraction of the incident intensity that makes it through n slabs. In each slab a fraction (1 − f ) of the impinging light is transmitted. Here 0 <= f <= 1.Chapter 12 Absorption A geometric derivation of exponential attenuation Here we derive the standard exponential attenuation law of light being absorbed by a solid. After the first slab you have a residual intensity of I = (1 − f )Io. Picture an incident beam of light with an initial intensity of Io . So initially we have Io . The general expression for the residual intensity after n slabs is therefore I = (1 − f )nIo To derive the exponential attenuation law. After the second slab you have a residual intensity of I = (1 − f )2Io and so on. Next. The whole solid consists of n slabs placed back to back such that the total thickness of the solid is x.

Alternative derivation There is an alternative derivation of the exponential attenuation law.1) where α is called the absorption coefficient. so is lny. This gives  −x I = elny Io Next. since y is a constant. I = Io e−αx (12. to simplify the notation let p = (1 − f ) yielding x I = p ∆x Io where 0 < p < 1. ∆x is a constant. ∆I ∝ −∆x . This can be rearranged as I Io  1 x p ∆x !−x 1 1 ∆x = p = where note that 1p > 1.166 CHAPTER 12. call the whole thing some other constant. This leads to the popular expression for the exponential attenuation of light in a solid. Call it α giving I = e−αx Io where α > 0. say y. The argument goes as follows: (a) The change in intensity of the light is proportional to the thickness of the solid. In addition. ABSORPTION Next. Now note that p is a constant. Therefore. I = y −x Io where recall that y can be expressed alternatively as y = elny .

if one lets ∆I and ∆x become infinitesimally small one gets dI dx = −αI (12. . the absolute change in intensity on attenuation is also proportional to the intensity of the light.3) This is our exponential attenuation law again.4) where α is the “absorption coefficient” and is generally expressed in units of cm−1 when l is expressed in cm units.2) which is a first order differential equation that is easily solved. (b) Next. the change in intensity is also proportional to the product of both independent variables ∆I ∝ −I∆x Call the proportionality constant α giving ∆I = −αI∆x ∆I = −αI ∆x Next. In bulk solids it is common to see I = Io e−αl (12. ∆I ∝ I (c) As a consequence. Conventions In dealing with the absorption of materials there are various conventions.167 The negative sign is to indicate attenuation of the light in the solid through absorption. dI I lnI = −αdx = αx + const or I = Io e−αx (12.

By comparison one can see that if one needs to relate α and σ α = σn (12. ABSORPTION In more molecular-like systems it is common to see I = Io e−σnl (12.  is the molar extinction coefficient in units of (M −1cm−1 ). A = cl (12. logeσnl = cl (σnl)loge = cl . Finally. c is the concentration in units of (M) and finally l is the pathlength in units of cm.7) where A is the “absorbance” or “optical density (OD)”. Thus A = log( I Io is the transmittance Io ) = cl I and from the exponential attenuation law we know that I = e−σnl Io Note that c is in units of M and n is in units of number/cm3 . when dealing with molecular-like materials it is common to see the everyone’s favorite Beer’s law.168 CHAPTER 12.6) but this shouldn’t be necessary since in one case one has a bulk solid and in the other case the assumption is of a molecular-like material (quantum dots for example).5) where σ is the “cross section” having units of cm2 and n is the concentration of absorbers in the sample in units of number/per cm3 and l is the pathlength expressed in units of cm. Sometimes it is of interest to convert between extinction coefficient () and the absorption cross section (σ). This can be achieved in the following fashion A = cl where we also know that A = log( T1 ) and where T = of the sample.

5 nm and with a power of P = 10nW . c(mol/l) → cNa =n 1000cm3 therefore c= n1000 Na This means that our previous equivalence becomes (σn)loge = n 1000 Na or σ= σ= where the 2. A solution with a dye concentration of c = 4 × 10−6 M has a molar 1 absorptivity of  = 1.8) 1 loge . It is illuminated with green laser light at λ = 514. .5 × 105 M ·cm . Next convert from dm to cm by recalling that 1dm = 10cm.303 comes from (1000) loge(Na ) (2. c(mol/l) → cNa dm3 where recall that moles times Avogadros number is the number of absorbers and that 1l = dm3 . (b) Calculate the number of photons per second absorbed by the sample. Exercises 1. (c) Calculate the cross section of the molecule in units of cm2 and ˚ A2 .169 Drop the l terms on either side since they are both the same.303)(1000) Na (12. (σn)loge = c Let’s now convert c to the same units as n which is in terms of number of absorbers per cm3 . (a) What is the percentage of light absorbed by the sample after 1 micron.

P? (c) Next. What is the incident excitation power. diIC12 is a popular dye in single molecule experiments. . ABSORPTION 2.5 nm (a green HeNe).170 CHAPTER 12. (b) This diI sample is then illuminated with 543. (a) If 10 microliters of a diI solution spread evenly on a 4 mm diameter region achieves this coverage. Colloidal CdSe QDs overcoated with ZnS are all the rage today. what is the molar concentration of the diI solution. a single diI molecule being illuminated absorbs 2. Look up a paper by XG Peng in Chemistry of Materials and from his data provide the range of absorption cross sections (σ) at the band edge for QD sizes between 2 nm and 6 nm in diameter. Let’s say no more than 10 molecules per square micron. 3. Express the same range in ˚ A2 and then compare this to the physical size of the dot. If one wants to image individual diI molecules one needs sufficiently low coverages on a microscope coverslip.3 × 105 p/s. A uniform 100 nm diameter spot is illuminated with an excitation rate of 3 × 1010 photons per second. Calculate the absorption cross section and the molar extinction coefficient of diI (a 1 cm cell may be assumed).

171 ov .

172 CHAPTER 12. ABSORPTION .

The transition which promotes an electron from the valence band to the conduction band through the absorption of a photon is referred to as an “interband” transition.1) where |M | is a matrix element linking the initial and final states and ρj () is the joint density of states. Now in what follows we will assume that we are dealing with direct gap semiconductors. The transition rate for absorption is given by Fermi’s (second) Golden Rule. Its basically a quantum mechanical way to derive a rate for a process and is analogous to a classical rate. 173 . given by Fermi’s Golden Rule is Ri→f = 2π 2 h |M | ρj () ¯ (13. Notice that the transition rate is proportional to the joint density of states which you have just calculated for bulk. Note this is also sometimes abbreviated and expressed as the “Golden rule”. These bands are composed of accessible states to the electron and hole. The initial state (i) to final state (f) transition rate. namely the conduction band and the valence band. 1D and 2D materials. transitions within a given band are called “intraband” transitions. Ri→f . Alternatively. We generally speak of electrons in the conduction band and holes in the valence band. there are so-called bands in semiconductors. 0D.Chapter 13 Interband transitions Interband absorption in semiconductors Although we haven’t discussed the concept of bands yet. Here is where your density of states calculations start to pay off.

174 CHAPTER 13. The general Schrodinger equation is actually written as i¯h 2 ∂ Ψ(r.4) You will note that the left hand side is dependent only on time and that the right hand side is dependent only on position.  2 2  ∂ ¯h ∇ i¯h (Ψf ) = − + V Ψf ∂t 2m i¯hΨ ¯ 2 ∇2 h Ψf + V Ψf 2m ¯h2 00 = − f Ψ + V Ψf 2m ∂f ∂t i¯hΨf = − 0 Consolidate the f and Ψ terms giving ¯h2 00 f =− Ψ +VΨ f 2m 0 i¯hΨ or 0 i¯h ff = 1 Ψ h i h2 00 ¯ Ψ +VΨ − 2m (13. We look for solutions of the form Ψ(r. INTERBAND TRANSITIONS Time dependent perturbation theory We will work in what is referred to as the Schrodinger representation. the interaction representation is preferred. Usually when doing time dependent perturbation theory. t) = Ψf . t) ∂t (13. Replace this into the general time dependent Schrodinger equation.3) To keep the notation more compact.2) 2 ∇ + V and V is a potential perturbation which can be time where H = − ¯h2m dependent. To be equivalent both sides must equal a constant which we call . Note for future reference that there are three picture of quantum mechanics. the Heisenberg representation and the interaction representation. But we will get the same expressions this way as well. t) = Ψ(r)f (t) (13. (turns out to be the eigenenergy or eigenvalue) . t) = HΨ(r. Note that the wavefunctions depend on time now. the Schrodinger representation. reduce this to Ψ(r.

the general form of the wavefunction is Ψ(r. Putting everything together. Solving this we get  and Ψ(r). t) = Ψ(r)e −it h ¯ (13.5) Right hand side   1 h2 00 ¯ Ψ +VΨ =  − Ψ 2m ¯ 2 00 h Ψ + V Ψ = Ψ 2m   h2 2 ¯ ∇ + V Ψ = Ψ − 2m − or H0 Ψ = Ψ (13. t) = Ψ(r)f (t) or more explicitly Ψ(r.175 Left hand side 0 i¯ h f f 0 f f lnf =  i h ¯ which we integrate to get it = − ¯h = − The desired time dependent function is f (t) = e− it h ¯ (13.7) .6) This is our usual eigenvalue and eigenvector problem.

t) = X Cn (t)e− (0) in t h ¯ Ψ(0) n (r) n = X Cn (t)Ψn(r. t)  (0) = H + H (1) Ψ(r. then in the absence of a perturbation we would have for all times (0) X in t Ψ(t > 0) = Cn e− h¯ Ψ(0) n (r) n where again Cn is a constant. t) = HΨ(r. t) dt where H = H (0) + V (t) or H (0) + H (1) (t). i¯h dΨ(r. the eigenfunctions appropriate to the problem would be given by the equation H (0) Ψ(0) = (0) Ψ(0) If the initial state is expanded in terms of these unperturbed eigenfunctions (0) X in t Ψ(t = 0) = Cn e− h¯ Ψ(0) n (r) n where Cn is a constant. Plug our time dependent form of the wavefunction back into the general time dependent Schrodinger equation.176 CHAPTER 13. Use whatever notation that suits you. If the perturbation is present however. We then have  dΨ(r. t) i¯ h dt . But we can still express whatever true solution exists as a linear combination of the un-perturbed zero order wavefunctions provided that the coefficients are now time dependent. INTERBAND TRANSITIONS Time dependent perturbation theory now The basic idea of time-dependent perturbation theory is simple. t) n So let’s go ahead and start deriving some results of the time dependent perturbation theory. Ψ(r. the above linear combination is no longer a valid solution to the Schrodinger equation. First understand that if the perturbation (call it V ) were absent.

t) = i¯h dCn n dt Ψn (r. t) dt (0) i t − 1h ¯ = i¯hC1 Ψ1 (r)e (0) i − 1 ¯h ! cancel common terms (0) H (0) Ψ1 (r) = 1 Ψ1 (r) which is our favorite eigenvalue and eigenvector relation. Furthermore. t) = i¯hC1 (0) i t − 1h ¯ H (0) C1 Ψ1 (r)e dΨ1 (r. t) n Cn Ψn (r. t) P (13. t) dΨn (r. t) dt dt n The first terms on the left and right side of the last expression will cancel. t) Cn Ψn (r. t) dt n n n Expand the right hand side using the Chain rule. t) = i¯h H +H dt n n " # X X X d (1) (0) Cn Ψn (r. The sample applies for all n so that we are left with H (1) P n Cn Ψn (r. " # h iX d X (0) (1) Cn Ψn (r. t) + H (1) Cn # n n = i¯ h X X dCn dΨn (r. t) X dCn Cn + Ψn (r. t) = i¯h Cn Ψn (r. H (0) X Cn Ψn (r. H (0) C1 Ψ1 (r. t) + H H Cn Ψn (r. t) + i¯h Ψn (r. . Just understand that the time dependence is implied.8) Now we desire to solve this expression for the time dependent coefficients of some state say k.177 Ok since the notation gets wild. we’re going to drop the (t) in Cn (t). To illustrate let n = 1 for example. t) + H (1) n = i¯ h " X n H (0) n X Cn Ψn (r. just work with Ψn (r. t) dt dt n X X Cn Ψn (r. t) instead of the full expression.

.. .. t) Ψk Ψk (0) in (0)∗ − h ¯ Ψk (r)Ψ(0) n (r)e where n = k on right side X Cn < k|H (1) n |n > e Multiply both sides by e X n n (0) (0) i t k h ¯ (0) dCn − ik e h¯ = i¯h dt t to simplify.. .. . . 3  dt   .. .9) differential equations.. t) = i¯h n X (0)∗ Cn Ψk n (r)H (1) Ψn (r. ..     . .  . . t) (0)∗ (r)Ψn (r.. . . . INTERBAND TRANSITIONS (0)∗ Multiply both sides by Ψk same basis are orthogonal.  C3   . Cn < k|H (1)|n > e− X Let ωkn = (0) t i − nh ¯ (0) i t k h ¯ = i¯h dCk dt (0) (0) i( −n )t k h ¯ = i¯h dCk dt (0) in t h ¯ Cn < k|H (1) |n > e e (0) k −n h ¯ since ∆ = h ¯ ω yielding X n Cn < k|H (1) |n > eiωkn t = i¯h dCk dt or better yet k i¯ h dC dt = P n Cn < k|H (1) |n > eiωkn t This is basically a system of simultaneous linear In matrix notation    C1 V11 V12 eiω12 t  C2  V21 e−iω12 t V22  d     i¯ h  C = . . . . . t) = i¯h n X X dCn X dCn n (0) i (0)∗ − nh ¯ Cn Ψk (r)H (1) Ψ(0) n (r)e t = i¯h n dt dt X dCn n dt (0)∗ (r)Ψn (r... (13. . (0)∗ Ψk X (r)H (1) (r) and recall that all states n 6= k of the Cn Ψn (r. Ck . .. .178 CHAPTER 13....   . C1 C2  . . .

Then C1 (t = 0) = 1 C2 (t = 0) = 0 C3 (t = 0) = 0 and so forth. Replacing this into the above expression gives Z 0 1 t 0 < k|H (1) |n > eiωkn t dt Ck (t) = Ck (0) + i¯ h 0 . . If there are no perturbations. suppose the system starts out in state 1. . Cn (t) = 0 To get our first order approximation insert these zeroth order values into our matrix yielding i¯ h dCk dt dCk dt = Cn=1 < k|H (1)|n = 1 > eiωkn=1 t = 1 Cn=1 < k|H (1)|n = 1 > eiωkn=1 t i¯ h where cn=1 ≈ 1 yielding Ck (t) = 1 i¯ h Rt 0 0 < k|H (1)|n = 1 > eiωkn=1t dt 0 (13..10) k Note however that when integrating dC dt there is actually a constant of integration that comes out. then the system stays in state 1 forever. Our zeroth order approximation is therefore C1 (t) = 1 C2 (t) = 0 C3 (t) = 0 . are). So to overcome this problem you apply time dependent perturbation theory (which is basically making some assumptions about the size of the perturbation and in turn what C1 .179 Now what? This expression is pretty tough to solve. . C3 . C2 . To illustrate.. Z 0 1 t 0 Ck (t) = const + < k|H (1)|n > eiωkn t dt i¯ h 0 where const = Ck (t = 0) = δkn .

t) = HΨ(r.12) (13.13) Two level system. TLS). t) dt i¯ h Now recall that a general solution to this time dependent Schrodinger equation can be expressed as a linear combination of the unperturbed zero order wavefunctions. t) + C2 (t)Ψ2 (r. t) = Ψ2 (r)e− (0) i1 t h ¯ (0) i t 2 h ¯ Now let’s be more specific with what H (1) (t) looks like. t) = Ψ1 (r)e− Ψ2 (r. We want to consider the interaction of a molecule with dipole moment µ with light where the electric field of light can be written as  = Eo cos2πνt . Schrodinger representation Let’s illustrate our results with a more concrete example and for simplicity keep it to two levels (a two level system. Recall that our time dependent Schrodinger equation is d Ψ(r. provided time dependent coefficients. Ψ(r. sinusoidal perturbation. t) = C1 (t)Ψ1 (r.180 CHAPTER 13. INTERBAND TRANSITIONS or your standard textbook expression Ck (t) = δkn + 1 i¯ h Summary Rt 0 0 < k|H (1)|n > e1ωkn t dt 0 (0) Ck (t) = δkn (1) Ck (t) = 1 i¯ h Rt 0 0 < k|H (1) |n > e1ωkn t dt 0 (13.11) (13. t) dt where H = H (0) + H (1) d i¯ h Ψ(r. t) = (H (0) + H (1) )Ψ(r. t) where Ψ1 (r.

t) into the Schrodinger equation.14) This is going to be our sinusoidal perturbation. t) + C2 Ψ2 (r. t)) dt dt (0) (1) = (H + H ) [C1 Ψ(r. t) dt dt dt dt = H (0) (C1 Ψ1 (r. t) + i¯hΨ2 (r. Insert Ψ(r. t) + C2 Ψ2 (r. t)) + H (1) (C1 Ψ1 (r.15) H = H (0) − µ · E0 cosωt (13. t) + C1 H Ψ1 (r. t) C1 i¯ h We’ve seen previously that the first terms on either side of the last equation will cancel since they are equivalent.181 where ν is the frequency of the radiation. t)] dt = (H (0) + H (1) ) [C1 ψ1 (r. t) dt dt dt dt (0) (0) (1) (1) = C1 H Ψ1 (r. dΨ1 (r. t) dΨ2 (r. t) + C2 Ψ2 (r. d [C1 Ψ1 (r. t) = i¯h .16) Such that We are now going to re-derive our previous expression for the coefficients. Also for shorthand let the time dependence of the coefficients be understood C1 (t) → C1 and C2 (t) → C2 . t) dC2 + i¯ hΨ1 (r. t) + C2 Ψ2 (r. t) + C2 H Ψ2 (r. t) + i¯hC2 + i¯hΨ2 (r. t)] i¯ h dΨ1 (r. t)) + i¯ h (C2 Ψ2 (r. Alternatively  = E0 cosωt (13. You can see that it will oscillate periodically with time. t) dC1 dC2 + C2 i¯ h + i¯hΨ1 (r. t) dC1 dΨ2 (r. t) = i¯h dt H (0) Ψ1 (r. t) + C2 Ψ2 (r. t) + C2 H Ψ2 (r. t) H (0) Ψ2 (r. t)] i¯ h d d (C1 Ψ1 (r. The interaction between a molecule’s dipole moment and the electric field is then H (1) = −µ ·  H (1) = −µ · E0 cosωt (13. t)) i¯ hC1 dΨ1 (r. t) dt dΨ 2 (r.

17) 2 1 hΨ2 (r.19) What now? Well. INTERBAND TRANSITIONS leaving (1) Ψ (r. t) − i¯h < 2|1 > e (0) i1 t h ¯ (0) i2 t dC dC1 2 + i¯h < 2|2 > e− h¯ dt dt = C1 < 2|H (1) |1 > e− − i¯ he (0) i2 t h ¯ dC2 dt = C1 < 2|H (1) − |1 > e (0) i1 t h ¯ (0) i1 t h ¯ + C2 < 2|H (1) |2 > e− + C2 < 2|H multiply both sides by e i¯ h dC2 dt = C1 < 2|H (1) |1 > e (0) as before let ω21 = (1) − |2 > e (0) i2 t h ¯ (0) i2 t h ¯ (0) i2 t h ¯ (0) (0) i(2 −1 )t h ¯ + C2 < 2|H (1) |2 > (0) 2 −1 h ¯ . t) (0)∗ i¯hΨ2 (r)Ψ1(r. dC2 dC1 (0)∗ + i¯hΨ2 (r)Ψ2(r. t) + C2 Ψ2 (r)H Ψ2 (r. t) dt dt (0)∗ (0)∗ (1) (1) = C1 Ψ2 (r)H Ψ1 (r.182 CHAPTER 13. The exact expression we get then is (1) |1 > eiω21 t + C < 2|H (1) |2 > 2 i¯ h dC 2 dt = C1 < 2|H (13. Recall that H (1) = −µ · Eo cosωt . t) dC 1 2 2 dt + i¯ dt = C1 H (0)∗ Now like before.18) Now we invoke time dependent perturbation theory and assume that since H (1) is small C1 (t) ' 1 C2 (t) ' 0 giving (1) |1 > eiω21 t 2 i¯h dC dt =< 2|H (13. multiply both sides by Ψ2 (r) and integrate. t) + C H (1) Ψ (r. t) dC i¯ hΨ1 (r. t) (13. we haven’t really dealt with H (1) yet.

Remember that when you integrate dC dt you get an additional constant of integration which is C2 (0).183 For convenience.   dC2 Eo i¯ h = − < 2|µz |1 > ei(ω+ω21 )t + e−i(ω−ω21 t) dt 2   dC2 iEo = < 2|µz |1 > ei(ω+ω21 )t + e−i(ω−ω21 )t dt 2¯ h Integrate this to find C2 (t) but before we do this.22) 2 Now integrate this to get C2 (t). dC2 dt dC2 i¯ h dt dC2 i¯ h dt i¯ h = < 2| − µz Eo iωt (e + e−iωt )|1 > eiω21 t 2 Eo < 2|µz (eiωt + e−iωt )|1 > eiω21 t 2  Eo = − < 2|µz |1 > eiωt+iω21 t + e−iωt+iω21 t 2 = − (13.  h¯ ω+¯hω21  h ¯ ω−¯ hω21 dC2 iEo = < 2|µz |1 > ei h¯ t + e−i h¯ t dt 2¯ h where ¯hω = hν and h ¯ ω21 = h ¯ ω2 − ¯hω1 = 2 − 1   hν+(2 −1 ) hν−2 +1 iEo dC2 t t h ¯ = < 2|µz |1 > ei + e−i h¯ dt 2¯ h dC2 dt = iEo 2¯ h   (2 −1 +hν) (2 −1 −hν) t t h ¯ h ¯ + ei < 2|µz |1 > ei (13. let’s flip back to  notation so that our final expression will be consistent with everyone elses. let’s throw this into our previous expression and see what happens.20) Ok. Z 0 0 i(2 −1 −hν)t 0 iEo < 2|µz |1 > t i(2 −1 +hν)t h ¯ h ¯ +e dt e C2 (t) = C2 (0) + 2¯ h 0 where C2 (0) = δ12 Z 0 0 i(2 −1 −hν)t iEo < 2|µz |1 > t i(2 −1 +hν)t 0 h ¯ h ¯ C2 (t) = δ12 + e +e dt 2¯ h 0 . choose the z direction Hz(1) = −µEo cosωt or (1) Hz = −µz Eo  eiωt +e−iωt 2  (13.21) The term < 2|µz |1 > is the transition dipole moment.

Under this resonance condition where the monochromatic light energy equals the energy difference between states 1 and 2 you get (1) C2 (t) ' Eo <2|µz |1> 2  e i(2 −1 −hν)t h ¯ −1 (2 −1 −hν)  (13. INTERBAND TRANSITIONS but since states 1 6= 2 we are left in practice with the 1st order contribution. (1) C2 (t) = iEo <2|µz |1> 2¯ h Rt 0 e i(2 −1 +hν)t h ¯ 0 +e i(2 −1 −hν)t h ¯ 0 dt 0 (13. " i(2 −1 +hν)t # i(2 −1 −hν)t h ¯ h ¯ (e E < 2|µ |1 > (e − 1)¯ h − 1)¯ h (1) o z C2 (t) = + 2 i(2 − 1 + hν) i(2 − 1 − hν) or (1) C2 (t) = Eo <2|µz |1> 2 " (e i(2 −1 +hν)t h ¯ −1) (2 −1 +hν) + (e i(2 −1 −hν)t h ¯ −1) (2 −1 −hν) # (13.26) . The one with (2 − 1 + hν) is called the nonresonant term.184 CHAPTER 13. Extract out e i(2 −1 −hν)t −i(2 −1 −hν)t ! i(2 −1 −hν)t 2¯ h 2¯ h e E < 2|µ |1 > − e o z (1) 2¯ h C2 (t) ' e 2 2 − 1 − hν −i(2 −1 −hν)t ! i(2 −1 −hν)t 2¯ h 2¯ h −e iEo < 2|µz |1 > i(2 −1 −hν)t e (1) 2¯ h e C2 (t) ' (2 − 1 − hν) 2i resulting in (1) C2 (t) ' iEo <2|µz |1> i(2 −1 −hν)t 2¯ h sin (2 −1 −hν) e  2 −1 −hν 2¯ h  t (13. So let’s just work with it.25) Note that this approximation is also called the “Rotating Wave Approximation”. let’s keep simplifying. another approximation is made here. You have two terms. The one with (2 − 1 − hν) in the denominator is called the resonant term. i(2 −1 −hν)t 2¯ h Since we like math.23) So go on and integrate this expression. When hν = (2 − 1 ) the resonant term will dominate.24) Now at this point.

29) H (1) = −µz Eo sinωt (13. let’s keep going. Recall that the probability of being in a particular state k is basically Pk = |Ck |2 Therefore the transition probability into the final state 2 is P2 = |C2 (t)|2   Eo2 | < 2|µz |1 > |2 2 2 − 1 − hν sin = t (2 − 1 − hν)2 2¯ h The final transition probability into state 2 is P2 = Eo2 | < 2|µz |1 > |2   − −hν)t sin2 ( 2 12¯h ) (2 −1 −hν)2  (13.28) since ¯hω = hν.30) Note that it doesn’t matter which one we use. we want to consider the transition probability and more specifically the rate of transition into a group of final states k (not just into one state). Like above. It depends on both time and frequency.27) Note that P2 is P2 (hν.185 Ok. t). Fermi’s Golden Rule Recall that we are dealing with a harmonic perturbation of the form H (1) = −µz Eo cosωt (13. In general express this as ω Pk = −ω Eo2 |<k|µz |n>|2 sin2 kn2 t (ωkn −ω)2 h2 ¯ (13. both work fine. From our previous work we derived Pk = Eo2 | < k|µz |n > |2 sin2 ωkn2−ω t (ωkn − ω)2 h2 ¯ where again ωkn = (ωk − ωn ). .

assume that ρ(ωk ) is constant or rather that the distribution is uniform.31) Pk.33) Again. INTERBAND TRANSITIONS So we integrate over the spread of final states dictated by the density ρ(k ) or ρ(ωk ).group = = Z Z ∞ −∞ ∞ −∞ Eo | < k|µz |n > |2 sin2 ωkn2−ω ρ(ωk )dωk (ωkn − ω)2 ¯h2 Eo | < k|µz |n > |2 sin2 ωk −ω2n −ω ρ(ωk )dωk (ωk − ωn − ω)2 ¯h2 Now as before.group = dx x2 ¯h2 −∞ As before.group . This is called Fermi’s Golden Rule. dt R= π 2 2¯ h Eo ρ(k )| < k|µz |n > |2 (13. the rate is constant with respect to time.group = πEo2 ρ(k ) | 2¯ h < k|µz |n > |2 t (13.group = −∞ Pk (ωk . flip ρ(ωk ) into ρ(k ) to keep the notation consistent with other texts (ρ(ωk ) = h ¯ ρ(k )). . Pk. R∞ Pk. It can be removed from the integral. t)ρ(ωk )dωk (13. Also remember to change the limits of integration.186 CHAPTER 13.group = Eo2 | < k|µz |n > |2 πt 2 ¯h2 Also like before.32) This is the desired form of the transition probability. rather than stop here. from a table of integrals the definite integral is Z ∞ sin2 xt πt 2 dx = 2 x 2 −∞ The resulting transition probabilty is then Pk. It is the quantum mechanical version of a classical rate. dP Now the desired rate is R = k. Z E 2 ρ(ωk )| < k|µz |n > |2 ∞ sin2 ωk −ω2n −ω t dωk Pk.group = o 2 ¯h2 −∞ (ωk − ωn − ω) Let x = (ωk − ωn − ω) and dx = dωk . We then get Z Eo2 ρ(ωk )| < k|µz |n > |2 ∞ sin2 xt 2 Pk.

For the interaction of light and matter we have seen the Hamiltonian H (1) = −µ · E One can also express this interaction through the following Hamiltonian H (1) = − q (A · P ) m where A is the vector potential associated with the magnetic field of the electromagnetic wave and P is the quantum mechanical operator. We carry out the following derivations in a solid state physics approach to stay consistent with existing texts. φ. Then q Ao eˆcos(ωt − kr)P m q e · P) = − Ao cos(ωt − kr)(ˆ m i qAo h i(ωt−kr) = − e + e−i(ωt−kr) (ˆ e · P) 2m H (1) = − Next it can be shown that generally the transition probability can be expressed as (this is best done through the interaction representation) . Proceed using the second Hamiltonian. It can be shown that A = Ao cos(ωt − kr) where we ignore a potential phase factor.187 Putting it all together Interband transitions (or a whole lot of work) There are various ways to proceed with this. and A = Ao eˆ where the second Ao is a scalar amplitude and eˆ is a unit vector showing the orientation of the electric field polarization.

2 .

Z .

1 .

.

t 0 iωkn t0 (1) < k|H |n >.

.

Pk = 2 .

dt e h ¯ to (13. .34) where k is the final state (the conduction band in our case) and n is the initial state (the valence band in our state).

188 CHAPTER 13. INTERBAND TRANSITIONS We will shove into this transition probability the above expression for H (1) to get .

Z  E.

.

2 0 1 .

.

t 0 iωkn t0 D  i(ωt0 −kr) −i(ωt −kr) Pk = dt e k| e +e (ˆ e · P )|n .

.

h2 .

to ¯  .

Z   E.

.

2 0 1 q 2 A2o .

.

t 0 iωkn t0 D  i(ωt0 −kr) −i(ωt −kr) dt e k| e +e (ˆ e · P )|n .

.

= h2 4m2 .

So to simplify life just drop the first term. ¯ to Next there are two terms sandwiched by < k| and |n >. The one on the left will eventually lead to a “non-resonant” term related to stimulated emission. The second term on the right will lead o the absorption term that we are interested in. This is the “resonant” term. These are the 0 0 terms ei(ωt −kr) and e−i(ωt −kr) . We now have   .

Z E.

.

2 1 q 2 A2o .

.

t iωkn t0 D  −i(ωt0 −kr)  e k| e (ˆ e · P )|n .

.

Pk ' 2 4m2 .

  . h ¯ to Pull out the time dependent term from the bra-ket since this is only a spatial integral.

Z E.

.

2 1 q 2 A2o .

.

t 0 iωkn t0 −iωt0 D ikr Pk ' dt e k|e (ˆ e · P )|n .

.

h2 4m2 .

to ¯   .

Z E.

.

2 1 q 2 A2o .

.

t 0 i(ωkn −ω)t0 D ikr dt e k|e (ˆ e · P )|n .

.

' h2 4m2 .

¯ to Next we have the Bloch forms of the wavefunctions 1 |k > = √ uk (r)eikk r N 1 |n > = √ un (r)eikn r N where √1N is a normalization constant reflecting normalization of both wavefunctions over the entire volume of the solid being considered. . N is the number of unit cells in the crystal considered.

2  .

Z  Z .

1 q 2 A2o .

.

t 0 i(ωkn −ω)t0 1 −ikk r ikr ikn r 3 .

u e e (ˆ e · P )u e d r dt e Pk ' 2 n k .

.

. 4m2 N V h ¯ to where the interior integral is over the entire volume of the solid considered.

2    .

Z Z .

1 q 2 A2o 1 .

.

t 0 i(ωkn −ω)t0 −ikk r ikr ikn r 3 .

Pj ' 2 u e e (ˆ e · P )u e d r dt e n k .

.

4m2 N2 h ¯ V to .

189 Now in principle. we (as an approximation) ignore P operating on eikn r . P operates on both un and eikn r but since we understand that eikn r varies very slowly. .

2   .

Z  Z .

1 .

.

t 0 i(ωkn −ω)tt 1 q 2 A2o i(k+kn −kr )r 3 .

e u (ˆ e · P )u d r dt e Pk ' 2 n k .

.

4m2 N2 h ¯ V to Next ei(k+kn −kr )r is a slowly varying term that can be approximated by a constant in a given unit cell. We approximate this now as .

2    .

Z Z .

1 q 2 A2o 1 .

.

t 0 i(ωkn −ω)t0 X i(k+kn −kr )r 3 .

Pk ' 2 dt e e u (ˆ e · P )u d r .

.

n k .

4m2 N 2 .

to h ¯ V N We take what is sometimes called the long wavelength approximation eikr ' 1 and get .

2   .

Z .

1 q 2 A2o 1 .

.

0 i(ωkn −ω)t0 3 .

Pk ' 2 dt e N uk (ˆ e · P )und r.

.

2 2 4m N h ¯ V Next cancel common terms and choose a projection for P . say Px (ˆ e·P ) = Px giving .

2   .

Z .

1 q 2 A2o .

.

t 0 i(ωkn −ω)t0 dt e < k|Px|n >.

.

Pk ' 2 .

2 4m h ¯ to Extract the spatial integral since it does not depend on time. .

Z t .

2   0.

.

0 i(ω 1 q 2 A2o 2.

−ω)t .

| < k|Px|n > | .

dt e kn Pk ' 2 .

4m2 h ¯ to where Z t 0 0 i(ωkn −ω)t dt e to Let to = 0 to get 0 .

t ei(ωkn −ω)t .

.

= .

i(ωkn − ω) .

to ei(ωkn −ω)t − 1 i(ωkn − ω) Re-express this as e i(ωkn −ω)t 2  i(ωkn −ω)t  −i(ωkn −ω)t 2 2 −e e i(ωkn − ω) .

INTERBAND TRANSITIONS or as 2e i(ωkn −ω)t 2 sin  (ωkn −ω)t 2 (ωkn − ω)  t Put it all together now to get 1 Pk ' 2 h ¯  q 2 A2o 4m2o  This simplifies to 1 h2 ¯ Pk ' .190  CHAPTER 13.

.

.

i(ω −ω)t sin ωkn −ω  t .

2 kn .

.

2 2 | < k|Px |n > |2 .

2e .

.

(ωkn − ω) .

first convert the transition probability over from frequence to energy using  = h ¯ω    2 kn − t 1 q 2 A2o 2 sin 2¯ h Pk () ' 2 | < k|Px |n > | ¯h2 m2 (kn − )2 h ¯ giving Pk () '  q 2 A2o m2  | < k|Px|n > | 2 2 sin (kn kn −  t 2¯ h − )2 (13.37)  .36) Next at long enough times (t → ∞). However. we can simplify the last term above into a delta function using limt→∞ sin2 at = δ(a) πa2 t So that  πt 4¯ h2 " π giving Pk () '  kn −  t 2¯ h  2 kn − t 2¯ h sin2 q 2 A2o m2  πt 4¯ h2  # πt → 2δ 4¯ h  kn −  2¯ h 2 | < k|Px|n > | δ   kn −  2¯ h (13.  q 2 A2o m2  | < k|Px|n > |2 ωkn −ω 2 (ωkn −ω)2 sin2 t (13.35) Now to account for the available states we eventually integrate over the joint density of states.

where δ kn 2¯ h The desired transition probability is 2π h ¯ Pk '  q 2 A2o 4m2  | < k|Px|n > |2 δ(kn − )t (13. . multiply the above expression by two 1 ρj. R. we integrate the rate expression over the joint density of states Rtot = 2π h ¯  q 2 A2o 4m2  | < k|Px|n > |2 In the 3D bulk case we have 1 ρj.41) This is the total transition rate per unit volume of a bulk 3D material including spin degeneracy.191  − = 2¯ hδ(kn − ).40) If we account for the degeneracy of the spin. R = giving 2π h ¯ R=  q 2 A2o 4m2  | < k|Px |n > |2 δ(kn − ) dPk dt (13. is the time derivative of the probability.38) The transition rate.3D () = 2 2π  2µ ¯h2 3 2 Shove this into the above expression for the total transition rate (per unit volume) to get Rtot ' '  3 2µ 2 p 1  − g δ(kn − )d | < k|Px|n > | 2 2π ¯h2      3 Z 2π q 2 A2o 1 2µ 2 p 2  − g δ(kn − )d | < k|P |n > | x 2π 2 h2 3m2o ¯ ¯h2 2π h ¯  q 2 A2o 4m2o  2 Z The final expression we have been waiting for is Rtot = 2π h ¯  q 2 A2o 4m2o  | < k|Px|n > |2  1 2π 2   2µ  32 √  − g h2 ¯  (13.39) Finally.3D () = 2 4π  2µ ¯h2 3 2 R ρj ()δ(kn − )d p  − g p  − g (13.

α= 2¯ hω Rtot nco ω 2 A2o . The number of photons per unit area per unit time is np = I ¯ω h (13. I= nr co ω 2 A2o nr co E 2 = 2 2 (13. I. replace it into the expression for α above.43) Next we need an alternative expression for the incident intensity.192 CHAPTER 13.44) This was obtained from I=  1 E×B µo  but in many cases (nonmagnetic materials) µ ' µo giving   1 E ×B I= µ where B = nc E. INTERBAND TRANSITIONS The absorption coefficient Now let’s close the loop and connect this total rate to the absorption coefficient.42) and the absorption coefficient can be expressed as α= ¯ ωRtot h Rtot = np I (13. (13.45) Then using this alternative expression for I. This leads to     1 n E2 I = µ c 1 n 1 2 E = µ c 2 o 1 and finally cµ = and since c2 = o1µo we have c = co1µo or co = cµ1o ' cµ Replace the last expression into our main equation to get I= nco ω 2 A2o 2 1 co .

48) The absorption spectrum therefore is shown to be proportional to the joint density of states.193 After simplifying this you get α=  q2 2πm2o co  1 nω  2µ ¯h2 3 2 | < k|Px |n > |2 Often there is a change of notation. This is the 3D case.46) (13. The final expression is α3D () =  q2 ¯ h 4πmo co n  Ep hω ¯  2µ h2 ¯ 3 √ 2  − g (13.47) which has a value of approximately 20 eV for many semiconductors. . Let Ep = p  − g 2|<k|Px |n>|2 mo (13.

194 CHAPTER 13. INTERBAND TRANSITIONS .

Chapter 14 Emission In the last section. absorption with total rate: B12 ρN1 2. Preliminaries: Einstein A and B coefficients Picture a two level system like that shown in the figure. a complementary process to absorption. stimulated emission with total rate: B21ρN2 195 . alternatively probability of being in the excited state • g1 =degeneracy of ground state • g2 =degeneracy of excited state • ρ =the energy density containing thermal and/or external contributions Three processes were considered 1. In turn. Here • N1 =population in the ground state or alternatively probability of being in the ground state • N2 =population in the excited state. the absorption coefficient of materials with different dimensionality were shown to be proportional to the calculated joint density of states. one could predict that the absorption spectrum of these materials would look like. Here we work out the background behind spontaneous emission.

196 CHAPTER 14.1: Two level system showing various transitions considered for Einstein A and B coefficients . EMISSION Figure 14.

the upward and downward rates are equivalent ( dN dt = 0).197 3. So using either of the above expressions dN2 dt = B12 ρN1 − B21ρN2 − AN2 = 0 or B12 ρN1 (upward) = B21 ρN2 + AN2 (downward) solving for ρ gives ρ = = AN2 B12N1 − B21 N2 A N1 B12 N2 − B21 where N1 and N2 are Boltzman distributed N1 = g1 1 X − ε1 −εF g1 ε1 −ε0 e kT = e− kT N N i=1 N2 = g2 1 X − ε2 −εF g2 ε2 −ε0 e kT = e− kT N N i=1 and N is a normalization constant. spontaneous emission with total rate: AN2 Putting everything together one obtains the following rate equations for populating either the ground or excited states dN1 dt dN2 dt = −B12 ρN1 + B21 ρN2 + AN2 = B12ρN1 − B21 ρN2 − AN2 1 In equilibrium. This leads to g1 ε1 −ε2 N1 g1 hν = e− kT = e kT N2 g2 g2 where hν = ε2 − ε1 is the energy of the photon or transition.1) . Replace this into the above expression for ρ giving ρ(ν) = A hν B12 gg21 e kT − B21 (14.

EMISSION This is the expression for the energy density derived by Einstein. Note that in general. Expressed in terms of ν rather than λ. the Planck energy density is (derived in the next section) ρp (ν) = 8πhν 3 c3  1 hν e kT −1  (14. the Planck distribution is (standard textbook expression) ρp (λ) = 8πhc λ5  1 hc e λkT −1  (14.2) Units: J/(unit volume unit wavelength). the index of refraction must be taken into account when considering the speed of light.198 CHAPTER 14. Expressed in terms of wavelength. This expression is interesting from a historical point of view because when the derivative of this expression is set to zero one obtains the Wein displacement law for blackbody radiation. in the absence of vacuum. though not necessary. Likewise the integral of this expression gives what’s called the Stefan Boltzman law for blackbody radiation. just for convenience) and kx = ky = kz = 2π Lx 2π Ly 2π Lz . knowing or unknowingly.3) Units: J/(unit volume unit frequency). imagine a sphere of radius k with volume Vk = 4 3 πk 3 where the volume of a given mode is kxky kz (assume periodic boundary conditions. Next he realized that this expression had to be equivalent to the Planck distribution for blackbody radiation. Now back to the main discussion. Interlude: Derivation of Planck energy density Just like in the density of states section. In a sense. Apparently he nearly suffered a nervous breakdown doing it. Planck basically explained everything in one shot.

4) Here is where we diverge a little from the past.5) hν < ε >= e hν kT −1 Note the Bose Einstein form of the equation (photons are bosons). Therefore 2πν c if the index of refraction is different from 1. Then ρ k= alternatively 2πν nc becomes 8π ν 3 k3 = 3π 2 3 c3 Now like before. To get the frequency density.199 The number of modes in the given sphere is then (look familiar?) N = 4 πk 3 Vk = 3 3 Lx Ly Lz kx ky kz 8π For a photon now. to get the frequency density (alternatively called the mode density) ρ= dρ 8πν 2 = 3 number per unit volume per unit frequency dν c Planck next showed that the average energy per mode is 0 ρ = (14. c first let k = 2π λ where λ = ν . multiply this by 2 to account for two possible polarizations 4 πk 3 = 2N = 2 3 3 Lx Ly Lz 8π k3 Lx Ly Lz total number of modes = 3π 2 Consider the volume density N 0 0 N k3 k3 ρ= = 2 = 2 number of modes/unit volume Lx Ly Lz 3π 3π (14. This leads to the Planck distribution 0 ρp (ν) = ρ < ε >   hν 8πν 2 = hν c3 e kT − 1   1 8πhν 3 = hν c3 e kT − 1 .

The volume is V = 4πk 2dk where kx . kz all equal nπ Lx.y.200 CHAPTER 14. We get . N is then = 4πk 2dk kx ky kz = 4πk 2dk Lz Ly Lz π3 Since there are two polarizations.z . N= 8πk 2 dk Lx Ly Lz π3 Now divide by 8 since each corner of an imaginary cube is shared by 8 other neighboring cubes. EMISSION Alternative derivation Consider the volume of a thin shell in k space. ky . N = k2 dkLxLy Lz π2 Now get the volume density by dividing by Lx Ly Lz giving 0 ρ= where k = 2π λ = 2πν c k 2 dk N = Lx Ly Lz π2 4π 2 ν 2 c2 such that k 2 = ρ = = 4πν 2 (2π)dν c3 8πν 2 dν c3 Now consider the frequency density 0 ρ dν = ρ such that 0 ρ = and also dk = 8πν 2 c3 2π c dν. multiply by 2. N The number of modes.

in terms of wavelength. The first.6) More usually you will see g1 = g2 such that the expression reduces to B12 = B21. Word of caution The Planck distribution is often written a number of ways. The above final expression are what are referred to as the relationship between Einstein A and B coefficients. Upon (14. has units of: number per unit volume per unit wavelength. Different texts will have what on the surface appear to be completely different expressions.201 Now back to where we left off.7) Again. Alternatively sometimes what people mean is the same expression but in units . note that if g1 = g2 you will commonly see written A= 8πhν 3 B12 c3 (14. has units of: number per unit volume per unit frequency. The second.8) Also if the index is not 1 (not vacuum) then replace c with nc in the above expressions. in terms of frequency. These difference actually arises because of differing definitions for the “density” being used by the various authors. to achieve equivalence B12 g1 = B21 g2 c n. Previously we saw two expression for the Planck distribution. Equate the Einstein and Planck relations   1 A 8πhν 3 = hν hν c3 B12 gg21 e kT − B21 e kT − 1 where note again that if the index is not 1 then replace c with examination of the above equation. Remember that A is the one associated with spontaneous emission. Furthermore 8πhν 3 A = B21 c3 yielding the standard textbook expression A = = 8πhν 3 B21 c3 8πhν 3 g1 B12 c3 g2 (14.

This is potentially very confusing. EMISSION of: number per unit volume per unit energy.202 CHAPTER 14. c3 h3 We (14. Rearrange the expression to 8πn3 ν 3 ρ= 3c3  h3 h3  = 8πn3 (hν)3 8πn3 ε3 = 3c3 h3 3c3 h3 Now as before take the derivative with respect to energy to get the energy density 0 ρ = dρ dε 8πn3 (3ε2 ) 3c3 h3 8πn3 ε2 c3 h3 8πn3 (hν)2 c3 h3 8πn3 ν 2 c3 h = = = = 0 Some authors will leave it at the second step which is just ρ = will go with 0 ρ = 8πn3 ν 2 c3 h 8πn3 ε2 .10) Now earlier we had the Planck derived average energy to be < ε >= hν e hν kT −1 This leads to an expression for the average number of photons which is < p >= 1 <ε> = hν hν e kT − 1 (14. One should look very carefully at what is meant by “density”. replace c with nc ).9) where this expression comes from the previous section where we derived the Planck distribution and where also the index of the medium has been explicitly considered (recall.11) . Starting with ρ= 8πν 3  n 3 3 c (14. Here we derive this third form of the Planck distribution with units: number per unit volume per unit energy.

stimulated emission and spontaneous emission. Let . let’s rederive the expression is a slightly different manner. In this section. but in a way that will be useful a little later on.203 The resulting Planck photon density is   1 8πn3 ν 2 ρp = number per unit volume per unit energy hv c3 h e kT − 1 (14. Number per unit volume per unit energy (as done here) or number per unit volume per unit frequency (as done previously).12) This then leads to another set of relationships between Einstein A and B coefficients B12 = B21 (14. Einstein A and B coefficients revisited In the last section we derived the relationships between the Einstein coefficients for absorption.14) The only difference between these relations and the previous ones was how the Planck distribution was defined. Let R12 be the unit transition rate from the ground state to the excited state (basically the rate constant) R12 = Pabs ρdε ≡ B12 ρ where • ρ is the number of photons per unit volume per unit energy (note the units!) • ρdε is the number of photons per unit volume • Pabs is the probability for absorption per unit time • R12 is the absorptions per unit volume per unit time and where B12 = Pabs dε As before set up the rate equations except now consider explicitly the probability of occupied and unoccupied states in the valence and conduction bands.13) A= 8πn3 ν 2 c3 h B12 or = 8πn3ε2 c3 h3 B12 (14.

abs = B12ρf1 (1 − f2 ) 2 → 1 transition (stimulated emission) requires • valence band state empty (1 − f1 ) • conduction band state occupied (f2) resulting in the joint probability being f2 (1 − f1 ) such that R21.spont = Af2 (1 − f1 ) In all three cases f1 and f2 are Fermi Dirac distributions f1 = f2 = 1 e ε1 −εF kT +1 1 e ε2 −εF kT +1 . EMISSION • f1 =probability of occupied valence band state • f2 =probability of occupied conduction band state • 1 − f1 =probability of unoccupied valence band state • 1 − f2 =probability of unoccupied conduction band state 1 → 2 transition (absorption) requires • valence band state occupied (f1) • conduction band state empty (1 − f2 ) resulting in the joint probability being f1 (1 − f2 ) such that R12.stim = B21 ρf2 (1 − f1 ) 2 → 1 transition (spontaneous emission) requires • valence band state empty (1 − f1 ) • conduction band state occupied (f2) resulting in the joint probability being f2 (1 − f1 ) such that R21.204 CHAPTER 14.

At equilibrium the upward and downward rates equal resulting in R12.abs = R21. A= .205 At this point. assume that g1 = g2 .stim + R21.spont B12 ρf1 (1 − f2 ) = B21 ρf2 (1 − f1 ) + Af2 (1 − f1 ) Rearrange this to solve for ρ ρ[B12 f1 (1 − f2 ) − B21 f2 (1 − f1 )] = Af2 (1 − f1) which gives ρ = = Af2 (1 − f1 ) B12 f1 (1 − f2 ) − B21 f2 (1 − f1 ) A (1−f2 ) B12 ff21(1−f − B21 1) Now introduce the explicit expression for f1 and f2 . for simplicity. To simplify    1 1 1 − ε2 −εF ε1 −εF f1 (1 − f2 ) e kT +1 e kT +1     = f2 (1 − f1 ) 1 1 1 − ε1 −εF ε2 −εF e = e kT +1 e kT +1 ε2 −ε1 kT hν = e kT Replace this into the main expression for ρ to get ρ= A (1−f2 ) B12 ff21 (1−f 1) − B21 = A B12 e hν kT − B21 But ρ equals the Planck distribution for photon density   8πn3 ν 2 1 A = ρ= hν hν c3 h B12e kT − B21 e kT − 1 meaning that for this to be true B12 = B21 and 8πn3 ν 2 B12 c3 h which are exactly the same Einstein A and B relations we found before.

stim = B21 ρf2 (1 − f1 ) (transition rate per unit volume) such that the net upwards rate is Rnet = R12.16) Equating the two expressions for ρ gives Pabs ρdε = B12 ρ[f1 (1 − f2 ) − f2 (1 − f1 )] Pabs dε = B12 [f1 (1 − f2 ) − f2 (1 − f1 )] Here we relate Pabs to the absorption coefficient as follows c Pabs = α n How? If P is the absolute probability for an absorption event and   dP dP dz = dt dz dt c Pabs = α n dP dt ≡ Pabs . EMISSION Emission spectrum Here we will calculate the emission spectrum of a 3D material using the Einstein A and B coefficients. Rnet = Pabs ρdε (14.206 CHAPTER 14. Define the (net) unit transition rate from the ground state to the excited state.stim (transition rate per unit volume) = B12 ρf1 (1 − f2 ) − B21 ρf2 (1 − f1 ) Since we have already shown that B12 = B21 Rnet = B12 ρ[f1 (1 − f2 ) − f2 (1 − f1 )] (14.abs = B12 ρf1 (1 − f2 ) (transition rate per unit volume) R21.abs − R21.15) where as before • ρ is the number of photons per unit volume per unit energy • ρdε is the number of photons per unit volume • Pabs is the probability for absorption per unit time • Rnet is the absorptions per unit volume per unit time Now consider the net rate upwards from 1 to 2 including stimulated emission R12.

207 Insert this Pabs expression into the above equivalence to get c α dε = B12 [f1 (1 − f2 ) − f2 (1 − f1 )] n Rearrange this to solve for B12 c dε B12 = α n [f1 (1 − f2 ) − f2 (1 − f1 )] Relate B12 to A using our derived Einstein A and B coefficients.spont = Af2 (1 − f1 ) such that 8πν 2  n 2 f2 (1 − f1 )dε h c [f1(1 − f2 ) − f2 (1 − f1 )]   2 2 dε 8πν n = α f (1−f 1 2) h c −1 R21. A = = 8πn3 ν 2 dε 8πν 2  n 3  c  α B = 12 c3 h h c n [f1 (1 − f2 ) − f2 (1 − f1 )] dε 8πν 2  n 2 α h c [f1 (1 − f2 ) − f2 (1 − f1 )] Now recall that R21.spont = α f2 (1−f1 ) Now for convenience define rspont(ε)dε = R21 where rspont (ε) is the transition rate per unit volume per unit energy. By inspection rspont (ε) = α 8πν 2  n 2 h c 1 f1 (1−f2 ) f2 (1−f1 ) and f1 = f2 = 1 e ε1 −εF kT +1 1 e ε2 −εF kT +1 −1 .

This leads to our final expression for the emission spectrum √  n 2 hν A ε − εg (8πν 3 ) e− kT I3D = αexc c 0 I3D = A hν √ ε − εg e− kT (14. I3D (ε) = ε rspont (ε) αexc (ε) where αexc (ε) is the absorption coefficient at the particular excitation position.  2  hν n 2 − kT α 8πν e h c I3D (ε) = (hν) αexc (ε)  n 2 hν α (8πν 3) e− kT = (14.17) We can stop here or continue to express this as an intensity.19) .18) αexc c Finally. EMISSION We’ve solved the ratio in the denominator before giving hν f1 (1 − f2 ) = e kT f2 (1 − f1 ) Putting everything together rspont (ε) = α  8πν 2 h     n 2 1 hν c e kT − 1 hν If hν > kT the term in parenthesis ∼ e− kT and the main expression becomes rspont (ε) = α  8πν 2 h   n 2 − hν e kT c (14. This value is a constant.208 CHAPTER 14. for a 3D material the absorption coefficient is proportional to the joint density of states p α ∝ A ε − εg where A is a constant.

2: Derived emission spectrum for a 3D material using Einstein A and B coefficients Figure 14. . Only the (downward) radiative transition is considered.209 Figure 14.3: Two level system considered in pulsed laser experiment.

210 CHAPTER 14. Basically this is a pulsed experiment. Imagine populating the excited state of the system and immediately (i. . These include energy transfer or non-radiative decay through defect states. there are other de-excitation pathways.20) 1 is called the lifetime of the The decay will be exponential and τ = krad excited state. EMISSION Quantum yields and lifetimes Here we discuss some complementary aspects to the emission. . instantaneously) turning off the light. In general. The depopulation of the excited state occurs by spontaneous emission since stimulated emission only occurs in the presence of the excitation.e. since we don’t live in a perfect world. giving dN2 dt = −(krad + k1 + k2 + k3 + . . . The relevant rate equation is dN2 = −AN2 dt or if A = krad dN2 = krad N2 dt dN2 = −krad dt N2 ln N2 = −krad t + (const) N2 = Ce−krad t At t = 0 C = N2 (0) resulting in N2 (t) = N2 (0)e−krad t (14. So in general the total decay rate out of the excited state is the sum of all rates ktot = krad + k1 + k2 + k3 + . )N2 ! n X ki N2 = − krad + i = −ktot N2 . however.

211 Figure 14.4: Linear and semilogarithmic plots of the excited state decay profile. .5: Linear and semilogarithmic plots of the excited state decay in the absence and presence of multiple decay pathways. Figure 14.

The first part of this chapter described three processes: (stimulated) absorption. What are the Wein and Stefan Boltzman laws anyway? . Compare this to the 1/e lifetime. 3. 5. 2. Rationalize why there isn’t a fourth process called spontaneous absorption. Show that at 300 K. and spontaneous emission. Which mechanism dominates for visible light.5. For example. fluorescent tags and so forth. Show mathematically how one extracts the average decay time of an exponential process. thermal stimulation dominates for frequencies well below 5x1012 Hz whereas spontaneous emission dominates for frequencies well above 5x1012 Hz. Exercises 1. 4. 7. stimulated emission. 6. See Figure 9. Therefore. Spontaneous emission competes with thermally stimulated emission (Planck radiation is the source). Show how the Plank expression reduces to both the Wein displacement law and the Stefan Boltzman law for blackbody radiation.21) For most applications. light emitting diodes. the efficiency of emission is called the quantum yield (QY ) and is defined by QY = krad ktot or 1 τrad 1 τtot = τtot τrad (14. Assume an exponential decay and consider its pure radiative decay as well as its radiative decay in the presence of other non-radiative recombination pathways. EMISSION Only the krad pathway gives you emission. Explain how one measures the quantum yield of something in real life. Come up with an alternative means of measuring the quantum yield of a system based on its fluorecence decay profile.212 CHAPTER 14. Derive a relation between the half life (50% mark) and the lifetime of the state. one desires a QY that is as close to unity as possible (QY = 1). such applications could involve lasers.