Chapter 9. Electromagnetic Radiation

Similar documents
INTRODUCTION TO QUANTUM ELECTRODYNAMICS by Lawrence R. Mead, Prof. Physics, USM

Phys 622 Problems Chapter 5

Semi-Classical Theory of Radiative Transitions

Quantization of scalar fields

Maxwell s equations. based on S-54. electric field charge density. current density

CHAPTER 1. SPECIAL RELATIVITY AND QUANTUM MECHANICS

Physics 221B Spring 2018 Notes 34 The Photoelectric Effect

16. GAUGE THEORY AND THE CREATION OF PHOTONS

2 Canonical quantization

Time dependent perturbation theory 1 D. E. Soper 2 University of Oregon 11 May 2012

Notes on x-ray scattering - M. Le Tacon, B. Keimer (06/2015)

The Particle-Field Hamiltonian

Electric and Magnetic Forces in Lagrangian and Hamiltonian Formalism

in terms of the classical frequency, ω = , puts the classical Hamiltonian in the form H = p2 2m + mω2 x 2

Lecture notes for QFT I (662)

From Particles to Fields

B2.III Revision notes: quantum physics

H ( E) E ( H) = H B t

Microscopic electrodynamics. Trond Saue (LCPQ, Toulouse) Microscopic electrodynamics Virginia Tech / 46

Preliminaries: what you need to know

Light - Atom Interaction

Molecular spectroscopy

Maxwell s equations. electric field charge density. current density

Representation of the quantum and classical states of light carrying orbital angular momentum

PHY 396 K. Solutions for problems 1 and 2 of set #5.

van Quantum tot Molecuul


Mechanics Physics 151

B = 0. E = 1 c. E = 4πρ

ψ s a ˆn a s b ˆn b ψ Hint: Because the state is spherically symmetric the answer can depend only on the angle between the two directions.

The 3 dimensional Schrödinger Equation

Molecules in Magnetic Fields

Quantum Mechanics: Fundamentals

(a) Write down the total Hamiltonian of this system, including the spin degree of freedom of the electron, but neglecting spin-orbit interactions.

1 Fundamentals of laser energy absorption

4. Spontaneous Emission october 2016

FYS 3120: Classical Mechanics and Electrodynamics

The interaction of light and matter

Second Quantization: Quantum Fields

PHYSICAL SCIENCES PART A

CONTENTS. vii. CHAPTER 2 Operators 15

QUANTUM MECHANICS. Franz Schwabl. Translated by Ronald Kates. ff Springer

Quantum Physics III (8.06) Spring 2007 FINAL EXAMINATION Monday May 21, 9:00 am You have 3 hours.

PHY 396 K. Problem set #7. Due October 25, 2012 (Thursday).

Inverse Problems in Quantum Optics

Non-stationary States and Electric Dipole Transitions

C. Show your answer in part B agrees with your answer in part A in the limit that the constant c 0.

Quantization of the E-M field

2. The Schrödinger equation for one-particle problems. 5. Atoms and the periodic table of chemical elements

Optical Lattices. Chapter Polarization

Multipole Fields in the Vacuum Gauge. June 26, 2016

PHYS3113, 3d year Statistical Mechanics Tutorial problems. Tutorial 1, Microcanonical, Canonical and Grand Canonical Distributions

Light - Atom Interaction

Quantization of Scalar Field

List of Comprehensive Exams Topics

Part I. Many-Body Systems and Classical Field Theory

Atomic Structure and Processes

MODERN OPTICS. P47 Optics: Unit 9

Physics 5153 Classical Mechanics. Canonical Transformations-1

Quantum Electronics/Laser Physics Chapter 4 Line Shapes and Line Widths

Quantum Mechanics II Lecture 11 ( David Ritchie

The Helmholtz Decomposition and the Coulomb Gauge

a = ( a σ )( b σ ) = a b + iσ ( a b) mω 2! x + i 1 2! x i 1 2m!ω p, a = mω 2m!ω p Physics 624, Quantum II -- Final Exam

Fundamentals of Spectroscopy for Optical Remote Sensing. Course Outline 2009

d 3 k In the same non-relativistic normalization x k = exp(ikk),

Identical Particles. Bosons and Fermions

Quantization of the E-M field

6. Molecular structure and spectroscopy I

Second quantization: where quantization and particles come from?

Section 5 Time Dependent Processes

Approximation Methods in QM

Quantization of the E-M field. 2.1 Lamb Shift revisited 2.1. LAMB SHIFT REVISITED. April 10, 2015 Lecture XXVIII

Introduction to particle physics Lecture 3: Quantum Mechanics

Curves in the configuration space Q or in the velocity phase space Ω satisfying the Euler-Lagrange (EL) equations,

Quantum Light-Matter Interactions

Physics 139B Solutions to Homework Set 5 Fall 2009

2. Electric Dipole Start from the classical formula for electric dipole radiation. de dt = 2. 3c 3 d 2 (2.1) qr (2.2) charges q

arxiv:physics/ v3 [physics.gen-ph] 2 Jan 2006

Physics 221A Fall 1996 Notes 21 Hyperfine Structure in Hydrogen and Alkali Atoms

J10M.1 - Rod on a Rail (M93M.2)

( ) /, so that we can ignore all

Retarded Potentials and Radiation

Molecular Magnetic Properties

Lecture 6 Photons, electrons and other quanta. EECS Winter 2006 Nanophotonics and Nano-scale Fabrication P.C.Ku

E = 1 c. where ρ is the charge density. The last equality means we can solve for φ in the usual way using Coulomb s Law:

Lecture 3: Helium Readings: Foot Chapter 3

Module I: Electromagnetic waves

Quantum Dynamics. March 10, 2017

Spectral Broadening Mechanisms

Plan of the lectures

1 Lagrangian for a continuous system

STRUCTURE OF MATTER, VIBRATIONS AND WAVES, AND QUANTUM PHYSICS

Lecture 0. NC State University

PH575 Spring Lecture #26 & 27 Phonons: Kittel Ch. 4 & 5

1 The postulates of quantum mechanics

FREE ELECTRON LASER THEORY USING TWO TIMES GREEN FUNCTION FORMALISM HIROSHI TAKAHASHI. Brookhaven Natioanal Laboratory Upton New York, 11973

Set 5: Classical E&M and Plasma Processes

So far we have derived two electrostatic equations E = 0 (6.2) B = 0 (6.3) which are to be modified due to Faraday s observation,

Many-Body Problems and Quantum Field Theory

B(r) = µ 0a 2 J r 2ρ 2

Transcription:

Chapter 9. Electromagnetic Radiation 9.1 Photons and Electromagnetic Wave Electromagnetic radiation is composed of elementary particles called photons. The correspondence between the classical electric field and the quantum picture of photons is that the intensity of light is proportional to the number of photons. Properties of photon: (rest) mass m = 0, energy ε = ω = cp, momentum p = k, angular frequency ω = ck, spin s = 1, so that photons are bosons. Light has two choices of polarization, which are given the label λ (λ = 1, ). The frequency is the same one that is associated with an oscillating electric field, which satisfying the electromagnetic wave equation (in vacuum): Monochromatic electromagnetic wave: E t = c E, and B t = c B. (9.1) E(r,t) = Re[E(r)e iωt ] = E 0 cos(ωt k r + ϕ 0 ), B(r,t) = Re[B(r)e iωt ] = B 0 cos(ωt k r + ϕ 0 ). (9.) A field may have classical meaning only after averaging over some spatial volume Ω, for example, the Poynting vector (directional energy flux density) S = c 4π E B, Note: we use the Gaussian units throughout this class. On the other hand, S = c N ω Ω therefore, E 0 = N 4π ω Ω and S = c 4π E. (9.3) 0, where Ω is the volume of the normalization big box;, and the field is quantized. When N is very large, the classical electromagnetism works extremely well, i.e., you can treat macroscopic electromagnetic fields as being continuous. However, at the microscopic level, the electromagnetic fields must be quantized, leading to phenomena which are not consistent with the classical theory. Examples: (a) Spontaneous radiation; (b) Photoelectric effects. The classical theory of electromagnetism fails completely. We have treated the photoelectric effect semi-classically, but the semi-classical theory in which the field is continuous (NOT an operator) cannot explain the spontaneous radiation. 148

9. Gauge Invariance Let s begin with the classical theory. Then we use the correspondence principle to construct the quantum mechanical version of the electromagnetic theory, i.e., field quantization. Maxwell s equations E = 4πρ, B = 0, E = 1 c B t, B = 1 E c t + 4π c j. (9.4) Here ρ the charge density and j the current density. The problem is to convert the above classical equations into a quantum mechanical description, i.e., to derive the Hamiltonian (operator) at microscopic level compatible to Maxwell s equations at macroscopic level. There is an important theorem: any vector function of position can be written as the sum of two terms; one is the gradient of a potential and the other is the curl of a vector: S(r) = g + m = S l + S t. (9.5) Here S l and S t are longitudinal and transverse parts of S. Assuming B(r) has this form, then Eq. (9.4) becomes We can simply set g(r) = 0, so that Then If S = E + 1 c A t, one can write ( g + A) = g = 0. (9.6) B = A. (9.7) E + 1 c A t = 0. (9.8) ( φ + m) = 0 = ( m). (9.9) The above equation is satisfied if we set m = 0, therefore E = φ 1 A c t, (9.10) B = A. (9.11) When these two Eqs. [(9.10) and (9.11)] for E(r) and B(r) are put into Maxwell s equations (9.4), the equations for the scalar and vector potentials are φ + 1 c ( A) t = 4πρ, (9.1) 149

( A)+ 1 A c t + 1 c A t = 4π j. (9.13) c E and B are observables, while A and φ are mathematical tools for computing E, B and other observables. However, the four unknown functions (A x, A y, A z, φ) are not uniquely determined by E and B, since E and B remain unchanged under the gauge transformation: E 1 c A A + Λ (A + Λ) t and φ φ 1 c φ 1 c Λ t = 1 c Λ t, (9.14) A t φ, (9.15) B (A + Λ) = A, (9.16) where Λ(r,t) is an arbitrary scalar function. Thus it is necessary to impose one additional condition (or constraint), which is called gauge. Coulomb gauge is widely used in condensed matter physics and quantum chemistry: A = 0. (9.17) Coulomb gauge is also called transverse gauge because Eq. (9.17) implies that the longitudinal part of A (A l ) is zero. Coulomb gauge also implies that the scalar potential acts instantaneously: φ(r,t) = ρ( r,t) r r d 3 r, (9.18) because Eq. (9.1) becomes φ = 4πρ. (9.19) Using this identity, ( A) = A + ( A), (9.0) one can show that under Coulomb gauge, Eq. (9.13) for vector potential becomes where the transverse part of current density j t is defined by A 1 A = 4π c t c j (r,t), (9.1) t j t (r,t) = 1 4π j( r,t) r r d 3 r. (9.) 150

In the Coulomb gauge, since the vector potential A is purely transverse, it should respond only to the transverse part of the current, and the longitudinal component of A does not occur. Another widely used one (especially in quantum field theory) is the Lorentz gauge: A + 1 c which causes Eqs. (9.1) and (9.13) change to φ t = 0, (9.3) φ 1 ψ = 4πρ, (9.4) c t A 1 A = 4π c t c j. (9.5) In the Lorentz gauge, both the vector and scalar potentials obey the retarded wave equation. They combine to produce a four-vector that is invariant under a Lorentz transformation. Another gauge that is often used is the condition that the scalar potential is set equal to zero, φ(r) = 0. In this case it is found that the longitudinal vector potential is not zero, which leads to an interaction between charges, the instantaneous Coulomb interaction. We use Coulomb gauge in this course, so that the Coulomb interaction is un-retarded. 9.3 Semi-Classical Approximation The choice of gauge won t affect E and B, but it affects the Hamiltonian. Classically, H = p m H = 1 m p e c A +eφ, (9.6) with e the charge (negative for an electron). Semi-classically, Ĥ = 1 m ˆp e c A +eφ. (9.7) Here A is a classical (continuous) field. As usual in the following we omit ^ except for the case in which confusion might rise. The quadratic term e mc A can be omitted for linear optical properties. In addition, we consider the case of the free electromagnetic field ( ρ = 0, j = 0 ), in which the gauge invariance allows we choose the simplest Coulomb gauge: 151

A = 0; φ = 0. (9.8) Then H = 1 m p e c A p e c p A = p m e mc A p. (9.9) Here H = H 0 + H cr, and the charge-radiation interaction Hamiltonian. H cr = e mc A p. (9.30) Within electric dipole approximation, i.e., for electromagnetic wave with small wave number k (long wavelength), Ĥ cr = ee r. (9.31) 9.4 Hamiltonian in Classical Mechanics Field (second) quantization follows the same approach used to quantize energies of bound states of particles employing Bohr s correspondence principle. In classical mechanics, H(q i, p i,t) = (p i q i ) L(q i, q i,t). (9.3) i Here L =T V is the Lagrangian, with T and V kinetic and potential energies, respectively. q i are generalized coordinates, q i = dq i dt, and the generalized momenta, which are also called the conjugate momenta. p i = L(q i, q i,t) q i, (9.33) Consider a particle in a conservative force field and use the Cartesian coordinates as generalized coordinates, i.e., q i = x i (i = 1,,3), p i = m x i, and Then in quantum mechanics, H = 1 m 3 p i +V(x i ). (9.34) i=1 Ĥ = 1 m ˆp i + ˆV(ˆx i ). (9.35) In construction of classical Hamiltonian, we need canonical coordinates q i and p i, which satisfy the following Poisson bracket relations: 3 i=1 15

where the Poisson bracket is defined as {q i,q j } = 0; {p i, p j } = 0; {q i, p j } = δ ij, (9.36) A {A(p,q),B(q, p)} PB B A B s q s p s p s q s. (9.37) The quantum mechanical version of the commutators: ˆq i, ˆq j = 0; ˆp, ˆp i j = 0; ˆq, ˆp i j = i δ, (9.38) ij which look almost exactly the same as Eq. (9.36). The fundamental commutators in Eq. (9.38) leads to first quantization; therefore, we seek classical conjugate variables in electromagnetic field, converting them into quantum mechanical operators to accomplish second quantization. 9.5 Lagrangian and Hamiltonian of Classical Electromagnetic Field Lagrangian for a particle (mass: m, velocity: v, and charge e) in an electromagnetic field (electric field E and magnetic field B): L = 1 mv [eφ(r) e c v A(r)]+ 1 8π ( E B )d 3 r = 1 mv [eφ(r) e c v A(r)]+ 1 ( 1 8π!A φ A )d 3 r. c (9.39) Denote the conjugate momenta to variables r, A and φ as p, and P φ, respectively. p = L r = mv e c A. (9.40) which is essentially the electric field. = L A = 1 4πc (1 c A + φ) = E 4πc, (9.41) P φ = L φ = 0, (9.4) i.e., the generalized conjugate momentum P φ vanishes. Then the corresponding Hamiltonian: H = ( r p + A + φp φ ) L = 1 m (p e c A) +eφ + 1 (16π c + A )d 3 r 8π. (9.43) 153

Since P φ = 0, we can use the gauge φ = 0 so that only variables and A are in H. This gauge imposes the transversality constraints of A = 0 and = 0. 9.6 Field Quantization In order to quantize the field, we require and A to be canonical, i.e., Classically, {q i, p j } = δ ij ; {A i (r,t),π j ( r,t)} = δ ij δ(r r ). (9.44) In quantum mechanics, the correspondence principle suggests that [q i, p j ] = i δ ij ; [A i (r,t),π j ( r,t)] = i δ ij δ(r r ). (9.45) First, let s work in the classical theory of electrodynamics. The Fourier transformations: Define dimensionless variables a: A(r) = [b(k)e ik r + b * (k)e ik r ]d 3 k, (9.46) (r) = 1 k[b(k)e ik r 4πic b * (k)e ik r ]d 3 k. (9.47) 3 c 1/ b(k) = ξ() a(). (9.48) 4π ω Imposing the transversality constraints of A = 0 and = 0, one can derive k [b(k)+ b * (k)] = 0, and k [b(k) b * (k)] = 0, respectively; thus k b(k) = 0 a(k3) = 0. (r) = 1/ c A(r) = ξ() [a()e ik r +a * ()e ik r ]d 3 k, (9.49) 4π ω 1 i The Poisson brackets are: ω 1/ ξ() 64π 4 c [a()e ik r a * ()e ik r ]d 3 k = E(r) 4πc. (9.50) {A i (r),a j ( r )} = 0, {a(),a( k λ )} = 0 {Π i (r),π j ( r )} = 0, {a * (),a * ( k λ )} = 0 {A i (r),π j ( r )} = δ ij δ(r r ), {a(),a * ( k λ )} = iδ λ λ δ(k k ) (9.51) 154

The Hamiltonian for electromagnetic field is Define H field = 1 8π (16π c + A )d 3 r = ω[a()a * ()]d 3 k. (9.5) q() 1 [a()+a * ()], (9.53) ω and p() i ω / [a() a * ()], (9.54) the familiar form consisting oscillators is recovered from Eq. (9.5): 1 H field = p ()+ 1 ω q () d 3 k. (9.55) Now simply change the classical variables p and q to the quantum mechanical operators ˆp and ˆq, we can quantize the electromagnetic field! First, the fundamental commutator is: [ˆq(), ˆp( k λ )] = i {q(), p( k λ )} = i δ λ λ δ(k k ). (9.56) Defining the corresponding lowering ( â ) and raising ( â ) operators based on the generalized coordinate ( ˆq ) and momentum ( ˆp ) operators: â() = ω ˆq + i 1 ω â () = ω ˆq i 1 ω The fundamental commutator becomes: ˆp, (9.57) ˆp. (9.58) [â(),â ( k λ )] = δ λ λ δ(k k ) (9.59) The quantized vector potential A and its conjugate momentum (essentially the electric field E) are: 1/ c Â(r) = ξ() [â()e ik r +â ()e ik r ]d 3 k, (9.60) 4π ω 155

1 ω 1/ ˆ (r) = ξ() [â()e ik r â ()e ik r ]d 3 k. (9.61) i 64π 4 c Finally the Hamiltonian for the quantized electromagnetic field The field operators: Notes: Ĥ field = â ()â()+ 1 ω d 3 k (9.6) 1/ ω Ê(r,t) = i ξ() [â()e i(k r ωt) â ()e i(k r ωt) ]d 3 k, (9.63) 4π 1/ c ˆB(r,t) = i [k ξ()] [â()e i(k r ωt) â ()e i(k r ωt) ]d 3 k. (9.64) 4π ω (1) In Eqs. (9.60) and (9.61), the time-dependency is omitted, and the full expression is where A k = c Â(r,t) = A kξ()[â()e i(k r ωt) +â ()e i(k r ωt) ]d 3 k (9.65) 4π ω = c 4π k and ω = ck. () The above expression uses infinite volume and thus Dirac δ(k k ) to normalize the planewave e ik k. In literature, the big-box normalization with volume Ω is also widely used: 1 Ω k 1 (π) 3 d 3 k, (9.66) which is normalized to δ k, k. But here we shall use 1 Ω k 1 (π) 3 d 3 k, (9.67) Why? Then where A k = π c Â(r,t) = A kξ()[â()e i(k r ωt) +â ()e i(k r ωt) ], (9.68) λ,k Ωω = π c Ωk. 156

9.7 Photon States and Wave Functions Photons are bosons, behaving as independent simple harmonic oscillator. The ket state for photons of wave vector k and polarization λ is denoted as, where is an integer that is the number of photons in that state. The state 0 is the (photon) vacuum. â 0 = 0, â 0 = = 1, (9.69) â = 1, â = +1 +1, (9.70) ( = â )! ( ) n 0, 0 = â!, (9.71) ˆN â â, ˆN = (9.7) Here ˆN is the photon number operator. Energy of the vacuum state [derived from Eq. (9.6)]: 1 E 0 = ωd 3 k = c kd 3 k. (9.73) Thus vacuum has infinite energy! Q: Does this even make sense? How to experimentally verify? A ket state with photons of different momenta is designated as n k1 λ 1 n k λ n km λ m Classical momentum, then Ĥ n k1 n λ k n 1 λ km = (n λ m + 1 ) ω n k1 n λ k n 1 λ km. (9.74) λ m p = 1 (E B)d 3 r 4πc, (9.75) Momentum operator in quantum mechanics using the correspondence principle: ˆp = 1 4πc (Ê ˆB)d 3 r = â â kd 3 k = ˆN kd 3 k, (9.76) Which is obviously corrected and expected, but mathematically it needs some work to prove it. ˆp = k, (9.77) Summary: quantum state of electromagnetic field ˆp n k1 n λ k n 1 λ km = n λ m k n k1 n λ k n 1 λ km. (9.78) λ m quantum state of oscillators n : number of photons at (k,λ) 157

Next we derive the wave functions of photons. A general photon state by superposition: The normalization condition Φ Φ = 1 Here φ is the scalar photon wave function. Φ = c = c â 0. (9.79) c = 1. (9.80) Let s focus on a photon state with specific (k, λ): k λ = δ δ (Big-box volume Ω) λ λ k, k δ λ λ δ(k k ) (Ω ) (Q: how to derive it?) The Real-space representation: (9.81) φ (r) r = eik r Ω, (9.8) because ˆp = k and in the r representation, ˆp = i. (Q: does λ play a role in photon wave function?) We can t use a scalar wave function to fully describe a particle with spin; e.g., we use spinors for electron wave functions. Here for spin-one photons, we use vector wave functions: where ξ indicates that photon has none-zero spin. φ (r) ξ φ (r) = ξ e ik r, (9.83) Ω A spin-one particle usually has three values along an arbitrary direction; however, it is not true for photons. Helicity of photons: the components of spin are parallel to momentum. Mathematically, the tranversality condition ( k ξ = 0 ) restricts the photon spin to be ± along k. (Q: Why? Hint: angular momentum including spin is the generator operator for rotation.) Physically, it is irrelevant to the chosen gauge; instead, it is the consequence of ultra-relativity, and a full understanding requires quantum field theory. The electromagnetic fields associated with a photon state Φ : 158

9.8 Spontaneous Radiation E(r,t) 0 Ê(r,t) Φ = i π ω ξ Ω c e i(k r ωt), (9.84) B(r,t) 0 ˆB(r,t) π c Φ = i ωω (k ξ )c e i(k r ωt). (9.85) We consider the spontaneous decay of H atom from lm to 100. Semi-classically, the transition rate is zero; however, experimentally the lifetime τ = 1 / R = 1.60 ns. Now we use the full quantum-mechanical treatment, i.e., field quantization, to compute τ. The perturbing Hamiltonian is H (t) = e  ˆp, (9.86) c Â(r,t) = A kξ() [â()e i(k r ωt) +â ()e i(k r ωt) ]d 3 k, with A k c 4π ω The initial state: i = lm 0, E i = E lm. The final state: f = 100, E f = E 100 + ω. 1/. (9.87) This is a periodic perturbation because H (t) ~ H cos(ωt), so we use Fermi s Golden Rule to calculate the transition rate R i f lim t d dt P i f (t) = π The (time-independent) perturbation matrix element where f H i δ(e f E i ). (9.88) H fi f H i = e c f Â(r) ˆp i, (9.89) f Â(r) ˆp i = 100 Â(r) 0 ˆp lm. (9.90) Because Â(r) 0 = A ξ k e ik r, (9.91) we obtain that f Â(r) ˆp i = A k ψ * 100 (r)e ik r ξ ( i ) ψ lm (r)d 3 r. (9.9) 159

Using the electric dipole approximation, i.e., e ik r 1, and replacing ξ ˆp by (i ω) ξ r : We need to evaluate: f Â(r) ˆp i = A k (i ω) ψ * 100 (r)[ ξ r] ψ lm (r)d 3 r. (9.93) I = ψ * 100 (r)[ ξ! r] * ψ lm (r)d 3 r = 100! ξ r lm. (9.94) n l m ξ! r nlm = sin θ cosφ n l m x nlm + sin θ sinφ n l m y nlm + cosθ n l m z nlm (9.95) One can use the selection rule (under the electric dipole approximation) to exclude the zero matrix elements of n l m ξ r nlm : (1) Δl = ±1 (9.96) () n l m x nlm : Δm = ±1; n l m y nlm : Δm = ±1; n l m z nlm : Δm = 0. (9.97) For the lm 100 (or specifically p 1s ) transition, we need to evaluate the following: 100 z 10 = 15 Similarly, 100 x 1 ±1 = ± 7 3 5 a 0, where a 0 is the Bohr radius. 3 5 a 0, and 100 y 1 ±1 = i 7 3 5 a 0. Now we can evaluate the associated transition rate using FGR by replacing the delta-function by (i) integrating over k and (ii) summing over λ = 1, : all R i f = π! = π! = e π λ λ λ H fi () δ(e f E i )d 3 k e A c k i ω I δ(e 1s +!ck E p )dωk dk 1 I 4π dω ωδ(e +!ck E )4πk dk 1s p (9.98) Let s do the angular part first, and define I 1 I 4π dω. (9.99) Here I is taken as the average over (1) m = 1,0,1; () all the orientations of ξ : 160

where I = 1 4π [I x sin (θ)cos (φ)+ I y sin (θ)sin (φ)+ I z cos (θ)]sin(θ)dθdφ, (9.100) I = 1 x 3 7 + 0 + 7 a 3 5 3 5 0 = 15 3 a, 11 0 I = I = I y z x = 15 3 a. 11 0 (9.101) This shows that I is independent of the orientation, thus I = 15 3 a. The transition rate 11 0 for this spontaneous radiation (p 1s ) thus is (Note: below is due to summation over λ ): all R i f = e 15 π 3 a ωδ(e 11 0 1s +!ck E p )4πk dk. (9.10) Note that in the above expression is due to summation over λ. We evaluate the integral: where ω = ck and ωδ(e 1s +!ck E p )4πk dk = 4πc( k ) 3, (9.103)!c k = E E p 1s!c = 3e 8a 0!c. (9.104) Finally we obtain the spontaneous transition rate from p to 1s of H atom, all R i f where the fine structure constant α = e = 8 α m c 5 e 3! = 6.7 108 s 1, (9.105) c = 1 137.036. The corresponding lifetime τ = 1 / R = 1.60 ns, in excellent agreement with experimental data τ = 1.600 ± 0.004 ns. [Phys. Rev. 148, 1 (1966)] 161

9.9 Optical Absorption and Stimulated Radiation Now let s consider the reverse process: optical absorption. The optical absorption by solids is a crucial topic in condensed matter physics. Here we only study the absorption by gas. There are a number of processes whereby gases can absorb electromagnetic radiation: 1) Collisions of atoms. ) Rotation and vibration. 3) Inelastic processes such as Raman scattering. 4) Two-photon absorptions. But we omit all the above processes, focusing only on the simplest and most crucial case: optical absorption by a single static atom. The results for gas in a box are multiplied by the number of atoms N A. The perturbing Hamiltonian still is H (t) = e  ˆp. c The initial state: I = i, The final state: F = f 1. Fermi s Golden Rule: R = π e c f ; 1 Â(r) ˆp i; δ(e f E i ω), (9.106) with E i and E f the initial and final energies for electronic states i and f, respectively. Here R is the rate for the specific transition associated with a photon mode (), and M fi f ; 1 Â(r) ˆp i; = f 1 Â(r) ˆp i, (9.107) 1 Â(r) = A k ξ e ik r. (9.108) Thus M fi = A k ( ξ ), (9.109) where p fi f e ik rˆp i, the optical transition matrix element. Note that semi-classically, A is NOT an operator, and the electromagnetic field is classical no quantum states, instead it is a continuous field describe by the vector potential A(r,t): 16

A(r,t) = A 0 [e i(k r ωt) +e i(k r ωt) ], with A 0 a constant vector potential. The matrix elements M if have exactly the same form as the quantum version: M fi = A 0 ( ξ ), (9.110) therefore, A 0 = A k. The total transition rate for a gas is obtained by (1) summation of all possible transitions and () multiplying the number of atoms N A : R = N A R. (9.111) 9.9.1 Absorption Coefficient and Cross Section Now we need to consider physical quantities, which are measurable. An important parameter is the flux F of photons in the gas, which is proportional to the Poynting vector S. F has the dimensional units of number of photons/(cm s): Fig. 9.1: Geometry for Beer s law. where Ω is the renormalization volume. Then F = S ω = c Ω, (9.11) = F Ω c, (9.113) π e R = A c k ξ δ(e f E i ω) = F Ω c π e A c k ξ δ(e f E i ω) (9.114) F A = number of photons with mode () per second going through the slab. Note: A is area. AdF = N A R, (9.115) (Adx) df dx = N A R, (9.116) 163

Ω df dx = N A R = N A F Ω c π e A c k ξ δ(e f E i ω), (9.117) Eq. (9.118) is Beer s law for optical absorption: Here the absorption coefficient df dx = α()f. (9.118) F(x) = F 0 e αx. (9.119) π e α() = N A A c c k ξ δ(e f E i ω) = n A πe ξ cω c δ(e f E i ω). (9.10) We used A k π c ωω 1/ and n A N A Ω. Then the total absorption coefficient α(ω) = α()d 3 k. (9.11) Without losing generality, set the direction of matrix element p fi along z-axis, then ξ = sin (θ)p fi (9.1) where θ is the angle between k and p fi. The absorption coefficient Absorption cross section (per atom): α(ω) = n A c 4 3 σ(ω) = α(ω) n A = 4 3 e m c p (E 3 fi i E f ). (9.13) e m c p (E 4 fi i E f ). (9.14) 9.9. Stimulated Radiation Radiation (emission) is a complimentary experiment to absorption. The atom must first be excited so that some of the electrons are in excited states. This step is usually accomplished by 164

optical absorption or by bombarding the atoms with a beam of energetic electrons. After an electron is in an excited state, it can return to the ground state by the emission of a photon. Fermi s Golden rule: R = π e c f ; +1 Â(r) ˆp i; δ(e f + ω E i ), (9.15) M fi F Â(r) ˆp I = f +1 Â(r) ˆp i (9.16) where p fi f e ik rˆp i. So the FGR is +1 Â(r) = A k ξ e ik r +1 (9.17) M fi = +1 A k ( ξ ), (9.18) R = ( +1) π e m c A ξ k δ(e f + ω E i ). (9.19) e Comments: 1) The optical transition matrix elements (p fi ) for stimulated radiation are identical to those for the corresponding absorption. ) Transition rate (R ) for stimulated radiation is proportional to the number of available photons ( ), while R for spontaneous radiation is independent of. Consider two electronic states i and f, with E f = E i + ω. Optical absorption rate: abs R i f = B i f N i ρ(ω), (9.130) where N i the number of atoms in the i state, ρ(ω) the energy density of photons with frequency ω, and B i f the Einstein B coefficient. The radiation rate is R rad f i = B f i N f ρ(ω)+ A f i N f, (9.131) where N f the number of atoms in the f state, and A f i the Einstein A coefficient. From comment (1), B i f = B f i B. (9.13) 165

Let A A f i. The equilibrium condition is: abs R i f = R rad f i, (9.133) and one can derive that ρ(ω) = A B 1 (N i / N f ) 1. (9.134) Statistical mechanics show that N i N f = e Ei /kbt e E f /k B T = e (Ei Ef )/kbt = e ω/k B T, (9.135) where T the temperature and k B the Boltzmann constant. Bose factor: we find that n ω = Based on Einstein relations, one can write down the balance as 1 e ω/k B T 1, (9.136) ρ(ω) = Ωω π c 3 1 e ω/k B T 1, (9.137) A B = Ωω π c 3. (9.138) di dx = const I(N N ), (9.139) f i where I is light density. Since E f > E i, N f < N i at equilibrium, thus di dx < 0, light is diminishing. In order to make a laser (light amplification by stimulated emission of radiation), inverse population (N f > N i ) is required, which means negative temperature. A simplest implementation is a three-level system with a forbidden transition between two states. 166