MecanicaQuanticaScript PDF

.
Graduate courses in Physics
Quantum Mechanics
applied to Atoms and Light
Ph.W. Courteille
Universidade de São Paulo
Instituto de Fı́sica de São Carlos
03/05/2019
2
Content
0 Preface 1
0.1 Possible topics of partial courses . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
0.2 Organization of the course . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
0.3 Recommended bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
I Quantum Mechanics 5
1 Antecedents of quantum mechanics 7

1.1 Constants in atomic physics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.1.1 Atomic units . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.1.2 Clebsch-Gordan symbols . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.1.3 Functions and polynomials . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.2 The discovery of the atom . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.2.1 Democrit’s model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.2.2 Thomson’s model and Rutherford’s experiment . . . . . . . . . . . . . . . 14
1.2.3 Emission of radiation in the planetary model . . . . . . . . . . . . . . . . 18
1.2.4 Zeeman effect in the planetary model . . . . . . . . . . . . . . . . . . . . 19
1.2.5 Bohr’s theory and its limitations . . . . . . . . . . . . . . . . . . . . . . . 20
1.3 The discovery of the photon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
1.3.1 Radiation in a conductive cavity . . . . . . . . . . . . . . . . . . . . . . . 21
1.3.2 Black body radiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
1.3.3 Planck’s distribution of modes . . . . . . . . . . . . . . . . . . . . . . . . 25
1.3.4 The corpuscular nature of the photon . . . . . . . . . . . . . . . . . . . . 26
1.3.5 Einstein’s transitions rates . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
1.3.6 Propagation of light in dielectric media . . . . . . . . . . . . . . . . . . . 29
1.3.7 Propagation of light in dilute gases . . . . . . . . . . . . . . . . . . . . . . 30
1.3.8 Spectral line profiles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
1.3.9 Specific heat of solids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
1.4 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
1.4.1 The discovery of the atom . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
1.4.2 The discovery of the photon . . . . . . . . . . . . . . . . . . . . . . . . . . 35
2 Foundations of quantum mechanics 37

2.1 Basic notions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
2.1.1 Dispersion relation and Schrödinger equation . . . . . . . . . . . . . . . . 37
2.1.2 Relativistic particle waves . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
2.1.3 Born’s interpretation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
2.1.4 Continuity equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.1.5 Distributions in space and time . . . . . . . . . . . . . . . . . . . . . . . . 39
2.1.6 Eigenvalues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
2.1.7 Temporal evolution of eigenvalues . . . . . . . . . . . . . . . . . . . . . . 41
2.2 Postulates of quantum mechanics . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3
4 CONTENT
2.2.1 Superposition principle (1. postulate) . . . . . . . . . . . . . . . . . . . . 42
2.2.2 Interpretation of the wave function (2. postulate) . . . . . . . . . . . . . . 42
2.2.3 Dirac bra-ket notation and vector representation . . . . . . . . . . . . . . 43
2.2.4 Observables (3. postulate) . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
2.2.5 Representation of operators as matrices . . . . . . . . . . . . . . . . . . . 44
2.2.6 Correspondence principle (4. postulate) . . . . . . . . . . . . . . . . . . . 45
2.2.7 Schrödinger equation and quantum measurements (5. postulate) . . . . . 45
2.2.8 Stationary Schrödinger equation . . . . . . . . . . . . . . . . . . . . . . . 46
2.3 Abstract formalism of quantum mechanics . . . . . . . . . . . . . . . . . . . . . . 47
2.3.1 Lie algebra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
2.3.2 Complete bases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
2.3.3 Degeneracy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
2.3.4 Bases as unitary operators . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
2.3.5 Complete set of commuting operators . . . . . . . . . . . . . . . . . . . . 50
2.3.6 Uncertainty relation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
2.3.7 Representations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
2.3.8 Setting up a Hilbert space with several degrees of freedom . . . . . . . . . 53
2.4 Time evolutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
2.4.1 Unitary transformations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
2.4.2 Schrödinger picture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
2.4.3 Heisenberg picture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
2.4.4 Interaction picture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
2.4.5 Ehrenfest’s theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
2.5 Symmetries in quantum mechanics . . . . . . . . . . . . . . . . . . . . . . . . . . 63
2.5.1 Translation and rotation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
2.5.2 Galilei and Lorentz boosts . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
2.5.3 Gauge transformations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
2.5.4 Noether’s theorem and conservation laws . . . . . . . . . . . . . . . . . . 67
2.6 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
2.6.1 Basic notions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
2.6.2 Postulates of quantum mechanics . . . . . . . . . . . . . . . . . . . . . . . 70
2.6.3 Abstract formalism of quantum mechanics . . . . . . . . . . . . . . . . . . 71
2.6.4 Time evolutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
2.6.5 Symmetries in quantum mechanics . . . . . . . . . . . . . . . . . . . . . . 73
3 Linear motion / Separable potentials 75

3.1 Translational motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
3.1.1 Quadratic integrability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
3.1.2 Separation of dimensions . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
3.2 Rectangular potential . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
3.2.1 Box potential . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
3.2.2 Multidimensional box potential . . . . . . . . . . . . . . . . . . . . . . . . 77
3.2.3 Potentials with several sections of constant depths . . . . . . . . . . . . . 77
3.2.4 Potential well . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
3.3 Potential barrier . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
3.3.1 T -scattering matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
3.3.2 S-scattering matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
3.3.3 Quantum reflection at a potential step . . . . . . . . . . . . . . . . . . . . 81
CONTENT 5
3.3.4 Continuity of probability flow . . . . . . . . . . . . . . . . . . . . . . . . . 82

3.3.5 Tunneling and quantum reflection at a potential well . . . . . . . . . . . . 82
3.3.6 The delta-potential . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
3.4 Harmonic oscillator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
3.4.1 Factorization of the Hamiltonian and Fock states . . . . . . . . . . . . . . 87
3.4.2 Harmonic oscillator in spatial representation . . . . . . . . . . . . . . . . 89
3.4.3 Properties of the harmonic oscillator . . . . . . . . . . . . . . . . . . . . . 90
3.4.4 Time evolution of the harmonic oscillator . . . . . . . . . . . . . . . . . . 91
3.4.5 Multidimensional harmonic oscillator . . . . . . . . . . . . . . . . . . . . . 92
3.4.6 Coherent states . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
3.4.7 Quantization of the electromagnetic field . . . . . . . . . . . . . . . . . . 95
3.5 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
3.5.1 Translational motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
3.5.2 Rectangular potential . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
3.5.3 Potential barrier . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
3.5.4 Harmonic oscillator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
4 Rotations / Central potentials 101

4.1 Particle in a central potential . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
4.1.1 Transformation to relative coordinates . . . . . . . . . . . . . . . . . . . . 101
4.1.2 Particle in a cylindrical potential . . . . . . . . . . . . . . . . . . . . . . . 102
4.1.3 Hamiltonian in spherical coordinates . . . . . . . . . . . . . . . . . . . . . 103
4.1.4 Separation of radial motion . . . . . . . . . . . . . . . . . . . . . . . . . . 105
4.2 Quantum treatment of hydrogen . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
4.2.1 Bohr’s model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
4.2.2 The virial theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
4.3 Angular momentum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
4.3.1 The orbital angular momentum operator . . . . . . . . . . . . . . . . . . . 112
4.3.2 SU(2) algebra of angular momentum and spin . . . . . . . . . . . . . . . . 113
4.3.3 The electron spin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
4.4 Coupling of angular momenta . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
4.4.1 Singlet and triplet states with two electrons . . . . . . . . . . . . . . . . . 114
4.4.2 Coupling two spins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
4.4.3 Decoupled and coupled bases . . . . . . . . . . . . . . . . . . . . . . . . . 118
4.4.4 Clebsch-Gordan coefficients . . . . . . . . . . . . . . . . . . . . . . . . . . 120
4.5 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
4.5.1 Particle in a central potential . . . . . . . . . . . . . . . . . . . . . . . . . 122
4.5.2 Quantum treatment of hydrogen . . . . . . . . . . . . . . . . . . . . . . . 123
4.5.3 Angular momentum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
4.5.4 Coupling of angular momenta . . . . . . . . . . . . . . . . . . . . . . . . . 126
5 Approximation methods 129

5.1 Stationary perturbations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
5.1.1 Time-independent perturbation theory . . . . . . . . . . . . . . . . . . . . 129
5.1.2 TDPT with degenerate states . . . . . . . . . . . . . . . . . . . . . . . . . 131
5.2 Variational method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
5.2.1 The Rayleigh fraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
5.2.2 Rayleigh-Ritz method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
6 CONTENT
5.3 WKB approximation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
5.3.1 WKB approximation applied to the Schrödinger equation . . . . . . . . . 135
5.3.2 Connection formulas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
5.4 Time-dependent perturbations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
5.4.1 Two-level systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
5.4.2 The time-dependent perturbation method . . . . . . . . . . . . . . . . . . 143
5.4.3 Numerical methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
5.4.4 Transition rates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
5.4.5 Periodic perturbations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
5.4.6 Einstein transition probabilities . . . . . . . . . . . . . . . . . . . . . . . . 148
5.5 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
5.5.1 Stationary perturbations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
5.5.2 Variational method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
5.5.3 WKB approximation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
5.5.4 Time-dependent perturbations . . . . . . . . . . . . . . . . . . . . . . . . 151
6 Periodic systems 153

6.1 The Bloch model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
6.1.1 Approximation for quasi-bound electrons . . . . . . . . . . . . . . . . . . 154
6.1.2 Approximation for quasi-free electrons . . . . . . . . . . . . . . . . . . . . 156
6.1.3 Application to one-dimensional optical lattices . . . . . . . . . . . . . . . 157
6.1.4 Bloch oscillations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
6.2 The Kronig-Penney model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
6.3 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
6.3.1 The Bloch model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
6.3.2 The Kronig-Penney model . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
II Atomic and Molecular Physics 165
7 The electron spin and the atomic substructure 167

7.1 The Dirac equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
7.1.1 The Klein-Gordon equation for bosons . . . . . . . . . . . . . . . . . . . . 167
7.1.2 The Dirac equation for fermions . . . . . . . . . . . . . . . . . . . . . . . 168
7.1.3 The relativistic electron in a central Coulomb field . . . . . . . . . . . . . 173
7.1.4 The Pauli equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
7.2 Fine structure of hydrogen-like atoms via TDPT . . . . . . . . . . . . . . . . . . 177
7.2.1 Correction for relativistic velocities . . . . . . . . . . . . . . . . . . . . . . 178
7.2.2 Correction due to spin-orbit coupling . . . . . . . . . . . . . . . . . . . . . 179
7.2.3 Non-local electron-core interaction . . . . . . . . . . . . . . . . . . . . . . 181
7.2.4 Summary of the corrections . . . . . . . . . . . . . . . . . . . . . . . . . . 182
7.2.5 Lamb shift . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
7.3 Hyperfine structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
7.3.1 Coupling to the nuclear spin . . . . . . . . . . . . . . . . . . . . . . . . . 183
7.3.2 Electric quadrupole interaction . . . . . . . . . . . . . . . . . . . . . . . . 186
7.4 Exotic atoms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
7.4.1 Positronium and muonium . . . . . . . . . . . . . . . . . . . . . . . . . . 187
7.4.2 Hadronic atoms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
CONTENT 7
7.4.3 Muonic hydrogen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189

7.4.4 Rydberg atoms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
7.5 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
7.5.1 Fine structure and the Dirac equation . . . . . . . . . . . . . . . . . . . . 191
7.5.2 Fine structure of hydrogen-like atoms via TDPT . . . . . . . . . . . . . . 193
7.5.3 Hyperfine structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
7.5.4 Exotic atoms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
8 Atoms with spin in external fields 195

8.1 Charged particles in electromagnetic fields . . . . . . . . . . . . . . . . . . . . . . 195
8.1.1 Lagrangian and hamiltonian of charged particles . . . . . . . . . . . . . . 195
8.1.2 Minimal coupling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
8.2 Interaction with magnetic fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
8.2.1 Normal Zeeman effect of the fine structure . . . . . . . . . . . . . . . . . . 196
8.2.2 Anomalous Zeeman effect . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
8.2.3 Paschen-Back effect and intermediate magnetic fields . . . . . . . . . . . . 197
8.2.4 Zeeman effect of the hyperfine structure . . . . . . . . . . . . . . . . . . . 198
8.2.5 Paschen-Back effect of the hyperfine structure . . . . . . . . . . . . . . . . 200
8.2.6 Hyperfine structure in the intermediate field regime . . . . . . . . . . . . 200
8.2.7 Landau levels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
8.3 Interaction with electric fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
8.3.1 Stark Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
8.4 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205
8.4.1 Charged particles in electromagnetic fields . . . . . . . . . . . . . . . . . . 205
8.4.2 Interaction with magnetic fields . . . . . . . . . . . . . . . . . . . . . . . . 205
8.4.3 Interaction with electric fields . . . . . . . . . . . . . . . . . . . . . . . . . 206
9 Atoms with many electrons 209

9.1 Symmetrization of bosons and fermions . . . . . . . . . . . . . . . . . . . . . . . 209
9.1.1 Pauli’s Principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210
9.1.2 Consequences for quantum statistics . . . . . . . . . . . . . . . . . . . . . 212
9.2 Helium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
9.2.1 The ground state . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
9.2.2 Excited states . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214
9.3 Electronic shell structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
9.3.1 Thomas-Fermi model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
9.3.2 Hartree method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
9.3.3 Hartree Fock method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
9.4 The periodic system of elements . . . . . . . . . . . . . . . . . . . . . . . . . . . 226
9.4.1 Electronic shell model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
9.4.2 Alkalines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228
9.4.3 LS and jj-coupling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
9.4.4 Summary of the atomic degrees of freedom . . . . . . . . . . . . . . . . . 231
9.5 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
9.5.1 Symmetrization of bosons and fermions . . . . . . . . . . . . . . . . . . . 233
9.5.2 Helium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
9.5.3 Electronic shell structure . . . . . . . . . . . . . . . . . . . . . . . . . . . 234
9.5.4 The periodic system of elements . . . . . . . . . . . . . . . . . . . . . . . 234
8 CONTENT
III Quantum Optics 237
10 The Bloch equations 239

10.1 Density matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
10.1.1 The density operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
10.1.2 Matrix formalism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
10.1.3 Temporal evolution of the density operator . . . . . . . . . . . . . . . . . 245
10.2 Bloch equations for two-level atoms . . . . . . . . . . . . . . . . . . . . . . . . . . 246
10.2.1 The matrix elements of the density operator . . . . . . . . . . . . . . . . . 246
10.2.2 Rotating wave approximation . . . . . . . . . . . . . . . . . . . . . . . . . 247
10.2.3 Pauli matrices and the atomic Bloch vector . . . . . . . . . . . . . . . . . 249
10.2.4 Manipulation of the state by sequences of radiation pulses . . . . . . . . . 250
10.3 Bloch equations with spontaneous emission . . . . . . . . . . . . . . . . . . . . . 252
10.3.1 Phenomenological inclusion of spontaneous emission . . . . . . . . . . . . 252
10.3.2 Susceptibility and density operator . . . . . . . . . . . . . . . . . . . . . . 254
10.3.3 Liouville equation for two levels . . . . . . . . . . . . . . . . . . . . . . . . 255
10.3.4 Saturation effects by the effective hamiltonian . . . . . . . . . . . . . . . . 255
10.4 Line broadening mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256
10.4.1 Saturation broadening . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256
10.4.2 Collision broadening . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257
10.4.3 Doppler broadening . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
10.4.4 Voigt profile . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260
10.4.5 Bloch equations with phase modulation . . . . . . . . . . . . . . . . . . . 260
10.5 Multi-level systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
10.5.1 Liouville equation for many levels . . . . . . . . . . . . . . . . . . . . . . . 262
10.5.2 Bloch equations for three levels . . . . . . . . . . . . . . . . . . . . . . . . 263
10.5.3 Numerical treatment of Bloch equations . . . . . . . . . . . . . . . . . . . 264
10.5.4 General rules for setting up multilevel Bloch equations . . . . . . . . . . . 266
10.6 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271
10.6.1 Density matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271
10.6.2 Bloch equations for two-level atoms . . . . . . . . . . . . . . . . . . . . . 272
10.6.3 Bloch equations with spontaneous emission . . . . . . . . . . . . . . . . . 273
10.6.4 Line broadening mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . 276
10.6.5 Multi-level systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278
11 Atoms in quantized radiation fields 283

11.1 Quantization of the electromagnetic field . . . . . . . . . . . . . . . . . . . . . . . 283
11.1.1 Field operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
11.1.2 Interaction of quantized fields with atoms . . . . . . . . . . . . . . . . . . 285
11.1.3 Dressed states . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287
11.2 Quantum correlations in the Jaynes-Cummings model . . . . . . . . . . . . . . . 287
11.2.1 The classical and quantum limits . . . . . . . . . . . . . . . . . . . . . . . 289
11.2.2 Quantum correlations in light modes . . . . . . . . . . . . . . . . . . . . . 293
11.2.3 The Jaynes-Cummings model with dissipation . . . . . . . . . . . . . . . 299
11.3 Spontaneous emission and light scattering . . . . . . . . . . . . . . . . . . . . . . 300
11.3.1 Interaction of atoms with vacuum modes . . . . . . . . . . . . . . . . . . 300
11.3.2 Resonance fluorescence and coherent and incoherent light scattering . . . 304
11.3.3 Correlation functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 306
CONTENT 9
11.3.4 The spectrum of resonance fluorescence . . . . . . . . . . . . . . . . . . . 309

11.4 Beam splitting and quantum amplification . . . . . . . . . . . . . . . . . . . . . . 313
11.4.1 The beam splitter in various representations . . . . . . . . . . . . . . . . . 314
11.4.2 Fock and Glauber states at a beam splitter . . . . . . . . . . . . . . . . . 316
11.4.3 Shot-noise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 318
11.4.4 Quantum amplifier . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 318
11.4.5 Homodyne detection and inverse Radon transform . . . . . . . . . . . . . 320
11.5 The laser . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 321
11.5.1 Features and operation of lasers . . . . . . . . . . . . . . . . . . . . . . . . 323
11.5.2 Applications of lasers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 325
11.6 Squeezed states . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 326
11.6.1 Squeezing operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 326
11.6.2 Coherence of squeezed states . . . . . . . . . . . . . . . . . . . . . . . . . 328
11.6.3 Homodyne signature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 328
11.6.4 Hamiltonian for squeezing . . . . . . . . . . . . . . . . . . . . . . . . . . . 328
11.6.5 Multimode squeezing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 329
11.7 Photon counting statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 330
11.7.1 Photon counting with classica fields . . . . . . . . . . . . . . . . . . . . . 330
11.7.2 Photon counting with quantum fields . . . . . . . . . . . . . . . . . . . . . 330
11.7.3 Waiting time distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . 333
11.8 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333
11.8.1 Quantization of the electromagnetic field . . . . . . . . . . . . . . . . . . 333
11.8.2 Quantum correlations in the Jaynes-Cummings model . . . . . . . . . . . 333
11.8.3 Spontaneous emission and light scattering . . . . . . . . . . . . . . . . . . 334
11.8.4 Beam splitting and quantum amplification . . . . . . . . . . . . . . . . . . 334
11.8.5 The laser . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 334
11.8.6 Squeezed states . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 334
11.8.7 Photon counting statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . 335
12 Atomic motion in electromagnetic fields 337

12.1 Electromagnetic forces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 338
12.1.1 Forces on charges and electric dipole moments . . . . . . . . . . . . . . . 339
12.1.2 Forces on magnetic dipole moments . . . . . . . . . . . . . . . . . . . . . 339
12.1.3 Adiabatic potentials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 340
12.2 Optical forces on free and confined atoms . . . . . . . . . . . . . . . . . . . . . . 341
12.2.1 The dipolar gradient force and the radiation pressure force . . . . . . . . 342
12.2.2 Semiclassic calculation of dipole force and radiative pressure . . . . . . . 345
12.2.3 Force exerted by a quantized radiation field . . . . . . . . . . . . . . . . . 345
12.2.4 Refraction of atoms by light and of light by atoms . . . . . . . . . . . . . 345
12.2.5 Inelastic scattering from free thermal atoms . . . . . . . . . . . . . . . . . 347
12.2.6 Optical forces on confined atoms . . . . . . . . . . . . . . . . . . . . . . . 348
12.3 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351
12.3.1 Electromagnetic forces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351
12.3.2 Optical forces on free and confined atoms . . . . . . . . . . . . . . . . . . 351
10 CONTENT
13 Quantum measurement 353
13.1 The observer and the reality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353
13.1.1 Schrödinger’s cat . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353
13.1.2 The quantum jump . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 357
13.1.3 Weak measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 362
13.2 Open systems and the master equation . . . . . . . . . . . . . . . . . . . . . . . . 364
13.2.1 Born approximation for weak coupling . . . . . . . . . . . . . . . . . . . . 364
13.2.2 Assumption of an initial product state . . . . . . . . . . . . . . . . . . . . 365
13.2.3 Markov approximation for short memory . . . . . . . . . . . . . . . . . . 365
13.2.4 Example: Two level system with spontaneous emission . . . . . . . . . . . 365
13.2.5 Example: Damped harmonic quantum oscillator . . . . . . . . . . . . . . 366
13.2.6 Thermalization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 367
13.3 Repeated measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 367
13.3.1 The quantum Zeno effect . . . . . . . . . . . . . . . . . . . . . . . . . . . 367
13.3.2 Quantum non-demolition measurement . . . . . . . . . . . . . . . . . . . 370
13.4 Welcher Weg information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 370
13.4.1 The Elitzur and Vaidman bomb testing problem . . . . . . . . . . . . . . 370
13.5 Noisy measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 370
13.5.1 Quantum projection noise . . . . . . . . . . . . . . . . . . . . . . . . . . . 370
13.6 Topological phases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 371
13.6.1 Aharonov-Bohm effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 374
13.7 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 377
13.7.1 The observer and the reality . . . . . . . . . . . . . . . . . . . . . . . . . 377
13.7.2 Open systems and the master equation . . . . . . . . . . . . . . . . . . . . 377
13.7.3 Repeated Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . 378
13.7.4 Welcher Weg information . . . . . . . . . . . . . . . . . . . . . . . . . . . 379
13.7.5 Noisy measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 379
13.7.6 Topological phases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 379
IV Collective Scattering of Light 381
V Atom Optics 383
14 Manipulation of atomic gases 385

14.1 The atomic motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 386
14.1.1 The atom as a matter wave . . . . . . . . . . . . . . . . . . . . . . . . . . 386
14.1.2 Localized atoms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 389
14.1.3 Density-of-states of a trapping potential . . . . . . . . . . . . . . . . . . . 389
14.2 Optical cooling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 391
14.2.1 Optical molasses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 391
14.2.2 Sub-Doppler cooling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 393
14.2.3 Raman cooling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 397
14.2.4 Stimulated Raman sideband cooling . . . . . . . . . . . . . . . . . . . . . 399
14.2.5 Adiabatic decompression . . . . . . . . . . . . . . . . . . . . . . . . . . . . 399
14.3 Optical and magneto-optical traps . . . . . . . . . . . . . . . . . . . . . . . . . . 399
14.3.1 The magneto-optical trap . . . . . . . . . . . . . . . . . . . . . . . . . . . 399
CONTENT 11
14.3.2 Optical dipole traps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 403

14.4 Magnetic traps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 406
14.4.1 Quadrupolar traps and Majorana spin-flips . . . . . . . . . . . . . . . . . 407
14.4.2 Magnetic Ioffe-type traps . . . . . . . . . . . . . . . . . . . . . . . . . . . 408
14.4.3 Radiative coupling and evaporative cooling . . . . . . . . . . . . . . . . . 409
14.4.4 Sympathetic cooling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 413
14.5 Other traps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 415
14.5.1 Ion traps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 415
14.5.2 Micromotion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 418
14.5.3 QUEST . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 420
14.6 Analysing techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 421
14.6.1 Time-of-flight imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 421
14.6.2 Absorption imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 423
14.6.3 Dispersive imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 423
14.6.4 Reconstrução de column-integrated absorption images . . . . . . . . . . . 425
14.6.5 Condensable atomic species . . . . . . . . . . . . . . . . . . . . . . . . . . 425
14.7 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 426
14.7.1 The atomic motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 426
14.7.2 Optical cooling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 426
14.7.3 Optical and magneto-optical traps . . . . . . . . . . . . . . . . . . . . . . 427
14.7.4 Magnetic traps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 428
14.7.5 Others traps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 429
14.7.6 Analysing techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 429
15 Thermodynamics of quantum gases 431

15.1 Quantum statistics of an ideal Bose gas . . . . . . . . . . . . . . . . . . . . . . . 432
15.1.1 Condensation of a free gas confined in a box potential . . . . . . . . . . . 434
15.1.2 Condensation of a harmonically confined gas . . . . . . . . . . . . . . . . 436
15.1.3 Energy and heat capacity . . . . . . . . . . . . . . . . . . . . . . . . . . . 437
15.1.4 Distribution functions for a Bose gas . . . . . . . . . . . . . . . . . . . . . 439
15.2 Quantum statistics of an ideal Fermi gas . . . . . . . . . . . . . . . . . . . . . . . 440
15.2.1 Chemical potential and Fermi radius for a harmonic trap . . . . . . . . . 441
15.2.2 Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 442
15.2.3 Entropy and heat capacity . . . . . . . . . . . . . . . . . . . . . . . . . . . 443
15.2.4 Distributions of a Fermi gas . . . . . . . . . . . . . . . . . . . . . . . . . . 444
15.2.5 Equipartition theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 447
15.2.6 Density and momentum distribution for anharmonic potentials . . . . . . 448
15.2.7 Classical gas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 449
15.2.8 Intensive and extensive parameters . . . . . . . . . . . . . . . . . . . . . . 450
15.2.9 Signatures for quantum degeneracy of a Fermi gas . . . . . . . . . . . . . 451
15.2.10 Fermi gas in reduced dimensions . . . . . . . . . . . . . . . . . . . . . . . 453
15.3 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 454
15.3.1 Quantum statistics of Bose-condensates . . . . . . . . . . . . . . . . . . . 454
15.3.2 Quantum statistics of a Fermi gas . . . . . . . . . . . . . . . . . . . . . . 456
0 CONTENT
Chapter 0
Preface
This script represents a synthesis of several postgraduate courses given at the Institute of Physics
of São Carlos (IFSC) of the University of São Paulo (USP). The courses are Quantum Mechanics
(SFI5774), Atomic and Molecular Physics (SFI5814), Quantum Mechanics B (SFI5707), Inter-
action of Light and Matter (SFI5877), and Atomic Optics (SFI5887). The topics of the courses
are, of course, closely intertwined. The purpose of this composite script is to emphasize the
interconnection of topics and facilitate the understanding of how they are related. In part I we
introduce the quantum mechanics, which represents the fundamental theory for the rest of the
book. In the second part we focus on the structure of the atom. In the third and fourth part
we study the properties of light, its interaction with individual atoms and atomic ensembles and
how the interaction is influenced by cavities and surfaces. Finally, in part V we introduce the
optics of matter wave.
The course is intended for masters and PhD students in physics. The script is a preliminary
version continually being subject to corrections and modifications. Error notifications and sug-
gestions for improvement are always welcome. The script incorporates exercises the solutions of
which can be obtained from the author.
0.1 Possible topics of partial courses
The courses in Quantum Mechanics A (QM/SFI5774), of Quantum Mechanics B (QO/SFI5708),

of Atomic and Molecular Physics (AM/SFI5814), and of Atom Optics (OA/SFI5887) may have
the following contents:
1
2 CHAPTER 0. PREFACE
Sections QM QO AM AO
Antecedents 1.1.1-1.3.7 x
Repetition of basic notions of quantum mechanics 2.1.1-2.5.4 x x x
Motion in separable potentials 3.1.1-3.3.5 x
The harmonic oscillator 3.4.1-3.4.7 x x
Rotations, central potentials and spin coupling 4.1.1-4.4.4 x x
Stationary perturbation theory 5.1.1-5.3.2 x x
Time-dependent perturbation theory 5.4.1-5.4.6 x x x
Periodic systems 6.1.1-6.1.6 x
Spin of the electron and the (hyper-)fine structure 7.1.1-7.4.4 x x
Charged particles in electromagnetic fields 8.1.1-8.1.2 x x x
Atoms in stationary and electromagnetic fields 8.2.1-8.3.1 x x
Atoms with multiple electrons 9.1.1-9.4.4 x
Dimeric molecules 10.1.1-10.3.3 x
Collisions 11.1.1-11.2.3 x
Absorption and emission of radiation and spectral lines 12.1.1-12.3.2 x
Susceptibility and polarization 12.4.1-12.4.4 x x
Density operator and Bloch equations 13.1.1-13.2.4 x
Spontaneous emission and saturation 13.3.1-13.3.4 x
Line broadening processes 13.4.1-13.4.5 x
Multilevel systems 13.5.1-13.5.4 x
Quantization of radiation and dressed states 14.1.1-14.1.3 x
Jaynes-Cummings model and quantum correlations 14.2.1-14.2.3 x x
Spontaneous emission and the Markov approximation 14.3.1-14.3.1 x
Fluorescence and Rayleigh and Raman scattering 14.3.2-14.3.4 x
Beam splitting and quantum amplification 14.4.1-14.4.6
The laser 14.5.1-14.5.2
Squeezed states 14.6.1-14.6.5
Photon counting statistics 14.7.1-14.7.3
Forces exerted by electromagnetic fields 15.1.1-15.1.3 x x x
Cooling and trapping 15.2.1-15.2.6 x x
Quantum measurement 16.1.1-16.7.1 x x
Quantum correlations and entanglement 17.1.1-17.1.3 x
Creation of quantum correlations 17.2.1-17.2.6 x
Quantum gates 17.3.1-17.3.5 x
The structure factor 18.1.1-18.1.1 x
Coupled dipoles model 18.2.1-18.1.2 x
Bragg scattering and photonic band gaps 18.2.1-18.3.3 x
Interaction of atoms with cavities 19.1.1-19.2.5
CARL & consorts in the classical and quantum regime 20.1.1-20.3.4 x
Statistical models of CARL 21.1.1-21.3.2 x
Manipulation of atomic gases 22.1.1-22.6.5 x
Thermodynamics of degenerate gases 23.1.1-23.2.8 x
Bose-Einstein condensation 24.1.1-24.6.4 x
Superfluid properties of condensed gases 25.1.1-25.4.3 x
Coherent properties of condensed gases 25.5.1-25.6.3 x
Theories on the interaction of light and BECs 26.1.1-26.2.2 x
Bragg diffraction by BECs and matter wave superradiance 26.3.1-26.4.4 x
BECs in superpositions of states coupled by light 26.5.1-26.5.2 x
Interaction of BECs with optical cavities 26.6.1-26.6.10 x
Condensates in reduced dimensions 27.1.1-27.1.3 x
0.2. ORGANIZATION OF THE COURSE 3
0.2 Organization of the course

Information and announcements regarding the course will be published on the website:
http://www.ifsc.usp.br/ strontium/ − > Teaching − > Semestre
The student’s assessment will be based on written tests and a seminar on a special topic
chosen by the student. In the seminar the student will present the chosen topic in 15 minutes.
He will also deliver a 4-page scientific paper in digital form. Possible topics are:
- Observation of super- and subradiant spontaneous emission of two ions [PRL 76, 2049 (1996)],
- Squeezed states,
- The Jaynes-Cummings model,
- Quantum projection noise,
- Quantum gates,
- The method of quantum Monte-Carlo wavefunction simulation,
- The quantum Zeno effect,
- Bloch equations: derivation and interpretation,
- The quantum jumps, its history and observation,
- Schrödinger’s cat,
- The Einstein-Podolski-Rosen hypothesis and its experimental falsification,
- Elitzur and Vaidman bomb testing problem,
- Topological phases and the Aharonov-Bohm effect,
- Quantum non-demolition measurements,
- Calculation of photoelectric effect from Fermi’s golden rule,
- Quantum correlations and the experiments of Young and Hanbury-Brown-Twiss,
- Bose-Einstein condensation.
0.3 Recommended bibliography

J. Weiner and P-T. Ho, Light-Matter Interaction: Fundamentals and Applications (Springer-
Verlag, Berlin, 2003)
R. Loudon, The quantum theory of light (Oxford Science Publications, Oxford, 1973)
P. Meystre and M. Sargent III, Elements of Quantum Optics (Springer-Verlag, Berlin, 1990)
I.I. Sobelman, Atomic Spectra and Radiative Transitions (Springer Verlag, Berlin, 1977)
M. Weissbluth, Photon-Atom Interactions (Academic Press, Boston, 1989)
C. Cohen-Tannoudji, B. Diu, F. Laloe, Quantum mechanics, vol. 1, (Wiley Interscience)
L.I. Schiff, Quantum mechanics (McGraw-Hill Book Company)
Ch.J. Foot, Atomic physics, (Oxford Master Series in Atomic, Optical and Laser Physics, 2005)
J.J. Sakurai, J.J. Napolitano, Modern Quantum Mechanics, 2nd ed., (2011)
4 CHAPTER 0. PREFACE
Part I
Quantum Mechanics
5
Chapter 1
Antecedents of quantum mechanics

1.1 Constants in atomic physics
Physical constants:
Velocity of light c = 299792458 m/s
Planck’s constant ~ = 1.05457266 × 10−34 Js
Atomic mass unit u = 1.6605402 × 10−27 kg
Boltzmann constant kB = 1.380658 × 10−23 J/K
Faraday constant F = 96485.309 C/mol
Vacuum permeability µ0 = 10−7 Vs/Am
Gravitational constant γ = 6.67259 × 10−11 m3 kg−1 s−2
Electron mass me = 9.1096 × 10−12 kg
Electron charge e = 1.60217733 × 10−19 C
g-factor of the electron g = 2.002319304386
Derived constants:
Fine-structure constant α = e2 /4πε0 ~c ≈ 1/137
Avogadro constant NA = 1/uA × 1g/mol = 6.0221367 × 1023 1/mol
Gas constant R = NA kB = 8.314510 J/mol K
Vacuum permittivity ε0 = 1/µ0 c2 = 8.8542 × 10−12 As/Vm
Bohr radius aB = 4πε0 ~2 /me e2 = 0.529 × 10−10 m
Bohr magneton µB = e~/2me = 9.27 × 10−24 J/T
Classical electron radius re = α2 aB
Rydberg constant R∞ = me cα2 /2h = 13.7 eV
Compton wavelength λC = h/me c
Thomson cross section σe = (8π/3)re2
Muon mass mµ = 105.658389 MeV
Proton mass mp = 938.27231 MeV
Neutron mass mp = 939.56563 MeV
Deuteron mass md = 1875.61339 MeV
1.1.1 Atomic units

A system of units commonly used in atomic physics is the one of atomic units. This system is
based on the system of Gaussian units (CGS) 1 defined by,
√
ecgs = e/ 4πε0 , aB = 1/α × ~/me c = ~2 /me e2cgs , ~=1. (1.1)
1
See the script Eletrodinamics by the same author [55] .
7
8 CHAPTER 1. ANTECEDENTS OF QUANTUM MECHANICS
With this fixing we give the energy in terms of e2cgs /aB , the wavevector in terms of 1/aB , the
distance in terms of aB and the mass in terms of me , such that,
Ẽ = E/(e2cgs /aB ) , (1.2)

k̃ = kaB ,
R̃ = R/aB ,
µ̃ = µ/me .
This notation simplifies the formulas. For example:

q q
k = 2µ ~2
(E − V ) fica k̃ = 2µ̃(Ẽ − Ṽ ) , (1.3)
e2cgs a5B C6
V = C6 R6
fica Ṽ = R̃6
.
1.1.2 Clebsch-Gordan symbols

The Clebsch-Gordan coefficient describes spin coupling 2 . The Clebsch-Gordans are related to
Wigner’s {3j}-symbols,

j1 j2 j j1 j2 j j1 −j2 +m
p
= = (−1) ∆(j1 j2 j3 )× (1.4)
m1 m2 m m1 m2 −m
p
× (j1 + m1 )!(j1 − m1 )!(j2 + m2 )!(j2 − m2 )!(j + m)!(j − m)!
X (−1)t
,
t
t!(m1 − j2 + j + t)!(−j1 − m2 + j + t)!(j1 + j2 − j − t)!(j1 − m1 − t)!(j2 + m2 − t)!
with the abbreviation,
(j1 + j2 − j3 )!(j1 − j2 + j3 )!(−j1 + j2 + j3 )!

∆(j1 j2 j3 ) ≡ . (1.5)
(j1 + j2 + j3 + 1)!
The {6j}-symbol describes the recoupling of of two spins,

X (−)t (t + 1)!
j1 j2 j3 p
= ∆(j1 j2 j3 )∆(j1 J2 J3 )∆(J1 j2 J3 )∆(J1 J2 j3 ) , (1.6)
J1 J2 J3 f (t)t
where f (t) = (t − j1 − j2 − j3 )!(t − j1 − J2 − J3 )!(t − J1 − j2 − J3 )!(t − J1 − J2 − j3 )!(j1 + j2 +

J1 + J2 − t)!(j2 + j3 + J2 + J3 − t)!(j3 + j1 + J3 + J1 − t)!.
The {9j}-symbol can be evaluated by,
 
 j1 j2 J12  X
2g j1 j2 J12 j3 j4 J34 J13 J24 J
j3 j4 J34 = (−) (2g + 1) . (1.7)
  J34 J g j2 g J24 g j1 j3
J13 J24 J g
{9j}-symbols satisfy the following orthogonality relation,

  
X  j1 j2 J12   j1 j2 J12 
Jˆ12 Jˆ34 Ĥ13 Ĥ24 j3 j4 J34 j3 j4 J34 = δJ13 H13 δJ24 H24 . (1.8)
  
J12 ,J34 H13 H24 J J13 J24 J
2
Ver [11], p.111 ou [241], p.119.
1.1. CONSTANTS IN ATOMIC PHYSICS 9
1.1.3 Functions and polynomials

1.1.3.1 The Gauss function
Indefinite integrals:
Z p Z
∞
−ax2
∞
2 Γ( n+1
2 )
e dx = π/a e xn e−ax dx = n+1 . (1.9)
−∞ 0 2a 2
Higher momenta:
Z x1 x1 Z x1
−ax2 −ax2 2
e dx = xe + 2a x2 e−ax dx . (1.10)
x0 x0 x0
1.1.3.2 Hermite polynomials
Definition:
Z ∞
2 dn −x2 2n 2
Hn (x) = (−1)n ex n
e =√ (x + it)n e−t dt (1.11)
dx π −∞
Hen (x) ≡ 2−n/2 Hn (x) .
Orthogonality:
Z ∞
2
e−x Hm (x)Hn (x)dx = 0 for m 6= n . (1.12)
−∞
Recursion:
d
Hn (x) = 2Hn−1 (x) (1.13)
dx
d −x2 2
e Hn (x) = e−x Hn+1 (x)
dx
Hn+1 (x) = 2xHn (x) − 2nHn−1 (x) .
Particular values:
H2n+1 (0) = 0 (1.14)

n n
H2n+1 (0) = (−1) 2 (2n − 1)!!
Series:
∞
X (2n)! (−1)n−k
H2n (x) = (2x)2k (1.15)
(2k)! (n − k)!
k=0
X∞
(2n + 1)! (−1)n−k
H2n−1 (x) = (2x)2k+1
(2k + 1)! (n − k)!
k=0
int(n/2)
X 1 (−1)k
Hn (x) = n! (2x)n−2k .
k! (n − 2k)!
k=0
1.1.3.3 Laguerre polynomials

Definition:
ex x−m dn −x n+m
L(m)
n (x) ≡ (e x ) (1.16)
m! dxn
Ln ≡ L(0)
n (x) .
Series:
n
X
n + m (−x)k
L(m)
n (x) = . (1.17)
n−m k!
k=0
Recursion:
d (m) (m+1)
L (x) = −Ln−1 (x) . (1.18)
dx n
Related functions:
q
2
umn (ε) ≡ e−ε · (iε)n−m · m! n−m 2
n! · Lm (ε ) (1.19)
q
umn (0) ≈ (iε)n−m · m!(n−m)!
n!
2
√
un+1,n (0) ≈ iε · n + 1 .
Fourier transforms:
Z ∞ √ √
2 (−1)k π 2 /4a
e−ax x2k cos xp · dx = √ · ep · He2k (p/ 2a) (1.20)
2k a2k+1
−∞
Z ∞ √
2 /2 (−1)m π 2 /2
e−x x2m L2m 2
n (x ) cos xp · dx =
√
2n!
· e−p · Hen (p)Hen+2m (p)
−∞
Z ∞ √
2 /2 (−1)m π 2 /2
e−x x2m+1 L2m+1
n (x2 ) sin xp · dx = √
2n!
· e−p · Hen (p)Hen+2m+1 (p)
−∞
Z ∞ Z ∞ Z ∞
−ax−bp 1 −ax −bp
e f (|x − p|) · dxdp = a+b e f (x)dx + e f (p)dp .
−∞ −∞ −∞
1.1.3.4 Legendre polynomials

Definition:
1 dn
Pn (x) ≡ (x2 − 1)n (1.21)
2n n! dxn
dm
Pn(−m) (x) ≡ (1 − x2 )m/2 Pl (x) .
dxm
Series:
" (n−m)!(m+n+1)! 1−x
#
(−1)m (n + m)! 1 − +
1!(m+1)! 2
Pn(m) (x) = m (1 − x2 )m/2
1−x 2
. (1.22)
2 m!(n − m)! + (n−m)!(n−m+1)!(m+n+1)!(m+n+2)!
2!(m+1)!(m+2)! 2 − ...
1.1.3.5 The Gamma function

Definition:
√
Γ(x + 1) = xΓ(x) , Γ(1/2) = π. (1.23)
1.2. THE DISCOVERY OF THE ATOM 11
1.1.3.6 Riemann zeta-function

The definition of the Riemann zeta-function is,
gη (1) = ζ(η) , (1.24)
where,
∞
X Z ∞
(±z)t 1 xη−1 dx
gη (z) = ± = , (1.25)
tη Γ(η) 0 z −1 ex ± 1
t=1
is also called the Bose/Fermi function. The upper sign holds for bosons, the lower for fermions.
The Sommerfeld expansion,
Z ∞ Z y Z ∞ Z x
g(x)dx g(y + x)η−1 dx g(y − x)η−1 dx
x−y
= g(x)dx + x
− (1.26)
0 e +1 0 0 e +1 0 ex + 1
Z y
π2
≈ g(x)dx + g 0 (x) + ...
0 6
holds for z 1 and yields,

xη π 2 η(η − 1) 7π 4 η(η − 1)(η − 2)(η − 3)
fη (ey ) ≈ 1+ + + ... . (1.27)
Γ(η + 1) 6x2 360x4
For small z both functions converge towards,
Z ∞ η−1
1 x dx
cη (z) = = cη−1 (z) = z . (1.28)
Γ(η) 0 z −1 ex
The derivative is,
∞ ∞ ∞
∂fη (Z) ∂ X (−Z)t X (−Z)t−1 1 X (−Z)t fη−1 (Z)
= − η = η−1
= − η−1 = , (1.29)
∂Z ∂Z t t Z t Z
t=1 t=1 t=1
or,
Z ∞ Z ∞ η−1 −2 x
∂fη (Z) 1 ∂ xη−1 dx 1 x Z e dx
= −1
= (1.30)
∂Z Γ(η) ∂Z 0 Z e + 1 x Γ(η) 0 (Z −1 ex + 1)2
Z ∞ Z
1 d −1 η−1 ∞ xη−2 fη−1 (Z)
= xη−1 −1 x
dx = −1
dx = .
ZΓ(η) 0 dx Z e + 1 ZΓ(η) 0 Z ex + 1 Z
1.2 The discovery of the atom

The fundamental idea of quantum mechanics is the assumption of the existence of entities that,
reaching a limit, can no longer be subdivided. Examples are the mass of a body, the speed of
an electron orbiting an atom, or the intensity of a beam of light. This idea was first uttered
by Leucippus 500 years a.c. and his student Democritus, who imagined matter being made
of smallest particles which they called atoms. These atoms move freely, collide, combine, and
separate: ’There is nothing more than atoms and free space.’ The microscopic atoms would
have the same characteristics as the macroscopic objects they form when they combine, for
example, color and shape. The idea of the atom resurfaced and was refined in the course of the
18th century (see Tab. 1.1 below). Today, we know that the basic idea was good, but the reality
is a little more complicated.
Table 1.1: Historical sketch of the quantization of matter
500 a.c. Democritus invention of the atom

1800 Avogadro, Dalton reinvention of the atom
1897 Thomson charge transport, model of raisins in a cake
1909 Rutherford, Geiger, Marsden α-scattering, concentration of charge in nucleus
1911 Rutherford planetary model
1900 Bohr quantized orbitals
1923 de Broglie matter has characteristics of waves
1927 Davisson, Germer, Stern electron and atoms diffraction experiments
At the end of the 19th century, the physical world seemed simple: matter and light was all
that existed. Matter was made up of atoms and light was a wave. Therefore, to describe a real
system, it was enough to calculate the trajectories of its elementary particles, the propagation
of light and the way they interact. We now know that life is not as simple, and that atoms are
also waves and light also behaves like particles.
Frictions between the old notions and new observations appeared in the late 19th century,
such as the ultraviolet divergence of black-body radiation. The pioneer of the new ideas was
Max Planck who, in 1905, with a little help from Einstein quantized the electromagnetic field,
and therefore the light, into small harmonic oscillators. This was the starting point for the
development of a new theory called quantum mechanics. Soon, this theory was applied to
explain the photoelectric effect. The second important step was initialized by Niels Bohr, who
quantized the hydrogen atom in 1913 into discrete excitation levels.
Table 1.2: Historical sketch of the quantization of light
1801 Young light is diffracted like a wave

1860 Maxwell unified theory of electrodynamics including light
1888 Hertz detection of radio waves
∼ 1890 accurate measurements of the black-body radiation spectrum
1900 Planck quantum hypothesis: E = hν
1905 Einstein photoelectric effect, light behaves like a particle
1.2.1 Democrit’s model

’The principles of reality are atoms and emptiness while other things are mere opinions.’ This is
a quotation from the Greek philosopher Democritus 400 years before Christ and before Socrates.
Together with his teacher Leucippus, he formed the first idea of indivisible particles: atoms.
Democritus’ work only survived as second-hand accounts, the major part of it having been
written down by Aristotle, who also, defending the idea of the continuum, was the greatest critic
of Democritus’ theory. Aristotle said that the reasoning that guided Democritus to affirm the
existence of atoms was as follows. For a body to change its shape, it is necessary that its parts
can move. This presupposes an emptiness (or vacuum) in which the matter moves. But if matter
were divided infinitely into ever smaller parts, it would loose its consistency. Nothing could be
formed because nothing could arise from the ever more infinitely deep dilution of matter into
emptiness. Hence, he concluded that the division of matter can not be infinite, that is, there is
an indivisible limit: the atom. ’There is only atoms and emptiness’, he said.
Figure 1.1: Democritus and dust in a sun ray.
Observing dust particles in a whirling motion within a ray of sunlight, Democritus was led to
the idea that atoms would behave in the same way, randomly colliding, some crowding, others
dispersing, others never yet joining with another atom.
The consistency of clusters of atoms, which makes something look solid, liquid, gaseous, or
animated (which is the state of the soul) would then be determined by the shape of the atoms
involved and their spatial arrangement. In this sense, water atoms are smooth and slippery;
the atoms of steel have shapes with sharp edges that hold them solidly together; the atoms of
salt, as their taste shows, are harsh and pointed; the atoms of air are small and little connected,
penetrating all other materials; and the atoms of soul and fire are spherical and very delicate.
Figure 1.2: Atoms of steel and air, atoms of the soul, and Bohr’s atom model.
We know nowadays that Democritus’ first theory of the structure of matter was very close
to the truth: There really are indivisible particles called atoms composed of a nucleus and an
electronic shell, and the space between the atomic nuclei is, in fact, quite empty.
The atomic hypothesis came to be reborn in the modern age with the scientists Boyle,
Clausius, Maxwell, and Boltzmann due to their successful explanations of the properties of
gases based on the so-called kinetic theory, where they assumed a gas being constituted of
identical molecules that collided elastically with each other and with the walls of the recipient
containing them. The discovery of the atom through the laws of proportions in chemistry and
the establishment of Avogadro’s number considerably strengthened the atomic hypothesis. The
hypothesis was definitely consecrated with the various experiments that established the charge
of the electron and the mass ratio between electrons and protons.
By the beginning of the 19th century the atomic nature of matter had definitely been es-
tablished, and the basic composition of the atoms was already relatively well known. It was
known, through experiments, that electrons could be removed from neutral atoms thus creating
positively charged ions and that only a certain number of electrons could be removed from each
atom. This number proved to be dependent on the atomic species and was called the atomic
number Z. This information was fundamental for establishing the basic composition of atoms.
The question that arose at this point concerned the dimensions and configurations of the atomic
system. How would loads and masses be distributed in this entity?
1.2.2 Thomson’s model and Rutherford’s experiment

The internal structure of a body can be studied by throwing beams of small particles against it.
The detection of the angular distribution of the scattered particles gives access to the structure
factor of the body. In crystallography we throw X-rays into super-complicated molecules to
learn the architecture e.g. of proteins. And in medicine, X-rays reveal the internal structure
of the human body. Obviously, the scattering technique is an extremely powerful tool, used in
many areas of modern physics.
In a series of experiments done before 1911, Ernest Rutherford analyzed the internal structure
of gold atoms using α-particles, i.e., He2+ atoms. The experiments carried out by Geiger,
Marsden, and Rutherford consisted of observing the deflection of particles from a collimated
beam when scattered by a thin metallic sheet (gold of thickness ∼ 1 µm) carefully obtained by
electroplating [see Fig. 1.3(cd)].
Figure 1.3: Comparison of Rutherford scattering by free electrons and electrons strongly bound
to small nuclei. (a) Thomson’s ’raisin-in-a-pudding’ type atom; (b) Rutherford’s ’planetary’
atom. (c) Rutherford scattering by a raisin pudding atom and (d) by a planetary atom.
The atomic model proposed by Joseph John Thomson suggests a structure resembling a pud-
ding with raisins: the electrons would be homogeneously distributed within an extended nucleus
(size 0.1 nm) of positive charge thus compensating for the negative charge of the electrons. The
α-particles would penetrate the gold nucleus, perceived as almost homogeneous, but would suf-
fer multiple deflections due to collisions with the disordered electrons within the nucleus. Since
electrons are very light, the angle of deflection θ would be small, even after many collisions.
For this model we expect a Gaussian dependence of the particles’ deflection angle given by the
scattering cross section [see Fig. 1.3(a-b)],
dσ 2 2
∝ e−θ /θ0 , (1.31)
dΩ
where θ0 is a small angle.
However, the measurements performed on this Rutherford scattering showed different results:
• For a fixed scattering angle, the amount of particles scattered into a solid angle element
dΩ is proportional to the thickness of the metal foil.
• For a given fixed angle and a given metal sheet the amount of scattered particles in dΩ
2 , where E
varies inversely with Ekin kin is the kinetic energy of the α-particles.
• For a given energy and a given metal sheet, the number of particles scattered into dΩ is
proportional to (sin 2θ )−4 .
• For a given energy and sheet thickness, the number of particles scattered into dΩ in a
2 , where Z is the atomic number of the element that
given direction is proportional to Ztg tg
constitutes the sheet.
The extremely rare deflection of α-particles and their angular distribution can be understood
by the assumption that the positive charge is concentrated in a very small volume (∼ 1 fm, that
is 10000 times less than the size of the atom itself). This volume is called the atomic nucleus,
hence the denomination of nuclear model. Since most of the particles pass through the gold
sheet without hindrance, there must be a large gap between the nuclei. The electrons, which
move within a large (in comparison with the diameter of the nucleus) empty space (the vacuum)
around the nucleus, shield the positive nuclear charge, so that the atom appears outwardly
neutral.
Figure 1.4: (Left:) Trajectory of an α-particle. (Right:) Illustration of the scattering cross
section.
We now derive Rutherford’s scattering formula from the hypothesis of a point-like nucleus.
Due to the repulsive action of the Coulomb force,
Zα Ztg e2
F = , (1.32)
4πε0 r2
we have for the trajectory of the α-particle (Zα = 2) a hyperbola [see Fig. 1.4(a)]. The large
half-axis of the hyperbola can be determined from the following ansatz,
Zα Ztg e2 1
Ekin = , (1.33)
4πε0 2a
where 2a is the minimum distance of the particle α, when it collides with the nucleus in a central
collision 3 . The distance a depends on the kinetic energy and can also be used for non-central
3
In a central collision, when the α-particle reaches the minimum distance 2a, its initial kinetic energy, Ekin is
fully converted into potential energy.
collisions. The collision parameter b is the minimum distance of the α-particle to the nucleus, if
it continued to fly in a straight line. In fact the α-particle will be deflected by an angle θ. From
the geometry of the hyperbola, as 2φ + θ = 180◦ , we obtain the following equation:

tan φ = ab = tan 90◦ − 2θ = cot 2θ , (1.34)
and therefore
b 8πε0 Ekin
cot 2θ = = b, (1.35)
a Zα Ztg e2
replacing a with the formula (1.33). Taking the derivative of this latter formula, we obtain a
relation between the width db of the hollow cone and the pertinent width dθ of the deflection
angle θ,
1 8πε0 Ekin
− 2 θ
dθ = db . (1.36)
2 sin 2 Zα Ztg e2
N
Let ntg = Vtg be the density of the particles in the target (Ntg atoms per volume V ) and x
the film thickness. Then σ = NAtg = VN/x tg
= ntg1 x is the average cross-section per atom sensed by
the α-particle on its way through the film. The probability P (θ)dθ for the α-particle of being
within a ring at distance b from the nucleus (whose area is 2πbdb) and being scattered into the
angle θ is then given by,
2πbdb
P (θ)dθ = = ntg x2πbdb . (1.37)
σ
These particles, i.e., dN of the N particles, are deflected into the hollow cone with the probability,
dN Zα Ztg e2 θ Zα Ztg e2 1 Zα2 Ztg

2 e4
cos 2θ
= P (θ)dθ = ntg x2π cot · · dθ = n tg x · dθ ,
N 8πε0 Ekin 2 8πε0 Ekin 2 sin2 2θ 64πε20 Ekin
2
sin3 2θ
(1.38)
where we replaced the parameters b and db with the expressions (1.35) and (1.36). The solid
angle of the cone can be expressed by,
dΩ = 2π sin θdθ = 4π sin 2θ cos 2θ dθ . (1.39)
Thus, the number dN of particles scattered to the solid angle dΩ remains,
dN Zα2 Ztg
2 e4
1
= ntg x 2 2 · dΩ . (1.40)
N 256π ε0 Ekin sin4 2θ
2
That is Rutherford’s scattering formula. Often, the formula is expressed with the differential
dσ
cross section dΩ . We get,
dN dσ
= = ntg xdσ , (1.41)
N σ
and therefore 2
dσ Zα Ztg e2 1
= , (1.42)
dΩ 4πε0 · 4Ekin sin4 2θ
with
dN dσ
= N ntg x . (1.43)
dΩ dΩ
Here, we have to make some comments:
• The angle θ = 0 is not defined, since there exists a minimum deflection angle θmin . This
angle is reached, when the α-particle moves at the distance b = bmax from the atom, that
is, at the edge of the circular area of the cross section. For a greater collision parameter
b, the α-particle traverses the field of the next neighboring atom, and the deflection angle
increases again. We have:
A θmin Zα Ztg e2
σ= = πb2max e 2 ' tan θmin
2 = , (1.44)
Ntg 8πε0 Ekin · bmax
simply by inverting the formula (1.35). For very large impact parameters, that is, when
the α-particle passes the atom outside its electronic layer, the electrons of the atom shield
the charge of the nucleus, an effect called screening.
• For very high energies, the distribution of the nuclear charge over a finite volume influences
the scattering, calling for corrections in the Rutherford formula. Moreover, at short inter-
nuclear distances, nuclear forces appear additionally to the electromagnetic interaction.
• The integral over the probability distribution P (θ)dθ is normalized,
Zπ
P (θ)dθ = 1 . (1.45)
θmin
Similarly, we have for the surface integrals,

Z
dσ
dΩ = σ . (1.46)
dΩ
θ>θmin
-25
10
N
-30
10
0 50 100 150
φ
Figure 1.5: (Code: QM Antecedents Rutherspectrum.m) Angular dependence of the cross-

section corresponding to Thomson’s (green) and Rutherford’s (red) models.
Rutherford derived the formula (1.42) describing the scattering of α-particles within classical
physics. A derivation from the laws governing quantum mechanics using the Born approximation
shows that Rutherford’s formula describes scattering correctly in first order, and that purely
quantum effects present only minor corrections. We will review the Rutherford scattering in
Excs. 1.4.1.1 and 1.4.1.2 and discuss the screening effect in Exc. 1.4.1.3.
1.2.3 Emission of radiation in the planetary model

The planetary model proposed by Rutherford suggests electrons spinning around a positively
charged nucleus in circular orbits 4 . This motion of electrons should obey the laws of Maxwell’s
electrodynamic theory. Let us now calculate some consequences of this picture.
We now treat the atom as a rotor where the negative particle, the electron, orbits the positive
particle. The dipole moment is,
p0 = −er . (1.47)
We calculate in the Exc. 1.4.1.4 the power emitted by the acceleration a = ω 2 r of the electron
on its circular trajectory,
µ0 ω 4 p20
P = . (1.48)
12πc
The initial energy of the electron spinning around the nucleus (for a hydrogen atom Z = 1),
p2 e2 me ω 2 r2 e2
E= − = − , (1.49)
2me 4πε0 r 2 4πε0 r
is dissipated by radiation of the power (1.48), i.e.,
dE dr e2 dr dr
−P = = me ω 2 r + 2
= 2me ω 2 r . (1.50)
dt dt 4πε0 r dt dt
The latter equation supposes an equilibrium between the centrifugal force and the Coulomb
force,
e2
me ω 2 r = , (1.51)
4πε0 r2
allowing link the revolution frequency ω to the instantaneous radius of the orbit r(t). Resolving
the Eq. (1.50) by ṙ and replacing the power by the relation (1.48) and the frequency ω by the
relation (1.51), we obtain,
dr P µ0 ω 2 e2 e4 1
=− 2
= − r = − 2 . (1.52)
dt 2me ω r 24πme c 96π ε0 me c r2
2 2 3
Integration of this equation gives,
32π 2 ε20 m2e c3 3

t − t0 = − [r − r3 (t0 )] . (1.53)
e4
Now inserting t0 = 0 and assuming r(t0 ) = aB , the time τ within which the loss of energy due
to radiation emission decreases the radius of the electronic orbit to r = 0, is,
32π 2 ε20 m2e c3 a3B

t=τ = . (1.54)
e4
Insertion of the numerical values gives the decay time τ ∼ 10−10 s. This is the effect called
radiation collapse of the classical atomic model.
4
This type of model had already been proposed by Jean Perrin in 1901 and by Hantaro Nagaoka in 1903,
around the same time when Thompson developed his model. The planetary model was later on rescued by John
William Nicholson in 1911.
1.2.4 Zeeman effect in the planetary model

The orbital motion of the electron generates a ring current I = e/T = eω/2π, which produces
an orbital magnetic moment which, as shown in Exc. 1.4.1.5, can be calculated following the
laws of electromagnetism,
eω 2
µ
~ ` = IAn̂ = πr n̂ , (1.55)
2π
where A = πr2 is the area of the trajectory. Introducing the angular momentum L = me ωr2 n̂
we get in vector notation,
e
µ
~` = L. (1.56)
2me
We now imagine this atom in the presence of a magnetic field B oriented in the direction
that we will call z. This results in a precession of the magnetic moment around the field (similar
to the precession of a spinning top in the presence of a gravitational field) governed by the
equation,
dL e
=µ ~` × B = L × B = −ΩL × L ,
dt 2me
e
with ΩL = 2m e
B representing the precession frequency and being called Larmor frequency. It
is evident that the presence of the magnetic field considerably alters the state of the atom, even
producing profound modifications in the frequency of the orbit of the electron ω0 and therefore
in the energetic state of the atom. This change is called Zeeman effect.
The Zeeman effect can be calculated by imagining that the field has an arbitrary direction
with respect to L. In this case, the equation describing the electronic motion as resulting from
an equilibrium between the centrifugal force and the Coulomb force needs to be complemented
by a Lorentz force,
me r̈ + me ω02 r = FL = −ev × B . (1.57)
where mr̈ is the centrifugal force due to the circular motion of the electron and me ω02 r the
centripetal force due to the Coulomb attraction exerted by the nucleus. Assuming the direction
of the magnetic field given by B = Bêz with B = 2me ΩL /e, the equations of motion can be
decomposed into,
ẍ + ω02 x + 2ΩL ẏ = 0 (1.58)

ÿ + ω02 y − 2ΩL ẋ = 0
z̈ + ω02 z = 0 .
The z-direction is not influenced. With the ansatz x = aeiωt e y = beiωt we obtain the system
of equations,
a(ω02 − ω 2 ) + 2iΩL ωb = 0 (1.59)

b(ω02 2
− ω ) − 2iΩL ωa = 0 ,
which has a non-trivial solution for a and b only when the determinant of the coefficients of a
and b vanishes: 2
ω0 − ω 2 2iΩL ω
0= = ω 4 − (2ω02 + 4Ω2L )ω 2 + ω04 . (1.60)
−2iΩL ω ω02 − ω 2
We get, r q
1 Ω2L
ω = ω1,2 = ω02 + 2Ω2L ± 2ΩL ω02 + Ω2L = ω0 ± ΩL + + ... , (1.61)
2 ω0
or, as ΩL ω, we get ω1,2 = ω0 ∓ ΩL . The result is a splitting of the energy levels proportional
to the magnetic field,
~e
∆E = 2~ΩL = B = 2µB B , (1.62)
me
where the abbreviation µB = e~/2me ' 9.27 · 10−24 JT−1 is called the magneton Bohr.
Although the classical derivation shows quantitative deviations from experimental observa-
tions, it is quite interesting, as it illustrates several aspects which have a quantum mechanical
equivalence.
Example 1 (Stern-Gerlach experiment): Among several historical experiments carried

out to unravel the atomic structure, one of the most important is the experiment carried out
by Otto Stern and Walter Gerlach in 1922 to measure the magnetic moment of atoms. The
results of this experiment once again demonstrated the need for new concepts to explain the
observations.
Using Bohr’s quantization rule, L = n~, within the formula (1.56) we get,
L
~ = −µB
µ .
~
In the presence of a magnetic field the dipole undergoes an interaction W = −~
µ · B, and
therefore a feels a force,
F = −~µ · ∇B .
By subjecting beams of atoms to the gradient of a magnetic field and detecting this force,
Stern and Gerlach were able to measure the magnetic moment produced by the rotation of
the electrons around the atomic nuclei.
1.2.5 Bohr’s theory and its limitations

The classical model of the planetary atom provides a mechanical illustration of the microscopic
world but fails to quantitatively explain experimental observations such as the discrete nature
of atomic spectra.
The radiation emitted by hydrogen atoms is characterized by discrete, spectrally very thin
lines. The observed lines are grouped in series named after Lyman, Ballmer and others,

1 µ 1 1
= RH 2
− 2 , (1.63)
λ me m n
where m and n are integers. RH = (1/4π)2 (me e4 /4π~3 c) is the Rydberg constant and µ =
me mat /(me + mat ) the reduced mass.
The discrete nature of spectral lines and the problem of the radiation collapse led Niels Bohr
to formulate the following postulates:
1. There are specific stationary orbits, where electrons do not emit energy.
2. Each emission or absorption of radiation energy by electrons comes with a transition

between stationary orbits. The radiation emitted during this transition is homogeneous.
3. The laws of mechanics can describe the dynamic equilibrium of electrons in stationary
states, but fails to describe the transition of electrons between stationary orbits.
1.3. THE DISCOVERY OF THE PHOTON 21
Thus, Bohr’s model predicts the quantization of energy levels, known as first quantization of
quantum mechanics. The radii of the possible orbits can be calculated from the postulate that
the orbital angular momentum be quantized in units of ~, that is, the electrons form stationary
de Broglie waves along the orbits 5 . We discuss Bohr’s model in Excs. 1.4.1.6 and 1.4.1.7.
In the picture proposed by Bohr, the radiative decay happens as an abrupt transition of an
electron between an outer (more energetic) orbit and an inner (less energetic) orbit. Since the
energies of stationary orbits are very well defined, the emitted radiation is mono-energetic, i.e.,
the spectrum consists of discrete characteristic lines.
We note here that the picture of an abrupt transition of electrons between discrete states,
called the quantum jump, did not receive Schrödinger’s blessing. He rather imagined for elec-
trons, within his theory of quantum wave mechanics, wave-shaped orbitals instead of planetary
trajectories, thus avoiding the problem of radiation due to charge deceleration and the quantum
jump concept. According to him, during a transition between electronic orbits, the energy is
transformed into radiation gradually 6 .
1.3 The discovery of the photon

The concept of the nature of light has a variable history. Newton proposed around ∼1650 a
corpuscular model to explain Snellius’ law on the refraction of a light beam penetrating a crystal.
Around the same time Huygens found a wave-based interpretation. The two models predicted
different speeds of light within the dense medium. Newton found, that the speed of light is
greater in the medium than outside, while Huygens found the opposite 7 . In the late 1800’s
the wave nature of light was established through observations of interference effects confirming
Huygens’ hypothesis. At that time, the world appeared to be simple: Light was a wave and
matter was composed of particles. However, some observations made were incompatible with
this simplistic ideas, for example, the spectrum of blackbody radiation, the Compton effect,
the specific heat of the solid, the radiation pressure, and the photoelectric effect. All these
observations are readily understood by assuming a corpuscular nature of the light 8 .
Nowadays, knowing the theory of quantum mechanics, we know that both ideas have their
range of validity and that the electromagnetic radiation is dual: In general, propagation and
interference effects are best described by waves. However, when interacting with matter, light
tends to localize into small particles called photons.
1.3.1 Radiation in a conductive cavity

In the age of lasers a classical treatment of the emission and absorption of light may seem an
atavism. However, even with coherent and monochromatic radiation sources, the most commonly
used physical picture is that of a classical optical field interacting with an atom or a molecule
whose energetic structure is treated quantum mechanically. And even the atomic or molecular
5
A generalization of Bohr’s theory was provided by Arnold Sommerfeld. Assuming elliptical orbits for the
electrons he managed to explain some features of the fine structure, provided the motion of the electron was
treated relativistically. The basic premises were 1. stable orbits when the Coulomb attraction is balanced by
= nq ~, and 3. quantization of angular momentum
R
the centrifugal force, 2. quantization of phase space r q dq
Ldθ = nθ ~.
R
6
We note here, that quantum jumps were observed much later!
7
Note, until today there remain doubts about the correct value of the momentum of light in dielectric media
[181].
8
The corpuscular hypothesis is now called the second quantization of quantum theory or quantization of the
electromagnetic field.
dipole is often treated like a classical oscillator. The exposition of such a dipole to simple
boundary conditions prepares the analogous development of a quantum oscillator and provides
a direct path to quantization of the radiation field.
Even if we rarely do experiments by throwing light into a small hole in a metallic box, the
electromagnetic fields obtained by solving Maxwell’s equation are particularly simple for bound-
ary conditions, where the fields disappear on the inner surfaces of the box. Before discussing the
physics of radiation in a perfectly conducting cavity, we have to introduce some basic relations
between electromagnetic amplitudes, stored energy, and intensity.
The electric field of a plane wave oscillating with frequency ω and propagating through
vacuum in the direction of propagation defined by the wave vector,
2π
k= k̂ , (1.64)
λ
can be written,
E = E0 ei(k·r−ωt) , (1.65)
where E0 = E0 ê consists of an amplitude E0 and a polarization ê. Since the field E0 is transverse
to the direction of propagation, the polarization has two components perpendicular to k. The
magnetic induction field associated with the wave is,
1
B0 = E0 . (1.66)
c
For a propagating wave E and B are in phase, while for a standing wave they are out of phase.
For a given cavity mode we can express the standing wave in this mode as,
E = E0 (r)e−iωt . (1.67)
The energy of the electromagnetic field of a standing wave, averaged over one oscillation of the
frequency ω is, Z
1 ε0 2 1
Ū = |E| + |B|2 dV . (1.68)
2 2 2µ0
Now, the energy density of the oscillating electromagnetic field is given by,

dŪ 1 1
ū = = ε0 |E| + |B|2
2
. (1.69)
dV 4 µ0
From the equation (1.66) we can see that the contributions of the electric and magnetic fields
are equal. Therefore, Z
1 1
Ū = ε0 |E|2 dV e ū = ε0 |E|2 . (1.70)
2 2
Another important quantity is the flux of electromagnetic energy through a surface. The
Poynting vector describing this flux is defined by,
1
I= E×B . (1.71)
µ0
Again using the equation (1.66), we find the value averaged over a time period,
1
I¯ = ε0 c|E|2 . (1.72)
2
This quantity, called intensity, describes the fact that the flux is a density of energy multiplied
with the velocity of propagation in vacuum,
1
ūc = ε0 c|E|2 = I¯ . (1.73)
2
The intensity can also be written,

r
1 ε0
I¯ = |E|2 . (1.74)
2 µ0
p
where the factor µ0 /ε0 is called impedance of free space, because it has the unit of a resistance
and the last equation has the same form as the power dissipated in a resistor,
1V2
W = . (1.75)
2 R
1.3.2 Black body radiation

We now want to calculate the energy density inside the cavity before using the result to describe
the interaction between light and a sample of two-level atoms located inside the cavity. The
basic idea is to say that the electrons inside the conducting surface of the cavity oscillate because
of thermal motion. The oscillation generates a dipolar radiation leading to stationary waves
developing within the cavity. As the walls of the cavity are conducting, the electric field E must
disappear inside the wall and on its surfaces. The task is now twofold: first count the number
of possible standing waves, which satisfy the boundary conditions as a function of frequency;
second, determine the energy for each wave and then calculate the spectral distribution of the
energy within the cavity.
The equations describing the radiated energy in free space are,
1 ∂2E
∇2 E = and ∇·E=0 . (1.76)
c2 ∂t2
The stationary waves solutions separate into terms oscillating in time and in space. Now,
Figure 1.6: (Left) Cavity in position space showing the thermal motion of the electrons inside
the walls. (Center and right) Density-of-states in a cavity in momentum space.
respecting the boundary conditions for a three-dimensional box of length L, we have for the
components of E 9 ,
E(r, t) = e−iωt [êx cos(kx x) sin(ky y) sin(kz z) (1.77)
+êy sin(kx x) cos(ky y) sin(kz z)
+êz sin(kx x) sin(ky y) cos(kz z)] ,
with the components,
πnx
kx = para nx = 0, 1, 2, ... (1.78)
L
and similar for ky e kz . Note, that for each component Ex,y,z the transverse amplitudes disappear
in 0 and L. By inserting this solution into Helmholtz’s equation (1.76), we obtain,
ω2
kx2 + ky2 + kz2 =
. (1.79)
c2
The states kx,y,z (enumerated by nx,y,z ) form a three-dimensional orthogonal lattice of points in
space k separated by a distance along the axes kx , ky , kz of Lπ , as shown in Fig. 1.6. In principle,
the number of states that can be placed within a sphere of radius k in the momentum space is,
Z
N= dnx dny dnz . (1.80)
sphere
However, the periodic boundary conditions for |k| limit the components kx , ky , kz to positive
values (n ≥ 0), that is, the volume under consideration is limited to an octant. On the other
hand, we must multiply the number of states by two because of the degeneracy of polarizations,
Z n 3 Z k
2 L 4L3 k 3 4L3 ω 3
4N = 4πn dn = 4πk 2 dk = 2 = . (1.81)
0 π 0 π 3 3π 2 c3
With this, we obtain the mode density,
N ω3
= . (1.82)
L3 3π 2 c3
The spectral density of modes % can be given in several units,
Z Z Z
N
%(n)dn = %(k)dk = %(ω)dω = 3 , (1.83)
L
such that,
πn2 k2 ω2
%(n) =
3
ou %(k) = 2 or %(ω) = 2 3 . (1.84)
L π π c
The density of oscillating modes within the cavity grows like the square of the frequency.
Now, the mean energy per mode in a sample of oscillators in thermal equilibrium is, following
the equipartition law, equal to,
Ē = kB T , (1.85)
where kB is the Boltzmann constant. We conclude that the spectral energy density J in the
cavity is,
ω2
JRJ (ω)dω = kB T %(ω)dω = kB T 2 3 dω . (1.86)
π c
This law is known as the Rayleigh-Jeans law of black-body radiation. As seen in Fig. 1.7, this law
suggests the physically impossible fact, called ultraviolet catastrophe, that the energy storage in
the cavity grows without limits like the square of frequency.
9
See the script Electrodynamics by the same author [55].
2000
T = 3000 K
1500
u (eV)
1000
500
0
0 200 400 600
ν (THz)
Figure 1.7: (Code: QM Antecedents RayleighJeans.m) Spectral energy density following

Rayleigh-Jeans’ and Planck’s laws.
1.3.3 Planck’s distribution of modes

We obtained the result (1.85) by multiplying the number of modes with the mean energy per
mode. As there is no doubt about our method of counting the modes, the problem with the
ultraviolet catastrophe can only root in the use of the equipartition principle for assigning energy
to the oscillators.
Planck’s idea to solve this problem was to first consider the probability distribution for excit-
ing the modes (thermal states) for a sample of oscillators in thermal equilibrium at temperature
T . This probability distribution p comes from mechanical statistics P and can be written in terms
of the Boltzmann factor, e −E n /kB T , and the partition function q = ∞n=0 e
−En /kB T as,
e−En /kB T
pn = . (1.87)
q
Now Planck hypothesized that the energy be quantized, that is, it must be assigned in discrete
portions, proportional to the frequency, such that,
En = n~ω , (1.88)
−~ω/k T
P∞ ~ isn called Planck’s
where n = 0, 1, 2, .. and the proportionality constant
−1
constant. With the
abbreviation U ≡ e B and using the rule n=0 U = (1 − U ) , we find the average
number,
X X ∂ X n U 1
n̄ = npn = (1 − U ) nU n = (1 − U )U U = = ~ω/k T . (1.89)
n n
∂U n 1−U e B −1
The probability of occupancy of state n is,
n̄n
pn = (1 − U )U n = , (1.90)
(1 + n̄)1+n
and the average energy is,

X X ~ω
Ē = En pn = n~ωe−n~ω/kB T = , (1.91)
n n
e~ω/kB T −1
in contrast to the initial assumption (1.86).

Finally, we obtain Planck’s expression for the energy density inside the cavity by substituting
¯ for the factor kB T in Rayleigh-Jeans’ law (1.85),
ω2 ~ω
J (ω)dω = Ēρω dω = 2 3 ~ω/k
dω . (1.92)
π c e BT − 1
This result, drawn in Fig. 1.7, is much more satisfactory, because now the energy density has an
upper bound, and coincides with the results of experiments. Note, that for high temperatures or
low excitation energies, ~ω kB T , Planck’s distribution converges to that of Rayleigh-Jeans’,
J (ω) → JRJ (ω).
Solve the Excs. 1.4.2.1 to 1.4.2.4.
1.3.4 The corpuscular nature of the photon

1.3.4.1 The photoelectric effect
Light incident on a metallic surface can expel electrons. For this to occur, the light must have a
minimum frequency. If the frequency is below this value, there is no point in increasing the light
intensity: the electrons won’t be expelled. The main experimental observations are: 1. Electrons
are ejected without apparent delay, i.e. it is not necessary (and it doesn’t help) to accumulate a
certain amount of energy. 2. Higher light intensities increase the number of electrons, but not
their kinetic energy after expulsion. 3. Red light does not eject electrons, even at high intensities.
4. Weak ultraviolet light only ejects few electrons, but with high kinetic energy.
These observations challenge the classical electromagnetic model according to which the
Lorentz acceleration of the electrons should be proportional to the field amplitude. The obser-
vations were explained by Einstein’s theory of the photoelectric effect, which assumes the light
to be quantized (unlike Planck, who preferred to quantize the process of light absorption),
E = hν . (1.93)
Assuming a fixed exit work A for the extraction of an electron, we can measure the constant ~:
mv 2 eV
hν = A + 2 = A + eV → h= . (1.94)
ν − νg
The energy of the fastest electrons is measured through the decelerating voltage by varying ν
and I. We will discuss the photoelectric effect quantitatively later in the Exc. 5.5.4.5.
1.3.4.2 Bremsstrahlung and the Franck-Hertz experiment

Bremsstrahlung is, in a way, the inverse process of the photoelectric effect. Here, electrons are
accelerated toward a cathode. Finding a target they are rapidly decelerated, a process in which
they emit a continuous spectrum of X-rays (in addition to characteristic lines attributed to
electronic transitions in the target atoms). For any given kinetic energy the spectra have a red
threshold corresponding to photons that receive the entire energy of the electron.
In the Franck-Hertz experiment free electrons produced in a plasma are accelerated by a
strong electric field. Having traveled a sufficiently long distance they have acquired enough
kinetic energy to excite electronic transitions in the atoms of the plasma. When an excitation
occurs, the electron suddenly loses all its energy and must be accelerated again, starting from
rest, before it can excite another atom.
1.3.4.3 Radiative pressure and Compton scattering

When light is scattered from a particle, it transfers momentum to it called photonic recoil. This
effect, known as radiation pressure, occurs for example in Compton scattering.
X-rays scattered by the electrons of a carbon target are red-shifted by an amount, which
increases with the scattering angle. This is the Compton effect. The data are understood
assuming a corpuscular nature of light and applying the laws of conservation of energy and
momentum to the collision processes between photons and electrons. The scattered photon sees
its energy reduced and therefore its wavelength increased.
In a material where there are free electrons, this effect will occur at all photon energies.
In other materials, it is only observed with high energy photons. For high energy photons,
exceeding the atomic binding energy, the electrons can be considered free such that, in the
scattering process, the photon is able to eject the electron from its atom. The photon receives
the remaining energy and is deviated, such that the overall momentum of the system is conserved.
The loss of energy for the photon results in a spectral shift to the red during its passage through
the material.
Photons of visible light, on the other hand, do not have enough energy to eject bound
electrons. In this case, the mass in the Compton formula must be replaced by the atomic mass,
such that the spectral displacement becomes much smaller. This limit, which involves bound
electrons, is that of Thomson and Rayleigh scattering.
The relevance of this effect lies in the fact that it shows that light exhibits properties com-
monly attributed to corpuscles, since Thomson’s scattering model, based on the classical theory
of charged particles accelerated by electromagnetic fields, can not explain any spectral shift.
1.3.5 Einstein’s transitions rates

Bohr’s atom model 10 explained for the first, time how light interacts with matter: Atoms have
discrete excitation levels. They absorb and emit energy packets ~ω. Unfortunately, Bohr’s
model can not predict transition rates. Here, Einstein helped out by developing a useful theory
(see Fig. 1.8).
Figure 1.8: Bohr model and Einstein rate diagram.
We consider a two-level atom or a sample of atoms within a conducting cavity. We have

N1 atoms in the lower energy state E1 and N2 in the upper state E2 . Light interacts with
these atoms through stimulated resonant absorption and emission. The rates, B12 J (ω) and
B21 J (ω) are proportional to the energy spectral density J (ω) of the cavity modes. The central
idea of Einstein is to postulate that atoms in the higher state can emit light spontaneously
at a rate A21 , which depends only on the density of modes of the cavity, i.e. the volume of
the cavity, but not the energy of the field of radiation. With the Einstein coefficients we can
formulate valid rate equations in situations, where the spectral distribution of the radiation is
10
now called the first quantization or quantization of the excitation energy of the electrons of the atom
wider than the spectral width of the atomic transition and where the spectral distribution of
the light flux from the source, I¯ω , is weak compared to the saturation intensity of the atomic
transition. Even if modern light sources generally have very narrow and intense spectral emission
bands, Einstein’s coefficients are often used in the spectroscopic literature to characterize the
light-matter interaction with atoms and molecules.
The Einstein rate equations describe the energy flux between atoms and the optical modes
of the cavity,
dN1 dN2
=− = −R1→2 + R2→1 + S2→1 (1.95)
dt dt
= −N1 B12 J (ω) + N2 B21 J (ω) + N2 A21 .
R1→2 is the absorption rate, R2→1 the stimulated emission rate and S2→1 the spontaneous emis-
sion rate. The assumption of a third type of transition, called spontaneous emission, is necessary,
if B12 = B21 but N1 > N2 in thermal equilibrium. In thermal equilibrium we have the condition
of stationarity, dN dN2
dt = − dt = 0 for a given energy density value J (ω) = Jth (ω), such that,
1
A21
Jth (ω) = . (1.96)
N1
N2 B12 − B21
The Boltzmann distribution law controlling the distribution of the number of atoms in the lower
and upper states is given by,
N1 g1
= e−(E1 −E2 )/kB T , (1.97)
N2 g2
where g1,2 are the degeneracies of the lower and upper states and E2 − E1 = ~ω0 . We find,
A21
Jth (ω) = g1 ~ω0 /kB T . (1.98)
g2 e B12 − B21
But this result must be consistent with the Planck’s distribution (1.92). Therefore, by comparing
this equation with the equation (1.98), it must be,
g1 B12
=1. (1.99)
g2 B21
and also,
A21 ~ω 3
= 2 03 . (1.100)
B21 π c
This equation shows that, once we know one of the three transition rates, we can always calculate
the others.
It is useful to compare the rate A21 with B21 from the equation (1.98) inserting the equation
(1.99),
A21
= e~ω0 /kB T − 1 . (1.101)
B21 Jth (ω)
This expression shows that, when ~(ω2 − ω1 ) kB T , that is, for optical, UV, or X-ray frequen-
cies, spontaneous emission dominates. But in low-frequency regimes, that is, IR, microwave, or
radio waves, stimulated emission is more important. Note that even when stimulated emission
dominates, spontaneous emission is always present and plays an important role, for example, in
processes ultimately limiting the emission bandwidth of lasers.
1.3.6 Propagation of light in dielectric media

Let us now use Einstein’s theory to calculate the interaction of light with a gas being so diluted
that we can treat the interaction as occurring with individual atoms. We first consider the prop-
agation of light through a continuous (non-conducting) dielectric medium. The interaction of
light with such a medium allows us to introduce important quantities, such as the polarization,
the susceptibility, the index of refraction, the extinction coefficient, and the absorption coeffi-
cient. We shall see later that the polarization can be considered as densities of dipole moments
induced into the dielectric medium by the oscillating light field. But for the moment, we simply
begin by defining the polarization P̃ with respect to an applied electric field Ẽ as,
P̃ = ε0 χ̃e Ẽ . (1.102)
The tilde ∼ ornamenting the quantities means that they are complex, and that measurable
quantities must be taken from the real parts. χ is the linear complex electric susceptibility,
which is an intrinsic property of the medium responding to the light field. It is related to the
medium’s dielectric constant,
ε̃ = ε0 (1 + χ̃) . (1.103)
Near resonances, the susceptibility is a strong function of frequency and can be spatially
anisotropic. It is a complex quantity with a real dispersive part, χ0e , and an imaginary absorptive
part, χ00e ,
χ̃e = χ0e + iχ00e . (1.104)
Several expressions that we already know become modified in a dielectric medium,
2
kc
= 1 + χ̃e , (1.105)
ω
with χe = 0 in free space. In a dielectric medium, kc

ω becomes a complex quantity, conventionally
expressed by,
kc
= η + iκ , (1.106)
ω
where η is the refraction index and κ the extinction coefficient of the dielectric medium. The
relationships between the refractive index, the extinction coefficient and the two components of
the susceptibility are,
η 2 − κ2 = 1 + χ0e and 2ηκ = χ00e . (1.107)
Note, that in a transparent dielectric medium there is no absorption, such that, η 2 = 1+χ0e = εε0 .
Within a dielectric medium, we obtain the propagating wave solutions of Maxwell’s equations
by substituting k with the equation (1.106)
ηz
−t)−ω κc z
E = E0 eiω( c . (1.108)
The relationship between the amplitudes of the electric and magnetic fields is,11
(χ) √ √ p 1
B0 = εµE0 = ε0 µ0 1 + χ̃e E0 = (η + iκ)E0 = (η + iκ)B0 . (1.109)
c
We use the subscript (χ) to mark quantities within the dielectric medium. The average energy
density is,
1
ū(χ) = ε0 η 2 |E|2 = η 2 ū . (1.110)
2
11
In a dielectric medium, µ ' µ0 .
Now, the intensity of the light beam in a dielectric medium is attenuated,

1 1 1 (χ)
I¯(χ) = |E × B| = ε0 cη|E|2 = ε0 cηE20 e−2ωκz/c = I¯0 e−Kz , (1.111)
µ0 2 2
where
(χ)
I¯0 = 21 ε0 cηE20 (1.112)
is the intensity at the point, where the light enters the medium, and
ωκ ω
K=2 = χ00e (1.113)
c ηc
is called absorption coefficient. Note, that the energy flow I¯(χ) in the dielectric medium is always
the product of energy density and propagation velocity c/η. Note also, since the frequency ω
of the light propagating through the dielectric remains the same, the wavelength shrinks like
λ = c/ην [164].
1.3.7 Propagation of light in dilute gases

We are often interested in the attenuation of the intensity of a beam of light traversing a
dilute gas of resonant scattering atoms. The equation (1.111) describes this attenuation via the
properties of the dielectric material. What we are looking for in fact is a microscopic equivalent
description in terms of absorption and light emission rates. The Einstein rate equation yields
the temporal transition rates, but does not say how they relate to the spatial attenuation length
of the light beam. We now consider a beam propagating through a cell containing an absorptive
gas and assume that, along the optical axis, absorption and emission reach a steady state. We
begin with the Einstein equation (1.95), and write
0 = −N1 B12 ūω + N2 B21 ūω + N2 A21 . (1.114)
We now use the result (1.99) to write
g1
N2 A21 = ūω B12 (N1 − N2 ). (1.115)
g2
In steady state, the number of the excited atoms is,
ūω B12 N1
N2 = . (1.116)
A21 + gg12 ūω B12
Now, we must treat the refractive index of the dielectric medium carefully. The expression for
the energy density must be modified following Eq. (1.110)
(χ)
ūω
ūω = 2 . (1.117)
η
To use the Einstein coefficients, which suppose propagation at the speed of light in the vacuum 12 ,
we must correct the energy density in the dielectric medium before using it in the equation (1.116)
(χ)
[164]. Therefore, we should express ūω in this equation by ūω /η 2 ,
(χ)
ūω
N2 A21 = B12 (N1 − N2 gg12 ) . (1.118)
η2
12
In Einstein’s model, what really matters for inducing transitions is the flux of photons, that is, the number of
photons that traverse an atom per unit of time. This flux can not depend on the index of refraction. Therefore,
we must use the spectral energy density calculated in the vacuum ūω .
Multiplying the two sides with ~ω0 , we see that the left side describes the rate of energy scattered
out of the beam by spontaneous emission, while the right side describes the energy loss of the
beam, i.e., the difference between energy removed by absorption and energy returned to the
beam by stimulated emission,
(χ)
dŪ (χ) ūω
= −N2 A21 ~ω0 = − 2 B12 (N1 − N2 gg12 )~ω0 . (1.119)
dt η
1.3.8 Spectral line profiles

Every light source has a certain spectral width. Conventional light sources, such as incandes-
cent bulbs or plasmas have relatively broad emission bands compared to atomic or molecular
absorbers, at least when the latter ones are studied in dilute gases. Even when we use pure
spectral sources, such as a laser tuned to the peak of a resonance, the transition line always
exhibits an intrinsic width associated with the interruption of the phase evolution of the excited
state. Phase interruptions such as spontaneous or stimulated emission and collisions are com-
mon examples of line broadening mechanisms. The emission or absorption of radiation occurs
within a frequency distribution centered about ω0 ≡ ω2 − ω1 , and we must account for this
spectral distribution in our calculation of the energy transfer. Instead of using the expression
N2 A21 ~ω0 , we should express the rate of energy loss by spontaneous emission as,
Z ω0 +∆/2
N2 A21 ~ ωL(ω − ω0 )dω , (1.120)
ω0 −∆/2
R
where L(ω − ω0 ) is the function of the atomic absorption profile, usually normalized by L(ω −
ω0 )dω = 1. A common function in atomic spectroscopy is the Lorentzian,
γ0 dω
Lγ 0 (ω − ω0 )dω = 0 2 , (1.121)
2π γ
(ω − ω0 )2 + 2
with the spectral width γ 0 . The differential L(ω − ω0 )dω can be considered as the probability
of finding the emitted light in the frequency range between ω and ω + dω. In fact, we will
retrieve the Lorentzian profile, when discussing other contributions to the spectral width, such
as excitations with strong radiation fields (power broadening) or broadening due to collisions.
The width γ 0 is generally caused by several physical sources.
Note that L(ω − ω0 )dω is dimensionless. Now, by generalizing the equation (1.119), the
energy loss rate of the light beam becomes,
dŪ (χ) B12 g1

= −ū(χ)
ω ? N1 − N 2 g2 ~ωL(ω − ω0 ) (1.122)
dt η2
Z ∞
B12 g1

=− ū(χ)
ω (ω − ωL ) · N1 − N2 g2 ~ωL(ω − ω0 )dω . (1.123)
−∞ η2
The absorption probability distribution is convoluted with the energy distribution of the light
(χ)
source, ūω = dū(χ) (ω)/dω. The distribution of energy density is,
dū(χ) dŪ (χ)

= , (1.124)
dt V dt
Figure 1.9: Absorption spectrum (blue) and spectral energy distribution of the source (red).
where V is the volume of the cavity. Assuming that light propagates toward z and converting
the time dependence into a spatial dependence,
dū(χ) dū(χ) dI (χ)

= ·c= . (1.125)
dt dz dz
With this, looking at the equations (1.110) and (1.111), we see
c
Iω(χ) = ū(χ)
ω , (1.126)
η
and replacing (1.125) and (1.126) in (1.124), we finally get,
Z ∞
dI¯(χ) B12
=− Iω(χ) (ω − ωL ) · N1 − N2 gg12 ~ωL(ω − ω0 )dω . (1.127)
dz −∞ cηV
Now, if the light only weakly excites the gas, such that N2 N1 , we have,
Z ∞
dI¯(χ) B12
=− Iω(χ) (ω − ωL ) · n~ωL(ω − ω0 )dω , (1.128)
dz −∞ cη
where n = N/V ' N1 /V is the density of the gas. For a field of weak light and a dilute
gas, we can obtain a simple expression for the intensity behavior by approximating the spectral
distribution of the absorption by a Lorentzian centered in ω0 and having the width A21 /2π,
I (χ)
Iω(χ) = . (1.129)
A21 /2π
Now, Eq. (1.128) becomes, with a bit of algebra,

dI¯(χ) B12 2π~ω0
=− ndz . (1.130)
I¯(χ) A21 cη
The right-hand parenthesis of Eq. (1.130) can be considered as a cross-section for absorption
of resonant light. Using Eq. (1.100) we can express this cross section as,
g2 πλ20
σ0 = , (1.131)
g1 2η
such that Eq. (1.130) becomes
dI¯(χ)
= −σ0 ndz (1.132)
I¯(χ)
and
I¯(χ)
(χ)
= e−σ0 nz0 . (1.133)
I¯
0
Here, z0 is the total distance, over which absorption takes place. The last equation is the
integral version of the Lambert-Beer law for light absorption. It is very useful for measuring
atomic densities in gas cells or of atomic beams. Comparing the Eqs. (1.113) and (1.133), we
see that the absorption coefficient can be written as a product of the absorption cross section
and the density of the gas,
2ωκ ω
K= = χ00e = σ0 n . (1.134)
c ηc
Solve the Excs. 1.4.2.5 to 1.4.2.8.
1.3.9 Specific heat of solids

The Debye model applies Planck’s law on the distribution of energy in electromagnetic radiation,
which treats radiation as a gas of photons, to the energy distribution of atomic vibrations in
a solid, treating them as a gas of phonons in a box (the box being the solid). Most of the
steps of the calculation are identical, as both are examples of a massless bosonic gas with linear
dispersion relation.
According to the equipartition theorem, every atom has 3 degrees of freedom due to its
translational motion. Thus, in a crystal lattice with N atoms, we expect the total energy
E = 3N kB T and the specific heat should be, following the Dulong-Petit law,

∂E
CV = = 3N kB , (1.135)
∂T V
for all solids regardless of temperature.

It was observed, however, that the specific heat of solids decreases like CV ∝ T 3 as T
approaches zero. It was Einstein’s idea to apply Planck’s formula by treating the N atoms
as three-dimensional harmonic oscillators vibrating in a lattice. The discrete energies n~ω are
identified with quasi-particles called phonons. The quantum nature of atoms does not matter,
they just provide the medium supporting the phonons. Following the Bose-Einstein statistics,
we must replace,
kB T → ~ω/(e~ω/kB T − 1) , (1.136)
such that the derivative of the energy,
2
∂ 3N kB ~ω ~ω e~ω/kB T
CV = = 3N kB , (1.137)
∂T e~ω/kB T − 1 V kB T (e~ω/kB T − 1)2
gives the specific heat. The disappearance of the specific heat at low temperatures,
3N (~ω)2 −~ω/(kB T )
CV ' e , (1.138)
kB T 2
which is related to the finite localization energy of harmonic oscillators, does not describe ex-
perimental observations very well, and the model had to be refined by Debye, later on.
1.3.9.1 Debye model for the specific heat

While Einstein assumed monochromatic lattice vibrations, Debye’s approach was to allow a
spectrum of vibrational frequencies. With the density of states,
ρ(ν)dν = (4πV v 3 )ν 2 dν , (1.139)

where v is the velocity of sound propagation, the formula is totally equivalent to the density-
of-states for photons in a cavity. Assuming
R νm that there is an upper bound νm for the vibrational
frequencies, we normalize as 3N0 = 0 ρ(ν)dν. The energy now is 13 ,
Z νm Z θ/T
~ω 4πV T4 x3 dx
E= νdν = 9N k B . (1.140)
0 e~ω/kB T − 1 v3 θ3 0 ex − 1
The Debye temperature θ = hνm /kB is characteristic for the metal. The derivative is then,
" Z #
T 3 θ/T x3 dx θ 1
CV = 9N kB 4 − . (1.141)
θ 0 ex − 1 T eθ/T − 1
At low temperatures this formula reproduces the Debye law,

12π 4
CV = N kB (T /θ)3 . (1.142)
5
1.4 Exercises
1.4.1 The discovery of the atom
1.4.1.1 Ex: Analysis of Rutherford scattering
a. What conclusions can be drawn from the observation that Rutherford’s formula describes well
the scattering of charged particles traveling through matter over a wide range of parameters?
b. Why do we see a deviation from Rutherford’s formula for large energies?
c. The scattering of protons with energy E crossing a thin film of thorium is well described up
to energies of E = 4.3 MeV by Rutherford’s formula. Estimate for this case the range of nuclear
forces.
d. For small scattering angles θ we observe large deviations from Rutherford’s formula. Explain
why?
e. Assume the thorium atoms of item (c) to form a periodic crystal with the lattice constant
d = 10aB . At which minimum angle θ Rutherford’s formula loses its validity.
1.4.1.2 Ex: Rutherford scattering

a. A beam of α-particles with energy Ekin = 3 MeV and flux I = 5 · 103 s−1 impinges on a thick
gold film x = 1 µm. Using Rutherford’s formula, calculate how many particles are scattered in
∆t = 10 minutes in the range of angles 10◦ ≤ θ ≤ 30◦ .
b. Now, the gold film is replaced with an aluminum film of the same thickness. How many
α-particles are scattered under equal circumstances?
1.4.1.3 Ex: Screening of electrons

Consider thin layer of charge −Ztg e with radius R. This screening causes a scattering angle,
p
D 1 − (b/R)2
tan 2θ = ,
2b 1 + D/2R
3Ze2 dσ
with D ≡ m2 v 2 /2
for b < R. Verify how the screening changes the differential cross section dΩ .
13
The fact that the electron gas also has a heat capacity is neglected.
1.4. EXERCISES 35
1.4.1.4 Ex: Radiation of an oscillating dipole

Calculate the angular distribution of the power radiated by an oscillating electric or magnetic
dipole from expressions for the emitted electric and magnetic fields found in literature.
1.4.1.5 Ex: Magnetic moments

R
~ L = 12 R3 r × j(r0 )d3 r0 of classical electrodynamics and an appro-
a. Derive from the expression µ
priate parametrization of the current density j the relation between the magnetic dipole moment
µ
~ due to the orbiting electron and the angular momentum L.
b. The length of the angular momentum vector being given by |L| = ~, calculate the magnetic
moment for an electron and for a proton.
1.4.1.6 Ex: Bohr’s atom

In 1913, Niels Bohr presented his atomic model adapting Rutherford’s model to the quantization
ideas proposed by Max Planck.
a. Impose the quantization rule for the angular momentum (L = n~) of an electron orbiting an
atom of atomic number Z to find an expression for the radii of the allowed orbits.
b. According to Bohr’s model, the transition between different orbits is accompanied by the
emission (or absorption) of a photon. Determine the energy of a photon emitted during a tran-
sition between the first excited state and the ground state of a hydrogen atom.
c. Consider an electron trapped in an infinite one-dimensional box potential of width a. Deter-
mine an expression for the electronic energy levels.
d. What should be the width a of this potential, in terms of the Bohr radius, so ensure that a
photon emitted during a transition between the first excited state and the ground state equals
that obtained in item (b)?
1.4.1.7 Ex: The hydrogen atom

The hydrogen atom can be seen as a point-like proton and an electron distributed over space
with charge density ρ = Ae−2r/aB around the proton that is in the center. Here, A is a constant
and r is the distance from the center.
a. Calculate A considering the fact that the atom is electrically neutral.
b. Calculate the amplitude of the electric field at a radius r = aB .
1.4.2 The discovery of the photon

1.4.2.1 Ex: Resistance of vacuum
p
Show that µ0 /0 has the dimension of a resistance and the value of 376.7 Ω.
1.4.2.2 Ex: Planck’s formula

~ω P∞ −~ω/kB T )ni
Demonstrate the equation ¯ = e~ω/kB T −1
using the closed form for the geometric series, ni =0 (e
dsni
and ds = ni sni −1 , where s =e −~ω/k T
B .
1.4.2.3 Ex: The laws of Planck and Rayleigh-Jeans

Show that Planck’s law reproduces the Rayleigh-Jeans law in the low-frequency limit.
1.4.2.4 Ex: The laws of Wien and Stefan-Boltzmann

According to Wien’s displacement law the maximum emission occurs
−3
R ∞at λmax T = 42.898 ×
10 Km. The law of Stefan-Boltzmann integrate Planck’s formula π 0 u(ν)dν = σT , where
4 /60c2 ~3 . Determine the spectrum of the 3 K background radiation of the universe.
σ = π 2 kB
1.4.2.5 Ex: Photons in a resonator

The power of a light field (λ = 633 nm) in a symmetrical optical resonator (length L = 10 cm)
be P = 1 nW. How many photons are in the resonator mode? How many photons are inside
the resonator on average at ambient temperature, when there is no incident light?
How many photons pass through a transversal cross section of a laser beam having a power of
P = 1 nW?
1.4.2.6 Ex: Atoms in an optical cavity

a. Consider a closed optical cavity at T = 600◦ C. The cavity has the shape the form of a
1-m-long, 3-cm-diameter tube. Calculate the total energy of the black-body radiation inside the
cavity.
b. Inside the cavity there is a gas with strontium atoms (1 fundamental level and 3 degenerate
excited levels, λ = 461 nm). Using the expression (1.92), assuming thermal equilibrium, calculate
the number of excited atoms for a partial pressure of the strontium gas of 10−3 mbar.
c. Calculate the optical density for a laser in resonance with the transition traversing the cavity
along the symmetry axis.
1.4.2.7 Ex: Lambert-Beer’s law

Show the validity of the Eq. (1.130).
1.4.2.8 Ex: Saturation

Assume that a beam of light with intensity I0 enters a cell filled with a gas at the position z0 .
Show from Eq. (1.119) using the generalization Eq. (1.120) that, with high power such that
B21 uω A21 , the beam intensity decreases linearly with the distance such that,
"Z #
ω0 +∆/2
(χ) g2
I (χ) − I0 = − nA21 ~ ωL(ω)dω (z − z0 ) .
g1 ω0 −∆/2
Chapter 2
Foundations of quantum mechanics

In the following, we will introduce step by step the formalism of quantum mechanics by gradually
increasing the degree of abstraction. Applications of the formalism will be shown in consecutive
chapters.
2.1 Basic notions

2.1.1 Dispersion relation and Schrödinger equation
A fundamental problem in physics is the issue of the propagation of physical entities. On one
hand, we have the light, whose propagation in the vacuum is described by the dispersion relation
ω = ck or,
ω 2 − c2 k 2 = 0 . (2.1)
Since light is a wave, in the most general form, assuming the validity of the superposition
R i(k·r−ωt)
principle, it can be described by a wave packet, A(r, t) = e a(k)d3 k. It is easy to verify
that the wave equation,
∂2
A − c2 ∇2 A = 0 , (2.2)
∂t2
reproduces the dispersion relation.
On the other hand, we have slow massive particles possessing kinetic energy,
p2
E= . (2.3)
2m
With the hypothesis of de Broglie that even a massive particle has wave quality, we can try
an ansatz 1 of a wave equation satisfying the dispersion relation (2.3). From Planck’s formula,
E = ~ω, and
R the formula of Louis de Broglie, p = ~k, describing the particle by a wave packet
ψ(r, t) = ei(k·r−ωt) ϕ(k)d3 k not subject to external forces, it is easy to verify that the equation,

∂ ~2 2
i~ ψ = − ∇ ψ, (2.4)
∂t 2m
reproduces the dispersion relation. If the particle is subject to a potential, its total energy is
E = p2 /2m + V (r, t). This dispersion relation corresponds to the famous Schrödinger equation,

∂ ~2 2
i~ ψ = − ∇ + V (r, t) ψ . (2.5)
∂t 2m
1
Trial, working hypothesis.
37
38 CHAPTER 2. FOUNDATIONS OF QUANTUM MECHANICS
2.1.2 Relativistic particle waves

Despite the similarities between light particles and material particles, there are notable differ-
ences: The photon is a relativistic particle with no rest mass. How can we establish a relationship
between such different objects?
To clarify this relationship we now consider particles that are similar to light in the sense
that they have high velocities, that is, relativistic particles. From the relativistic principle of
the equivalence of mass and energy, we obtain for a massive particle E 2 = m2 c4 + c2 p2 or,
m2 c4
ω 2 − c2 k 2 = . (2.6)
~2
This dispersion relation can be obtained from the differential equation,
∂2 2 2 m2 c4
A − c ∇ A = − A, (2.7)
∂t2 ~2
R
inserting, for example, the already proposed wave packet A(r, t) = ei(k·r−ωt) a(k)d3 k, supposed
not to be subject to external forces. The equation (2.7) is a wave equation called Klein-Gordon
equation. For particles without rest mass, as in the case of photons, the equation is reduced to
the wave equation of light (2.2).
Now, making the transition to non-relativistic velocities, v c, we can expand the dispersion
relation,
p
2 v2 ~2 k 2
2 4 2 2 2
E = m c + c m v = mc 1 + 2 + .. or ~ω ' mc2 + . (2.8)
2c 2m
In analogy with the Klein-Gordon equation we can derive the approximate dispersion relation
(2.8) from a wave equation,
∂ 2 ~2 2
i~ A = mc − ∇ A. (2.9)
∂t 2m
2 t/~
With the transformation ψ = e−imc A, we rediscover the Schrödinger equation (2.4),
∂ ~2
i~ ψ=− ∆ψ (2.10)
∂t 2m
as the non-relativistic limit of the Klein-Gordon equation.
It is interesting to note that in all cases discussed, obviously the dispersion relations and the
differential equations can be interconverted by the substitutions,
∂
E −→ i~ and p −→ −i~∇ . (2.11)
∂t
We will discuss this later in the context of Ehrenfest’s theorem in Secs. 2.1.6, 2.1.7, and 2.4.5.
2.1.3 Born’s interpretation

The first part of this script is devoted to individual particles or systems of distinguishable
massive particles, and we will only turn our attention to light and indistinguishable particles
when discussing the (second) quantization of fields.
According to our current conviction, the complete reality (neglecting relativistic effects) on
any system is contained in the Schrödinger equation (2.5). That statement does not make us
2.1. BASIC NOTIONS 39
smarter without having to explaining the meaning of the wavefunction ψ. In an attempt to

marry the concepts of particles and waves, Max Born proposed in 1926 the interpretation of the
quantity Z
|ψ(r, t)|2 d3 r (2.12)
V
as probability of finding the particle inside the volume V .
If |ψ(r, t)|2 has the meaning of a probability density or probability distribution, the square of
the wavefunction must be integrable,
Z
2
kψ(r, t)k ≡ |ψ(r, t)|2 d3 r < ∞ . (2.13)
R3
This allows us to proceed to a normalization of the wave function,

ψ(r, t)
ψ̃(r, t) ≡ qR , (2.14)
2 3
R3 |ψ(r, t)| d r
such that kψ̃(r, t)k = 1.
2.1.4 Continuity equation

In quantum mechanics we associate the wavefunction that describes a quantum system to a
probability wave. As the Schrödinger equation describes a time evolution, in order to be useful,
the wavefunction must allow for probability flows. We define the probability density and the
probability flow by,
ρ(r, t) ≡ ψ ∗ (r, t)ψ(r, t) , (2.15)

~
j(r, t) ≡ [ψ ∗ (r, t)∇ψ(r, t) − ψ(r, t)∇ψ ∗ (r, t)] .
2mi
Starting from the Schrödinger equation we can easily derive the continuity equation (see Exc. 2.6.1.1),
ρ̇(r, t) + ∇ · j(r, t) = 0 , (2.16)
or in the integral form,

Z Z I
d 3 3
− ρd r = ∇ · jd r = j · dS , (2.17)
dt V V ∂V
R
using Gauß’ law. With I ≡ S j · dS, Rthe probability current which flows through the surface S
delimiting the probability charge Q ≡ V ρ(r, t)d3 r, we obtain,
−Q̇ = I . (2.18)
The continuity equation is obviously similar to that of electromagnetism.
2.1.5 Distributions in space and time

So far we only spoke of spatial distributions, ψ(r, t). But we could also consider velocity or
moment distributions. In classical mechanics, a particle has a well-defined position and velocity.
Knowing the position and velocity, Newton’s equations allow predicting its coordinates at future
times. Let us now investigate whether the Schrödinger equation allows this as well.
The more general solution of the Schrödinger equation can be written as a superposition of
plane waves ei(r·k−ωt) with frequencies ω = p2 /2~m and wave vectors k = p/~. Each plane
wave has an individual amplitude ϕ(p), such that,
Z Z
2
1
ψ(r, t) = h3/2 d3 pϕ(p)ei(r·k−ωt) = d3 p h3/2
1
ϕ(p)ei(r·p/~−p t/2m~) , (2.19)
with h ≡ 2π~. At time t = 0, this expansion is nothing more than a Fourier transform,
Z
ψ(r, 0) = h3/2 d3 pϕ(p)eir·k ,
1
(2.20)
that we can reverse, Z

ϕ(p) = 1
h3/2
d3 rψ(r, 0)e−ir·k . (2.21)
In the absence of forces the momentum distribution becomes stationary. We can now use
the momentum distribution ϕ(p) as coefficients of the expansion of the temporal wavefunction
ψ(r, t), as shown above. Thus, the expansion represents a general solution of the time-dependent
Schrödinger equation. The magnitude |ϕ(p)|2 is the probability density in momentum space.
Example 2 (Normalization of the wave function in momentum space): It is easy
to show that the probability density in momentum space is also normalized:
Z Z Z Z
0
∗
2 3 1
|ϕ(p)| d p = h3 3 3
d p d rψ (r)e ir·k
d3 r0 ψ(r0 )e−ir ·k
Z Z Z
0
= d3 r d3 r0 ψ ∗ (r)ψ(r0 ) (2π)
1
3 d3 keik·(r−r )
Z Z Z
3 0 ∗ 0 3 0
= d r d r ψ (r)ψ(r )δ (r − r ) = |ψ(r)|2 d3 r = 1 ,
3
knowing that the Fourier transform of a plane wave is nothing more than the Dirac distri-
bution.
Since the probability distributions |ψ(r)|2 and |ϕ(p)|2 are interconnected by Fourier trans-
form, we already know that we can not localize 2 both simultaneously. If one is well localized,
the other is necessarily delocalized. Do the Exc. 2.6.1.2.
2.1.6 Eigenvalues
We have already seen that the position and momentum distributions of a particle are spread.
We calculate the mean values of these distributions, denoted by hri e hpi, as first moments of
the respective distributions:
Z Z
3 2
hri = d r|ψ(r, t)| r and hpi = d3 p|ϕ(p, t)|2 p . (2.22)
Using the expansions (2.19) and (2.20), we can calculate,

Z Z Z
∗
hpi = ϕ (p)pϕ(p)d p = 3 1
h3/2
ψ ∗ (r)eik·r d3 rpϕ(p)d3 p
Z Z
∗
= h3/2 ψ (r) ϕ(p)peik·r d3 pd3 r
1
Z Z Z
= h3/2 ψ (r) i ∇ ϕ(p)e d pd r = ψ ∗ (r) ~i ∇ψ(r)d3 r .
1 ∗ ~ ik·r 3 3
2
Localize: Restrict the distribution volume indefinitely.
2.1. BASIC NOTIONS 41
This calculation shows that the expectation value, called eigenvalue, of the momentum can be
expressed through an operator p̂ ≡ (~/i)∇ acting on the wavefunction 3,4 .
More generally, we can compute the eigenvalue of a function in r e p via,
Z
hf (r, p)i = d3 rψ ∗ (r)f (r, p̂)ψ(r) . (2.23)
However, it is important to note that the operators r̂ and p̂ do not necessarily commute.
Example 3 (Non-commutation of space and momentum): Considering a one-dimensional

motion, we verify,
~ d ~ ~ d ~ d
p̂x xψ = xψ = ψ + x ψ 6= x ψ = xp̂x ψ .
i dx i i dx i dx
2.1.7 Temporal evolution of eigenvalues

We now consider the temporal R evolutionHof the position
R of a particle.
R We will use in the following
the partial integration rule V ψ∇ξ = ∂V ψξ − V ∇ψξ = − V (∇ψ)ξ, assuming that at least
one of the functions, ψ or ξ, disappears at the edge of the volume, which can be guaranteed by
choosing the volume large enough. To begin with, we will concentrate on the x-component of
the position, the time derivative of which is computed using the continuity equation (2.16),
Z Z Z Z
d 3 d
R 0
hxi = d r |ψ| x = − d r x∇ · j = − dS · j x + d r j · ∇x = d3 r jx , (2.24)
2 3 3
dt dt
Generalizing to three dimensions, we can write,
Z Z
d ~
hmri = m d3 r j = m d3 r [ψ ∗ ∇ψ − ψ∇ψ ∗ ] (2.25)
dt 2mi
Z Z
= 12 d3 r[ψ ∗ p̂ψ + ψp̂ψ ∗ ] = d3 rψ ∗ p̂ψ = hp̂i ,
since the eigenvalue of p̂ is a real quantity.

Now, we define the abbreviation:
~2 2
Ĥ ≡ − ∇ + V (r, t) , (2.26)
2m
called the Hamilton operator or Hamiltonian and we calculate the second derivative of the
position using the Schrödinger equation (2.5),
Z h ∗ i Z
d
hp̂i = d3 r i~1
Ĥψ p̂ψ + ψ ∗ p̂ i~
1
Ĥψ = ~i d3 r ψ ∗ (Ĥ p̂ − p̂Ĥ)ψ = ~i h[Ĥ, p̂]i , (2.27)
dt
introducing the commutator [â, b̂] ≡ âb̂ − b̂â as an abbreviation. After that,
Z h i Z
~ ~
i i i
~ h[Ĥ, p̂]i = ~ h[V̂ , p̂]i = ~ d rψ V̂ i ∇ψ − i ∇(V ψ) = − d3 rψ ∗ ψ∇V = hF̂i .
3 ∗
(2.28)
3
From now on, the hat over a physical magnitude will denote quantum operators.
4
We note here that the rules hψ|x̂|ψi ↔ hφ|− ~i ∇p |φi and hψ| ~i ∇r |ψi ↔ hφ|p̂|φi from the Fourier transformation
are useful
2 for numerical simulations of the Schrödinger equation: Instead of calculating the spatial derivative
~
i
∇ of the wavefunction, one makes a Fast Fourier Transform (FFT) to momentum space, multiplies with p,
and transforms back.
In summary, we found a law,

d2
hF̂i = hmr̂i , (2.29)
dt2
much like the Newton law, but instead of applying to localized particles, the law applies to the
eigenvalues of probability distributions. Similar laws can be derived for angular momentum and
energy conservation.
The observation made by Paul Ehrenfest, that in quantum mechanics the mean values follow
the same laws of classical mechanics, is called Ehrenfest’s theorem.
2.2 Postulates of quantum mechanics

In this section we will introduce the fundamentals and main methods of quantum mechanics. We
will learn what are observables and get to know the postulates which establish the foundation
of quantum mechanics, as well as Heisenberg’s famous principle of uncertainty.
2.2.1 Superposition principle (1. postulate)

A physical system can be found in several states. For example, a particle may be at rest or in
motion, an atom may be excited or deexcited. In quantum mechanics, every possible state is
described by a wavefunction ψ. Wavefunctions can be functions of various types of coordinates,
for example, of position ψ = ψ(r), of momentum ψ = ψ(p), or of energy ψ = ψ(E). The choice
of the coordinates is called representation.
One peculiarity of quantum systems is that they may be in a superposition of states. That
is, if ψ1 , ψ2 , ..., ψk are possible states with amplitudes ck , automatically the functions,
X Z
ψ= ck ψk or ψ = dkc(k)φ(k) (2.30)
k
are possible states as well. This is called superposition principle, and means, for example, that
a particle may be simultaneously in several places or that an atom may be at the same time
excited and deexcited.
There are systems that can only exist in a restricted number of states, such as the two-level
atom. Others may exist in an infinite number of states or even in a continuous distribution of
states.
2.2.2 Interpretation of the wave function (2. postulate)

A state function (or wavefunction) characterizes a system of which we may calculate various
properties. The function can adopt complex values devoid of immediate physical interpretation.
In fact, the wavefunction is above all a mathematical construct. On the other hand, the norm
|ψ|2 has the meaning of a probability of the system to be in the state ψ. This is the famous
interpretation of Max Born of the wave function (see Sec. 2.1.3).
If ψk with k = 1, 2, . . . are all possible states of a system, the interpretation as a probability
requires, X
|ψk |2 = 1 . (2.31)
k
Analogically, for a continuous distribution, for example, in spatial representation,
Z ∞
|ψ(x)|2 dx = 1 . (2.32)
−∞
2.2. POSTULATES OF QUANTUM MECHANICS 43
That is, the probability needs normalization.
2.2.3 Dirac bra-ket notation and vector representation

In order to distinguish more easily the amplitudes (which are complex numbers) and the wave-
functions we will now use the Bra-Ket notation introduced by Paul Dirac. The functions are
represented by kets, X
|ψi = ck |ki . (2.33)
k
The complex transpositions of these states are represented by bras,

X
hψ| = |ψi† = c∗k hk| . (2.34)
k
But the notation has other advantages. For example, let us suppose that we know the three
possible states of a system, |1i, |2i, and |3i, which are linearly independent. Then we can define
the states as vectors:
     
1 0 0
|1i = 0 , |2i = 1 , |3i = 0 . (2.35)
0 0 1
These three states can be interpreted as the basis of a vector space representing the system.
Now, each wavefunction can be expanded on this basis and expressed by a vector. An arbitrary
ket state of this system will then be,
 
c1
|ψi = c2  . (2.36)
c3
The corresponding bra state will be,

hψ| = c∗1 c∗2 c∗3 . (2.37)
Now we can easily calculate the probability for a system to be in a state |ψi,
 
c1
||ψi|2 = hψ|ψi = c∗1 c∗2 c∗3 · c2  = |c1 |2 + |c2 |2 + |c3 |2 . (2.38)
c3
2.2.4 Observables (3. postulate)

The only way to get information about a system is to measure the values of characteristic
quantities of the system, e.g. energy or linear momentum. In classical mechanics we have learned
that a system can be completely characterized by a set of measurable physical quantities. For
example, the motion of a rigid body of mass m and inertial moment I is defined by its position
r, its moment p, and its angular momentum L. In quantum mechanics we describe observable
physical quantities by operators acting on the Hilbert space of wavefunctions, |ψi 7→ p̂|ψi, where
p̂ would be the operator of the linear momentum. To better distinguish the observables, we
decorate their symbols with a hat. We will see more ahead (see Sec. 2.3.5) that every quantum
system is completely described by a complete set of observables.
To find the current values aψ of any observable Â in a specific situation given by a wave
function ψ, we need to solve an equation of eigenvalues,
Â|ψi = aψ |ψi . (2.39)
We can rewrite the equation as aψ = hψ|Â|ψi. The values an are real numbers, if the observable
is a Hermitian operator, that is,
Â = Â† =⇒ aψ = a∗ψ . (2.40)
We leave proof of this for the Exc. 2.6.2.1.

Thus, we postulate the substitution of the dynamic variables characterizing a classical sys-
tem by abstract objects called operators. These operators can be understood as mathematical
prescriptions, e.g., differential operators acting on a state of the system. The expectation value
of any operator Â characterizing a system in a state |ψi is aψ ≡ hÂiψ ≡ hψ|Â|ψi/hψ|ψi. Such
operators are specific for a system, but independent of its state. The dynamical variables for a
specific state are obtained as eigenvalues of the respective variable in that specific state. The
temporal evolution of the operators or of the states is governed by equations of motion (see
Sec. 2.4) 5 .
2.2.5 Representation of operators as matrices

In the same way as we already represented wavefunctions by vectors, we can also represent
operators by matrices,
   
X : :
Â ≡ |iiaij hj| = .. aij .. = .. hj|Â|ii .. . (2.41)
i,j : :
To extract components from a matrix we do, hi|Â|ji, for example,

 
1
h1|Â|1i = 1 0 .. · Â · 0 = a11 .
 (2.42)
:
Projectors are particular operators defined by,

 
0 : 0
P̂k ≡ |kihk| = .. 1 .. . (2.43)
0 : 0
The eigenvalue of a projector, hP̂k i = hψ|P̂k |ψi = |hk|ψi|2 , is nothing more than the probability
of finding a system, whose general state is |ψi, in the particular state, since expanding as done
in (2.33), we have,
X
hP̂k i = c∗m cn hm|kihk|ni = |ck |2 . (2.44)
m,n
5
Note that there are theoretical attempts to generalize the concept of observables to non-Hermitian operators
[24, 25] only displaying PT -symmetry.
2.2. POSTULATES OF QUANTUM MECHANICS 45
Using the matrix formalism we can define other interesting operators and verify their prop-
erties,

1 0 0 0
|1ih1| = , |2ih2| = ,
0 0 0 1 . (2.45)
0 1 0 0
|1ih2| = = σ− , |2ih1| = = σ+
0 0 1 0
The rising and lowering operators, σ ± , are also called Pauli spin matrices, because they were
introduced by Wolfgang Pauli. The vector,
 + 
σ + σ−
~σ ≡ i(σ − − σ + ) (2.46)
+ −
[σ , σ ]
is called Bloch vector 6,7 . The eigenvalue of the Bloch vector has a fixed length (see Exc. 2.6.2.2).
The representation of physical quantities by matrices allows the description of quantum
superposition states. The purpose of Exc. 2.6.2.3 is to illustrate this at the example of a particle
passing through a double slit.
2.2.6 Correspondence principle (4. postulate)

Operators do not necessarily commute. We have already seen in Sec. 2.1.6, that in one dimen-
sion the position and the momentum operators do not commute. We can generalize to three
dimensions via,
[p̂j , x̂k ] = −i~δjk and [p̂j , p̂k ] = 0 = [x̂j , x̂k ] , (2.47)
which is easily verified by replacing the operators with x̂k = xk and p̂k = ~i ∇ and allowing the
commutators to act on a wavefunction ψ(x).
Conversely, quantum mechanics follows from classical mechanics with the prescription 8 ,
A(qk , pk , t) −→ A(q̂k , p̂k , t) = Â. Letting the smallest amount of energy possible go to zero,
~ −→ 0, the commutator disappears, the energy spectrum becomes continuous, and we recover
classical mechanics.
2.2.7 Schrödinger equation and quantum measurements (5. postulate)

The time evolution is given by the Schrödinger equation,
∂
i~ |ψi = Ĥ|ψi . (2.48)
∂t
A closed system, disconnected from the rest of the world (we will now call the rest of the
world reservoir) is not subject to dissipation, i.e., it does not lose energy to the reservoir. Such
a system is always described by a hermitian Hamiltonian. Unfortunately, this system also does
not allow information leakage, that is, we can not measure the system. This is reflected in
the fact that the Schrödinger equation does not allow to describe the process of a quantum
measurement. This is because before the measurement, the system can be in several states or
6
The Bloch vector is widely used in describing the interaction of a two-level system with a light field.
7
Schrödinger invented the wave mechanics when he derived his wave equation from the dispersion relation
for massive particles. Heisenberg invented a mechanics (detailed in later sections), which he called mechanics of
matrices. Later, he showed the formal equivalence of both theories.
8
Considering the Weyl order.
even in a superposition of states, while after the measurement we know exactly the state. This
amounts to a reduction of entropy, which is not allowed in a closed system.
The famous postulate of state reduction or projection of the wavefunction formulated by John
von Neumann describes the quantum measurement process as a sequence of two distinct steps 9 .
In a first step, the measuring apparatus projects the measured operator Â on an eigenvector
basis. That is, if the measurement is compatible with the operator 10 , we obtain a distribution
of probability amplitudes of the results,
X X X
Â y hÂi = hψ|Â|ψi = hψ|Â| ck |ki = ak ck hψ|ki = ak |ck |2 , (2.49)
k k k
P
with hψ|ψi = k |ak |2 = 1. Therefore, we can understand |hk|ψi|2 as the probability of the
system to be in the eigenstate |ki 11 .
In a second step, the observing scientist will read the measuring device and note the result,
which will necessarily be one of the possible ak ,
hÂi y ak . (2.50)
If the state is stationary, it will never change any more. That is, each subsequent measurement
will yield the same result. The Exc. 2.6.2.4 illustrates the process of quantum measurement at
the example of a measurement of the excitation energy of a two-level atom.
2.2.8 Stationary Schrödinger equation

The general form of the Schrödinger equation in one dimension is,
∂
ĤΨ(t, x) = i~ Ψ(t, x) , (2.51)
∂t
p̂ 2
∂
with Ĥ ≡ 2m + V (x, t) and p̂ ≡ −i~ ∂x . If the potential is independent of time, V (x, t) = V (x),
we can do the following ansatz, Ψ(x, t) ≡ ψ(x)f (t). Insertion into the Schrödinger equation
yields,
1 ~2 d2 i~ d
− 2
+ V (x) ψ(x) = f (t) = const. ≡ E . (2.52)
ψ(x) 2m dx f (t) dt
The solution of the right-hand side of the equation is i~(ln f − ln f0 ) = E(t − t0 ). Hence,
f (t) = f (0)e−iE(t−t0 )/~ . (2.53)
Obviously, |Ψ(x, t)|2 = |ψ(x)|2 .

Now, we can see that the stationary Schrödinger equation,
Ĥψ(x) = Eψ(x) , (2.54)
is nothing more than an eigenvalue equation. This means that the Schrödinger wave mechanics
is equivalent to the mechanics of the Heisenberg matrices. The Excs. 2.6.2.5 and 2.6.2.6 are first
simple calculations of the eigenvalues and eigenvectors of a two-level system.
9
For simplicity, we only consider pure state, here.
10
To understand the meaning of compatible, we must establish a more complete theory of measurement including
the reservoir in the quantum description.
11 P
Alternatively, Â −→ k |kihk|Â|kihk|.
2.3. ABSTRACT FORMALISM OF QUANTUM MECHANICS 47
2.3 Abstract formalism of quantum mechanics

The formal development of quantum mechanics will be the subject of this section. We will learn
how to find a complete set of observables characterizing a system, discuss the role of symmetries
in quantum mechanics and show how to switch between several representations of the same
system.
2.3.1 Lie algebra

The quantum mechanical operators form a Lie algebra L2 . This means that L2 is at the same
time a complex and linear vector space with respect to addition and scalar multiplication and
a non-commutative ring with scalar internal product. In particular, L2 is unitary, normalized,
and complete and acts on a Hilbert space of quantum states,
(Â + B̂)|ψi = Â|ψi + B̂|ψi, (2.55)

(αÂ)|ψi = α(Â|ψi) ,
(ÂB̂)|ψi = Â(B̂|ψi) .
The properties of the Hilbert space are,
Â|ψ + ϕi = Â|ψi + Â|ϕi , (2.56)

Â|aψi = aÂ|ψi .
For a Hermitian operator, Â = Â† , we have hψ|Â|ψi = hÂψ|ψi or hÂi ≡ hψ|Â|ψi = hÂi∗ , using
the Dirac bra-ket notation,
hψ|† ≡ |ψi . (2.57)
There are identity and nullity operators,
1̂|ψi = |ψi and 0̂|ψi = 0 . (2.58)
We define the (anti-)commutator as,
[Â, B̂]∓ ≡ ÂB̂ ± B̂ Â , (2.59)
which can be 6= 0. The sum of two hermitian operators is hermitian, but the product is not,
since,
(Â + B̂)† = Â† + B̂ † = Â + B̂ but (ÂB̂)† = B̂ † Â† = B̂ Â 6= ÂB̂ . (2.60)
On the other hand, the following relations of hermitian operators are always hermitian,
ÂB̂ + B̂ Â and i(ÂB̂ − B̂ Â) . (2.61)
We define the scalar product as,

hψ|ϕi . (2.62)
Two states are called orthogonal, if hψ|ϕi = 0. The norm is written as,
|ψ|2 = hψ|ψi1/2 , (2.63)
the deviation is, q

∆A ≡ hÂ2 i − hÂi2 . (2.64)
A unitaryoperator is defined by Â−1 = Â† .
2.3.2 Complete bases

If it is impossible to find a set of amplitudes cn ,
X
@{cn } such that cn |ni = 0 , (2.65)
n
the functions are called linearly independent. A set of linearly independent functions may form
a basis. The space opened by a set of linearly independent functions is called Hilbert space.
An operator Â is completely characterized by its eigenvalues and eigenfunctions. If a set
of eigenfunctions |ni is complete, every allowed state of the system can be expanded in these
eigenfunctions, X
|ψi = cn |ni and Â|ni = an |ni . (2.66)
n
To calculate properties of a specific system, we often want to find a matrix representation

for the operator Â. For this, we solve the stationary Schrödinger equation, that is, we calculate
the eigenvalues and eigenvectors. When all eigenvalues are different, an 6= am , we know that the
corresponding eigenvectors are orthogonal, hn|mi = 0,
Â|ni = an |ni , Â|mi = an |mi , ∀{n, m} an 6= am (2.67)

=⇒ ∀{n, m} hn|mi = δm,n .
Exc. 2.6.3.1 asks for demonstrating this.

Frequently, for example, in the case of a particle confined to a potential, there exist discrete
eigenvalues (for E < 0) simultaneously with continuous eigenvalues (for E > 0). Assuming
hm|m0 i = δm,m0 , hm|ki = 0 and hk|k0 i = δ (3) (k − k0 ), with a complete base,
X Z
|mihm| + d3 k|kihk| = 1̂ , (2.68)
m
an arbitrary vector can be expanded on an orthogonal basis,

X Z
|ψi = |mihm|ψi + d3 k |kihk|ψi . (2.69)
m
This also applies to observables,

X Z
Â = |mihm|Â|nihn| + d3 kd3 l |kihk|Â|lihl| , (2.70)
m,n
and functions of observables,

X Z
f (Â) = |mif (hm|Â|ni)hn| + d3 kd3 l |kif (hk|Â|li)hl| . (2.71)
m,n
2.3.3 Degeneracy
The eigenvectors form a natural basis for the Hilbert space. However, a problem arises in the case
of degeneracy, that is, when some eigenvalues are equal, an = am . In this case, the eigenvectors
that correspond to degenerate eigenvalues are not completely defined, and we have to construct
a basis verifying that all constructed eigenvectors are orthogonal. For this, there exists the
method of orthogonalization by Schmidt, which works like this: We assume that we have already
solved the eigenvalue equation, that we found a degenerate eigenvalue, Â|ak i = a|ak i for every
k = 1, .., gk , where gk is the degree of degeneracy, and that we also found a complete basis of
eigenvalues |am i, but which is not orthogonal, that is, ∃{m, n} with han |am i 6= 0. The task is
to build another basis |bm i satisfying hbn |bm i = δn,m .
The first vector of the orthogonal base can be chosen freely, e.g.,
|b1 i ≡ |a1 i . (2.72)
Since de basis {|ak i} is assumed to be complete, the second vector is necessarily a linear combina-
tion of vectors |ak i, that is, |b2 i = |a2 i+λ|b1 i. With the condition hb1 |b2 i = 0 = hb1 |a2 i+λhb1 |b1 i
we can determine the parameter λ, and obtain for the second vector,
hb1 |a2 i
|b2 i ≡ |a2 i − |b1 i . (2.73)
hb1 |b1 i
In the same way, we can derive for a third vector, |b3 i = |a3 i + µ|b1 i + ν|b2 i, the conditions,
hb1 |b3 i = 0 = hb1 |a3 i + µhb1 |b1 i and hb2 |b3 i = 0 = hb2 |a3 i + νhb2 |b2 i, and obtain,
hb1 |a3 i hb2 |a3 i
|b3 i ≡ |a3 i − |b1 i − |b2 i . (2.74)
hb1 |b1 i hb2 |b2 i
An overall way of writing this down is,

|b1 ihb1 | |b2 ihb2 | |bk−1 ihbk−1 |
|bk i ≡ 1 − − − ... − |ak i . (2.75)
hb1 |b1 i hb2 |b2 i hbk−1 |bk−1 i
In the Exc. 2.6.3.2 we practice the orthogonalization of a set of three linearly independent
but non-orthogonal vectors, and in the Exc. 2.6.3.3 we find an orthogonal basis for a partially
degenerate three-level system.
2.3.4 Bases as unitary operators

One way to formulate the eigenvalue problem is as follows: Let |ni be an orthonormal basis with
the respective eigenvalues an of an operator Â:
Â|ni = an |ni with hn|mi = δmn . (2.76)
We construct the matrices,
 
a1 0 · · ·
 
U ≡ |1i |2i · · · and Ê ≡  0 a2  . (2.77)
.. ..
. .
With the definition of U † we have,

   
h1| h1|1i h1|2i · · ·
h2|  
†
U =  and U † U = h2|1i h2|2i · · · = 1̂ . (2.78)
.. .. .. ..
. . . .
Therefore,
U † U = 1̂ =⇒ U † U U −1 = 1̂U −1 =⇒ U † = U −1 (2.79)
† † −1 −1 †
U U = 1̂ =⇒ UU UU = U 1̂U =⇒ U U = 1̂ .
Also,
Â|ni = Ê|ni and ÂU = U Ê . (2.80)
That is, by knowing the unitary matrix (or transformation matrix) U , we can solve the eigenvalue
problem simply by Ê = U −1 ÂU .
Note, that this does not apply to a non-orthonormal basis. In this case, we need to do a
Schmidt orthogonalization and use the condition det Û = 1. We apply the technique detailed in
this section to solve Excs. 2.6.3.4 and 2.6.3.5.
2.3.5 Complete set of commuting operators

Even for simple systems, we can ask various types of questions (measurements). Considering, for
example, a particle flying freely in space, we can gather its position or its velocity. Let a be the
result of a measurement of the observable Â, that is, a = hψa |Â|ψa i. Due to the measurement
we know that the system is in the state |ψa i. Immediately after this first measurement we
perform another measurement of another observable B̂ giving hψa |B̂|ψa i. The result of this
measurement can only yield an eigenstate, b = hψa |B̂|ψa i, if the operators commute, [Â, B̂] = 0.
That is, if two operators Â and B̂ commute, and if |ψi is an eigenvector of Â, then B̂|ψi is also
an eigenvector of Â with the same eigenvalue:
[Â, B̂] = 0 , a = hψ|Â|ψi (2.81)

=⇒ Â(B̂|ψi) = a(B̂|ψi) and hψ|B̂|ψi ∈ R .
In addition, we observe that if two operators commute, the orthonormal basis constructed
for one of the operators is also orthonormal for the other. That is, if two operators Â and B̂
commute and if |ψ1 i and |ψ2 i are two eigenvectors of Â with different eigenvalues, then the
matrix element hψ1 |B̂|ψ2 i is equal to zero:
[Â, B̂] = 0 , a1 = hψ1 |Â|ψ1 i =

6 hψ2 |Â|ψ2 i = a2 (2.82)
=⇒ hψ1 |B̂|ψ2 i = 0 .
Finally, we affirm that, if two operators Â and B̂ commute, we can construct an orthonormal
basis {|ψa,b i} with common eigenvectors of Â and B̂:
[Â, B̂] = 0 (2.83)

=⇒ ∃ {|ψa,b i} tal que Â|ψa,b i = a|ψa,b i and B̂|ψa,b i = b|ψa,b i .
The statements (2.81) to (2.83) are verified in Exc. 2.6.3.6.
The fact that commuting operators have a common system of eigenvectors authorizing sharp
eigenvalues can be used to construct and characterize a state.
Example 4 (Measuring momenta in orthogonal directions): For example, the obvious

solutions of the eigenvalue equations,
d d
p̂x |ψpx i = ~
i |ψp i = px |ψpx i and p̂y |ψpy i = ~
i |ψp i = py |ψpy i
dx x dy y
are the plane waves eipx x/~ and eipy y/~ . Therefore, the total state of the particle can be
described by,
|ψpx ,py ,pz i = |ψpx i|ψpy i = e(i/~)(px x+py y) f (z) .
However, these eigenfunctions are infinitely degenerate, since the linear momentum in z-
direction is not specified. A third operator p̂z |ψi = pz |ψi commutes with the others,
[p̂k , p̂m ] = 0 .
Hence,
|ψpx ,py ,pz i = e(i/~)(px x+py y+pz z) ,
is a possible state of the system.
∂2
On the other hand, choosing p̂2z = −~2 ∂z 2
2 as the third operator, giving the eigenvalues pz ,
the state would have been,
|ψpx ,py ,p2z i = e(i/~)(px x+py y) cos p~z z or |ψpx ,py ,p2z i = e(i/~)(px x+py y) sin p~z z . (2.84)
Therefore, there are two solutions with the same eigenvalues, px , py , p2z . To lift this degener-
acy, we need to introduce yet another observable. This observable can be, for example, the
parity P̂ , that is, the behavior of the wave function upon mirroring z −→ −z in the x-y plane.
The fact that the set of operators px , py , pz on one hand and px , py , p2z , P̂ on the other are
equivalent, shows that the required number of observables for a complete characterization
depends on their judicious choice.
Also, the number needed for a complete set of commuting operators (CSCO) depends on the
number of degrees of freedom and the symmetry of the system. In the case of the free particle
in one dimension it is enough to consider one observable only, for example, x̂ or p̂. In three
dimensions, we already need at least three commuting observables. In Exc. 2.6.3.7 we will try
to find a CSCO for a matrix with partially degenerate eigenvalues.
2.3.6 Uncertainty relation

We have already learned that observables that do not commute can not be measured with
arbitrary precision. This principle can be quantified as follows: If Â and B̂ are two observables,
then,
∆Â∆B̂ ≥ 12 |h[Â, B̂]i| . (2.85)
This is Heisenberg’s famous uncertainty principle. For example, [p̂, x̂] = −i~, and hence,
∆p∆x ≥ ~/2. We will see later (see Sec. 4.3.1), that [ˆlx , ˆly ] = i~ˆlz such that ∆lx ∆ly ≥ ~hlz i/2.
More difficult to show, since time has no simple quantum operator, is ∆E∆t ≥ ~/2. In the
Exc. 2.6.3.8 we will show the Schwartz inequality, and in the Exc. 2.6.3.9 we ask for a formal
derivation of Heisenberg’s uncertainty principle.
2.3.7 Representations
2.3.7.1 Spatial representation
A Hilbert space can be discrete or, as in the case of the momentum of a free particle, continuous.
In this latter case, the eigenvalues are continuously distributed, since the equation,
−i~∇r ψ(r) = pψ(r) , (2.86)
has solutions for each value of E. The eigenfunctions are ψ(r) = aeip·r/~ . Eq. (2.86) clearly has
the form of an eigenvalue equation, for which we have already introduced the Heisenberg matrix
formalism. The question now is how these descriptions combine.
Observables that do not commute correspond to expansions on different bases and generate
alternative representations. For example, we can represent quantum mechanics in position space
or linear momentum space. If |ri is a basis of the space of the particles’ state,
Z
0 3 0
r̂|ri = r|ri , hr |ri = δ (r − r) , |rihr|d3 r = 1̂ , (2.87)
R3
we can expand a state vector on a position basis as,

Z
|ψ(t)i = |riψ(t, r)d3 r . (2.88)
R3
The quantities hr|ψ(t)i = ψ(t, r) Schrödinger wave functions. We can also say that the wave-
functions are the coordinates of the state in the particular base |ri. Consequently,
hr|r̂|r0 i = rδ 3 (r − r0 ) (2.89)
0 3 0
hr|f (r̂)|r i = f (r)δ (r − r ) . (2.90)
It is also true that,

Z
hr|Â|ψ(t)i = A(r, r0 )ψ(t, r0 )d3 r0 , (2.91)
R3
where the quantity A(r, r0 ) ≡ hr|Â|r0 i is called kernel of the operator. The transition from
Heisenberg’s abstract mechanics to Schrödinger’s wave mechanics is done by the substitutions
|ψ(t)i → ψ(t, r) and Â → A(r, r0 ).
2.3.7.2 Momentum representation

The uncertainty relation is symmetric in r̂ and p̂. Nothing prevents us from choosing as a basis,
Z
0 3 0
p̂|pi = p|pi , hp |pi = δ (p − p) , |pihp|d3 p = 1̂ , (2.92)
R3
with the wavefunctions, Z

|ψ(t)i = |piϕ(p, t)d3 p , (2.93)
R3
where hp|ψ(t)i = ϕ(t, p). The formulas are analogous to the ones in the spatial representation.
In particular, in the momentum representation the position operator is r = i~∇p .
The representations follow from one another by Fourier transformation. Since −i~∇r hr|pi =
phr|pi, we know,
1
hr|pi = ~3/2
exp( ~i r · p) , (2.94)
where the prefactor ~−3/2 is introduced to take account of the unit of the states 12 . ψ and ϕ are
different representations of the same quantum state related by,
Z Z
3
hr|ψ(t)i = 1
hr|pihp|ψ(t)id p = h3/2 eir·p/~ ϕ(p, t)d3 p = ψ(r, t) (2.95)
ZR3
Z R3
hp|ψ(t)i = hp|rihr|ψ(t)id3 r = h3/2

1
e−ir·p/~ ψ(r, t)d3 r = ϕ(p, t) .
R3 R3
12
Note that the units of the wavefunctions are defined by normalization: hr0 |ri = δ 3 (r − r0 ). Introducing the
parenthesis [...] to extract the unit of a physical quantity, we find, [|ri] = [ψ(r)] = [r−3/2 ] e [|pi] = [ϕ(p)] = [p−3/2 ].
We do not assign a unit to the abstract state |ψi, that is, [|ψi] = 1.
Normalization ensures that ψ = F −1 Fψ with the relation,

Z t
1
δ(x) = lim 2π eikx dk . (2.96)
t→∞ −t
Using the wavevector ~k = p we can also write,

Z Z
ψ(r) = (2π)13/2 eir·k ϕ̃(k)d3 k and ϕ̃(k) = 1
(2π)3/2
e−ir·k ψ(r)d3 r , (2.97)
R3 R3
defining the function ϕ̃(k) ≡ ~3/2 ϕ(p). Applying the Fourier transform to functions of operator
we can calculate,
Z Z
hr|G(p̂)|r i = d phr|G(p̂)|pihp|r i = d3 pG(p)hr|pihp|r0 i
0 3 0
(2.98)
Z
0
1
= ~3/2 d3 pG(p)eik·(r−r ) = ~13 (FG)(r − r0 ) .
In Exc. 2.6.3.10 we will show hr|p̂|ψi = (~/i)∇r hr|ψi, thus justifying that we can understand
an operator as a rule to determine what happens to a function. For example, the rule p̂x asks
for a derivation of the wavefunction by x.
2.3.8 Setting up a Hilbert space with several degrees of freedom

All systems analyzed up to this point were characterized by a single degree of freedom (e.g.,
energy, momentum, or angular momentum), which could have a continuous or discrete spectrum.
Even when we treated systems exhibiting various degrees of freedom (motion of a particle in 3D
space, electron orbitals in the hydrogen atom), we always found a way to separate the degrees of
freedom into orthogonal Hilbert spaces, which allowed us to treat the dynamics of the degrees
of freedom separately. In this chapter, we will establish the theoretical foundations allowing us
to analyze systems, where degrees of freedom can not be separated because they are entangled
or interact. In particular, we will consider the system of two spins and the coupling of angular
momenta in general.
2.3.8.1 Projection and internal sum

A projector is an operator which reduces the domain of an operator, originally acting on a Hilbert
space H to the subspace defined by the projector. We consider an operator Â with the matrix
representation,  
X :
Â ≡ |iiAij hj| = .. Aij .. , (2.99)
i,j
|ii,|ji∈H
:
acting on wavefunctions |ψi ∈ H which can be expanded on a basis |ii of H. Now, we consider
a subspace R ⊂ H defined by the base |ki. Then the projector P̂R can be represented by,
 
0  0  0
 1 
X  
P̂R ≡ |kihk| = 0  ..  0

 . (2.100)
k  1 
|ki∈R
0 0 0
Applied to the operator Â,

 
0 0 
 0
 : 
X  
ÂR ≡ P̂R Â = P̂R ÂR = 
|kiAkl hl| = 0 .. Akl .. 0
 . (2.101)
k,l  : 
0 0 0
Applied to a state |ii,

 
0
 : 
X  
|ψiR = P̂R |ψi = P̂R |ψiR = ck |ki = 

ck  .
 (2.102)
k  : 
0
We study an example in the 2.6.3.11

Consequently, we can understand the Hilbert space as the sum of its subspaces,
M M
Â = ÂR and P̂R = I . (2.103)
R R
The dimensions of the subspaces are additive,

X
dim Â = dim ÂR .
R
Example 5 (Projection for a three-level atom): The Hamiltonian of a three-level atom

with excitation of two transitions is given by,
 
ω1 Ω12 0
Ĥ = Ω12 ω2 Ω23  .
0 Ω23 ω3
The projector,
 
1 0 0
P̂ = 0 1 0
0 0 0
reduces the Hamiltonian to a two-level transition,

 
ω1 Ω12 0
ĤR = Ω12 ω2 0 .
0 0 0
Obviously, the concatenation (2.103) only serves to increase the Hilbert space of a given
degree of freedom described by a given observable, e.g., when we add one more level of energy
to the spectrum of an atom described by a Hamiltonian. If, in contrast, we want to add another
degree of freedom, we need the external sum or external product discussed below.
2.3.8.2 External product

We have previously worked with systems exhibiting more than one degree of freedom and there-
fore having to be characterized by more than one observable with its spectrum of eigenstates.
One example are the electronic orbitals of the hydrogen atom |n`mi, which need three quantum
numbers to be labeled unambiguously. Obviously, each quantum number increases the dimen-
sionality of the Hilbert space. Another example is the system |αβi of two particles with spin 12 ,
each spin being defined on its respective space,

α1 β1
|αi = = (αi )i ∈ HA and |βi = = (βk )k ∈ HB . (2.104)
α2 β2
The combined state is,
|αβi ∈ HA ⊗ HB with dim HA ⊗ HB = dim HA dim HB . (2.105)
The symbol ⊗ denotes the external tensorial product of two vectors (states). Now, in order to
represent the multidimensional space HA ⊗ HB by a matrix, we use the fact that it is isomorphic
to the space HI ⊗ HA⊗B , that is, we proceed to a reorganization of the quantum numbers
identifying,
 
α 1 β 1
α1 |βi α1 β2 
|γi ≡ |αi|βi = |αi ⊗ |βi = |αβi = = 
α2 β1  = (γm )m ∈ HA ⊗ HB , (2.106)
α2 |βi
α2 β 2
where m = 1, 2, 3, 4 is identified with (i, k) = (1, 1), (1, 2), (2, 1), (2, 2). The new vector is element
of the 4-dimensional vector space HA ⊗ HB . If {|αii } and {|βik } are bases in their respective
spaces HA and HB , then {|γim } is a basis of the product space HA ⊗ HB .
Figure 2.1: Illustration of the isomorphism between HA ⊗ HB and HI ⊗ HA⊗B .
For observables we proceed in the same way: The external product of two commutators
spans a Hilbert product space with the dimension corresponding to product of the dimensions
of the sub-spaces. Assuming that,
X X
Â ≡ |iiAij hj| and B̂ ≡ |kiBkl hl| and [Â, B̂] = 0 . (2.107)
i,j k,l
then X
Â ⊗ B̂ ≡ |ikiAij Bkl hjl| , (2.108)
(ik)(jl)
such that
dim Â ⊗ B̂ = dim Â dim B̂ .
For example, |iihj| ⊗ |kihl| = |ikihjl|.
For two two-dimensional operators Â and B̂, the external product is defined by,
 
A11 B11 A11 B12 A12 B11 A12 B12
A11 B21 A11 B22 A12 B21 A12 B22 
Â ⊗ B̂ = 
A21 B11 A21 B12 A22 B11 A22 B12  ,
 (2.109)
A21 B21 A21 B22 A22 B21 A22 B22
and can be decomposed as,

  
A11 A12 B11 B12
 A11 A12   
Â ⊗ B̂ = (Â ⊗ I)(I ⊗ B̂) =   B21 B22  . (2.110)
 A21 A22  B11 B12 
A21 A22 B21 B22
Obviously, the external product is associative (Â⊗ B̂)⊗C = Â⊗(B̂ ⊗C), but does not commute.
Nevertheless, we can reverse the order of the product of two operators using,
 
1
 0 1 
Â ⊗ B̂ = Ŝ(B̂ ⊗ Â)Ŝ com Ŝ = 
 1 0  .
 (2.111)
1
We note, that it is important to distinguish from what space the vector came from. In our
notation, the vector before the symbol of the tensorial product (⊗) is belongs to the space HA ,
and the one after the ⊗ belongs to the space HB . With the definition (2.107) we can verify that
the operators only act on their respective states:
(A ⊗ B)(|αi ⊗ |βi) = A|αi ⊗ B|βi . (2.112)
We can check the relationship (2.112) by the definitions (2.106) and (2.109) of the external
product,
  
A11 B11 A11 B12 A12 B11 A11 B12 α11 β11
A11 A12 B11 B12 α1 β1 A11 B21 A11 B22 A12 B21 A11 B22   
⊗ ⊗ =  α11 β21 
A21 A22 B21 B22 α2 β2 A21 B11 A11 B12 A22 B11 A22 B12  α21 β11 
A21 B21 A11 B22 A22 B21 A22 B22 α21 β21
 
A11 B11 α11 β11 + A11 B12 α11 β21 + A12 B11 α21 β11 + A11 B12 α21 β21
A11 B21 α11 β11 + A11 B22 α11 β21 + A12 B21 α21 β11 + A11 B22 α21 β21 
=A21 B11 α11 β11 + A11 B12 α11 β21 + A22 B11 α21 β11 + A22 B12 α21 β21 
 (2.113)
A21 B21 α11 β11 + A11 B22 α11 β21 + A22 B21 α21 β11 + A22 B22 α21 β21
 
(A11 α1 + A12 α2 )(B11 β1 + B12 β2 )
(A11 α1 + A12 α2 )(B21 β1 + B22 β2 )
=(A21 α1 + A22 α2 )(B11 β1 + B12 β2 )

(A21 α1 + A22 α2 )(B21 β1 + B22 β2 )

A11 α1 + A12 α2 B11 β1 + B12 β2 A11 A12 α1 B11 B12 β1
= ⊗ = ⊗ .
A21 α1 + A22 α2 B21 β1 + B22 β2 A21 A22 α2 B21 B22 β2
Example 6 (External product of two spins): As an example we can consider a system

with two spins L · S.
2.3.8.3 Direct external sum

Using the nomenclature (2.107) we define the external direct sum by,
X
Â ⊕ B̂ ≡ |iki(Aij + Bkl )hjl| , (2.114)
(ik)(jl)
that is,  
A11 + B11 A11 + B12 A12 + B11 A12 + B12
A11 + B21 A11 + B22 A12 + B21 A12 + B22 
Â ⊕ B̂ = 
A21 + B11
 . (2.115)
A21 + B12 A22 + B11 A22 + B12 
A21 + B21 A21 + B22 A22 + B21 A22 + B22
It can be decomposed as,

1 1 1 1
Â ⊕ B̂ = Â ⊕ O + O ⊕ B̂ = Â ⊗ + ⊗ Â . (2.116)
1 1 1 1
Again, using the definition (2.111) of the unitary operator Ŝ, we can reverse the order of the
operator by,
Â ⊕ B̂ = Ŝ(B̂ ⊕ Â)Ŝ . (2.117)
Example 7 (Direct external sum of two diagonal Hamiltonians): As an example we
consider a two-level atom excited by radiation and trapped in an external harmonic potential.
We assume that the degrees of freedom do not interact. As the Hamiltonian of the HO is
diagonal, the total Hamiltonian is organized into a diagonal matrix of quadratic subspaces,
 
..

~ω
2
~ω
2 + ~Ω 0 0 . 
 ~ω .. 
 2 + ~Ω ~ω
+ ~∆ 0 0 . 
2 
0 ~Ω  . 
Ĥ = ~ω(n + 2 ) ⊕
1
= .
.  .
2 + ~Ω
3~ω 3~ω
~Ω ~∆  0 0 2 
 .. 

2 + ~Ω 2 + ~∆
3~ω 3~ω
 0 0 . 
..
··· ··· ··· ··· .
It acts on the product state |ni|ii, where the first ket denotes the vibrational level and the
second ket the electronic excitation of the atom.
Another example is studied in 2.6.3.12.
2.3.8.4 Trace
The trace of an operator over a subspace reduces its domain to the remaining dimensions (the
·-symbol is a place holder for the dimension over which we do NOT want to trace):
X X
TrB Â ⊗ B̂ = h· m|ikiAij Bkl hjl| · mi = |iiAij Bkl hj|δkm δlm (2.118)
(ik)(jl)(·m) (ik)(jl)(·m)
X X
= |iiAij Bmm hj| = Â Bmm = ÂTrB B̂ .
(i)(j)(m) m
For example, Trρ Â ⊗ ρ̂ = Â. See the Excs. 2.6.3.14 and 2.6.3.15.
It can be shown,
TrÂB̂ = TrB̂ Â . (2.119)
2.4 Time evolutions

2.4.1 Unitary transformations
The best we can do to characterize a system is, obviously, to measure all its observables. However,
neither the state functions nor the observables are fixed unambiguously, since defining a unitary
operator, Û † = Û −1 , we can do,
hψ|Â|ψi = hψ|Û † Û ÂÛ † Û |ψi = hÛ ψ|Û ÂÛ † |Û ψi . (2.120)
That is, exchanging |ψi by Û |ψi and at the same time Â by Û ÂÛ † , we obtain quantities
describing the same physical reality, since the eigenvalues are unchanged. This allows us to
choose the best mathematical representation for a specific problem. As an example, we will
apply the temporal unitary transformation to solve the dynamics of a coupled two-level system
in Exc. 2.6.4.1.
2.4.2 Schrödinger picture

Important examples of how the same system can be represented in different ways (related by
unitary transformations) are the Heisenberg, Schrödinger, and interaction pictures.
The Schrödinger picture, denoted by the subscript S, is defined by the choice of a Hamilto-
nian,
d d
ĤS = Ĥ(t, p̂S , r̂S ) with p̂S = r̂S = 0 . (2.121)
dt dt
That is, the observables of the system ÂS (t, p̂S , r̂S ) can only depend explicitly on time, but not
via the operators p̂S and r̂S , which are stationary,
d ∂ ÂS 0 ∂ Â 0 ∂ Â
S S
ÂS = + p̂˙S + r̂˙S . (2.122)
dt ∂t ∂pS ∂rS
This is,
d ∂
ÂS (t) = ÂS (t) . (2.123)
dt ∂t
In this case, the formal solution of the Schrödinger equation,
d
i~ |ψS (t)i = ĤS |ψS (t)i , (2.124)
dt
can be written,
|ψS (t)i = e−(i/~)ĤS t |ψS (0)i ≡ Û (t)|ψS (0)i . (2.125)
Apparently, the temporal dynamics is completely within the wave functions.
Example 8 (The time evolution operator ): Generalizing to an arbitrary initial time t0
we write the temporal translation operator,
U (t, t0 )|ψ(t0 )i = |ψ(t)i . (2.126)
By the expression (2.125) we find immediately, with t0 < t1 < t2 ,
U (t2 , t0 ) = U (t2 , t1 )U (t1 , t0 ) and U (t0 , t) = U † (t, t0 ) = U −1 (t, t0 ) = U (t, t0 )−1 .
The conjugate operator of time evolution acts on the vector ’bra’,
hψ(t)| = hψ(t0 )|U † (t, t0 ) .
2.4. TIME EVOLUTIONS 59
2.4.3 Heisenberg picture

As unitary transformations do not change the physics, the system described by,
|ψS (t)i −→ Û (t)† |ψS (t)i ≡ |ψH i and ÂS (t) −→ Û (t)† ÂS (t)Û (t) ≡ ÂH (t) (2.127)
with the transformation defined by equation (2.125), is equivalent. The subscript H means the
Heisenberg picture. In particular, we obviously have,
ĤS = ĤH ≡ Ĥ . (2.128)
Thus, the matrix element of the operator ÂS in Schrödinger’s picture with the time-dependent
base {|ψS i} is equal to the matrix element of the operator ÂH = U † ÂS U in Heisenberg’s picture
with the time-independent base {|ψH i}. In this picture the wavefunctions are independent of
time,
d d
|ψH i = |ψS (0)i = 0 , (2.129)
dt dt
but the operators depend im- and explicitly on time,
d d dÛ † dÛ ∂ ÂS (t)

ÂH (t) = Û (t)† ÂS (t)Û (t) = ÂS (t)Û (t) + Û (t)† ÂS (t) + Û (t)† Û (t)
dt dt dt dt ∂t
i −i ∂ ÂS (t)
= Ĥ † Û (t)† ÂS Û (t) + Û (t)† ÂS Ĥ Û (t) + Û † (t) Û (t) . (2.130)
~ ~ ∂t
That is,
d i ∂ ÂH (t)
ÂH (t) = [Ĥ, ÂH (t)] + . (2.131)
dt ~ ∂t
This so-called Heisenberg equation, which describes the temporal evolution of an operator acting
on time-independent states in the Heisenberg picture, is equivalent to the Schrödinger equation,
which expresses the temporal evolution of a quantum state in Schrödinger’s picture.
According to equation (2.131), the rate temporal variation of an operator in the Heisen-
berg representation is given by the commutator of that operator with the total Hamiltonian
of the system. Note that if an operator representing a dynamic variable commutes with the
Hamiltonian in the Schrödinger representation, it will also commute with the Hamiltonian in
the Heisenberg representation and thus with the complete set of commutating observables,
[Ĥ, ÂS ] = 0 ⇐⇒ [Ĥ, ÂH ] = 0 . (2.132)
We will show this in the Exc. 2.6.4.2.
Example 9 (Position and momentum operators in the Heisenberg picture): We

know that in Schrödinger’s picture (2.121), the operators p̂S and r̂S are stationary. Using
this fact in derivation (2.130), we can show for example for the momentum operator,
∂ ∂ d i
p̂S = 0 =⇒ p̂H = 0 =⇒ p̂H = [Ĥ, p̂H ] .
∂t ∂t dt ~
In the Exc. 2.6.4.3 we will use the Heisenberg picture to derive the equations of motion for
a particle confined to a potential.
2.4.4 Interaction picture

The interaction picture deals with problems where the total Hamiltonian is composed of an
time-independent part and a time-dependent part,
Ĥ = Ĥ0 + V̂ (t) . (2.133)
Analogously to Eq. (2.125), we define a time evolution operator in terms of the time-independent
part of the total Hamiltonian,
|ψI (t)i = eiĤ0 t/~ |ψS (t)i and AI (t) = eiĤ0 t/~ ÂS e−iĤ0 t/~ . (2.134)
Now we are interested in the temporal dependence of quantum states and operators in the
interaction picture. Replacing the inverse function |ψS (t)i = e−iĤ0 t/~ |ψI (t)i in the Schrödinger
equation (2.124) we immediately see,
∂
V̂ (t)|ψI (t)i = i~ |ψI (t)i . (2.135)
∂t
Apparently, in the interaction picture, only the perturbative term in Hamiltonian controls the
temporal evolution. Taking the time derivative of both sides of the equation by setting the
operator ÂI in the interaction representation to Eq. (2.134) results in,
dÂI i
= [Ĥ0 , ÂI ] . (2.136)
dt ~
Therefore, we see that the time derivative can be expressed in the form of a commutator, resem-
bling the Heisenberg equation (2.131), except that only the unperturbed term of the Hamiltonian
appears in the argument of the commutation operator.
Example 10 (Schrieffer-Wolff transformation): The Schrieffer-Wolff transformation

is a unitary transformation used to perturbatively diagonalize the system Hamiltonian to
first order in the interaction. As such, the Schrieffer-Wolff transformation is an operator
version of second-order perturbation theory. The Schrieffer-Wolff transformation is often
used to project out the high energy excitations of a given quantum many-body Hamiltonian
in order to obtain an effective low energy model. The Schrieffer-Wolff transformation thus
provides a controlled perturbative way to study the strong coupling regime of quantum-many
body Hamiltonians.
Consider a quantum system evolving under the time-independent Hamiltonian operator Ĥ
of the form Ĥ = Ĥ0 + V̂ , where H0 is a Hamiltonian with known eigenstates |mi and
corresponding eigenvalues Em , and where V is a small perturbation. Moreover, it is assumed
without loss of generality that V̂ is purely off-diagonal in the eigenbasis of Ĥ0 , i.e.,
hm|V̂ |m >= 0 (2.137)
for all m. Indeed, this situation can always be arranged by absorbing the diagonal elements
of V̂ into Ĥ0 , thus modifying its eigenvalues to,
0
Em = Em + hm|V̂ |mi . (2.138)
The Schrieffer-Wolff transformation is a unitary transformation which expresses the Hamil-

tonian in a basis (the ’dressed’ basis) where it is diagonal to first order in the perturbation
V̂ . This unitary transformation is conventionally written as:
Ĥ 0 = eiS Ĥe−iS . (2.139)

2.4. TIME EVOLUTIONS 61
When V̂ is small, the generator S of the transformation will likewise be small. The trans-
formation can then be expanded in S using the Baker-Campbell-Haussdorf formula,
Ĥ 0 = Ĥ + [iS, Ĥ] + 12 [iS, [iS, Ĥ]] + . . . . (2.140)
In terms of Ĥ0 and V̂ , the transformation becomes,
Ĥ 0 = Ĥ0 + V̂ + [iS, Ĥ0 ] + [iS, V̂ ] + 21 [iS, [iS, Ĥ0 ]] + 12 [iS, [iS, V̂ ]] + . . . . (2.141)
The Hamiltonian can be made diagonal to first order in V̂ by choosing the generator S such
that,
[Ĥ0 , iS] = V̂ . (2.142)
This equation always has a definite solution under the assumption that V̂ is off-diagonal in
the eigenbasis of Ĥ0 . Substituting this choice in the previous transformation yields:
Ĥ 0 = Ĥ0 + 21 [iS, V̂ ] + O(V̂ 3 ) . (2.143)
This expression is the standard form of the Schrieffer-Wolff transformation. Note that all the
operators on the right-hand side are now expressed in a new basis ’dressed’ by the interaction
V̂ to first order.
In the general case, the difficult step of the transformation is to find an explicit expression
for the generator S. Once this is done, it is straightforward to compute the Schrieffer-Wolff
Hamiltonian by computing the commutator [S, V̂ ]. The Hamiltonian can then be projected
on any subspace of interest to obtain an effective projected Hamiltonian for that subspace. In
order for the transformation to be accurate, the eliminated subspaces must be energetically
well separated from the subspace of interest, meaning that the strength of the interaction V̂
must be much smaller than the energy difference between the subspaces. This is the same
regime of validity as in standard second-order perturbation theory.
2.4.4.1 Hamiltonian in the transformed system

We have seen that the unitary transformation,
|ψU i = U † |ψi , ÂU = U † ÂU , (2.144)
leaves the physics of a system unchanged. The question is now, how the Schrödinger equation,
d
Ĥ|ψi = i~ |ψi (2.145)
dt
transforms into the new system, that is, what will the Hamiltonian Ĥ 0 look like in the trans-
formed equation,
? d
Ĥ 0 |ψU i = i~ |ψU i . (2.146)
dt
We calculate,
d d
i~ |ψU i = i~U † |ψi + i~U̇ † |ψi = (U † Ĥ + i~U̇ † )|ψi (2.147)
dt dt
= (U † Ĥ + i~U̇ † )U |ψU i = (U † ĤU + i~U̇ † U )|ψU i = Ĥ 0 |ψU i . (2.148)
Hence,
Ĥ 0 = U † ĤU + i~U̇ † U . (2.149)
Example 11 (Interaction picture): The above derivation is general and holds for any
unitary transformation. We will now apply it to transform the Hamiltonian Ĥ = Ĥ0 + V̂ (t)
into the interaction picture via the transformation U = e−(i/~)Ĥ0 t . From
i i
U̇ † = Ĥ0 e(i/~)Ĥ0 t = Ĥ0 U † (2.150)
~ ~
we calculate,
h i i
Ĥ 0 = U † ĤU + i~U̇ † U = U † Ĥ0 + V̂ (t) U + i~ Ĥ0 U † U (2.151)
h i ~
= U † Ĥ0 + V̂ (t) U − Ĥ0 = U † V̂ (t)U , (2.152)
which confirms the validity of the Schrodinger equation ... in the interaction picture provide
the Hamiltonian is taken to be the perturbation part V̂ (t), only. In the Heisenberg picture
V̂ (t) = 0, such that,
Ĥ 0 = 0 . (2.153)
2.4.5 Ehrenfest’s theorem

For linear operators satisfying [Â, B̂] = i we can give a generalization of the commutation
relation:
δF (Â, B̂)
[Â, F (Â, B̂)] = i . (2.154)
δ B̂
This can be verified by a Taylor expansion of F (Â, B̂) by B̂ around B̂ = 0, as will be shown in
Exc. 2.6.4.4. An immediate consequence of [p̂, r̂] = −i~ is,
δF (r̂)
[p̂, F (r̂)] = −i~ . (2.155)
δr̂
The momentum observable is not singularly defined by the commutation relation, because
each unitarily transformed operator satisfies the relation as well. We can expand a unitarily
equivalent momentum as p̃ = U pU + = eiF (r) pe−iF (r) = p + i[F (r), p] + 2!1 [F (r), [F (r), p]] + ...
using the relation (2.199).
The observables in the Heisenberg picture follow the same equations of motion as the cor-
responding classical quantities. This correspondence principle is called Ehrenfest theorem. For
~2 2
example, when working with position and momentum variables [x̂, k̂] = i and Ĥ = 2m k̂ + V (x̂),
we obtain,
δ Ĥ δ Ĥ
[x̂, Ĥ] = i~ and [p̂, Ĥ] = −i~ , (2.156)
δ p̂ δ x̂
and using the Heisenberg equation (2.131),
δ Ĥ δ Ĥ
x̂˙ = and p̂˙ = − . (2.157)
δ p̂ δ x̂
We will demonstrate this in Exc. 2.6.4.5 for the case of a harmonic potential.
In the Schrödinger picture the equation of motion for the eigenvalues of the observables takes
the form,
d ∂ i
hÂS i = h∂t ψ|ÂS |ψi + hψ|∂t ÂS |ψi + hψ|ÂS |∂t ψi = hÂS i + h[Ĥ, ÂS ]i . (2.158)
dt ∂t ~
2.5. SYMMETRIES IN QUANTUM MECHANICS 63
The eigenvalues behave as Heisenberg observables in Eq. (2.130), that is, they follow the laws
of Hamilton’s and Newton’s mechanics.
The important result now is that the equations that govern the eigenvalues of the observables
are identical in the both pictures, since from the Heisenberg picture we obtain with Eq. (2.130),
d ∂ i
hÂH i = hÂH i + h[Ĥ, ÂH ]i .
dt ∂t ~
2.5 Symmetries in quantum mechanics

We already saw in Sec. 2.3.4 that, beyond observables, there is another category of operators
that does not correspond to measurable physical quantities, but is very useful in the quan-
tum formalism. These are the unitary transformationunitary operator In this section we will
encounter some interesting examples.
2.5.1 Translation and rotation

2.5.1.1 Temporal translation operator
The temporal evolution of a system is described by the Schrödinger equation whose formal
solution can be written as follows,
|ψ(t)i = e−iĤt/~ |ψ(0)i . (2.159)
With this we can define an evolution operator or temporal translation,
Utp (τ ) ≡ e−iĤt/~ tal que Utp (τ )|ψ(t)i = |ψ(t + τ )i . (2.160)
The temporal evolution has already been discussed extensively in Sec. 2.4.
2.5.1.2 Spatial translation operator

In this section we look for a unitary translation operator,
Ttr r ≡ a + r . (2.161)
Before this, we need to derive the following calculation rule for commutators, which will be done
in Exc. 2.6.5.1:
eÂ B̂e−Â = B̂ + [Â, B̂] + 2!1 [Â, [Â, B̂]] + ... . (2.162)
Applying this formula to the two operators p̂ and r̂ related by the commutation rule (2.47),
we obtain,
0
e(i/~)a·p̂ r̂e(−i/~)a·p̂ = r̂ + [(i/~)a · p̂, r̂] + 1
2! [(i/~)a · p̂, a] + ... = r̂ + a . (2.163)
That is, the operator

Utr (a) ≡ e(−i/~)a·p̂ (2.164)
performs a spatial translation of the position operator. The operator is unitary,
Utr (a)−1 = Utr (a)† , (2.165)

and forms a group since Utr (a)Utr (b) = Utr (a + b). Summarizing the impact of the translation
on the operators of space,
† †
Utr (a)r̂Utr (a) = r̂ + a , Utr (a)p̂Utr (a) = p̂ , (2.166)
where the second relation is obvious.

To demonstrate how the translation acts on a state, let us calculate,
r̂e(−i/~)a·p̂ |ri = e(−i/~)a·p̂ (r̂ + a)|ri = (r + a)e(−i/~)a·p̂ |ri . (2.167)
Hence,
Utr (a)|ri = e(−i/~)a·p̂ |ri = |r + ai . (2.168)
Finally, comparing the expansion of the translation operator

(−i/~)a·p̂ i 1 (a · p̂)2
e |ri = 1 − a · p̂ − 2 + .. |ri , (2.169)
~ ~ 2!
with the Taylor expansion of the translated state,

(a · ∇)2
|r + ai = 1 + a · ∇ + + .. |ri . (2.170)
2!
we obtain
~
p̂|ri = − ∇|ri . (2.171)
i
2.5.1.3 Rotation operator

In this section we look for the unitary transformation corresponding to the rotation operator,
Trt r ≡ eα~ × r . (2.172)
We calculate,
X (~
α×)n
eα~ × r = ~ × r + 21 α
r=r+α ~ × (~
α × r) + .. (2.173)
n
n!
= êα (êα · r) + êα × r sin α − êα × (êα × r) cos α ,
as we will see in Exc. 2.6.5.2. We define the unitary rotational transformation by,
†
Urt (~ α) = eα~ × r̂
α)r̂Urt (~ , α)|ri = |eα~ × ri .
Urt (~ (2.174)
To derive the explicit form of the rotation operator, we consider two rotations about the
same axis α
~ = λ1 êα + λ2 êα , such that
Urt (λ1 êα )Urt (λ2 êα ) = Urt (λ1 êα + λ2 êα ) . (2.175)
Calculating the derivative of this equation by λ1 and then setting λ1 = 0, we have,

dUrt (λ1 êα ) dUrt (λ1 êα +λ2 êα ) d(λ1 +λ2 )
dλ1 U rt (λ2 êα ) = d(λ1 +λ2 ) dλ1
λ1 =0 λ1 =0 λ1 =0

=⇒ dλdλ1 ê1α · ∇α~ Urt (~ α)|α~ =0 Urt (λ2 êα ) = dUrtdλ(λ2 êα )
2
, (2.176)
λ1 =0
L dUrt (λ2 êα )
=⇒ êα · i~ Urt (λ2 êα ) = dλ2
where we define the angular momentum operator L̂ ≡ i~ ∇α~ Urt (~

α)|α~ =0 . The solution of the last
~ |λ1 =0 ,
differential equation is, with λ2 êα = α
α) = e(−i/~)L·~α .
Urt (~ (2.177)
The explicit form of L follows from its action on a state |ri. Comparing the expansion of the
operator (2.177),
Urt (α)|ri = 1 − ~i L · α
~ + ... |ri (2.178)
with the Taylor expansion of the state,
|eα~ × ri = |r + α
~ × r + ...i = |ri + (~
α × r) · ∇r |ri + ... , (2.179)
we find,
− ~i L̂ · α
~ |ri = (~ ~ · (r × ∇r )|ri = − ~i r̂ × p̂|ri ,
α × r) · ∇r |ri = α (2.180)
that is,
L̂ = r̂ × p̂ . (2.181)
Therefore, the observable L̂ is the orbital angular momentum of the particle producing the
rotations.
2.5.2 Galilei and Lorentz boosts

The Galilei transform (or Galilei boost) is defined by,
Tv r = r + vt and Tv p = p + mv . (2.182)
Holds Tv1 Tv2 = Tv1 +v2 . We derive the expression for the operator of this transformation from
its actions on the position and momentum states. For the position we have with (2.164),
UG (v)|ri ∝ e(−i/~)p̂·vt |ri = |r + vti . (2.183)
For the momentum we can proceed analogously,
UG (v)|pi ∝ e(i/~)r̂·vm |pi = |p + mvi . (2.184)
That is, by defining G = p̂t − r̂m = i~∇v UG (v)|v=0 , the unitary transformation
UG (v) = e(−i/~)v·G (2.185)
meets the requirements (2.183) and (2.184) 13 . We just need to check,

Z Z
e(i/~)v·rm |ri = e(i/~)v·rm |pihp|rid3 p = h3/2
1
|pie−(i/~)r·(p−vm) d3 p = |ri (2.186)
R3 R3
and analogously,
Z Z
(−i/~)v·pt (−i/~)v·pt 3
e |pi = e |rihr|pid p = 1
h3/2
|rie(i/~)p·(r−pt) d3 p = |ri . (2.187)
R3 R3
13
We need to keep in mind that, following Glauber’s rule, eÂ eB̂ = eÂ+B̂+[Â,B̂]/2 with Â ≡ v · p̂t and B̂ ≡ v ·r̂m.
However, the commutator [v · p̂t, v · r̂m] = −i~mtv2 is independent of p̂ and r̂, such that it only contributes an
unimportant phase.
With the commutator of p̂ and r̂ we derive,
[G · a, G · b] = 0 , (2.188)
and with that, we verify,

2
UG (v1 )UG (v2 ) = e(−i/~)(v1 +v2 )·G−[v1 ·G,v2 ·G]/~ = UG (v1 + v2 ) . (2.189)
Obviously, for very high velocities, the Galilei-boost should be replaced by the Lorentz trans-
form (or Lorentz boost) [139]. Here, we only note, that the additivity of velocities expressed by
equation (2.188) does not hold for non-collinear relativistic velocities 14
2.5.3 Gauge transformations

We learn in electrodynamics 15 , that the motion of a particle carrying the charge q and interacting
with an electrical potential Φ(r, t) and a magnetic vector potential A(r, t) is governed by the
electric and the magnetic field,
E(r, t) = −∇Φ − ∂t A and B(r, t) = ∇ × A . (2.190)
Also, we know that the fields are invariant under the substitution,
Φ → Φ0 ≡ Φ − ∂t χ and A → A0 ≡ A + ∇χ , (2.191)
where χ(r, t) is a scalar field called gauge field.

In quantum mechanics thegauge transformation defined by,
Ucl (χ) = e−iqχ(r,t)/~ (2.192)
obviously keeps the Schrödinger equation invariant. Transforming operators and wave functions
as,
Ĥ → Ucl ĤUcl−1 ≡ ĤU and |ψi → Ucl |ψi ≡ |ψU i , (2.193)
we calculate for the energy,

d −1 −iq −1 dχ
ĤU |ψU i = Ucl i~ dt Ucl |ψU i = Ucl i~Ucl−1 dt
d
|ψU i + Ucl i~ ~ Ucl dt |ψU i

= i~ dtd
− iq~ dχdt |ψU i , (2.194)
and for the momentum,

−iq −1
p̂U |ψU i = Ucl (−i~∇)Ucl−1 |ψU i = Ucl (−i~)Ucl−1 (∇|ψU i) + Ucl (−i~) ~ Ucl ∇χ |ψU i
h i
= (−i~) ∇ − iq~ (∇χ) |ψU i , (2.195)
This corresponds to the substitutions 16 ,
d −1 d dχ
Ucl i~ Ucl = i~ + q and Ucl p̂Ucl−1 = p̂ − q∇χ . (2.196)
dt dt dt
14
See the script Electrodynamics of the same author [55].
15
See the script Electrodynamics of the same author [55].
16
In quadrivetorial notation i~∂µ −→ i~∂µ + q∂µ χ.
This shows that the gauge transformation applies to the minimum coupling rule,
Ucl Ucl
Ĥ = Ĥkin + qΦ y Ĥkin + qΦ + q∂t χ and mv = p − qA y p − qA − q∇χ , (2.197)
confirming the rules (2.191). That is, the Hamiltonian of a particle carrying the charge q and
interacting with an electric potential Φ and a potential magnetic vector A is,
Ĥ = 1
2m (p − qA − q∇χ)2 + qΦ + q∂t χ . (2.198)
2.5.4 Noether’s theorem and conservation laws

The fundamental laws of physics are often expressed as symmetries. The knowledge of sym-
metries allows the characterization of a system and its behavior without the need to know its
details. We can often deduce the differential equation of motion from the symmetries. The fun-
damental symmetries define the fundamental laws of physics. Following Noether’s theorem each
symmetry corresponds to a conserved quantity, that is, a quantities that remains invariant for all
time. The invariance of a system under symmetry transformation represents a conservation law.
For example, the homogeneity of space corresponds to the conservation of linear momentum.
In quantum mechanics, a symmetry transformation is defined by,
|ψi −→ U |ψi and Q̂ −→ U Q̂U † . (2.199)
Therefore, to find a conservation law, i.e., an invariable observable (also called constant of
motion), we must verify that the observable and the transformed wavefunctions simultaneously
satisfy the same fundamental equations (that is, Schrödinger’s or Heisenberg’s equation) as
the original observable and wavefunctions. For example, if the wavefunction |ψi satisfies the
Schrödinger equation, the wave function U |ψi must do this too,
! d dU d dU
ĤU |ψi = i~ U |ψi = i~ |ψi + i~U |ψi = i~ |ψi + U Ĥ|ψi . (2.200)
dt dt dt dt
Consequently, we obtain the relation,
[Ĥ, U ] = i~U̇ . (2.201)
As shown in (2.158) and (2.159), an operator that commutes with the Hamiltonian does not
explicitly depend on time, that is, it is conserved.
2.5.4.1 Temporal homogeneity

Temporal homogeneity means invariance under translation in time, that is, under the unitary
temporal transformation,
U (τ ) ≡ |ψ(τ )ihψ(0)| = e(i/~)Êτ . (2.202)
d (i/~)Êτ
Since dt e = 0, this means [e(i/~)Êτ , Ĥ] = 0, which implies conservation of energy [Ê, Ĥ] =
0. This will be verified in the Exc. 2.6.5.3.
Example 12 (Homogeneity of time): We imagine the following mental experiment or

Gedankenexperiment: We consider two attractive bodies that move away from each other
until they reach the perihelia. At this point, before the bodies reapproach, we change the
laws, for example, by modifying the force of attraction. As a consequence, when the bodies
arrive at the initial point, the total energy is non-zero. Therefore, the conservation of energy
indicates that the laws are invariant.
2.5.4.2 Temporal isotropy

The fundamental laws of classical physics and quantum mechanics are all symmetrical under
time reversal. That is, they are remain invariant when we change the arrow of time, t → −t.
2.5.4.3 Spatial homogeneity

Spatial homogeneity means invariance under spatial translation, that is, under the unitary trans-
lational transformation,
Utr (a) ≡ |r + aihr| = e(−i/~)p·a . (2.203)
This is equivalent to momentum conservation [p̂, Ĥ] = 0 17 .
Example 13 (Homogeneity of space): Ehrenfest’s theorem says [p̂, H] = −i~ ∂H

∂p̂ . There-
fore, the commutator is not zero when there is a potential, Ĥ = p̂2 /2m + V (r̂). This is obvi-
ous, because the potential introduces an energy inhomogeneity to a particle interacting with
the potential. However, this does not mean that the space itself is inhomogeneous, because
in order to verify the translational invariance of space, we must displace the entire system,
that is, the particle together with the potential. For example, if the potential is generated
by another particle we must consider the Hamiltonian Ĥ = p̂21 /2m1 + p̂22 /2m2 + V (r̂1 − r̂2 ).
2.5.4.4 Spatial isotropy

Spatial isotropy means invariance under rotation, that is, under rotational unitary transforma-
tion,
Urt (φ) ≡ e(−i/~)L̂φ . (2.204)
This is equivalent to the conservation of angular momentum [L̂, Ĥ] = 0.
2.5.4.5 Parity conservation

Besides continuous symmetry transformations there exist discrete transformations. Discrete
symmetries are important in elementary particle physics. The parity conservation means invari-
ance to spatial reflection: r → −r. A parity transformation is defined by the mirroring of the
wavefunction through a point in space, for example r = 0,
P̂ |ψ(r)i ≡ |ψ(−r)i . (2.205)
with
P̂ 2 = P̂ . (2.206)
We talk about even parity when P̂ |ψ(r)i = |ψ(r)i and odd parity when P̂ |ψ(r)i = −|ψ(r)i. See
Exc. 2.6.5.4.
17
Imagine that the forces attracting two bodies to each other are not equal: Contrary to Newton’s third law,
body A attracts body B, more than the body B attracts the body A. In that case after a while the two bodies
have different momenta. With the unitary transformation Utr (a) = e−ip·a/~ ' 1 − ip · a/~ + ... we have,
Utr H|ψi = Utr Eψ |ψi = Eψ U |ψi = H|ψ(r + a)i =? = H|ψ(ri .
Since, [H, p] = 0, Heisenberg’s equation yields,

∂ 1
hψ|p · a|ψi = hψ|[p · a, H]|ψi = 0 .
∂t i~
2.5.4.6 Invariance to the velocity of the inertial system

The Galilei boost asks for Galilei invariance regarding the transformation,
UG (v) ≡ |r + vt, p + mvihr, p| , (2.207)
that is, the independence of the inertial system on its velocity v.
2.5.4.7 Charge conservation

Let us consider again the gauge transform (2.191). We know that the Lagrangian density in free
space is given in terms of the potentials by,
L(xµ ) = 1 µν µ ε0 2 1 2
4µ0 F Fµν − Aµ j = 2 E − 2µ0 B − Aµ j
µ
(2.208)
ε0 2 1 2
= 2 [∇Φ + ∂t A] − 2µ0 [∇ × A] − Φρ + A · j ,
and the action is simply the fourth-dimensional integral,

Z
S = L(xµ )dV dt . (2.209)
From the Lagrangian formulation, Maxwell’s equations can be derived by requiring the action
to be minimal, δS = 0, which yields the Euler-Lagrange equations. As the field equations do
not change under gauge transformation, this implies that the action is also unchanged.
To find the relation with charge conservation, we simply have to compare the actions in
different gauges. First, we express the Lagrangian transformed into the old gauge,
0
L0 (xµ ) = ε0
2 {∇[Φ − ∂t χ ] + ∂t [A + ∇χ ]}2 − 1
2µ0 {∇ × [A + ∇χ ]}2 (2.210)
− [Φ − ∂t χ]ρ + [A + ∇χ] · j
= L + (∂t χ)ρ + ∇χ · j .
With this result, we can calculate the difference between the actions under gauge transformation
and recall, that they can not be different:
Z Z
! 0
0 = S − S = [(∂t χ)ρ + ∇χ · j]dV dt = − χ[∂t ρ + ∇ · j]dV dt , (2.211)
using partial integration 18 and choosing volumes so large, that every charge is inside. This is
the continuity equation derived from the gauge invariance of the action. The calculation really
is nothing more than an application of Noether’s theorem from which we could have derived
directly the continuity equation, ∂µ j µ = 0.
In summary, the conservation of charge means invariance with respect to gauge transforma-
tions,
Ucl (χ) ≡ e−iqχ(r,t)/~ , (2.212)
where χ is the gauge field. We note that q and χ are conjugated observables. Therefore, if
[χ, Ĥ] = 0, then the charge q is a conserved quantity.
Transformations can be combined. For example, we believe that nowadays all laws are
invariant with respect to CPT transformation, that is, a combination of charge conjugation,
parity inversion, and θ transform.
18
R
Think about the argument, because ∂t |χρdt = 0!
2.6 Exercises
2.6.1 Basic notions
2.6.1.1 Ex: Conservation of probability
Demonstrate the conservation of local probability through the definitions of probability densities,
ρ(r, t), and probability current j(r, t).
2.6.1.2 Ex: Fourier theorem

The spatial distribution of a particle is given by a Gaussian function with the width ∆x. Calcu-
late the momentum distribution and its width ∆p. Just consider one spatial dimension. Show
that ∆x∆p = ~ using the rms definition for the widths.
2.6.2 Postulates of quantum mechanics

2.6.2.1 Ex: Reality of eigenvalues
Show that the eigenvalues of an observable are real.
2.6.2.2 Ex: Normalization of the Bloch vector

Calculate the eigenvalue of the length of the Bloch vector (2.46).
2.6.2.3 Ex: Quantum superposition

Discuss how matrices can describe the creation of superposition states at the example
of a particle
passing through a double slit. Identify the slits with the states h1| = 1 0 and h2| = 0 1
and construct an observable for the position of the particle. How this observable must behave
in the classical limit?
2.6.2.4 Ex: Quantum measurement

Explain the idea of quantum measurement at the example of a measurement of the excitation
energy of a two-level atom.
2.6.2.5 Ex: Two-level atom

Consider a two-level atom. The Hamiltonian is given by,

0 0
Ĥ = .
0 ~ω0
Using the stationary Schrödinger equation, calculate the eigenvalues and eigenvectors.
2.6.2.6 Ex: The ammonium molecule

Consider the two states |1i and |2i of the ammonium molecule outlined in the figure. Suppose
they are orthormal, hi|ji = δij , and that only these two states are accessible to the system, so
2.6. EXERCISES 71
that we can describe it using the basis formed by |1i and |2i. On this basis the Hamiltonian Ĥ
of the system is given by,
E0 −E1
Ĥ = .
−E1 E0
a. If the system is initially in state |1i, will it remain in that state at a later time? How about
if the initial state is |2i?
b. Obtain the eigenvalues EI and EII and the respective eigenvectors |Ii and |IIi of Ĥ, express-
ing them in terms of |1i and |2i.
c. What is the probability of measuring an energy EI in the following state,
|ψi = √1 |1i − √2 |2i ?

5 5
d. Based on the above result, we can predict at least one possible electromagnetic radiation
emission frequency for an ammonia sample. What is this frequency?
EUF 2014, 1 Semestre, Mecânica Quântica - Q1 1
/ )
)
)
)
)
/
)
|1 > |2>
Figure 2.2: The two states of the ammonium molecule.
(a) Não, já que estes estados não são autoestados do hamiltoniano. A evolução temporal do sistema
mistura os dois e temos duas combinações específicas com energia bem definida, os autoestados de
H. Começando em |1> ou |2> teremos a possibilidade de medir qualquer uma das duas energias
possíveis dependendo do tempo esperado.
2.6.3 Abstract formalism(b)of quantum mechanics

2.6.3.1 Ex: Orthogonality
Show that two eigenvectors of a Hermitian operator associated with two different eigenvalues
are orthogonal.
2.6.3.2 Ex: Orthonormalization

Orthonormalize the base ha1 | = 1 −1 0 , ha2 | = 0 1 0 , ha3 | = 0 1 1 .
2.6.3.3 Ex: Orthonormal base

Construct an orthonormal basis for the following operator describing a partially degenerate
three-level system,  
1 1 1
Â = 1 1 1 .
1 1 1
2.6.3.4 Ex: Eigenvalue equation

1 −i
Calculate the unitary matrix U transforming the Hamiltonian Ĥ = into a diagonal
i 1
matrix E = U † ĤU .
2.6.3.5 Ex: Eigenvalues and eigenvectors

 
1 1 1
Find the eigenvalues and -vectors of the operator Â = 1 1 1 and construct the unitary
1 1 1
matrix which transforms this operator into a diagonal matrix.
2.6.3.6 Ex: Commuting operators

a. Show that if two operators Â and B̂ commute and if |ψi is an eigenvector of Â, B̂|ψi also is
an eigenvector of Â with the same eigenvalue.
b. Show that if two operators Â and B̂ commute and if |ψ1 i and |ψ2 i are two eigenvector of Â
with different eigenvalues, the matrix element hψ1 |B̂|ψ2 i is equal to zero.
c. Show that if two operators Â and B̂ commute, we can construct an orthonormal basis of
eigenvectors common to Â and B̂.
2.6.3.7 Ex: Eigenvalues

 
1 0 1
a. Find the eigenvalues and eigenvectors of the operator Â = 0 µ 0 for 0 < µ < 2.
1 0 1
b. Write down the unitary matrix U satisfying the eigenvalue equation: ÂU = U EA , where EA
is the matrix that has all eigenvalues of Â in its diagonal.
c. Now consider the case µ = 0. Find a complete set of commuting operators (CSCO). That is,
calculate the components of a second operator B̂ which commutes with Â as a function of its
eigenvalues λ1 , λ2 , and λ3 , and verify [Â, B̂] = 0. Find the most general form of operator B̂.
2.6.3.8 Ex: Schwartz inequality

Demonstrate the Schwartz inequality |hu|vi|2 ≤ hu|uihv|vi.
2.6.3.9 Ex: Heisenberg’s uncertainty principle

Develop the formal derivation of Heisenberg’s uncertainty principle.
2.6.3.10 Ex: Fourier transform

Show that hr|P̂|ψi = ~i ∇hr|ψi reproduces the Schrödinger equation in position representation.
2.6.3.11 Ex: Projection of the motion of a particle

Project the Hamiltonian of the motion of a free particle onto the plane x-y at the position z = z0 .
2.6.3.12 Ex: Complete system of commuting operators

Construct the Hilbert space of two independent two-level systems.
2.6.3.13 Ex: Liouville equation

1 0 0 0
Show that ⊗ Â + ⊗ B̂ = Â ⊕ B̂.
0 0 0 1
2.6. EXERCISES 73
2.6.3.14 Ex: Liouville equation

Show at the example of a two-level system that the von Neumann equation, ρ̇ = Lρ̂ = − ~i [Ĥ, ρ̂],
~˙ = − ~i (Ĥ ⊗ I − I ⊗ Ĥ)~
can be written, ρ ρ, using the definition of the external product.
2.6.3.15 Ex: Unitary transformation of singlet states

Consider two spins a and b that do not interact. Applying to each spin the same trans-
formation to another base, show that the singlet state has in each base the following form:
|ψi = √12 (| ↑ia | ↓ib − | ↓ia | ↑ib ).
2.6.4 Time evolutions

2.6.4.1 Ex: Coupled two-level atom
Calculate the time evolution of an atom with two levels coupled by a light field using the
Hamiltonian,
1
0 2 ~Ω
Ĥ = 1 ,
2 ~Ω ~∆
where ∆ = ω − ω0 is the detuning between the frequency of the light and the frequency of the
transition and Ω the Rabi frequency. Help: Determine the matrix of the eigenvalues Ê and the
unitary transformation U given by U † ĤU = Ê and use the formal solution of the Schrödinger
†
equation: |ψ(t)i = e−iĤt/~ |ψ0 i = e−iU ÊU t/~ |ψ0 i = U † e−iÊt/~ U |ψ0 i 19 .
2.6.4.2 Ex: Commutator in Schrödinger’s and Heisenberg’s picture

Show that operators which commute with the Hamiltonian in the Schrödinger picture also do it
in the Heisenberg picture. Use the rule [Ĥ, ÂB̂] = Â[Ĥ, B̂] + [Ĥ, Â]B̂.
2.6.4.3 Ex: Motion in Heisenberg’s picture

p̂ 2
Consider the Hamiltonian Ĥ = 2m +m 2 2
2 ω x̂ . Using the relation [p̂, x̂] = −i~ calculate in the
Heisenberg picture the equations of motion for the observables p̂, x̂, and p̂x̂.
2.6.4.4 Ex: Commutator of a function of operators

Prove the relationship (2.154).
2.6.4.5 Ex: Ehrenfest’s theorem

Compare the equations of Ehrenfest’s theorem with those of Hamilton-Jacobi for a classical
particle subject to a time-independent potential. Discuss the classical limit, that is, when the
Hamilton-Jacobi equations approach those of Ehrenfest.
2.6.5 Symmetries in quantum mechanics

2.6.5.1 Ex: Calculus with commutator
Derive the rule (2.162) via a Taylor expansion of the operator Ĝ(τ ) ≡ eτ Â B̂e−τ Â .
19
The MATLAB code for calculating time evolution (QM Fundaments Evolucao.m) can be found on the web
page of this course.
2.6.5.2 Ex: Rotation operator

P ×)n
Derive the rule eα~ × r = n (~αn! r = êα (êα · r) + êα × r sin α − êα × (êα × r) cos α.
2.6.5.3 Ex: Constants of motion

Show at the example of energy conservation using the relation (2.201), that energy commutes
with the Hamiltonian if Ė = 0.
2.6.5.4 Ex: Parity

Show that the eigenfunctions of the Hamiltonian Ĥ = −(~/2m)(d2 /dx2 )+V (x) have well-defined
parity, i.e., parity is an good quantum number in cases where the energy is an even function of
position, V (x) = V (−x).
Chapter 3
Linear motion / Separable potentials

In this chapter we will analyze the translational and vibrational motion of a quantum particle.
We will give special consideration to the rectangular potential and the harmonic oscillator.
3.1 Translational motion

In one dimension the Hamiltonian of a free particle is,
~2 d2
Ĥ = − . (3.1)
2m dx2
Therefore, the general solution of the Schrödinger stationary equation,
Ĥψ(x) = Eψ(x) , (3.2)
is, q
ψ(x) = Aeikx + Be−ikx with k= 2mE
~2
. (3.3)
R∞ R∞
Note that the eikx functions are not quadratically integrable, since −∞ |eikx |2 dx = −∞ dx → ∞.
On the other side, they do not represent actual physical systems. In practice, we need to consider
wave packets or specify a finite volume for the particle. Note also that the spectrum of eigenvalues
is continuous. Do the Exc. 3.5.1.1.
3.1.1 Quadratic integrability

To allow for an interpretation as probability density we need to ask for quadratic integrability,
Z
|ψ|2 d3 r = 1 . (3.4)
This means that the wavefunction can not be infinite inside a finite volume. But it can be
infinite within an infinitely small volume. Also, since the Schrödinger equation contains the
second derivative by position, the wavefunction must be continuous and have a continuous
derivative.
3.1.2 Separation of dimensions

Frequently, a 3D potential can be written in the way,
V (x, y, z) = Vx (x) + Vy (y) + Vz (z) . (3.5)
75
76 CHAPTER 3. LINEAR MOTION / SEPARABLE POTENTIALS
This is the case, for example, for a rectangular well with Vx (x) = Vy (y) = Vz (z) = V0 /3 inside
the well and V (x, y, z) = 0 outside. It also holds for a harmonic potential,
m 2 2
V (r) = ωx x + ωy2 y 2 + ωz2 z 2 . (3.6)
2
In these cases, the following ansatz for the wavefunction is generally useful,
ψ(r) = ψx (x)ψy (y)ψz (z) , (3.7)
since inserting the ansatz into the Schrödinger equation,

2
~2 d d2 d2
− + + + Vx (x) + Vy (y) + Vz (z) ψx (x)ψy (y)ψz (z) = Eψx (x)ψy (y)ψz (z) ,
2m dx2 dy 2 dz 2
(3.8)
the equation separates into three independent one-dimensional equations,
~2 ψx00 (x)
− + Vx (x) = const. ≡ Ex , (3.9)
2m ψx (x)
and the same for y and z. Since, E = Ex + Ey + Ez may have the same value for different
combinations of Ex , Ey e Ez , multidimensional systems are often degenerate.
3.2 Rectangular potential

3.2.1 Box potential
Let us now place the particle into a rectangular potential well, such that the Hamiltonian is,

~2 d2 0 para x ∈ [0, L]
Ĥ = − + V (x) com V (x) = . (3.10)
2m dx 2 ∞ para x ∈ / [0, L]
As the potential barriers are high, the walls are hard, that is, the particle, even being a quantum
particle, can not penetrate. The wavefunction and the possible energy values are,
r
2 nπx n2 ~2 π 2
ψ(x) = sin and En = . (3.11)
L L 2mL2
The Exc. 3.5.2.1 asks to demonstrate the result (3.11) illustrated in Fig. 3.1 1 .
Obviously the spectrum of eigenvalues is now discrete. They can be enumerated by an integer
n called quantum number. Note that the energy levels are not equidistant.
2 2
~ π
Example 14 (Localization energy ): There is a minimal energy E1 = 2mL 2 which is called
zero point energy or localization energy. This energy can be understood as a consequence of
Heisenberg’s uncertainty principle. We can make the following gross estimation of the zero
point energy. Obviously, the particle is localized with an uncertainty lower than ∆x < L.
Hence, ∆p > ~/∆x > ~/L. The average kinetic energy is,
hp2 i hpi2 + ∆p2 ∆p2 ~2

= = > .
2m 2m 2m 2mL2
The fact that the numerical value is different from the value calculated by the formula (3.11)
comes from the particular geometry of the box potential.
1
The MATLAB code for calculating the time evolution can be found on the webpage dedicated to this course.
3.2. RECTANGULAR POTENTIAL 77
40
35
30
25
E , ψ (μm)
20
15
10
0
0 0.5 1
x / L (μm)
Figure 3.1: (Code: QM M otion SquareW ell.m) Wavefunctions and energies in the rectangular
potential well.
3.2.2 Multidimensional box potential

In a multidimensional well there can be degeneracy if the well exhibits symmetries. In the case
of a 2D quadratic well Lx = Ly , the eigenenergies are doubly degenerate, since Enx ,ny = Eny ,nx .
In the case of a 3D cubic well Lx = Ly = Lz , the eigenenergies are 6-fold degenerate, because
Enx ,ny ,nz = Eny ,nz ,nx = Enz ,nx ,ny = Enz ,ny ,nx = Eny ,nx ,nz = Enx ,nz ,ny . The states and energies
of the 2D well are calculated in Exc. 3.5.2.2.
3.2.3 Potentials with several sections of constant depths

To find the global wavefunction in potentials with several sections of constant depths, we solve
Schrödinger’s equations separately for each section,

~2 d2
− + Va ψa (x) = Eψa (x) . (3.12)
2m dx2
The general solution for a section a with potential energy Va is,
ψa (x) = Aa eika x + Ba e−ika x , (3.13)

p
where ka = ~1 2m(E − Va ). If E > Va , the wave is propagating. ka is the Broglie wavevector
of the wave. If E < Va , the wave is evanescent. That is, the wave decays within a distance
κa = −ika .
If the particle is confined, that is, if E < V (x → ±∞), the possible energy levels are quantized
and the spectrum is discrete.
For every transition between two sections a = 1 and a = 2 we require the boundary condi-
tions,
ψ1 (x) = ψ2 (x) e ψ10 (x) = ψ20 (x) . (3.14)
R∞
Together with the normalization, 1 = −∞ |ψ|2 dx, these conditions are sufficient to determine
the wavefunction unambiguously.
Figure 3.2: Scheme of a potential with several sections of constant depths.
3.2.4 Potential well

Consider a particle with energy E and a potential well of finite depth such that V (x) = V0 < 0
for −L/2 > x > L/2 and V (x) = 0 otherwise, as illustrated on the left side of Fig. 3.3. The
particle be confined, E < 0.
Figure 3.3: Scheme of a two-sided (left) and one-sided (right) potential well.
The wavevectors are

√ p p
k1 = k3 = ~1 2mE = i ~1 2m|E| = iκ1 and k2 = 1
~ 2m(E − V0 ) . (3.15)
with κ1 ∈ R+ . The boundary conditions yield,
A1 e−ik1 L/2 + B1 eik1 L/2 = A2 e−ik2 L/2 + B2 eik2 L/2 (3.16)

−ik1 A1 e−ik1 L/2 + ik1 B1 eik1 L/2 = −ik2 A2 e−ik2 L/2 + ik2 B2 eik2 L/2
A2 eik2 L/2 + B2 e−ik2 L/2 = A3 eik1 L/2 + B3 e−ik1 L/2
ik2 A2 eik2 L/2 − ik2 B2 e−ik2 L/2 = ik1 A3 eik1 L/2 − ik1 B3 e−ik1 L/2 .
For confined particles, E < 0, the problem is totally symmetric. In addition, the wavefunction
must disappear for x → ±∞. Therefore, we can simplify,
A1 = 0 = B 3 and A3 = B1 . (3.17)
The first two equations (3.16) now give,
k2
B1 eik1 L/2 = A2 e−ik2 L/2 + B2 eik2 L/2 = −A2 e−ik2 L/2 + B2 eik2 L/2 . (3.18)
k1
We now consider the quotient B2 /A2 . Using the right part of equation (3.18),
B2 e−ik2 L/2 (k2 + k1 ) e−ik2 L (k2 + iκ1 )2

= ik L/2 = . (3.19)
A2 e 2 (k2 − k1 ) k22 + κ21
3.3. POTENTIAL BARRIER 79
Since the amplitudes are real, the imaginary part of the quotient (3.19) should disappear, which
is the case when,
0 = Im e−ik2 L (k2 + iκ1 )2 = 2κ1 k2 cos k2 L + (κ21 − k22 ) sin k2 L (3.20)

2κ1 k2
=⇒ tan k2 L = .
−κ21 + k22
In order to construct graphically the values of the momentap k2 of the particle associated
with the allowed energy levels, we introduce a constant β ≡ ~/(L 2m|V0 |). Hence,
p p
1p 2 |E/V0 | 1 − |E/V0 | 2κ1 k2
tan k2 L = tan 1 − |E/V0 | = = . (3.21)
β 1 − 2|E/V0 | −κ21 + k22
−0.2
E , En (V0)
−0.4
−0.6
−0.8
−1
−10 0 10
2κ1 k2
tan k2 L ,
−κ21 + κ22
Figure 3.4: (Code: QM M otion SquareF inite.m) Graphical solution for a finite bilateral po-
tential well. The red dotted curves represent the tangents (left side of the equation (3.21)),
the solid green curves the hyperbolas (right side of the equation), the circles in cyan are the
eigenenergies. When 0 < E − V0 E, they converge to the eigenenergies of the infinitely deep
well (black crosses and vertical black line).
At the bottom of deep potentials, that is, when 0 < E − V0 E, or equivalently, E ' V0 ,
we have k2 κ1 and hence, tan k2 L → 0 =⇒ k2 L = nπ. The energies are then,
~2 ~2 π 2 2
E − V0 = 2m = n . (3.22)
k22 2mL2
Apply the notions obtained in this section to solve Exc. 3.5.2.3.
3.3 Potential barrier

The linear momentum of a particle described by ψ(x, t) = Aeikx is,
~ d
hψ|p̂|ψi = hψ| |ψi = ~k . (3.23)
i dx
Therefore, this particle propagates towards +∞. On the contrary, the particle Be−ikx propagates
towards −∞. Thus, the two solutions (3.13) of the Schrödinger equation (3.12) correspond to
propagating particle waves. From here on we will use the letter A (B) to denote the amplitudes
of waves propagating in direction ∞ (−∞).
In locations where the potential changes abruptly, the particle can be partially reflected.
3.3.1 T -scattering matrix

As we have already shown in the previous section, we can write the transformation of the
amplitudes due to a potential step at position L as,
A2 eik2 L + B2 e−ik2 L = A1 eik1 L + B1 e−ik1 L (3.24)

ik2 L −ik2 L ik1 L −ik1 L
ik2 A2 e − ik2 B2 e = ik1 A1 e − ik1 B1 e .
We can summarize these two equations in a matrix formalism,

A2 A1
=T , (3.25)
B2 B1
with the scattering matrix T for a particle with energy E (see Fig. 3.2),
 
1 + kk12 ei(k1 −k2 )L 1 − kk21 ei(−k1 −k2 )L
T = 12   (3.26)
1 − kk21 ei(k1 +k2 )L 1 + kk21 ei(−k1 +k2 )L
−ik L !
k1 k1
1 e 2 0 1 + k 1 − k eik1 L 0
=2 2 2 .
0 eik2 L 1 − kk21 1 + kk21 0 e−ik1 L
If there are more zones with different depths, we may concatenate the scattering matrices.
Denoting by Tm→n the scattering matrix describing a transition at position Lm,n of a potential
of the depth Vm to another potential Vn , we write,
T = T2→3 T1→2 . (3.27)
3.3.2 S-scattering matrix

Another common definition is the scattering matrix S,

A2 B2
=S . (3.28)
B1 A1
To see how the scattering matrices are interconnected, we start with

A2 A1 T11 A1 + T12 B1
=T = , (3.29)
B2 B1 T21 A1 + T22 B1
Multiplying the first line with T22 and the second with −T12 and adding them,
T22 A2 − T12 B2 = (T11 T22 − T12 T21 ) A1 . (3.30)
This equation resolved by A2 along with the second equation (3.29) resolved by B1 give,

A2 B2 T12 /T22 T11 − T12 T21 /T22 B2
=S = . (3.31)
B1 A1 1/T22 −T21 /T22 A1
The matrix S describes the causality of scattering process more adequately: The amplitude
A2 in region (2) results from the superposition of a wave B2 being reflected by the barrier and
a wave A1 being transmitted by the barrier. The amplitude B1 in region (1) results from the
superposition of a wave A1 being reflected by the barrier and a wave B2 being transmitted by
the barrier. Therefore, the matrix S is more appropriate for the description of the quantum
reflection, as we will discuss in the next section. However, it has the disadvantage that it can
not be concatenated in the same way as the T matrices.
Unlike the T matrix the S matrix is unitary, since
T11
det S = S11 S22 − S12 S21 = − = −e2ik1 L . (3.32)
T22
Also, it is possible to show,
∗ ∗

S11 S21 S11 S12 1 0
S †S = ∗ ∗ = . (3.33)
S12 S22 S21 S22 0 1
The Exc. 3.5.3.1 asks to calculate the transmission and reflection of a particle at a potential
barrier.
3.3.3 Quantum reflection at a potential step

The quantum reflection is a non-classical property of the motion of a particle. An example is
the reflection of a quantum particle by an attractive potential. To study this effect, we consider
a plane wave eik1 x propagating in region (1) (E1 > V1 ) encountering a potential step up or down
at position x = 0 leading to another region (2). Using the S matrix formalism introduced in the
previous section,
1 k2 − k1 2k1
S= , (3.34)
k1 + k2 2k2 k1 − k2
we find that one part of the wave is reflected into the region (1), another is transmitted into the
region (2),
(1+k1 /k2 )2 −(1−k1 /k2 )2
!
A2 0 T11 − T12 T21 /T22 2(1+k1 /k2 )
=S = = 1−k1 /k2 (3.35)
B1 1 −T21 /T22 − 1+k 1 /k2

1 2k1
= .
k1 + k2 k1 − k2
We use B2 = 0, since no wave comes from the side of region (2), and A1 = 1, because it simplifies
the formulas and does not affect the generality of the results. The interesting results are:
• Even when E −κ2 x

1
p2 < V2 , the particle enters the classically prohibited region: ψ2 (x) ∝ e
with κ2 = ~ 2m(V2 − E2 ), i.e. the transmission is non-zero, |A2 | > 0.
• Even with E2 > V2 , the particle has a probability of being reflected at the step, |B1 | > 0.

Example 15 (Contrast of a partially reflected wave): Defining K± ≡ 21 max |ψ1 |2 ± min |ψ1 |2 ,
the contrast of the wavefunction in region (1) is given by K− /K+ . Writing the function as
ψ1 = eik1 x + B1 e−ik1 x it is easy to show, that
p p
K+ + K− − K+ − K− K−
|B1 | = p p ' . (3.36)
K+ + K− + K+ − K− 2K+
This formula can be understood as an analogue of Fresnel’s formula for matter waves 2 .
In Exc. 3.5.3.2 we calculate the behavior of a Broglie wave passing through a potential step
and entering a classically forbidden region. In Exc. 3.5.3.3 we investigate a model describing the
collision between attracting or repelling particles via a partial reflection at a potential step.
3.3.4 Continuity of probability flow

The continuity equation (2.16) requires that the probability flux be preserved in stationary sit-
uations,
dj d ~ ∗ d d ∗
0= = ψ ψ − ψ ψ . (3.37)
dx dx 2mi dx dx
Applying this to a potential step separating the regions n = 1, 2, we find,

~ ∗ d d ∗
jn = ψ ψ−ψ ψ (3.38)
2mi dx dx
~ h ∗ −ikn x
= (An e + Bn∗ eikn x )(ikn An eikn x − ikn Bn e−ikn x )
2mi i
− (An eikn x + Bn e−ikn x )(−ikn A∗n e−ikn x + ikBn∗ eikx )
~kn
= (|An |2 − |Bn |2 ) .
m
Hence, j1 = j2 implies k1 |A1 |2 − k1 |B1 |2 = k2 |A2 |2 − k2 |B2 |2 . Assuming that the particle comes
from side 1 and B2 = 0, we have,
1 = |B1 |2 + k2
k1 |A2 |
2
=R+T , (3.39)
defining the transmission T and the reflection R as,

2 2
T ≡ k2
k1 |S12 | = k2
k1 |A2 | e R ≡ |S22 |2 = |B1 |2 . (3.40)
3.3.5 Tunneling and quantum reflection at a potential well

Particles thrown with a kinetic energy E against potential barriers can cross them even if V0 > E
or be reflected even when V0 < E. This can be verified by considering a particle propagating
from x = −∞ towards x = +∞ through a potential well located at x ∈ [0, a]. We determine the
concatenation T = T2→3 T1→2 . Then we find the S matrix that corresponds to the T matrix and
solve the problem in the same way as in the previous section. For example, we can calculate the
transmission and reflection probabilities (see Fig. 3.5).
3.3.6 The delta-potential

In quantum mechanics the δ-potential can be used to simulate situations, where a particle is
free to move in two regions of space with a barrier in between. For example, an electron can
move almost freely in a conducting material, but when two conducting surfaces are put close
together, the interface between them acts as a barrier for the electron that can be approximated
by a δ-potential. The δ-potential is a limiting case of the finite potential well when we decrease
2
In this sense light reflection at an optical interface (with typical losses of 4% for glass) can be interpreted as
quantum reflection of light
3 3
2.5 2.5
2 2
E/V0
E/V0
1.5 1.5
1 1 β = 10
β=3
0.5 0.5
0 0
0 0.5 1 0 0.5 1
R,T T
Figure 3.5: (Code: QM Motion Reflection.m) Left: Tunnel effect and quantum reflection at a
potential barrier. Right: Coefficients of transmission and reflection (horizontal) through the
shown potential barrier as a function of the energy normalized√to the height of the barrier E/V0 .
The dashed red curve corresponds to a low barrier, β ≡ ~1 L 2mV0 = 3, the blue solid curve
corresponds to a deep barrier β = 10.
its width while maintained the product of its width and its depth constant. Here, for simplicity,
we only consider a one-dimensional potential well, but the analysis can be expanded to more
dimensions.
The time-independent Schrödinger equation for the wavefunction ψ(x) of a particle in one
dimension is,
~2 d2 ψ
− (x) + αδ(x)ψ(x) = Eψ(x) , (3.41)
2m dx2
The potential is called a δ-potential well if α is negative and a δ-potential barrier if α is positive.
The potential splits the space in two parts (x < 0 and x > 0). In each of these parts the
potential energy is zero, and the Schrödinger equation reduces to,
d2 ψ 2mE
=− 2 ψ . (3.42)
dx2 ~
This is a linear differential equation with constant coefficients, whose solutions are linear
√ com-
binations of eikx and e−ikx , where the wavenumber k is related to the energy by k = 2mE ~ . In
general, due to the presence of the δ-potential in the origin, the coefficients of the solution need
not be the same in both half-spaces:
(
ψ1 (x) = A1 e−ikx + B1 eikx for x < 0
ψ(x) = , (3.43)
ψ2 (x) = A2 e−ikx + B2 eikx for x > 0
where, in the case of positive energies (real k), eikx represents a wave traveling to the right, and
e−ikx one traveling to the left. One obtains a relation between the coefficients by imposing that
the wavefunction be continuous at the origin,
ψ(0) = ψ1 (0) = ψ2 (0) = A1 + B1 = A2 + B2 . (3.44)

A second relation can be found by studying the derivative of the wavefunction. Normally,
we could also impose differentiability at the origin, but this is not possible because of the δ-
potential. However, if we integrate the Schrödinger equation around x = 0 over an interval
[−, +]: Z + Z + Z +
~2
− ψ 00 (x)dx + V (x)ψ(x)dx = E ψ(x)dx . (3.45)
2m − − −
In the limit → 0, the right-hand side of this equation vanishes; the left-hand side becomes,
~2 0
− [ψ (0) − ψ10 (0)] + αψ(0) . (3.46)
2m 2
Substituting the definition of ψ into this expression, we obtain,
~2
− ik(−A1 + B1 + A2 − B2 ) + α(A1 + B1 ) = 0 . (3.47)
2m
The boundary conditions thus give the following restrictions on the coefficients,
A1 + B1 − A2 − B2 = 0
2mα . (3.48)
−A1 + B1 + A2 − B2 = (A1 + B1 )
ik~2
Figure 3.6: (a) The δ-potential (green) and the bound state wavefunction (blue). (b) Double
δ-potential (green).
3.3.6.1 Bound states

The graph of the bound state wavefunction solution to the δ-function potential is continuous
everywhere, but its derivative is not defined at x = 0.
In any one-dimensional attractive
p potential there will be a bound state. To find its energy,
note that for E < 0, k = i 2m|E|/~ = iκ is imaginary and the wavefunctions which were
oscillating for positive energies in the calculation above, are now exponentially increasing or
decreasing functions of x (see above). Requiring that the wave functions do not diverge at
infinity eliminates half of the terms: A1 = B2 = 0. The wavefunction is then,
(
ψ1 (x) = B1 eκx for x < 0
ψ(x) = −κx
. (3.49)
ψ2 (x) = A2 e for x > 0
From the boundary conditions and normalization conditions, it follows that,
√ mα
A2 = B1 = κ and κ=− 2 , (3.50)
~
from which follows that α must be negative, that is the bound state only exists for the well, and
not for the barrier. The Fourier transform of this wavefunction is a Lorentzian function. The
energy of the bound state is then,
~2 κ2 mα2
Eb = − =− 2 . (3.51)
2m 2~
The δ-potential well and its wavefunction are exhibited in Fig. 3.6(a).
3.3.6.2 Scattering
For positive energies, the particle is free to move in either half-space: x < 0 or x > 0, but it may
be scattered at the delta function potential. The quantum case can be studied in the following
situation: a particle incident on the barrier from the left side (B1 ). It may be reflected (A1 ) or
transmitted (B2 ). To find the amplitudes for reflection and transmission for incidence from the
left, we put in the above equations B1 = 1 (incoming particle), A1 = r (reflection), A2 = 0 (no
incoming particle from the right), and B2 = t (transmission), and solve for r and t even though
we do not have any equations in t. The result is,
1 1
t= , r= . (3.52)
mα i~2 k
1− 2 −1
i~ k mα
Due to the mirror symmetry of the model, the amplitudes for incidence from the right are the
same as those from the left. The result is that there is a non-zero probability,
1 1
R = |r|2 = = . (3.53)
~4 k 2 2~2 E
1+ 1+
m2 α 2 mα2
for the particle to be reflected. This does not depend on the sign of λ, that is, a barrier has
the same probability of reflecting the particle as a well. This is a significant difference from
classical mechanics, where the reflection probability would be 1 for the barrier (the particle
simply bounces back), and 0 for the well (the particle passes through the well undisturbed).
The probability for transmission is,
1 1
T = |t|2 = 1 − R = = . (3.54)
m2 α 2 mα2
1+ 4 2 1+ 2
~ k 2~ E
An application example regards the interfaces between two conducting materials. In the bulk
of the materials, the motion of the electrons is quasi-free and can be described by the kinetic
term in the above Hamiltonian with an effective mass m. Often, the surfaces of such materials
are covered with oxide layers or are not ideal for other reasons. This thin, non-conducting layer
may then be modeled by a local δ-function potential. Electrons may then tunnel from one
material to the other giving rise to a current.
Example 16 (Double delta potential ): The δ-function model is actually a one-dimensional

version of the hydrogen atom. The model becomes particularly useful when applied to the
hydrogen molecule ion, as shown in the following.
Figure 3.7: (Code: QM M otion DeltaT unnelling.m) Transmission (red) and reflection (blue)
probability of a δ-potential well. The energy E > 0 is in units of Eb = mα2 /2~2 .
The double-well δ-function models a diatomic hydrogen molecule by the corresponding

Schrödinger equation:
~2 d2 ψ
− (x) + V (x)ψ(x) = Eψ(x) ,
2m dx2
where the potential is now:

V (x) = −qλ δ x + R2 + δ x − R2
where 0 < R < ∞ is the ’internuclear’ distance with δ-function (negative) peaks located at
x = ±R/2 (shown in brown in the diagram). Keeping in mind the relationship of this model
with its three-dimensional molecular counterpart, we use atomic units and set ~ = m = 1.
Here 0 < λ < 1 is a formally adjustable parameter. From the single well case, we can infer
the ’ansatz’ for the solution to be:
ψ(x) = Ae−d|x+ 2 | + Be−d|x− 2 | .

R R
Matching of the wavefunction at the δ-function peaks yields the determinant:

q−d qe−dR d2
= 0 where E = − .
qλe−dR
qλ − d 2
Thus, d is found to be governed by the pseudo-quadratic equation:

n o1/2
d± (λ) = 12 q(λ + 1) ± 1
2 q 2 (1 + λ)2 − 4λq 2 [1 − e−2d± (λ)R ] ,
which has two solutions d = d± . For the case of equal charges (symmetric homonuclear
case), λ = 1 and the pseudo-quadratic reduces to:
d± = q[1 ± e−d± R ] .
The ’+’ case corresponds to a wave function symmetric about the midpoint (shown in red
in the diagram) where A = B and is called gerade. Correspondingly, the ’-’ case is the
wavefunction that is anti-symmetric about the midpoint where A = −B is called ungerade
(shown in green in the diagram). They represent an approximation of the two lowest discrete
energy states of the three-dimensional H+2 and are useful in its analysis. Analytical solutions
for the energy eigenvalues for the case of symmetric charges are given by:
d± = q + W (±qRe−qR )/R ,
3.4. HARMONIC OSCILLATOR 87
where W is the standard Lambert function. Note that the lowest energy corresponds to
the symmetric solution d+ . In the case of unequal charges, and for that matter the three-
dimensional molecular problem, the solutions are given by a generalization of the Lambert
function (see section on generalization of Lambert function and references herein).
One of the most interesting cases is when qR ≤ 1, which results in d− = 0. Thus, one has a
non-trivial bound state solution with E = 0. For these specific parameters, there are many
interesting properties that occur, one of which is the unusual effect that the transmission
coefficient is unity at zero energy.
3.4 Harmonic oscillator

Many systems oscillate. Common examples are vibrations of atoms bound in a molecule or
in a crystalline lattice, of particles trapped in applied electric or magnetic fields, or light in
an electromagnetic mode. Most periodic movements are approximately harmonic for small
amplitude vibrations and can be treated in a way that we will detail now.
We start with the unidimensional harmonic oscillator (OH),

−~2 d2 m 2 2
− + V (x) − E ψ(x) = 0 where V (x) = ω x . (3.55)
2m dx2 2
3.4.1 Factorization of the Hamiltonian and Fock states

i
Respecting the fact that the operators p̂ and x̂ do not commute, ~ [p̂, x̂] = 1, we can rewrite the
Hamiltonian of the harmonic oscillator in the following way,
~2 d2 m
Ĥ = − 2
+ ω 2 x̂2 (3.56)
2m
" dx 2 ! ! #
r r r r
mω 1 mω 1
= ~ω x̂ − i p̂ x̂ + i p̂ + 12 = ~ω â† â + 12 ,
2~ 2m~ω 2~ 2m~ω
p mω q
1
with the abbreviation â ≡ 2~ x̂ + i 2m~ω p̂ and its Hermitian transposition â† . Now let’s try
to find out the properties of the operators â† and â. First of all, the commutator is,
"r r r r #
mω 1 mω 1 i i
[â, â† ] = x̂ + i p̂, x̂ − i p̂ = [x̂ + p̂, x̂ − p̂] = [p̂, x̂] = 1 .
2~ 2m~ω 2~ 2m~ω 2~ ~
(3.57)
Knowing Ĥ|ψi = E|ψi is it clear that â â is an observable with the eigenvalue n ≡ ~ω − 12 ,
† E

â† â|ψi = E
~ω − 1
2 |ψi ≡ n|ψi =⇒ |ψi = |ni . (3.58)
Now, we show that the states â|ψi are eigenstates of the operator defined as n̂ ≡ â† â, since,
â† ââ|ψi = (ââ† − [â, â† ])â|ψi = (ââ† â − â)|ψi = â(â† â − 1)|ψi = (n − 1)â|ψi (3.59)
=⇒ â|ψi ∝ |n − 1i
=⇒ n = hn|â† â|ni = C 2 hn − 1|n − 1i
√
=⇒ C = n .
We note that the quantum number of the new |n − 1i is decreased by 1. Similarly, we show for
the state â† |ψi,
â† ââ† |ψi = â† ([â, â† ] + â† â)|ψi = â† (1 + â† â)|ψi = (n + 1)â† |ψi (3.60)
=⇒ â† |ψi ∝ |n + 1i
=⇒ n + 1 = hn|â† â + [â, â† ]|ni = C 2 hn + 1|n + 1i
√
=⇒ C = n + 1 .
Therefore, this new state is also an eigenvector |n + 1i, with a quantum number increased by
one unit. â† e â are creation and annihilation operators of an energy packet,
√ √
â† |ni = n + 1|n + 1i e â|ni = n|n − 1i . (3.61)
The matrix representation of the field operators is,

X√ X√
â† = n + 1|n + 1ihn| and â† = n|n − 1ihn| . (3.62)
n n
Now it is clear, that n̂ can be understood as a number operator 3 . The energy spectrum of the
harmonic oscillator is equidistant,

En = ~ω n + 21 . (3.63)
The state with n quanta can be created from the vacuum,

â† â†n
|ni = √ |n − 1i = √ |0i . (3.64)
n n!
The state |ni is called number state or Fock state.
3.4.1.1 Uncertainty in Fock states

We consider an OH of mass m and angular frequency ω prepared in the stationary state |ni which
is an eigenstate of the Hamiltonian Ĥ with eigenvalue (n + 12 )~ω. Defining the characteristic
p
size of the OH, aho = ~/mω, the annihilation and creation operators can be written,

1 x̂ aho † 1 x̂ aho
â = √ +i p̂ and â = √ −i p̂ . (3.65)
2 aho ~ 2 aho ~
Therefore, the position and momentum operators are,
√ 1 √ aho
2 x̂ = â + â† and 2i p̂ = â − â† . (3.66)
aho ~
The mean squared deviations of the position x̂ and the momentum p̂ are,
a2ho a2 a2
∆x2 = hn|x̂2 |ni = hn|ââ + ââ† + â† â + â† â† |ni = ho hn|2n̂ + 1|ni = ho (2n + 1) (3.67)
2 2 2
−~2 −~2 ~2
∆p2 = hn|p̂2 |ni = 2 hn|ââ − ââ† − â† â + â† â† |ni = 2 hn| − 2n̂ − 1|ni = 2 (2n + 1) .
2aho 2aho 2aho
(3.68)
3 P
Also, we can define phase operators by ed
xp(∓iφ) = n |n ∓ 1ihn|.
From the results of the previous item we obtain the uncertainty relation ∆x∆p for the OH
in the state |ni,
~
∆p∆x = (2n + 1) . (3.69)
2
Example 17 (Localization energy ): The non-vanishing energy of the fundamental state of
the harmonic oscillator, E0 = ~ω/2, is an immediate consequence of the Heisenberg principle
∆x∆p ≥ ~, because in analogy with Example 14 we calculate,
hp2 i ∆p2 ~2 ~2 ~ω
= > > = .
2m 2m 2m∆x2 2ma2ho 2
In the case of an electromagnetic field this energy is called vacuum fluctuation.
3.4.2 Harmonic oscillator in spatial representation

To simplify the Schrödinger equation in spatial representation,

~2 d2 m 2 2 1
− + ω x ψ(x) = ~ω n + ψ(x) , (3.70)
2m dx2 2 2
p
we use the scale x̃ ≡ x/aho , where aho = ~/mω is the spatial extent of the ground state.
Therefore,

2 ~2 d2 m 2 2 2 ~ω d2 ~ω 2
− + ω (a ho x̃) ψ̃(x̃) = − + x̃ ψ̃(x̃)
~ω 2m d(aho x̃)2 2 ~ω 2 dx̃2 2

d2 2
= − 2 + x̃ ψ̃(x̃) = (2n + 1)ψ̃(x̃) .
dx̃
Now we start looking for asymptotic solutions. For x̃ → ±∞, that is, when the particle
enters the classically forbidden region, we can neglect the total energy of the particle,

d2 2
− 2 + x̃ ψ̃∞ (x̃) ' 0 . (3.71)
dx̃
2
The solution of this equation is ψ̃∞ (x̃) = Ce−x̃ /2 , since

d2 2 d 2 2 2 2 2 2
− 2 + x̃ e−x̃ /2 = − (−x̃)e−x̃ /2 +x̃2 e−x̃ /2 = −x̃2 e−x̃ /2 +e−x̃ /2 +x̃2 e−x̃ /2 = e−x̃ /2 ' 0 .
2
dx̃ dx̃
(3.72)
2
This motivates the ansatz ψ̃(x̃) = e−x̃ /2 H(x̃) for the complete differential equation (3.70),
2 2
d2 −x̃2 /2
2
−x̃2 /2 d H(x̃) de−x̃ /2 dH(x̃) d2 e−x̃ /2 2
− + x̃ 2
e H(x̃) = −e − 2 − H(x̃) + x̃2 e−x̃ /2 H(x̃)
dx̃2 dx̃2 dx̃ dx̃ dx̃2
(3.73)
2
2
d H(x̃) 2 dH(x̃) h 2 2
i 2
= −e−x̃ /2 2
− 2(−x)e−x̃ /2 + −x̃2 e−x̃ /2 + e−x̃ /2 H(x̃) + x̃2 e−x̃ /2 H(x̃)
dx̃ dx̃
2
≡ (2n + 1)e−x̃ /2 H(x̃) .
Thus, the functions H(x̃) must satisfy the differential equation,
H 00 (x̃) = 2x̃H 0 (x̃) − 2nH(x̃) . (3.74)

We can verify that the Hermite polynomials defined by,

2 dn −x̃2
Hn (x̃) = (−1)n ex̃ e , (3.75)
dx̃n
transform the differential equation into a recursion formula,
Hn+1 (x̃) = 2x̃Hn (x̃) − 2nHn−1 (x̃) , (3.76)
which allows us to easily calculate the polynomials,
H0 (x̃) = 1 , H1 (x̃) = 2x , H2 (x̃) = 4x2 − 2 , ... (3.77)
In summary, the eigenfunction of a harmonic oscillator in the state of excitation n is,

2 /2a2
hx|ni = ψn (x) = Ce−x ho Hn (x/aho ) , (3.78)
where the constant C is determined by the normalization condition, hψm |ψn i = δm,n . The
Hermite functions, Hn , are found in mathematical tables. Here we will only show the graphical
representation of |ψ|2 in Fig. 3.8. The Exc. 3.5.4.1 asks to evaluate OH in a classically forbidden
region and in Exc. 3.5.4.2 we will calculate the spectrum of a semi-harmonic OH.
7
5
E/h̄ωho , ψ(x)
0
−5 0 5
x/aho
Figure 3.8: (Code: QM M otion Harmonic.m) Wavefunctions and energies for a rectangular
well.
3.4.3 Properties of the harmonic oscillator

We note that there are regions where ψ(x̃) 6= 0 even though V (x) > E. This effect is purely
quantum. Classically, we can not find a particle in regions where its energy is below the potential.
We also note that for high quantum numbers, n → ∞, we expect to recover the classical
predictions, i.e.,
lim |ψ(x)|2 = PE (x) , (3.79)
n→∞
where PE is the probability density of finding the oscillating particle at position x. The proba-
bility of finding the particle in a range dx close to the location x is easily calculated,
m 2 m 2 2 t(x + dx) − t(x) dx dx 1

E= v + ω x ⇒ PE (x)dx = = = p . (3.80)
2 2 T vT T 2E/m − ω 2 x2
We see that for high energy values the wavefunction approaches the classical expectation.
We already mentioned that there exist solutions only for certain energies En = ~ω(2n + 1).
Consequently, the energy levels are equidistant, En+1 − En = ~ω, as if there were a box into
E
n+3
n+2
n+1
n
Figure 3.9: Ladder of levels.
which we add, one after the other, particles with the energy ~ω until we have accumulated
n portions of energy. These particles are called phonons in the case of vibrations of massive
particles, and photons in the case of a radiation field.
The fact that the energy distribution is the same as the one proposed by Planck for the black-
body radiation suggests the use of the harmonic oscillator to describe the second quantization.
3.4.4 Time evolution of the harmonic oscillator

Here we study the temporal evolution of a population distribution in a harmonic oscillator. The
formal solution of the Schrödinger equation is,
|ψ(t)i = e−iĤt/~ |ψ(0)i . (3.81)
As the Hamiltonian is diagonal in the basis |ni,

1

Ĥ = ~ω n̂ + 2 . (3.82)
we can write, X
e−iĤt/~ = |nie−iωt(n+1/2) hn| . (3.83)
n
P
If the initial state is |ψ(0)i = m cm |mi, the final state and the eigenvalue of any observable
will be,
X X
|ψ(t)i = |nie−iωt(n+1/2) hn|ψ(0)i = e−iωt(n+1/2) cn |ni (3.84)
n n
X X X
hψ(t)|Â|ψ(t)i = hm|eiωt(m+1/2) c∗m |Â| e−iωt(n+1/2) cn |ni = c∗m cn eiωt(m−n) hm|Â|ni .
m n m,n
If the oscillator is initially in an eigenstate, |ψ(0)i = |ki, we obtain,
|ψ(t)i = e−iωt(k+1/2) |ki and hψ(t)|Â|ψ(t)i = hk|Â|ki , (3.85)

that is, the state remains stationary. Motion needs non-diagonal elements of Â.
Another observation is that the populations do not change, even in the case of an initial
superposition, since,
Pk (t) = |hk|ψ(t)i|2 = |e−iωt(k+1/2) ck |2 = |ck |2 . (3.86)
We conclude that
• movement of an observable Â is possible, but only due to variations of the phase factors;
• to carry out transitions between the vibrational states is necessary to perturb the oscillator,
e.g. by applying fields of electromagnetic radiation.
Example 18 (Motion of a harmonic oscillator ): We now consider some specific ex-

amples. If the studied observable is the Hamiltonian and the initial state an arbitrary
superposition, then
X X
hψ(t)|Ĥ|ψ(t)i = ~ω c∗m cn eiωt(m−n) hm|n̂ + 21 |ni = ~ω |cn |2 n + 1
2 .
m,n n
That is, the total energy of the oscillator is the sum of the energies of the states weighted
with the populations of those states. In the case of the position operator,
X
hψ(t)|x̂|ψ(t)i = a√ho
2
c∗m cn eiωt(m−n) hm|â + â† |ni
m,n
X √ √
= a√ho
2
c∗n−1 cn e−iωt n + c∗n+1 cn eiωt n + 1
n
m,n→∞ √ X√
−→ aho 2 n|cn |2 cos ωt .
n
That is, the particle can only oscillate, if there are populations in consecutive states. If this
is not the case, hψ(t)|x̂|ψ(t)i = 0. The oscillation frequency is always ω, independent of
the energy of the particle. The Excs. 3.5.4.3 and 3.5.4.4 analyze the temporal evolution of
oscillators subject to a sudden perturbation.
3.4.5 Multidimensional harmonic oscillator

The 3D harmonic potential is given by
m 2 2 m 2 2 m 2 2
Vho (r) = ω x + ωy y + ωz z . (3.87)
2 x 2 2
Making the ansatz

ψ(r) = ψx (x)ψy (y)ψz (z) , (3.88)
we can separate the spatial directions and obtain a one-dimensional equation for each coordinate,
such that the coordinates can be considered separately. Each function ψk (xk ) is of the form (3.78)
and the energies are,
Ek = ~ωk (nk + 21 ) , (3.89)
where k = x, y, z.
3.4.6 Coherent states

3.4.6.1 Glauber’s Formula
A useful formula that we will use later is the Glauber formula (or Baker-Hausdorff),
eÂ eB̂ = eÂ+B̂+[Â,B̂]/2 , (3.90)
which holds when Â and B̂ commute with their commutator, i.e., [Â, [Â, B̂]] = [B̂, [Â, B̂]] = 0.
Example 19 (The Baker-Hausdorff formula): In order to prove the Baker-Hausdorff

formula, we consider the operator,
Ĝ(τ ) ≡ eτ (Â+B̂) e−τ B̂ e−τ Â .
The derivative is,
Ĝ0 (τ ) = (Â + B̂)eτ (Â+B̂) e−τ B̂ e−τ Â − eτ (Â+B̂) B̂e−τ B̂ e−τ Â − eτ (Â+B̂) e−τ B̂ Âe−τ Â
h i h i
= eτ (Â+B̂) Âe−τ B̂ − e−τ B̂ Â e−τ Â = eτ (Â+B̂) Â − e−τ B̂ Âeτ B̂ e−τ B̂ e−τ Â
h i
= eτ (Â+B̂) Â − Â + [−τ B̂, Â] + 2! 1
[−τ B̂, [−τ B̂, Â]] + .. e−τ B̂ e−τ Â ,
using the formula (2.162). If now [Â, [Â, B̂]] = 0 = [B̂, [Â, B̂]], then,
Ĝ0 (τ ) = eτ (Â+B̂) τ [B̂, Â]e−τ B̂ e−τ Â = −τ [Â, B̂]eτ (Â+B̂) e−τ B̂ e−τ Â = −τ [Â, B̂]Ĝ(τ ) .
The solution of this differential euation is,

2
Ĝ(τ ) ≡ e−(τ /2)[Â,B̂]
Ĝ(0) .
With Ĝ(0) = 1 we obtain at the point τ = 1,
eÂ+B̂ e−B̂ e−Â = e−(1/2)[Â,B̂] .
3.4.6.2 Displacement operator

We now consider the so-called displacement operator 4 ,
† −α∗ â
D̂(α) ≡ eαâ , (3.91)
and try to discovers its features.

D̂(α) is a unitary operator, since using Glauber’s formula, eÂ eB̂ = eÂ+B̂+[Â,B̂]/2 , we get,
∗ â−αâ† † −α∗ â ∗ â−αâ† +αâ† −α∗ â+[α∗ â−αâ† ,αâ† −α∗ â]/2
D̂† (α)D̂(α) = eα eαâ = eα (3.92)
[α∗ â−αâ† ,αâ† −α∗ â]/2 [α∗ â,αâ† ]/2+[−αâ† ,αâ† ]/2+[α∗ â,−α∗ â]/2+[−αâ† ,−α∗ â]/2
=e =e
|α|2 [â,â† ]/2+|α|2 [â† ,â]/2
=e = e0 = 1̂ .
We can rewrite the displacement operator like this:

† −α∗ â † ∗ † ,−α∗ â]/2 † ∗ 2 [â† ,â]/2 † ∗ 2 /2
D̂(α) = eαâ = eαâ e−α â e−[αâ = eαâ e−α â e|α| = eαâ e−α â e−|α| . (3.93)
4
The operator acts on the phase space spanned by the operators â and â† , that is, x̂ ∝ Re â and p̂ ∝ Im â.
The state resulting from the action of the operator D̂(α) onto the estado fundamental of the
HO is,
∞
X
2 /2 † ∗ 2 /2 (αâ† )n
|αi ≡ D̂(α)|0i = e−|α| eαâ e−α â |0i = e−|α||0i (3.94)
n!
n=0

−|α|2 /2 † (αâ† )2 −|α|2 /2 α√ α2 √
=e 1 + αâ + + .. |0i = e |0i + 1|1i + 2!|2i + .. , (3.95)
2! 1! 2!
that is, the state |αi is a superposition distributed according to the Poisson distribution,
X∞
2 /2 αn
|αi = e−|α| √ |ni . (3.96)
n=0 n!
Applying the step-down operator â onto the state |αi, we find,
X∞ X∞ X∞
−|α|2 /2 αn −|α|2 /2 αn √ −|α|2 /2 αn
â|αi = e √ â|ni = e √ n|n − 1i = e p |n − 1i ,
n=0 n! n=0 n! n=0 (n − 1)!
(3.97)
that is,
â|αi = α|αi . (3.98)
We can also write,

hα|â† = (â|αi)† = (α|αi)† = hα|α∗ .
The state |αi is called coherent state or Glauber state 5,6 . We note that, in spite of its appearance,
the equation (3.98) is not an eigenvalue equation, since â is not observable.
Using the formula (2.162), we verify immediately,
D̂† (α)âD̂(α) = â + α , D̂† (α)â† D̂(α) = â† + α∗ . (3.99)
3.4.6.3 Uncertainty in Glauber states

a√ho †
Consider a HO prepared in a state |αi. The eigenvalues of the observables x̂ ≡ 2
(â + â) and
i~√
p̂ ≡ aho 2
(â† − â) are,
√ √
† ∗ iaho 2
2
aho hα|x̂|αi = hα|â + â |αi = α + α e ~ hα|p̂|αi = hα|â − â† |αi = α − α∗ . (3.100)
5
We can also define a Bargmann state as the eigenstate corresponding to the step-up operator using the
notation â† ||αi = α||αi.
6
For the differential calculus with Glauber states it is practical to use the Wirtinger derivative,
∂ ∂
â† − α∗ = e â† − α = .
∂α ∂ α∗
−
→ ←−
With the bosonic operators we can construct the observables,
√
x̂ + ip̂ = 2â , ∂x − i∂p = 2∂α , ∂x + i∂p = 2∂α∗ .
−
→ ←−
−
In two dimensions [87], d2 α = d(Re α)d(Im α) = dxdp.
With this the eigenvalues of the quadratures become,

2
a2ho
hα|x̂2 |αi = hα|(â + â† )2 |αi = hα|ââ + 1 + 2â† â + â† â† |αi (3.101)
= α2 + 1 + 2|α|2 + α∗2 = 1 + (α + α∗ )2 = 1 + 2
a2ho
hα|x̂|αi2
−a2ho 2
~2
hα|p̂2 |αi = hα|(â − â† )2 |αi = hα|ââ − 1 − 2â† â + â† â† |αi
2a2ho
= α2 − 1 − 2|α|2 + α∗2 = −1 + (α − α∗ )2 = −1 − ~2
hα|p̂|αi2 .
The uncertainties defined in (2.64) become,
a2ho ~2
∆x2 = hα|x̂2 |αi − hα|x̂|αi2 = 2 and ∆p2 = hα|p̂2 |αi − hα|p̂|αi2 = 2a2ho
. (3.102)
And finally, we find the Heisenberg relation,

~
∆p∆x = 2 . (3.103)
Comparing with the uncertainty relation (3.69) derived for Fock states, we conclude that the
uncertainty is always smallestfor Glauber states. In this sense, the Glauber states are the ones
which are closest to classical states characterized by the absence of uncertainty.
3.4.6.4 Orthogonality of Glauber states

Glauber are not orthogonal, since,
2
|hα|βi|2 = e−|α−β| . (3.104)
We leave the demonstration for Exc. 3.5.4.5, but we note here already that for |α − β| 0
the states are approximately orthogonal. The reason for this is, that the respective population
distributions through the Fock states, |hn|αi|2 and |hn|βi|2 , do not overlap and hence do not
interfere. The state |αi + | − αi is sometimes called Schrödinger cat state. In Exc. 3.5.4.6 we
will show why such states are very difficult to detect.
3.4.7 Quantization of the electromagnetic field

Historically the quantization of light by Max Planck (also called second quantization) was first.
This second quantization resolved the problem of the ultraviolet divergence and explained the
photoelectric effect. The quantization of the atom by Niels Bohr (also calledfirst quantization
primeira) explained the internal structure of the atom.
The operator for the electric field of a laser mode is given by,
Ê = iEm [âeik·r−iωt − â† e−ik·r+iωt ] , (3.105)
p
where Em = ~ω/2ε0 V and V is the mode volume. Exc. 3.5.4.7 asks to calculate the eigenvalues
hÊi e ∆Ê.
It is sometimes convenient to represent the light field by its quadratures. With the definition
â ≡ x̂1 + ix̂2 , where x̂1,2 are non-commuting operators ([x̂1 , x̂2 ] = i/2), we can write the field as,
Ê = −2Em [x̂1 sin(k · r − ωt) + x̂2 cos(k · r − ωt)] . (3.106)
Heisenberg’s uncertainty relations requires,
1
∆x1 ∆x2 ≥ 4 . (3.107)
For coherent states, ∆x1 = ∆x2 = 21 .
Figure 3.10: Illustration of the Glauber states. Here, x̂1 ≡ â + â† e x̂2 ≡ i(â − â† ).
3.5 Exercises
3.5.1 Translational motion
3.5.1.1 Ex: Trapped particle
Consider the problem of a particle of mass m forced to move in a single direction and completely
confined to a box, with walls placed at the positions x = 0 and x = a. a. The particle be in the
ground state, what is its energy and its wavefunction?
b. Suppose the particle has the following wavefunction:
π π
1 2π
ψI (x) = √ 2 cos (6x − a) − 3i sin x + cos (2x − a) ,
7a 2a a 2a
~ 2 2
what is the probability that a measurement of the energy and yields the result E = 2π ma2
?
c. Considering again the state of item (a) (the ground state), what is the probability distribution
for the momentum of this particle?
d. Still starting from the ground state, suppose we remove (instantaneously) the walls, leaving
the particle free (Ĥ = p̂2 /2m). What is the energy of this free particle?
Formulae:
Z L nπx nπL[1 − (−1)n eiBL ]
eiBx sin dx = para n = 1, 2, 3, ...
0 L n2 π 2 − B 2 L2
Z ∞
x2 2 πx π2
2 2
cos 2 dx =
−∞ (1 − x ) 4
3.5.2 Rectangular potential

3.5.2.1 Ex: Particle in a box
Obtain the wavefunctions and associated energy levels of a particle confined in a box, where
V (x) = 0 for 0 ≤ x ≤ l and V (x) = ∞ outside.
3.5.2.2 Ex: Particle in a two-dimensional box

Obtain the wavefunctions and associated energy levels of a particle trapped in a two-dimensional
box inside which the particle is confined to a rectangular surface with dimensions L1 in x-
direction and L2 in y-direction, V (x, y) = 0 for 0 ≤ x ≤ L1 and 0 ≤ y ≤ L2 and V (x, y) = ∞
else.
3.5. EXERCISES 97
3.5.2.3 Ex: Particle in a well

Obtain the energies of the bound states of a particle in the potential well in which V (x) = ∞
for x < 0, V (x) = −V0 for 0 ≤ x ≤ L/2 and V (x) = 0 to x > L/2. Compare the obtained values
with those of the symmetrical well discussed in Sec. 3.2.4 and the well with infinitely high walls
discussed in Sec. 3.2.1.
3.5.3 Potential barrier

3.5.3.1 Ex: Energy barrier
Consider a particle with energy E thrown (in the direction êx ) against a potential energy barrier
of finite height and width, such that V (x) = 0 for x < 0 or x > L and V (x) = V0 for 0 ≤ x ≤ L.
a. Obtain the reflection and transmission coefficients R and T for the case E > V0 . Discuss the
result.
b. Do the same for the case E < V0 .
3.5.3.2 Ex: Tunneling

A rubidium-87 atom moves in free space (region 0) with velocity v = 1 cm/s (see diagram).
Suddenly it encounters a gap with depth V1 = −kB · 1µK.
a. What is the particle’s Broglie wavelength in region 1?
b. Now the atom encounters a barrier of height V2 = −V1 . What is the probability that the
particle will enter region 2?
c. What is the probability of finding the particle inside region 2 up to a depth of x2 = 10 nm?
Figure 3.11: Particle in a potential landscape.
3.5.3.3 Ex: Collisions

A collision between attractive or repulsive particles can be described by the Schrödinger equation
as a one-dimensional scattering,
~2 00
− ψ (x) + αδ(x)ψ(x) = Eψ(x) .
2m
The energy spectrum may be a discrete spectrum of bound states and a continuum of free states.
a. Calculate the transmission coefficient for the case of a particle with energy E thrown against
the potential energy barrier V (x) = αδ(x). Does the result change for the case when V (x) =
−αδ(x), with α > 0?
b. For this last potential, find the energy of the bound state and its corresponding wavefunction.
3.5.4 Harmonic oscillator

3.5.4.1 Ex: Ground state of a harmonic oscillator
Equating the ground state energy of quantum HO to that of its classical analog, obtain the
maximum elongation xm . Now, knowing that the ground state wavefunction is proportional to
2 2
the Gaussian ψ0 ∝ e−x /2xm , obtain the expression for the probability of finding the HO outside
the classical limits and estimate its value.
3.5.4.2 Ex: Particle in a semi-harmonic well

Find the energy levels of a particle in a potential energy well of the form V (x) = ∞ for x < 0
2 2
and V (x) = mω2 x for x > 0. What is the parity of the allowed states?
3.5.4.3 Ex: Vibration of a harmonic oscillator

Consider Pa HO of mass m and angular frequency ω. At time t = 0 the oscillator’s state is
|ψ(0)i = n cn |ni, where |ni are the stationary states of the HO with energy (n + 1/2)~ω.
a. What is the probability P for measuring, at an arbitrary time t > 0, an energy of the HO
higher than 2~ω? For the case when P = 0, what are the non-zero coefficients cn ?
b. From now on, we assume that only c0 and c1 are nonzero. Write down the normalization
condition for |ψ(0)i and the mean value hĤi of energy in terms of c0 and c1 . With the additional
requirement hĤi = ~ω, calculate |c0 |2 e |c1 |2 .
c. Given that the normalized state vector |ψ(0)i is defined to less than an overall phase factor, we
determine this factorpby choosing the real and positive coefficients c0 and c1 = |c1 |eiθ . Assuming
hĤi = ~ω e hx̂i = 21 ~/mω, calculate θ.
d. With |ψ(0)i determined (according to the previous item), write down |ψ(t)i for t > 0 and
calculate the value θ at this time t. Deduce the average value hx̂i(t) of the position at time t.
3.5.4.4 Ex: Shifted harmonic oscillator

Consider a HO of mass m, angular frequency ω, and electric charge q immersed in a uniform
electric field oriented parallel to the axis êx of the oscillator.
a. Get the energies of the stationary states of the HO and show how to get the corresponding
eigenstates.
b. Calculate the expectation values hxi and hpi for the displaced oscillator now using Glauber
states (or arbitrary superpositions of states) and taking advantage of the formulas (3.66), (3.99),
and (2.162).
c. Now, the electricity field is suddenly turned off. Calculate the time evolution of the oscillator.
Solution: We consider an electron in a harmonic potential suddenly subject to an electric field,

(
p2
Ĥ (1) = 2m +m 2 2
2 ω x + eEx para t < 0 .
Ĥ = p 2
Ĥ = 2m +m 2 2
2ω x para t ≥ 0
eE
a. With the abbreviation b ≡ mω 2
we can rewrite the Hamiltonian
p2 m m
Ĥ (1) = + ω 2 (x − b)2 − ω 2 b2 ,
2m 2 2
3.5. EXERCISES 99
with the obvious solution

m 2 2
Ĥ (1) |ψn(1) i = En(1) |ψn(1) i , |ψn(1) (x)i = |ψn (x − b)i , En(1) = En − ω b .
2
b. With the expansion of Taylor
(−b)2 00
ψn (x − b) = ψn (x) − bψn0 (x) + ψn (x) + .. = e−b(d/dx) ψn (x) ,
2!
b√ (1) †
and the abbreviation β ≡ aho 2
we can rewrite, |ψn (x)i = e−β(â−â ) |ψn i. The expectation value
of â is therefore,
† −â) †
hψn(1) |â|ψn(1) i = hψn |e−β(â âe−β(â−â ) |ψn i .
Using the formula (2.162) we have
† −â) †
e−β(â âe−β(â−â ) = â + β[â − â† , â] = â + β ,
(1) (1)
which verifies the formula postulated in class (3.99). Hence, hψn |â|ψn i = hψn |â + β|ψn i e
(1) (1)
hψn |â† |ψn i = hψn |â† + β|ψn i, tal que,
hψn(1) |x̂|ψn(1) i = b e hψn(1) |p̂|ψn(1) i = 0 .
c. The temporal evolution is given by the time-dependent Schrödinger equation. Since the jump
is finite, the solution must be well behaved in time t = 0,
( (1)
(1) (1) (1)
(1) e−(i/~)Ĥ t ψn (x) = e−(i/~)En t ψn (x) para t < 0
ψn (x, t) = (1) .
e−(i/~)Ĥt ψn (x) para t ≥ 0
For times t ≥ 0 we write,
† †
ψn(1) (x, t) = e−(i/~)Ht e−β(â−â ) ψn (x) = e−(i/~)Ĥt e−β(â−â ) e(i/~)Ĥt e−(i/~)Ĥt ψn (x) .
We now use the relationship

X e−Â B̂ n eÂ X (e−Â B̂eÂ )n −Â B̂eÂ
e−Â eB̂ eÂ = = = ee ,
n
n! n
n!
which is easy to show by expansion of eB̂ . The expression e−Â B̂eÂ can be evaluated through its
action on the complete system of eigenfunctions,
(0) (0) (0)
e−(i/~)Ĥt âe(i/~)Ĥt |ψn i = e−(i/~)Ĥt âe(i/~)En t |ψn i = e−(i/~)En−1 t e(i/~)En t â|ψn i .
Hence,
e−(i/~)Ĥt âe(i/~)Ĥt = âeiωt e e−(i/~)Ĥt â† e(i/~)Ĥt = â† e−iωt .
Now we can write the temporal solution,
iωt −â† e−iωt ) † )−iβ (0)
sin ωt(â+â† ) −(i/~)En t
ψn(1) (x, t) = e−β(âe e−(i/~)Ĥt ψn (x) = e−β cos ωt(â−â e ψn (x) .
Using Glauber’s formula (3.90) we find,

† † 2 (0)
ψn(1) (x, t) = e−iβ sin ωt(â+â ) e−β cos ωt(â−â ) eiβ sin ωt cos ωt −(i/~)En t
e ψn (x)
(0)
sin ωt(â+â† ) iβ 2
= e−iβ e sin ωt cos ωt −(i/~)En t
e ψn (x − b cos ωt) .
Finally,
|ψn(1) (x, t)|2 = |ψn (x − x̄(t))|2 ,
(1)
where x̄(t) ≡ b cos ωt. This means that the spatial distribution of ψn around x̄(t) is the same
as of ψn around x̄ = 0. The entire distribution oscillates without deformation. The momentum
distribution follows from the Fourier transform,
Z Z
1 −(i/~)px (1) 1
(1)
φn (p, t) = √ dxe ψn = √ due−(i/~)pu e−(i/~)umωb sin ωt eiγ(p,t) ψn
2π~ 2π~
= eiγ(p,t) φn (p + mωb sin ωt) ,
with γ = γ ∗ . We obtain,
|φ(1) 2 2
n (p, t)| = |φn (p − p̄(t))| ,
where x̄(t) ≡ −mωb sin ωt.
3.5.4.5 Ex: Harmonic oscillator and coherent states

a. Verify whether the Glauber states of a harmonic oscillator are orthogonal.
b. Show that hα|n̂|αi = |α|2 , hα|n̂2 |αi = |α|4 + |α|2 , and ∆n̂ = |α|.
c. What is the population of the state |ni of a harmonic oscillator in a Glauber state?
3.5.4.6 Ex: Schrödinger cat state

Calculate the probability of finding n photons in Schrödinger’s cat state |ψi = 2−1/2 (|αi±|−αi).
3.5.4.7 Ex: Glauber state

Calculate hÊi and ∆Ê.
Chapter 4
Rotations / Central potentials

4.1 Particle in a central potential
Many potentials do not have Cartesian symmetry, but fortunately, many problems have some
kind of symmetry, cylindrical, spherical or periodic. Those with cylindrical or spherical sym-
metry can be solved by separating the curvilinear coordinates, as we will show in the follow-
ing. Particularly important are spherical potentials caused by central forces, for example, the
Coulomb force between the proton and the electron in the hydrogen atom.
4.1.1 Transformation to relative coordinates

The hydrogen atom represents a two-body problem. We consider the two masses m1,2 of a
proton and an electron separated by a distance r and interacting through a potential V (r). The
Hamiltonian is
−~2 2 −~2 2
Ĥ = ∇r1 + ∇ + V (r1 − r2 ) , (4.1)
2m1 2m2 r2
where r1,2 are the positions of the proton and the electron. With the ansatz Ξ(t, r1 , r2 ) =
Ξ(r1 , r2 )e−iEtot t/~ , the time-dependent Schrödinger equation
d
ĤΞ(t, r1 , r2 ) = i~ Ξ(t, r1 , r2 ) , (4.2)
dt
becomes stationary,

−~2 2 −~2 2
∇ + ∇ + V (r1 − r2 ) Ξ(r1 , r2 ) = Etot Ξ(r1 , r2 ) . (4.3)
2m1 r1 2m2 r2
Now we transform into the center-of-mass system making for the total wavefunction the ansatz
Ξ(r1 , r2 ) = e−iP·R/~ Ψ(r) with R ≡ m m2
M r1 + M r2 and r ≡ r1 −r2 and introducing the abbreviation
1
M = m1 + m2 . This corresponds to a product of a plane wave, describing the linear motion of

the center of the masses, and a radial wave function, which describes the relative motion of the
atom. The kinetic energy of one mass is:
−~2 2 −iP·R/~
∇ e Ψ(r) (4.4)
2m1 r1
−~2 h −iP·R/~ 2 −iP·R/~ 2 −iP·R/~
i
= e ∇r1 Ψ(r) + 2(− im
~M
1P
)e ∇ r1 Ψ(r) + Ψ(r)(− im1
~2 M
P) e
2m1
2
−~ 2 i~P m 1 P2
= e−iP·R/~ ∇r1 Ψ(r) + ∇r1 Ψ(r) − Ψ(r) .
2m1 M 2M 2
101
102 CHAPTER 4. ROTATIONS / CENTRAL POTENTIALS
Hence, for two atoms,
Etot Ξ(r1 , r2 ) − V (r)Ξ(r1 , r2 ) (4.5)

2
−iP·R/~ −~ 2 −~2 2 i~P P2
=e ∇ Ψ(r) + ∇ Ψ(r) + (∇r1 + ∇r2 )Ψ(r) + Ψ(r) .
2m1 r1 2m2 r2 M 2M
Using ∇r1 = −∇r2 = ∇r , we see that the third term cancels, such that,
P2 −~2 2 −~2 2
Ψ(r) + ∇r Ψ(r) + ∇ Ψ(r) + V (r)Ψ(r) = Etot Ψ(r) . (4.6)
2M 2m1 2m2 r
P2
Subtracting the energy of the center-of-mass motion with E = Etot − 2M and introducing the
abbreviation m−1 = m−1 −1
1 + m2 , we finally get,

−~2 2
∇ + V (r) Ψ(r) = EΨ(r) . (4.7)
2m r
4.1.2 Particle in a cylindrical potential

The equation (4.7) is three-dimensional because Ψ(r) is a scalar field and the momentum oper-
ator in Cartesian coordinates is given by,
∂2 ∂2 ∂2
∇2r = 2
+ 2+ 2 . (4.8)
∂x ∂y ∂z
However, in some situations, the symmetry of the system allows to reduce dimensionality sim-
ilarly to the cases of the box potential and the three-dimensional harmonic oscillator. Let us
now discuss the cases of cylindrical and spherical symmetry.
Electrons in magnetic fields are subject to the Lorentz force, which keeps them in a rotating
motion. We can rewrite the momentum operator in cylindrical coordinates,
x = ρ cos ϕ , y = ρ sin ϕ , z=z , (4.9)
as
∂2 1 ∂ 1 ∂2 ∂2
∇2r = + + + . (4.10)
∂ρ2 ρ ∂ρ ρ2 ∂ϕ2 ∂z 2
Now, with the assumption that the potential only depends on ρ, we can try ansatz Ψ(r) =
R(ρ)ξ(ϕ)ζ(z),
2
1 ~2 ∂ 1 ∂ ~2 1 ∂ 2 ~2 1 ∂2
− + + V (ρ) R(ρ)− ζ(z)− ξ(ϕ) = E . (4.11)
R(ρ) 2m ∂ρ2 ρ ∂ρ 2m ζ(z) ∂z 2 2mρ2 ξ(ϕ) ∂ϕ2
First, we separate the axial motion,
ζ 00 2mEz
− = const ≡ , (4.12)
ζ ~2
the solution of this equation being a superposition of two plane waves counterpropagating along
the axis z, ζ(z) = Aeikz z + Be−ikz z . Now, we separate the azimuthal motion,
ρ2 ∂R2 (ρ) ρ ∂R(ρ) 2mρ2 2 2 ξ 00

+ + [E − V (ρ)] − ρ k z = − = const ≡ m2ϕ . (4.13)
R(ρ) ∂ρ2 R(ρ) ∂ρ ~2 ξ
4.1. PARTICLE IN A CENTRAL POTENTIAL 103
The solution of the right-hand part of the equation is ξ(ϕ) = Ceimϕ ϕ + De−imϕ ϕ . Finally, we
have the radial equation,
1 ∂R(ρ)2 1 ∂R(ρ) 2m 2
m2ϕ
+ − [E − V (ρ)] − k z − =0, (4.14)
R(ρ) ∂ρ2 ρR(ρ) ∂ρ ~2 ρ2
~2 m2
with the effective potential Vef f = V (ρ) + 2mρϕ2 . For a homogeneous potential, V (ρ) = V0 , the
solution will be a superposition of Bessel functions.
Example 20 (Rigid rotor in cylindrical coordinates): To give an example, we disregard
the potential, V (ρ) = 0, and we consider for the particle an orbit with constant radius,
ρ = const such that R0 (ρ) = 0. In this case, we only need to treat the orbital motion
described by the right part of Eq. (4.13). For the solution of this equation, ξ(ϕ) = Aeimϕ ϕ ,
to be well-defined, we need ξ(ϕ) = ξ(ϕ + 2π). This implies,
mϕ = 0, ±1, ±2, ..
and
~2 m2ϕ
Eϕ = .
2mρ2
The allowed energies Emϕ = Eϕ can be obtained by letting the Hamiltonian
~2 ∂ 2
Ĥ = − ,
2I ∂ϕ2
with the moment of inertia I = mρ2 actuate on the azimuthal wavefunction ξ(ϕ). We now
define the operator,
ˆlz = ~ ∂ .
i ∂ϕ
This operator acts on the wavefunction ξ as follows,
ˆlz ξ(ϕ) = ~mϕ ξ(ϕ) .
It is easy to show that wavefunctions with different values ml are orthogonal.

Note: 1. The state mϕ = 0 has zero energy; that is, it has no zero-point energy. 2. The
particle is delocalized within a ring of radius r: ∆lz ∆ sin ϕ ≥ ~2 |hcos ϕi|.
4.1.3 Hamiltonian in spherical coordinates

We can rewrite the momentum operator in spherical coordinates,
x = r sin ϑ cos ϕ , y = r sin ϑ sin ϕ , z = r cos ϑ , (4.15)
as

1 ∂ 2 ∂ 1 L̂2 L̂2 1 ∂ ∂ 1 ∂2
∇2r = 2 r + 2 2 where ≡ sin ϑ + , (4.16)
r ∂r ∂r r ~ ~ 2 sin ϑ ∂ϑ ∂ϑ sin2 ϑ ∂ϕ2
is an abbreviation called Legendre operator. For an isotropic potential, V (r) = V (r), we can try
the ansatz,
Ψ(r) = R(r)Y (ϑ, ϕ) (4.17)
to solve the Schrödinger equation (2.54),

r2 ~2 1 ∂ 2 ∂ −1 L̂2 Y (ϑ, ϕ) ~2
− r + V (r) − E R(r) = = const ≡ − `(`+1) , (4.18)
R(r) 2m r2 ∂r ∂r 2m Y (ϑ, ϕ) 2m
where we choose a separation constant, `(` + 1), the significance of which we shall soon learn.
Considering only the angular part,
L̂2 Y (ϑ, ϕ) = ~2 `(` + 1)Y (ϑ, ϕ) , (4.19)
and making another separation ansatz,
Y (ϑ, ϕ) = Θ(ϑ)Φ(ϕ) , (4.20)
we obtain,

2 1 1 ∂ ∂ 1 ∂2
sin ϑ sin ϑ Θ(ϑ) + `(` + 1) = − Φ(ϕ) = const ≡ m2 , (4.21)
Θ(ϑ) sin ϑ ∂ϑ ∂ϑ Φ(ϕ) ∂ϕ2
where we choose a separation constant, m2 . Introducing another abbreviation,

~ ∂
L̂z ≡ , (4.22)
i ∂ϕ
the azimuthal equation takes the form
L̂z Φ(ϕ) = ~mΦ(ϕ) . (4.23)
As in the case of the cylindrical potential, the solution of the azimuthal equation is, using the
normalization,
Φ(ϕ) = √12π eimϕ , (4.24)
with the magnetic quantum number m = 0, ±1, ±2, ...
The polar equation,
1 1 ∂ ∂ m2
sin ϑ Θ(ϑ) + `(` + 1) = , (4.25)
Θ(ϑ) sin ϑ ∂ϑ ∂ϑ sin2 ϑ
is called Legendre’s differential equation and can be solved by a power series in cosk ϑ. For
m = 0, the solutions are the Legendre polynomials, P` (cos ϑ) with
1 d`
P` (z) = [(z 2 − 1)` ] . (4.26)
2` `! dz `
The first polynomials are,
P0 (z) = 1 , P1 (z) = z , P2 (z) = 12 (3z 2 − 1) , P3 (z) = 12 (5z 3 − 3z) . (4.27)
For m > 0, the solutions are the associated polynomials,
dm (−1)m 2 m/2 d
`+m
P`m (z) = (−1)m (1 − z 2 )m/2 P ` (z) = (1 − z ) [(z 2 − 1)` ] (4.28)
dz m 2` `! dz `+m
(` − m)! m
P`−m (z) = (−1)m P (z) .
(` + m)! `
The polar function must still be normalized,
s
2` + 1 (` − m)!
Θm m
` (ϑ) = P` (cos ϑ) . (4.29)
2 (` + m)!
4.1. PARTICLE IN A CENTRAL POTENTIAL 105
l=0 l=1
90 1 90 1
120 60 120 60
150 0.5 30 150 0.5 30
180 0 180 0
210 330 210 330
240 300 240 300

270 270
l=2 l=3
90 1 90 1
120 60 120 60
150 0.5 30 150 0.5 30
180 0 180 0
210 330 210 330
240 300 240 300

270 270
Figure 4.1: (Code: QM Rotation Legendre.m) Angular wavefunctions. Shown are the Legendre
polynomials Plm (cos ϑ) for ` = 0, 1, 2, 3 and m = 0, .., `. Red: m = 0, green: |m| = 1, blue:
|m| = 2, and magenta: |m| = 3.
The functions Y`m (ϑ, ϕ) are the spherical harmonics. They form an orthonormal system,
Z π Z 2π
Y`∗0 m0 (ϑ, ϕ)Y`m (ϑ, ϕ) sin ϑdϑdϕ = δ`0 ` δm0 m . (4.30)
0 0
Finite solutions only exist when the angular momentum quantum number is ` = 0, 1, .. and for
|m| ≤ `.
The solutions of the angular part of the Schrödinger equation for the hydrogen atom are
finally,
s
1 m 2` + 1 (` − m)! imϕ
Y`m (ϑ, ϕ) = √ P` (cos ϑ) e . (4.31)
2π 2 (` + m)!
The spherical harmonics are simultaneously eigenfunctions of the operators L2 , as can be seen
from Eq. (4.19), and of the operator Lz according to Eq. (4.23). The quantities represented by
the quantum operators Ĥ, L̂2 , L̂z are conserved in the hydrogen system. The conservation of
the angular momentum is due to the spherical symmetry of the Coulomb potential.
We will verify the parity of the spherical harmonics in Exc. 4.5.1.1.
4.1.4 Separation of radial motion

In Sec. 4.1.3 we derived, after having separated the motion of the center-of-mass (that is, of
the heavy nucleus) and the angular coordinates, the radial equation (4.18) describing the radial
component of the electronic motion,

1 ~2 1 ∂ 2 ∂ L2
− r + V (r) − E R(r) = − , (4.32)
R(r) 2m r2 ∂r ∂r 2mr2
Now, we make the substitution R(r) = u(r)/r and the radial equation becomes,

~2 ∂ 2 L2
− + + V (r) u(r) = Eu(r) . (4.33)
2m ∂r2 2mr2
This equation is very similar to a one-dimensional Schrödinger equation, but there is an addi-
tional potential term called centrifugal potential,
L̂2
V` (r) ≡ . (4.34)
2mr2
For example, for the potential of an electron orbiting a proton, we have,

~2 ∂ 2 Ze2 ~2 `(` + 1)
− − + − E uE` (r) = 0 . (4.35)
2m ∂r2 4π0 r 2mr2
We will discuss this equation intensely in the context of the hydrogen atom.
V(r)
0 5 10
r/aB
Figure 4.2: (Code: QM Rotation Centrif ugal.m) Sum of a Coulomb potential and centrifugal
potential for ` = 0 (lower curve), ` = 1 (center curve), and ` = 2 (upper curve).
In Exc. 4.5.1.2 we derive the radial Gross-Pitaevskii equation for a Bose-Einstein condensate
trapped in a spherical potential. In the Exc. 4.5.1.3 we will study particles inside a central
potential of zero depth, in the Excs. 4.5.1.4 and 4.5.1.5 we consider 3D spherical box potentials
and in Exc. 4.5.1.6 a spherical harmonic potential.
Example 21 (Rigid rotor in spherical coordinates): We continue the discussion of the

rigid rotor, now in spherical coordinates. In the case that the orbit of the particle is fixed
to a radius R, we can neglect the kinetic energy due to the radial motion and the potential,
both being constant. In this case the radial Schrödinger equation is,
2
~ `(` + 1)
uE` = E` uE` .
2mr2
The energies of the rigid rotator are
~2 `(` + 1)
E` = ,
2I
with the momentum of inertia I = mR2 .
4.2 Quantum treatment of hydrogen

According to Rutherford’s and Bohr’s planetary atomic model we may imagine an atom as
a very heavy nucleus having a positive electric charge surrounded by a very light negatively
charged charge electronic cloud. Since the nucleus is very small compared to the electronic
4.2. QUANTUM TREATMENT OF HYDROGEN 107
cloud, we treat it as an entity with mass M and charge Ze, where Z is the number of protons
and corresponds to the order of the element in the periodic system.
The canonical procedure for calculating all properties of an atom is to establish its Hamil-
tonian, that is, to determine the kinetic energies of all components and all interaction energies
between them, and to solve the Schrödinger equation. For each component we write the kinetic
energy,
XZ
P2 p2i
Tncl = and Tele = . (4.36)
2M 2m
i=1
Here, (R, P) are the nuclear coordinates and (ri , pi ) those of the electrons. The energies that
corresponds to the interactions, that is, Coulombian attraction or repulsion, between the com-
ponents of the atom are,
Z
X Z
X
Ze2 e2
Vncl−ele = − and Vele−ele = . (4.37)
4πε0 |R − ri | 4πε0 |ri − rj |
i=1 i6=j=1
There are also interactions due to the spin of the particles, which we will deal with later.
Obviously, the solution to this many-body problem is very complicated. For this reason, we
will in this chapter, based on the Schrödinger equation, calculate the complete spectrum of the
simplest possible atom, hydrogen. This atom consists of a proton and an electron, only.
e-
Ze -
(Z-1)e-
Figure 4.3: The hydrogen model applies to other atoms having a single valence electron occupying
a sufficiently large space, that it sees the nucleus together with rest of the electrons shielding
the nucleus as a single positive charge.
4.2.1 Bohr’s model

Let us now turn our attention to the radial part of the Schrödinger equation describing a particle
in a radial potential. We expect that the quantum solutions for the hydrogen atom are similar
to the predictions of Bohr’s model. Following this model, the orbit is stable when the attraction
force is equal to the centrifugal force. But in addition, Bohr postulated, that only certain
energies are allowed. For the hydrogen atom he found,
1 Ze2 1 Z 2 ~2 1 Z 2 e2 1 Z2
En = − =− = − = − 13.6 eV , (4.38)
2 4πε0 rn 2ma2B n2 4πε0 2aB n2 n2
with the Bohr radius

~2
aB ≡ 4πε0 . (4.39)
me2
With this equation he was able to explain the spectral observations. Electrons can only jump
from one level to another, while emitting or absorbing a photon. The series observed in the
Paschen
Brackett
Ballmer
Lyman
abgd
Figure 4.4: The hydrogen transitions.
hydrogen spectrum (En − Em )/~ are the Lyman (m = 1), the Balmer (m = 2), the Paschen
(m = 3) an the Brackett series (m = 4).
The discussion of the hydrogen atom within quantum mechanics can start from the radial
Schrödinger equation (4.35) with the Coulomb attraction potential,

~2 ∂ 2 Ze2 ~2 `(` + 1)
− − + − E uE` (r) = 0 . (4.40)
2m ∂r2 4πε0 r 2mr2
In order to facilitate comparison with Bohr’s classical model, let us express the energy in terms
of Bohr’s energy, E ≡ En = E1 /n2 , and write the radius in units of aB , that is, r̃ ≡ Zr/aB .
This yields,

00 `(` + 1) 2 1
un,` (r̃) + − + − 2 un,` (r̃) = 0 . (4.41)
r̃2 r̃ n
To ensure that for large radii, r → ∞, the solution is finite, we need an asymptotic behavior
like un,` (r̃ → ∞) = e−r̃/n . To ensure that for small radii, r → 0, the solution is finite, we
need un,` (r̃ → 0) = r̃`+1 . We derive the asymptotic solutions in Exc. 4.5.2.1. The resulting
differential equation only has solutions for an integer and positive main quantum number n
and when ` = 0, 1, .., n − 1. That is, in the relation E = E1 /n2 the parameter n is integer
and positive, such that energy levels remain degenerate in ` and m. This means that Bohr’s
postulate of discrete (i.e. quantized) energy levels is valid (uff!)
Substituting the ansatz,
un` (r̃) = Dn` r̃`+1 e−r̃/n L(r̃) , (4.42)
it’s easy to show (see Exc. 4.5.2.2), that the differential equation (4.41) reduces to,

r̃L00 (r̃) + 2 (` + 1) − n1 r̃ L0 (r̃) + 2 1 − n1 (` + 1) L(r̃) = 0 . (4.43)
Still with the abbreviation ρ ≡ 2r̃/n = 2Zr/naB the ansatz
un` (ρ) = Dn` ρ`+1 e−ρ/2 L(ρ) , (4.44)

Figure 4.5: Level scheme.
leads to the differential equation 1
ρL00 (ρ) + [2(` + 1) − ρ] L0 (ρ) + [n − ` − 1]L(ρ) = 0 . (4.45)
(2`+1)
The solutions of this differential equation, Ln−`−1 (r̃), are the Laguerre polynomials. These
polynomials are listed in mathematical tables. Using the properties of these polynomials it is
possible to show that the radial functions are orthogonal and can be normalized (see Exc. 4.5.2.3).
Fig. 4.6 shows the curves for the lowest orbitals.
Finally, we can write the total solutions,
un,` (r) ~2 Z 2
ψn,`,m (r, θ, φ) = Y`,m (θ, φ) with En = − , (4.46)
r 2ma2B n2
where n = 1, 2, 3, .. e ` = 0, 1, .., n − 1 e m = −`, −` + 1, .., `. Of course, each energy level n is,
n−1
X
(2` + 1) = n2 (4.47)
`=0
times degenerate.
1
Laguerre’s associated differential equation is,
ρ∂ρ2 Lν(α) + (α + 1 − ρ)∂ρ L(α)

ν + νL(α)
ν =0.
The Laguerre polynomials are generated by
eρ ρ−α dν
Lν(α) (ρ) = e−ρ ρν+α .

α! dρν
0.6 0.05
0.5 〈 r10 〉
0.04
0.4
2
aB3/2 R(r)
0.03
a [rR(r)]
0.3
0.2
B
0.02 〈 r20 〉
0.1
〈 r30 〉
0.01
0
−0.1 0
0 5 10 15 20 25 0 5 10 15 20 25
r/aB r/aB
0.12 0.025
〈 r32 〉
0.1
0.02
0.08
2
aB3/2 R(r)
0.015
aB [rR(r)]
0.06
〈 r31 〉
0.04 0.01
〈 r30 〉
0.02
0.005
0
−0.02 0
0 5 10 15 20 25 0 5 10 15 20 25
r/aB r/aB
Figure 4.6: (Code: QM Rotation Laguerre.m) Radial wavefunctions, left R; right u. Top for
(n, `) = (1..3, 0); bottom (n, `) = (3, 0..2).
Here is a list of the first wavefunctions of the hydrogen atom,
3/2
ψ100 = √1
π
Z
aB e−r̃ (4.48)
3/2
ψ200 = √1
4 2π
Z
aB (2 − r̃)e−r̃/2
3/2
ψ210 = √1
4 2π
Z
aB r̃e−r̃/2 cos θ
3/2
ψ21±1 = √1
8 2π
Z
aB r̃e−r̃/2 sin θe±iϕ
3/2
ψ300 = √1
81 3π
Z
aB (27 − 18r̃ + 2r̃2 )e−r̃
√ 3/2
ψ31±1 = √2
81 3π
Z
aB (6 − r̃)r̃e−r̃/3 sin θe±iϕ
3/2
ψ320 = 1
√
81 6π
Z
aB r̃2 e−r̃/3 (3 cos2 θ − 1) ,
where we use the abbreviation r̃ ≡ Zr/aB . Using these wavefunctions we can now calculate
important eigenvalues such as, for example,

h1in`m = 1 (4.49)

2 1 `(` + 1)
hr̃in`m =n 1+ 1−
2 n2
" !#
3 `(` + 1) − 31
hr̃2 in`m 4
=n 1+ 1−
2 n2

6 35 35 15 3
hr̃3 in`m =n − 2 − 2 (` + 2)(` − 1) + 4 (` + 2)(` + 1)`(` − 1)
8 8n 4n 8n

63 35 5 12
hr̃4 in`m = n8 + 2 (2`2 + 2` − 3) + 4 5`(` + 1)(3`2 + 3` − 10) + 8
8 8n 8n n

1 1
= 2
r̃ n
n`m
1 1
= 3
r̃2 n`m n (` + 12 )

1 n
= 4
r̃3 n`m n `(` + 12 )(` + 1)
3 2 1
1 2 n − 2 `(` + 1)
= .
r̃4 n`m n5 (` + 32 )(` + 1)(` + 21 )`(` − 12 )
These results will become important later. In Exc. 4.5.2.4 we will calculate the eigenvalue hri
for several orbitals |Ψn`m i.
4.2.2 The virial theorem

Originally derived for classical mechanics, the virial theorem also holds for quantum mechanics,
as shown for the first time by Vladimir Aleksandrovich Fock. We evaluate the commutator
between the Hamiltonian
Ĥ = p̂2 /2m + V (r̂) , (4.50)
and the product of the position operator r̂ with the momentum operator p̂ = −i~∇ of the
particle:
p̂2
[Ĥ, r̂ · p̂] = [Ĥ, r̂] · p̂ + r̂ · [Ĥ, p̂] = −i~ + i~r̂ · ∇V , (4.51)
m
using the theorems of Ehrenfest. Therefore, we find for the operator Q̂ = r̂ · p̂ the commutator,
i
[Ĥ, Q̂] = 2Ekin − r̂ · ∇V . (4.52)
~
The left side of this equation is precisely −dQ̂/dt, following the Heisenberg equation of motion.
The eigenvalue hdQ̂/dti of the temporal derivative vanishes in steady state, therefore we obtain
the virial theorem,
2hEkin i = hr̂ · ∇V i . (4.53)
Example 22 (Virial theorem applied to a central potential ): For example, for a
central potential V (r) ∝ rs we obtain,
∂V
2hEkin i = hr̂ · êr i = shV i .
∂r
In Exc. 4.5.2.5 we calculate the eigenvalues hr−1 i and hp2 i and we verify the virial theorem.
Finally, in Exc. 4.5.2.6 we calculate transition matrix elements between different orbitals.
4.3 Angular momentum

4.3.1 The orbital angular momentum operator
The definition of orbital angular momentum is adopted from classical mechanics:

êx êy êz

ˆ = −i~ x y z .
l̂ = r̂ × p̂ = −i~r̂ × ∇ (4.54)

∂x ∂y ∂z
To better understand the properties of the angular momentum operator in quantum mechanics
we will derive in the Excs. 4.5.3.1 and 4.5.3.2 some of its properties.
Figure 4.7: Illustration of angular momentum in quantum mechanics.
4.3.1.1 Constants of motion

The preceding chapter dealt with the resolution of the radial and angular equations for the
case of a radial potential. The radial equation allowed to calculate the eigenenergies of the
Hamiltonian Ĥ,
Ĥ|ψi = En` |ψi . (4.55)
We also found the common eigenvalues and eigenfunctions of operators l̂2 and ˆlz [see Eqs. (4.19)
and (4.23)]. We now use the notation |`, mi ≡ Y`m (θ, φ) for the eigenfunctions,
l̂2 |`, mi = ~2 `(` + 1)|`, mi and ˆlz |`, mi = ~m|`, mi . (4.56)
With this we have,
[Ĥ, ˆlz ]|ψi = Ĥ~m|ψi − ˆlz E|ψi = 0 and [Ĥ, l̂2 ]|ψi = Ĥ~2 `(` + 1)|ψi − l̂2 E|ψi = 0 . (4.57)
Therefore, the operators l̂2 and ˆlz are constants of motion,
[Ĥ, ˆlz ] = 0 = [Ĥ, l̂2 ] . (4.58)
Exc. 4.5.3.3 asks to show explicitly, at the example of an isotropic three-dimensional harmonic
oscillator, that l̂2 and ˆlz are constants of motion.
4.3. ANGULAR MOMENTUM 113
4.3.2 SU(2) algebra of angular momentum and spin

So far, we have solved the angular eigenvalue equation in the spatial representation for an orbital
angular momentum, l̂ = r̂ × p̂. But it is not clear, whether every angular momentum has this
representation, which is derived from classical notions. In fact, we will see that the electron
has an intrinsic spin with no orbiting charges. What we must show now is that for any spin ĵ
satisfying
ĵ × ĵ = i~ĵ , (4.59)
or [ĵm , ĵn ] = i~kmn jk using the Levi-Civita symbol, we obtain a consistent algebra.
Since ĵ2 and ĵz commute (we show this from Eq. (4.59) in Exc. 4.5.3.4), they have common
eigenfunctions |j, mi. We can write the eigenvalues as,
ĵ2 |j, mi = ~2 j(j + 1)|j, mi and ĵz |j, mi = ~m|j, mi , (4.60)
where, for now, we only know that m is real and j ≥ 0. But since hj, m|ĵ2 |j, mi ≥ hj, m|ĵz2 |j, mi,
it is clear that j(j + 1) ≥ m2 .
4.3.2.1 Creation and annihilation operator

Now we introduce the rising operator ĵ+ and the lowering operador ĵ− via
†
ĵ± ≡ ĵx ± iĵy such that ĵ− = ĵ+ .
It is easy to check the following relationships
[ĵz , ĵ± ] = ±~ĵ± and [ĵ2 , ĵ± ] = 0 and ĵ∓ ĵ± = ĵ2 − ĵz2 ∓ ~ĵz . (4.61)
With this we find
ĵz ĵ± |j, mi = ([ĵz , ĵ± ] + ĵ± jz )|j, mi = ~(m ± 1)ĵ± |j, mi (4.62)
ĵ2 j± |j, mi = ĵ± ĵ2 |j, mi = ~2 j(j + 1)ĵ± |j, mi .
That is, ĵ± |j, mi is a eigenstate of ĵ2 and ĵz with the eigenvalues j and m ± 1, respectively, if
j± |j, mi =
6 0. Hence,
ĵ+ |j, mi ∝ |j, m + 1i . (4.63)
In order not to violate the condition m2 ≤ j(j + 1), we need to fix ĵ± |j, ±ji = 0. Therefore, for
a specified j, the m can have only one of the 2j + 1 possible values m = −j, −j + 1, .., j. Since
2j + 1 is an integer, j can only have values j = 0, 12 , 1, 32 , ... Thus, the eigenvalue equation of the
observables ĵ2 , ĵ is solved since we could have chosen instead of ĵz any one of the components of
ĵ, knowing that the others do not commute with the chosen one.
All spin components ĵz and the scalar ĵ2 can only have discrete eigenvalues. The smallest
unit is ~/2. With the normalization hj, m|j 0 , m0 i = δj,j 0 δm,m0 we have,
hj, m|ĵ∓ ĵ± |j, mi = hj, m|(ĵ2 − ĵz2 ∓ ~ĵz )|j, mi = ~2 [j(j + 1) − m(m ± 1)] , (4.64)
and p
ĵ± |j, mi = ~ j(j + 1) − m(m ± 1)|j, m ± 1i . (4.65)
In Exc. 4.5.3.5 we calculate the uncertainty of the angular momentum components, in Exc. 4.5.3.6
we write the operator ĵx in a matrix form, and in Exc. 4.5.3.7 we calculate projections of the
spin of the electron in different directions of the quantization axis.
4.3.3 The electron spin

Every angular momentum l̂ generates a dipole moment µ ~ ` ∝ l̂, which interacts with external
magnetic fields, V (B) = µ ~ ` · B. Inhomogeneous magnetic fields exert forces on dipole moments,
F = −∇(~ µ` · B), which are detected by the Stern-Gerlach experiment. This experiment reveals
not only the quantization of angular momentum, but also the presence of semi-integral values
for the magnetic quantum number.
In 1925 Uhlenbeck and Goudsmit proposed that the electron could have an intrinsic angular
momentum with the quantum number s = 1/2. This angular momentum, called spin, would
not correspond to any orbiting mass or charge distribution within the classical radius of the
electron of the type l = r × p. The spin is a purely quantum phenomenon because it disappears
when ~ → 0. It is believed nowadays that the electron is actually point-like with no detectable
deviation from Coulomb’s law at any distance. The spin of the electron does not follow from
the Schrödinger equation, but can be included, ad hoc. On the other hand, it is interesting that
it is a necessary consequence of the stringent relativistic derivation of quantum mechanics by
Paul Dirac.
To characterize the spin, we can use the whole SU(2) formalism of the quantum mechanics
of angular momentum:
ŝ × ŝ = i~ŝ , (4.66)
and
ŝ2 | 21 , ± 21 i = ~2 34 | 21 , ± 12 i , ŝz | 12 , ± 21 i = ± ~2 | 21 , ± 12 i , ŝ± = σ̂± = | 12 , ± 12 ih| 12 , ∓ 12 | . (4.67)
The operators σ̂± are the Pauli spin matrices.
4.4 Coupling of angular momenta

4.4.1 Singlet and triplet states with two electrons
In this section we first consider the spin states of two electrons, which can be combined into
two groups with well-defined total spin. With this we can understand the energy spectrum
of helium, which is very much dominated by Pauli’s principle and quantum statistics. The
introduced concepts can be extended to atoms with many electrons.
Angular momentum is an important quantum number in the treatment of the internal struc-
ture of atoms. The two electrons in the helium electronic shell each contribute a spin of S = 12 ,
which couple to a total angular momentum. Let us consider, for simplicity, two free electrons.
The state of the two-particle system is an element of the product space of the two Hilbert spaces
in which the individual electrons are described. We will now apply the formalism of Sec. 2.3.8
explicitly to a pair of electrons. The states that the two electrons can occupy are:
 
1
1 1  0
|γ1 i = ⊗ =

 ≡ | ↑↑i , |γ2 i = | ↑↓i , |γ3 i = | ↓↑i , |γ4 i = | ↓↓i . (4.68)
0 0 0
0
The Pauli matrices act on the spin of the individual electrons a and b. They can be extended
4.4. COUPLING OF ANGULAR MOMENTA 115
to the product Hilbert space as follows,

~ ~ 0 I2 ~ ~ σx 0
2 σ̂x ⊗ I2 = 2 I 0
, 2 I2 ⊗ σ̂x = 2 0 σ̂x
(4.69)
2

~ ~ 0 iI2 ~ ~ σy 0
2 σ̂y ⊗ I2 = 2 −iI 0
, 2 I2 ⊗ σ̂y = 2 0 σ̂y
2

~ ~ I2 0 ~ ~ σz 0
σ̂
2 z ⊗ I2 = , I
2 2 ⊗ σ̂z = .
2 0 I2 2 0 σ̂z
With these operators we can now build other operators. We first consider the three components
of the total angular momentum,
Ŝk = ~2 (σ̂k ⊗ I2 + I2 ⊗ σ̂k ) such that (4.70)
     
0 1 1 0 0 −1 −1 0 1 0 0 0
 1 0 0 1 1 0 0 −1 0 0 0 0
Ŝx = ~2 
1
 , Ŝy = i~   , Ŝz = ~   .
0 0 1 2 1 0 0 −1 0 0 0 0
0 1 1 0 0 1 1 0 0 0 0 −1
The operator for the square of the absolute value of the total angular momentum is calculated
as follows:  
2 0 0 0
 0 1 1 0
Ŝ2 = Ŝx2 + Ŝy2 + Ŝz2 = ~2 
0 1 1
 . (4.71)
0
0 0 0 2
Now we look for the eigenvalues of the total angular momentum. The equation for the eigenvalues
of Sz ,
Ŝz |γk i = MS |γk i , (4.72)
is already diagonal in the introduced basis {γk } with the eigenvalues,
MS = ~, 0, 0, −~ . (4.73)
For Ŝ2 the situation is more interesting: The states |γ1 i and |γ4 i are eigenstates of S2 for the
eigenvalue 2~2 , but the states |γ2 i and |γ3 i are not eigenstates. On the other hand, we know
that the linear combination of two eigenstates with the same eigenvalue is also a eigenstate.
Therefore, the states
|γa i = √1 (|γ2 i − |γ3 i) and |γs i = √1 (|γ2 i + |γ3 i) , (4.74)

2 2
are eigenstates of Ŝz , but they also are eigenstates of Ŝ2 , since we can easily verify,
Ŝ2 |γs i = 2~2 |γs i and Ŝ2 |γa i = 0~2 |γa i , (4.75)
using the matrices (4.70). In summary, for the eigenvalue hŜ2 i = 2~2 there exist the following
three states: 
|γ1 i ms = 1 
|γ4 i ms = −1 triplet , S = 1 (4.76)

|γs i ms = 0
For hS2 i = 0 there is only one state:
|γa i ms = 0 singlet , S = 0 . (4.77)
By exchanging the two electrons, the vectors |γ1 i and |γ4 i retain their shape, while the mixed
vectors change their shape: γ2 ←→ γ3 . Under particle exchange |γa i reverses its sign, that is, it
is antisymmetric, while |γ1 i, |γ4 i and |γc i conserve their signs, thas is, they are symmetrical.
In summary, the triplet states have the quantum number of the total angular momentum
(with the expected value for Ŝ2 of ~2 S(S + 1) = 2~2 ), and they are symmetrical about the
exchange of particles. The singlet state has the quantum number of the total angular momentum
S = 0, and it is antisymmetric about the exchange of particles.
A similar treatment can be done with bosons, as will be discussed in Sec. 9.1.
4.4.2 Coupling two spins

We now consider a perturbation of the system which, for some reason, only affects the first spin.
In the absence of the second atom we would have,

0 Ω
Ĥ1 = . (4.78)
Ω∗ 0
Including the second atom,

 
Ω
 Ω
Ĥ = Ĥ1 ⊗ I = 
Ω∗
 .
 (4.79)
Ω∗
In this case, the perturbation Hamiltonian does not commute with the total angular momentum,
[Ŝ2 , Ĥ] 6= 0] . (4.80)
Another type of perturbation affects both spin symmetrically (e.g., superradiance Dicke with
two atoms in the same radiative mode or two counterpropagating modes in a ring cavity). The
interaction Hamiltonian is now the sum of the individual perturbations,
 
Ω Ω
Ω∗ Ω
Ĥ = Ĥ1 ⊗ I + I ⊗ Ĥ1 = 
Ω∗
 . (4.81)
Ω
Ω∗ Ω∗
This Hamiltonian commutes with the total angular momentum,
[Ŝ2 , Ĥ] = 0 . (4.82)
S now it’s a good quantum number. Singlet states do not couple with triplets. This is the idea
behind Dicke’s superradiance. The absolute value of the total angular momentum is conserved.
The quantum number S is called Dicke cooperatividade [74].
In Sec. ?? we will discuss the coupling of two counterpropagating modes in a ring cavity. In
Sec. ?? we will discuss Dicke states.
4.4.2.1 Two atoms interacting through their dipole moments
As an example of a system exhibiting coupling of the type described in (4.81) we consider

two atoms j = 1, 2 interacting through their dipole moments. First, without interaction the
Hamiltonian will be,
Ĥ = ~ω0 σ̂1z ⊗ I + ~ω0 I ⊗ σ̂2z . (4.83)
P
We note the rules σ̂jz = 2σ̂j+ σ̂j− − 1, σ z = j σjz , and σ̂ 2 = (σ̂ x )2 + (σ̂ y )2 + (σ̂ z )2 .
Now, the atoms interact via their dipole moments with an electromagnetic field assumed to
be the same for both atoms,
Ĥint = −A(r1 ) · (êx σ̂1x ⊗ I + êy σ̂1y ⊗ I) − A(r2 ) · (êx I ⊗ σ̂2x + êy I ⊗ σ̂2y ) , (4.84)
with r1 ' r2 . By the rules (4.69) we find in matrix notation a Hamiltonian equivalent to (4.81),
 
0 Ax − iAy Ax − iAy 0
Ax + iAy 0 0 Ax − iAy 
Ĥ = 
Ax + iAy
 . (4.85)
0 0 Ax − iAy 
0 Ax + iAy Ax + iAy 0
±
Note, that the Hamiltonian is Hermitian, although it consists of sums of operators σ̂1,2 . Thus,
the coupling is inherently different from a beamsplitter-type couplings generated by terms ex-
pressions like σ̂1+ σ̂2− .
Using MAPLE we find the eigenvector and eigenvalue matrices,
   q 
Ax −iAy A −iA Ax −iAy
Ax +iAy − Axx +iAyy 0 −2 A2x + A2y 0 0 0
 q qAx +iAy   
− Ax −iAy 0 −1
Ax −iAy 
 
  0 0 0 0
U=  q Ax +iAy q Ax +iAy  and Ê = 

.

− Ax −iAy Ax −iAy 
 0 0 0 q 0 
 Ax +iAy 0 1 Ax +iAy 
1 1 0 1 0 0 0 2 A2x + A2y
(4.86)
Now we get to the evolution of the various states,
 q q 
−A +iA
  − Axx+iAyy sin 2t A2x + A2y    
0  q  0 0
 2 + A2 
   cos 2t A   1 1
 
iÊt −1 1 x y
Ue U   =  
q 
 and U eiÊt U −1  
−1 = −1 .
1  cos 2t A2x + A2y 
0 q q  0 0
Ax +iAy 2 + A2
−Ax +iAy sin 2t A x y
(4.87)
4.4.2.2 Dicke states
For 2J atoms,
p
hJ, M − 1|σ − |J, M i = J(J + 1) − M (M + 1) . (4.88)
For 2J = 2 atoms,
I = |h1, M − 1|σ − |1, M i|2 = 2 − M (M + 1) . (4.89)
Obviously, for M = 1 the transition is not possible, whereas for M = −1, 0 the I rate is doubled.
J is called cooperation number,
| + +i = |1, 1i (4.90)
| + −i = |1, 0i
| − +i = |1, 0i
| − −i = |1, −1i .
Consider superposition states,
√1 (| + +i + | − −i) = |1, 0i (4.91)

2
1
√ (|
2
+ +i − | − −i) = |0, 0i
1
√ (|
2
+ −i + | + −i) = |1, 0i
1
√ (|
2
+ −i − | + −i) = |0, 0i .
4.4.3 Decoupled and coupled bases

Electrically charged orbiting particles produce a magnetic field. This field can influence the
motion of other particles. In the same way, the spin of an electron can influence its own orbital
motion. That is, angular momenta can couple and interact in a complicated way. Even to
describe the behavior of an atom as simple as hydrogen in an external field, we need to construct
the eigenstates of the total angular momentum resulting from a coupling of the electron’s intrinsic
spin and its orbital motion.
On the other side, we have hitherto considered predominantly hydrogen and hydrogen-like
atoms, that is, atoms with a nucleus and a single (valence) electron. But in fact atoms can have
more than 100 electrons, which complicates the exact description. In atoms with many electrons,
one of the most common coupling schemes P is when the angular momenta of all electrons
P couple
to a total angular momentum, L̂ = k l̂k , which then couples to the total spin, Ŝ = k ŝk , to
form the complete total angular momentum, Ĵ = L̂ + Ŝ. We generally assign total momenta to
capital letters.
Adopting an unbiased notation we study some properties of the total angular momentum,
ĵ ≡ ĵ1 + ĵ2 . In Exc. 4.5.4.1 we find that the addition of angular momenta produces a quantity
which is also an angular momentum, but not the subtraction.
The angular momenta of two particles or two angular momenta of different origins in a single
particle represent independent degrees of freedom, [j1 , j2 ] = 0. Without interaction between
angular momenta the Hilbert spaces are orthogonal:

H1 0
H1 ⊗ H 2 = . (4.92)
0 H2
The eigenfunctions act on a space of dimension, dim H1 + dim H2 :
|j1 , mj1 ; j2 , mj2 i . (4.93)
That is, there is a complete set of commuting operators {ĵ21 , ĵ1z , ĵ22 , ĵ2z }. Therefore, we can
specify quantum numbers j1 , j2 , mj1 , and mj2 simultaneously. On the other hand, the group
{ĵ21 , ĵ22 , ĵ2 , ĵz } also represents a complete set of commuting operators, as shown in Exc. 4.5.4.2.
It has the basis
|(j1 , j2 )j, mj i . (4.94)
In Exc. 4.5.4.3 we derive the matrix representation of two spins in the decoupled and the coupled
base.
To describe the two angular momenta simultaneously we must opt between the decoupled
picture |j1 , mj1 ; j2 , mj1 i and the coupled picture |(j1 , j2 )j, mj i. For now, the choice of the
picture makes no difference, but we will see later that there may be an energy associated with
the coupling 2 . In this case, as we will show, that the choice of the coupled base is more natural,
because the energy commutes like [Ĥ, ĵ2 ] = 0 = [Ĥ, ĵz ], but [Ĥ, ĵ21 ] 6= 0 6= [Ĥ, ĵ22 ].
Figure 4.8: Illustration of the coupling of two angular momenta.
4.4.3.1 Allowed values of total angular momentum

Since we do not specify an interaction energy between the spins or between spins and external
fields, all states are energetically degenerate. In the decoupled image the degeneracy is easily
calculated,
Xj1 j2
X
#= 1 = (2j1 + 1)(2j2 + 1) . (4.95)
mj1 =−j1 mj2 =−j2
Now we need to find the possible values of j and mj in the coupled picture. The values of
mj follow immediately from ĵ1 + ĵ2 = ĵ,
mj = mj1 + mj2 . (4.96)
With |mj1 | ≤ j1 and |mj2 | ≤ j2 the values of mj are limited to
|mj | ≤ j1 + j2 . (4.97)
We often know the two angular momenta j1 and j2 and all their projections in the decoupled
base,
|mj1 | ≤ j1 and |mj2 | ≤ j2 . (4.98)
To find the quantum numbers in the coupled base, we arrange the states ordering them by their
total magnetic quantum number mj . We can, without loosing generality, concentrate on the
situation j1 ≥ j2 . The following table reproduces the possible combinations. The x-symbols
represent Clebsch-Gordan coefficients.
2
That is, the Hamiltonian of the system does not contain terms of type ĵ1 · ĵ2 , but may have terms proportional
to ĵ1 + ĵ2 .
j j j j−1 j j−1 j−2 j j−1 j

mj1 +mj2 = mj j j−1 j−1 j−2 j−2 j−2 −j + 1 −j + 1 −j
j1 j2 x
j1 j2 − 1 x x
j1 − 1 j2 x x
j1 j2 − 2 x x x
j1 − 1 j2 − 1 x x x
j1 − 2 j2 x x x
..
.
−j1 + 1 −j2 x x
−j1 j2 + 1 x x
−j1 −j2 x
The possible values for j are all those allowing for j ≥ |mj | = |mj1 + mj2 |, that is,
|j1 − j2 | ≤ j ≤ j1 + j2 . (4.99)
Each value of j has the degeneracy 2j + 1. Therefore, as will be verified in Exc. 4.5.4.4, the total
degeneracy is,
jX
1 +j2
2j + 1 = (2j1 + 1)(2j2 + 1) . (4.100)

j=|j1 −j2 |
Example 23 (L · S coupling ): As an example we consider two electrons. The first electron

has s1 = 21 and `1 = 0, the second has s2 = 21 and `2 = 1. As illustrated in Fig. 4.9, the
coupling first gives S = s1 + s2 = 0, 1 and L = `1 + `2 = 0, 1. Then we determine the possible
values of the total angular momentum J = L + S = 0, 1, 2 depending on the values of L and
S.
Figure 4.9: Spin-orbit coupling L · S for two electrons.
4.4.4 Clebsch-Gordan coefficients

For now, we will only describe how to add two angular momenta, ĵ1 and ĵ2 . Since they act on
different degrees of freedom,
α1 · ĵ1 , α
[~ ~ 2 · ĵ2 ] = 0 (4.101)
for arbitrary vectors α ~ k . We have a system of common eigenvectors, |η, j1 , j2 , m1 , m2 i, where η
are the eigenvalues of other observables commuting with ĵ1 and ĵ2 . These eigenvectors give the
values ~2 j1 (j1 + 1) and ~2 j2 (j2 + 1) for the observables ĵ21 and ĵ22 , as well as ~m1 and ~m2 for the
observables jz1 and jz2 . The number of states is (2j1 + 1)(2j2 + 1). Now we want to construct
the eigenstates of the total angular momentum ĵ = ĵ1 + ĵ2 . Since
[ĵ, ĵ21 ] = 0 = [ĵ, ĵ22 ] , (4.102)

there exist common eigenstates |j1 , j2 , j, mi for the set of observables ĵ21 , ĵ22 , ĵ2 and ĵz . These
eigenstates are linear combinations of the individual states,
X
|(j1 , j2 )j, mi = |j1 , j2 , m1 , m2 ihj1 , j2 , m1 , m2 |(j1 , j2 )j, mi (4.103)
m1 ,m2

X j1 j2 j
= |j1 , j2 , m1 , m2 i .
m1 m2 m
m1 ,m2
The matrix coefficient is called Clebsch-Gordan coefficient. The Clebsch-Gordans disappear

when the conditions 3
|j1 − j2 | ≤ j ≤ j1 + j2 and m = −j1 − j2 , −j1 − j2 + 1, .., j1 + j2 (4.104)
are not satisfied.

The unitary transformation matrices between decoupled and coupled bases,
|(j1 , j2 )j, mi = UCGC |j1 , m1 ; j2 , m2 i , (4.105)
are listed in tables of the Clebsch-Gordan coefficients.
Example 24 (Clebsch-Gordans for the coupling of two spins 21 ): For example, for
the system consisting of two 12 spins we have,
 
 1 1  1 q0 0 0 1 1 1 1 
|( 2 , 2 )1, +1i q
 1 1  |2, +2; 2, +2i
 |( 1 , 1 )1, 0i  0 2 2 0 | 1 , − 1 ; 1 , + 1 i
 12 21   q q   21 21 12 12  .
 |( , )0, 0i  =  1 1  | , + ; , − i
2 2
1 1
 0 2 − 2 0  21 21 12 12
|( 2 , 2 )1, −1i |2, −2; 2, −2i
0 0 0 1
In the Excs. 4.5.4.5 and 4.5.4.6 we write all possible states of two angular momenta in
decoupled and coupled bases. In Excs. 4.5.4.7, 4.5.4.8, 4.5.4.9, 4.5.4.10, and 4.5.4.11 we practice
the transformation between decoupled and coupled bases, and in Exc. 4.5.4.12 we verify a rule
guaranteeing the unitarity of the Clebsch-Gordan transformation.
4.4.4.1 Coupling of three angular moments

Three angular momenta can be coupled in three different configurations: First j1 with j2 ,
then the total spin (j1 , j2 )j12 with the third one j3 . We use the notation |[(j1 , j2 )j12 , j3 ]Ji or
|[(j1 , j3 )j13 , j2 ]Ji or |[(j2 , j3 )j23 , j1 ]Ji. The recoupling of three spins
j1 + j2 = j12
+ +
j3 j3 (4.106)
= =
j13 + j2 = J
is described by {6j}-symbols, for example,

X
|[(j1 , j2 )j12 , j3 ]Ji = {6j}|[(j1 , j3 )j13 , j2 ]Ji . (4.107)
j13
3
The Clebsch-Gordans are related to the (3j) de Wigner symbols.
4.4.4.2 Notation for atomic states with LS-coupling

P
In an atom, the spins of the electrons often couple to a total spin, S =P k sk , and separately
the orbital angular momenta to a total orbital angular momentum, L = k lk . These two total
spins now couple to a total angular momentum, J = L + S. When this LS-coupling happens,
the following notation is used to characterize the electronic states in atoms:
2S+1
LJ . (4.108)
4.4.4.3 jj-coupling
There is also the case that for each electron its spin couples to its own orbital angular P
momentum,
jk = lk + sk , before coupling to the total angular momenta of other electrons, J = k jk . This
is called jj-coupling. In the case of two electrons the recoupling of the four involved spins
l1 + l2 = L
+ + +
s1 + s2 = S (4.109)
= = =
j1 + j2 = J
 
 l1 l2 L
is described by {9j} = s1 s2 S -symbols,
 
j1 j2 J
X
|[(l1 , s2 )j1 , (l2 , s2 )j2 ]Ji = {9j}|[(l1 , l2 )L, (s1 , s2 )S]Ji . (4.110)
L,S
4.5 Exercises
4.5.1 Particle in a central potential
4.5.1.1 Ex: Parity of the spherical harmonic functions
P
We consider the parity transformation P with (x, y, z) −→ (−x, −y, −z). Use spherical coordi-
P
nates to show that Y`m −→ (−1)` Y`,m , and therefore that a spherical surface function has even
parity when ` is even, and odd parity, when ` is odd.
4.5.1.2 Ex: Bose-Einstein condensate in an isotropic potential
The Gross-Pitaevskii equation describing the wavefunction of a Bose-Einstein condensate is,

∂ψ(r) ~2 2
i~ = − ∆ + Vtrp (r) + g|ψ(r)| ψ(r) ,
∂t 2m
where the factor g depends on the force of the interatomic interaction and Vtrp is the potential
trapping the atoms. For V (r) = V (r) the wave function will have radial symmetry, ψ(r) = φ(r)
r .
Rewrite the Gross-Pitaevskii equation for the function φ.
4.5. EXERCISES 123
4.5.1.3 Ex: Motion of a free particle in spherical coordinates

Obtain the eigenfunctions of a free particle as the limiting case of its motion in a central force
field with V (r) −→ 0. Compare the derived eigenfunctions - associated to the complete set of
observables H, L2 and Lz - to those described by plane waves - associated with the motion
characterized by the observables px , py , pz and Ĥ = P2 /2m -, which also constitute a complete
set of observables.
4.5.1.4 Ex: Particle in a spherical box

Find the energy levels and wavefunctions of a particle confined in a spherical box described by
potential energy, V (r) = 0 for r < a and V (r) = ∞ for r ≥ a considering the angular momentum
` = 0.
4.5.1.5 Ex: Finite spherical 3D potential well

a. Derive the possible energy levels and associated wavefunctions for a particle trapped in a
spherical 3D potential well of depth V0 and radius a. Note that this problem is analogous to
Mie scattering of scalar waves.
b. Discuss the case of a well surrounded by infinitely high walls.
4.5.1.6 Ex: Particle in a spherical harmonic potential

A quantum particle of mass m is subject to a potential
V = 12 mω 2 (x2 + y 2 + z 2 ) .
a. Obtain the energy levels of this particle. That is, determine the eigenvalues of
~2 2
− ∇ ψ + V ψ = Eψ .
2m
b. Consider the fundamental level and the first two excited levels. Set up a table showing for
each of these three levels the energy value, the degeneracy, and the respective states in terms of
the quantum numbers.
c. Using
2 1 ∂ 2 ∂ L2
∇ ψ= 2 r − 2 2 ψ
r ∂r ∂r ~ r
and remembering L̂2 Y`m (θ, φ) = ~2 `(` + 1)Y`m , write down the differential equation of item (a)
for the radial part of the wavefunction (it is not necessary to solve it). Identify in this equation
the effective potential Vef f (r).
d. Solve the differential equation of the previous item for the case where ` = 0 and determine
2
the corresponding eigenvalue. To do this, allow for a solution of the type e−αr and determine
α.
4.5.2 Quantum treatment of hydrogen

4.5.2.1 Ex: Asymptotes of Laguerre’s polynomials
Derive the asymptotic solutions of equation (4.41).
4.5.2.2 Ex: Laguerre equation

Show that the equation (4.41) transforms with the ansatz (4.42) into equation (4.43).
4.5.2.3 Ex: Laguerre functions

Using the orthogonality relation of associated Laguerre polynomials,
Z ∞
Γ(n + α + 1)
ρα e−ρ L(α) (α)
n (ρ)Lm (ρ)dρ = δn,m
0 n!
Z ∞
(n + α)!
ρα+1 e−ρ L(α) 2
n ρ dρ = (2n + α + 1) ,
0 n!
and the recursion formula,

(α+1) (α)
nL(α+1)
n (ρ) = (n − ρ)Ln−1 (ρr) + (n + α)Ln−1 (ρ)
(α)
ρL(α+1)
n (ρ) = (n + α)Ln−1 (ρ) − (n − ρ)L(α)
n (ρ) ,
a. calculate the normalization constant Dn,l for a hydrogen-like atom with atomic number Z;
b. calculate the mean value

n2 aB 1 `(` + 1)
hrinlm = 1+ 1− ;
Z 2 n2
c. calculate the mean value

1 Z
= 2 .
r n`m n aB
4.5.2.4 Ex: Orbital radii in Bohr’s model

Using the results of 4.5.2.3, obtain the expectation values hri for the states Ψ100 , Ψ210 and Ψ320
of the hydrogen atom. Compare the results with those of Bohr’s model.
4.5.2.5 Ex: The virial theorem and Bohr’s model

Calculate, for the state Ψ320 of the hydrogen atom, the expectation values h 1r i and hp2 i.
From the results, obtain the expectation values for the kinetic and potential energies, hT i and
hV i, and show that, consistent with the virial theorem, hT i = −(1/2)hV i. Compare the results
with Bohr’s model.
4.5.2.6 Ex: Transition matrix elements

Using the following (non-normalized) wavefunctions of hydrogen, ψ100 (r) = e−r̃ , ψ210 (r) =
r̃e−r̃/2 cos θ e ψ21±1 (r) = r̃e−r̃/2 sin θe±iφ , calculate the matrix elements (a) hψ100 |z̃|ψ210 i, (b) hψ100 |z̃|ψ211 i,
(c) hψ100 |x̃ − iỹ|ψ210 i, and (d) hψ100 |x̃ − iỹ|ψ211 i using the formulae:
Z ∞ Z π Z π Z π
4 −3x/2 3 2
x e 256
dx = 81 , sin xdx = 3 , 4
cos x sin xdx = 0 , cos2 x sin xdx = 23 .
0 0 0 0
Try to interpret the results.

4.5. EXERCISES 125
4.5.3 Angular momentum

4.5.3.1 Ex: Properties of the angular orbital momentum
Show that l̂ × l̂ = i~l̂ and [ˆlx , ˆly ] = i~ˆlz .
4.5.3.2 Ex: Levi-Civita tensor

Demonstrate [lk , rm ] = i~rn εkmn where the Levi-Civita tensor is defined by εkmn = 1 when
(kmn) is an even permutation of (123), εkmn = −1 for an odd permutation, and εkmn = 0 when
two of the indices are equal.
4.5.3.3 Ex: Angular orbital momentum of a harmonic oscillator

Show for an isotropic three-dimensional harmonic oscillator: [Ĥ, l̂2 ] = [Ĥ, ˆlz ] = 0. Make explicit
calculations, that is, show
h 2 i hm i h 2 i hm i
p ˆ 2 2 ˆ p 2 2 2 2
2m , lz = 0 = ω r , lz e 2m , l̂ = 0 = ω r , l̂ .
2 2
4.5.3.4 Ex: Commutation of the absolute value and the components of the orbital
angular momentum
Show [ĵ2 , ĵ] = 0.
4.5.3.5 Ex: Uncertainty of angular momentum components

Shows that if ĵz is precise, then ĵx and ĵy are imprecise.
4.5.3.6 Ex: Matrix representation of the components of the angular momentum

Calculate the matrix elements of ĵx and ĵx2 in the base where ĵz is observable.
4.5.3.7 Ex: Spin-1/2-particle in a magnetic field

Consider a spin-1/2-particle whose magnetic moment is µ ~ = γS (where γ is a constant). We can
describe the quantum state of this particle in terms of the space generated by the eigenvectors
|+i and |−i of the operator Ŝz , which measures the spin projection in z-direction:
Ŝz |+i = ~2 |+i , Ŝz |−i = − ~2 |−i
Initially (t = 0) the particle is in the state ψ(t = 0)i = |+i and is subject to a uniform magnetic
field B = Bêy , so that:
Ĥ = −~µ · B = −γB Ŝy .
a. What are the possible measurements of the spin projection on the y-axis?
b. Find the eigenvectors of Ŝy .
c. Get |ψ(t)i at t > 0 in terms of the eigenvectors |+i and |−i defined above.
d. Obtain the mean expectation values of the observables Sx , Sy and Sz as a function of time.
4.5.4 Coupling of angular momenta

4.5.4.1 Ex: Addition/subtraction of angular momenta
Show that ĵ1 + ĵ2 is an angular momentum, but not ĵ1 − ĵ2 .
4.5.4.2 Ex: CSCO for coupled angular momenta

Be ĵ = ĵ1 + ĵ2 . Show that {ĵ21 , ĵ22 , ĵ2 , ĵz } is a CSCO; that is, show that
a. ĵ2 commutes with ĵ21 e ĵ22 ;
b. ĵ2 does not commute with ĵ1z or ĵ2z and that we can not specify mj1 or mj2 together with j.
4.5.4.3 Ex: Spin-orbit coupling

a. Show that the operator L · S associated with the spin-orbit coupling, satisfies the relation
L · S = Lz Sz + (L+ S− + L− S+ )/2.
Obtain the matrix representation of the operator L · S, considering the bases:
b. {|mL i ⊗ |mS i} of the eigenstates which are common to the operators L2 , S2 , Lz , Sz ;
c. {|J, M i}, which is associated with the operators L2 , S2 , J2 , Jz .
d. Give the explicit matrices for the case L = 1 and S = 21 in the representations (b) and (c)
and verify that the two representations yield the same eigenvalue spectrum.
4.5.4.4 Ex: Multiplicity of coupled angular momenta

Verify # = (2j1 + 1)(2j2 + 1) within the coupled representation.
4.5.4.5 Ex: Possible states of two (de-)coupled angular momenta

Find all possible states with the angular momenta j1 = 1 and j2 = 1/2 in decoupled and coupled
pictures.
4.5.4.6 Ex: Fine and hyperfine structure of the rubidium atom 85 Rb
1. The rubidium atom 85 Rb has one valence electron. In the first excited state this electron has
the orbital angular momentum, L = 1. What are the possible states?
2. In the fundamental state of this atom the total electronic angular momentum J couples with
the spin of the nucleus, I = 5/2, to form the total angular momentum F = J + I. Determine
the possible values for the angular momentum F and the magnetic quantum number mF .
4.5.4.7 Ex: Expansion of the hyperfine structure of the rubidium atom 87 Rb
Determine for the states S1/2 and P3/2 of an atom with nuclear spin I = 3/2 and hyperfine
coupling Ĵ · Î how the eigenstates of the coupled base expand into the decoupled base. Do not
consider external magnetic fields.
4.5.4.8 Ex: Transition amplitudes between Zeeman sub-states

a. We consider the atom of 87 Rb having the nuclear angular momentum I = 3/2. What are the
possible hyperfine states F resulting from a coupling of I with the total electronic state angular
momentum of the ground state 2 S1/2 ? What are the possible Zeeman sub-states of F ?
4.5. EXERCISES 127
b. What are the possible hyperfine states F 0 resulting from a coupling of I with the total elec-
tronic angular momentum of the excited state 2 P3/2 , F 0 = 2? What are the possible Zeeman
sub-states of F 0 ?
c. A transition between a ground hyperfine state and an excited hyperfine state can be described
by a coupling of the total angular momentum F with the angular momentum of the photon κ
forming the angular momentum of the excited state F 0 . To see this, we now consider the levels
F = 1 and F 0 = 2. Expand the coupled angular momentum |(F, κ)F 0 , mF 0 i = |(1, 1)2, mF 0 i
on a decoupled basis for every possible value m0F . Use the table in Fig. 4.10 to determine the
Clebsch-Gordan coefficients.
Note: The Clebsch-Gordans only compare the oscillator strengths of transitions between Zee-
man sub-states of a given set (F, F 0 ). In order to compare the oscillator strengths to other
transitions (F, F 0 ) it is necessary to calculate 6j-coefficients.
4.5.4.9 Ex: Expansion of the spin-orbit coupling

Consider the problem of adding the orbital angular momentum ` and a spin 1/2. Obtain the
2l + 1 states |` + 1/2, mj i, in addition to the 2l states |` − 1/2, mj i (which constitute a common
basis for the operators `21 , s22 , j2 , jz ), expanded in the base |m1 , m2 i of the eigenstates of the
operators l2 , s2 , `z , sz . You can simplify the procedure by deriving two recurrence relationships
from which the desired states follow 4 .
4.5.4.10 Ex: Gymnastics of angular momentum operators

Consider the problem of adding angular momenta j1 = 1 and j2 = 1/2:
a. What are the possible values of m and j, in which ĵ 2 |j, mi = j(j + 1)~2 |j, mi and jz |j, mi =
m~|j, mi?
b. What are the degeneracy gj1 ,j2 (m)?
c. Find the base states {|j, mi}, which are common to the operators j21 , j22 , j, jz , expanded in
the base {|j1 , m1 i ⊗ |j2 , m2 i} of the eigenstates of j21 , j22 , j1z , j2z .
4.5.4.11 Ex: (Un-)coupled bases of the spherical harmonics

Expand the triplet state 3 PJ of strontium in a decoupled basis and write down the transformation
matrix between the bases.
4.5.4.12 Ex: Properties of Clebsch-Gordan coefficients

Given
P the momenta j1 and j2 , and Cm1 ,m2 denoting the Clebsch-Gordan coefficients, prove that
2
m1 ,m2 |Cm1 ,m2 | = 1.
4
See Cohen-Tannoudji, Vol.2, Complement A X.
36. Clebsch-Gordan coefficients 1
12836. CLEBSCH-GORDAN COEFFICIENTS,

CHAPTER 4. ROTATIONS / CENTRAL
SPHERICAL POTENTIALS
HARMONICS,
AND d FUNCTIONS
p
Note: A square-root sign is to be understood over every coefficient, e.g., for −8/15 read − 8/15.
r
3
Y10 = cos θ
4π
r
3
Y11 =− sin θ eiφ
8π
r ³
5 3 1´
Y20 = cos2 θ −
4π 2 2
r
15
Y21 =− sin θ cos θ eiφ
8π
r
1 15
Y22 = sin2 θ e2iφ
4 2π
Yℓ−m = (−1)m Yℓm∗ r hj1 j2 m1 m2 |j1 j2 JM i

4π
d ℓm,0 = Y m e−imφ = (−1)J−j1 −j2 hj2 j1 m2 m1 |j2 j1 JM i
2ℓ + 1 ℓ
j ′ j j
d m′ ,m = (−1)m−m d m,m′ = d −m,−m′ 1/2 θ 1 + cos θ
d 10,0 = cos θ d 1/2,1/2 = cos d 11,1 =
2 2
1/2 θ sin θ
d 1/2,−1/2 = − sin d 11,0 = − √
2 2
1 − cos θ
d 11,−1 =
2
3/2 1 + cos θ θ
d 3/2,3/2 = cos
2 2
√ 1 + cos θ ³ 1 + cos θ ´2
3/2 θ d 22,2 =
d 3/2,1/2 = − 3 sin 2
2 2
√ 1 − cos θ θ 1 + cos θ
3/2
d 3/2,−1/2 = 3 cos d 22,1 =− sin θ
2 2 2
√ 1 + cos θ
3/2 1 − cos θ θ 6 d 21,1 = (2 cos θ − 1)
d 3/2,−3/2 = − sin d 22,0 = sin2 θ 2
2 2 4 r
3/2 3 cos θ − 1 θ 1 − cos θ 3
d 1/2,1/2 = cos d 22,−1 = − sin θ d 21,0 = − sin θ cos θ
2 2 2 2
3 cos θ + 1 θ ³ 1 − cos θ ´2 1 − cos θ ³3 1´
3/2
d 1/2,−1/2 = − sin d 22,−2 = d 21,−1 = (2 cos θ + 1) d 20,0 = cos2 θ −
2 2 2 2 2 2
Figure 4.10: Clebsch-Gordan coefficients.

Figure 36.1: The sign convention is that of Wigner (Group Theory, Academic Press, New York, 1959), also used by Condon and Shortley (The
Theory of Atomic Spectra, Cambridge Univ. Press, New York, 1953), Rose (Elementary Theory of Angular Momentum, Wiley, New York, 1957),
and Cohen (Tables of the Clebsch-Gordan Coefficients, North American Rockwell Science Center, Thousand Oaks, Calif., 1974).
Chapter 5
Approximation methods
Virtually every problem going beyond the potential well, the harmonic oscillator, or the hydrogen
atom without spin and external fields is impossible to solve analytically. In this chapter we will
talk about techniques to solve approximately problems in more realistic situations. There are
a number of methods of which we will discuss the only following: 1. The stationary or time-
dependent perturbation method is useful for evaluating small perturbations of the system, for
example, caused by external electric or magnetic fields; 2. the variational method, which serves
to find and improve trial wavefunctions, the initial shapes of which are generally motivated by
the symmetries of the system; 3. the semi-classical WKB method; 4. and finally the method of
self-consistent fields, which is an iterative method of solving the Schrödinger equation.
5.1 Stationary perturbations

5.1.1 Time-independent perturbation theory
We first introduce time-independent perturbation theory (TIPT) for multilevel systems. We
separate the Hamiltonian into an unperturbed part,
Ĥ (0) |ψ (0) i = E (0) |ψ (0) i , (5.1)
and perturbations, which are proportional to a small parameters λ,
Ĥ = Ĥ (0) + λĤ (1) + λ2 Ĥ (2) + .. . (5.2)
The perturbed wavefunctions are,
|ψi = |ψ (0) i + λ|ψ (1) i + λ2 |ψ (2) i + .. , (5.3)
and the energies

E = E (0) + λE (1) + λ2 E (2) + .. . (5.4)
The contributions ∝ λn are the corrections of order n. The equation we need to solve now is,
Ĥ|ψi = E|ψi . (5.5)
By inserting all the expansions above and segregating all orders of λk , we find the following
system of equations,
Ĥ (0) |ψ (0) i = E (0) |ψ (0) i (5.6)

(0) (0) (1) (1) (1) (0)
(Ĥ −E )|ψ i = (E − Ĥ )|ψ i
(0) (0) (2) (2) (2) (0)
(Ĥ −E )|ψ i = (E − Ĥ )|ψ i + (E (1) − Ĥ (1) )|ψ (1) i
... .
129
130 CHAPTER 5. APPROXIMATION METHODS
5.1.1.1 First order energy correction

(1)
We now consider eigenstates |ψn i of the perturbed system and expand the first-order correction
(0)
of the wavefunction in a linear combination of unperturbed eigenvectors |ψn i ≡ |ni,
X
|ψn(1) i = |mihm|ψn(1) i . (5.7)
m
We insert this expansion into the second equation (5.6) and multiply with hn|,
X
hn|(Ĥ (0) − En(0) ) |mihm|ψn(1) i = 0 = hn|En(1) − Ĥ (1) |ni . (5.8)
m
We obtain for the first order correction of the energy of unperturbed states,
En(1) = hn|Ĥ (1) |ni . (5.9)
As a first example we will calculate in Exc. 5.5.1.1 the first order correction for the energy
of a slightly deformed one-dimensional box potential.
5.1.1.2 First order correction for the wavefunction
Now let us have a look at the first-order correction for the wavefunction again considering the
second equation (5.6),
hm|Ĥ (0) − En(0) |ψn(1) i = hm|En(1) − Ĥ (1) |ni . (5.10)
(1)
When n = m, the left side of this equation disappears. Therefore, En − hn|Ĥ (1) |ni = 0, and
(1)
we can restrict to the terms n 6= m discarding the terms in En ,
(1)
En δmn − hm|Ĥ (1) |ni hm|Ĥ (1) |ni
hm|ψn(1) i = (0) (0)
= (0) (0)
. (5.11)
Êm − En En − Em
We obtain for the first-order correction for the energy of the states,
X X hm|Ĥ (1) |ni

|ψn(1) i = |mihm|ψn(1) i = |mi (0) (0)
. (5.12)
m m6=n Ên − Em
This procedure simulates the distortion of the state by blending it with other states. The
perturbation induces virtual transitions to other states. The perturbation is large when the
blended levels are close.
See Exc. 5.5.1.2. In Exc. 5.5.1.3 we calculate the first order correction due to the finite
extension of the hydrogen nucleus. In Exc. 5.5.1.4 we treat the coupling of the energy levels of
a two-level system as a first order perturbation, and compare the result with the exact solution.
The Stark effect for an electron confined in a box can be discussed (see Exc. 5.5.1.5) in first
order TIPT.
5.1. STATIONARY PERTURBATIONS 131
5.1.1.3 Second order correction for the energy

To calculate the second order correction for the energy we expand the second order correction,
X
|ψn(2) i = |mihm|ψn(2) i , (5.13)
m
import it into the third equation (5.6) and multiply with hn|,
X X
hn|(Ĥ (0) −En(0) ) |mihm|ψn(2) i = hn|(En(2) − Ĥ (2) )|ni+hn|(En(1) − Ĥ (1) ) |mihm|ψn(1) i . (5.14)
m m
Now,
X X
(2) (0) (0) (2) (2) (1) (1) (1)
hm|ψn i(En − Em )δnm = 0 = En − hn|Ĥ |ni + hm|ψn i En δnm − hn|Ĥ |mi .
m m
(5.15)
The left-hand side of this equation disappears. Also, on the right-hand side, for n 6= m, the term
(1)
En δnm disappears, and for n = m the whole parenthesis disappears. Therefore, we can discard
(1) (1)
the term En and restrict the sum to terms with n 6= m. Inserting the coefficients hm|ψn i
calculated in (5.11), we finally obtain,
X hn|Ĥ (1) |mihm|Ĥ (1) |ni

En(2) = hn|Ĥ (2) |ni + (0) (0)
. (5.16)
m6=n En − Em
The first term is similar to the first order correction; the eigenvalue of the second order pertur-
bation calculated in the base of the unperturbed states. The second term describes the shift of
the energies through possible temporary transitions to other states.
In Exc. 5.5.1.6 we treat a system of three coupled levels up to the second perturbative order.
The Stark effect discussed in Exc. 5.5.1.7 needs the TIPT calculation up to the second order.
5.1.2 TDPT with degenerate states

Exact calculations show that the effect of a perturbation is larger – but finite – for degenerate
states. On the other hand, from the above expressions for the corrections of both energies and
wavefunctions, we would infer that these corrections can become very large for small perturba-
tions or even diverge.
Fortunately, the fact that every linear combination of degenerate wavefunctions is an eigen-
function of the Hamiltonian as well gives us the freedom to choose the combination, which is
most similar to the final form of the perturbed wavefunctions. For example, considering a per-
turbation by a magnetic field it may be advantageous to expand the spherical functions Ylm on a
basis of cylindrical coordinates 1 . We will see in the following that we can solve both problems,
the selection of the initial combination and the prevention of divergent denominators at once,
without explicitly specifying the expansion.
(0)
We consider eigenstates |n, νi with the energy En being r times degenerate with respect to
the quantum number ν, where ν = 1, .., r. All states satisfy
Ĥ (0) |n, νi = En(0) |n, νi . (5.17)

1
Another example would be the preference for the coupled base |(l, s)j, mj i in comparison to the decoupled
base |l, ml , s, ms i knowing that the degeneracy in j is lifted, when there is an energy associated with interacting
angular momenta and the degeneracy in mj is lifted, when we apply a magnetic field.
We construct linear combinations that most resemble the perturbed states

r
X
(0)
|ψnµ i = cµν |n, νi . (5.18)
ν=1
(0)
When the perturbation Ĥ (1) is applied, we assume that the state |ψnµ i is distorted towards the
(0)
similar state |ψnµ i, and the energy changes from En to Enµ . We now need the index µ to label
the energy, since the degeneracy can be removed by the perturbation. As before, we write now,
Ĥ = Ĥ (0) + λĤ (1) + .. (5.19)

(0) (1)
|ψnµ i = |ψnµ i + λ|ψnµ i + ..
(0) (1)
Enµ = En + λEnµ + .. .
The replacement of these expansions in Ĥ|ψnµ i = Enµ |ψnµ i, and a collection of the terms in λ
up to first order gives,
Ĥ (0) |ψnµ
(0)
i = En(0) |ψnµ
(0)
i (5.20)
(En(0) − Ĥ (0) (1)
)|ψnµ i = (1)
(Enµ − Ĥ (1) (0)
)|ψnµ i .
As before, we try to express the first-order corrections for the wavefunctions through degen-
(0) (0)
erate unperturbed wavefunctions, |ψnµ i, and non-degenerate wavefunctions 2 , |ψm i:
X X
(1) (0) (0)
|ψnµ i= bµν |ψnν i+ anm |ψm i. (5.21)
ν m
Inserting this into the first order equation (5.20), we obtain,

X X
bµν (En(0) − En(0) )|ψnν
(0)
i+ (0)
anm (Em − En(0) )|ψm
(0) (1)
i = (Enµ − Ĥ (1) )|ψnµ
(0)
i. (5.22)
ν m
The first term disappears. Inserting the expansion (5.18),

X X
(0)
anm (Em − En(0) )|ψm
(0) (1)
i = (Enµ − Ĥ (1) ) cµν |n, νi , (5.23)
m ν
and multiplying the two sides with hn, µ|, we get zero on the left-hand side, since we can choose
the non-degenerate states to be orthogonal hn, ν|mi = δm,n . Hence,
X h i
(1)
cµν Enµ hn, µ|n, νi − hn, µ|H (1) |n, νi = 0 . (5.24)
ν
This secular equation (one for each µ) represents, in fact, a set of r linear equations for the
coefficients cµν . The condition for having non-trivial solutions is,

det hn, ν|H (1) |n, µi − Enµ
(1)
δµ,ν =0. (5.25)
µ,ν
(1)
The solution of this secular determinant yields the solicited energies Eµ . Now, the solution of
the secular equation (5.24) for each energy value produces those coefficients, which represent
2 (1)
Note that we label all states which are not degenerate with the state under investigation |ψnµ i with the index
m, even if there are degeneracies between them.
5.1. STATIONARY PERTURBATIONS 133
the best linear combinations adapted to the perturbation. Unlike in previous calculations with
degenerate states, here we consider linear combinations of vectors of the degenerate subspace
prior to switching on the perturbation.
In practice, we apply perturbation theory only to the lowest relevant order. That is, we only
calculate the second order correction if first order corrections vanish. One famous example is
the quadratic Stark effect discussed in Sec. 8.3. In the case of eigenvalues, which are degenerate
in the absence of perturbation, the first order will always produce a remarkable correction, as
in the example of the linear Stark effect, also discussed in Sec. 8.3. For this reason, we need not
discuss higher perturbation orders in the case of degenerate eigenvalues.
Example 25 (Perturbation in a system with two degenerate states): As an example,

we consider the following Hamiltonian,

∆ Ω
Ĥ = .
Ω ∆
The exact solution gives the eigenvalues and eigenvectors,

1 −1
E1 = ∆ + Ω , E2 = ∆ − Ω , |ψ1 i = √12 , |ψ2 i = √1 .
1 2 1
Now we divide the Hamiltonian into an unperturbed part and a perturbation,

(0) (1) ∆ 0 0 Ω
Ĥ ≡ Ĥ + Ĥ = + .
0 ∆ Ω 0
We get in zero order,

(0) (0) 1 0
E1 =∆= E2 , |1i = , |2i = ,
0 1
The application of non-degenerate perturbation theory in first order would give,
(1) h1|Ĥ (1) |2i (1)

h1|Ĥ (1) |1i = 0 = h2|Ĥ (1) |2i , |ψ1 i = |1i (0) (0)
→ ∞ ← −|ψ2 i .
E1 − E2
That is, the correction of the energy vanishes in first order, while the correction of the
wavefunction diverges. Obviously, the |νi obtained by the diagonalization of the matrix
Ĥ (0) is not adapted to the calculation of the matrix elements Ĥ (1) .
Now, applying degenerate perturbation theory, we obtain by the secular determinant,
!
h i −Eµ
(1)
Ω
(1) (1)
0 = det hν|Ĥ )|µi − Eµ δµ,ν = det (1) = (Eµ(1) )2 − Ω2 ,
Ω −Eµ
(1) (1)
eigenvalues are E1 = Ω and E2 = −Ω allowing the establishment of the secular equation,
h i
(1)
c11 E1 − h1|Ĥ (1) |1i − c12 h1|Ĥ (1) |2i = c11 [Ω − 0] − c12 Ω = 0
h i
(1)
−c21 h2|Ĥ (1) |1i + c22 E2 − h2|Ĥ (1) |2i = −c21 Ω + c22 [−Ω − 0] = 0 .
We obtain c11 = c12 e c21 = −c22 and with this,

X
(0) 1 1 (0)
√1
−1
|ψ1 i = c1ν |νi = c11 |1i + c12 |2i = 2
√ , |ψ2 i = c21 |1i + c22 |2i = .
1 2 1
ν
Thus, we can verify that the corrections for the eigenenergies,

(0) (0) (0) (0) (0) (0)
E1 = E1 + hψ1 |Ĥ (1) |ψ1 i = ∆ + Ω , E2 = E2 + hψ2 |Ĥ (1) |ψ2 i = ∆ − Ω ,
(0)
coincides with the exact calculation made at the beginning. The eigenfunctions |ψ1 i should
have been already corrected in first order, which we verify by calculating,
(0) (0)
(1) (0) hψ1 |Ĥ (1) |ψ2 i (1)
|ψ1 i = |ψ1 i = 0 = |ψ2 i .
E1 − E2
In Exc. 5.5.1.8 we study a partially degenerate three-level system and the breakdown of the
degeneracy due to a perturbation. And in Exc. 5.5.1.9 we will treat a perturbation in a box
potential with degenerate energy levels.
5.2 Variational method

5.2.1 The Rayleigh fraction
Let us assume that we want to calculate the ground state energy Eg of a system described by
a Hamiltonian Ĥ, but we do not know the wavefunction, and we do not know how to solve
the Schrödinger equation. If at least we had a good idea of the generic form of the solution
(Gaussian, sinusoidal, ..), we could choose a trial function with a free parameter and optimize
this parameter minimizing the energy, which ought to be minimal for the ground state. This is
precisely the idea of the variational method. Note that the variational method only works for
the ground state.
For any function ψ we know that the Rayleigh fraction E satisfies,
hψ|Ĥ|ψi
Eg ≤ ≡E , (5.26)
hψ|ψi
not only when ψ is the wavefunction of an excited state, but even when it represents a (imperfect)
trial to the ground state. Assuming normalized wavefunctions we can discard the denominator
hψ|ψi = 1. To verifyPthe theorem, we expand the function ψ into orthonormal (unknown)
eigenfunctions, |ψi = n cn |ψn i. Since ψ is normalized,
X X
1 = hψ|ψi = hψm |c∗m cn |ψn i = |cn |2 . (5.27)
m,n n
In the same way, X X

hψ|Ĥ|ψi = hψm |c∗m Ĥcn |ψn i = En |cn |2 . (5.28)
m,n n
As the ground state is that of the lowest energy, Eg ≤ En , we have demonstrated the relationship
(5.26) X X
Eg = Eg |cn |2 ≤ En |cn |2 = hĤi . (5.29)
n n
In practice, the ansatz ψα for the ground state allows us to calculate an energy that must
be minimized via
∂hψα |Ĥα |ψα i
=0 . (5.30)
∂α
In the Excs. 5.5.2.1 and 5.5.2.2 we will approach the fundamental state of a quartic potential
and a harmonic oscillator, respectively, by trying several trial wavefunctions and optimizing
their free parameters.
5.3. WKB APPROXIMATION 135
5.2.2 Rayleigh-Ritz method

A modification of the variational method is the Rayleigh-Ritz method. Here, instead of using
aPtrial function, we use a linear combination of eigenfunctions with variable coefficients: |ψi =
k ck |ki. These variables are then optimized to minimize the Rayleigh fraction,
P
k,m ck cm hk|Ĥ|mi
Eg ≤ P =E , (5.31)
k,m ck cm hk|mi
where we assume real coefficients and eigenfunctions. For this, the derivatives with respect to
all coefficients must vanish:
P P P P P
∂E k ck hk|Ĥ|qi + m cm hq|Ĥ|mi k,m ck cm hk|Ĥ|mi ( k ck hk|qi + m cm hq|mi)
= P − P 2
∂cq k,m ck cm hk|mi
k,m ck cm hk|mi
(5.32)
P P
k ck (hk|Ĥ|qi − Ehk|qi) + m cm (hq|Ĥ|mi − Ehq|mi)
= P =0,
k,m ck cm hk|mi
using the definition of E (5.31). The equation is satisfied when the numerator disappears:
X
0= cm (hq|Ĥ|mi − Ehq|mi) . (5.33)
m
The condition for the existence of solutions is that the secular determinant disappears,
0 = det(hq|Ĥ|mi − Ehq|mi) . (5.34)
The solution of this equation leads to a set of values E, and the lowest value, Emin , is the best
approximation for the ground state energy. The coefficients of the wavefunction are obtained
by solving the eigenvalue equation (5.33) with Emin .
In Exc. 5.5.2.3 we will use the Rayleigh-Ritz method to estimate the effect of a finite nuclear
mass of the hydrogen atom on the energy levels. In Exc. 5.5.2.4 we will use the Rayleigh-Ritz
method to find the maximum number of atoms allowing for a stable Bose-Einstein condensate
made of atoms subject to an attractive interatomic force.
5.3 WKB approximation

The WKB approximation (from Wentzel-Kramers-Brillouin) [37, 155, 254] is a method to find
approximate solutions for linear differential equations with spatially variable coefficients. It is
typically used for calculations in quantum mechanics where the wavefunction is reformulated as
an exponential semi-classically expanded function, and then the amplitude or phase is slowly
changed. In the following, we present the WKB approximation applied to the Schrödinger
equation and exemplify it in some canonical systems.
5.3.1 WKB approximation applied to the Schrödinger equation

Starting from the time-independent Schrödinger equation,
~2 d2
− ψ(x) + V (x)ψ(x) = Eψ(x) , (5.35)
2m dx2
and rewrite it as follows,

d2 ψ
= k(x)ψ(x) . (5.36)
dx2
p
with k(x) = 2m[E − V (x)]/~2 . For now, we will restrict ourselves to energies E > V (x). In
this scheme, the wavefunctions are usually complex functions, so that we can write them in polar
coordinates, containing an amplitude A(x) and a phase φ(x), which are both real numbers:
ψ(x) = A(x)eiφ(x) . (5.37)
Substituting this function into the Schrödinger equation we obtain a system of coupled equations
in terms of A(x) and φ(x),
h i
A00 = A (φ0 )2 − k 2 and (A2 φ0 )0 = 0 . (5.38)
The equations (5.39) and (5.37) are completely equivalent to the Schrödinger equation. The
second Eq. (5.38) is easy to solve,
C
A= √ 0 , (5.39)
φ
being C a real constant. We can not say the same thing about the solution of the first Eq. (5.38).
In order to solve it we are going to use the WKB approach, assuming that A varies slowly, so
the term A00 → 0. By doing this approximation we can rewrite Eq. (5.39) in this way:
(φ0 )2 = k 2 . (5.40)
Solving this last expression we obtain two linearly independent solutions, φ0 = ±k. So we get
the expression for the phase: Z
φ(x) = ± k(x)dx . (5.41)
We write this indefinite integral, because the constant term can be absorbed by the constant C.
Finally, we obtain the expression for the wavefunction in the WKB approximation:
C R
ψ(x) = p e± |k(x)|dx . (5.42)
|k(x)|
Here, taking the absolute value of the wavevector, we have already generalized for the case that
the energy E of the particle is lower than the potential V (x) (classically forbidden region).
Example 26 (WKB approximation): The WKB approach is a semiclassical method

for the solution of the Schrödinger equation that does not require the potential to be a
perturbation of a soluble problem. Instead, it only assumes that certain classical quantities
having the dimension of an action (energy per time) are much larger than Planck’s constant.
Inserting the ansatz
ψ(x) = AeiS(x)/~ ,
into the one-dimensional time-independent Schrödinger equation, we find,
i~ 00 1 0 2
− 2m S (x) + 2m S (x) + V (x) − E = 0 .
Now we expand the exponent in orders of ~,

~2
S(x) = S0 (x) + ~S1 (x) + 2 S2 (x) + ... ,
and insert it in the above equation. Collecting the orders in ~, we find in the first orders,
1 0 2
0
2m S0 (x) + V (x) − E ~ = 0
i~ 00 1 0

− 2m S0 (x) + m S0 (x)S10 (x) ~1 = 0
1 0 2 00
2
2m S1 (x) − 2m S1 (x) ~ = 0 .
i
Rxp
The solution of the zeroth order equation, S0 (x) = ± 2m[E − V (x0 )dx0 , gives
Rx√
i
2m[E−V (x0 )dx0
ψ(x) = Ae± ~ .
The WKB approximation can be used to describe continuous potentials (or barriers) by
stepwise constant potentials. The transmission |T |2 through these parts can be obtained by
multiplying the individual tunneling probabilities,
Z
2
ln |T | ' −2 κ(x)dx ,
barrier
p
with κ(x) = ~1 2m[V (x) − E].
5.3.2 Connection formulas

Now let us derive the connection formulas that interconnect solutions with E above and below
V (x) at the turning points, precisely those regions where WKB fails. We will apply the derivation
to a generic confining potential shown in Fig. 5.1.
Figure 5.1: (a) Potential for which we want to obtain the connection formula. (b) Turning point.
Let us start with the right turning point Fig. 5.1(a)]. First, we shift the coordinate system
so that the turning point coincides with zero, as shown in Fig. 5.1(b). As seen above, the WKB
solutions will be given by the following equations:
 h R0 0 0
R0 0 0
i
√1 Bei x k(x )dx + Ce−i x k(x )dx if x≤0
k(x)
ψ(x) ≈ Rx . (5.43)
 √ 1 De− 0 |k(x0 )|dx0 if x≥0
k(x)
In the vicinity of the turning point we approximate the potential by a straight line (Taylor series
expansion up to first order) with the following functional dependence,
V (x) ≈ E + V 0 (0)x . (5.44)

The Schrödinger equation for this potential acquires the following format,
d2 ψt
= α3 xψt , (5.45)
dx2
with α = [ 2m
~2
V 0 (0)]1/3 . Through a change of variables, z = αx, we fall back on Airy’s equation,
d2 ψt
= zψt , (5.46)
dz 2
having as solution a linear combination of the two solutions of the Airy equation,
ψt (x) = aAi(αx) + bBi(αx) . (5.47)
Now let’s have a look at the WKB solutions in the two regions √ in the vicinityRof the turning
x
point. In the classically forbidden region we have k(x) = α3/2 −x, thus being 0 |k(x0 )|dx0 =
2 3/2 . Thus, the WKB solution in the classically forbidden region near the turning point
3 (αx)
will be given by:
D 2 3/2
ψ(x) ≈ √ e− 3 (αx) . (5.48)
~α x
3/4 1/4
Using the asymptotic forms of Airy functions in the solution (5.47) we obtain the following
expression for ψt (x),
a 2 3/2 b 2 3/2
ψp (x) ≈ √ 1/4
e− 3 (αx) + √ 1/4
e 3 (αx) , (5.49)
2 π(αx) π(αx)
q
4π
which when compared to equation (5.48) shows us that a = α~ D and b = 0. Repeating the
previous steps in the negative region we see that the WKB solution in the asymptotic forms of
the Airy solutions for approximately linear potentials takes the following format (with b = 0):
1 h 2 3/2 2 3/2
i
ψ(x) ≈ √ Bei 3 (−αx) + Ce−i 3 (−αx) , (5.50)
~α3/4 (−x)1/4
and
a 1 h iπ/4 i 2 (−αx)3/2 −iπ/4 −i 23 (−αx)3/2
i
ψp (x) ≈ √ e e 3 − e e . (5.51)
π(−αx)1/4 2i
a
When compared, 2i√ π
eiπ/4 = √B~α and − 2i√ a
π
e−iπ/4 = √C~α . Having all this information we can
rewrite the WKB solutions for all positions in the potential, including the turning points 3 :
 hR i
 √2D sin xx2 k(x0 )dx0 + π4 , if x ≤ x2
k(x)
ψ(x) ≈ − x |k(x0 )|dx0
R . (5.52)
√ D
e x2 , if x ≥ x2
k(x)
Repeating the process for a decreasing turning point [left turning point of the potential of
Fig. 5.1(a)], we obtain the following expression:
 0 Rx
1 0 0
 √D e− x |k(x )|dx , se x ≤ x1
k(x) hR i
ψ(x) ≈ . (5.53)
 √2D sin x k(x0 )dx0 + π , if x ≥ x1
0
k(x) x1 4
3
Note that we shifted the turning point to an arbitrary position x2 .
Example 27 (Harmonic oscillator ): Now we apply the WKB method to a well-known

system: the harmonic oscillator. We will calculate its energy levels and the respective eigen-
functions.
Eigenenergies: First, note that for a confining potential, and more specifically in the region
where E ≥ V (x), we have the solutions obtained for the left and right turning point, these
two solutions must match each other, that is,
2D h Z x2 πi 2D0 hZ x πi
0 0
p sin k(x )dx + ' p sin k(x0 )dx0 + ,
k(x) x 4 k(x) x1 4
and hence the zeros of these functions, so the arguments of those sines must be equal (except
for a multiple of π),
Z x2 Z x1
π π
kdx0 + = − kdx0 − + nπ (5.54)
x 4 x1 4
Z x Z x2 1
+ kdx0 = n − π
x1 x 2
Z x2 1
kdx0 = n − π,
x1 2
with n = 1, 2... 4 . With this information we take a harmonic potentialq of the type
q V (x) =
1 2 2E 2E
2 κx . In this case, the turning points for a given energy E will be at − κ and κ . For
q
this potential we will have that k(x) = 2m 1 2
~2 (E − 2 κx ). Calculating the integral of k(x)
between these two turning points we get,
Z x2 Z √2E/k r
2m 1 m 1
k(x)dx = ~ √ E − κx2 dx = πE = n − π~ ,
x1 − 2E/k 2 κ 2
p
isolating E and taking ω = κ/m we have,
1
E = n− ~ω ,
2
with n = 1, 2..., the exact spectrum of the harmonic oscillator, but this is just a coincidence.
Eigenstates: Now we will calculate the eigenstates of the harmonic oscillator. The eigen-
functions were calculated on a computer. The first graph (Fig. 5.2) compares the first exact
excited state with that obtained using the WKB method.
Figure 5.2: First excited state calculated accurately and through the WKB approximation.
Note that the WKB approach is very good when x → 0 and x → ∞, regions where the
4
Note that n 6= 0, because the integral (5.54) has to be greater than zero.
difference between the oscillator energy and the potential are large (E V (x → 0) and
E V (x → ∞)), because in these regions the wavelength λ(x) acquires the lowest values,
p
since it is proportional to |1/ E − V (x)|. Hence, the spatial region in which the potential
needs to be practically constant is smaller, which explains why the approximation is closer
to the exact solution. In the intermediate regions the difference between E and V (x) begins
to decrease, and the WKB approximation delivers its worst results.
As we increase the energy of the harmonic oscillator, the approximation becomes better (for
the same reason as discussed in the previous paragraph). The following graph illustrates this
effect for n = 10.
Figure 5.3: Wavefunction of the vibrational state n = 10 calculated exactly and using the WKB
approximation.
Example 28 (Hydrogen atom): Eigenenergies: For the hydrogen atom the effective
potential is given by,
e2 1 ~2 l(l + 1)
V (x) = − + .
4π0 r 2m r2
Note that the WKB method for this case obeys the relation (5.54), hence we get,
Z r2 √ Z r2 s
e2 1 ~2 l(l + 1)
p(r)dr = 2m E+ − dr
r1 r1 4π0 r 2m r2
Z r2 s
√ 1 e2 ~2
= −2mE −r2 − r+ l(l + 1)dr .
r1 r 4π0 E 2mE
Notice that E < 0. Let us make the following substitution to facilitate algebraic manipula-
tions,
e2 ~2
B=− and C=− l(l + 1) .
4π0 E 2mE
The turning points r1 and r2 are given by the following expressions,
√ √
B − B 2 − 4C B + B 2 − 4C
r1 = and r2 = .
2 2
Thus, returning to the integral we will have the following:
Z r2 Z r2 p
√ 1 √ π √ √
p(r)dr = −2mE (r − r1 )(r2 − r)dr = −2mE ( r2 − r1 )2
r1 r1 r 2
√ π √ π √
= −2mE (r1 + r2 − 2r1 r2 ) = −2mE (B − 2 C)
2√ 2
π e2 2m p 1
= − √ − 2~ l(l + 1) = n − π~ .
2 4π0 −E 2
5.4. TIME-DEPENDENT PERTURBATIONS 141
Isolating E we obtain the energy spectrum of the hydrogen atom in the WKB approximation:
m e2 2 1 13.6
E=− h i2 = − h i2 eV .
2~ 4π0
n − 1/2 + l(l + 1) n − 1/2 + l(l + 1)
For high energies (n l), we recover Bohr’s expression.
5.4 Time-dependent perturbations

Temporal perturbations typically occur when we suddenly switch on an external field that in-
fluences the motion or spin of the particles, or when the field varies over time, for example, an
electromagnetic field. Let us first study a two-level system subject to a temporal perturbation.
5.4.1 Two-level systems

We write the perturbation as
Ĥ = Ĥ (0) + Ĥ (1) (t) . (5.55)
As in the case of a stationary perturbation, we write the eigenenergies and -functions of the
unperturbed system as
Ĥ (0) |ni = En |ni . (5.56)
Recalling that this stationary Schrödinger equation was obtained from the time-dependent
Schrödinger equation via a separation ansatz (2.53), the temporal evolution of these self-functions
is given by,
|ψn(0) (t)i = |nie−iEn t/~ . (5.57)
For small perturbations we can expect that the ansatz,
(0) (0)
|ψ (1) (t)i = a1 (t)|ψ1 (t)i + a2 (t)|ψ2 (t)i , (5.58)
be good. Note that not only do eigenfunctions oscillate, but the coefficients also depend on time,
because the composition of the states can change. The instantaneous probability of finding the
system in state n is |an (t)|2 . Importing the above linear combination into the Schrödinger
equation,
h i ∂
Ĥ (0) + Ĥ (1) (t) |ψ (1) (t)i = i~ |ψ (1) (t)i , (5.59)
∂t
we find,
(0) (0) (0) (0)
a1 Ĥ (0) |ψ1 i + a2 Ĥ (0) |ψ2 i + a1 Ĥ (1) |ψ1 i + a2 Ĥ (1) |ψ2 i
" #
(0) (0)
∂a1 (0) ∂a2 (0) ∂|ψ1 i ∂|ψ2 i
= i~ |ψ i + |ψ i + a1 + a2 (5.60)
∂t 1 ∂t 2 ∂t ∂t
(0) (0) (0) (0)
=⇒ a1 Ĥ (1) |ψ1 i + a2 Ĥ (1) |ψ2 i = i~ȧ1 |ψ1 i + i~ȧ2 |ψ2 i ,
because the other terms satisfy the Schrödinger equation of zero order. Replacing the stationary
eigenfunctions,
a1 e−iE1 t/~ Ĥ (1) |1i + a2 e−iE2 t/~ Ĥ (1) |2i = i~ȧ1 e−iE1 t/~ |1i + i~ȧ2 e−iE2 t/~ |2i , (5.61)
and multiplying this equation with h1|× and h2|×, we find with the abbreviation ~ω0 ≡ E2 − E1 ,
i~ȧ1 = a1 h1|Ĥ (1) |1i + a2 e−iω0 t h1|Ĥ (1) |2i e i~ȧ2 = a1 eiω0 t h2|Ĥ (1) |1i + a2 h2|Ĥ (1) |2i . (5.62)
Frequently, the perturbation induces only a coupling, but does not directly influence the energies,
hn|Ĥ (1) |ni = 0,
e−iω0 t eiω0 t
ȧ1 = a2 h1|Ĥ (1) |2i and ȧ2 = a1 h2|Ĥ (1) |1i . (5.63)
i~ i~
Without perturbation, hm|Ĥ (1) |ni = 0, no dynamics develops; the eigenfunctions evolve
independently.
Example 29 (The Rabi formula): Now, a perturbation be suddenly switched on at time
t = 0, h1|Ĥ (1) |2i ≡ ~Ω · Θ(t). With the equations of motion we can, starting from the initial
situation a1 (0) and a2 (0), calculate the temporal evolution,
ȧ1 = −iΩa2 e−iω0 t and ȧ2 = −iΩ∗ a1 eiω0 t .
We solve this system of differential equations by differentiating one and substituting the
other,
ä2 = −iȧ1 Ω∗ eiω0 t + ω0 a1 Ω∗ eiω0 t = −a2 |Ω|2 + iω0 ȧ2 .
We find solutions via the ansatz a2 = eiω0 t/2 (AeiGt + Be−iGt ). The equation for a2 yields,
(iG + 2i ω0 )2 AeiGt+iω0 t/2 + (−iG + 2i ω0 )2 Be−iGt+iω0 t/2
= −|Ω|2 (AeiGt+iω0 t/2 + Be−iGt+iω0 t/2 )
h i
+ iω0 (iG + 2i ω0 )AeiGt+iω0 t/2 + (−iG + 2i ω0 )Be−iGt+iω0 t/2 .
Separating the parts in A and in B we obtain two equations with the same result,
G2 = |Ω|2 + 14 ω02 .
Ω is called Rabi frequency and G is the generalized Rabi frequency. Using the initial con-
ditions, a1 (0) = 1 and a2 (0) = 0, we can fix one of the coefficients A and B, since
a2 (0) = A + B = 0,
a2 = 2iAeiω0 t/2 sin Gt .
We now import this solution into the differential equation for a1 ,
ȧ1 = 2ΩAe−iω0 t/2 sin Gt .
The integral is,
Z t
0 2A −iω0 t/2
a1 (t) = 2ΩAe−iω0 t /2 sin Gt0 dt0 = − ∗
e G cos Gt + 12 iω0 sin Gt .
0 Ω
Using the normalization condition,

2A 2 2

1 = |a1 |2 + |a2 |2 = − ∗ e−iω0 t/2 G cos Gt + 12 iω0 sin Gt + A2ieiω0 t/2 sin Gt
Ω
4A2 G2
= 2
G2 cos2 Gt + 14 ω02 sin2 Gt + 4A2 sin2 Gt = 4A2 2 .
|Ω| |Ω|
Hence, A = |Ω|/2G. In general, we can choose Ω real, and the final solution is,

iω0 iΩ iω0 t/2
a1 (t) = −e−iω0 t/2 cos Gt + sin Gt and a2 (t) = e sin Gt . (5.64)
2G G
The Rabi formula (5.64) can be derived exactly without perturbative calculus. In Exc. 5.5.4.1
we show a derivation using the Laplace transformation method.
When the energies En are degenerate, under the influence of the perturbation, the popula-
tions of the system oscillate with the Rabi frequency Ω. When the energies are different, the
oscillation frequency is higher, but the amplitude decreases as well. The initially empty state
never reaches unitary population. In Exc. 5.5.4.2 we calculate the time required to allow the
perturbation to invert the population of a two-level system.
5.4.2 The time-dependent perturbation method

Now let’s study systems with many levels.
In time-dependent perturbation theory (TDPT) we separate the Hamiltonian into a stationary
part and a time-dependent part 5,6 ,
Ĥ(t) = Ĥ (0) + λĤ (1) (t) . (5.65)
As usual, this Hamiltonian satisfies the Schrödinger equation,
∂
i~ |ψ(t)i = Ĥ(t)|ψ(t)i . (5.66)
∂t
(0)
Now we do a unitary transformation into the interaction picture with S(t) = e−iĤ t/~ substi-
tuting |ψ(t)i ≡ S(t)|ψI (t)i and Ĥ (1) (t) ≡ S(t)Ŵ (t)S −1 (t) in the Schrödinger equation. This
procedure removes the stationary part, as shown in Sec. 2.4.4,
∂
i~ |ψI (t)i = λŴ (t)|ψI (t)i . (5.67)
∂t
If W (t) were also independent of time, the solution would simply be |ψI (t)i = e−iŴ t/~ |ψI (0)i.
Otherwise, we integrate the equation,
Z
λ t
|ψI (t)i = |ψI (0)i + Ŵ (τ )|ψI (τ )idτ . (5.68)
i~ 0
Substituting |ψI (τ )i by |ψI (t)i we iterate this equation,
Z Z
λ t λ τ1
|ψI (t)i = |ψI (0)i + Ŵ (τ1 ) |ψI (0)i + Ŵ (τ2 )|ψI (τ2 )idτ2 dτ1 (5.69)
i~ 0 i~ 0
Z 2 Z t Z τ1
λ t λ
= |ψI (0)i + Ŵ (τ1 )dτ1 |ψI (0)i + Ŵ (τ1 ) Ŵ (τ2 )|ψI (τ2 )idτ2 dτ1 i
i~ 0 i~ 0 0
X n Z t Z τ1 Z τn−1
N λ
= Ŵ (τ1 ) Ŵ (τ2 )... Ŵ (τn )dτ1 dτ2 ...dτn |ψI (0)i + o(λN +1 ) .
n=1 i~ 0 0 0
This is called the Dyson series. For N = 1, we get the first order of the perturbation series 7 ,
Z
λ t
|ψI (t)i = 1 + Ŵ (τ )dτ |ψI (0)i . (5.70)
i~ 0
5
See Becker-Sauter II, p.118ff e [241], p.104ff. An alternative treatment is found in [164], p.191ff or in
Blochinzew, p.332ff.
6
Note that by substituting W by Ĥ (1) , the equation (5.67), i~∂t |ψI (t)i = Ĥ (1) (t)|ψI (t)i, corresponds to a first-
order perturbative approximation, i.e., the perturbation eigenvalues Ĥ(1) are calculated with the eigenvectors of
the unperturbed system. Thus, in first order TDPT we can substitute W for Ĥ (1) .
7
For higher orders,
X n Z t n Z t
N λ λ
|ψI (t)i ≈ Ŵ (τ )dτ |ψI (0)i = T exp Ŵ (τ )dτ |ψI (0)i .
n=1 i~ 0 i~ 0
(0)
The stationary states of the unperturbed Hamiltonian are P given by Ĥ |f i = Ef |f i. Now,
the perturbed states are expanded on this basis, |ψI (t)i = f |f iaf (t). The expansion coeffi-
cients are 8 ,
Z t
λ
af (t) = hf |ψI (t)i = hf |ψI (0)i + hf | S −1 (τ )Ĥ (1) (τ )S(τ )|ψI (0)idτ . (5.71)
i~ 0
Now, we assume that the system be initially in the eigenstate |ψI (0)i = |ii. The amplitudes
then are,
Z
λ t iEf τ /~
ai→f (t) = hf |ii + e hf |Ĥ (1) (τ )|iie−iEi τ /~ dτ (5.72)
i~ 0
Z
λ t
= δif + hf |Ĥ (1) (τ )|iieiωif τ dτ .
i~ 0
The varying potential is considered a perturbation, and a variation of the system’s state is
observed. As the energy is not conserved, [∂t , Ĥ(t)] 6= 0, the time-dependence is not separable
and the system exchanges energy with the potential. In first-order perturbation theory we only
consider weak perturbations, i.e. the initial state is gradually emptied, ai→i (dt) ' ai→i (0) = 1.
For an initially empty state the growth is obviously considerable. For i 6= f we have,
λ
dai→f (t) = ai→f (t + dt) − ai→f (t) = hf |Ĥ (1) (t)|iieiωif t dt . (5.73)
i~
This formula is nothing more than a generalization of the formula (5.63) obtained for a two-level
system assuming that the initial state does not deplete considerably. In Exc. 5.5.4.3 we calculate
the dynamics of a harmonic oscillator perturbed by a decaying force.
5.4.3 Numerical methods

The softwares ’Maple’ or ’Mathematics’ are useful for analytical calculations, that is, multiplying
matrices or determining eigenvalues. For numerical calculations the software ’Matlab’ is more
adapted. For example, the time evolution of a Schrödinger equation,
|ψ(t)i = e−iĤt/~ |ψ(0)i , (5.74)
can be calculated in a single command line using the Matlab ’expm’ function.
When the system varies temporally, Ĥ(t), we may divide time into small units dt and prop-
agate the wavefunction as,

|ψ(t + dt)i = e−iĤ(t)dt/~ |ψ(t)i ' |ψ(t)i 1 − i Ĥ~ dt , (5.75)
continuously reinserting the solution into the equation. This Newton method does not converge
quickly (dt should be chosen small enough, when Ĥ(t) varies rapidly), but there are other more
sophisticated methods like the Runge-Kutta method.
A variation of this method is called steepest descent method. This method is similar to
the Newton Eq. (5.75), but replaces the time dt with an imaginary time. Thus, the coherent
8
We could define the coefficients in Schrödinger’s picture, af ≡ hf |ψi, but this would only introduce a phase
factor, ai→f −→ ai→f ei(Ef −Ei )t/~ , which is unimportant for absolute values |ai→f |2 . This corresponds to a
transformation to a rotating system, which will be discussed in Sec. 10.2.1.
temporal evolution of the Schrödinger equation is replaced by a dissipative evolution. The loss
of energy automatically takes the system to the ground state. The method also applies to
more complicated equations than the Schrödinger equation, for example, the Gross-Pitaevskii
equation.
Another numerical method often used in quantum mechanics is the method called the quan-
tum Monte Carlo simulation of the wavefunction [184]. This method simulates trajectories of
quantum systems treating intrinsic quantum noise as random processes disrupting the uniformity
of the trajectory. The advantage of this method is that it also applies to dissipative systems.
5.4.4 Transition rates

To begin with, we consider a constant perturbation Ĥ (1) suddenly switched on at t = 0. In
Schrödinger’s picture we can write,
Z t
λ λ −1 + eiωif t
ai→f (t) = δif + hf |Ĥ (1) |ii eiωif τ dτ = δif + hf |Ĥ (1) |ii . (5.76)
i~ 0 i~ iωif
We obtain for i 6= f ,
2
λ2 2 sin (ωif t/2)
|ai→f (t)|2 = |hf | Ĥ (1)
|ii| . (5.77)
~2 (ωf i /2)2
For long times we calculate the rate 9 ,
d λ2 sin ωf i t t→∞ 2πλ2

|ai→f (t)|2 = 2 |hf |Ĥ (1) |ii|2 −→ |hf |Ĥ (1) |ii|2 δ(ωf − ωi ) , (5.78)
dt ~ ωf i /2 ~2
where we use the representation of the Dirac function,

Z t
δ(x) = lim 2π1
eikx dk = lim t
sinc xt . (5.79)
t→∞ −t t→∞ π
The generalization of this transition rate for all perturbation orders is,

1 2πλ2 X (1) λ X hf |Ĥ (1) |lihl|Ĥ (1) |ii
= hf |Ĥ |ii + + ... (5.80)
τ ~2 f ~ l ωi − ωl
2
λn X hf |Ĥ (1) |l1 ihl1 |...|ln−1 ihln−1 |Ĥ (1) |ii
+ n δ(ωif ) ,
~ l1 ,...,ln−1 (ωi − ωl1 ) ... ωi − ωln−1
where several decay channels f have been considered 10 .

In practice, the changes applied to a system are often slow and the observation times are
long because the frequencies of the transitions are high ωf i /2π ' THz. Let us assume that the
perturbation be switched on within a time constant γ −1 . In Exc. 5.5.4.4 we will study how the
rapidity of a perturbation influences the transition rate. We will see via a temporal analysis
of |ai→f (t)|2 , that for slow variations, γ ωf i , the system adiabatically approaches the final
situation. For γ ' ωf i , the system receives a shock and exhibits oscillating transients. For
γ > ωf i , we observe violent oscillations with largest amplitudes.
9
We use the trigonometric rule sin x = 2 sin x2 cos x2 .
10
The second order corresponds to the Kramers-Heisenberg formula, which serves to describe Thomson, Rayleigh
and Raman scattering.
Figure 5.4: Graphical illustration of the various transitions orders.
5.4.5 Periodic perturbations

We now consider the case of an oscillatory perturbation, for example an electromagnetic field.
In principle, knowledge of the system’s response to periodic perturbations allows us to treat
arbitrary perturbations, since we can expand them in Fourier series. We first treat transitions
between discrete levels, before considering states embedded in continua,

(1) 0 for t < 0
Ĥ (t) = . (5.81)
2~Ω̂0 cos ωt for t ≥ 0
With the abbreviation Ωf i ≡ hf |Ω̂0 |ii the transition rate is,

Z " 0
#
t i(ω +ω)t
−1 ei(ωf i −ω)t − 1
ai→f (t) = −iΩf i 2eiωf i τ cos ωτ dτ = −iΩf i e fi
i(ωf i +ω) + . (5.82)
0 i(ωf i − ω)
The first term being small, we neglect it in the rotating wave approximation (RWA). We obtain,
sin2 12 (ωf i − ω)t

|ai→f (t)|2 = |Ωf i |2 1 2
. (5.83)
4 (ωf i − ω)
This result coincides with the Rabi formula (5.64), except that the energy difference between
the states ωf i is shifted by the frequency of the perturbation ω. The quantity ∆f i ≡ ω − ωf i
is called a detuning. The transition probability is maximal, when we are at resonance, that is
∆f i = 0. In this case,
|ai→f (t)|2 −→ |Ωf i |2 t2 . (5.84)
This can be seen by expanding the numerator in a Taylor series for small (ωf i − ω)t.
Note, that the probability exceeds 1 for long times, which can not be. In fact, the restriction
to the first order in the Taylor expansion used in the derivation of the last equation is no longer
valid for long times, when (ωf i − ω)t > 1, and we need to take into account higher orders.
5.4.5.1 Transitions to continuous levels

When there are several final states, f ∈ F , the formula (5.83) must be generalized. The total
transition probability, X
Pi→F (t) = |ai→f (t)|2 , (5.85)
f ∈F
corresponds to the probability of the initial state |ii to be depleted. When the final state lies
within a continuum, the sum in (5.85) must be replaced by an integral. With the density-of-
states written in the form ρ(E), where ρ(E)dE is the number of states found in the energy range
between E and E + dE, the transition probability is 11 ,
Z Emax
Pi→F (t) = |ai→f (t)|2 ρ(Ef )dEf , (5.86)
Emin
where E ∈ [Emin , Emax ] is the regime of energies within reach of the perturbation. Now,
Z 1
Emax sin2 − E)t
2~ (Ef i
Pi→F (t) = |Ωf i |2 1 ρ(Ef )dEf . (5.87)
Emin 4~2
(Ef i − E)2
Using the representation of the Dirac function,
δ(x) = lim t
sinc2 xt , (5.88)
t→∞ π
with the substitution x ≡ (Ef i − E)/2~, we obtain after sufficiently long times 12 ,
Z Emax
π E i −E
Pi→F (t) = |Ωf i |2 t2 δ( f2~ )ρ(Ef )dEf = 2π~t|Ωf i |2 ρ(Ei + E) . (5.89)
Emin t
The transition rate is,

dPi→F (t)
= π|Ωf i |2 ρ(Ei + E) . (5.90)
dt
For narrow distributions we may substitute the density of states ρ by a δ-distribution,
dPi→F 2π
= |hf |Ĥ (1) |ii|2 δ(Ef i − E) , (5.91)
dt ~
where we went back to the definition of the perturbation Hamiltonian (5.81). This expression
is called Fermi’s Golden Rule. In Exc. 5.5.4.5 we will calculate the photoelectric effect.
5.4.5.2 Continuous frequency distribution
To derive Eq. (5.83), we considered perturbations with fixed oscillation frequencies. To handle
frequency distributions %(ω), we must generalize this equation by calculating the integral,
Z
2 2 sin2 1 (ωf i − ω)t
|ai→f (t)| = |Ωf i | %(ω) 1 2 2
dω (5.92)
4 (ωf i − ω)
Z ∞
2
' |Ωf i | t%(ωf i ) sinc2 xdx = πt|Ωf i |2 %(ωf i − ω) ,
−∞
again using the representation (5.88) of the Dirac function. The approximation %(ω) = %(ωf i )
can be used if the width of the sinc function is much narrower than the frequency distribution,
which is the case for sufficiently long times, t π/2∆f i .
11 P
With ρ(E) ≡ f ∈F δ(Ef − E) the integral is converted back into a sum.
12
Remember δ(ax) = a1 δ(x).
5.4.6 Einstein transition probabilities

We now consider the problem of energy transfer between an electromagnetic field and a sample
of atoms. The rate of absorption of a light field is,
π 2
Ri→f ≡ 13 Ṗi→f = E |df i |2 %(ωf i ) , (5.93)
6~2 0
with W0 = E0 df i and df i being the transition matrix element between atomic states. The factor
1
3 comes from the fact that the vector E of the electric field can have any polarization, but only
polarizations along the direction of the oscillation of the dipole moment contribute.
For a single atom, the result (5.93) is symmetric with respect to an exchange of the initial
and final states, that is, the rates for absorption and induced emission of light are the same. For
a sample of atoms being in thermal equilibrium, the populations Ni of the ground state and Nf
of the excited state are unequal according to Boltzmann’s law. Therefore, as we have shown in
Sec. 1.3.5,
Nf Rf →i 6= Ni Ri→f . (5.94)
Thus, Einstein came to the conclusion that Fermi’s golden rule correctly describes absorption,
but does not contain all contributions of emission. The rates being related to the Einstein
coefficients by the equation (1.95), we find,
Rf →i = Bf i Nf J (ωf i ) (5.95)
and
~ωf3i ~ωf3i Ri→f ωf3i
Sf →i = Af i Nf = 2 3 Bif Nf = 2 3 = |df i |2 , (5.96)
π c π c J (ωf i ) 3π0 ~c3
exploiting the relation (1.100). We shall return to the subject of transitions in chapter ??.
5.5 Exercises
5.5.1 Stationary perturbations
5.5.1.1 Ex: One-dimensional well with a deformation in the centre
Consider a one-dimensional potential well between −L/2 and L/2 with infinitely high walls. In
the center of the well is a small deformation,

(1) ε for −a
2 ≤x≤ 2
a
H =
0 outside that region .
Calculate the correction for the eigenenergies in first order and discuss the limits a L and
a → L.
5.5.1.2 Ex: Perturbation

(0) (1)
Show that the scalar product hψn |ψn i (from the first-order correction to the state of the
’perturbed’ system with the n-th state of the free Hamiltonian), cancels out when we impose
(0)
that the ’perturbed’ state |ψ(λ)i be normalized and the the product hψn |ψ(λ)i be real.13
13
See [48], Cap XI, A-2.
5.5. EXERCISES 149
5.5.1.3 Ex: Extended nucleus

The expression V (r) = −e2 /4π0 r for the potential energy of an electron in the hydrogen atom
implies that the nucleus (the proton) is treated as a point particle. Now suppose that, on the
contrary, the charge of the proton +e is evenly distributed over a sphere of radius R = 10−13 cm.
a. Derive the modified potential Vm , which corresponds to this distribution of the nuclear charge.
b. Assume that the wavefunction of the hydrogen atom does not change much due to the modified
potential. Calculate in lowest order in R/aB the average energetic displacement h∆V i for the
state (n = 1, ` = 0, m = 0). How will the energy displacement be in comparison to the states
(n = 2, ` = 0, m = 0) and (n = 2, ` = 1, m = 0)?
c. Calculate in the same way h∆V i for muonic hydrogen in the ground state.
5.5.1.4 Ex: Perturbation of a 2-level system

We consider a two-level system. Without perturbation the system would have the Hamiltonian
(0) (0)
H (0) , the eigenenergies E1,2 and the eigenfunctions ψ1,2 . Now we switch on a stationary pertur-
bation of the form H (1) = (|1ih2| + |2ih1|).
a. Calculate the eigenenergies directly solving the perturbed Schrödinger equation.
b. Calculate the perturbed energies using TIPT and compare to the exact calculation of the
eigenenergies.
c. Calculate the eigenstates directly solving the perturbed Schrödinger equation.
d. Calculate the perturbed states using TPIT and compare to the exact calculation of the eigen-
functions.
5.5.1.5 Ex: Stark effect for an electron in a box

Consider an electron in a one-dimensional box, that is, in a well inside the interval x ∈ [0, a]
delimited by infinite walls. When a uniform electric field E is applied, also in x-direction, the
electron experiences a force equal to −eE, being −e the electron charge, so that the potential
energy inside the box becomes eEx.
a. What is the energy of the ground state of the electron (in first order approximation)? We
can assume that eEa is much smaller than the ground state energy the electron would have in
the absence of electric fields.
b. Use first-order TIPT to get an approximation for the ground state wavefunction by calculating
the first term of the correction.
5.5.1.6 Ex: Perturbed 3-level system until second order TIPT

Consider the following perturbed Hamiltonian:
   
E1 0 0 0 λ 0
H = H0 + Hλ =  0 E2 0  + λ 0 λ .
0 0 E3 0 λ 0
a. Determine the perturbed eigenvalues and eigenfunctions in first order TIPT.
b. Determine the eigenvalues in second order TIPT.
5.5.1.7 Ex: Stark effect for a charge in a harmonic oscillator

Consider a charged harmonic oscillator, immersed in a uniform electric field E, described by
the Hamiltonian H (1) = H + eEx, being H = p2 /2m + mω 2 x2 /2 the Hamiltonian of the free
one-dimensional oscillator, and e the charge of the oscillator.

a. Obtain, through TIPT, the eigenenergies (first and second order corrections). Compare the
results obtained by TIPT with the analytical ones.14
b. Same thing for a perturbation of the form ρmω 2 x2 /2.
c. Same thing for a perturbation σ~ωx3 .
5.5.1.8 Ex: Three-level system with degeneracy

Consider the following Hamiltonian Ĥ (0) and its perturbation Ĥ (1)
   
∆ 0 0 0 Ω 0
Ĥ (0) + Ĥ (1) =  0 ∆ 0  + Ω 0 Ω .
0 0 ∆0 0 Ω 0
Calculate the corrections for the eigenvalues and eigenfunctions up to first order.
5.5.1.9 Ex: Perturbation in a 3D well with degeneracy

Consider a particle confined to a three-dimensional, infinite cubic well described by the poten-
tial energy V (x, y, z) = 0 for 0 < x < a, 0 < y < a and 0 < z < a and V (x, y, z) = ∞
(0)
outside this region. We know that the particle’s stationary states are Ψnx ,ny ,nz (x, y, z) =

2 3/2
n π
a sin nxaπ x sin ya y sin naz π z , being nx , ny , nz positive integers. The associated en-
(0) π ~
2 2
2 2 2
ergies are Enx ,ny ,nz = 2ma 2 (nx + ny + nz ). Note that the ground state is not degenerate while
the first excited state is three times degenerate. Consider that the particle in this box is subject
to a perturbation of the shape H (1) = V0 for 0 < x < a/2 and 0 < y < a/2 and H (1) = 0 outside
this region.
a. Obtain the first-order correction for the ground state energy.
b. Obtain the first-order correction for the (degenerate) energy of the first excited state, in
addition to the optimal base (which follows from the linear combinations of degenerate states)
which most closely approximates the perturbed states.
5.5.2 Variational method

5.5.2.1 Ex: Variational method applied to a quartic potential
Determine the ground state energy of the quartic potential V (x) = bx4 making the variational
2
ansatz ψα (x) = (α/π)1/4 e−αx /2 . Formulae:
Z ∞ Z ∞ Z ∞
2 √ 2 √ 2 √
e−x dx = π , x2 e−x dx = 12 π , x4 e−x dx = 34 π
−∞ −∞ −∞
5.5.2.2 Ex: Variational method applied to the harmonic oscillator

Obtain, through the variational method, the ground state energy of the one-dimensional har-
~2 d2
monic oscillator described by the Hamiltonian H = − 2m dx2
+ 12 mω 2 x2 , and the corresponding
wavefunction from the test functions
2
a. ψ(x) = Ae−αx being α a constant;
b. ψ(x) = A/(x2 + β 2 ) being β a constant;
c. ψ(x) = A cos(πx/a) between the limits ±a/2 being a a constant.
14
See [48], Complement A XI.
5.5. EXERCISES 151
5.5.2.3 Ex: Effect of finite nuclear mass on hydrogen via Rayleigh-Ritz

Use the Rayleigh-Ritz method to estimate the impact of the finite mass of the nucleus of the
hydrogen atom. To do this, calculate the ground state energy using the exact Hamiltonian, but
a basis of wavefunctions assuming an infinitely heavy nucleus. Only take into account the states
ψ100 and ψ200 .
5.5.2.4 Ex: Collapse of a condensate with attractive interactions

A Bose-Einstein condensate of 7 Li may become unstable due to attractive interatomic force, the
scattering length being as = −27.3aB . Consider the radial Gross-Pitaevskii Hamiltonian derived
in Exc. 4.5.1.2 with an external harmonic potential with the oscillation frequency ωtrp /(2π) =
50 Hz. Using the variational method to determine the maximum number of atoms allowing
for a stable condensate. (Note that the derived minimization condition must be evaluated
numerically.)
5.5.3 WKB approximation

5.5.3.1 Ex:
Use the WKB approach to calculate the energy levels of the hydrogen atom.
5.5.4 Time-dependent perturbations

5.5.4.1 Ex: Two-level atom via Laplace transformation
Solve the problem of a two-level atom interacting with a laser using the Laplace transformation
method.
5.5.4.2 Ex: Rabi oscillation

The population of a degenerate two-level system be initially in state |1i. What should be the
duration of a perturbation to transfer the population to state |2i?
5.5.4.3 Ex: Perturbed harmonic oscillator

Consider a one-dimensional harmonic oscillator (HO) initially prepared (t = −∞) in the ground
state |0i of the unperturbed Hamiltonian H (0) = ~ωâ† â, such that H (0) |ni = En |ni with En =
n~ω. R
1 tf iωf i t dt, and the perturbative Hamiltonian W (t) =
a. Through the expression, af (t) ≈ i~ ti Wf i e
2 2
−eExe−t /τ (x is the position operator of the HO), applied between t = −∞ and t = +∞,
calculate the probability of the system to be in the excited state |ni, specifying n, at t = +∞.
Analyze the result.
2 2
b. Do the same for a shape-changing perturbation, W (t) = Λx2 e−t /τ .
5.5.4.4 Ex: Impact of the rapidity of a perturbation

Here we consider a slow variation,

0 for t<0
Ŵ (t) = ,
W0 (1 − e−γt ) for t ≥ 0
with γ ωf i . Calculate the transition rate for long times, t γ −1 .
5.5.4.5 Ex: Photoelectric effect

A hydrogen atom ground state in the ground state 1s is placed in an electric field E(t) =
E0 cos ωt, such that W (t) = −er · E(t) = W0 e−iωt + W0† eiωt with W0 = er · E0 /2. Find, via
Fermi’s Golden rule,
2π
R= |hf |W (t)|ii|2 ρ(Ef − Ei ∓ ~ω) ,
~
using the density of states ρ(Ek )dEk = V /(2π)3 k 2 dkdΩ, the probability per unit of time for
the atom to be ionized, by exciting from the ground state ψ100 (r) = e−r/aB /(πa3B )1/2 to the
state described by the plane wave ψk (r) = e−ik·r /V 1/2 . Simplify the calculation by assuming
E0 = E0 êz and k = kêz .
Chapter 6
Periodic systems
Many physical systems treat quantum particles in periodic potential. Examples are electrons in
crystals or cold atoms in optical lattices. The periodicity adds a wealth of new phenomena.
6.1 The Bloch model

The motion of an electron inside a crystal is ruled by a spatially periodic potential V (r) origi-
nating from the positively charged crystal atoms and the mean field produced by the quasi-free
electrons,
V (r) = V (r + R) , (6.1)
where R is a vector connecting two arbitrarily chosen atoms of the lattice. With the Hamiltonian
~2 2
Ĥ = − ∇ + V (r) (6.2)
2m
we can write the Schrödinger equation,
[Ĥ + V (r)]ψ(r) = Eψ(r) . (6.3)
Since V and ∇ are invariant under translations Utr (R)ψ(r) ≡ ψ(r + R) by a fixed distance R,
where the translation operator has been defined in Eq. (2.168), we have,
ĤUtr (R)ψ(r) = EUtr (R)ψ(r) . (6.4)
That is, for a non-degenerate eigenvalue 1 ,
ψ(r + R) = f (R)ψ(r) . (6.5)
This relation holds for all vectors R of the lattice, such that,
f (R1 + R2 )ψ(r) = ψ(r + R1 + R2 ) = f (R1 )ψ(r + R2 ) = f (R1 )f (R2 )ψ(r) . (6.6)
The relationship f (R1 + R2 ) = f (R1 )f (R2 ) is satisfied by the ansatz f (R) ≡ eik·R , where k is
an arbitrary vector of reciprocal space. We get the famous Bloch theorem,
ψk (r + R) = eik·R ψk (r) , (6.7)
which represents a necessary condition for any eigenfunction ψk of the Schrödinger equation
with periodic potential. Bloch’s theorem simply postulates that, apart from a phase factor, the
wavefunction has the same periodicity as the potential.
1
This also holds true for degenerate eigenvalues if we choose suitable basis of eigenvectors.
153
154 CHAPTER 6. PERIODIC SYSTEMS
The Bloch function,
ψk (r) ≡ uk (r)eik·r with uk (r + R) = uk (r) , (6.8)
automatically satisfies Bloch’s theorem. That is, the wave function of the electron ψ is a plane
wave eik·r modulated by a function uk having the same periodicity as the lattice [151]. Although
the vector of the electronic wave is arbitrary, it is possible (and useful) to restrict its value to the
first Brillouin zone defined by k ∈ [−π/a, π/a], where a is an elementary vector of the lattice.
The reason is that we can reduce a wavevector k in a wavefunction trespassing the first Brillouin
zone by an appropriate vector G of reciprocal lattice,
k0 = k + G , (6.9)
yielding,
0
ψk (r) = uk (r)eik·r = uk (r)e−iG·r eik ·r . (6.10)
We now define another function uk0 (r) ≡ uk (r)e−iG·r , which also satisfies the requirement (6.8),
knowing that G · R = n2π, we see,
uk0 (r + R) = uk (r + R)e−iG·(r+R) = uk (r)e−iG·r = uk0 (r) . (6.11)
Hence,
0
ψk (r) = uk0 (r)eik ·r = ψk0 (r) . (6.12)
6.1.1 Approximation for quasi-bound electrons

We now assume that the behavior of the electron near an atom is not influenced by atoms farther
apart,
X
ψk (r) = ci (k)φ(r − Ri ) . (6.13)
i∈lattice
That is, we neglect superposition states of the electron at various sites of the lattice. The atom
is subject to a potential Uat (r − Ri ) located near the atom at the position Ri , and it is described
by the eigenfunction φ(r − Ri ) (only defined for the site i) with energy E0 ,

~2 2
− ∇ + Uat (r − Ri ) φ(r − Ri ) = E0 φ(r − Ri ) . (6.14)
2m
Even so, the function ψk (r) must satisfy Bloch’s theorem. This is the case when ci (k) = eik·Ri
and therefore,
X
ψk (r) = eik·Ri φ(r − Ri ) . (6.15)
i∈lattice
Example 30 (Ansatz for a quasi-bound electron wavefunction): The ansatz (6.15)

satisfies Bloch’s theorem because,
X X
ψk (r+Rj ) = eik·Ri φ(r−(Ri −Rj )) = eik·Rj eik·(Ri −Rj ) φ(r−(Ri −Rj )) = eik·Rj ψk (r) .
i i
6.1. THE BLOCH MODEL 155
We now calculate the energy E(k) of an electron with the wavevector k inserting the function
ψk (r) of (6.15) in the Schrödinger equation and obtain,
X X
~2 2
− ∇ + U (r) eik·Ri φ(r − Ri ) = E(k) eik·Ri φ(r − Ri ) . (6.16)
2m
i i
U (r) is the potential energy of the electron illustrated in Fig. 6.1 together with the energy
Uat (r − Ri ) of a free electron.
Substituting the kinetic energy term of (6.16) by the kinetic energy of (6.14), we calculate,
X
eik·Ri [−Uat (r − Ri ) + E0 + U (r) − E(k)]φ(r − Ri ) = 0 . (6.17)
i
P ik·Rj ∗
Now, multiplying this equation with ψk∗ (r) = je φ (r − Rj ) and integrating over the
volume of the crystal, we obtain,
X Z
ik·(Ri −Rj )
[E(k) − E0 ] e φ∗ (r − Rj )φ(r − Ri )dV (6.18)
i,j
X Z
= eik·(Ri −Rj ) φ∗ (r − Rj )[U (r) − Uat (r − Ri )]φ(r − Ri )dV .
i,j
The functions φ∗ (r − Rj ) and φ(r − Ri ) overlap only a little, even for adjacent atoms, such
that we can neglect the terms i 6= j on the left side. The sum then corresponds to the number
N of sites in the lattice. On the right side we can not neglect the terms involving other sites,
because even if the wavefunctions of adjacent sites overlap little, the contribution of the potential
difference |U (r) − Uat (r − Ri )| is much lower for r = Ri than for r = Rj . On the other hand,
as the wavefunctions φ(r − Ri ) disappear quickly when |r − Ri | > |Rm − Ri |, we can focus on
adjacent sites (called Rm ),
Z
N [E(k) − E0 ] =N φ∗ (r − Ri )[U (r) − Uat (r − Ri )]φ(r − Ri )dV (6.19)
X Z
+N eik·(Ri −Rm ) φ∗ (r − Rm )[U (r) − Uat (r − Ri )]φ(r − Ri )dV .
m=adjacent
Figure 6.1: Potential energy U (r) of a crystal electron (red) and potential energy UA (r − Ri ) of
the electron of a free atom (blue).
Now we further suppose that the eigenfunction φ exhibits radial symmetry corresponding to
an s orbital. We obtain for the eigenvalues from the Schrödinger equation,

X
E(k) = E0 − α − γ eik·(Ri −Rm ) (6.20)
m adjacent of i
Z
with α= φ∗ (r − Ri )[Uat (r − Ri ) − U (r)]φ(r − Ri )dV
Z
and γ= φ∗ (r − Rm )[Uat (r − Ri ) − U (r)]φ(r − Ri )dV .
The interpretation is as follows: The combination of the atoms in a lattice produces an energy
displacement α. In addition, it generates a spitting into a continuous band of energies as a
function of reduced wavevector k...
6.1.2 Approximation for quasi-free electrons

Here we assume an essentially homogeneous potential acting on the free electrons and con-
sider the imapct of the periodic lattice as a small perturbation. The periodic potential can be
decomposed into a Fourier series by the vectors G of the reciprocal lattice,
X
U (r) = UG eiG·r . (6.21)
G
Consequently, we can make for Bloch functions (6.10) the following periodic ansatz,
X
ψk (r) = uk (r)eik·r with uk (r) = √1V uG (k)eiG·r , (6.22)
c
G
where Vc is the volume of the crystal.

Without periodic potential, the eigenfunctions would be those of a free particle,
1 ik·r
ψk (r) = Vc e (6.23)
with the eigenenergies

~2 k 2
E0 (k) = V0 +. (6.24)
2m
Inserting the functions (6.21) and (6.22) in the Schrödinger equation, we obtain,
" #
X 00
X 0
X 0
~2
− 2m ∇2 + UG00 eiG ·r √1V eik·r uG0 (k)eiG ·r = E(k) √1V eik·r uG0 (k)eiG ·r , (6.25)
c c
G00 G0 G
that is,
Xh 2 i 0
X 00 ·r
X 0
~ 0 2
√1
V
− 2m (k + G ) − E(k) uG0 (k)ei(k+G )·r + √1
Vc
UG00 eiG uG0 (k)ei(k+G )·r = 0 .
c
G0 G00 G0
(6.26)
Now multiplying with √1V ei(k+G)·r and integrating over the volume of the crystal (knowing
1
R iG·r c
Vc Vc e dV = δG,0 ), we obtain,
h i X
~2
2m (k + G)2 − E(k) uG (k) + UG−G0 uG0 (k) = 0 , (6.27)
G0
for any value of G.

To estimate the dependence of the Fourier components uG (k) for G 6= 0 we insert the
unperturbed eigenenergies into the equation (6.27) only considering, in the sum over G0 , the
terms of the first perturbative order, that is, those containing U0 or u0 (k),
~2
[(k + G)2 − k 2 ]uG (k) − U0 uG (k) + U0 uG (k) + UG u0 (k) = 0 (6.28)
2m
UG u0 (k)
uG (k) = ~2 . (6.29)
2 2
2m [k − (k + G) ]
Since the Fourier coefficients UG have, for G 6= 0, small values, the function uG (k) is not
negligible only for k 2 ' (k + G)2 that is,
−2k · G ' |G|2 . (6.30)
We now want to find out the meaning of this condition ...

For the coefficients u0 (k) and uG (k) we obtain,
h 2 i
~ 2
2m k − E(k) u0 (k) + U0 u0 (k) + U−G (k)uG (k) = 0 (6.31)
h 2 i
~ 2
2m k − E(k) uG (k) + UG u0 (k) + U0 (k)uG (k) = 0 .
From this follows,

h i2
~2 2
2m k + U0 − E(k) = UG U−G = 0 . (6.32)
∗ . Therefore, introducing the eigenenergies E (k) of
Since the potential U (r) is real, U−G = UG 0
free electrons (6.24),
E(k) = E0 (k) ± |UG | . (6.33)
Under the influence of the periodic perturbation potential we find at the surfaces of a Brillouin
zone an energy splitting developing a forbidden gap in the spectrum. We can understand this
observation as follows: In the crystal all electronic waves with wavevectors ending on a surface
of a Brillouin zone are reflected by Bragg reflection. In the example of a one-dimensional lattice
we understand that the superposition of an incident wave (k = nπ/a) with the reflected one
(k = −nπ/a) produces a standing electronic probability density wave ρ being proportional to
ρ1 ∝ cos2 nπ/a or ρ2 ∝ sin2 nπ/a. The charge density ρ1 is maximal at the location of the atom
in this site, which corresponds to an increased interaction energy; the density ρ2 is minimal at
the location of the atom. This explains the splitting.
The Bloch model can explain many properties of metals, semiconductors and insulators.
6.1.3 Application to one-dimensional optical lattices

In the following, we restrict ourselves to a one-dimensional potential, V (z) = V (z + a), acting on
(the center-of-mass of) atoms. Such a potential can be generated by two counterpropaganting
plane wave laser beams with wavevectors kL and −kL and tuned to the red side of an atomic
transition. In this situation the atoms are attracted to the maxima of the light intensity, the
antinodes. Therefore, we can write the potential as V (z) = − V20 |eikL z + e−ikL z |2 = −V0 (1 +
cos 2kL z) or, by letting K = 2kL ,
V (z) = −2V0 cos2 Kz . (6.34)

P
In the Fourier expansion, V (z) = K UK eiKz , this potential corresponds to the Fourier coeffi-
cients U0 = −V0 and U±K = − V20 . We also expand the wavefunction into plane waves, ψ(z) =
P iqz
q cq e , and we insert these expansions into Schrödinger’s stationary equation Ĥψ = εψ,
yielding, " #
−~2 ∂ 2 X X X
iKz
2
+ UK e cq eiqz = ε cq eiqz . (6.35)
2m ∂z q q
K
Defining q = k + nK, where k ∈ [−K/2, K/2] and n ∈ Z,

h 2 i
~ 2 1 1
2m (nK + k) − V 0 cnK+k − 2 V0 cnK+k−K − 2 V0 cnK+k+K = εcnK+k . (6.36)
In matrix notation,
M̂ c = εc . (6.37)
where the matrix is around n = .., −1, 0, +1, ..:
.   
.. ..
   . 
~2


2
2m (k − K) − V0 − 12 V0 

ck−K 
 
M̂ = 
 − 21 V0 ~2 2
2m k − V0 − 12 V0 
 with c=  ck   .
 ~2   
ck+K 
1 2
 − 2 V0 2m (k + K) − V0 
.. .
..
.
(6.38)
For shallow potentials, V0 ~ K /2m, we can neglect the coefficients V0 in the Eq. (6.36)
2 2
and we find,
ε ≈ ~2 q 2 /2m , (6.39)
which corresponds to the dispersion relation for free particles. On the other hand, looking at
the bottom of deep potentials, V0 ~2 K 2 /2m,pwe can harmonically
√ approximate the cosine
m 2 2 −1
potential by V (z) ≈ −2V0 + 2 ω z with ω = K V0 /m = ~ 2V0 Er . For this case we expect,

ε ≈ −2V0 + ~ω n + 21 . (6.40)
The exact spectrum of eigenvalues ε can be calculated by numerically determining the eigenvalues
of the matrix (6.38) for the first Brillouin zone, k ∈ [−K/2, K/2], and the above limits are
confirmed.
To estimate the width of the forbidden band, we cut out a 2 × 2 matrix within the matrix
M̂ and neglect its coupling with the others submatrices,
!
~2 2−V 1
(k − K) 0 − V
2 0
M̂s = 2m ~2 2 . (6.41)
− 12 V0 2m k − V0
At the edges of the Brillouin zone, k = 12 K, we get the eigenvalues ε = ~ mK − V0 ± V20 , that is,
2 2
the band gap is ∆ε = V0 2 . Bloch’s theorem says that Schrödinger’s equation can be solved for
any Bloch states. These are superpositions of plane wave momentum states [10],
ψk (z) = eikz uk (z) , (6.42)

2
For Bose-Einstein condensates, the procedure should be generalized taking into account the energy of the
mean field.
Figure 6.2: (Code: QM Lattices BlochBands.m) (Continuous red line:) Bloch Bands. (Dot-
ted black line:) Without potential, V0 = 0. The parameters are ωr = (2π) 20 kHz,
ωho = (2π) 12 kHz, λL = 689 nm, and V0 = 0.2~ωr .
with uk (z) = uk (z + a).

The requirement that ψ(z) satisfies the Schrödinger equation is equivalent to the condition
that c satisfies an eigenvalue equation. Let U be the matrix of the eigenvectors of M̂ and Ê the
diagonal matrix of eigenvalues: M̂ = U −1 ÊU gives ÊU c = εU c, such that U c can be understood
as eigenvectors.
Alternatively, we define dnK+k ≡ cnK+k+K /cnK+k , consequently Eq. (6.36) becomes,
V0
dnK+k−K = . (6.43)
~2
m (nK + k)2 − 2ε − V0 (2 + dnK+k )
6.1.4 Bloch oscillations

A Bloch oscillation is a phenomenon in solid state physics. It is the oscillation of a particle
(e.g., an electron) confined to a periodic potential (e.g., a crystal), when a constant force (e.g.,
generated by a continuous electric field) acts on it. This phenomenon is very difficult to observe
in solid crystals because, due to electron scattering by defects of the lattice [57, 203], the coherent
evolution is limited to a small fraction of the Brillouin zone. However, Bloch oscillations were
observed in semiconducting superlattices, in ultrathin Josephson junctions and with cold atoms
in optical lattices [115, 169].
Let us first show a simple treatment for electrons subject to a constant electric field E. The
one-dimensional equation of motion is,
dk
~ = −eE , (6.44)
dt
with the solution,
eE
k(t) = k(0) − t. (6.45)
~
The velocity v of the electron is given by,

1 dE
v(k) = , (6.46)
~ dk
where E(k) denotes the dispersion relation for a given energy band. We now assume that it has
the following form (tight-binding limit),
E = A cos ak , (6.47)
where a is the lattice parameter and A a constant. Then, v(k) is given by,
Aa
v(k) = − sin ak , (6.48)
~
and the position of the electron by,
Z
A aeE
x(t) = v(k(t))dt = − cos t . (6.49)
eE ~
This shows, that the electron is oscillating in real space. The oscillation frequency is given by,
ae|E|
ωB = . (6.50)
~
6.1.4.1 Bloch oscillations of atoms in optical lattices

Neutral atoms in a lattice can be accelerated by gravitation. To treat this case, we simply
replace the electric force −eE in the above calculation by the gravitational force mg, and we
obtain the result,
mgλL
ωB = , (6.51)
2~
with the wavelength of the retroreflected laser λL = 2a generating the standing wave.
To reproduce the dynamics of the matter wave, we start from the time-dependent Schrödinger
equation with the same periodic potential. We now expand the time-dependent wavefunction
into plane waves. By inserting this ansatz into the Schrödinger equation, we obtain a set of
equations for the expansion coefficients cn , which can be simulated numerically.
Now, we also allow an external force whose potential can be added to Schrödinger’s potential.
The additional term can be removed by a transformation into the moving frame. This modifies
the equations of motion for the population amplitudes of the momenta cn ,
∂ ψ̂ ~2 ∂ 2 ψ̂ ~W0
i~ =− + sin(2kl x)ψ̂ − mgxψ̂ . (6.52)
∂t 2m ∂x2 2
The additional term, which contains the frequency of the Bloch oscillation νb , increases
linearly over time. As time goes by, a resonance is crossed when t = −nτb , and the crossing
is periodically repeated at every n = −1, −2, 0, ... Tracing the matter wave evolution in the
laboratory system, we see that whenever the resonance is crossed, the momentum undergoes
a change of sign corresponding to a reflection of its motion. We expand the population of the
momentum states into plane (Bloch) waves with |cn (t)|2 [219, 220],
∞
X
ψ̂(x, t) = cn (t)e2inkl x · eimgxt/~ (6.53)
n=−∞
transform into the moving frame, and obtain the time dependent solution with the usual defini-
~k2
tion of the recoil frequency ωr = 2ml and
g
νb = . (6.54)
ωr
The equations of motion are now written,

dcn
dt = −4iωr (n + νb t)2 cn + W0
2 (cn+1 − cn−1 ) (6.55)
and the center-of-the mass momentum is,

X
hpilab = n|cn (t)|2 + νb t (6.56)
n
Figure 6.3: (Code: QM Lattices BlochDynamics.m) Dynamics of Bloch’s oscillations.
Here we give another pictorial representation of the Bragg reflection process. The require-
ment for commensurability of the Broglie wavelength and wavelength of the standing light wave
is equivalent to saying that the matter wave momentum is equal to the recoil of a single photon.
In other words, the matter wave is on the edge of a Brillouin zone. In fact, the dispersion relation
of a free particle is distorted due to the periodicity of the potential generated by the standing
light wave such as to open a forbidden band. As a consequence, instead of being accelerated
without limits, the atom enters the second Brillouin zone, which is to say that it is reflected to
the other side of the first Brillouin zone.
An important question concerns the origin of the interaction between the matter wave and
the standing light wave? An interaction is only possible if the atom has an internal transition
capable of scattering photons from the light beams. As any absorption and emission process
transfers a recoil momentum to the atom, we can understand the process as a Raman scattering
process: a photon of the laser beam coming from the left is absorbed and re-emitted to the left.
This transfers twice the photonic recoil to the atom, which explains the Bragg reflection of the
atom by the optical lattice.
Of course there are some conditions that need to be met to observe Bloch oscillations. The
transfer of momentum is efficient only in the rapid adiabatic passage (ARP) regime characterized
by the conditions 2(νb /ωr ) (W0 /4ωr )2 16. The first condition requires that the force that
drives the atoms to perform the Bloch oscillations must be weak enough to avoid transitions
between Bloch bands, which guarantees the adiabaticity of the process. The other condition
requires that the optical lattice be weak enough so that the dynamics involves only two adjacent
momentum states at the same time and the transfer between the two is successful.
1.8
1.6
1.4
1.2
En /h̄ωr
0.8
0.6
0.4
0.2
0 W0
-0.2
-0.5 0 0.5
q/2ki
Figure 6.4: Illustration of the first Brillouin zone.
6.2 The Kronig-Penney model

The Kronig-Penney model describes the band structure of a lattice. Let us assume a periodic
potential of rectangular wells with valleys of widths a and peaks of widths b,
V (x) = U0 θmod(x,a+b)∈[a,a+b] . (6.57)
Inserting into the Schrödinger equation the plane wave ansatz ψ = AeiKx + Be−iKx for the
wavefunction in the valley, 0 < x < a, and ψ = CeQx + De−Qx in the peak, −b < x < 0,
we obtain ε = ~2 K 2 /2m and U0 − ε = ~2 Q2 /2m. Choosing the constants A, B, C, D such
that ψ and ψ̇ are continuous in x = 0, a, we derive, using the periodicity of the Bloch wave
ψ(a < x < a + b) = ψ(−b < x < 0)eik(a+b) ,
  
1 1 −1 −1 A
 iK −iK −Q Q  B 
   = 0 . (6.58)
 eiKa e−iKa −e−Qb+ik(a+b) −eQb+ik(a+b)   C 
iKeiKa −iKe−iKa −Qe−Qb+ik(a+b) QeQb+ik(a+b) D
The determinant of the matrix must be zero, or,
Q2 − K 2
sinh Qb sin Ka + cosh Qb cos Ka = cos k(a + b) . (6.59)
2QK
For δ-shaped peaks, we let b = 0 and U0 = ∞ such that Q2 ba/2 = P , this simplifies to,
P
sin Ka + cos Ka = cos ka . (6.60)
Ka
The dispersion relation for light is different. According to [172],
√ √ −1
ε+ ε √ √
− sin( εωa/c) sin(ωa/c) + cos( εωa/c) cos(ωa/c) = cos ka . (6.61)
2
For ε = 1 the equation simplifies to ka = 2ωa/c.
6.3. EXERCISES 163
6.2.0.1 Photonic density of states

The photonic density of states in free space in three dimensions is evaluated from [245],
3 Z 3 Z
L 3 L 2 L3 ω 2
dN = 2 d k=2 k dk dφd cos θ = 2 3 dω ≡ D(ω)dω . (6.62)
2π 2π π c
In one dimension,
L L
dN = 2 dk = dω . (6.63)
2π πc
R πc/L
The density is normalized 0 D(ω)dω = 1 and the total energy, if all states are populated,
R πc/L
is E ≡ 0 ωD(ω)dω = πc/L. However, this applies only if ω = ck. If the dispersion relation
is more complicated, for example, inside a cavity or a forbidden photonic band, ω = ω(k), we
must generalize,
L dk
D(ω) = . (6.64)
π dω
Assuming that the dispersion relation is given by the Kronig-Penney model, we obtain gaps in
the density-of-states for those values of ω which do not belong to any k.
ω0
Re θ ρ(ω) ω0
Figure 6.5: Dispersion ratio and state density for a one-dimensional optical lattice.
6.3 Exercises
6.3.1 The Bloch model
6.3.1.1 Ex: Gaussian approximation for Wannier function
Consider a stationary light wave producing a dipolar potential of the V (x) = W0 sin2 (kx) =
W0 W0 2
2 − 2 cos 2kx com W0 Er = (~k) /2m.
a. Approximate the potential of a single site around x = 0 for a harmonic potential.
b. Calculate for which depth of the dipolar potential the approximate potential supports at
least one connected state. c. Calculate the spacing of the levels and the length of the harmonic
oscillator aho .
d. Use the appropriately normalized ground state of this harmonic oscillator as an approximation
for the Wannier function ω0,0 (x). It assumes the same depth of the network in all directions of
3/4
a 3D cubic network. Derive the formula U 3d = π8 kad W Er
0
starting from,
Z
3D 3D 4
U =g ω(0,0) (x, y, z)d3 r
4π~ad
com g 3D = m .
(a) (b)
hw
(c)
E0
W0
Figure 6.6: (a) Approximation of a periodic potential by a potential harmonic. (b-c) An infinite
periodic potential should be considered as free or confining?
6.3.1.2 Ex: Perturbative treatment of a weak lattice

A weak lattice potential with V0 < Er can be treated in perturbation theory to motivate the
resulting opening of a gap in the refolded energy parabola. The unperturbed Hamiltonian
Ĥ0 = p2 /2m contains only the kinetic energy and the perturbation is V (x) = V0 sin2 (kx) =
1 1 2ikx + e−2ikx ).
2 V0 − 4 V0 (e
a. Calculate V̂ (x)φp (x) and show that hφp±~k |V̂ |φp i are the only non-zero matrix elements of
the perturbation V (x) between the eigenstates of Ĥ0 (which are the orthonormal plane waves
φp = eipx/~ ). Neglect the constant term of the potential, which only yields a global energy shift.
b. This coupling is relevant around those momenta p, where φp has the same energy φp+~k or
φp−~k . Show that these momenta are p = ∓~k.
c. Consider the perturbed system restricted to the basis {|p = −~ki, |p = +~ki} and give the
Hamiltonian as 2x2 matrix.
d. Diagonalize the matrix and consider the difference of the eigenenergies. Use them to estimate
the size of the gap, that the lattice opens between the two lowest bands.
e. Calculate the eigenstates and interpret them by comparing the probability density to the
lattice potential.
6.3.2 The Kronig-Penney model

Part II
Atomic and Molecular Physics
165
Chapter 7
The electron spin and the atomic

substructure
The energy structure of hydrogen calculated by Bohr’s model from the non-relativistic Hamilto-
nian agrees very well with the experimental measurements. However, in high-resolution exper-
iments, small deviations were observed as energy shifts and splittings of spectral lines. These
deviations, called fine structure, were not predicted by theory, which suggests that there are
weak additional effects that do not strongly affect the position of the spectral lines but remove
the energy degeneracy of the orbital quantum number `: E = En,` .
As a possible explanation we have the fact that the electrons present relativistic mass and
momentum. In order to estimate the relevance of relativistic corrections let us estimate the
electron velocity in the fundamental hydrogen states given by E1 = −~2 /2me a2B . Using the
definitions of the Bohr radius, aB = 4πε0 ~2 /(me e2 ), and the fine structure constant
e2 1
α≡ ' , (7.1)
4π0 ~c 137
we obtain, r
2E1 ~ e2
v= = = = αc , (7.2)
me me aB 4πε0 ~
which shows that the electron velocity is very high and that relativistic effects may indeed be
not negligible.
7.1 The Dirac equation

7.1.1 The Klein-Gordon equation for bosons
The Schrödinger equation for a free particle is based on the non-relativistic energy-momentum
dispersion relation,
p2
E= . (7.3)
2me
and the definitions of the quantum operators for energy and momentum,
∂
E = i~ and p = −i~∇ . (7.4)
∂t
As already discussed in Sec. 2.1.2 we can, in order to find a relativistic wave equation, try the
approach of inserting the quantum operators into the relativistic energy-momentum relation 1 .
E 2 = c2 p2 + m2e c4 . (7.5)
1
Using the covariant notation with pµ ≡ (E/c, p): pµ pµ = E 2 /c2 − p2 = m2e c2 is a Lorentz invariant.
167
168 CHAPTER 7. THE ELECTRON SPIN AND THE ATOMIC SUBSTRUCTURE
We obtain,
" 2 #
1 ∂2 me c2
− ∇2 + ψ=0 . (7.6)
c2 ∂t2 ~
This is the Klein-Gordon equation. The stationary solution of this equation is a spherical
wave,
1
ψ = ψ0 e−2πr/λC , (7.7)
r
where λC = h/me c is the Compton wavelength. We show this in Exc. 7.5.1.1. For example, in
the case of heavy bosonic particles, such as a field of π-mesons, ψ is the Yukawa potencial.
In the framework of the standard model, it is believed that matter is composed of two
fundamental types of particles, bosons and fermions. Bosons are exchanged between fermions
conveying the interaction between them. A typical example is the one of two electrons whose
Coulomb interaction is mediated by the exchange of photons. Bosons obey the Klein-Gordon
equation, fermions the Dirac equation derived in the following section.
7.1.2 The Dirac equation for fermions

In 1928 Paul Dirac, at the age of 26, developed an approach to a relativistic wave equation
which differed from the Klein-Gordon equation. Motivated by the observation that the photon,
being the relativistic particle par excellence, obeys a linear energy-momentum relation of the
form ω = ck, he attempted to derive a linear dispersion relation in E and p for heavy particles
via the following ansatz:
E = α0 me c2 + α1 cpx + α2 cpy + α3 cpz . (7.8)
Replacing energy and momentum with their respective operators 2 ,

∂ 2 ∂ ∂ ∂
i~ φ = α0 me c φ − ic~ α1 + α2 + α3 φ. (7.9)
∂t ∂x ∂y ∂z
We must now ensure that the relativistic energy-momentum condition (7.5) be satisfied.
Example 31 (Derivation of the Dirac equation): Taking the square on the right-hand
side of the equation (7.9),
[α0 me c2 − ic~(α1 ∂x + α2 ∂y + α3 ∂z )][α0 me c2 − ic~(α1 ∂x + α2 ∂y + α3 ∂z )]

= m2e c4 α02 − ic~me c2 [(α0 α1 + α1 α0 )∂x + (α0 α2 + α2 α0 )∂y + (α0 α3 + α3 α0 )∂z ]
− c2 ~2 [α12 ∂x2 + α22 ∂y2 + α32 ∂z2 ]
− c2 ~2 [(α1 α2 ∂x ∂y + α2 α1 ∂y ∂x ) + (α2 α3 ∂y ∂z + α3 α2 ∂z ∂y ) + (α3 α1 ∂z ∂x + α1 α3 ∂x ∂z )] .
For this expression to be identical to the relativistic energy-momentum condition (7.5),
m2e c4 − c2 ~2 [∂x2 + ∂y2 + ∂z2 ] ,
we need to postulate for all i = 0, .., 4, that αi αj + αj αi = 2δij .

2 ∂
We introduce the abbreviation ∂k ≡ ∂xk
7.1. THE DIRAC EQUATION 169
Obviously, the condition

[αi , αj ]+ = 2δij (7.10)
can not be satisfied if the αi are numbers. The idea of Dirac was to interpret the variables αi
as matrices. These matrices act as operators on appropriate states, which are no longer scalar
wavefunctions but vectors. Each component of the vector is a wavefunction in the usual sense.
The Hilbert space is extended to be the product space of the usual spatial wavefunctions and a
finite-dimensional vector space.
Example 32 (Calculation with matrices of operator ): To give an idea of how the
algebra works we consider a general situation. As the operator we choose the product,

0 1 ∂
1 0 ∂x
and as the wavefunction vector we choose,

eik1 x
.
eik2 x
Applying the operator on the state vector we get,
∂
ik x ∂ ik2 x

0 1 ∂ eik1 x 0 ∂x e 1 0 + ∂x e ik2 eik2 x
= ∂ = ∂ ik1 x = .
1 0 ∂x eik2 x ∂x 0 eik2 x ∂x e +0 ik1 eik1 x
The matrices αi must satisfy the condition (7.10). It is possible to show that this requires
at least four-dimensional matrices of the following form:

I 0 0 σj
α0 = and αj = , (7.11)
0 −I σj 0
where j = x, y, z = 1, 2, 3. In this notation the components of the matrices are also matrices:

1 0 0 1 0 −i 1 0
I= , σx = , σy = , σz = , (7.12)
0 1 1 0 i 0 0 −1
called Pauli spin matrices. The state vector must also have four dimensions,

~ t)
~~
Φ(r, t) =
φ(r,
with ~ t) = φ1 (r, t)
φ(r, and χ
~ (r, t) =
χ1 (r, t)
. (7.13)
χ
~ (r, t) φ2 (r, t) χ2 (r, t)
Φj are called large components, χj are called small components. This designation is explained
later. Combining the matrices αj to a three-dimensional vector α~ , we can now write the Dirac
equation (7.9) like,
~ ~
~ t) = me c2 α0 + c~
i~∂t Φ(r, ~ t) .
α · p Φ(r, (7.14)
Or, using the notation (7.11) and combining the Pauli matrices σj to a three-dimensional vector
~σ , we can write the Dirac equation as:

∂ φ ~ 2 I 0 0 ~σ · p ~
φ
i~ = me c +c . (7.15)
∂t χ
~ 0 −I ~σ · p 0 χ
~
The non-diagonal matrix,

0 ~σ · p
(7.16)
~σ · p 0
couples large and small components. In Exc. 7.5.1.2 we derive an important rule for the calcu-
lation with Dirac and Pauli matrices.
Example 33 (Covariant and relativistically invariant form of Dirac’s equation): To

demonstrate its relativistic invariance it is useful to rewrite the Dirac equation in a way in
which time and space appear on equal footings. For this we introduce new matrices,
γ 0 ≡ α0 and γ k = γ 0 αk . (7.17)
We obtain,
I 0 0 σk
γ0 = and γk = . (7.18)
0 −I −σk 0
We also define another important matrix by,

0 I
γ5 ≡ iγ0 γ1 γ2 γ3 = . (7.19)
I 0
With this, using Einstein’s notation 3 , the Dirac equation (7.13) adopts the form,
i~γ µ ∂µ ψ − mcψ = 0 . (7.20)
The complete system is summarized in the Minkowski metrics of time-space in the form,
[γ µ , γ ν ]+ = 2η µν , (7.21)
for µ, ν = 0, .., 5, that is, all matrices γk anticommute.

The Dirac equation can now be interpreted as an eigenvalue equation, where the rest mass
is proportional to the eigenvalue of a momentum quadrivector, the proportionality constant
being the speed of light:
Pop ψ = mcψ , (7.22)

Using ∂ in the Feynman slash notation, which includes the γ-matrices, as well as a summation
over the components of the spinor in the derivative, the Dirac equation becomes:

i~∂ ψ − mcψ = 0 . (7.23)
A fundamental theorem states that, if two distinct sets of matrices are given, which both
satisfy Clifford’s relations, then they are connected to each other by a similarity transforma-
tion:
γ 0µ = S −1 γ µ S . (7.24)
If, in addition, the matrices are all unitary, as is the case of Dirac’s set, then S is unitary,
γ 0µ = U † γ µ U . (7.25)
7.1.2.1 Anti-particles
Disregarding for a moment the non-diagonal matrix, the Dirac equation separates into two
independent equations,
~
∂φ ∂~
χ
i~ ~
= me c2 φ and i~ = −me c2 χ
~ . (7.26)
∂t ∂t
These are eigenenergy equations with the eigenvalues me c2 and −me c2 . The state with negative
energy is interpreted as anti-particle. Therefore, the non-diagonal matrix mixes particles and
anti-particles.
3
∂0 ≡ 1c ∂t
7.1.2.2 Particles and anti-particles in the non-relativistic limit

To reduce the Dirac equation to the non-relativistic Schrödinger equation, we first need to get
rid of the rest energy. To do so, we separate a fast oscillation, whose frequency corresponds to
the rest mass of the electron via the following ansatz, where u and v vary slowly in time:

~
~ −iω0 t u(r, t)
Φ(r, t) = e , ~ω0 = me c2 , (7.27)
v(r, t)
with the temporal derivative,

˙
~ 2 u u̇
~
i~Φ = me c + i~ e−iω0 t . (7.28)
v v̇
We insert this into the Dirac equation,

2 u u̇ −iω0 t u v
me c + i~ e = me c2
+ c~σ · p e−iω0 t (7.29)
v v̇ −v u
finally obtaining,
i~u̇ = c(~σ · p)v , i~v̇ = c(~σ · p)u − 2me c2 v . (7.30)
Since u and v only vary slowly in time, the derivatives on the left-hand side are small quantities.
However, the condition that both derivatives must zero is too strong, because it leads to the
trivial solution u = 0 and v = 0. We find the first non-trivial solution by the condition v̇ = 0.
The second equation then becomes,
1
v= (~σ · p)u . (7.31)
2me c
Inserted into the first equation,
(~σ · p)2
i~u̇ = c u. (7.32)
2me c
We need, therefore, to evaluate the expression (~σ · p)2 ,

pz px − ipy 2 2 1 0
~σ · p = and (~σ · p) = p . (7.33)
px + ipy −pz 0 1
Inserted into the differential equation (7.32) for u we obtain precisely the Schrödinger equation
for a free particle,
p2
i~u̇ = u. (7.34)
2me
Let us return to the question, why we call u the strong component. We have from the equation
(7.31),
1 1 p2 †
v† v = (~
σ · p) 2 †
u u = uu, (7.35)
(2me c)2 2me c2 2me
2
p
and since 2m e
me c2 follows immediately v† v u† u.
In this non-relativistic approximation the components u are much larger than the components
p2
v. The mixture between particles and antiparticles only matters when 2m e
' me c2 , resp.,
1 2 2
2 me v ' me c or |v| ' c. The electron only receives small positronic contributions as it
approaches the speed of light. In the ground state of the hydrogen atom the electron has a
velocity of of v = αc ' c/137. That is, the contribution of the weak components is small, but
present.
Example 34 (Vanishing rest mass): Let us note that for the case of vanishing rest mass,
me = 0, the Dirac equation (7.15) dramatically simplifies. Taking the time derivative of the
upper equation (7.15) and inserting the lower equation (7.15), we find,
1 ∂2 ~ 1 c~σ · p ∂ (~σ · p)2 ~ p2 ~ ~.

φ = χ
~ = − φ = − φ = ∇2 φ (7.36)
c2 ∂t2 c2 i~ ∂t ~2 ~2
I.e. we recover a Helmholtz type wave equation.
7.1.2.3 The spin

We consider the operator defined by [76, 77],
ŝ ≡ ~2 ~σ , (7.37)
and we calculate the commutation relations between its components. From the definitions (7.12)
we obtain the rule,

~2 0 1 0 −i ~2 0 −i 0 1 ~2 2i 0
[ŝx , ŝy ] = 4 − 4 = = i~Ŝz . (7.38)
1 0 i 0 i 0 1 0 4 0 −2i
In general terms the following holds true: [si , sj ] = ijk i~sk . It is interesting to compare this with
the commutation relation for the orbital angular momentum [ˆli , ˆlj ] = ijk i~ˆlk . The coincidence
suggests a generalization of the concept of angular momentum: We now call angular momentum
operator every three-dimensional vector operator satisfying this commutation relation 4 . We
consider the eigenvalue equation for Ŝz ,

~ 1 0 φ1 φ
~
Ŝz φ = 2 = ms ~ 1 . (7.39)
0 −1 φ2 φ2
The eigenvalues are obviously ms = ± 12 . The angular momentum related to the matrices S is
obviously half-integer. We are dealing here with a new type of angular momentum, which is
not included in the usual definition of orbital angular momentum L = r × p. The new angular
momentum is called intrinsic angular momentum or spin of the particle. The spin represents
a new structure or dimension additional to space comparable to the polarization of light. The
photons of a circularly polarized light beam also contribute to an intrinsic angular momentum,
which however in this case is integer.
In Exc. 7.5.1.3 we will see that neither ˆlz nor ŝz are constants of motion of the Hamiltonian
(7.15), but the sum ĵz ≡ ˆlz + ŝz ,
[ĵz , Ĥ] = 0 . (7.40)
7.1.2.4 The stationary Dirac equation

By a similar treatment as in the Schrödinger equation one can deduce a stationary Dirac equation
(7.14) via a separation of the time variable. Making for the time an exponential ansatz,
~ t) = φ(r)e
φ(r, ~ −iEt/~
and χ ~ (r)e−iEt/~ ,
~ (r, t) = χ (7.41)
we obtain coupled stationary equations for the large and small components,
~
(E − me c2 )φ(r) = cσ · p~
χ(r) and (E + me c2 )~ ~
χ(r) = cσ · pφ(r) . (7.42)
4
This concept can be derived from the requirement of symmetry under rotation of space as discussed in Sec. 2.5.
7.1.3 The relativistic electron in a central Coulomb field

7.1.3.1 Minimal coupling
In atomic physics we are mainly interested in electrons bound to a potential (e.g., generated by
an atomic nucleus), that is, we must introduce electromagnetic forces into the Dirac equation.
Therefore, we now consider the interaction of a charged particle with an electromagnetic field
given by the vector potential A and by the electrostatic potential U , such that the electric and
magnetic fields,
∂A
E = −∇U − and B=∇×A , (7.43)
∂t
allow to calculate the Coulomb-Lorentz force. In the Hamiltonian formulation of electrodynamics
the interaction can be described simply by the transition 5 ,
p̂ −→ p̂ − qA ≡ ~π and Ĥ −→ Ĥ + qU . (7.44)
called the minimal coupling. These rules will be derived in Sec. 8.1. In addition to the substitu-
tion of the momentum, we must add the scalar potential qU , and we obtain the Dirac equation
for a particle inside an applied electromagnetic field,
˙
~ ~
~ = me c2 α0 + c~
i~Φ ~ .
α · ~π + qU Φ (7.45)
7.1.3.2 Solving the stationary Dirac equation

Let us, for now, disregard external magnetic fields, A = 0. Then, the stationary Dirac equation
becomes,
~
[E − qU (r) − me c2 ]φ(r) = cσ · p~
χ(r) and [E − qU (r) + me c2 ]~ ~
χ(r) = cσ · pφ(r) . (7.46)
For the Coulomb potential,

1 e2
qU (r) = − (7.47)
4π0 r
the Dirac equation can be solved algebraically [90] 6 . The calculation is more complicated
than the resolution of the Schrödinger equation for hydrogen derived in Secs. 4.1.4 and 4.2.1.
Therefore, we only give the result here:
q
me c2
En,j = r 2 with δj ≡ j + 1
2 − (j + 12 )2 − α2 and j = ` ± 12 , (7.48)
α
1 + n−δ j
with ` = 0, 1, ...
The energy now depends on two quantum numbers. The degeneracy of the orbital angular
momentum j is lifted, and the new quantum number besides the main quantum number n is that
of the total angular momentum j. The intransparent expression for the energy can be expanded
by α,
2 α2 α4 1 3
En,j ' me c 1 − 2 − 3 − . (7.49)
2n 2n j + 1/2 4n

5 E/c U/c
In quadrivetorial notation: πµ = pµ − qAµ with pµ = and Aµ = .
p A
6
See also http://einstein.drexel.edu/∼bob/Term Reports/Whitehead 3.pdf
The second term reproduces the energy of Bohr’s model, but there are correction terms pro-
portional to α4 . We will show in Secs. 7.1.4, that the energy levels, called fine structure, result
from several relativistic corrections of different origins. We will transform the Dirac equation
into spherical coordinates in Exc. 7.5.1.4 and derive its solution (7.48) in 7.5.1.5 and 7.5.1.6.
In the expression (7.49) for the electron energy in the Coulomb potential, the last term is
positive and proportional to 1/n4 . It describes relaxation of the binding due to the contribution
of weak components. The term containing the quantum number j is called the spin-orbit cou-
pling. To better understand this contribution we must first analyze more deeply the matrices
~σ .
7.1.3.3 Dirac’s Hamiltonian in the sub-relativistic limit

Defining the energy E 0 = E − me c2 , the stationary Dirac equation (7.46) for an electron of
charge q = −e in an external electrostatic potential U (r) can be written,
~ = c~σ · p~
[E 0 − qU (r)]φ χ and [E 0 − qU (r) + 2me c2 χ ~.
χ = c~σ · pφ
~ ]~ (7.50)
resolving the second equation for the wavefunction χ

~ and substituting it into the first,

0~ ~ + ~σ · p 1 E 0 − qU (r) −1 ~ .
E φ = qU (r)φ 1+ ~σ · pφ (7.51)
2me 2me c2
In the non-relativistic limit,

p2
E 0 − qU ' me c2 , (7.52)
2me
we get,
0 ~ + ~σ · p 1 E 0 − qU (r) ~.
E ' qU (r)φ 1− ~σ · pφ (7.53)
2me 2me c2
Now, ~σ · p is an operator entity, which acts on the subsequent operators and wavefunctions. We
thus have to apply the product rule, (~σ · p)V ψ = V (~σ · p)ψ + [(~σ · p)V ]ψ, to the first occurrence
of operator this operator in equation (7.51),

E0φ ~ + 1 1 − E 0 −qU 2(r) (~σ · p)2 φ
~ ' qU (r)φ ~ + q2 2 [(~σ · p)U (r)](~σ · p)φ
~. (7.54)
2me 2me c 4m c e
In the following we will make use of the general relationship,
(~σ · B)(~σ · C) = (B · C) + i~σ · (B × C) , (7.55)
which holds for [~σ , B] = 0 = [~σ , C] and has been demonstrated in Exc. 7.5.1.2. The relationship
yields,
(~σ · p)2 = p2 and [~σ · pU (r)](~σ · p) = pU (r) · p + i~σ · [pU (r) × p] , (7.56)
so that,

~ ' qU (r)φ
~+ E 0 −qU (r) ~− ~2 q ~+ ~ ~ . (7.57)
E0φ 1
2me 1− 2me c2
p2 φ 4m2e c2
∇U (r) · ∇φ 4m2e c2
~σ · [∇U (r) × p]φ
Also, with U (r) = U (r),

∂U ∂U r ∂U ∂U ∂
∇U (r) = ∇r = and ∇U (r) · ∇ = êr · ∇ = . (7.58)
∂r ∂r r ∂r ∂r ∂r
We get, again applying the non-relativistic approximation (7.52),

0~ ~ 1 E 0 − qU (r) 2~ ~2 ∂qU ∂ ~ ~q 1 ∂U ~
E φ = qU (r)φ + 1− p φ− φ+ ~σ · r × pφ
2me 2me c2 4m2e c2 ∂r ∂r 4m2e c2 r ∂r
2
p p4 q 1 ∂U ~2 q ∂U ∂
' + qU (r) − + s·l− . (7.59)
2me 8m3e c2 2m2e c2 r ∂r 4m2e c2 ∂r ∂r
where we made use of the definitions s = ~2 ~σ and l = r × p. The term in the bracket can be
used as the Hamiltonian allowing to calculate the fine structure as first-order perturbations to
the non-relativistic energy levels obtained from non-relativistic theory,
p2 1 e2 p4 e 1 ∂U ~2 e ∂U ∂ ~
Ĥ ' − − − s · l − φ . (7.60)
2me 4πε0 r 8m3e c2 2m2e c2 r ∂r 4m2e c2 ∂r ∂r
The first two terms are those arising from Bohr’s atom model, the third one is a correction due to
the relativistic velocity of the electron, the forth comes from the electron’s spin-orbit coupling,
and the fifth is called the Darwin term. All contributions represent perturbations to the non-
relativistic Schrödinger theory of Bohr’s atom and will be discussed extensively in Secs. 7.2. We
will show in 7.5.1.7 that l̂2 , ŝ2 , and ĵ2 are constants of motion of the above Hamiltonian.
7.1.4 The Pauli equation

When we calculated the electron’s energy in the Coulomb potential (7.60), we only considered
the electrostatic potential of the nucleus, letting the potential vector A be zero. As long as we
do not apply an external magnetic field this is correct, because the internal magnetism of the
atom is already completely enclosed in the Dirac equation. On the other hand, we know that the
atom contains moving charges, that is, currents which generate magnetic fields 7 . Furthermore,
the spins of the electron and of the proton produce magnetic moments, which ought to interact
with the magnetic fields. Hence, the existence of magnetic effects in an atom is to be expected.
These magnetic effects can be discussed in a more transparent way applying the Schrödinger
equation with minimal coupling to electromagnetic fields (7.44) to a two-component spinor
~ This Schrödinger equation can be obtained from Dirac’s equation (7.50) via a stronger
φ.
non-relativistic approximation, which consists in completely neglecting the weak component
[E 0 − qU (r)]~
χ. The equation for the strong component (7.51) then becomes,
2
E0φ ~ + (~σ · ~π ) .
~ = [qU (r)]φ (7.61)
2me
We can again apply the formula (7.55) to calculate,
(~σ · ~π )2 ψ = ~π 2 ψ + i~σ · (~π × ~π )ψ = ~π 2 ψ + iq~σ · [−p × A(r) − A(r) × p]ψ (7.62)
= ~π 2 ψ − ~q~σ · {∇ × [A(r)ψ] + A(r) × ∇ψ}
= ~π 2 − ~q~σ · [∇ × A(r)]ψ = [p − qA(r)]2 ψ − ~q~σ · B(r)ψ .
In the case of an electron (e = −q) we obtain the so-called Pauli equation,

0~ 1 2 e~ ~,
Eφ= (−i~∇ + eA) + ~σ · B − eU (r) φ (7.63)
2me 2me
7
The spin of the electron does not generate a magnetic field, in contrast to the angular momentum caused by
its orbital motion. It only interacts with the environment through the requirement of symmetrization for being a
fermion.
~ with the Hamiltonian,

which corresponds to a Schrödinger equation for a two-component spinor φ
p2 1 e2 i~e e
Ĥ ' − − (∇ · A + A · ∇) + s·B , (7.64)
2me 4πε0 r 2me me
neglecting terms in A2 .
The Pauli equation serves for a classical (non-relativistic) approach to the electron’s spin-
orbit coupling, as we will see below and in the discussion of the fine structure in Sec. 7.2.2. In
also serves to calculate the coupling of the nuclear spin to the electron, as will be shown in the
discussion of the hyperfine structure in Secs. 7.3.
7.1.4.1 Dipole moment of the angular momentum and the spin

The rotational motion of a charge, −e, creates a current I, corresponding to a current density,
v
j(r0 ) = Iêφ δ(r − r0 )δ(z 0 ) = −e δ(r − r0 )δ(z 0 ) . (7.65)
2πr
Hence, the dipole moment caused by the circular motion of an electron is,
Z Z 2π Z ∞ Z ∞
1 0 3 0 1 0 0 −ev
µ
~l = r × j(r )d r = r× dφ dz r0 dr0 δ(r − r0 )δ(z 0 ) (7.66)
2 V 4π 0 −∞ 0 r
−1 −e
= 2 er × v = l,
2me
with the angular momentum l = r × me v. The quotient γe ≡ −e/2me is called gyromagnetic
ratio of the electron. We often use the Bohr magneton, µB ≡ ~e/2me , which represents the
elementary unit of spin,
µ
~l l
= −gl , (7.67)
µB ~
e~
where we introduced the Bohr magneton µB ≡ 2m e
as an abbreviation. The g-factor of a system
having any angular momentum l is defined as a proportionality constant between the normalized
dipole moment and the normalized angular momentum. gl ≡ `µµlB = 1 takes into account possible
corrections between our classical derivation and quantum mechanics.
7.1.4.2 Pauli’s model of spin-orbit coupling

The aim of this section is to demonstrate the relationship between the spin-orbit coupling term
in Dirac’s Hamiltonian (7.60) and the spin-magnetic field coupling term in Pauli’s Hamiltonian
(7.64).
A comparison of Pauli’s expression with the energy of a magnetic moment in the field B,
Ĥ`s = −~
µs · B , (7.68)
suggests the following connection between the spin and the magnetic moment:
e~ e
−~
µs · B = ~σ · B = s·B . (7.69)
2me me
We conclude, that the electron carries, besides mass, charge and spin, also a magnetic dipole
moment,
µ
~s e s
=− s = −2 , (7.70)
µB m e µB ~
7.2. FINE STRUCTURE OF HYDROGEN-LIKE ATOMS VIA TDPT 177
For the g-factor of the electron, we obtain ge = 2 8 . Neutron and proton are also fermions
with spin 21 , but they do not obey the Dirac equation! Their g-factors are gproton = 5.5858
and gneutron = −3.8261. The large deviation from g = 2 points to the existence of an internal
structure.
The rapid motion of the electron within the electrostatic field E of the nucleus produces,
following the theory of relativity, in the electron’s reference frame a magnetic field B0 with which
the electronic spin can interact. As we will show in 7.5.2.1, the field seen by the electron can be
approximated in first order in v/c by,
v
B0 ' ×E . (7.71)
c2
With this the interaction energy (7.68) becomes,
e e
µs · B0 =
Ĥ`s = −~ 2
s · (v × E) = − 2 2 s · (p × ∇U ) (7.72)
me c me c

e r dU 1 dV (r)
= − 2 2s · p × =− 2 2 s·l ,
me c r dr me c r dr
with V (r) = −eU (r).

The resulting interaction energy coincides, apart from a factor 12 [242], with the one obtained
in the from Dirac’s equation (7.60). The deviation, called Thomas factor, is due to the necessity
to transform back into the inertial system of the nucleus. This transformation, called Thomas
Precession, must be done by a Lorentz transformation, which is not trivial with electron contin-
uously changing its propagation direction on its circular orbit. The transformation introduces
an additional factor of 12 called Thomas factor 9 .
7.2 Fine structure of hydrogen-like atoms via TDPT

The wave equation that simultaneously satisfies the requirements of quantum mechanics and
special relativity is the Dirac equation. In free space including electromagnetic interactions it
describes all massive particles of semi-integer spin with parity as a symmetry, such as electrons
and quarks. It was the first theory to fully explain special relativity in the context of quantum
mechanics. The Dirac equation describes the fine structure of the hydrogen spectrum in a
completely rigorous manner. The equation also implied the existence of a new form of matter,
antimatter, previously unsuspected and unobserved. The equation also justifies a posteriori the
introduction of spinors, that is, of the vector wavefunctions introduced by Pauli in a heuristic
way. We have seen in the last section that, in the limit of high but non-relativistic velocities,
the Dirac equation adopts the form of a Schrödinger equation with the modified Hamiltonian
(7.60),
Ĥ = Ĥ0 + Ĥrl + Ĥ`s + Ĥdw + Ĥlamb (7.73)

2 2
4 2 2
p Ze p 1 1 dV π~ Ze 3
= − − 3 2
+ 2 2
l·s+ δ (r) + Ĥlamb .
2me 4πε0 r 8me c 2me c r dr 2m2e c2 4πε0
8 µs 2
The exat value is gs ≡ sµ B
= 2.002319314... The deviation gs − 2 ' α π
− 0.164 α
π2
is due to the coupling of
the spin to the fluctuations of the electromagnetic vacuum. We need to use quantum electrodynamical methods
to calculate the corrections.
9
This is a kinematic effect in space-time: the Lorentz transformations for systems moving with non-collinear
velocities can not simply be concatenated, but must be rotated, too [90, 139].
We will discuss the various terms in the following sections. Note that the expression for the
Darwin term differs from that of (7.60). We will see in Exc. 7.5.1.8, that they are, in fact,
equivalent.
7.2.1 Correction for relativistic velocities

The first correction in the expression, Ĥrl in Eq. (7.73), comes from the expansion of the rela-
tivistic energy for small velocities up to second order,
p p2 p4
Ekin = p2 c2 + m2e c4 ' me c2 + − + ... . (7.74)
2me 8m3e c2
The correction is of the order of magnitude,
p4
Hrl 8m3e c2 v2 α2
= p2
= ' ≈ 0.01% . (7.75)
H0 4c2 4
2me
Due to the degeneracy of these states, it would be appropriate to use perturbation theory
with degenerate states. However, as Ĥrl only depends on spatial coordinates commuting with l
and s, the degeneracy is not very important, since Ĥrl is already diagonal in the base |n, `, mi,
that is, hn, `, m|n0 , `0 , m0 i = δ``0 δmm0 . Starting from,
2 2 2
p4 1 p 1 Ze2
Ĥrl = − 3 2 = − =− Ĥ0 + (7.76)
8me c 2me c2 2me 2me c2 4πε0 r
2
1 2En n2
=− Ĥ0 − ,
2me c2 r̃
Zr
with r̃ ≡ aB and using as an abbreviation the energies of hydrogen following Bohr’s model,
Z 2 e2 1 me c2 Z 2 α2
En = hn, `|Ĥ0 |n, ì = − = − . (7.77)
4πε0 2aB n2 2 n2
We have
" 2 #
1 4E n n 2E n n
∆Erl = hn, `|Ĥrl |n, ì = − hn, `|Ĥ02 |n, ì − hn, `| Ĥ0 |n, ì + hn, `| |n, ì
2me c2 r̃ r̃
" #
Z 2 α2 2 2 2 1 2 4 1
= En − 4En n 2 + 4En n 3 , (7.78)
4En n2 n n (` + 21 )
using the eigenvalues calculated in (4.49). Finally, we obtain the following relativistic correction,
" #
2 1 3
∆Erl = En (Zα) 1 − 4n2 . (7.79)
n(` + 2 )
Obviously, the degeneracy with respect to the angular momentum ` is lifted by this correction.
7.2.2 Correction due to spin-orbit coupling

The second correction, Ĥ`s in the expression (7.73), called spin-orbit interaction, is a relativistic
correction due to the fact that the electron moves rapidly within the electrostatic field E gener-
ated by the nucleus. Considering the fundamental orbit and the fact that the angular momenta
are of the order of ~ we can estimate the importance of this effect,
1 e2 1 e2 1 2
H`s
1
2m2e c2 r 4πε0 r2
l ·s 2m2e c2 4πε0 a3B
~ 1 ~2 α2
= 2 ' 2 = 2 = ≈ 0.01% . (7.80)
H0 p e 2m2e c2 aB 2
2me 4πε0 aB
In the following we will derive the expression from classical arguments borrowed from electro-
dynamic theory.
7.2.2.1 Classical derivation of the spin-orbit interaction

Seen from the rest system of the electron being at position x = 0, it is the proton that orbits
around the electron. This orbit creates a current, −j(r0 ), which generates a magnetic field.
Following the Biot-Savart law the potential vector and the amplitude of the field are,
Z
µ0 −j(r0 )d3 r0
A(x) = , (7.81)
4π V |x − r0 |
respectively,
Z
µ0 (x − r0 ) × j(r0 ) 3 0
B(x) = ∇x × A(x) = d r (7.82)
4π V |x − r0 |3
Z Z Z
µ0 ∞ 0 ∞ 0 0 2π (x − r0 ) × v Ze Zeµ0 (x − r) × v
=− dz r dr dφ 0 |3 2πr
δ(r − r0 )δ(z 0 ) = ,
4π −∞ 0 0 |x − r 4π |x − r|3
where we replaced the expression for the current density (7.65). With the expression for the
Coulomb potential between the electron and the proton and its radial derivative,
−Ze2 1 dV (r) Ze2

V (r) = , = , (7.83)
4πε0 r r dr 4πε0 r3
we have at the position of the electron,
Zeµ0 −r × v ε0 µ0 r × v dV (r) 1 r × v dV (r) 1 dV (r)

B(0) = 3
=− =− 2 =− 2
l . (7.84)
4π r e r dr ec r dr eme c r dr
The advantage of introducing the potential V is, that this expression also holds for more com-
plicated atoms with many electrons, where the potential may deviate considerably from the
colombian potential. Note, that the magnetic field is very strong, B ' ξ(aB )~/µB ≈ 5 T. In-
serting the magnetic field into Pauli’s expression (7.68) together with the magnetic moment of
the spin (7.70),
1 1 dV (r)
Ĥ`s = −~µs · B(0) = 2 2 s · l . (7.85)
me c r dr
Hence, for the case of a Coulomb potential and applying the corrective Thomas factor of 12 , the
interaction operator can be written,
Ĥ`s = ξ(r)l · s , (7.86)

with the abbreviation,

−1 dV Ze2 1 En Z 2 α2 n2 1
ξ(r) ≡ =− = , (7.87)
2 2
2me c r dr 2 2
8πε0 me c r 3 ~2 r̃3
with r̃ ≡ Zr/aB and using the formulas (4.49).
7.2.2.2 Calculation of the energy correction

After the introduction of the spin, the Hilbert space of the particles’ wavefunctions must be
extended. The wavefunctions are now products of spatial wavefunctions and spin eigenvectors:

s
|n, `, m` , ms i = Rn` (r)Y`m (θ, φ) 1 . (7.88)
s2
The new Hilbert space is the tensorial product of position space and spin space. The radial
Hamiltonian for the hydrogen atom including the centrifugal term and the spin-orbit coupling
now takes the form:
p2 l2
Ĥ = + V (r) + + ξ(r)l · s . (7.89)
2m 2me r2
We may again consider the energy term V`s as a small perturbation, and calculate it using
unperturbed wavefunctions,
∆E`s = hn, `, s, m` , ms |V`s |n, `, s, m` , ms i = hn, `|ξ(r)|n, ìh`, s, m` , ms |s · l|`, s, m` , ms i . (7.90)
Assuming a coulombian potential, we first look at the radial part (7.87), which can easily be
calculated using the formulae (4.49),
En Z 2 α2 n2 1
hn, `|ξ(r)|n, ì = . (7.91)
~ 2 n `(` + 21 )(` + 1)
3
To diagonalize the angular part of the Hamiltonian, we need the common wavefunctions of l2
and l · s. We can rewrite the coupling term as:
l · s = 12 (j2 − l2 − s) . (7.92)
In the common eigensystem of j2 , l2 , and s2 the Hamiltonian, therefore, is diagonal. We know
the basis of this system from the theory of the addition of angular momenta. The states of the
basis are linear combinations of the functions |n, `, m` , ms i. Since the spins precess around each
other, `z and sz are not good observables, the non-coupled base is not appropriate. But s2 , l2 ,
and j2 are good observables. In the coupled basis {n, (`, s)j, mj },
~2
hn, (`, s)j, mj |s · l|n, (`, s)j, mj i = 2 [j(j + 1) − `(` + 1) − s(s + 1)] . (7.93)
Since j = ` ± 1/2, we find that every level splits into two levels, one with the energy En` + `ζn`
and the degeneracy 2`+2 and the other with the degeneracy En` −(`+1)ζn` with the degeneracy
2`, where we introduced the abbreviation,
~2
ζn` ≡ 2 hξ(r)i . (7.94)
All in all, we get an energy correction due to the spin-orbit interaction of,
j(j + 1) − `(` + 1) − 43
∆E`s = −En (Zα)2 . (7.95)
2n`(` + 1/2)(` + 1)
Note, that the coupling l · s lifts the degeneracy with respect to l, but not with respect to `z
(see Fig. 7.1). As we have already seen in Exc. 7.5.1.7, in the presence of an energy associated
with the coupling l · s, only the total angular momentum l + s is a constant of motion.
7.2.3 Non-local electron-core interaction

Let us now discuss the third correction in the expression (7.6). The electron-nucleus interaction
that we have considered so far is local, that is, the interaction at the point r sensed by the
electron depends essentially on the field at that point in space. However, when relativistic
theory is correctly applied, the electron-nucleus interaction becomes non-local, and the electron
is then affected by all values of the nuclear field in a region around r 10 . The size of this region
is of the order of the Compton wavelength of the electron, λC /2π ≡ ~/me c. This correction
was introduced by Sir Charles Galton Darwin through a substitution in the Dirac equation that
solved the problem of normalization of the wavefunction.
Imagine that instead of the potential V (r), the potential of the electron is given by the
integral, Z
f (r0 )V (r + r0 )d3 r0 , (7.96)
where f (r0 ) is a radially symmetric and normalized density-type function that takes significant
values only in the vicinity of r within a volume (λC /2π)3 centered at r0 = 0. Expanding the
potential V (r + r0 ) near the origin,
V (r + r0 ) = V (r) + [r0 · ∇r ]V (r) + 1 0

2! [r · ∇r ]2 V (r) + ... , (7.97)
and inserting into the integral,

Z Z Z Z
0 0 3 0 0 3 0 0 0 3 0
f (r )V (r + r )d r = V (r) f (r )d r + r f (r )d r · ∇r V (r) + 1
2! r02 f (r0 )[êr0 · ∇r ]2 d3 r0 V (r) + ...
Z
∂2
= V (r) + 0 + 2! r02 f (r0 )d3 r0 2 V (r) + ... .
1
(7.98)
∂r
The second term is null due to the parity of f (r0 ) and the third produces the Darwin correction
using V (r) = V (r). Letting the function be constant within the volume, f (r) ' f0 , and with
the normalization,
Z ~/2me c Z ~/2me c Z ~/2me c 3
~
1= f (r)dxdydz = f0 , (7.99)
−~/2me c −~/2me c −~/2me c me c
we get the integral

Z Z ~/2me c Z ~/2me c Z ~/2me c 2
~
r2 f (r)d3 r = f (r)r2 dxdydz = . (7.100)
−~/2me c −~/2me c −~/2me c 2me c
Also,
Ze %(r) Ze2 δ 3 (r)
∇2 V (r) = −e∇2 = −e =− . (7.101)
4πε0 r ε0 ε0
Hence,
Z
Ze2 π~2 Ze2 3
f (r0 )V (r + r0 )d3 r0 = − + δ (r) + ... , (7.102)
4πε0 r 2m2e c2 4πε0
which is precisely the electrostatic energy with the Darwin correction in the expressions (7.60)
and (7.73).
10
The smearing out of the electron’s position is also known as Zitterbewegung.
To estimate the importance of this effect we consider the ground state, inserting its wave-
function (4.48),
Z
∗ π~2 Ze2 3 π~2 Ze2 π~2 Ze2 1
hĤdw i = d3 rψ100 (r) 2 2 δ (r)ψ100 (r) = |ψ(0)|2
= .
2me c 4πε0 2m2e c2 4πε0 2m2e c2 4πε0 πa3B
(7.103)
We obtain,
π~2 Ze2 1
Hdw 2m2e c2 4πε0 πa3B ~2 Z α2
= = = ≈ 0.01% . (7.104)
H0 e2 2m2e c2 a2B 2
4πε0 aB
Darwin’s correction does not depend on the angular momentum l nor on nthe spin s, such
that,
∆Edw = hĤdw i = −En (Zα)2 . (7.105)
7.2.4 Summary of the corrections

Combining the LS and relativistic corrections, we obtain,
∆Ef s = ∆Erl + ∆E`s + ∆Edw (7.106)

" #
3
1 3 2 j(j + 1) − `(` + 1) − 4
= En (Zα)2 1 − 4n2 − En (Zα) 1 − En (Zα)2
n(` + 2 ) 2n`(` + 2 )(` + 1)
 j(j+1)−(j− 2 )(j+ 2 )− 34
1 1
1
 nj − 4n3 2 − 2n(j− 12 )j(j+ 21 )
−1 para ` = j − 12
2
= En (Zα) 1 3 3
 1 − 3 − j(j+1)−(j+ 2 )(j+ 2 )− 4 − 1 para ` = j + 21
n(j+1) 4n 2 1 3
2n(j+ 2 )(j+1)(j+ 2 )
" #
2 1 3
= En (Zα) − −1 .
n(j + 12 ) 4n2
That is, the levels are now degenerate in j (see Fig.. 7.1)11 . Obviously the levels which are most
affected by relativistic corrections are those with low values of n and `.
Figure 7.1: Hydrogen levels.
The levels are labeled by n`j . For example, the state 3d5/2 has the main quantum number
n = 3, the orbital angular momentum ` = 2, and the total angular momentum j = 5/2.
11
It is interesting that the quantum treatment presented here, including relativistic corrections, coincidentally
agrees with the corrections of Arnold Johannes Wilhelm Sommerfeld,
α2

1 3
En,j = En 1 + − .
n j + 1/2 4n
7.3. HYPERFINE STRUCTURE 183
States with equal j are degenerate. For large n or j the fine structure disappears. The new
energy scheme is shown in Fig. 7.1. We note that, taking into account all relativistic corrections
(but without the Lamb shift), we still have a partial degeneracy of the quantum number j. For
example, the states 2 s1/2 and 2 p1/2 have the same energy. This is a particularity of the hydrogen
atom.
7.2.5 Lamb shift

Only remains to discuss the fourth correction, Ĥlamb in the expression (7.6). The origin of
the Lamb shift lies in quantum electrodynamics. Being due to the quantum nature of the
electromagnetic field, this correction is not predicted within the Dirac equation.
We may imagine the Coulomb force between charged particles being mediated by a con-
tinuous exchange of virtual photons. But each isolated charge also continuously emits and
reabsorbs virtual photons, with the result that the position of the electron is smeared over a
region of 0.1 fm. This reduces the overlap between the electronic orbits and the nucleus. Hence,
the Lamb shift causes corrections that are stronger for small n and small `. For example in
hydrogen, the 2p1/2 is 4.4 · 10−6 eV = 1 GHz below the 2s1/2 (see Fig. 7.1).
7.3 Hyperfine structure

Rutherford’s measurements suggested a point-like and infinitely heavy atomic nucleus. In fact,
the mass is finite and the nuclear charge is distributed over a finite volume and often in a non-
isotropic manner, which leads to multipolar interactions with the electrons. In addition, many
nuclei have a spin that can interact with the magnetic moment of the electrons. The energy
corrections due to these effects are called hyperfine structure 12 .
7.3.1 Coupling to the nuclear spin

7.3.1.1 Dipole moment of the nuclear spin
The nucleus may also have an angular momentum interacting with the angular momentum of
the electrons. However, the momentum depends inversely on the masses. That is, the angular
momentum of the nucleus is µN /µB = me /mp ' 10−3 times smaller, where µN = ~e/2mp is
an abbreviation called nuclear magneton. So we can assume that the interaction between the
nucleus and the atoms will not interfere with the L · S-coupling between the orbital angular
momentum and the spin of the electrons. The spin of the nucleus will be oriented along the
total momentum of the electrons J. However, this interaction will have the ability to lift the
hydrogen degeneracy, even though the splitting will only be hyperfine. Indeed, the order of
magnitude of hyperfine splitting is 10−6 eV.
Analogously to equation (7.67), we write the dipole moment of the nucleus,
e µN
µ
~I = gp I = gp I , (7.107)
2mp ~
where gp ≡ µI /I is once again a factor taking into account possible corrections between the
classical derivation and quantum mechanics 13 .
12
See [48] p. 1229 and [246] p. 23 for further reading.
13
In fact, the proton factor g is anomalous, gp = 5.58, which reduces the faction µl /µI . For the neutron we
have: gp = −3.83
7.3.1.2 Hyperfine splitting

We calculate the hyperfine structure starting (as we did for the fine structure) from the Pauli
equation (7.64),

p2 1 e2 e e
Ĥ ' Ĥ0 + ĤLI + ĤSI = − + A·p+ s·B , (7.108)
2me 4πε0 r 2me me
with µS = − mee s.
The nuclear spin produces, at the position of the electrons, a magnetic vector potential,
~I × r
µ0 µ
Adp (r) = , (7.109)
4π r3
interacting with the angular momentum of the electron L in the form,
e e µ0
ĤLI = A · p̂ = µI × r) · p
(~ (7.110)
me me 4πr3
e µ 0 µN µ0 µ B µN
= gp (I × r) · p = gp L · I ,
me 4πr3 ~ 2πr3 ~ ~
using the definition of Bohr’s magneton.
In addition, the potential vector generated by the nuclear spin produces a magnetic field,
µ0
B=∇×A= µI · r̂)r̂ − µ
[3(~ ~I] , (7.111)
4πr3
as will be shown in Exc. 7.5.3.1 14 . This field interacts with the spin of the electron S of the
form,
µ0
ĤSI = −~
µS · B = − µI · r̂)(~
[3(~ µS · r̂) − (~
µS · µ
~ I )] (7.112)
4πr3
µ 0 µB µN
= gs gp [3(I · r̂)(S · r̂) − (S · I)] ,
4πr3 ~ ~
using equation (7.70).
Combining the two terms (7.110) and (7.112), we obtain,
µ 0 µB µN
ĤJI = ĤLI + ĤSI = gs gp [3(I · r̂)(S · r̂) + L · I − S · I] (7.113)
4πr3 ~ ~
µ0 µB µN
= gs gp N · I ,
4πr3 ~ ~
introducing
N ≡ 3(S · r̂)r̂ + L − S . (7.114)
Defining the complete total angular momentum of the atom
F≡I+J , (7.115)
is useful for calculating the coupling I · J = 21 (F2 − I2 − J2 ). Now, as the coupling L · S is strong,
we project the two angular momenta on the total electronic angular momentum J,
N·J J I·J J
N→ , I→ . (7.116)
|J| |J| |J| |J|
We consider two cases:
14
We present here a simplified calculation short-circuiting the Fermi contact term.
7.3. HYPERFINE STRUCTURE 185
7.3.1.3 Orbital angular moments L = 0

In the case L = 0 we can approximate,
N · S ' S2 , (7.117)
and replace the J with S in the projections (7.116). With this the coupling between the spins
of the electronic layer and the nucleus becomes,
(N · S)(I · S) S2 (F2 − I2 − S2 )
N·I= = . (7.118)
~ |J|
2 2 2~2 |S|2
We calculate
`=0
∆Ehf s = h(S, I)F, mF |ĤJI |(S, I)F, mF i (7.119)

µ0 µ B µ N N·I
= gs gp
4π ~ ~ r3
3
µ0 µB µN
3 8π Z 1
= gs gp F (F + 1) − I(I + 1) − 4 .
8π ~ ~ 3 aB πn3
As an example consider the hyperfine structure of the state 1s1/2 of the hydrogen atom.
With J = I = 12 and Z = n = 1 we obtain,
L=0 `=0 2µ0 µB gs µN gp 2gs gp m2e c2 4

∆Ehf s (F = 1)−∆Ehf s (F = 0) = = α ≈ (2π~)·1.420 GHz . (7.120)
3πa3B 3mp
The experimental value is 1.4204057518 GHz. This frequency corresponds to the spectral line
used in radio astronomy, where the measurement of the angular distribution of this radiation
allows the mapping of the spatial distribution of interstellar hydrogen.
7.3.1.4 Orbital angular momenta L 6= 0

In the case L 6= 0 we get for the coupling between the spins of the electronic layer and the
nucleus,
(N · J)(F2 − I2 − J2 )
N·I= . (7.121)
2|J|2
We calculate
L6=0
∆Ehf s = h((L, S)J, I)F, mF |ĤJI |((L, S)J, I)F, mF i (7.122)

µ0 µB µN N·I
= gs gp
4π ~ ~ r3
3
µ0 µB µN N · J[F (F + 1) − I(I + 1) − J(J + 1)] Z n
= gs gp .
4π ~ ~ 2J(J + 1) aB n4 L(L + 12 )(L + 1)
Introducing the interval factor,
3
µ0 µB µN Z N·J n
AJ ≡ gs gp , (7.123)
4π ~ ~ aB 2J(J + 1) n4 L(L + 21 )(L + 1)
we can write
L6=0 AJ
∆Ehf s = 2 [F (F + 1) − J(J + 1) − I(I + 1)] . (7.124)
Note, that the J · I-coupling breaks the degeneracy of J in the hydrogen atom, but not of Jz .
We can derive the following interval rule,
∆EF +1 − ∆EF = AJ (F + 1) . (7.125)
Besides the magnetic interaction between the angular momenta of the nucleus and the elec-
tronic shell there is an interaction between the nucleus, when it is not spherically symmetric, and
the shell. This interaction causes deviations from the interval rule and an additional splitting
of the hyperfine states.
7.3.2 Electric quadrupole interaction

The fact that the nucleus is not perfectly spherical gives rise to new electron-nucleus corrections
that are called quadrupolar interaction. The starting point is,
1 e2 1 e2
Ĥqud = − − , (7.126)
4π0 |re − rN | 4π0 re
where re is the electronic coordinate and rN is the nuclear coordinate, both having their origin
in the center mass of the nucleus. For re > rN this interaction can be obtained after several
mathematical steps as,
3(I · J)(2I · J + 1) − 2I2 J2
Ĥqud = BJ , (7.127)
2I(I − 1)2J(J − 1)
where BJ is called the constant of the quadrupolar electron-nucleus interaction. With this
expression we can calculate,
3
2 K(K + 1) − 2I(I + 1)J(J + 1)
∆Equd = hIJKmK |Ĥqud |IJKmK i = BJ , (7.128)
2I(2I − 1)2J(2J − 1)
where K ≡ 2hJ · Ii = F (F + 1) − I(I + 1) − J(J + 1). It is important to remember that a

nucleus with I = 0 or I = 21 has no quadrupole moment, BJ = 0. Also for J = 12 there will be
no contribution.
Joining the contributions J · I of Eq. (7.124) and the quadrupolar contribution (7.128), the
hyperfine structure can be described by,
∆Ehf s = ∆EJI + ∆Equd (7.129)

AJ BJ
= K+ [3K(K + 1) − 4I(I + 1)J(J + 1)] ,
2 8I(2I − 1)(2J − 1)
where the constants AJ and BJ depend on the atom and the total electronic angular momentum.
7.4. EXOTIC ATOMS 187
Table 7.1: List of atomic data [?].
Element γD2 /2π D1 D2 νHF S [S1/2 ]

[ MHz] [ cm−1 ] [ cm−1 ] [ MHz]
1H 99.58 82264.000 82264.366
2H 99.58 82264.000 82264.366
6 Li 5.92 14901.000 14901.337 228.2
7 Li 5.92 14901.000 14901.337 803.5
23 Na 10.01 16956.000 16973.190 1771.6
39 K 6.09 12985.170 13042.876 461.7
40 K 6.09 12985.170 13042.876 -1285.8
41 K 6.09 12985.170 13042.876 254.0
85 Rb 5.98 12578.920 12816.469 3035.7
87 Rb 5.98 12578.920 12816.469 6834.7
133 Cs 5.18 11182.000 11737.000 9192.6
135 Cs 5.18 11182.000 11737.000
Table 7.2: Hyperfine constants of some alkaline atoms.
atom n AJ (n2 S1/2 ) AJ (n2 P1/2 ) AJ (n2 P3/2 ) BJ (n2 P3/2 )

[MHz·h] [MHz·h] [MHz·h] [MHz·h]
1 H, I = 12 1 1420 46.17 −3.07 −0.18
7 Li, I = 32 2 401.75 46.17 −3.07 −0.18
3 13.5 −0.96
23 Na, 3
I= 2 3 885.82 94.3 18.65 2.82
4 202 28.85 6.00 0.86
85 Rb, 5
I= 2 5 1011.9 120.7 25.029 26.03
6 239.3 39.11 8.25 8.16
87 Rb, 3
I= 2 5 3417.3 409.1 84.852 12.510
6 809.1 132.5 27.70 3.947
In Excs. 7.5.3.2 and 7.5.3.3 we determine the hyperfine structures of sodium and rubidium
atoms.
7.4 Exotic atoms

’Normal’ atoms consist of a nucleus made of protons and neutrons and an electronic shell. But
other two-particle systems are possible, e.g. where the nucleus or electron is replaced by another
hadron or lepton (anti-proton, positron, muon, etc.). Such a system is called exotic atom. Atoms
in Rydberg states also belong to this category.
7.4.1 Positronium and muonium

Positronium (e+ e− ) is a hydrogen-like system consisting of leptons, that is, an electron and a
positron, which is the antiparticle of the electron. The muonium (µ+ e− ) is similar to positron-
ium, except that here the positron is replaced by a muon whose mass is mµ+ = 207me . Leptons
are, according to the present understanding, particles without internal structure. Both systems
are unstable: the two particles annihilate each other producing γ-photons. The energy levels
and orbits of the two particles are similar to that of the hydrogen atom. However, because of
the reduced mass, the frequencies of the spectral lines are less than half of the corresponding
hydrogen lines.
The fundamental state of positronium, like that of hydrogen, has two possible configurations
depending on the relative orientation of the electron and positron spins. The singlet state with
antiparallel spins (S = 0, Ms = 0) is known as para-positronium (p-Ps) and denoted by 1 S0 . It
has an average lifetime of
2~
τ= = 124.4 ps (7.130)
me c2 α5
and decays preferably in two gamma rays with energy of 511 keV each (in the center-of-mass).
The triplet state with parallel spins (S = 1, Ms = −1, 0, 1) is known as ortho-positronium (o-
Ps) and denoted as 3 S1 . It has an average life of 138.6 ns, and the most common form of
decay produces three photons. Other forms of decay are negligible. For example, the decay
channel producing five photons is 10−6 times less likely. Measurements of these lifetimes and
the positronium energy levels have been used in precision tests of quantum electrodynamics.
While the precise calculation of the positronium energy levels is based on the Bethe-Salpeter
equation, the similarity between positronium and hydrogen allows for an approximate estimate.
In this approach, the energy levels are suppodsed to be different from those of hydrogen because
of the difference in the value of the reduced mass µ, used in the energy equation. Since µ = me /2
for positronium, we have
µqe4 1 1 me qe4 1 −6.8 eV

En = − 2 = − = . (7.131)
8h2 0 n 2 2 8h2 20 n2 n2
A di-positronium molecule, that is, a system of two bound positronium atoms, has already
been observed. Positronium in high energy states has been conjectured to become the dominant
form of atomic matter in the universe in the very distant future if the proton decay becomes
tangible.
7.4.2 Hadronic atoms

In contrast to leptons (such as the electron e− , the positron e+ and the muons µ+ and µ− ) that
participate only in electromagnetic interactions and weak interactions, hadrons also participate
in strong (nuclear type) interactions. There are two types of hadrons, baryons (such as the
proton p and antiproton p̄, the neutron n and antineutron n̄, hyperons Σ, Ξ, ...) that have semi-
integer spin and behave like fermions and mesons (like the π-meson, K-meson, ...) that have an
integer spin. Every negatively charged hadron can be used to form a hydrogen-type hadronic
atom. These systems contain a nucleus and negative hyperon and are known as hyperonic atoms.
All of these are unstable and due to the fact that they have a sufficiently long lifetime, some of
their spectral lines have now been observed.
Since the hadrons interact strongly with the nucleus, the theory developed for hydrogen
systems (in which only exist Coulomb interaction) can not be directly applied. In this way the
values shown in Tab. 7.3 give only an estimate of the ’radius’ and the ionization potential of the
hadronic atoms pπ − , pκ− , pp̄ and pΣ− .
7.4. EXOTIC ATOMS 189
Table 7.3: Main features of some exotic atoms.
system reduced mass radius ’a’ Ip

pe− 1836/1837 ≈ 1 ≈ aB = 1 e2 /2aB ≈ 0.5
e+ e− 0.5 2 0.25
µ+ e − 207/208 ≈ 1 1 0.5
pµ− ≈ 186 5.4 · 10−3 93
pπ − ≈ 238 4.2 · 10−3 119
pκ− ≈ 633 1.6 · 10−3 317
pp̄ ≈ 928 1.1 · 10−3 459
pΣ− ≈ 1029 9.7 · 10−3 515
7.4.3 Muonic hydrogen

The muon mass is mµ = 207me . When a muon is attached to a proton we have muonic hydrogen.
Its size is smaller because of the reduced mass aµ = aB 1/m1/m e
µ +1/mp
and the binding energy and the
energies of excitation are greater for the same reason. F.ex. while for H = p+ e− the transition
2S − 2P1/2 is at 10 eV , 121 nm, for p+ µ− it is at 1900 eV. Muonic atoms are interesting
because they have amplified Lamb shifts, hyperfine interactions, and quantum electrodynamical
corrections. Therefore, the displacement due to the finite distribution of charges in the proton
rp = 0.8 fm should influence the spectrum. While in p+ e− the 2S level is shifted upward by the
Lamb shift by a value of 4.4 × 10−6 eV, in p+ µ− it is shifted down by a value of 0.14 eV. In
Exc. 7.5.4.2 we calculate the spectrum of the muonic hydrogen and in Exc. 5.5.1.3 we compare
the energy corrections due to the finite extension of the nuclei for muonic and for standard
hydrogen in first order TIPT.
7.4.4 Rydberg atoms

An atom excited to a state whose main quantum number is very high is called Rydberg atom.
These atoms have a number of peculiar properties, including high sensitivity to electric and
magnetic fields, long decay times, and wavefunctions that approximate classical electron orbits.
The inner electrons protect the outer electron from the electric field of the nucleus such that,
from a distance, the electric potential looks identical to that seen by the electron of a hydrogen
atom.
Despite its flaws, Bohr’s atom model is useful in explaining these properties. In Exc. 1.4.1.6
we derive Bohr’s expression for the orbital radius in terms of the principal quantum number n:
4π0 n2 ~2
r= . (7.132)
e2 m
Thus, it is clear why Rydberg atoms have peculiar properties: the radius goes as n2 (such that
for example the state with n = 137 of hydrogen has a radius of ∼ 1 mm) and the geometric
cross section goes as n4 . Thus, Rydberg atoms are extremely large, with loosely bound valence
electrons that are easily perturbed or ionized by collisions or external fields.
Since the binding energy of a Rydberg electron is proportional to 1/r, and therefore falls as
1/n2 , the spacing between energy levels falls as

1 1 n→∞ 2 3
∆E = E1 − −→ E1 − 3 + 4 + ... (7.133)
(n + 1)2 n2 n n
leading to less and less spaced levels. These Rydberg states form the Rydberg series.
7.4.4.1 Correspondence principle in Rydberg atoms

To calculate the oscillation frequency of an electron confined to a proton, we use the classical
planetary model,
e2
mωn2 = and mωr2 = n~ . (7.134)
4π0 r
Eliminating r,
me4
ωn = . (7.135)
(4π0 )2 n3 ~3
Radiation of this frequency will be emitted by an atomic antenna. On the other hand, the Bohr
model predicts frequencies between orbitals,

En+1 − En me4 1 1 n→∞ me4 2
ωn = = − −→ . (7.136)
~ 2(4π0 ) ~
2 2 (n + 1) 2 n 2 2(4π0 )2 ~2 n3
7.4.4.2 Production of Rydberg atoms

In the hydrogen atom only the ground state (n = 1) is actually stable. Other states must be
excited by various techniques such as electron impact or charge exchange. In contrast to these
methods, which produce a distribution of excited atoms at various levels, the optical excitation
method allows to produce specific states, but only for alkali metals whose transitions fall into
frequency regimes which are accessible to lasers.
7.4.4.3 Potential in a Rydberg atom

The valence electron in a Rydberg atom with Z protons in the nucleus and Z − 1 electrons in
closed layers sees a spherically symmetric Coulomb potential:
e2
Ucou = − . (7.137)
4πε0 r
The similarity of the effective potential ’seen’ by the outer electron and the authentic hydrogen
potential suggests a classical treatment within the planetary model. There are three notable
exceptions:
• An atom can have two (or more) electrons in highly excited states with comparable orbital
radii. In this case, the electron-electron interaction gives rise to a significant deviation from
the hydrogen potential. For an atom in a multiple Rydberg state the additional term Uee
includes a sum over each pair of highly excited electrons:
e2 X 1
Uee = . (7.138)
4πε0 |ri − rj |
i<j
• If the valence electron has very low angular momentum (interpreted classically as an
extremely eccentric elliptical orbit), it can pass close enough to the nucleus to polarize it,
giving rise to an additional term,
e 2 αd
Upol = − . (7.139)
(4πε0 )2 r4
7.5. EXERCISES 191
• If the outer electron penetrates the inner electronic shells, it sees more of the charge of
the nucleus and therefore feels a larger force. In general, the modification of the potential
energy is not simple to calculate and should be based on some knowledge of the nucleus’
geometry.
In hydrogen the binding energy is given by:
E1
EB = − . (7.140)
n2
The binding energy is weak at high values of n, which explains the fragility of the Rydberg
states that can easily be ionized, e.g. by collisions.
Additional terms modifying the potential energy of a Rydberg state require the introduction
of a quantum defect, δ`, in the expression for the binding energy:
E1
EB = − . (7.141)
(n − δ` )2
The long lifetimes of Rydberg states with high orbital angular momentum can be explained
in terms of overlapping wavefunctions. The wavefunction of an electron in a state with high
` (large angular momentum, ’circular orbit’) has little overlap with the wavefunctions of the
internal electrons and therefore stays relatively unperturbed. Also, the small energy difference
between adjacent Rydberg states decreased the decay rate according to the result (??).
7.4.4.4 Rydberg atoms in external fields

The large distance between the electron and ionic nucleus in a Rydberg atom gives rise to an
extremely large electric dipole moment d. There is an energy associated with the presence of an
electric dipole in an electric field E, known as Stark shift,
ES = −d · E . (7.142)
Depending on the sign of the projection of the dipole moment onto the vector of the local electric
field, the energy of a state increases or decreases with the intensity of the field. The narrow
spacing between adjacent levels n in the Rydberg series means that the states can approach
degeneracy even for relatively weak fields. Theoretically, the force of the field in which a level
crossing would occur (assuming no coupling between the states) is given by the Inglis-Teller
limit,
e
FIT = . (7.143)
12πε0 a20 n5
In hydrogen the pure Coulomb potential does not couple the Stark states of an n level, which
results in a real crossover. In other elements, deviations from the ideal 1/r-potential allow for
avoided crossings.
7.5 Exercises
7.5.1 Fine structure and the Dirac equation
7.5.1.1 Ex: Yukawa potential
Show, that Yukawa’s potential satisfies the Klein-Gordon equation.
7.5.1.2 Ex: Commutators with Dirac matrices

Prove that, if [B, α
~ ] = 0 = [C, α
~ ] then,
α · B)(~
(~ α · C) = B · C+i~
α · (B × C) .
Note, that the same relation also holds for the Pauli matrices, which satisfy the same ant-
commutation relation [~σk , ~σm ]+ = 2δkm .
7.5.1.3 Ex: Constants of motion of Dirac’s Hamiltonian

~ ~ σz 0
Show that L̂z with L ≡ r × p and Ŝz ≡ 2 γ5 α3 = 2 defining γ5 ≡ −iα1 α2 α3 are not
0 σz
constants of motions, but J = L + S.
7.5.1.4 Ex: Commutators with Dirac matrices

a. Prove, using L ≡ r × p,
α · L + ~)]+ = 0 .
α · p, (~
[~
b. Prove, using L × L = i~L and defining J ≡ L + ~2 α
~,
~2
α · L + ~)2 = J2 +
(~ 4 .
c. Prove,
∂
r · p = −i~r .
∂r
7.5.1.5 Ex: Dirac equation in spherical coordinates

The goal of this exercise is to derive the Dirac equation for an electron in a central Coulomb
field. The starting point is the Dirac equation (7.15),
i~∂t Φ(r, t) = ĤΦ(r, t) ,
with the Hamiltonian in the minimal coupling (7.44),
e2
Ĥ ≡ me c2 α0 + c~
α · [p − qA(r)] + qU (r) with A=0 and U (r) = − .
4πε0 r
We adopt the standard procedure from non-relativistic physics, which consists in rewriting the
Hamiltonian in terms of observables, which commute with the Hamiltonian 15 . The final radial
Hamiltonian will be,

2 ~ ciα0 j 0 ~ e2
Ĥ = me c α0 + c pr − i + .
r r 4πε0 r
We proceed as follows:
a. Consider the total angular momentum defined by,
J ≡ L + ~2 α
~ with L≡r×p ,
and show that xxx commutes with the Hamiltonian ...
15
Typical examples are the Hamiltonian of the harmonic oscillator (??) written in terms of n̂ ≡ â† â or the
Hamiltonian of the hydrogen atom (??) written in terms of L2 .
7.5. EXERCISES 193
7.5.1.6 Ex: Solution of the radial Dirac equation
Solve the radial Dirac equation.
7.5.1.7 Ex: Constants of motion in the l · s-coupling

2~
Consider a particle of mass µ described by the Hamiltonian Ĥ = − 2µ ∇2 + V (r) + ξ(r)L · S,
being V (r) a central potential, L and S its orbital angular momentum and spin.
a. Obtain the commutation relations [L, H], [S, H] and [L + S, H] for the cases without and with
spin-orbit interaction ξ(r)L · S introduced by relativistic corrections.
b. Calculate [L2 , H], [S2 , H] and [J2 , H].
7.5.1.8 Ex: The Darwin term
Show that the expressions for the Darwin correction (7.60) and (7.73) are equivalent.
7.5.2 Fine structure of hydrogen-like atoms via TDPT

7.5.2.1 Ex: Magnetic field generated by the orbiting proton at the location of
the electron
Calculate the magnetic field generated by the orbiting proton as it is perceived by the electron.
7.5.3 Hyperfine structure

7.5.3.1 Ex: Field of a magnetic moment
a. Calculate the vector potential A(r) and the magnetic dipole moment µ ~ produced by an
0 −1
orbiting electron by Biot-Savart’s law using the expansion of |r − r | in Legendre polynomials.
b. Calculate the magnetic field B(r).
7.5.3.2 Ex: Hyperfine structure of sodium
Determine the hyperfine structure of the 2 S and 2 P states of the sodium atom including energy
shifts. See Tab. 7.2 for the hyperfine constants AJ e BJ .
7.5.3.3 Ex: Hyperfine structure of rubidium
Given the following energy distances νF,F 0 of the hyperfine levels of the rubidium isotopes 87 Rb
and 85 Rb [21],
87
Rb, S1/2 splits into ν1,2 = 6834.7 MHz
87
Rb, P3/2 splits into ν0,1 = 72.3 MHz, ν1,2 = 157.1 MHz, ν2,3 = 267.2 MHz
85
Rb, S1/2 splits into ν1,2 = 3035.7 MHz
85
Rb, P3/2 splits into ν1,2 = 29.4 MHz, ν2,3 = 63.4 MHz, ν3,4 = 120.7 MHz ,
calculate the positions of the barycenters.

7.5.3.4 Ex: Two particles

Consider a two-particle system of masses µ1 and µ2 , exposed to a central potential V (r) and
an interaction potential V (|r1 − r2 |) which only depends on the distance between the particles.
The Hamiltonian of the system in the interaction representation is H = H1 + H2 + V (|r1 − r2 |)
~2
with H` = − 2µ `
∇2` + V (r` ), ` = 1, 2, ... Show that the individual angular momenta L` are not,
in general, constants of the motion, unlike the total angular momentum L = L1 + L2 .
7.5.4 Exotic atoms

7.5.4.1 Ex: Positronium
Calculate and compare the fine and hyperfine structure of positronium.
7.5.4.2 Ex: Hidrogênio muônico

Muonic hydrogen consists of a proton and a negatively charged muon. Calculate the bind-
ing energy of the ground state of muonic hydrogen in eV and write down the ground state’s
wavefunction.
Chapter 8
Atoms with spin in external fields

8.1 Charged particles in electromagnetic fields
8.1.1 Lagrangian and hamiltonian of charged particles
A charge subject to an electromagnetic field feels the Lorentz force,
F = qE + q ṙ × B , (8.1)
where
∂A
E = −∇Φ − and B=∇×A , (8.2)
∂t
where Φ and A are called scalar and vector potential, respectively.
As we learned in electrodynamics it is possible to derive this force from a Lagrangian for the
electronic motion,
m
L(ri , ṙi ) = ṙ2 − qΦ(r) + q ṙ · A(r) . (8.3)
2
With this aim we first determine the momentum by
∂L
pi = = mṙi + qAi , (8.4)
∂ ṙi
and the Hamiltonian by,
X m m
H= pi ṙi − L(ri , ṙi ) = (mv + qA) · ṙ − ṙ2 + qΦ − q ṙ · A = v2 + qΦ . (8.5)
i 2 2
That is,
1
H(ri , pi ) = (p − qA)2 + qΦ . (8.6)
2m
The following equations hold,
∂H ∂H
ṙi = and ṗi = − . (8.7)
∂pi ∂ri
The first equation is easily verified by inserting the Hamiltonian (8.5). The second leads to the
Lorentz force,
∂H
Fi = mv̇i = ṗi − q Ȧi = − − q Ȧi = qEi + q(v × B)i , (8.8)
∂ri
where the last step of the derivation will be shown in the Exc. 8.4.1.1 using the Coulomb gauge
∇ · A = 0.
195
196 CHAPTER 8. ATOMS WITH SPIN IN EXTERNAL FIELDS
8.1.2 Minimal coupling

Note that the same result (8.6) can be obtained by a canonical substitution,
mv −→ p − qA and H −→ H + qΦ . (8.9)
This substitution rule, called minimal coupling, can be applied in quantum mechanics,
mv̂ −→ −i~∇ − qA and Ĥ −→ Ĥ + qΦ . (8.10)

Ze 2
In the case of the electron (q = −e) trapped in a central Coulomb potential qΦ = − 4πε0r
and in the presence of any magnetic potential A, we thus obtain,
m 2 −~2 2 i~e i~e e2 A2

Ĥ = v̂ + qΦ = ∇ − A·∇− ∇·A+ + qΦ . (8.11)
2 2m 2m 2m 2m
The fourth term called diamagnetic term is quadratic in A and usually so small that it can
be neglected. The second and third terms describe the interaction of the electron through its
momentum p̂ with the potential vector A produced by magnetic moments inside the atom or
outer magnetic fields. Within the Coulomb gauge we have (∇ · A)ψ = (A · ∇)ψ + ψ(∇ · A) =
(A · ∇)ψ, such that,
e
Ĥint = mA · p̂ . (8.12)
8.2 Interaction with magnetic fields

8.2.1 Normal Zeeman effect of the fine structure
The dipole moments of atoms can interact with external magnetic fields. The interaction leads
to a shift of levels, which depends on the magnetic quantum number. Thus, the ultimate
degeneracy in the energetic structure of the atom is lifted. This is called Zeeman splitting. We
consider a uniform magnetic field B = Bêz with the potential vector,
A = 12 B × r = − B2 (−yêx + xêy ) . (8.13)
Thus the interaction energy between the electron and the field is given by the Hamiltonian
(8.12),
i~e i~e i~e
Vzee (B) = − A·∇=− (B × r) · ∇ = − B · (r × ∇) (8.14)
m 2m 2m
e µB µB
= − 2m B · L̂ = − gL L̂ · B = −~ µL · B = − L̂z B ,
~ ~
e
with gL = 1 using the relation µ ~ L = 2m L between the angular momentum of the electron and
the resulting magnetic moment. This relationship holds for an atom without spin (two electrons
can couple their spins to a singlet state) and no hyperfine structure (or an unresolved hyperfine
structure). The energies are therefore,
µB
∆Ezee (B) = − Bhn, L, mL |L̂z |n, L, mL i = −µB mL B . (8.15)
~
In the Excs. 8.4.2.1 and 8.4.2.2 we represent the interaction between an atomic angular momen-
tum and a magnetic field in different bases characterized by different quantization axes.
8.2. INTERACTION WITH MAGNETIC FIELDS 197
8.2.2 Anomalous Zeeman effect

The anomalous Zeeman effect occurs when the ensemble of electrons has a spin. Using the
already known expressions for the dipole moments of the orbital momentum and the spin of the
electron, we obtain for the magnetic dipole moment,
µB µB µB
µ
~J = µ
~L + µ
~S = gL L + gS S = (L + 2S) , (8.16)
~ ~ ~
with gL = 1 and gS = 2. We can see that the dipole moment of the atom is not parallel to the
total momentum, J = L + S.
When the magnetic field is weak, Vls Vzee (B), the total momentum J will be a good
observable. Therefore, we must first project the momenta L and S onto J,

J J J J
L −→ L · |J| |J| and S −→ S · |J| |J| , (8.17)
before projecting the result over the B field. The potential is,
µB µB h J

J

J

J
i
Vzee (B) = −~µJ · B = − (L + 2S) · B −→ − L · |J| · B + 2 S · · B
~ ~ |J| |J| |J|
µB µB 1 2
=− [L · J + 2S · J] J · B = − J + L2 − S2 + 2(J2 + S2 − L2 ) J · B
~|J| 2 ~|J| 2
2
2
µB 1 3J − L + S 2 2
=− J·B . (8.18)
~ 2 |J|2
And the energy is,

µB J(J + 1) − L(L + 1) + S(S + 1)
∆Ezee (B) = 1+ Ĵ · B̂ . (8.19)
~ 2J(J + 1)
Introducing the Landé factor,
J(J + 1) + S(S + 1) − L(L + 1)]

gJ ≡ 1 + , (8.20)
2J(J + 1)
we can write
µB
∆Ezee (B) = − gJ hJˆz Bi = −µB gJ mJ B . (8.21)
~
This expression describes the anomalous Zeeman effect, for which S 6= 0. For the normal Zeeman
effect, for which the spin is zero, we find again gJ = 1.
8.2.3 Paschen-Back effect and intermediate magnetic fields

A very strong external magnetic field (> 1 T), such that Vls Vzee (B), can break the L · S-
coupling. Both spins L and S now couple separately to the field,

B B B B
L −→ L · |B| |B| and S −→ S · |B| |B| . (8.22)
Therefore,
µB µB h B

B

B

B
i
Vpb (B) = − (L̂ + 2Ŝ) · B −→ − L̂ · + 2 Ŝ · ·B , (8.23)
~ ~ |B| |B| |B| |B|
Figure 8.1: Coupling angular moments for the effect (a) Normal Zeeman effect, (b) anoma-
lous Zeeman effect, (c) Paschen-Back effect, (d) Zeeman effect of the hyperfine structure, and
(e) Paschen-Goudsmith effect.
such that
∆Epb (B) = −µB (mL + 2mS )B . (8.24)
This is the Paschen-Back effect.
The derivations we have made so far have focused on simple situations well described by
CSCOs in various coupling schemes. The projections on the different quantization axes [the
total spin (8.17) in the Zeeman case or the applied magnetic field (8.22) in the Paschen-Back
case] ensure that the Hamiltonians Vls and Vzee (B) in these CSCOs are described by diagonal
matrices. However, in regimes intermediate between Zeeman and Paschen-Back, Vls ' Vzee (B),
it is generally not possible to find a diagonal representation.
In order to calculate the energy spectrum in intermediate regimes we must, therefore, deter-
mine all the components of the matrix,
µB
Vls + Vzee (B) = ξ(r)L̂ · Ŝ + (L̂ + 2Ŝ) . (8.25)
~
Using L̂± ≡ L̂x ± iL̂y and Ŝ± ≡ Ŝx ± iŜy , we can easily rewrite the energy in the following way,
µ
B
Vls + Vzee (B) = ξ(r) L̂z Ŝz + 12 L̂+ Ŝ− + 21 L̂− Ŝ+ + (L̂ + 2Ŝ) · B . (8.26)
~
This operator acts on the uncoupled states,
∆Els + ∆Ezee (B) = hL0 m0L ; S 0 m0S |ξnl (L̂z Ŝz + 12 L̂+ Ŝ− + 12 L̂− Ŝ+ ) + µB (L̂z + 2Ŝz )B|LmL ; SmS i

= ~2 ξnl mL mS δmL ,m0L δmS ,m0S + 21 L+ S− δmL ,m0L −1 δmS −1,m0S + 21 L− S+ δmL −1,m0L δmS ,m0S −1
+ ~µB (mL + 2mS )BδmL ,m0L δmS ,m0S , (8.27)
p
with the abbreviations L± ≡ L(L + 1) − mL (mL ± 1). The energies are now the eigenvalues of
this matrix. The factor ξnl is usually determined experimentally by letting B = 0. In Exc. 8.4.2.3
we calculate the re-coupling of the spins of two electrons in an external magnetic field.
8.2.4 Zeeman effect of the hyperfine structure

When the energy of the interaction with the magnetic field is comparable to the hyperfine
interactions, but much weaker than that of the fine interactions, the fields do not disturb the
coupling between the total electronic momentum J and the spin of the nucleus I. Hence, J, I, F,
and mF are good quantum numbers. Therefore, to calculate the interaction energy,
Vhf s + Vzee (B) = Vhf s − µ
~F · B , (8.28)
Figure 8.2: Transition between the Zeeman regime and the Paschen-Back regime for the case
L = 1 and S = 1/2.
we project the nuclear spin and the total electronic momentum separately in the direction of F,

F F F F
J −→ J · |F| |F| and I −→ I · |F| |F| . (8.29)
The total magnetic momentum is,

µB µN
µ
~F = µ ~I = −
~J + µ gJ J + gp I . (8.30)
~ ~
Note the negative sign due to the negative charge of the electron. The Landé factor gJ (see
(8.20)) is the one caused by the coupling of the orbital angular momentum L and the electron
spin S and depends on the state under consideration. Thereby,
h µ µN F F i
B F F
Vzee (B) = − gJ J · |F| + gp I · |F| |F| B (8.31)
~ ~
|F|

µB µN
= − gJ J · F + gp I · F (B · Fz ) .
~|F| 2 ~|F|2
Using J · F = 21 (F2 + J2 − I2 ) e I · F = 21 (F2 − J2 + I2 ) we write,
µB F2 + J2 − I2 µN F2 − J2 + I2
Vzee (B) = − gJ BFz + gp BFz , (8.32)
~ 2|F| 2 ~ 2|F|2
such that
∆Ehf s + ∆Ezee (B) ' ∆Ehf s + µB gF mF B , (8.33)
using the Landé factor gF for the state F ,
F (F + 1) + J(J + 1) − I(I + 1) µN F (F + 1) − J(J + 1) + I(I + 1)

gF ' gJ − gJ , (8.34)
2F (F + 1) µB 2F (F + 1)
where the second term can be neglected.

The splitting of electronic states with the momentum F into 2F +1 sublevels mF = −F, .., F is
called Zeeman effect of the hyperfine structure. The result (8.31) only applies to weak fields. For
strong fields the Zeeman splitting becomes a Paschen-Back splitting of the hyperfine structure.
8.2.5 Paschen-Back effect of the hyperfine structure

When the interaction with the magnetic field exceeds the hyperfine interaction, the nuclear spin
I decouples from the total momentum J, and both couple separately to the external magnetic
field,
B B B B
J −→ J · |B| |B| and I −→ I · |B| |B| . (8.35)
The Zeeman effect of the hyperfine structure becomes a hyperfine structure of the Zeeman effect,
also called Paschen-Back effect of the hyperfine structure or Paschen-Goudsmith effect. We
can diagonalize the potential on a basis, where I, mI , J, and mJ are good quantum numbers.
Using the expression (7.129) but disregarding the quadrupolar contribution to the hyperfine
interaction, BJ ' 0, we obtain,
AJ
Vhf s + Vzee (B) = Vhf s − (~ ~I) · B ' 2 J · I + µ
µJ + µ ~J · B (8.36)
~
AJ B B B B AJ
−→ 2 J · · I· + µJz B = 2 Jz Iz + µJz B ,
~ B B B B ~
where we neglect the interaction of the dipole moment of the nucleus with the external magnetic
~ I ' 0. We obtain,
field, µ
∆Ehf s + ∆Ezee (B) ' AJ mJ mI + µB gJ mJ B . (8.37)
Here, we project the angular momentum J and the spin I separately onto the direction of the
magnetic field. The re-coupling of the state |F mF i to |mI mJ i in strong magnetic fields is
described by Clebsch-Gordan coefficients,
X
|F mF i = |mI mJ ihmI mJ |F mF i . (8.38)
mI +mJ =mF
8.2.6 Hyperfine structure in the intermediate field regime

Knowing the dipolar magnetic AJ and quadrupolar BJ interval factors, it is possible to calculate
the Zeeman displacement of the hyperfine structure in magnetic fields intermediate between the
Zeeman and Paschen-Back regimes. For this, we must determine all the components of the
matrix Vhf s + Vzee (B) and calculate the eigenvalues. The relevant terms of the Eqs. (7.129) and
the Eq. (8.35) are,
AJ BJ 6(I · J)2 + 3I · J − 2J2 I2

Vhf s + Vzee (B) = I · J + + gJ µB B · J − gI µN B · I . (8.39)
~2 ~2 2I(2I − 1)2J(2J − 1)
We develop the complete matrix representation of this Hamiltonian within the puncoupled
base, where mp J , mI are good quantum numbers, introducing the abbreviations I± ≡ I(I + 1) − mI (mI ± 1)
and I±± ≡ I(I + 1) − (mI ± 1)(mI ± 2). The SU(2) algebra provides useful expressions,
I · J = Iˆz Jˆz + 12 (Iˆ+ Jˆ− + Iˆ− Jˆ+ ). The elements of the matrix are,
hm0I m0J |Hhf s + HB |mI mJ i (8.40)

h in o
3BJ
= AJ + 2I(2I−1)2J(2J−1) mI mJ δm0I mI δm0J mJ + 12 I+ J− δm0I mI +1 δm0J mJ −1 + 12 I− J+ δm0I mI −1 δm0J mJ +1
2 h i
BJ 2I(I+1)J(J+1)
6BJ
+ 2I(2I−1)2J(2J−1) hm0I m0J | I·J
~ |mI m J i + − 2I(2I−1)2J(2J−1) + (gJ m J − gI µ N m I )µ B B δm0I mI δm0I mI
where

I·J 2

hm0I m0J | ~ |mI mJ i = (mI mJ )2 + 14 I−
2 2 2 2
J+ + 41 I+ J− δm0I mI δm0J mJ + (8.41)

+ 12 m0I m0J + mI mJ I+ J− δm0I mI +1 δm0J mJ −1 +

+ 12 m0I m0J + mI mJ I− J+ δm0I mI −1 δm0J mJ +1 +
+ 14 I+ J− I++ J−− δm0I mI +2 δm0J mJ −2 +
+ 14 I− J+ I−− J++ δm0I mI −2 δm0J mJ +2 .

The matrix hm0I m0J |Hhf s + HB |mI mJ i is divided into 2F + 1 diagonal blocks, each labeled mF .
The total number of levels is,
 
X X X
2F + 1 = (2I + 1)(2J + 1) =  1 .
F =|I−J|,..,I+J mF =|−F,..,F | mI =|−I,..,I|, mJ =|−J,..,J|, mI +mJ =mF
In this form the matrix can be programmed, e.g. using computational software such as
MATLAB, and all eigenvalues of the Hamiltonian for any state 2S+1 XJ and nuclear spin I can
be calculated numerically. Obviously, the eigenvalues follow from a diagonalization of the matrix
and do not depend on the chosen base. Fig. 8.3 shows the result obtained for 6 Li (I = 32 ) in the
state 2P3/2 knowing that AJ /h = −1.17 MHz and BJ = 0.
20
10
∆ν (MHz)
-10
-20
0 2 4
B (G)
Figure 8.3: (Code: AM Staticf ields HiperLithium.m) Hyperfine and Zeeman structure of the
state 2 P3/2 of 6 Li.
It is interesting to analyze the so-called fully stretched states. In this states, the interval
factor is A = Vhf s /2. By inserting F = I + J into the first formula, we obtain,
AJ 3(I + J)
Vhf s +Vzee (B) = I·J−BJ +gJ µB B·J−gI µN B·I ≈ AJ mJ mI +µB gJ mJ B .
~2 2(2I − 1)(2J − 1)
(8.42)
The energy displacement of the fully stretched√states is linear in the magnetic field. We can also
look at the matrix elements I+ = 0 and I− = 2I and note that all non-diagonal terms vanish.
When one of the spins, J or I, is equal to 1/2 only two possible hyperfine states exist:
F = I ± J, which are necessarily totally stretched. For this case there is an approximate
analytic formula called the Breit-Rabi formula [17], which will be derived in Exc. 8.4.2.4,
∆Ehf s + ∆Ezee (B) = h A~2J I · J + gJ µB B · J − gI µN B · Ii (8.43)
r
AJ AJ (I + 12 ) 4mF
=− + µN gN mF B ± 1+ x + x2 ,
4 2 2I + 1
with the abbreviation x ≡ 2(µB gJA−µ

J
N gI )B
.
Atoms with paired electrons have no spin and therefore no magnetic dipole moment. For
example, helium or strontium in their ground state 1 S0 . These systems are diamagnetic due to
the Hamiltonian term (8.11) being quadratic in B, as we shall show in Exc. 8.4.2.6.
8.2.7 Landau levels

Consider a two-dimensional system of non-interacting particles with charge q and spin S confined
to an area A = Lx Ly in the x-y plane. We apply a uniform magnetic field,
B = Bêz (8.44)
along the z axis. The Hamiltonian of this system is,

1
Ĥ = (p̂ − q Â)2 , (8.45)
2m
where p is the operator of the canonical momentum and Â is the potential vector, related to
the magnetic field by B = ∇ × Â. The vector potential,
 
−By
1
A =  Bx  (8.46)
2
0
reproduces the field (8.44). However, we have the freedom in choosing the potential vector,
given by the gauge transformation, to add the gradient of a scalar field, for example,
 
By
1 
1
χ ≡ 2 Bxy =⇒ ∇χ = Bx =⇒ A0 ≡ A + ∇χ = Bxêy . (8.47)
2
0
The potential vector A0 gives the same magnetic field and only changes the general phase of the
wave function, but the physical properties do not change. In this gauge, which is called Landau
gauge, the Hamiltonian is,

p̂2x 1 qB x̂ 2
Ĥ = + p̂y − . (8.48)
2m 2m c
The operator p̂y commutes with this Hamiltonian, since the y operator is absent due to the
choice of the gauge. Thus, the operator p̂y can be replaced by its eigenvalue ~ky . Hence, by
introducing the cyclotron frequency, ωc = qB/mc, we obtain,
2
p̂2 mωc2 ~ky
Ĥ = x + x̂ − . (8.49)
2m 2 mωc
This is exactly the Hamiltonian of the quantum harmonic oscillator, except that the minimum of
the potential is displaced in position space by the value x0 = ~ky /mωc . To find the energies, we
note that the translation of the potential of the harmonic oscillator does not affect the energies.
The energies of this system are therefore identical to those of the standard quantum harmonic
oscillator,
En = ~ωc n + 12 , (8.50)
for n ≥ 0. Since the energy does not depend on the quantum number ky , we will have degen-
eracy. For the wavefunctions, we remember that p̂y commutes with the Hamiltonian. Then the
8.3. INTERACTION WITH ELECTRIC FIELDS 203
wavefunction splits into a product of eigenstates of the momentum in y-direction and eigenstates
of the harmonic oscillator |φn i shifted by a value x0 in x-direction:
Ψ(x, y) = eiky y φn (x − x0 ) . (8.51)
That is, the state of the electron is characterized by two quantum numbers, n and ky .
Each set of wavefunctions with the same n is called Landau level. Effects due to Landau levels
are only observed when average thermal energy is lower than the separation of energy levels,
which means that low temperatures and strong magnetic fields are required. Each Landau level
is degenerate because of the second quantum number ky , which can adopt the values,
2πN
ky = , (8.52)
Ly
with N ∈ N. The allowed values of N are more restricted by the condition that the center of
force of the oscillator, x0 , must be physically inside the system, 0 ≤ x0 < Lx . This gives the
following range for N ,
mωc Lx Ly
0≤N < . (8.53)
2π~
For particles with charge q = Ze, the upper limit in N can simply be written as a ratio of
fluxes,
ZBLx Ly Φ
=Z , (8.54)
(hc/e) Φ0
where Φ0 = h/2e is the fundamental flux quantum and Φ = BA the flux through the system
(with area A = Lx Ly ). Thus, for particles with spin S, the maximum number of particles per
Landau level D is,
Φ
D = Z(2S + 1) . (8.55)
Φ0
In general, Landau levels are observed in electronic systems with Z = 1 and S = 1/2. As the
magnetic field increases, more and more electrons can fit a certain Landau level. The occupation
of the highest Landau level ranges from entirely full to entirely empty, leading to oscillations in
various electronic properties (see Haas-van Alphen effect, Shubnikov-de Haas effect and quantum
Hall effect).
8.3 Interaction with electric fields

8.3.1 Stark Effect
Electric fields interact with the electrons of the atom,
V̂stark = −ed̂ · E . (8.56)
This is the Stark effect. This effect is usually weak, and its observation requires strong fields or
high spectral resolution. Stationary perturbation theory TIPT gives,
Z
(1) (0) (0)
En = hψn | − d · E|ψn i = eEz · z|ψn(0) |2 d3 r = 0 . (8.57)
R3
This only applies when the states have well-defined parity and are NOT degenerate in `. When
they ARE degenerate in `, which is the case of hydrogen, the states have no defined parity. For
example, the states s and p contributing to the same state |ψn,j i have different parities. In this
case, the condition (8.57) does not have to be satisfied, and the first perturbation order yields a
value. This is the case of the linear Stark effect.
Other atoms do not have this degeneracy, and we must calculate the quadratic Stark effect
in second order TIPT,
X (0) hψ (0) (0)
0 |ẑ|ψn i
|ψn(1) i = eEz |ψn0 i n . (8.58)
0
En − En0
n 6=n
and
X |hψ (0) (0) 2
n0 |ẑ|ψn i|
En(2) = e2 Ez2 . (8.59)
0
En − En0
n 6=n
To simplify the matrix elements, we separate the radial part from the angular part,
Z ∞ Z
(0) 0 0 0
(0)
hψn0 |ẑ|ψn i = hn J mJ |ẑ|nJmJ i = r Rn0 J 0 RnJ dr YJ∗0 m0 zr YJmJ dΩ .
3
(8.60)
J
0
The radial part, written as

Z ∞
hn0J J 0 ||ẑ||nJ Ji ≡ r3 Rn0J J 0 RnJ J dr , (8.61)
0
and called the irreducible matrix element, no longer depends on the magnetic quantum number.
On the other hand, the angular part may be expressed by Clebsch-Gordan coefficients, as will
be discussed more extensively in Sec. ??. The result is called Wigner-Eckart theorem,

|hn0J J 0 m0J |ẑ|nJ JmJ i|2 1 J 1 J0
= . (8.62)
hn0 J 0 ||ẑ||nJ Ji2 2J 0 + 1 mJ 0 −m0J
J
With [ẑ, L̂z ] = 0, which was shown in Exc. 4.5.3.2 we find,

0 = hJ 0 m0J |[ẑ, Jˆz ]|JmJ i = (mJ − m0J )hJ 0 m0J |ẑ|JmJ i . (8.63)
This means that for mJ 6= m0J , the matrix elements hJ 0 m0J |ẑ|JmJ i should disappear. Therefore,
the matrix is diagonal in mJ . We consider dipole transitions with |J − J 0 | ≤ 1 1 ,

J 1 J +1 (J + 1)2 − m2J
= , (8.64)
mJ 0 −mJ (2J + 1)(J + 1)

J 1 J m2J
= ,
mJ 0 −mJ J(J + 1)

J 1 J −1 J 2 − m2J
= .
mJ 0 −mJ J(2J + 1)
States with the same |mJ | lead to the same quadratic Stark effect,
∆E ∼ A + B|mJ |2 . (8.65)
The factors A and B depend on the main quantum number n and also on L, S, J. Moreover,
they depend on the energy distance of all contributing levels, because of the denominator in the
perturbation equation (8.58). Only levels with different parity (−1)L contribute. In Excs. 8.4.3.1
and 8.4.3.2 we explicitly calculate the Stark energy shift for a hydrogen atom subject to an
electric field.
1
For it is possible to show that hn0J J 0 ||ẑ||nJ Ji = 0 for JJ − J 0 | > 1.
8.4. EXERCISES 205
8.4 Exercises
8.4.1 Charged particles in electromagnetic fields
8.4.1.1 Ex: Lagrangian of an electron in the electromagnetic field
a. Show that the Lagrangian (8.3) reproduces the Lorentz force (8.1).
b. Show that the Hamiltonian (8.5) reproduces the Lorentz force (8.1).
8.4.2 Interaction with magnetic fields

8.4.2.1 Ex: Zeeman effect with different quantization axes
The Zeeman effect can be described in several ways depending on the choice of the quantization
axis. Consider a magnetic field B = Bx êx and calculate the interaction Hamiltonian V (B) =
−~µJ · B
a. choosing the quantization axis êx in the direction of the magnetic field,
b. choosing the quantization axis êz perpendicular to the direction of the magnetic field.
8.4.2.2 Ex: Zeeman shift and quantization axes

Choosing the fixed quantization axis êz and a magnetic field B in an arbitrary direction, calculate
the Hamiltonian of the Zeeman interaction with an angular momentum J = 1 and show that
the energy shift depends only on absolute value |B|.
8.4.2.3 Ex: Coupling of two electrons

Consider a two-electron system.
a. Show that the operator (~A/~2 )ŝ1 · ŝ2 distinguishes the triplet from the singlet states.
b. Consider now, that the electrons are exposed to a magnetic field B applied in the direction êz ,
so that they acquire the interaction energies with the field (µB B/~)(g1 ŝ1z + g2 ŝ2z ). Obtain the
matrix associated with the total Hamiltonian and demonstrate that in the regime ~A µB B,
the representation that favors the total momentum is more adequate.
c. Show that in the regime ~A µB B, it is convenient to use the representation that privileges
the individual spins of the total momentum.
d. Analyze the intermediate regime ~A ' µB B.
8.4.2.4 Ex: Breit-Rabi formula

Derive the analytical Breit-Rabi formula supposing J = 12 .
8.4.2.5 Ex: Reciprocal pollution of the Paschen-Back and Zeeman regimes

a. Determine the interaction matrix hm̃J m̃I |V̂hf s + V̂zee (B)|mJ mI i of an atom with electron spin
J and nuclear spin J in the decoupled base without considering the quadrupolar terms.
b. Determine the interaction matrix explicitly for the case of 6 Li (I = 1) in its ground state
2S
1/2 (AJ = h · 152.137 MHz) for a magnetic field of B = 100 G.
c. For the system defined in (b) determine the eigenvalues E(B) of the interaction matrix and
the eigenvectors |α(B)i on the decoupled base |mJ mI i.
d. For the system defined in (c) determine the eigenvectors |α(B)i in the coupled base |F mF i.
e. How good are the selection rules for transitions S1/2 − P3/2 in the intermediate regime
between Zeeman and Paschen-Back? We start by calculating the Zeeman shifts for both levels
(s denotes the structure S1/2 , p the structure P3/2 )
s s
B hmJ mI |Hhf s + HB |msJ msI iB = E s (B)
p p
B hmJ mI |Hhf s + HB |mpJ mpI iB = E p (B) .
For the level P3/2 the interval factor is less. In particular for 6 Li it is so small that we are
immediately in the Paschen-Back scheme. This means that the matrix ∞ hm̃pJ m̃pI |mpJ mpI iB =
δmp ,m̃p δmp ,m̃p is diagonal. The element of the transition matrix is then,
J J I I
p p (Eκ)
X X p p p p p p (Eκ)
B hmJ mI |Tq |msJ msI iB = s s s s
∞ hm̃J m̃I |mJ mI iB ∞ hm̃J m̃I |mJ mI iB ∞ hm̃J m̃I |Tq |m̃sJ m̃sI i∞
m̃sJ m̃sI m̃pJ m̃pI
X p p (Eκ)
s s s s
= ∞ hm̃J m̃I |mJ mI iB ∞ hmJ mI |Tq |m̃sJ m̃sI i∞ .
m̃sJ m̃sI
The matrix elements in the pure Zeeman regime can be expressed by [Deh07, unpublished],

hF p mpF |Tq(Eκ) |F s msF i =0 mpJ mpI Tq(Eκ) |msJ msI i0
s 2 p 2
J κ Jp J J s κ (2F s + 1)(2J p + 1)(2κ + 1)
= .
msJ sign(mp − ms ) −mpJ Fs Fp I 2I + 1
p p (Eκ)
Discuss the pure Paschen-Back regime via ∞ hmJ mI |Tq |msJ msI i∞ .
8.4.2.6 Ex: Diamagnetism of the ground states of H and He atoms
a. Calculate the quadratic Zeeman effect for the ground state of the hydrogen atom caused by
the (usually neglected) diamagnetic term of the Hamiltonian in first order TPIT. Write down
the energy shift as ∆E = − χ2 B 2 in order to obtain the diamagnetic susceptibility χ.
b. Estimate the energy shift due to diamagnetism of the ground state of helium using the orbital
3
Zef f −Zef f (r1 +r2 )/aB 5
ψ(r1 , r2 ) = πa3B
e with Zef f = 2 − 16 obtained by the variational method. The
measured value is χ = 1.88 × 10−6 cm3 /mol.
Using the Hamiltonian for an atomic electron in a
magnetic field determine, for a state with angular orbital momentum 0, the energy shift in B 2
when the system is in a uniform magnetic field represented by the potential vector A = 12 B × r.
What is diamagnetic susceptibility χ for a helium atom in the ground state? Compare with the
measured value.
8.4.3 Interaction with electric fields

8.4.3.1 Ex: Stark effect in hydrogen
Consider the hydrogen atom immersed in a uniform electric field E applied along the êz -direction.
The term corresponding to this interaction in the total Hamiltonian is H (1) = −eEz. For typical
electric fields produced in laboratory, the condition H (1) H0 , which allows the use of TIPT, is
satisfied. The effect of the perturbation H (1) , called Stark effect, is the removal of the degeneracy
of some of the hydrogen atom states. Calculate the Stark effect for the state n = 2 of the
hydrogen atom.
8.4. EXERCISES 207
8.4.3.2 Ex: Stark effect in 1s hydrogen level

Calculate the Stark shift of the hydrogen ground state by taking into account the contributions
of the excited states n = 2, 3, ...
8.4.3.3 Ex: Stark effect

Derive the Eqs. (8.64) from the formula (1.4).
Chapter 9
Atoms with many electrons

9.1 Symmetrization of bosons and fermions
Quantum mechanics must be formulated in a way to avoid any possibility of distinguishing
identical particles. However, the language of mathematics automatically assigns a particle to a
wavefunction; for example, ψa (x1 ) is the wavefunction a of particle 1 and ψb (x2 ) the wavefunction
b of particle 2. In the absence of interactions, the total wavefunction, Ψ = ψa (x1 )ψb (x2 ), solves
the Schrödinger equation of two particles. Now, by changing the coordinates of the particles we
get a different state Ψ0 = ψa (x2 )ψb (x1 ) 1 . This erroneously suggests that the wavefunction of a
particle plays the role of a label (or ’soul’) characterizing the particle beyond its set of quantum
numbers. Why this is a problem, we will see in the following example.
Example 35 (Indistinguishability of particles): We consider a system of two non-

interacting spinless particles in an infinite potential well. The total wavefunction is,
na πx1 nb πx2
Ψ(1,2) ≡ ψa (x1 )ψb (x2 ) = C cos cos
L L
with the energy,
π 2 n2a π 2 n2b
Ea,b = + .
2mL2 2mL2
For observable quantities, such as |Ψ(1,2) |2 , we must ensure, |Ψ(1,2) |2 = |Ψ(2,1) |2 , that is,
nb πx2 nb πx1
C 2 cos2 na πx1
L cos2 L = C 2 cos2 na πx2
L cos2 L ,
but this is not valid for na 6= nb . If na = nb , we have ψa = ψb . That is, the particles stay in
the same state and we do not need to worry about indistinguishability:
Ψ(2,1) = ψa (x2 )ψb (x1 ) = Ψ(1,2) and Ea,b = Eb,a .
However, the fact that this state is never observed with two electrons shows, that theory
must be corrected to allow a true description of reality.
We need to construct the total wavefunction in another way. Let us consider linear combi-
nations of Ψ(1,2) ,
ΨS,A ≡ √1 (Ψ(1,2)
2
± Ψ(2,1) ) = √1
2
[ψa (x1 )ψb (x2 ) ± ψa (x2 )ψb (x1 )] . (9.1)
1
We note that states are orthogonal because
Z Z Z Z
Ψ∗(1,2) Ψ(2,1) dx1 dx2 = ψa∗ (x1 )ψb∗ (x2 )ψa (x2 )ψb (x1 )dx1 dx2 = ψa∗ (x1 )ψb (x1 )dx1 ψb∗ (x2 )ψa (x2 )dx2 = δna ,nb .
209
210 CHAPTER 9. ATOMS WITH MANY ELECTRONS
This symmetrized wavefunction (or anti-symmetrized) represents a trick to eradicate the label
sticking to the particles. For, under particle exchange described by the operator Px ψa (x1 )ψb (x2 ) ≡
ψa (x2 )ψb (x1 ), the (anti-)symmetrized functions behave like 2 ,
Px ΨS,A = ±ΨS,A while Px Ψ(1,2) = Ψ(2,1) 6= ∓Ψ(1,2) . (9.2)
The (anti-)symmetrized function solves the Schrödinger equation, as well. As [Ĥ, Px ] = 0, we

can say that the system exhibits an exchange symmetry or exchange degeneracy upon particle
exchange. Observables such as Ψ∗S,A ΨS,A stay conserved, for example, the probability

|ΨS,A |2 = 12 |ψa (x1 )ψb (x2 )|2 + |ψa (x2 )ψb (x1 )|2 (9.3)
± 21 [ψa∗ (x1 )ψb∗ (x2 )ψa (x2 )ψb (x1 ) + ψa∗ (x2 )ψb∗ (x1 )ψa (x1 )ψb (x2 )] = Px |ΨS,A |2
does not change, when we exchange x1 for x2 . For x1 = x2 , we observe,
|ΨS,A |2 = |ψa (x)ψb (x)|2 ± |ψa (x)ψb (x)|2 . (9.4)
That is, for a symmetric system, the probability of finding two particles at the same location is
doubled, whereas for an anti-symmetric system, this probability is zero.
Wolfgang Pauli showed that the (anti-)symmetric character is related to the spin of the
particles. Particles with integer spin called bosons must be symmetric. Particles with semi-
integer spin called fermions must be anti-symmetric. Electrons are fermions. Therefore, in an
atom, they can not be in the same state (location), but must be distributed over a complicated
shell of orbitals. We note, that this applies not only to elementary particles, but also to composed
particles such as, for example, atoms. We will determine in Exc. 15.3.1.1 the bosonic or fermionic
character of several atomic species.
9.1.1 Pauli’s Principle

Two electrons with anti-parallel spins can be separated by inhomogeneous magnetic fields, even
if they are initially in the same place. Therefore, they are distinguishable and the wavefunction
need not be anti-symmetric. But if we exchange the spin along with the position, the particles
must be indistinguishable. This must be taken into account in the wavefunction by assigning a
spin coordinate, ψa (x1 , s1 ). The exchange operator should now be generalized,
Px,s Ψ(1,2) ≡ Px,s ψa (x1 , s1 )ψb (x2 , s2 ) = ψa (x2 , s2 )ψb (x1 , s1 ) = Ψ(2,1) . (9.5)
We now assume that the electrons not only do not interact with each other, but there is
also no interaction between the position and the spin of each electron. That is, we will discard
L · S-coupling.3 We can then write the total wavefunction of an electron as the product of a
spatial function, ψ(x), and a spin function, χ(s) = α ↑ +β ↓, where α and β are probability
amplitudes of finding the electron in the respective spin state, such that,
ψ(x, s) = ψ(x)χ(s) . (9.6)
For two particles, the total spin function is,
X (1,2) = χa (s1 )χb (s2 ) . (9.7)

2
To guarantee |Px Ψ(1,2) |2 = |Ψ(1,2) |2 , we have Px Ψ(1,2) = eiφ Ψ(1,2) . From this, Px Px Ψ(1,2) = e2iφ Ψ(1,2) =
(1,2)
Ψ . Hence, Px Ψ(1,2) = ±Ψ(1,2) .
3
In the case of L · S-coupling, the total wavefunction can not be written as a product of spatial and spin
functions, but it must be anti-symmetric anyway.
9.1. SYMMETRIZATION OF BOSONS AND FERMIONS 211
The (anti-)symmetrized version is

X S,A = √1 (X (1,2)
2
± X (2,1) ) = √1 [χa (s1 )χb (s2 )
2
± χa (s2 )χb (s1 )] . (9.8)
Since there are only two spin directions, there are four possibilities to attribute the spins ↑ and
↓ to the functions χm (sn ),

 ↑↑ = χ1,1
S 1
X = √ (↑↓ + ↓↑) = χ1,0 and X A = √12 (↑↓ − ↓↑) = χ0,0 (9.9)
 2
↓↓ = χ1,−1
For the total wavefunction, which must be anti-symmetric for electrons, there are two possibili-
ties,
 S A 1

 Ψ X = Ψ(1,2) + Ψ(2,1) X (1,2) − X (2,1) = √1
[ψa (x1 )ψb (x2 ) + ψa (x2 )ψb (x1 )] χ0,0

 2 2 
ΘA =  χ1,1 .
1

 ΨA X S = Ψ(1,2) − Ψ(2,1) (X (1,2) + X (2,1) ) = √12 [ψa (x1 )ψb (x2 ) − ψa (x2 )ψb (x1 )] χ1,0
 2 
χ1,−1
(9.10)
That is, the two electrons may be in a triplet state with the anti-symmetric spin wavefunction,
or in a singlet state with the symmetric spin wavefunction 4 .
How to generalize these considerations to N electrons? The symmetric wavefunctions contain
all permutations of the label ak , where we understand by ak the set of quantum numbers
unambiguously specifying the state of the particle k,
X
ΘS = N ψa1 (x1 )...ψaN (xN ) , (9.11)
Px,s ak
with a normalization factor N 5. The (anti-)symmetrized wavefunction is obtained from the

Slater determinant,

ψa (x1 ) · · · ψa1 (xN )
1
A 1 1 .. .. ..
Θ = N ! det ψak (xn ) = N ! . . . . (9.12)

ψa (x1 ) · · · ψa (xN )
N N
This function satisfies

Px,s ΘA,(1,..,i,j,..,N ) = ΘA,(1,..,j,i,..,N ) . (9.13)
The Slater determinant is zero, when two sets of quantum numbers are identical, ai = aj . For
example, for two electrons in an electronic shell, |ni , li , mi , si i = |nj , lj , mj , sj i. This is Pauli’s
strong exclusion principle:
The total wavefunction must be antisymmetric with respect to the exchange of
any pair of identical fermions and symmetrical with respect to exchange any pair of
identical bosons.
Pauli’s weak exclusion principle (usually sufficient for qualitative considerations) says that two
fermions in identical states can not occupy the same region in space. That is, their Broglie waves
interfere destructively, as if Pauli’s principle exerted a repulsive interaction on the particles. This
’force’ has a great impact on the phenomenology of the bonds between atoms, as we will discuss
in the following sections.
4
In the coupled image, the total spin S = s1 + s2 can have the following values S = |s1 − s2 |, .., s1 + s2 = 0, 1.
In the case S = 0 the magnetic quantum number can only have one value (singlet), mS = 0. In the case S = 1 it
can have three values mS = −1.0,
q +1 (triplet) (see Exc. 8.4.2.3).
5 Πm
k=1 k
n !
It is possible to show N = N!
, where nk is the population of state ψak , that is, the number of particles
with the same set of quantum numbers ak .
9.1.2 Consequences for quantum statistics

The indistinguishability of quantum particles has interesting consequences on the statistical
behavior of bosons and fermions. This becomes obvious when we consider two particles 1 and 2
being able to adopt two different states a and b. Distinguishable particles can be in one of the
following four states,
Ψ = {ψa (x1 )ψa (x2 ), ψa (x1 )ψb (x2 ), ψb (x1 )ψa (x2 ), ψb (x1 )ψb (x2 )} (9.14)
with the same probability of p = 1/4. Bosonic indistinguishable particles can stay in one of the
following three states,
Ψ = {ψa (x1 )ψa (x2 ), √12 [ψa (x1 )ψb (x2 ) + ψb (x1 )ψa (x2 )], ψb (x1 )ψb (x2 )} (9.15)
with the same probability of p = 1/3. Finally, fermionic indistinguishable particles can only be
in a state,
Ψ = { √12 [ψa (x1 )ψb (x2 ) − ψb (x1 )ψa (x2 )]} (9.16)
with the same probability of p = 1. We see that a simple two-particle system already exhibits
qualitative modifications of its statistical behavior. These differences generate different physics
as we deal with systems of large numbers of particles, as we can see in the cases of the free
electron gas and the Bose-Einstein condensate. For a broader discussion, see Chap. ??.
We finally note a result of the standard model of particle physics assigning a fermionic
character to all fundamental constituent particles of matter while the mediators of fundamental
forces are always bosons.
9.2 Helium
The simplest atom to discuss Pauli’s principle is helium. The helium atom has a charged nucleus
Z = +2e and mass mHe ≈ 4mH .
9.2.1 The ground state

The ground state of the helium atom brings together the two electrons, that is, (1s)2 . To
treat the helium atom we can, as a first trial, describe the atom by the Bohr model, assuming
independent electrons. Neglecting the electronic repulsion term (which depends on r12 ), we can
separate the total wavefunction:
Ψ(r1 , r2 ) = Ψ1 (r1 )Ψ2 (r2 ) , (9.17)
and we get two Schrödinger equations, the Hamiltonian being equal to the one of hydrogen-like
atoms: 2
~ 2 e2 Z
− ∇i − Ψi (ri ) = En(i) Ψi (ri ) , (9.18)
2µ 4πε0 ri
with i = 1, 2. For hydrogen-like atoms we have,

1 1
E = En(1) + En(2) = EB Z 2 + , (9.19)
n21 n22
with EB = −13.6 eV. With this, we get the energy for the ground state:
EHe (1s) = −2Z 2 EB = −108.8 eV . (9.20)

9.2. HELIUM 213
The value predicted by Bohr’s model is far from experimental reality: The ionization energy
measured for the first electron is 24.6 eV, for the second 54.4 eV, totalizing a binding energy for
two electrons of −78.983 eV . This corresponds to an error of about 38%. The lower energy of
the first electron is due to the shielding of the nucleus by the second.
9.2.1.1 First-order perturbation of the energy

Treating the repulsion term between the electrons as a perturbation [11] and using the eigenfunc-
tions of hydrogen atoms |n, `, m` i, the total wavefunction is |n1 , `1 , m`1 ; n2 , `2 , m`2 i, we obtain
as first order TIPT correction for the energy:
e2
∆E = hn1 , `1 , m`1 ; n2 , `2 , m`2 | |n1 , `1 , m`1 ; n2 , `2 , m`2 i . (9.21)
4πε0 r12
This correction is called the Coulomb integral and has the value:
Z
e2 1
∆E = |Ψn1 ,`1 ,m`1 (r1 )|2 |Ψn2 ,`2 ,m`2 (r2 )|2 dV1 dV2 . (9.22)
4πε0 r12
This integral is always positive. The term |Ψn1 ,`1 ,m`1 (r1 )|2 dV1 is the probability of finding the
electron inside the volume element dV1 and, when multiplied by −e, gives the charge associated
with that region. Thus, the integral represents the coulombian interaction energy of the confined
loads within the two volume elements dV1 and dV2 . ∆E is the total contribution to the potential
energy. Calculating the Coulomb integral for the ground state, which will be done in Exc. 9.5.2.1,
we obtain,
5Z e2 5Z
∆E = = EB , (9.23)
4 4πε0 2aB 4
with aB the Bohr radius. ∆E corresponds to 34 eV. Thus, the ground state energy is EHe (1s) =
−108.8 eV + 34 eV = −74.8 eV. Comparing with the experimental value of −78.983 eV we have
an error around 5.3%.
9.2.1.2 Shielding of the nuclear charge

We can make the approximation in which we consider that each electron moves in a coulombian
potential, with respect to the nucleus, shielded by the charge distribution of the other electron
[71]. The resulting potential will be generated by an effective charge ζe ≡ (Z − B)e. The
quantity B ∈ [0, 1] is called the shielding constant.
The first electron feels a total nuclear charge Ze, while the second feels an effective nuclear
charge ζe. We exchange Z for ζ in the energy term for hydrogen-like atoms,
EB
En = −ζ , (9.24)
n2
and the energy for the ground state becomes, assuming total shielding, B = 1,
E = E1 + E2 = −Z 2 EB − ζ 2 EB = −4EB − EB = −5EB = −67.5 eV . (9.25)
Comparing with the experimental value of −78, 983 eV we have an error around 15%. For
a shielding constant of around B = 0.656 the experimental value is reproduced. This means
that the effective nuclear charge felt by the second electron is partly shielded by the former.
The TPIT method (9.21) and the shielding concept (9.24) can be combined in a variational
calculation, where the effective charge ζ is the variational parameter. In Exc. 9.5.2.2 we study
the reciprocal shielding of the electrons at the example of the helium-type ion H− .
9.2.2 Excited states
Let us now investigate the excited states of helium, in particular those, where only one electron
is excited, the other one being in the ground state, (1s)1 (2s)1 and (1s)1 (2p)1 . The energy of
the electron in the shell n = 2 is smaller than predicted by Bohr’s model with Z = 2 because of
the interaction with the other electron. Also, the (2s) and (2p) levels are no longer degenerate,
because the electrostatic potential is no longer Coulombian (see Fig. 9.1).
Figure 9.1: Helium levels and allowed singlet and triplet transitions. Note that the state (1s)↑↑
does not exist.
As we have seen in the discussion of the fine structure, the energy of the l · s-coupling
given by (??) is ∝ En (Zα)2 ∝ Z 4 . For helium which still has a small Z, the energy of the
coupling is weak, so that we can count on a direct coupling of the two spins. Since the orbits of
the electrons are now different, we can construct combinations of symmetric or antisymmetric
spatial wavefunctions ΨS,A , are therefore combinations X A,S of anti-parallel or parallel spins.
When the spins are parallel (S = 1), the spatial wavefunction is antisymmetric, when they are
antiparallel (S = 0), it is symmetric. From the symmetry of the wavefunction depends the
energy of the coulombian interelectronic interaction, because in the symmetric state the average
distance of the electrons is much smaller than in the antisymmetric state, where the total spatial
function disappears for zero distance. Consequently, the configuration (1s)1 (2s)1 has two states
with S = 0.1, with energy ES=0 > ES=1 . Likewise, all configurations are split, as shown in
Fig. 9.1. The energy difference (∼ 1 eV) is considerable and well above the energy of fine
structure interaction (∼ 10−4 eV). This explains why the two spins first couple to a total spin,
s1 + s2 = S, before this spin couples to the total orbital angular momentum, S + L = J. This
is the L · S-coupling.
9.2.2.1 Exchange energy
The energy difference between the two states S = 0.1 is called exchange energy. It comes out of a
first-order perturbation calculation. For example, for the (1s)1 (2s)1 we write the anti-symmetric
wavefunctions,
ΘA
± =
√1
2
[ψ100 (r1 )ψ200 (r2 ) ± ψ100 (r2 )ψ200 (r1 )] · χA,S , (9.26)
9.2. HELIUM 215
where the (+) sign holds for χA (S = 0) and the (−) sign for χS (S = 1). The energies are,
Z Z
e2
∆E S,A = 21 dr13 dr23 Θ∗A± ΘA,± (9.27)
4π0 |r1 − r2 |
Z Z
e2
= 2 dr1 dr23
1 3
[|ψ100 (r1 )|2 |ψ200 (r2 )|2 + |ψ100 (r2 )|2 |ψ200 (r1 )|2 ]
4π0 |r1 − r2 |
Z Z
e2
± 2 dr1 dr23
1 3
2ψ ∗ (r1 )ψ200 ∗
(r2 )ψ100 (r2 )ψ200 (r1 )
4π0 |r1 − r2 | 100
≡ ∆Ecoulomb ± ∆Eexchange .
The first integral,

Z Z
e2
∆Ecoulomb = dr13 dr23 |ψ100 (r1 )|2 |ψ200 (r2 )|2 , (9.28)
4π0 |r1 − r2 |
is the Coulomb energy between the electronic orbitals. We note that this part can be calculated
from the Hamiltonian using non-symmetrized orbitals. The second integral,
Z Z
e2
∆Eexchange = dr13 dr23 ψ ∗ (r1 )ψ200
∗
(r2 )ψ100 (r2 )ψ200 (r1 ) , (9.29)
4π0 |r1 − r2 | 100
called exchange energy corresponds to the interference terms of the symmetrization and must
be added or subtracted according to their symmetry character. It is interesting to note that up
to this point the spin does not enter directly into the helium Hamiltonian,
p21 p2
Ĥ S,A = + 2 + V (r1 ) + V (r2 ) + V (|r1 − r2 |) ± ∆Eexchange , (9.30)
2m 2m
but only through the symmetry character of the spatial wavefunction. On the other hand, on a
much smaller energy scale, the spin enters through the L · S-interaction.
The potential is not spherically symmetric, the term r12 depends on the angle between r1
and r2 . Thus, the total wavefunction Ψ(r1 , r2 ) is not separable into a radial and an angular part.
By consequence, unlike for hydrogen, the Schrödinger equation with the Hamiltonian (9.20) has
no analytical solution.
Example 36 (TDPT for excited helium states): We consider the two electrons of
a helium atom occupying different orbits described by wavefunctions denoted by ψa (1) ≡
ψn1 ,`1 ,m`1 (r1 ) and ψb (2) ≡ ψn2 ,`2 ,m`2 (r2 ). Applying the Hamiltonian without the interelec-
tronic interaction term, the total states Θ = ψa (1)ψb (2) and ψa (2)ψb (1) have the same energy
Ea + Eb . To calculate the energy correction, we use TIPT for degenerate states. We have to
(1)
calculate the secular determinant det(hn, ν|H (1) |n, µi − En,µ δµ,ν ). The terms of the matrix
H (1) are:
(1) e2
H11 = hψa (1)ψ( 2)| |ψa (1)ψb (2)i
4πε0 r12
(1) e2
H22 = hψa (2)ψb (1)| |ψa (2)ψb (1)i
4πε0 r12
(1) e2 (1)
H12 = hψa (1)ψb (2)| |ψa (2)ψb (1)i = H21 .
4πε0 r12
(1) (1) (1)
The terms J ≡ H11 = H22 are Coulomb integrals. The term K ≡ H12 is called exchange
integral:
e2 1
K= hψa (1)ψb (2)| |ψa (2)ψb (1)i .
4πε0 r12
Hence, as J and K are positive, the determinant is:

J − E K
=0,
K J − E
yielding,
E (1) = J ± K .
That is, the states that were previously degenerate with energy E = Ea + Eb are now split
into two states with energies E = Ea + Eb + J ± K. And the corresponding eigenfunctions
are:
1
ΨS,A (1, 2) = √ [ψa (1)ψb (2) ± ψb (1)ψa (2)] .
2
This result shows that the repulsion between the two electrons breaks the degeneracy (of
separable functions written in product form) into states with an energy difference 2K. Note
that the eigenfunctions are symmetric, which is discussed in the next section.
9.2.2.2 The spectrum of helium
So far we have seen that, if the electrons are in the same orbital, we have an energy term
E = 2Ea + J and, when they are in different orbitals, we have E = Ea + Eb + J ± K, with a
separation between levels of 2K.
In practice, we consider only the excitation of one electron, because the energy to excite the
two electrons exceeds the ionization energy of the helium atom. To find the selection rules for
transition between symmetric and antisymmetric states, we calculate the dipole moment of the
transition. For a two-electron system the dipole moment is d = −er1 − er2 , which is symmetric
with respect to a permutation of the two electrons. The dipole moment for transition is:
Z
d± = −e Ψ∗A (r1 , r2 )(r1 + r2 )ΨS (r1 , r2 )dV1 dV2 . (9.31)
If we exchange the electrons, the above integral changes sign, but the integral can not depend
on the nomenclature of the integration variables, so it must be zero. The transition between
a symmetric and an antisymmetric state can not occur. Considering the spin wavefunction,
Θ = ΨS χA or ΨA χS , we find that transitions are only allowed between singlet states or between
triplet states. That is, there is a selection rule for the spin postulating ∆S = 0 6,7 .
Because of the differences observed in the singlet and the triplet spectrum of helium, il-
lustrated in Fig. 9.1, it was first believed that they belong to different atomic species, called
para-helium e ortho-helium. A chemical analysis showed later that it was the same element.
6
Moreover, transitions between the states 1 S0 and 3 S1 are impossible, because they violate the selection rule
for the angular momentum, ∆L = ±1.
7
We can understand the selection rules as follows: As long as the wavefunction can be written as a product,
Θ = Ψ(x)χ(s), the symmetry character is preserved for the two functions separately. The eigenvalues of the
operators Px and Ps are then good quantum numbers. But this only holds for weak L · S-coupling. The electric
dipole operator for the transition does not act on the spin (which prevents the recoupling S = 1 ↔ S = 0
via E1-radiation) and also does not act on the symmetry character of the orbitals (which prevents transitions
ΨS ↔ ΨA ).
In principle, this holds for any species of atoms with two valence electrons. In reality however, the influence of
the L · S-coupling grows with Z, which weakens the interdiction of the intercombination transition. In this case,
only the operator Px,s yields good eigenvalues.
9.3. ELECTRONIC SHELL STRUCTURE 217
9.3 Electronic shell structure

The interelectronic interaction and the need to antissimetrize the wavefunction of the electrons
both contribute to excessively increase the complexity of multielectronic atoms. The Hamilto-
nian describing a multielectronic atom of atomic number Z,
Z
X Z
X Z
X
p2i Ze2 e2
Ĥ = Ekin + Vncl−ele + Vele−ele = − + , (9.32)
2m 4π0 |r − ri | 4π0 |ri − rj |
i=1 i=1 i<j=1
is extremely complicated to solve, even for the simplest case (Z = 2) we must use approximation
methods.
Note that, if we assume independent electrons (Vele−ele = 0), that is, each electron moves
independently of the others within the electrostatic potential generated by the nucleus and the
other Z − 1 electrons, the problem would be easily solvable: We would solve the Schrödinger
equation for a product state of all the electronic wavefunctions, and we would know the eigen-
functions and individual eigenenergies of each electron (as for the hydrogen atom). In principle,
we should use antisymmetric wavefunctions, but as a first approach we can choose to only respect
Pauli’s weak principle, that is, assign an individual and unique set of quantum numbers to each
electron. The total energy would be the sum of the energy of every electron, and the associated
physical eigentates would be obtained by means of an antisymmetrization of the tensor product
of the multielectronic state.
Example 37 (Fermi gas): We consider an infinite potential well, that we gradually fill up
with electrons. The Pauli principle allows us to place at most two electrons in each orbital,
Ψ = ψ1,↑ (x1 )ψ1,↓ (x2 )ψ2,↑ (x3 )ψ2,↓ (x4 ) · .. .
This total wavefunction satisfies the weak Pauli principle, but is obviously not antisymmet-
ric. The approximation is good, when the interaction between the electrons is negligible.
Otherwise, we need to consider the exchange energy terms.
This model, called Fermi gas model, is often used to describe the behavior of electrons that
can freely move within the conductance band of a metal. Let N be the number of electrons
placed in the potential successively filling all the orbitals with quantum number n up to
the maximum number nmax = N/2, provided each orbital can withstand two electrons with
antiparallel spins. For a one-dimensional well of size L, the maximum energy, called Fermi
energy is,
N2
EF ∝ 2 .
L
For a three-dimensional well of volume V ,
2/3
N
EF ∝ .
V
9.3.1 Thomas-Fermi model

Even if the real potential sensed by the electrons bound to a nucleus is very different from the
three-dimensional well, we can roughly imagine that the atom is subdivided into small volumes,
all filled with electrons following the Fermi gas model. From this we can calculate the distribution
of the electronic charge, such that the average local energy is homogeneous and the electronic
cloud in equilibrium. The distribution, in turn, serves to determine the shape of the electrostatic
potential which, when subdivided into small volumes filled with electrons, produces the same
charge distribution. This principle is called self-consistency.
The wavefunctions (spatial distributions of electrons) determined by this method often serve
as a starting point for the Hartree method discussed below. One of the important predictions of
the Thomas-Fermi model is that the average radius of an atom depends on the nuclear charge
as R̄ ∝ Z −1/3 .
The Thomas-Fermi model allows us to understand the electronic configuration of the fun-
damental states and provides the basis for the periodic system of elements. In this model, the
electrons are treated as independent particles, on one side forming an effective radial electric
potential, on the other side being subjected to this potential. Instead of requiring anti-symmetry
of the wavefunction, it is only necessary to ensure that all electrons are distinguished by at least
one quantum number. The functions of complex atoms are similar to the functions of hydrogen.
We can use the same quantum numbers n, `, m` , and ms for every electron.
However, the effective radial potential depends very much on the species and is quite different
from the Coulomb potential. So the degeneracy in ` is lifted. In general, electrons with small
` are more strongly bound, because they have a higher probability of being near the nucleus,
where the potential is deeper (see Fig. 9.7). The same argument explains why electrons with
small n are more strongly bound. We will discuss these effects in more depth in Sec. 9.4.2 by
comparing the excitation levels of the valence electron in different alkalis.
9.3.1.1 TPIT method

Thus, as a first approximation we use the states of individual electron (orbital approximation)
and consider Vele−ele (|ri − rj |) as a perturbation making use of time-independent perturbation
theory. However, this term is not small enough to justify this procedure, since approximating
Z 2 e2 Z(Z − 1)e2
Vncl−ele ' and Vele−ele ' , (9.33)
aB 2aB
we realize that Vele−ele /Vncl−ele varies between 14 for Z = 2 and 21 for Z 1/2. For this reason
the use of alternative methods to describe multielectronic atoms is necessary.
The Thomas-Fermi model is a semi-classical model that aims to roughly describe the total
energy of the electrons as a density functional of atomic/molecular electrons. Such a model
is the basis for more sophisticated methods to obtain the electronic structure, such as density
functional theory (DFT).
9.3.1.2 Electron gas model

Now we divide the atom into small volumes of size L (cells), containing N uniformly distributed
non-interacting electrons whose total number is Nt , and we analyze each cell individually. The
volume can be modeled by the following potential: V (r) = 0 for 0 ≤ x, y, x ≤ L and V (r) = ∞
in all other places. In this case we find the possible states {|nx , ny , nz i} with nx , ny , nz = 1, 2, 3
and the single electron energies,
π 2 ~2
Enx ,ny ,nz = (n2 + n2y + n2z ) . (9.34)
2me L2 x
For N electrons in the ground state, each particle occupies the energetically lowest state available,
obeying the Pauli exclusion principle and considering the spin. The total energy is given by the
sum of the energies of the N less energetic states, and the final physical state is given by the
antisymmetrization of the corresponding state. The energy of the N -th electron (the most
energetic one) is called the Fermi energy (EF ) 8 .
It is interesting to calculate the number of existing states n(E) whose energy is less than E.
For this we write the energy as follows,
~2 2
Enx ,ny ,nz = k , (9.35)
2me nx ,ny ,nz
n π
where knx ,ny ,nz = kx2 + ky2 + kz2 = ( nLx π )2 + ( Ly )2 + ( nLz π )2 . Each set of values k = (kx , ky , kz )
corresponds to a possible state which is accessible to the system, and each state is associated
with a volume element (π/L)3 in k-space. Note that the p number of states n(E) with energy less
given by the number of states in which k ≤ 2me E/~2 . This corresponds to a sphere
than E is p
of radius 2me E/~2 centered on the origin, thus
" 3/2 # 3/2

2me E 1 L3 2me E
n(E) = 2 18 4π
= . (9.36)
3 ~2 (π/L)3 3π 2 ~2
Fig. 9.2 illustrates the possible states in k-space and the sphere with energy E.
Figure 9.2: Available states for a box potential.
With the above expression we are able to calculate the Fermi energy, since the condition
n(EF ) = N implies,
2/3
~2 2N
EF = 3π 3 . (9.37)
2me L
It is interesting to note that for T = 0 K all states below EF are occupied and all those above
are unoccupied. In addition, we define the density of states ρ(E), where ρ(E)dE is the number
of states with energies between E and E + dE. Thus, the density of states becomes,
3/2
dn(E) L3 2me 3N E 1/2
ρ(E) = = 2 E 1/2 = . (9.38)
dE 2π ~2 2 E 3/2
F
8
Note, however, that for large N (e.g., 1 mol of electrons) this process turns out to be unfeasible.
9.3.1.3 Thomas-Fermi energy

The average kinetic energy of the electrons with the system being in its ground state is,
Z Z
EF
L3 2me 3/2 EF 3/2
Ec = Eρ(E)dE = 2 E dE (9.39)
0 2π ~2 0

L3 2me 3/2 2 5/2
= 2 5 EF
2π ~2

~2 35/3 π 4/3 3 N 5/3
= L = CL3 ρ5/3 ,
10me L3
where ρ ≡ N/L3 is the density of electrons per unit volume. Hence,

Z
Nt = ρ(r)d3 r . (9.40)
Thus, the kinetic energy density is given by,
ukin (r) = Cρ5/3 . (9.41)
Therefore, the total kinetic energy of the electrons in the electronic shell is,
Z
T [ρ] = C ρ5/3 d3 dr . (9.42)
The potentials associated with the interactions e− -e− and e− -p+ can also be written as func-
tionals of electronic density, such that,
Z Z
Ze2 ρ(r) 3 1 e2 ρ(r)ρ(r0 ) 3 3 0
Vep [ρ] = − d r and Vee [ρ] = d rd r . (9.43)
4π0 r 2 4π0 |r − r0 |
Thus, the total energy (Thomas-Fermi energy) can be written as a functional of the electronic
density of the atom,
HT H [ρ] = T [ρ] + Vncl−ele [ρ] + Vele−ele [ρ] . (9.44)
9.3.1.4 Electronic density and the Thomas-Fermi equation

Taking into account the variational principle, we are interested in the electronic density ρ(r)
which minimizes the Thomas-Fermi energy. We can perform this process via Lagrange multipli-
ers, with the restriction that the number of electrons remains constant in the atom. Thus,
Z
3
δ HT H [ρ] − µ ρ(r)d r − Nt =0 (9.45)
δHT H [ρ]
δHT H [ρ] − µδρ(r) = 0 =⇒ =µ.
δρ(r)
Inserting the Thomas-Fermi energy (9.44) we calculate,

Z
5 2/3 e2 Z e2 ρ(r0 ) 3 0
µ= 3 Cρ − ξef (r) with ξef (r) = − d r . (9.46)
4π0 r 4π0 |r − r0 |
Resolving for the electronic density,

3/2
3
ρ(r) = µ + ξef (r) . (9.47)
5C
The above expression describes the electron density of the atom in its ground state. It is
interesting to note that, as δHT H [ρ]/δρ(r) = µ, we can identify the Lagrange multiplier µ as a
chemical potential. In particular, for non-interacting neutral atoms, we have µ = 0, so,
3/2
3
ρ(r) = ξef (r)3/2 . (9.48)
5C
In addition, since for an atom both the potential and the electronic density must have spherical
symmetry, we can write,
Zχ(r)
ξef (r) = . (9.49)
r
Now, it is common to perform the variable transformation x = αr (α = 27/3 Z 1/3 /(3π)2/3 =
1.1295Z 1/3 ) and use atomic units, where ~2 /me = 1 such that C = 35/3 π 4/3 /10 = 2.871 [11].
Thus, the ’electronic density’ is,
3/2
Z 2 25 χ(x)
ρ(x) = 2 3 . (9.50)
3 π x
In addition, the potential must satisfy the Poisson equation ∇2 ξef = 4πρ, resulting in the
Thomas-Fermi equation:
d2 χ χ3/2
= . (9.51)
dx2 x1/2
It is important to note that the previous equation does not depend on the parameter Z,
thus being a general result for any neutral atom. The function χ(x) is determined numerically,
but we can analyze its asymptotic values given the expected behavior of the effective potential
ξef (r): for r → 0 we expect that ξef (r) = Zr , hence χ(0) = 1; on the other hand for r → ∞ we
expect ξef (r) = 0, hence χ(∞) = 0.
With χ(x) known we get the charge density ρ(x), and hence we are able to calculate the
total energy of the atom under investigation. Thus, it is possible to show that [11],
HT H [ρ] = −0.7687Z 7/3 . (9.52)
It is important to highlight some points: (i) the result holds for neutral atoms; (ii) there is no
electronic shell structure assumed; (iii) quantum statistical effects of identical particles are not
taken into account. A more refined model which deals with the criticism (iii) and, in addition,
is closer to density functional theory (DFT) is the Thomas-Fermi-Dirac model.
9.3.2 Hartree method

To calculate most of the atomic properties we need more realistic potentials. The most important
terms of the Hamiltonian are the coulombian potential between the nucleus and the electrons,
Vncl−ele , being naturally spherical, and the interaction potentials between the electrons, Vele−ele ,
which we will try to approximate by a spherical potential and treat the deviations caused by the
approach afterward. Knowing the effect of the shielding of the nucleus by electronic charges, we
already know the asymptotes (see Fig. 9.7),
Ze2 e2
V0 = − para r→0 and V0 = − for r → ∞. (9.53)
4π0 r 4π0 r
An effective potential, V0 , constructed to satisfy these limits serves as a first trial to establish
and numerically solve the Schrödinger equation for each electron independently,

~2 2
Ĥi = − ∇ + V0 ψi (ri ) = ei ψi (ri ) . (9.54)
2m i
With this we calculate all energies and eigenunctions (only the radial parts are of interest)
minimizing the total energy and respecting the weak Pauli principle, that is, all states are
successively filled with electrons. For the total wavefunction we obtain,
N
! N
X X
Ĥi ΨN = En ΨN com ΨN = ψ1 · .. · ψN and En = ei . (9.55)
i=1 i=1
With the eigenfunctions we calculate the charge densities e|ψj (rj )|2 .
We integrate the field to
obtain a potential that represents an improved estimation for the electronic mean field,
Ze2 XZ e2
Vi ←− − + d3 rj |ψj (rj )|2 . (9.56)
4π0 ri 4π0 |ri − rj |
j6=i
We replace that potential in the Schrödinger equation, and repeat the whole process from the be-
ginning. This self-consistent method is called Hartree method. Fock improved these calculations
using antisymmetric wavefunctions for the valence electrons. This method is called Hartree-Fock
method. The idea of the Hartree method is shown in the following diagram,
choose a common potential V (r) for
all electrons, f.ex. Thomas-Fermi
↓
(0)
calculate
the eigenfunctions
ψi with
−~2 (0) (0) (0)
2m ∆i + V0 (ri ) ψi = ei ψi
↓
new entry
←− ψi ←−
↓ ↓ ↑
↓ classify according to increasing energies ei ↑
↓ satisfying Pauli’s principle ↑
↓ ↓ ↑
↓ calculate the mean potential Vi acting ↑
↓ on electron i for all i ↑
P R
↓ Vi (ri ) = j6=i 4πεZe 0 rij
|ψi (ri )|2 d3 rj ↑
↓ ↓ ↑
↓ calculate
new ψ i with these
Vi ↑
−~2 Ze
↓ 2m ∆i − 4πε0 ri + Vi (ri ) ψi = ei ψi ↑
↓ ↓ ↑
bad
−→ compare with initial ψi −→
↓ good
result
O Método de Hartree-Fock é um método utilizado para tratar sistemas atômicos/moleculares de
muitos corpos que visa a obtenção da função da onda eletrônica do sistema. Ele é um refinamento
do Método
9.3. ELECTRONIC (SCF:
SHELL
de Hartree Self-Consistent-Field ), pois leva em conta a antissimetrização da223
STRUCTURE
função de onda. De maneira geral o método se baseia no princípio variacional e na suposição
9.3.3 Hartree Fock method
de que podemos escrever a função de onda global como um determinante de Slater, sendo que
The Hartree-Fock method used to treat atomic or molecular many-body systems aims at obtain-
ingcada
the elétron ocupa um estado orbital específico (spin-orbital) e interage com um potencial efetivo
electronic wavefunction of the system. It is a refinement of the Hartree method (SCF:
oriundo dos elétrons que
Self-Consistent-Field), ocupam
since outros
it takes orbitais.
into accountDesta
themaneira, ao invés de resolvermos
antisymmetrization a equação
of the wavefunction.
In general, the method is based on the variational principle and on the assumption that we
de Schrödinger devemos resolver um sistema de equações denominada de Equações de Hartree-
can write the global wavefunction as a determinant of Slater, with each electron occupying a
specific do tipo state
Fock orbital (spin-orbital)
F̂ ψk (1) O método
= εk ψk (1). and é realizado
interacting de maneira
with an effectivecíclica até convergência
potential stemming from dos the
electrons which occupy other orbitals. In this way, instead of solving the Schrödinger equation
orbitais atômicos e suas respectivas energias, por este motivo é denominado como auto-consistente:
we must solve a set of equations called Hartree-Fock equations of the type F̂ ψk (1) = k ψk (1).
Thea partir
methodda suposição inicial
is performed da funçãountil
cyclically de onda global é calculado
convergence o potencial
of the atomic efetivo
orbitals andemtheir
cada respective
orbital
energies is reached in a procedure called self-consistent: Starting from an initial trial global
e é obtido um novo conjunto de funções de onda que, por sua vez, geram um novo potencial efetivo;
wavefunction we calculate the effective potential in each orbital and a new set of wavefunctions
assim,ineste
which, novo
turn, potencial
generate é utilizado
a new em um
effective novo sistema
potential. This de
newequações de Hartree-Fock.
potential is then used inA Figura
a new set
of 2.1
Hartree-Fock equations. Fig. 9.3 illustrates
ilustra a característica cíclica do método.
the iterative procedure.
Definição dos
spin-orbitais
Não
Cálculo dos po-

Resolução das Equa-
Cálculo dos Potenciais tenciais com os Convergência?
ções de Hartree-Fock
novos spin-orbitais
Sim
Fim
2.1: Método
Fig.Figure de Hartree-Fock.
9.3: Hartree method.
9.3.3.1 Hartree-Fock
2.2 Equações equations
de Hartree-Fock -
We have seen that the Hamiltonian of a multielectronic atom is given by,
Vimos que o Hamiltoniano de um átomo multieletrônico é dado por
XZ 2 X X Z X
pi Ze2 1 e2 1
Ĥ = − + = ĥ i + V̂ij , (9.57)
Z 0ri 2 2
24π0 |ri X
2
i=1
2mi 4π
X pi Ze
i6= j 1 − rj | 2
ei=1 i6=j
Ĥ = − +
i=1
2mi 4π0 ri 2 4π0 |rj − ri |
i6=j
where ĥi is the Hamiltonian onlyZ of the electron i, and V̂ij is the interaction term between the
X 1 Xwe must suppose that the multielectronic state can
electrons i and j. To implement
= theĥmethod
i+ V̂ji
be written as the product of thei=1 2 states of each electron:
individuali6=j
Ψ0 (1, ..., Z) = ψ1 (1)ψ2 (2)...ψZ (Z) , (9.58)

8
where ψi (1) = φi (r1 χ(α) = ψiα (r1 ) represents the spin-orbital state of electron 1, that is, the
spatial wavefunction of the electron in the state i and with spin α. However, due to the sym-
metrization postulate, the physical state of the system must be expressed by a Slater determi-
nant,
1
Ψ(1, ..., Z) = √ det [ψ1 (1)ψ2 (2)...ψZ (Z)] . (9.59)
Z!
Now, we use the variational principle to minimize the expectation value of the ground state
energy by varying the functions ψk (n). In this way, the correct orbitals are those that minimize
the energy. The expectation value is written as,
Z
X X
E = hΨ|Ĥ|Ψi = hΨ| ĥi |Ψi + hΨ| 21 V̂ij |Ψi . (9.60)
i=1 i6=j
It is possible to show that,

Z
X Z
X
hΨ| ĥi |Ψi = hψi |ĥi |ψi i e (9.61)
i=1 i=1
X XZ
hΨ| 12 V̂ij |Ψi = 1
2 [hψi ψj |V̂ij |ψi ψj i − hψj ψi |V̂ij |ψj ψi i] .
i6=j i,j
Hence,
Z
X Z
X
1
E= ĥi |Ψi + 2 [hψi ψj |V̂ij |ψi ψj i − hψj ψi |V̂ij |ψj ψi i] . (9.62)
i=1 i,j
The above expression can be minimized via Lagrange multipliers under the constraint that the
states are orthogonal hψi |ψj i = δij ,
 
 X 
δ hΨ|Ĥ|Ψi − ij [hψi ψj |V̂ij |ψi ψj i − hψj ψi |V̂ij |ψj ψi i] (9.63)
 
i,j
Thus, we obtain the following set of Hartree-Fock equations:

( )
X
ĥ1 + (2Jî − K̂i ) ψk (1) = k ψk (1) (9.64)
i
F̂ ψk (1) = k ψk (1) ,
P
where, F̂ = ĥ1 + i (2Jî − K̂i ) is the Fock operator and k is the energy associated with the
spin-orbital ψk . The operator Jî , called Coulomb operator, represents the mean potential sensed
by electron 1 in the orbital k due to the presence of electron 2 in the orbital i:
Z
ˆ
Ji ψk (1) = ∗
ψi (2)V12 ψi (2)dr2 ψk (1) . (9.65)
The operator K̂i , denominated exchange operator, is a consequence of the symmetrization process
and therefore a purely quantum effect, that is, without classical analogue:
Z
∗
K̂i ψk (1) = ψi (2)V12 ψk (2)dr2 ψk (1) . (9.66)
Once we known all the wavefunctions, the energies of the orbitals can be obtained in the following
way: ( )
Z X Z
∗
dr1 ψk (1) ĥ1 + ˆ
(2Ji − K̂i ) ψk (1) = k dr1 ψk∗ (1)ψk (1) = k ,
∗
(9.67)
i
that is,
Z X
k = dr1 ψk∗ (1)ĥ1 ψk∗ (1) + (2Jˆki − K̂ki ) , (9.68)
i
where,
Z
Jˆki = dr1 ψk∗ (1)Jî ψk (1) is the Coulomb integral (9.69)
Z
K̂ki = dr1 ψk∗ (1)K̂i ψk (1) is the exchange integral .
The total atomic energy can be calculated by,

X X
E=2 k − (2Jˆki − K̂ki ) . (9.70)
k k,i
Furthermore, if we assume that, taking an electron away from the orbital ψk the electronic
distribution remains unchanged, it is possible to associate the energy k with the ionization
energy of the electron in this orbital. This equality is known as Koopman’s theorem,
Ik ' k . (9.71)
9.3.3.2 Hartree-Fock-Roothaan equations
Finally, it is worth emphasizing that a refinement of the self-consistent procedure can be ob-
tained, if we expand each spin-orbital in a certain basis of functions (not necessarily orthogonal),
such that,
X
ψk (n) = Clk φl (n) , (9.72)
l
where N is the number of functions in the basis. Then, the Hartree-Fock equations can be
written as,
N
X N
X
F̂ ψk (1) = k ψk (1) =⇒ F̂ Clk φl (1) = k Clk φl (1) (9.73)
l l
N
X Z N
X Z
=⇒ Clk dr1 φ∗m (1)F̂ φl (1) = k Clk dr1 φ∗m (1)φl (1)
m l
N
X N
X
=⇒ Fml Clk = k Sml Clk .
m l
In matrix representation the previous expression is given by the Hartree-Fock-Roothaan equa-

tion,
FC = SC , (9.74)
where is the diagonal matrix containing the orbital energies.

9.4 The periodic system of elements

Completely filled principal layers n, ` are isotropic, ΨN (r) = Ψ(r). Electronic orbits with small n
perceive less shielding, as their radii are approximately r̄n ' r̄H,n /(Z − 2), i.e small. In contrast,
orbits with large n are shielded, as their radii are approximately r̄n ' naB . Following Bohr’s
model, energies in −1/r coulombian potentials are degenerate in `. In many-electron systems the
deviation of the real potential from the coulombian law caused by shielding, lifts the degeneracy
and considerably decreases energy for small `. For large `, the centrifugal term dominates over
the coulombian potential and the energy decrease is much smaller. For example, in the sodium
atom the 3d orbital is lower than the 4s. Along the periodic system of the elements, the orbitals
are consecutively filled with electrons according to these shifted energies.
Figure 9.4: Illustration of Hund’s rule.
It is important to distinguish three different energetic sequences: 1. Tab. 9.6 shows, for a
given atom, the excited orbitals of the last electron. 2. The energy sequence shown in Tab. 9.4
tells us in which orbital the next electron will be placed, when we go to the next atom in the
periodic table 9.10. 3. The inner electrons are subject to different potentials and, hence, follow
a different sequence energetic sequence: While for the inner electrons we find,
En,l < En,l+1 En+1,l , (9.75)
the sequence is partially inverted for the outermost electron. Note that it is the outermost
electrons that determine the chemical reactivity of the atom. The sequence is illustrated in
Fig. 9.8.
Following Hund’s rule, the L · S-coupling is energetically favorable compared to the j · j-
coupling, which means that the spins of the outermost electrons, that is, the electrons outside
of filled subshells (n, `), prefer to orient their spins in parallel in order to anti-symmetrize the
spatial wavefunctions and thus maximize the distance between the electrons. Every sub-layer of
9.4. THE PERIODIC SYSTEM OF ELEMENTS 227
the series shown in Fig. 9.8 must be filled in the listed order before placing new electrons in the
next layer.
Noble gases have small radii, high excitation energies and high ionization energies. The
valence electron must fill the gap to higher major quantum numbers. Halogens have strong
electro-affinities, since the outer electron layer (nmax ) is incomplete and therefore malleable, such
that an electron approaching the halogen perceives the nucleus through a partially transparent
shield. Alkalis are similar to hydrogen and have excitation energies in the optical regime. Their
fundamental state 2 S1/2 is determined by a valence electron only in the ` = 0 orbital. Unlike
hydrogen, excitation energies are highly dependent on `, since orbits with small ` have higher
probabilities to be in the unshielded region −Z 2 e2 /r than orbits with large `, who spend more
time in the shielded region −e2 /r. For the same reason, energies corresponding to larger n
resemble more those of the hydrogen spectrum.
The interior shell structure of the atoms can be analyzed by X-ray scattering. Electrons
decelerated by atoms emit a continuous spectrum called Bremsstrahlung, but they can also
expel electrons from the inner layers leaving a hole behind. When a hole is filled by cascades of
electrons falling down from higher layers, the atom emits a specific X-ray spectrum (≈ 104 eV).
The selection rules ∆` = ±1 and ∆j = ±1 split the lines in two components. X-ray spectra
of neighboring elements in the periodic system of elements are very similar, because the inner
layers not being shielded, they see a potential close to ∝ Z 2 /r. Therefore, the Z-dependency of
the rows in the periodic table is more or less ω ∝ Z 2 , as predicted by Bohr’s atomic model.
9.4.1 Electronic shell model

In order to be able to say something about the general structure of a many-electron system, we
consider its Hamiltonian,
Z
X p2i Ze2 1 X e2 1
− + , (9.76)
2m 4π0 ri 4π0 |ri − rj |
i i>j
consisting of kinetic energy, potential energy of the nucleus, and interaction between electrons
Ĥww . The index cites the electrons and runs from 1 to the nuclear charge number Z. We neglect
the electron-electron interaction Ĥww . Then, each electron perceives a hydrogen atom with the
solution α replaced by Zα, because the Coulomb potential of the nucleus is now,
Ze2 1 Zα~c
= . (9.77)
4π0 r r
For the energies we obtain then:
Z
X 2
me 2 Zα
E=− c . (9.78)
2 ni
i
Each of the energy levels in this sum contains several states. Each of these states can be
occupied by a single electron, according to the Pauli principle. In this way, we obtain the
electronic configuration for the atoms of the periodic system. In this picture, the energies of the
ground states of the elements, normalized by 1/Z 2 , can be arranged in the scheme 9.5:
This only works for atoms with up to 18 electrons. When the layer 3p is completely filled,
the next to be occupied is not 3d but the 4s. The new scheme is illustrated in 9.6.
The anomalies beginning at Z = 18 arise due to electron-electron interaction. To understand
this, we consider an electron in the mean field generated by all other electrons. The electron
N 4s 19
K 20
Ca 4p 4d 4f
M 3s 11
Na 12Mg 3p 13
Al 14
Si 15
P 16
S 17
Cl 18
Ar 3d
L 2s 3
Li 4
Be 2p 5
B 6
C 7
N 8
O 9
F 10
Ne
K 1s 1
H 2
He
Figure 9.5: Periodic order.
is then located in an effective potential which evolves between two Coulomb potentials, one for
large and the other for small distances: for small distances, the electron sees only the positive
point charge of the nucleus and the potential is proportional to Zα,
Ze2 1
V (r) = − . (9.79)
4π0 r
For large distances, the nucleus and all other electrons together form a nearly spherical charged
source called the ’core’, and we can approximate the potential by,
e2 1
V (r) = − . (9.80)
4π0 r
The real potential evolves from one to the other coulombian potential, as distance from the
nucleus is increased:
Near the nucleus, the electrons shield the positive charge less than for large r. Thus, those
states that have a high probability near the nucleus are energetically lowered. That is,
E2s < E2p and E6s < E6p < E6d . (9.81)
The degeneracy of the orbital angular momentum in the Schrödinger model is thus lifted. The
shielding is, as can be seen in the example of the excited states of lithium, a large effect in the
range of some eV.
The shielding also accounts for the anomalies in the periodic system, such as in K or Ca.
Since E4s < E3d , the 4s state is filled before the 3d. Similar anomalies also occur in Rb (5s), Cs
(6s), and Fr (7s). In rare earths the shielding effect is even more pronounced. Here, the energy
of the state 6s is even below the energy of the 4f , which means that the shells 6s, 5s, 5p, and
5d protect the 4f shell very well 9 .
9.4.2 Alkalines
The electronic shell structure of alkalines consists of a completed noble gas shell and an additional
valence electron. Their spectrum is therefore very similar to hydrogen. An empirical approach
can be stressed to describe this feature,
c2
En,l = − 12 µEG c2 , (9.82)
n − ∆(n, l)
9
An example of this is Nd:YAG (Neodymium in Yttrium Aluminum Garnet). In this crystal, optical transitions
can be excited within the 4f shell of the Nd. However, these transitions are only allowed due to perturbations of
the crystalline field. The very strong shielding ensures a long life of the excited state. For this reason this crystal
is an excellent laser material.
Li Na K Rb Cs H
S P D F S P D F S P D F S P D F S P D F
0
7
6
5
5 5 5 5 5
5 5 6 5 7 6 8 7
4 4 4 4 4 4 4
-1 4 6 4 7 4
4 5 5 8
6
5 6 7
3
3 3 4 3
3 6 7
5 4
-2
3 4
5
5 6
4
-3
3
2
2
-4 6
5
4
-5
3
2
Figure 9.6: Comparison of the excitation energies of the valence electron for several alkaline
atoms.
where µEG is the reduced mass relative to the noble gas hull and ∆(n, l) is called quantum defect.
The quantum defect is tabulated for most alkaline states and is particularly important for low
energy states. For sodium, for example, the values are:
L n=3 n=4 n=5 n=6

s 1.37 1.36 1.35 1.34
p 0.88 0.87 0.86 0.86
d 0.10 0.11 0.13 0.11
f - 0.00 -0.01 0.008
For states with a large angular momentum, the quantum defect disappears. In these states,
the electron is far from the nucleus and the potential is similar to that of hydrogen. Alkalines
are currently widely studied in quantum optics laboratories, for being comparatively simple,
but having a sufficiently rich structure to be interesting. The fundamental electronic transitions
typically lie in the visible and near-infrared spectral range and can be excited with compara-
tively simple laser sources. The lifetime of excited states is typically longer than 20 ns, which
corresponds natural linewidths of approximately (2π) 10 MHz.
−2
V(r) / (e2/aB)
−4
−6
−8
−10
0 1 2 3
r/aB
Figure 9.7: (Code: AM M ultielectron P otencialEf etivo.m) External potential (shielded

2
Coulombian) Vcl ∝ er (blue, upper curve), interior potential (non-shielded Coulombian)
2
Vbl ∝ Zer (green, lower curve), and effective potential (red, middle curve).
Q 7s 7p
P 6s 6p 6d
O 5s 5p 5d 5f actinides
N 4s 4p 4d 4f lanthanides (terras raras)
M 3s 3p 3d
L 2s 2p
elementos de transição
K 1s
l= 0 1 2 3
Figure 9.8: Illustration of the sequence of filling the orbitals with electrons.
9.4.3 LS and jj-coupling

In the case of helium, we have seen that the Pauli principle first determines the relative orien-
tation of the electron spins. The spins si of the individual electrons therefore add up to a total
angular momentum S. The orbital angular momenta li also adopt a relative orientation. It is
determined by a residual spherically non-symmetric Coulomb interaction: A certain combina-
tion L of orbital angular momenta leads to a certain spatial distribution of the electrons and
thus to a certain electrostatic energy distribution.
The total spin S and the total orbital angular momentum L subsequently couple to a total
angular momentum J very similar to the l-s spin-orbit coupling in single electron systems.
States with different J then have the respective energies that the total spin S adopts in the field
generated by the total orbital angular momentum L.
In addition, there are the small contributions due to:
• li · lj -coupling
• si · sj -coupling
• relativistic corrections
1
P
1
4
S=0 D (3/2,5/2) 1,2,3,4 3
1 2
F
1
3
(3/2,3/2) 0,1,2,3 2
3
l =1, l =2
1 2 P 2 1
0
3
P 3
P 1
energia de 3
3
D P 0
1partícula
S=0
simetria e
(1/2,5/2) 2,3 3
energia de troca 3 2
F
interação residual
de Coulomb estrutura fina
(1/2,3/2) 1,2 2
1
(j , j )
1 2 J
Figure 9.9: Hierarchy of energies.
• hyperfine structure
The above coupling scheme is called Russel-Saunders coupling or LS-coupling. It works when
the spin-orbit coupling is small. In this case, intercombination is forbidden, which means that
there can be no electromagnetic transition between states with different spins (see metastable
helium).
Since ELS ' Zα4 ' Z 4 , for heavy atoms, the coupling of an electronic spin to its own orbital
momentum grows strongly with Z, as well as the symmetrization and the exchange energy, which
mutually orient the spins, and the residual Coulomb interaction, which mutually couples the
angular orbital momenta. In this case, the orientation of Li relative to Si delivers more energy
than the exchange energy and the residual energy cost. Hence, the spin and the orbital angular
momentum of an individual electron couple first,
ji = li + si . (9.83)
We obtain a new Hamiltonian of fine structure of the form,
HF S ∝ ji · jj . (9.84)
Pure jj-coupling only exists for very heavy nuclei. Normally, we have a so-called inter-
mediate coupling, which is a mixture of LS and jj-coupling. This can considerable relax the
intercombination prohibition. When the coupling is pure, we have the following dipolar selection
rules:
LS-coupling: ∆S = 0, ∆L = ±1, ∆` = ±1
jj-coupling: ∆j = 0, ±1 for one e− , ∆j = 0 for all others
In addition we have for the two couplings: ∆J = 0, ±1, but J, J 0 = 0 is forbidden, ∆mJ =
0, ±1 when ∆J = 0 but mJ , mJ 0 = 0 is forbidden.
9.4.4 Summary of the atomic degrees of freedom

The total Hamiltonian of a single atom is composed of the kinetic energy of the nucleus and
the electrons, of various interaction potentials between the nucleus and the electrons, and of
interactions with various types of external electromagnetic fields.

N
~2 2 X ~2 2
Ĥ = − ∇R + − ∇ri + V (r1 , s1 , .., rN , sN ) + Vext . (9.85)
2m 2m
i=1
Of course, with the presence of other atoms, other interactions may generate other relevant
contributions to the Hamiltonian.
The following interactions contribute to the potential V : The Coulomb interactions,
Z
X Z
X
Ze2 e2
Vncl−ele = − and Vele−ele = , (9.86)
4π0 |r − ri | 4π0 |ri − rj |
i=1 i<j=1
the antisymmetry of the wavefunction, that is, exchange integrals,
Vsym , (9.87)
the energies of spin-orbit couplings,

Z
X 1 1 dVcl
Vls = − (li · si ) , (9.88)
e2m2 c2 |r − ri | dri
i=1
the energies of spin-spin couplings,

Z
X
e2 σi · σj [σi · (ri − rj )][σj · (ri − rj )]
Vss = −3 , (9.89)
m2 |ri − rj |3 (ri − rj )5
i6=j=1
the energies of orbit-orbit couplings,

Z
X
Vll = cij (li · lj ) , (9.90)
i6=j=1
interactions between the spin of the electrons and the nuclear spin and between the orbital
angular momentum of the electrons and the nuclear spin,
A
Vhf s = J·I , (9.91)
~2
and relativistic corrections.
Vrel , (9.92)
In addition, static external fields may displace energy levels and can influence the internal
coupling of angular momenta and spins,
Vext = −d · E , − µ
~ ·B . (9.93)
What quantum numbers are good depends on the relative amplitudes of intra-atomic inter-
actions:
r
Case 1: fine structure with L·S-coupling plus Zeeman splitting of hyperfine structure: Vncl−ele , Vele−ele
a
Vele−ele , Vsym Vls Vhf s VB the quantum number are |ni , li , L, S, J, F, mF i.
r
Case 2: fine structure with j·j-coupling plus Zeeman splitting of hyperfine structure: Vncl−ele , Vele−ele
a
Vls Vele−ele , Vsym Vhf s VB the quantum number are |ni , li , ji , J, F, mF i.
9.5. EXERCISES 233
r
Case 3: fine structure with L·S-coupling plus hyperfine structure of Zeeman splitting: Vncl−ele , Vele−ele
a
Vele−ele , Vsym Vls VB Vhf s the quantum number are |ni , li , L, S, J,mJ , mI i.
r
Case 4: fine structure with L·S-coupling plus Paschen-Back splitting of fine structure: Vncl−ele , Vele−ele
a
Vele−ele , Vsym VB Vls Vhf s the quantum number are |ni , li , L, S,mL , mS , mI i.
Vncl−ele splitting in n
coarse structure ↓
Vele−ele splitting in l
↓
Vsym splitting in S &
↓ Vl s splitting in ji
fine structure Vee splitting in L ↓
↓ Vsym , Vele−ele splitting in J
VLS splitting in J .
↓
hyperfine structure Vhf s splitting in F
↓
Zeeman effect VLS splitting in mF
9.5 Exercises
9.5.1 Symmetrization of bosons and fermions
9.5.1.1 Ex: Bosonic and fermionic isotopes
Consulting an isotope table determine the bosonic or fermionic character of the following atomic
species: 87 Sr, 86 Sr, 87 Rb, 39 K e 40 K.
9.5.2 Helium
9.5.2.1 Ex: Helium atom
Compare the measured binding energy with the prediction of Bohr’s model considering the
inter-electronic interaction up to first order TIPT.
9.5.2.2 Ex: Shielding in helium

The helium atom (or helium-like atoms such as H − ) has two interacting electrons in its compo-
sition, which means that these systems have no exact solution. To circumvent this problem we
have to come up with a series of approximate methods for calculating their eigenstates and their
respective eigenenergies. Among these methods, a widely used one, due mainly to its ease and
practicality, is the variational method, in which we calculate the fundamental state of a given
problem through a test function that is not a solution of the original problem. This method,
when applied to a helium atom, uses as test function the solution of the problem without coulom-
bian interaction between the electrons, which only feel the interaction with the original charge
of the nucleus. However, this method could be further improved if we considered an effective
nuclear charge, due to its interaction with the electrons themselves, and then obtaining the test
function. Apply this correction to the case of helium. Interpret the result. Help:
Z p D1E
sin θ2 r12 + r22 − 2r12 r22 cos θ2 Z
p dθ 2 = and = .
2 2 2 2
r1 + r2 − 2r1 r2 cos θ2 r1 r2 r aB
9.5.3 Electronic shell structure

9.5.4 The periodic system of elements
9.5.4.1 Ex: Electronic excitation levels of alkaline
Explain why
a. state [Li] (2s)2 SJ has lower energy than [H] (2s)2 SJ ;
b. state [Li] (2s)2 SJ has lower energy than [Li] (2p)2 PJ ;
c. state [Na] (3d)2 DJ has lower energy than [Na] (4p)2 PJ .
Physics Standard Reference
Group P E R I O D I C T A B L E Laboratory Data Program
IA physics.nist.gov www.nist.gov www.nist.gov/srd VIII
1
1 2S1/2
Atomic Properties of the Elements 2 S0
U.S. DEPARTMENT OF COMMERCE
H Frequently used fundamental physical constants Technology Administration He
1 Hydrogen For the most accurate values of these and other constants, visit physics.nist.gov/constants National Institute of Standards and Technology Helium
1.00794 1 second = 9 192 631 770 periods of radiation corresponding to the transition 4.00260
133 2
1s between the two hyperfine levels of the ground state of Cs 1s
13.5984 IIA !1 IIIB IVB VB VIB VIIB 24.5874
speed of light in vacuum c 299 792 458 m s (exact)
2 1 !34 2 3 4 3 2 1
3 S1/2 4 S0 Planck constant h 6.6261 H 10 Js (h = h/2p) 5 P°1/2 6 P0 7 S°3/2 8 P2 9 P°3/2 10 S0
!19
elementary charge e 1.6022 H 10 C
9.5. EXERCISES
Li Be !31 B C N O F Ne
2 electron mass me 9.1094 H 10 kg
Lithium Beryllium 2 Boron Carbon Nitrogen Oxygen Fluorine Neon
mec 0.5110 MeV
6.941 9.01218 !27 10.811 12.0107 14.00674 15.9994 18.99840 20.1797
2 2 2 proton mass mp 1.6726 H 10 kg 2 2 2 2 2 2 2 3 2 2 4 2 2 5 2 2 6
1s 2s 1s 2s 1s 2s 2p 1s 2s 2p 1s 2s 2p 1s 2s 2p 1s 2s 2p 1s 2s 2p
5.3917 9.3227 fine-structure constant a 1/137.036 8.2980 11.2603 14.5341 13.6181 17.4228 21.5646
!1
2 1 Rydberg constant R4 10 973 732 m 2 3 4 3 2 1
11 S1/2 12 S0 15 13 P°1/2 14 P0 15 S°3/2 16 P2 17 P°3/2 18 S0
R4c 3.289 84 H 10 Hz
Na Mg R4hc 13.6057 eV Al Si P S Cl Ar
!23 !1
3 Sodium Magnesium Boltzmann constant k 1.3807 H 10 JK Aluminum Silicon Phosphorus Sulfur Chlorine Argon
22.98977 24.3050 26.98154 28.0855 30.97376 32.066 35.4527 39.948
2 2 2 2 2 3 2 4 2 5 2 6
[Ne]3s [Ne]3s VIIIA [Ne]3s 3p [Ne]3s 3p [Ne]3s 3p [Ne]3s 3p [Ne]3s 3p [Ne]3s 3p
5.1391 7.6462 IIIA IVA VA VIA VIIA IB IIB 5.9858 8.1517 10.4867 10.3600 12.9676 15.7596
2 1 5 4 3 2 2 3 4 3 2 1
19 S1/2 20 S0 21 2D3/2 22 3F2 23 4F3/2 24 7S3 25 6S5/2 26 D4 27 F9/2 28 F4 29 S1/2 30 1S0 31 P°1/2 32 P0 33 S°3/2 34 P2 35 P°3/2 36 S0
4
K Ca Sc Ti V Cr Mn Fe Co Ni Cu Zn Ga Ge As Se Br Kr
Potassium Calcium Scandium Titanium Vanadium Chromium Manganese Iron Cobalt Nickel Copper Zinc Gallium Germanium Arsenic Selenium Bromine Krypton
39.0983 40.078 44.95591 47.867 50.9415 51.9961 54.93805 55.845 58.93320 58.6934 63.546 65.39 69.723 72.61 74.92160 78.96 79.904 83.80
Period
2 2 2 2 3 2 5 5 2 6 2 7 2 8 2 10 10 2 10 2 10 2 2 10 2 3 10 2 4 10 2 5 10 2 6
[Ar]4s [Ar]4s [Ar]3d4s [Ar]3d 4s [Ar]3d 4s [Ar]3d 4s [Ar]3d 4s [Ar]3d 4s [Ar]3d 4s [Ar]3d 4s [Ar]3d 4s [Ar]3d 4s [Ar]3d 4s 4p [Ar]3d 4s 4p [Ar]3d 4s 4p [Ar]3d 4s 4p [Ar]3d 4s 4p [Ar]3d 4s 4p
4.3407 6.1132 6.5615 6.8281 6.7462 6.7665 7.4340 7.9024 7.8810 7.6398 7.7264 9.3942 5.9993 7.8994 9.7886 9.7524 11.8138 13.9996
2 1 2 3 6 7 6 5 4 1 2 1 2 3 4 3 2 1
37 S1/2 38 S0 39 D3/2 40 F2 41 D1/2 42 S3 43 S5/2 44 F5 45 F9/2 46 S0 47 S1/2 48 S0 49 P°1/2 50 P0 51 S°3/2 52 P2 53 P°3/2 54 S0
5
Rb Sr Y Zr Nb Mo Tc Ru Rh Pd Ag Cd In Sn Sb Te I Xe
Rubidium Strontium Yttrium Zirconium Niobium Molybdenum Technetium Ruthenium Rhodium Palladium Silver Cadmium Indium Tin Antimony Tellurium Iodine Xenon
85.4678 87.62 88.90585 91.224 92.90638 95.94 (98) 101.07 102.90550 106.42 107.8682 112.411 114.818 118.710 121.760 127.60 126.90447 131.29
2 2 2 2 4 5 5 2 7 8 10 10 10 2 10 2 10 2 2 10 2 3 10 2 4 10 2 5 10 2 6
[Kr]5s [Kr]5s [Kr]4d5s [Kr]4d 5s [Kr]4d 5s [Kr]4d 5s [Kr]4d 5s [Kr]4d 5s [Kr]4d 5s [Kr]4d [Kr]4d 5s [Kr]4d 5s [Kr]4d 5s 5p [Kr]4d 5s 5p [Kr]4d 5s 5p [Kr]4d 5s 5p [Kr]4d 5s 5p [Kr]4d 5s 5p
4.1771 5.6949 6.2171 6.6339 6.7589 7.0924 7.28 7.3605 7.4589 8.3369 7.5762 8.9938 5.7864 7.3439 8.6084 9.0096 10.4513 12.1298
2 1 3 4 5 6 5 4 3 2 1 2 3 4 3 2 1
55 S1/2 56 S0 72 F2 73 F3/2 74 D0 75 S5/2 76 D4 77 F9/2 78 D3 79 S1/2 80 S0 81 P°1/2 82 P0 83 S°3/2 84 P2 85 P°3/2 86 S0
6
Cs Ba Hf Ta W Re Os Ir Pt Au Hg Tl Pb Bi Po At Rn
Cesium Barium Hafnium Tantalum Tungsten Rhenium Osmium Iridium Platinum Gold Mercury Thallium Lead Bismuth Polonium Astatine Radon
132.90545 137.327 178.49 180.9479 183.84 186.207 190.23 192.217 195.078 196.96655 200.59 204.3833 207.2 208.98038 (209) (210) (222)
Figure 9.10: Periodic table.

2 14 2 2 14 3 2 14 4 2 14 5 2 14 6 2 14 7 2 14 9 14 10 14 10 2 2 3 4 5 6
[Xe]6s [Xe]6s [Xe]4f 5d 6s [Xe]4f 5d 6s [Xe]4f 5d 6s [Xe]4f 5d 6s [Xe]4f 5d 6s [Xe]4f 5d 6s [Xe]4f 5d 6s [Xe]4f 5d 6s [Xe]4f 5d 6s [Hg]6p [Hg]6p [Hg]6p [Hg]6p [Hg]6p [Hg]6p
3.8939 5.2117 6.8251 7.5496 7.8640 7.8335 8.4382 8.9670 8.9587 9.2255 10.4375 6.1082 7.4167 7.2856 8.417 ? 10.7485
2 1 3
87 S1/2 88 S0 104 F2 ? 105 106 107 108 109 110 111 112
Solids
7
Fr Ra Rf Db Sg Bh Hs Mt Uun Uuu Uub Liquids For a description of
Francium Radium Rutherfordium Dubnium Seaborgium Bohrium Hassium Meitnerium Ununnilium Unununium Ununbium
(223) (226) (261) (262) (263) (264) (265) (268) (269) (272) Gases the atomic data, visit
2 14 2 2
[Rn]7s [Rn]7s [Rn]5f 6d 7s ? physics.nist.gov/atomic
4.0727 5.2784 6.0 ?
Artificially Prepared
Atomic Ground-state 2 1 4- 5- 6 7 8 9 6 5- 4- 3 2 1 2
Number Level
57 D3/2 58 G°4 59 -I °9/2 60 -I 4 61 H°5/2 62 F0 63 S°7/2 64 D°2 65 H°15/2 66 -I 8 67 -I °15/2 68 H6 69 F°7/2 70 S0 71 D3/2
1
La Ce Pr Nd Pm Sm Eu Gd Tb Dy Ho Er Tm Yb Lu
Symbol 58 G°4 Lanthanum Cerium Praseodymium Neodymium Promethium Samarium Europium Gadolinium Terbium Dysprosium Holmium Erbium Thulium Ytterbium Lutetium
138.9055 140.116 140.90765 144.24 (145) 150.36 151.964 157.25 158.92534 162.50 164.93032 167.26 168.93421 173.04 174.967
2 2 3 2 4 2 5 2 6 2 7 2 7 2 9 2 10 2 11 2 12 2 13 2 14 2 14 2
Ce [Xe]5d6s [Xe]4f5d6s [Xe]4f 6s [Xe]4f 6s [Xe]4f 6s [Xe]4f 6s [Xe]4f 6s [Xe]4f 5d6s [Xe]4f 6s [Xe]4f 6s [Xe]4f 6s [Xe]4f 6s [Xe]4f 6s [Xe]4f 6s [Xe]4f 5d6s
Name 5.5769 5.5387 5.473 5.5250 5.582 5.6436 5.6704 6.1501 5.8638 5.9389 6.0215 6.1077 6.1843 6.2542 5.4259
Cerium 2 3 4 5 6 7 8 9 6 5- 4- 3 2 1 2
140.116 89 D3/2 90 F2 91 K11/2 92 L°6 93 L11/2 94 F0 95 S°7/2 96 D°2 97 H°15/2 98 -I 8 99 -I °15/2 100 H6 101 F°7/2 102 S0 103 P°1/2?
Atomic
†
Weight [Xe]4f5d6s 2 Ac Th Pa U Np Pu Am Cm Bk Cf Es Fm Md No Lr
5.5387 Actinium Thorium Protactinium Uranium Neptunium Plutonium Americium Curium Berkelium Californium Einsteinium Fermium Mendelevium Nobelium Lawrencium
(227) 232.0381 231.03588 238.0289 (237) (244) (243) (247) (247) (251) (252) (257) (258) (259) (262)
Ground-state Ionization 2 2 2 2 2 3 2 4 2 6 2 7 2 7 2 9 2 10 2 11 2 12 2 13 2 14 2 14 2
[Rn]6d7s [Rn]6d 7s [Rn]5f 6d7s [Rn]5f 6d7s [Rn]5f 6d7s [Rn]5f 7s [Rn]5f 7s [Rn]5f 6d7s [Rn]5f 7s [Rn]5f 7s [Rn]5f 7s [Rn]5f 7s [Rn]5f 7s [Rn]5f 7s [Rn]5f 7s 7p?
Configuration Energy (eV) 5.17 6.3067 5.89 6.1941 6.2657 6.0262 5.9738 5.9915 6.1979 6.2817 6.42 6.50 6.58 6.65 4.9 ?
† 12
235
Based upon C. ( ) indicates the mass number of the most stable isotope. For a description and the most accurate values and uncertainties, see J. Phys. Chem. Ref. Data, 26 (5), 1239 (1997). March 1999
Part III
Quantum Optics
237
Chapter 10
The Bloch equations

So far we mainly considered low amplitude light fields with broadband emission and incoherent
phases that interact weakly with an atom or a dilute gas. The expression (5.84) allowed us to
calculate the probability of finding a two-level atom exposed to a light field in the excited state.
But these expressions were obtained by averaging over the spectral width of the line, ignoring
any phase relationship between the incident field and the excited dipole, and basically assuming
a negligible depopulation of the ground state. For the first half of the 20-th century these
assumptions matched the available light sources, usually incandescent, arc or plasma discharge
lamps. After the invention of the laser in 1958, single mode and pulsed lasers quickly replaced
the lamps as a source for optical excitation. These new light sources initiated a revolution in
optical science, the consequence of which continue to reverberate through modern sciences and
applied technologies.
The characteristics of laser sources are far superior to the old lamps in all respects. They are
intense, collimated, spectrally narrow and phase coherent. The laser gave rise to a multitude
of new spectroscopic techniques and new disciplines, such as quantum electronics, the study
of statistical properties of light in quantum optics, optical cooling and trapping of microscopic
particles, the control of chemical reactivity, and new technologies for imaging and high resolution
microscopy.
To tackle the problem from the base we need first to examine what happens when our two-
level atom interacts with light sources, which are spectrally narrow in comparison to the natural
width of optical transitions, and what happens when the states have a well-defined polarization
and phase and the intensities are sufficient to significantly depopulate the ground state. We
will look out for an equation that describes the temporal evolution of atoms interacting strongly
with a single mode of the radiation field.
A first attempt to solve this problem could be to use the Schrödinger equation, which de-
scribes the temporal evolution of the state of any system. In fact, as long as we are only interested
in stimulated processes, such as the absorption of a monochromatic wave, the Schrödinger equa-
tion would suffice. A problem arises when we want to describe relaxation processes at the same
time as excitation processes, after all in most realistic situations the atom reaches a steady state,
where the rate of excitation and relaxation are equal. Spontaneous emission (and any other dis-
sipative process) must therefore be included in the physical description of the temporal evolution
of our light-atom system. In this case, however, our system is no longer restricted to a single
mode of the light field and the two atomic states of excitation. Spontaneous emission populates
a statistical distribution of states of the light field and leaves the atom in a superposition of
many momentum states. This situation can not be described by a single wavefunction, but only
by a distribution of wavefunctions, and we can only expect to calculate the probability of finding
the system within this distribution. The Schrödinger equation, therefore, no longer applies, and
we need to trace the time evolution of a system characterized by a density operator describing
a statistical mixture of quantum states. The equations which describe the time evolution of the
239
240 CHAPTER 10. THE BLOCH EQUATIONS
matrix elements of this density operator are the optical Bloch equations, and we must use them
instead of the Schrödinger equation. In order to appreciate the origin and the physical content
of the optical Bloch equations we begin by reviewing the rudiments of the density matrix theory.
10.1 Density matrix

10.1.1 The density operator
We define the statistical operator or density operator 1 ,
X
ρ̂ ≡ pk P̂k where P̂k ≡ |ψk ihψk | , (10.1)
k
where |ψk i is a complete set of orthonormal states of the system under study. We have a
statistical distribution
P of these states with pk being the probability of finding |ψk i in the set.
Obviously, k pk = 1. That is, the density operator acts on a member of the set |ψk i such as
to extract the probability of finding the system in |ψk i,
X
ρ̂|ψm i = pk |ψk ihψk |ψm i = pm |ψm i . (10.2)
k
If all members of the set are in the same state, for example |ψk i, the density operator reduces
to,
ρ̂ = |ψk ihψk | , (10.3)
and the system is in a pure state with P̂k = 1. Each time a quantum state can be expressed
by a single wave function, it is a pure state, but it does not have to be an eigenstate. Starting
from the equation (10.2) we find the diagonal elements of the density matrix, which are the
probabilities of finding the system in |ψm i,
hψm |ρ̂|ψm i = pm . (10.4)
Assuming that all |ψk i are orthonormal, the non-diagonal elements are necessarily zero. Besides
that, X
hψk |ρ̂|ψk i = 1 . (10.5)
k
So, ρ̂ contains all information about the system. When a state is unknown, ρ̂ describes the
probability of finding the system in each state. When the state is fully known, ρ̂ describes
a pure state, that is, a vector in the Hilbert space, which is unequivocally determined by a
complete set of observables with their respective quantum numbers.
1
In the presence of degeneracy or a continuous spectrum we can generalize the definition:
X Z X Z
ρ̂ ≡ pk P̂k + pλ P̂λ dλ where P̂k ≡ |kmihkm| and P̂λ ≡ |λµihλµ|dµ .
k m
Here, m and µ are degenerate quantum numbers, m, n are discrete, and λ, µ are continuous quantum numbers.
The set of quantum numbers is complete, when
X Z
|kmihkm| = 1̂ = |λµihλµ|dλdµ .
k,m
P
The degree of degeneracy of a state |ki is Tr P̂k = m 1. The probability of finding the system in the state |ki
P
is hP̂k i = pn m 1.
10.1. DENSITY MATRIX 241
The properties of the density operator are,
ρ̂ = ρ̂†
hρ̂i ≥ 0
Tr ρ̂ = 1
. (10.6)
Tr ρ̂2 ≤ 1
det ρ̂ = 0
ρ̂ = ρ̂2 for a pure state
10.1.1.1 Entropy
In a very general sense, the entropy determines in what direction a reversible process will take
place. It is related to the size of the available phase space on both sides of the reaction. For
example, the coupling of discrete and continuous modes is governed by entropy considerations.
Entropy measures of the lack of information about a system from which we only know hĤi,
S ≡ −kB hln ρ̂i = −kB Tr (ρ̂ ln ρ̂) . (10.7)
The information entropy (or von Neumann entropy) of statistically independent systems ρ̂ ≡
ρ̂1 ⊗ ρ̂2 is additive S = S1 + S2 . We can also define absolute temperatures by T −1 ≡ ∂S/∂hĤi.
The entropy of a pure state is 0. Hamiltonian processes conserve entropy, for they correspond to
non-dissipative unitary transformations. On the other side, relaxation increases the entropy and
the phase space volume. Another common definition is the so-called purity or Renyi entropy,
SR ≡ h1 − ρ̂i = 1 − Tr (ρ̂2 ) . (10.8)
Quantum states can exhibit coherences. For example, if we express a state |ψi on a basis of
eigenstates |1i and |2i:

|hψ|1i|2 h1|ψihψ|2i
ρ̂ = |ψihψ| = . (10.9)
h2|ψihψ|1i |hψ|2i|2
The evolution of such a state is described by the von Neumann equation,
i~∂t ρ̂(t) = [Ĥ, ρ̂(t)] . (10.10)
The measurement process is not described by this equation. A pure state will always remain
pure. If the eigenstates do not interact, the density operator will remain diagonal. The von
Neumann equation conserves the properties of hermiticity, ρ̂ = ρ̂† , completeness, Tr ρ̂ = 1, and
purity det ρ̂ = 0.
The density operator for a statistical mixture in a canonical ensemble (where S is maximum,
U is variable, and N is fixed) follows from a variational problem with the Lagrange parameters
δ(S − kB αh1̂i − kB βhĤi) = 0, since Tr ρ̂ and hĤi are fixed by boundary conditions. We find,
1 −Ĥ/kB T
ρ̂ = Ze with Z ≡ Tr e−Ĥ/kB T . (10.11)
We also have the expectation values, hHi = −∂ ln Z/∂(1/kB T ) and (∆H)2 = −∂hĤi/∂(1/kB T ).
All quantities are fixed, except the kinetic energy, which balances the interaction with a heat
bath. T is the only equilibrium parameter. The density operator satisfies a Boltzmann distri-
bution 2 ,
Z
p2 ∂ 2 kB
U = hĤi = =− ln e−p /2mkB T dp = T . (10.12)
2m ∂(1/kB T ) 2
10.1.2 Matrix formalism

The next step is to develop matrix representations of the density operator by expanding the
state vectors |ψk i in a complete orthonormal basis,
X X
|ψk i = cnk |ni = |nihn|ψk i , (10.13)
n n
P
using the completeness relation (2.68), that is, n |nihn| = I, and defining,
cnk ≡ hn|ψk i (10.14)
as the projection of the state vector |ψk i on the basis vector |ni. Now, we can write the density
operator matrix representation within the basis {|ni} using the definition of ρ̂ in Eq. (10.1) and
replacing the expansions of |ψk i and hψk | of Eq. (10.13):
X X X X X
ρ̂ = pk |ψk ihψk | = pk |nihn|ψk ihψk |mihm| = pk cnk c∗mk |nihm| . (10.15)
k k m,n k m,n
The matrix elements of ρ in this representation are

X
ρnm ≡ hn|ρ̂|mi = pk cnk c∗mk (10.16)
k
P
with the diagonal elements hn|ρ̂|ni = k pk |cnk |2 e,
X X
ρ∗nm = hn|ρ̂|mi∗ = pk c∗nk cmk = pk hm|ψk ihψk |ni = hm|ρ|ni = ρmn , (10.17)
k k
which means that the operator ρ̂ is hermitian.
Example 38 (Density operator for a single atom): For a very simple system such as
a single atom with several levels, that without spontaneous emission can be described by a
single wavefunction |ψ1 i, we can let pk = δ1k . That is, the equations (10.15) and (10.17)
reduce to,
X
ρ̂ = cn1 c∗m1 |nihm| and hn|ρ|mi = cn1 c∗m1 . (10.18)
m,n
2
The von Neumann entropy S of a mixture can be expressed in terms of the eigenvalues or in terms of the
trace and theP
logarithm of the density operator ρ̂. Since ρ̂ is a semi-definite positivePoperator, its spectrum λi ,
given by ρ = iP λi |ϕi ihϕi | where {|ϕi i} is an orthonormal basis, satisfies λi > 0 and λi = 1. Then the entropy
becomes S = − i λi ln λi = −Tr (ρ ln ρ).
10.1.2.1 Measurement and trace

The sum of the diagonal elements of a matrix representing an operator is called the trace. This
quantity represents a fundamental property of the density operator, since it is invariant with
respect to any unitary transformation:
X
Tr ρ̂ ≡ hn|ρ|ni . (10.19)
n
With the definition of the density operator (10.1) we can write the Eq. (10.19) as,
X
Tr ρ̂ ≡ pk hn|ψk ihψk |ni . (10.20)
n,k
Now, using the completeness relation,

X X
Tr ρ̂ ≡ pk hψk |nihn|ψk i = pk hψk |ψk i = 1 , (10.21)
n,k k
which shows that the trace of the density operator representation is always 1 regardless of the
basis of the matrix representation.
Expectation values of observables are expressed by,
X
hÂi = pk hψk |Â|ψk i . (10.22)
k
On the other side, X

ρ̂Â = pk |ψk ihψk |Â , (10.23)
k
and in the basis {|ni},

X X X
hn|ρ̂Â|mi = hn| pk |ψk ihψk |Â|mi = pk hn|ψk ihψk |Â|mi = pk hψk |Â|mihn|ψk i . (10.24)
k k k
Now, along the diagonal, we have,

X
hn|ρ̂Â|ni = pk hψk |nihn|Â|ψk i . (10.25)
k
With the closing relation in the basis {|ni}, we now have 3 ,

X
Tr ρ̂Â = pk hψk |Â|ψk i = hÂi . (10.26)
k
The Eq. (10.26) says that the ensemble average of any dynamic observable Â can be calculated
from the diagonal elements of the operator matrix ρ̂Â: Since the trace is independent of the basis
(this will be shown in Exc. 10.6.1.1), each unitary transformation taking the matrix representa-
tion from a basis {|ni} to another one {|ti} leaves the trace invariant. Using the definition of a
3
In the presence of degeneracy or a continuous part of the spectrum we can generalize the definition of the
expectation, X
hX̂i ≡ Tr ρ̂X̂ = hkm|ρ̂X̂|kmi .
k,m
unitary transformation we can easily show that the trace of a cyclic permutation of a product
is invariant. For example,
Tr [ÂB̂ Ĉ] = Tr [Ĉ ÂB̂] = Tr [B̂ ÂĈ] , (10.27)
and in particular
Tr [ρ̂Â] = Tr [Âρ̂] = hÂi . (10.28)
In the Excs. 10.6.1.2 and 10.6.1.3 we apply the density operator to pure and mixed states of a
two-level system. In Excs. 10.6.1.4 and 10.6.1.5 we study thermal mixtures.
10.1.2.2 Measurement process

P P
If an observable Â has a spectral representation Â = n an |an ihan | = n an Pn , with P̂n =
|an ihan |, the measurement process will transform the density operator to,
X
ρ̂0 = P̂n ρ̂Pn . (10.29)
n
That is, after the measurement, the operator becomes diagonal on the basis of the eigenvalues
of Â 4 .
We note that the density operator (10.29) describes the whole ensemble after the measure-
ment. The sub-ensemble corresponding to a particular result an of the measurement is described
by a different density operator,
P̂n ρ̂P̂n
ρ̂0n = . (10.30)
Tr [ρ̂P̂n ]
This is true, when |an i is the only eigenvector with the eigenvalue an . If not, P̂n in the expression
(10.30) should be replaced by the projection operator onto the sub-space of an 5
10.1.2.3 Systems and subsystems

Density operators are very useful for playing with systems and subsystems. Let us, for instance,
assume that we have two quantum systems defined on the Hilbert spaces H1 and H2 . The
composite system is then the tensor product H1 ⊗ H2 . We now suppose that the compound
system is in a pure state, |ψi ∈ H1 ⊗H2 . If the state can be written in the form |ψi = |ψ1 i⊗|ψ2 i,
this means that the state of the first subsystem is |ψ1 i. However, in general, |ψi does not
decompose like this. Of course, every vector in H1 ⊗ H2 is a linear combination of tensorial
products of H1 and H2 . If |ψi can not be decomposed as a tensor product, we say that the
two systems are entangled. In this case, there is no reasonable way of associating a pure state
4
A projective measure always increases entropy. The entropy of a pure state is zero, while that of a mixture
is always greater than zero. Therefore, a pure state can be converted into a mixture by a measurement, but
the reverse can not happen. Thus, the action of measuring induces an irreversible change in the density matrix
reminiscent of the collapse of the wavefunction. Strangely, the measurement reduces the amount of information
by quenching the quantum interference of the compound system in a process called quantum decoherence. A
subsystem can be taken from a mixed state to a pure state only at the price of increasing the von Neumann
entropy elsewhere in the global system.
5
In general, assuming that f is a function associating each observable Â with a number f (Â) (which we may
imagine as the expectation value), we can state the following: If f satisfies some natural properties (such as
the one to produce positive values for positive operators), then there exists a unique density matrix ρ̂, such that
f (Â) = Tr (ρ̂Â) for all Â. That is, every reasonable ’family’ of expectation values’ can be represented by a density
matrix, which suggests that the density matrix provides the most general description of a quantum state.
|ψ1 i ∈ H1 to the state ψ ∈ H2 ⊗ H2 . If, for example, in the case of a two particle wavefunction
Ψ(x1 , x2 ) there is no way to construct a wavefunction (i.e. a pure state) ψ1 (x1 ) describing the
state of the first particle, then |Ψ(x1 , x2 )i =
6 ψ1 (x1 )ψ2 (x2 ).
The point of the discussion is that, even if the total system is in a pure state, the various
subsystems that compose it will normally be in mixed states. On the other hand, regardless of
whether the composite system is in a pure or mixed state, we can perfectly construct a density
matrix that describes the state. Therefore, the use of density matrices is inevitable. Let ρ̂ be
the density matrix of the system composed of two subsystems. Then the state in H2 is described
by a reduced density operator given by the partial trace of ρ̂ over H2 . In the particular case,
where the state the density matrix has the form ρ̂ = ρ̂1 ⊗ ρ̂2 , where ρ̂1 and ρ̂2 are the density
matrices in H1 and H2 , then the partial trace is simply, Tr H2 ρ̂ρ̂1 .
10.1.3 Temporal evolution of the density operator

As shown in Secs. 2.4.2 to 2.4.4, the equations governing the temporal evolution of a quantum
system depend on the choice of the picture, i.e. Schrödinger’s (2.123), Heisenberg’s (2.129), or
the interaction picture (2.136). This, of course, also applies to a system represented by a density
matrix.
Returning to the density operator definition (10.1), we can express its temporal dependence
in terms of time-dependent quantum states and of the time evolution operator (2.126),
X X
ρ̂(t) = pk |ψk (t)ihψk (t)| = pk U (t, t0 )|ψk (t0 )ihψk (t0 )|U † (t, t0 ) . (10.31)
k k
Writing, X
ρ̂(t0 ) = pk |ψk (t0 )ihψk (t0 )| , (10.32)
k
we see immediately,
ρ̂(t) = U (t, t0 )ρ̂(t0 )U † (t, t0 ) , (10.33)
where, for the common case of a time-independent Hamiltonian,
U (t, t0 ) = e−iĤ(t−t0 )/~ . (10.34)
Now we find the time derivative of the density operator differentiating the two sides of (10.33)
and substituting the Eqs.
dU 1 dU † 1
= ĤU and = − U † Ĥ (10.35)
dt i~ dt i~
for the time derivatives U and U † . The result is
dρ̂(t) i
= [ρ̂(t), Ĥ] . (10.36)
dt ~
The commutator itself can be considered as a superoperator acting, not any more on states but
on operators, that is, we can write,
i
Lρ̂(t) ≡ [ρ̂(t), Ĥ] , (10.37)
~
where L is called Liouville operator. The equation (10.36) is called Liouville equation or von
Neumann equation. The Liouville equation describes the time evolution of the density operator
which, in turn, describes the distribution of an ensemble of quantum states. Even though the
form of the Liouville equation resembles the Heisenberg equation, Eq. (10.31) shows that ρ(t) is
in the Schrödinger picture.
For a two-level system perturbatively interacting with a light field, the Hamiltonian can be
decomposed as in (2.133) into a stationary part and a time-dependent part,
Ĥ = Ĥa + V̂ (t) = Ĥa − d̂ · E0 cos ωt , (10.38)
where Ĥa is the part of the Hamiltonian describing the atomic structure and V̂ (t) the interaction
of the dipole transition with the classical oscillating electric field. The interaction picture is the
natural choice for this type of problem. In this case, we can transform the density operator into
the interaction picture defined by (2.134),
ρ̃(t) = eiĤ0 (t−t0 )/~ ρ̂(t0 )e−iĤ0 (t−t0 )/~ , (10.39)
where the ’tilde’ decoration (replacing the ’hat’) emphasizes, that we are now in the interaction
picture. We look for the time evolution rate of ρ̃(t) analogously to the Liouville equation.
Calculating the time derivatives on both sides of (10.39) and substituting Eq. (10.36) for dρ dt
results in,
dρ̃(t) i
= [ρ̃(t), V̂ (t)] . (10.40)
dt ~
This equation shows that the time evolution of the density operator in the interaction picture
depends only on the time-dependent part of the total Hamiltonian.
10.2 Bloch equations for two-level atoms

In this chapter we will begin to apply the ideas and tools developed in the previous sections.
Let us first make use of the density matrix to describe a two-level atom coupled to a single-
mode light field without spontaneous emission. We will then introduce the atomic Bloch vector
as a convenient and suggestive method to describe the time evolution of a coupled two-level
atom. We will introduce spontaneous emission and the important concepts of polarization and
susceptibility emanating from an excited sample of oscillating dipoles. Optical Bloch equations
including spontaneous emission will be given and their stationary solutions will be discussed.
Dissipative processes always broaden transition lines, and thus we will discuss various broadening
mechanisms.
10.2.1 The matrix elements of the density operator

Since the optical Bloch equations are coupled differential equations relating the elements of the
density operator matrix, we must examine the temporal dependence of these matrix elements,
based on our knowledge of the operator’s properties. We begin with the Liouville equation
(10.36) and evaluate the elements of the matrix,
dρ̂(t)
hm| |ni = ~i hm|[ρ̂(t), Ĥ]|ni = ~i hm|[ρ̂(t), ĤA + V̂ (t)]|ni (10.41)
dt
= ~i (En − Em )hm|ρ̂(t)|ni + ~i hm|[ρ̂(t), V̂ (t)]|ni ,
10.2. BLOCH EQUATIONS FOR TWO-LEVEL ATOMS 247
where |mi and |ni are members of a complete set of vectors of a basis {|ki} which P are also
eigen-kets of ĤA and span the space of Ĥ. Now, we insert the closing expression k |kihk| in
the commutator on the right-hand side of Eq. (10.41):
X
hm|[ρ̂(t), V̂ (t)]|ni = [hm|ρ̂(t)|kihk|V̂ |ni − hm|V̂ |kihk|ρ̂(t)|ni] . (10.42)
k
For our two-level atom the complete set only includes two states: |1(t)i = |1i and |2(t)i =
e−iω0 t |2i.
In addition, the matrix elements of the dipole coupling operator V̂ are only non-
diagonal,
V ≡ h1|V̂ |2i = h2|V̂ |1i . (10.43)
Hence, Eq. (10.41) adopts the form,
dρ̂11 i
= ~ [ρ̂12 V − ρ̂21 V ]
dt
dρ̂22 i dρ̂11
= ~ [ρ̂21 V − ρ̂12 V ] = −
dt dt
, (10.44)
dρ̂12 i
= iω0 ρ̂12 + ~ [V (ρ̂11 − ρ̂22 )]
dt
dρ̂21 dρ̂∗12
= −iω0 ρ̂21 + ~i [V (ρ̂22 − ρ̂11 )] =
dt dt
remembering that the dash of the diagonal terms, called populations, must be unitary, and that
the non-diagonal terms, called coherences, must be complex,
ρ̂11 + ρ̂22 = 1 , ρ̂12 = ρ̂∗12 . (10.45)
The above set of equations constitutes the optical Bloch equations in the Schrödinger picture.
It does not include loss terms due to spontaneous emission. We transform the Bloch equations
to the interaction picture by replacing the Liouville equation (10.36) by (10.40), and calculating
the matrix elements. We obtain,
dρ̃22 i dρ̃12 i
= V (ρ̃21 − ρ̃12 ) and = V (ρ̃11 − ρ̃22 ) . (10.46)
dt ~ dt ~
We would also have obtained this expression via the substitution, ρ̂12 = ρ̃12 eiω0 t . The interaction
picture simplifies the expressions for the temporal dependence of the coherences by eliminating
the first term on the right-hand side. Transforming to the interaction picture removes the
temporal dependence of the basis vectors spanning the Hilbert space of the two-level atom.
We have derived the optical Bloch equations from the Liouville equation, which is the fun-
damental equation of motion of the density operator, and we have seen how a unitary transfor-
mation can be used to represent these equations in the Schrödinger, Heisenberg or interaction
picture. So far, the Bloch equations do not include the possibility of spontaneous emission. We
will learn later, how to include this phenomenon.
10.2.2 Rotating wave approximation

In the following, we will only consider exponentials rotating with the frequency ∆ ≡ ω −
ω0 , and we will neglect terms rotating like ∆ ≡ ω + ω0 . This approximation, called rotating
wave approximation (RWA) is good, when the Rabi frequency is sufficiently small, Ω ω.
Otherwise, we observe an energy correction of the levels called Bloch-Siegert shift. The RWA
can be implemented in the time dependence of the coupling operator,
V (t) = ~Ω cos ωt → ~2 Ωe−iωt , (10.47)
neglecting the part 12 ~Ωeiωt .

Once the RWA made, we can transform to the rotating system by the prescription,
ρ12 ≡ ρ̂12 e−iωt , ρ22 ≡ ρ̂22 , (10.48)
which, applied to the Bloch equations in the Schrödinger picture Eq. (10.44), yields,
dρ22 iΩ dρ12 iΩ
= (ρ21 − ρ12 ) , = −i∆ρ12 + (ρ11 − ρ22 ) . (10.49)
dt 2 dt 2
In Exc. 10.6.2.1 we derive the Bloch equations from the equations of motion for the population
amplitudes a1 and a2 .
For arbitrary starting conditions, the solution of these equations is not simple. To solve the
problem we write the equations in a matrix form,
   i 
ρ11 0 0 2Ω − 2i Ω
ρ22   0 0 − 2i Ω 2i Ω 
~≡
ρ 
ρ12  , A≡
 Ω − Ω −i∆
 , ~˙ = A~
ρ ρ . (10.50)
i
2 2
i
0 
ρ21 − 2i Ω 2i Ω 0 i∆
To solve this system of differential equations, we calculate the eigenvalues of the matrix,
det(A − λ) = λ2 (∆2 + Ω2 ) + λ4 = 0 (10.51)

λ = 0, ±iG ,
√
with the generalized Rabi frequency G ≡ ∆2 + Ω2 . Therefore, the general solution is,
(1) (2) (3)
ρ22 (t) = ρ22 + ρ22 eiGt + ρ22 e−iGt (10.52)
(1) (2) (3)
ρ12 (t) = ρ12 + ρ12 eiGt + ρ12 e−iGt .
The coefficients follow from the Bloch equations with particular starting conditions. With a
little algebra we get
(1) 1
2 ∗ ∗

ρ22 = ρ22 (0) + 2G 2 |Ω| (1 − 2ρ22 (0)) − ∆ (Ωρ12 (0) + Ω ρ12 (0)) (10.53)
(2) 1
2 ∗ ∗

ρ22 = 4G 2 −|Ω| (1 − 2ρ22 (0)) + (∆ + G)Ωρ12 (0) + (∆ − G)Ω ρ12 (0)
(3) 1
2 ∗ ∗

ρ22 = 4G 2 −|Ω| (1 − 2ρ22 (0)) + (∆ − G)Ωρ12 (0) + (∆ + G)Ω ρ12 (0)
(1)
ρ12 = 1
2G2
[∆Ω(1 − 2ρ22 (0)) + Ω (Ωρ∗12 (0) + Ω∗ ρ12 (0))]
(2)
ρ12 = ∆−G
4G2
−Ω(1 − 2ρ22 (0)) + (∆ + G) ΩΩ∗ ρ∗12 (0) + (∆ − G)ρ12 (0)
(3)
ρ12 = ∆+G
4G2
−Ω(1 − 2ρ22 (0)) + (∆ − G) ΩΩ∗ ρ∗12 (0) + (∆ + G)ρ12 (0) .
We derive this solution in Exc. 10.6.2.2.

To begin the discussion of this solution, let us consider a sample of atoms initially in the
ground state when the light field is switched on at time t = 0,
ρ11 (0) = 1 = 1 − ρ22 (0) , ρ12 (0) = 0 = ρ21 (0) . (10.54)
In this case, the conditions (10.53) simplify to,
(1) |Ω|2 (1) 1

ρ22 = 2G2
, ρ12 = 2G2
∆Ω (10.55)
(2) −|Ω|2 (2)
ρ22 = 4G2
, ρ12 = G−∆
4G2
Ω
(3) −|Ω|2 (3)
ρ22 = 4G2
, ρ12 = −G−∆
4G2
Ω ,
such that,
(1) (2) (3) |Ω|2

ρ22 = ρ22 + ρ22 eiGt + ρ22 e−iGt = (2 − eiGt − e−iGt ) (10.56)
4G2
(1) (2) (3) ∆Ω ∆ − G iGt ∆ + G −iGt i∆t
ρ12 = (ρ12 + ρ12 eiGt + ρ12 e−iGt )ei∆t = 2
− Ωe − Ωe e
2G 4G2 4G2
2Ω
= (∆ − ∆ cos Gt + iG sin Gt) ei∆t .
4G2
x
Using cos x = 1 − 2 sin2 2 e sin x = 2 sin x2 cos x2 , we finally obtain,

|Ω|2 Gt Ω Gt Gt Gt
ρ22 = 2 sin2 , ρ12 = 2 sin ∆ sin + iG cos ei∆t . (10.57)
G 2 G 2 2 2
10.2.3 Pauli matrices and the atomic Bloch vector

The internal structure of atoms is analyzed in atomic physics, where we find that the energy
levels are discrete (Bohr’s axiom). The center of mass motion of the atoms and collisions with
other atoms are ignored, and concerning the interaction of the atoms with light, we are only
interested in the aspect, that the interaction can induce transitions between internal states via
absorption or emission of photons. It is the duty of atomic physics to calculate the frequencies
and strengths of transitions (by Hartree-Fock or similar methods), as well as their behavior in
external electric and magnetic fields. The results of these calculations are visualized in energy
level schemes called Grotian diagrams. In quantum optics we do not care, how the energies of
the levels were calculated, but accept them as given. That is, we assume the Hamiltonian of the
unperturbed atom to be diagonalized, so that its internal structure can be written as,
X
Ĥelectron = ~ωj |jihj| . (10.58)
j
The electronic states are orthonormal hi|ji = δij , and we define the transition operators by
σ̂ij |ki = δjk |ii , (10.59)

+
and σ̂ij = σ̂ji satisfying the commutation relation,
[σ̂ij , σ̂lk ] = δjl σ̂ik − δik σ̂lj . (10.60)

Many times we will restrict ourselves to atoms of two or three levels. For a two-level system we
obtain the Pauli spin matrix defined in (2.45). Every 2 × 2 matrix can be expanded on a Pauli
matrix basis,

ρ11 ρ12
= |1iρ11 h1| + |1iρ12 h2| + |2iρ21 h1| + |2iρ22 h2| (10.61)
ρ21 ρ22
= ρ11 ( 12 + 12 σ̂z ) + ρ12 σ̂ − + ρ21 σ̂ + + ρ22 ( 12 − 21 σ̂z )
− +
+ − − + − + hσ̂ σ̂ i hσ̂ − i
= ρ11 σ̂ σ̂ + ρ12 σ̂ + ρ21 σ̂ + ρ22 σ̂ σ̂ = .
hσ̂ + i hσ̂ + σ̂ − i
This formalism can easily be extended to an atom with many levels. Solve the Exc. 10.6.2.3.
For the two-level case it is useful to introduce an alternative notation based on the Bloch
vector defined in (2.46),
     
2 Re ρ12 hσ − i + hσ + i hσx i
β~ ≡  2 Im ρ12  =  i(hσ − i − hσ + i)  = hσy i . (10.62)
ρ22 − ρ11 + −
hσ σ i − hσ σ i− + hσz i
We also define the torque vector,

 
Ω 2 p
G ≡ 0 com G = G = Ω2 + ∆2 , (10.63)
∆
the length of which is simply the Rabi frequency. With this, we can write the Bloch equations,
dβ~
= G × β~ , (10.64)
dt
as will be shown in Exc. 10.6.2.4. ρ12 describes the polarization and ρ22 − ρ11 the population
inversion of the atom. The equation is analogous to the equation of motion for a rigid rotor
or spinning top (for example, a dipole in a homogeneous field). It displays phenomena such as
precession and nutation. The physical content and usefulness of the Bloch vector will become
clearer when we use the formalism to analyze electric and magnetic couplings. In Exc. 10.6.2.5
we verify that the Bloch vector is normalized (as long as spontaneous emission is not considered).
10.2.4 Manipulation of the state by sequences of radiation pulses

The temporal dependence of the three components of the atomic Bloch vector provides a useful
illustration of the atom-field interaction. Resonant coupling, ∆ = 0 and G = Ω, puts the
solutions (10.57) into the form,
ρ22 (t) = 12 (1 − cos Ωt) , ρ12 (t) = i

2 sin Ωt , (10.65)
that is,  
0
~ =  sin Ωt  .
β(t) (10.66)
− cos Ωt
That is, a resonant pulse rotates a Bloch vector initially pointing in the direction −z within the
π π
plane z-y, until it arrives, at time t = 2Ω , at the +y direction and at time t = Ω at the +z
direction. This means that the entire population has been transferred to the excited state. The
Bloch vector continues to rotate (the movement is called nutation) around the torque vector G
which, as can be seen from Eq. (10.64), points at the +x direction when ∆ = 0. The nutation
frequency is proportional to the force Ω of the atom-field interaction. With the Eq. (10.57) we
see that the population oscillates between the ground and excited state with the frequency Ω.
This means that the energy ~ω is periodically exchanged between the atom and the field. A
pulse of resonant light of duration such that τ = π/2Ω is called a π/2-pulse. The nutation is
illustrated in Fig. 10.1(a).
Once the coherence has been excited by a detuned radiation, ∆ 6= 0, the Bloch vector does
not stand still, even after the radiation has been switched off. To see this, we consider again the
general solution (10.53) now entering Ω = 0. If the Bloch vector is initially at a point in the
unitary circle of the plane z-y, it will rotate according to the formula,
ρ22 (t) = ρ22 (0) , ρ12 (t) = ρ12 (0)e−i∆t , (10.67)
that is,
 
ρ12 (0) sin ∆t
~ = ρ12 (0) cos ∆t .
β(t) (10.68)
2ρ22 (0) − 1
That is, the Bloch vector performs a motion of precession around the symmetry axis. The
precession is illustrated in Fig. 10.1(b).
1 1
2ρ22-1
2ρ22-1
0 0
-1 -1
1 1
0 0
1 1
Re ρ -1 -1 0 -1 -1 0
12 Re ρ
Im ρ12 12 Im ρ12
Figure 10.1: (Code: LM Bloch V ectorRabi.m) (a) Nutation of the Bloch vector. The red circles
show the evolution of the Bloch vector on the Bloch sphere for a resonant π-pulse. (b) Precession
of the Bloch vector.
The evolution of the Bloch vector on the surface of the Bloch sphere under the influence
of radiation fields can be considered a coherent trajectory of the wavefunction of the atomic
state, which is therefore subject to interference phenomena [132]. Interferometers can be real-
ized by sequences of consecutive pulses splitting populations, exciting coherences, and remixing
populations.
Sensors based on interferometry of atomic excitation are nowadays among the most accurate
and most sensitive. We will discuss the method of radiation pulse sequences in several exercises:
In Exc. 10.6.2.6 we present the Rabi method and in the Excs. 10.6.2.7, 10.6.2.8, and 10.6.2.9 the
Ramsey method.
10.2.4.1 Atomic ensembles

While it is technically challenging to observe the dynamics of single atoms, it is relatively
easy monitor the dynamics of ensembles of atoms, provided that they react synchronously to
incident radiation. The concentration of a sufficient number of atoms in a small volume can,
however, introduce additional (desirable or undesirable) effects. Collisions, for instance, induce
(irreversible) decoherence. On the other hand, if the ensemble is sufficiently dense that the mean
distance between atoms is less than a resonant wavelength, then the transition dipoles of the
individual atoms will couple to produce a collective dipole moment and generate effects known
as superradiance.
Thermal motion of the atoms is another undesired effect, because every atom will inter-
act with the radiation on a different Doppler-shifted frequency. This leads to diffusion of the
individual atomic Bloch vectors in the x-y-plane, which in turn limits the resolution of interfer-
ometric applications. We will discuss in Exc. 10.6.2.10 the photon echo method, which allows to
circumvent this problem.
10.3 Bloch equations with spontaneous emission

10.3.1 Phenomenological inclusion of spontaneous emission
To find the Bloch equations including spontaneous emission, we insert the phenomenological
term −iγa2 into the Eqs. (5.63), as we have already done for Eq. (??),
da2
Ω∗ cos ωteiω0 t a1 − iγa2 = i
, (10.69)
dt
that is, the equations of motion can be corrected by simply replacing,

da2 d
y + γ a2 . (10.70)
dt dt
Knowing ρmn = a∗m an , it is easy to check,

dρ22 d dρ12 d
y + Γ ρ22 and y + γ ρ12 , (10.71)
dt dt dt dt
with γ = Γ/2, such that the Bloch equations become,
   i  
ρ11 0 Γ 2Ω − 2i Ω ρ11
d  ρ  
 22  =  i 0 −Γ − i
2 Ω i
2 Ω   
 ρ22  .
     (10.72)
dt ρ̃12 2Ω
i
− 2 Ω −i∆ − γ 0 ρ̃12 
i i
ρ̃21 −2Ω 2Ω 0 i∆ − γ ρ̃21
Example 39 (Langevin equation): The Heisenberg equation for the evolution of the
internal degrees of freedom, including the phenomenologically introduced decay, is also called
Langevin equation. It can be written as,
dσ̂
i = ~1 [σ̂, Ĥ] − 2i Γσ̂ ,
dt
and analogously for σ̂z . With the Hamiltonian Ĥ = ~∆σ̂ † σ̂ + 12 ~Ω(eiωt σ̂ + h.c.) we obtain,
using the Pauli spin matrices, exactly the Bloch equations,
iσ̂˙ = ∆[σ̂, σ̂ † σ̂] + 12 Ωe−iωt [σ̂, σ̂ † ] − 2i Γσ̂ = −∆σ̂z − 12 Ωe−iωt σ̂z − 2i Γσ̂
iσ̂˙ z = ∆[σ̂z , σ̂ † σ̂] + 1 Ωe−iωt [σ̂z , σ̂ † ] + 1 Ωeiωt [σ̂z , σ̂] − i Γσ̂z = −Ω(σ̂ † − σ̂) − i Γσ̂z .
2 2 2 2
10.3. BLOCH EQUATIONS WITH SPONTANEOUS EMISSION 253
10.3.1.1 Stationary solution of the Bloch equations

The dissipation introduced by the spontaneous emission allows the system to reach a steady
state. Letting the time derivatives be 0, we obtain the stationary solutions,
1 2 1 i
4 |Ω| 2 Ω(∆ − 2 Γ)
ρ22 (∞) = 1 1 2 and ρ12 (∞) = ei∆t . (10.73)
∆2 + 2
2 |Ω| + 4 Γ ∆2 + 21 |Ω|2 + 14 Γ2
This will be shown in Exc. 10.6.3.1. We see that the populations and coherences both have a
frequency dependence, which is similar but not identical to the one already found for the sus-
ceptibility χ, the absorption coefficient K, and the absorption cross section σ0a in the Eqs. (??),
(??), and (??). But here the denominators have an extra term 12 Ω2 contributing to an effective
widths of ρ22 and ρ12 , p
Γeff = 2|Ω|2 + Γ2 . (10.74)
This effect is called power broadening or saturation broadening. The phase factor ei∆t describes
the optical precession of the Bloch vector.
By introducing the saturation parameter,
2|Ω|2
s≡ , (10.75)
4∆2 + Γ2
we can rewrite the stationary dipole moment and the excited state population (10.73) as,
s/2 ∆ − iΓ/2 s
ρ22 (∞) = , ρ12 (∞) = ei∆t . (10.76)
1+s Ω 1+s
and
s/2
ρ12 (∞)2 = . (10.77)
(1 + s)2
Fig. 10.2(a) shows the Rabi oscillations damped by spontaneous emission. For long times the
population of the excited state ρ22 converges to the asymptote (10.76). Fig. 10.2(b) shows the
temporal evolution of the Bloch vector subject to spontaneous emission.
Example 40 (Resonant excitation and weak excitation): A case where the Bloch
equations can be analytically treated is under resonant excitation, ∆ = 0. In this case, for
the initial conditions, ρ12 (0) = ρ22 (0) = 0, the solution including decay is,

Ω2 −3Γt/4 3Γ
ρ22 (t) = 1 − e cos λt + sin λt and ρ12 (t) = 0 , (10.78)
2|Ω|2 + Γ2 4λ
√
where λ ≡ Ω2 − Γ2 . This solution (which will be derived in Exc. 10.6.3.2), describes the
optical nutation of the Bloch vector along the ρz axis. We note here that, due to spontaneous
emission, the norm of the Bloch vector is NOT conserved, i.e., the Bloch vector evolves to
the interior of the Bloch sphere.
Another case that can be solved analytically is the weakly excited atom, |Ω| γ,
Ω2 −iΩ −(i∆+Γ)t
ρ22 ' 2
1 + e−2Γt − 2 cos ∆t and ρ12 (t) ' e − 1 +ρ12 (0)e−(i∆+Γ)t .
4G 2i∆ + 2Γ
(10.79)
(a) (b)
1 1
2ρ -1
0.5 0
ρ22
22
0 -1
0 -1
5 0
Ω /Γ Im ρ
12 12 12
6 0.5 1
4 0
10 2 1 -0.5
0 -1
Γ t
12 Re ρ
12
Figure 10.2: (Code: LM Bloch T woLevelDecay.m) (a) Rabi oscillations damped by spontaneous
emission for Rabi frequencies between Ω/Γ = 0.2, .., 5. (b) Evolution of the Bloch vector subject
to spontaneous emission (Γ12 = 0.05Ω12 ) after of a resonant π-pulse (blue) and after a π-pulse
with detuning ∆12 = Ω12 /2.
10.3.2 Susceptibility and density operator

The frequency-dependent linear susceptibility completely describes the linear propagation of an
electromagnetic wave within a medium. It is related to the index of refraction and the absorption
coefficient. Nonlinear processes should be described by higher order susceptibilities. Electric
fields E = E0 eiωt + c.c. induce in media characterized by a given susceptibility χ the polarization
P = χE. The polarization is the sum of the dipole moments of the individual atoms, P = nhdi,
where n = N/V is the atomic density. The susceptibility can therefore be expressed by the
Hamiltonian interaction Ĥ = −d · E,
n
χ=− hĤi . (10.80)
|E|2
Using the two-level Hamiltonian (??) we obtain,
~Ω
χ = −n ρ12 ei∆t + c.c. (10.81)
2|E|2
for the polarization,

P = n dρ12 + c.c. (10.82)
and for the susceptibility,

r
2nd2 ∆ + iΓ 3πε0 ~Γ
χ(ω) = with d= . (10.83)
3ε0 ~ 4∆2 + 2|Ω|2 + Γ2 k3
We can insert the new expression (10.73) for ρ12 into our previous expression for hd12 i (??)
and get new expressions for the polarization P(t), (??) and (??), and the susceptibility χ (??).
The modified expression for the susceptibility is,
N d212 −∆ + 12 Γ
χ= . (10.84)
30 ~V ∆2 + 12 |Ω|2 + 14 Γ2
10.3. BLOCH EQUATIONS WITH SPONTANEOUS EMISSION 255
In the imaginary component we obtain the new absorption coefficient,
ω 00 πN e2 |hr12 i|2 ω0 Γ/2π

K= χ (ω) = , (10.85)
cη 30 ~cV ∆ + 12 Ω2 + 41 Γ2
2
and the absorption cross-section,
πe2 |hr12 i|2 ω0 Γ/2π

σ0a = . (10.86)
30 ~c ∆ + 12 Ω2 + 41 Γ2
2
The important new property is the effective width Γeff , which appears in χ, K, and σ0a . In
Exc. 10.6.3.3 and 10.6.3.4 we calculate the impact of the spontaneous emission on the determi-
nant of the density operator. Solve the Excs. 10.6.3.5, 10.6.3.6, 10.6.3.7, and10.6.3.8.
10.3.3 Liouville equation for two levels

In this chapter we chose to include spontaneous emission in the Bloch equations by phenomeno-
logical arguments. We will show more ahead, that dissipation can be treated from general
principles. This treatment, named after Weisskopf-Wigner, derives from a Liouville type equa-
tion (10.36), but which holds for a total density operator ρatom&f ield describing the atom and
the electromagnetic modes, an equation for the density operator of only the atom. The price
to pay for this simplification is an additional term appearing in the equation now called master
equation,
˙
ρ̂(t) = (L0 + Lsp )ρ̂(t) with
, (10.87)
i Γ +
L0 ρ̂(t) ≡ ~ [ρ̂(t), Ĥ] and Lsp = 2 (2σ̂ ρ̂σ̂ − σ̂ + σ̂ ρ̂ − ρ̂σ̂ + σ̂)
where σ̂ ± are the Pauli matrices. We show in Exc. 10.6.3.9, that the known Bloch equations
can be derived from the master equation.
10.3.4 Saturation effects by the effective hamiltonian

The eigenvalues of the effective Hamiltonian of a two-level system excited by radiation,
1

0 2 Ω
Ĥ = 1 i , (10.88)
2Ω ∆ − 2Γ
describe possible effects of line broadening and/or displacement due to coupling,
q 2
1 i 1
2 ∆− 2Γ ± 2 ∆ − 2i Γ + Ω2 . (10.89)
The real parts of the eigenvalues Re E describe shifts and/or splittings of the transition line.
The imaginary parts Im E describe broadening effects of the lines.
In the simplest case, ∆ = 0 and Γ > 4Ω, we find the saturation broadening already discussed
in (10.74), and we will deepen it in Exc. 10.6.3.10. For the case Γ < 4Ω, we observe a splitting
of the line called Autler-Townes splitting, which will be studied in Exc. 10.6.3.11. If Ω 6= 0,
the spectrum becomes asymmetrical. In the case of weak excitation, Γ 4Ω, we observe a
shift of the transition line with dispersive dependence (near the resonance) on the frequency
of the incident radiation. This is the dynamic Stark shift (or light-shift). In the case of strong
excitation, Γ 4Ω, we observe again at the split spectrum, but now the two lines exhibit
an avoided crossing-type dependence on the radiation frequency. We study these effects in
Exc. 10.6.3.12.
Obviously, these effects can be studied by the Bloch equation formalism containing the terms
of spontaneous relaxation.
10.4 Line broadening mechanisms

The resolution of atomic spectroscopy is generally limited by several perturbative effects, many
of them originating in the atomic motion. They manifest themselves as broadening and/or
shifts of atomic resonances. Free atoms, as well as atoms confined in potentials, have kinetic
energy and evolve on extended phase space trajectories. If the spatial localization is less than
the effective cross section of the exciting laser beam, then the interaction time is limited and the
resonance lines are broadened by the Fourier effect in a process called transit time broadening,
and the efficiency of fluorescence collection is reduced. The same happens with the Doppler
effect: Only those atoms that have a specific velocity along the optical axis defined by the laser
beam can interact. Free as well as confined atoms can only scatter when they are in specific
cells of the phase space.
There are two different fundamental types of broadening. The so-called homogeneous broad-
ening affects all atoms in the same way regardless of their positions or velocities. It usually give
rise to Lorentzian line profiles and can be included in the Bloch equations. It correspond to an
acceleration of the relaxation. Examples are the natural linewidth, saturation broadening, and
collision broadening.
The so-called inhomogeneous broadening is due to a displacement of atomic levels, which
may be different for each atom. Averaging over a large sample of atoms, the displacements
generate an effective broadening usually with a Gaussian line profile. It can not be included
in the Bloch equations, but only as an average over all trajectories of all atoms. It does not
correspond to an accelerated relaxation. Inhomogeneous broadening is often due to external
perturbations, e.g., Doppler broadening and broadening due to temporal fluctuations or spatial
inhomogeneities of external electric or magnetic fields. In Exc. 10.6.4.1 we calculate the optical
density of atomic clouds. In Exc. 10.6.4.2 we present a spectroscopic technique bypassing the
Doppler broadening called Doppler-free spectroscopy and calculate the Lamb-dip profile. Finally,
in Exc. 10.6.4.3, we discuss a cooling technique allowing for the reduction of Doppler broadening,
called Zeeman slower.
10.4.1 Saturation broadening

Eq. (10.74) shows that when the power of the incident light increases, the population of the
excited state saturates at a limit value of ρ22 = 21 . This property is analogous to Eq. (1.116),
which shows the same saturation behavior when a two-level atom is subjected to broadband
radiation. The saturation parameter defined in (10.75) measures the degree of saturation. When
the narrowband light source is tuned to resonance, the saturation parameter is basically a
measure for the ratio between the stimulated population transfer rate Ω and the spontaneous
decay rate A21 . We can rewrite the stationary population of the excited level as in (10.76). In
resonance and with the saturation parameter s = 1, we obtain
Ω= √1 Γ . (10.90)
2
10.4. LINE BROADENING MECHANISMS 257
We can use equation (10.90) to define the saturation intensity Isat for an atom with the transition
dipole d12 . From Eq. (1.112) we have,
s
2I¯
E0 = . (10.91)
0 c
Therefore, using the definition of the Rabi frequency, ~Ω = d12 E0 , and the relationship between
d12 and Γ given by Eq. (??), we have,6
g1 2π 2 c~
Isat = Γ , (10.92)
g2 3λ30
taking into account the degeneracies gj of the levels. In Exc. 10.6.4.4 we calculate the saturation
intensity of popular atomic transitions.
10.4.2 Collision broadening

The theory of atomic collisions covers a large area of research, including elastic and inelastic,
reactive and ionizing processes. In low-pressure gases at room temperature or hotter we need
only consider the simpler processes: long-range van der Waals interactions that result in elastic
collisions. The ’low pressure’ criterion requires that the average free path between collisions
be greater than any linear dimension of the gas volume. Under these conditions, collisions can
be modeled with straight trajectories, along which the interaction time is short and the time
between collisions is long in comparison with the radiative lifetime of the excited atomic state.
Then, the impact of a collision on the emission of a radiating atom causes a loss of coherence
due to a phase interruption of the excited state atomic wavefunction. The term ’elastic’ means
that the collision does not disturb the populations of the internal states, so we only need to
consider the off-diagonal elements of the density matrix,
dρ12 Ω0
= i ei(ω−ω0 )t (ρ11 − ρ22 ) − γ 0 ρ12 , (10.93)
dt 2
where γ 0 is the sum of the spontaneous emission γ and the collision rate γcol ,
γ 0 = γ + γcol . (10.94)
The inverse of the collision rate is simply the time between phase interruptions or the time
between collisions. Now, for collisions between hard cores of atoms of mass m (with reduced
mass mred = m/2) and with radius ρ in a gas with density n consisting of a single species,
a standard analysis based on the kinetic theory of dilute gases shows that the time between
collisions is given by the collision rate,
−1
γcol = τcol = σnv̄ , (10.95)
q
8kB T
where v̄ = πm is the average collision velocity in a homogeneous gas at the temperature T
√ 2
red
and σ = 8πρ the collision cross section. Thereby,
8ρ2 n
γcol = p . (10.96)
mred /πkB T
6
Some authors define the saturation for s = 2, as happens when Ω = Γ.
We can now relate this simple result of gas kinetics to the phase interruption rate by reinter-
preting the meaning of the collision radius. When an excited atom propagating through space
suffers a collision, the long-range interaction will produce a time-dependent perturbation of the
energy levels of the radiating atom and a phase shift in the radiation,
Z ∞ Z ∞
η= [ω(t) − ω0 ]dt = ∆ω(t)dt . (10.97)
−∞ −∞
The long-range van der Waals interaction is expressed by,

Cn
∆E = ~ω = , (10.98)
[b2 + (vt)2 ]n/2
where b is the impact parameter of the collision trajectory and v the collision velocity. The
phase shift is then Z
1 ∞ Cn
η= dt . (10.99)
~ −∞ [b + (vt)2 ]n/2
2
The integral is easily assessed for the two most frequent cases: non-resonant van der Waals
interactions n = 6 and resonant van der Waals interactions n = 3. The phase shifts are,
2π C6 4π C3
η6 (b) = and η3 (b) = √ . (10.100)
3~ b56 v 3 3~ b23 v
Now, if instead of using the hard core approximation, we define a collision as an encounter
causing a phase shift of at least 1 radians, we have a new condition for the collision radius,

2π C6 1/5 4π C3 1/2
b6 = and b3 = √ . (10.101)
3~ v 3 3~ v
Replacing these collision radiuses for the radius ρ in Eq. (10.96) and inserting the average
collision velocity, we find the collision rate,
√ 2 !2/5 3/2 2 3/10
2π C6 4πkB T 2 π C3
γc6 = 4n and γc3 = 4n . (10.102)
3~ µ 3 ~
Substituting the generalized γ 0 of (10.94) for γ in the Bloch equations (10.73), we find the
stationary solutions,
1 γ0 2 1 0
4 γ |Ω| 2 Ω(∆ − iγ )
ρ22 = and ρ12 = ei(ω−ω0 )t . (10.103)
1 γ0 0
∆2 + 2 γ |Ω|
2 + γ 02 ∆2 + 12 γγ |Ω|2 + γ 02
The effective linewidth (radiative and collisions) is,
q
0
Γ0eff = 2 γ 02 + 12 γγ |Ω|2 . (10.104)
When the excitation is sufficiently weak, so that power broadening can be neglected in compar-
ison to collision broadening, the second term can be discarded,
Γ0eff = 2(γ + γcol ) . (10.105)
The equations (10.74) and (10.105) express the linewidths in the limits of dominating power
and collision broadening, respectively. Note that the susceptibility, absorption coefficient, and
absorption cross-section retain their Lorentzian profile, but with a larger width due to collisions.
Since each atom is subject to the same broadening mechanism, the broadening is homogeneous.
Solve Excs. 10.6.4.5 and 10.6.4.6.
10.4. LINE BROADENING MECHANISMS 259
10.4.3 Doppler broadening

The Doppler broadening is simply the apparent frequency distribution of a sample of radiating
atoms at temperature T . The contribution of each atom to the radiation appears detuned by
the Doppler shift because of its velocity. The frequency shift for a non-relativistically moving
particle is ω = ω0 /(1 vc ), such that,
v
∆ ≡ ω − ω0 ' ω0 = k · v = kvz , (10.106)
c
where k is the wavevector of the light and v is the velocity of the atom. This distribution of
Doppler shifts of a gaseous sample in thermal equilibrium follows the probability distribution of
velocities,
2 2 2 2
P (vz )dvz ∝ e−mvz /2kB T dvz = e−mc ∆ /2ω0 kB T ωc0 dω . (10.107)
This frequency distribution is a Gaussian centered at ω = ω0 and with the width,
2
2kB T ln 2
FWHM = 2ω0 . (10.108)
mc2
A measure of the width is also the standard deviation,

r
2ω0 kB T FWHM
2σ = = . (10.109)
c m 1.177
From Eq. (10.107) we can see that the line profile is,
1 m −(ω−ω0 )2 /2σ2
D(ω − ω0 ) ≡ √ e dω . (10.110)
2π kB T
The profile compares with the Lorentzian profile Eq. (??) associated with natural, power, or col-
lision broadening. Doppler broadening is a property of the atomic ensemble, each atom suffering
a unique but different displacement than the other atoms. Hence, it is called inhomogeneous
broadening.
The Heisenberg equation used to derive the Bloch equations assumes immobile atoms. How-
ever, we can easily apply the Galilei transformation to a system, where the atoms move with
the given velocity v,
(∂t + v · ∇)ρ(r, t) = − ~i [Ĥ, ρ(r, t)] . (10.111)
Since the light fields propagate as ei(ωt−k·r) , the solution of the above equation simply follows
from the immobile solution with the substitution ∆ → ∆ − k · v.
2
For a cloud obeying Maxwell’s velocity distribution, P (v) ∼ e−mv kB T ,
Z
1 2 2
ρ̄(∆) = √ e−(k·v) /2δ ρ(∆ − k · v)d(k · v) . (10.112)
2πδ R
The average of the density operator over all velocities, ρ̄, therefore follows as the convolution
of the density operator ρ (obtained as the solution of the Bloch equation) and the Gaussian
2 2
function G(∆) = (2πδ 2 )−1/2 e−∆ /2δ ,
ρ̄(∆) = (G ? ρ)(∆) . (10.113)

10.4.4 Voigt profile

It is clear that in many practical circumstances homogeneous and inhomogeneous processes
simultaneously contribute to the broadening of lines. In these cases, we can consider that
the radiation of each atom, homogeneously broadened by phase-interruption processes (such
as spontaneous emission or collisions), is displaced by the Doppler effect within the Maxwell-
Boltzmann distribution corresponding to the temperature T . The profile of the gaseous sample,
therefore, is a convolution of homogeneous and inhomogeneous profiles. The resulting profile is
called Voigt profile:
Z ∞ Z ∞ 2 2
0 0 γ e−(ω−ω0 ) /2σ
V (ω − ω0 ) = L(ω − ω0 − ω )D(ω − ω0 )dω = √ 0 )2 + (γ/2)2
dω 0 .
−∞ 2σ 2π −∞ (ω − ω 0 − ω
(10.114)
This integral has no analytical solution, but it is easy to solve numerically. Resolve Excs. 10.6.4.7
and 10.6.4.8.
10.4.5 Bloch equations with phase modulation

In some situations, the atom vibrates thus producing an oscillating Doppler shift. Also, external
magnetic fields or oscillating laser frequencies can produce this effect. We incorporate this
temporal modulation (frequency Ωa ) of the light frequency shifts, induced by the Doppler effect,
into the optical Bloch equations via the substitution [226],
∆ij → ∆ij + kij · v cos Ωa t . (10.115)
The Bloch equations can then be brought into the form,
ρ̇ = (L + 2X cos Ωa t)ρ + b , (10.116)
where the matrix X contains the modulation index of the atomic motion kij v. The stationary
solution of the differential equation, averaged over the time of an oscillation period, can be
expressed as an infinite continuous fraction:
1
ρ(∞) = −(L + S+ + S− )−1 b where S± = −X 1 . (10.117)
L ± iΩa 1 − X L±iΩ 1 X
a 1−X ··· X
This solution replicates the correct excitation spectra even for a multilevel system.
Let us be more specific for a two-level system. Its Hamiltonian is, Ĥint = 12 ~Ωe−i[ωt−k·v/Ωa sin Ωa t] ,
such that the Bloch equation is,
         
ρ −Γ − 2i Ω 2i Ω 0 0 0 ρ22 0
d  22   i
ρ12 = − 2 Ω −Λ 0  + 2 cos Ωa t 0 − 2i kv 0  ρ12  +  2i Ω  .
dt i
ρ21 i
2Ω 0 −Λ∗ 0 0 2 kv ρ21 − 2i Ω
P (10.118)
We look for the stationary solution by letting ρ̇n = 0 and ρ = ∞ n=−∞ n ρ e −inΩa t and projecting
on e−inΩa t by,
(L + inΩa 1)ρn + X(ρn+1 + ρn−1 ) + bδn0 = 0 . (10.119)
Now we define ρn±1 = S± ρn for n R 0. Then, equation (10.119) becomes,
− −1
ρ0 = −[L + X(S+
0 + S0 )] b para n = 0, (10.120)
S±
n∓1 = −[L + inΩa 1 + XS± −1
n] b for n≷0.
10.5. MULTI-LEVEL SYSTEMS 261
Substituting the equation below into the equation above,

−1
−X| −X| −X| −X|
ρ0 = − L + + + ... + + + ... b . (10.121)
|L + iΩa |L + 2iΩa |L + iΩa |L + 2iΩa
0.2
0.15
22
0.1
ρ
0.05
0
-10 -5 0 5 10
Δ
12
Figure 10.3: (Code: LM Bloch M icromotion.m) Spectral broadening due to the periodic move-
ment of the atom.
10.5 Multi-level systems

The two-level system represents an idealization of the real atom, since at least one of the levels is
usually degenerate. Many important phenomena in quantum optics are not found in this system,
but conditioned to the existence of a third level. Examples are optical pumping (essential for
laser operation), quantum jumps or dark resonances [which are at the basis of the phenomenon
of electromagnetically induced transparency(EIT)].
To derive the Bloch equations for atoms with several levels excited by several lasers and
coupled to free space (i.e. without external cavity), we can use the same master equation (10.87),
but with a generalized Hamiltonian and Lindbladt operator,
X
Ĥatom = ~ωi σ̂ji σ̂ij (10.122)
i

Ĥclas = ~2 Ωij e−iωij t σ̂ij + eiωij t σ̂ji
X
+ + + + + +
Latom ρ = Γij [σ̂ij , ρ̂σ̂ij ] + [σ̂ij ρ̂, σ̂ij ] + 2βij [σ̂ij σ̂ij , ρ̂σ̂ij σ̂ij ] + [σ̂ij σ̂ij ρ̂, σ̂ij σ̂ij ] .
i,j
The levels have the energy ~ωi above the ground level.
Let us first have a look at the coherent part of the master equation. The Hamiltonian in the
semiclassical approximation (that is, the atom is quantized and consists of several levels |ii with
energies ~ωi , while the light fields are described by factors eiωij t , with frequencies ωij tuned near
the transitions |ii − |ji) includes the following contributions
X X
Ĥ = Ĥatom + Ĥatom−f ield = |ii~ωi hi| + |ii ~2 Ωij hj|eiωij t + c.c. . (10.123)
i
i<j com Ei <Ej
The Rabi frequency Ωij is a measure for the force at which the levels |ii and |ji are coupled
by the resonantly irradiated light field. The master equation can be simplified by applying the
rotating wave approximation and transforming to the coordinate system which rotates with the
light frequencies ωij :
ρij → ρ̂ij eiωij t , Ĥatom−f ield → e−iĤt/~ Ĥatom−f ield eiĤt/~ . (10.124)
Finally, the master equation can be reformulated by introducing a generalized Bloch vector, ρ,
and the matrix representation of the Liouville superoperator L as a linear system of n2 coupled
differential equations,
d
ρ = Lρ , ρ = (ρ11 .. ρnn ρ12 ρ21 .. ρn−1 n ρn n−1 ) . (10.125)
dt
Alternatively to the complex formulation, the differential equations can be written for the
real and imaginary part of the Bloch vector. The components ρii correspond to the population
probabilities of the levels |ii, the non-diagonal elements ρij describe the coherences between |ii
and |ji. Now, we must insert the Hamiltonian (10.123) and the density operator ρij into the
Liouville equation (10.36) in order to derive the generalized Bloch equations. In practice, these
calculations are simple but heavy. Therefore, we describe in Sec. 10.5.4 a simplified recipe for
compiling Bloch equations for arbitrary level systems for real atoms.
10.5.1 Liouville equation for many levels

The indices for the atomic levels are joined to a single index, such that the master equation
takes a simpler form after having introduced a Liouville operator:
X
ρ̂ = (...ρk ...) ≡ |iiρij hj| , (10.126)
i,j
ρ̂˙ = Lρ̂,
ρ̂ = eL0 t ρ̂0 .
P P
The relation with the von Neumann equation with Ĥ = i,j |iiHij hj| and ρ̂ = k,l |kiρkl hl| and
σ̂ij = |iihj| is:
X X
L0 ρ̂ = − ~i [Ĥ, ρ̂] = −i Hkl ρlj |kihj| + i Hlj ρkl |kihj| . (10.127)
k,l,j k,l,j
For example, for the two-level system with the definition of the external product (2.109):
L0 = −iĤ ⊗ I + iI ⊗ ρ̂ . (10.128)
The relaxation terms are,

X
Ld ρ̂ = γij (2σji ρ̂σij − σij σji ρ̂ − ρ̂σij σji ) (10.129)
i,j
X
= (2γji δkj ρii − γij ρkj − γik ρkj ) |kihj| .
i,j,k
Example 41 (Liouville equation for two levels): For example, for the two-level system,
   
2γ11 0 0 2γ21 2(γ11 + γ12 ) P 0 0 0
 0 0 0 0   0 (kj) γ(kj) 0 0 
L0 = 
 0
−
 
P  .

0 0 0 0 0 (kj) γ(kj) 0
2γ12 0 0 2γ22 0 0 0 2(γ22 + γ21 )
Here, we consider (kj) = (11 12 21 22) as a single index.

10.5.2 Bloch equations for three levels

In principle, three-level system can exist in there possible configurations, shown in Fig. 10.4.
Note that it is not possible to describe a three-level system with all levels pairwise coupled by
three lasers within the formalism of Bloch’s equations 7 .
Figure 10.4: Three level system (a) in Λ-configuration, (b) in V -configuration, and (c) in cascade
configuration.
Defining the Bloch vector by ρ ~, the Bloch equation matrix for three levels in Raman config-
uration (that is, in Λ-configuration) using the labeling of Fig. 10.4(a), is,
~˙ = L~
ρ ρ (10.130)
 i  
0 Γ12 Γ13 2 Ω12 − 2i Ω12 0 0 0 0 ρ11
 0 −Γ12 − Γ23 0 − 2i Ω12 i
2 Ω12 0 0 i
Ω
2 23 − 2i Ω23  ρ22 
 0 Γ23 −Γ13 0 0 0 0 − 2i Ω23 i  
 2 Ω23  ρ33 
 i
− 2i Ω12 −Λ12 i  
 2 Ω12 0 0 Ω
2 23 0 0 0  ρ12 
=
 − 2i Ω12 i
2 Ω12 0 0 −Λ∗12 0 − 2i Ω23 0 0 
 ρ21 

 0 0 0 i
Ω
2 23 0 −Λ13 0 - 2i Ω12 0  ρ13 
  
 0 0 0 0 − 2i Ω23 0 −Λ∗13 0 i
Ω
2 12
 ρ31 
 i  
0 2 Ω23 − 2i Ω23 0 0 − 2i Ω12 0 −Λ23 0 ρ23
0 − 2i Ω23 i
2 Ω23 0 0 0 i
Ω
2 12 0 −Λ∗23 ρ32
with Λmn = i∆mn + γmn and,
∆13 = ∆12 − ∆23 (10.131)

1 1 1
γ12 = 2 (Γ12 + Γ23 ) , γ23 = 2 (Γ12 + Γ23 + Γ13 ) , γ13 = 2 Γ13 .
In Exc. 10.6.5.1 we will derive the matrix (10.130).

The coherent terms of the same matrix can be used for the V - and the cascade configurations
shown in Figs. 10.4(b,c). Obviously, the incoherent terms, that is, the submatrix 3 × 3 separated
in the matrix (10.130) containing the population decay rates must be adjusted, as well as the
decay rates of the coherences on the diagonal. Finally, the definition of the Raman detuning
∆13 must be adjusted. For the system in V -configuration we have,
 
−Γ12 − Γ13 0 0
Lincoh = Γ12 0 Γ23  , ∆13 = ∆12 − ∆23 (10.132)
Γ13 0 −Γ23
γ12 = 21 (Γ12 + Γ13 ) , γ23 = 21 Γ13 , γ13 = 12 (Γ12 + Γ13 + Γ23 ) .
7
For the same reason that the three-body problem has no general analytic solution.
For the cascade system we have,

 
0 Γ12 Γ13
Lincoh = 0 −Γ12 Γ23  , ∆13 = ∆12 − ∆23 (10.133)
0 0 −Γ13 − Γ23
γ12 = 21 Γ12 , γ23 = 12 (Γ12 + Γ23 + Γ13 ) , γ13 = 12 (Γ13 + Γ23 ) .
These matrices serve to calculate, among others, the phenomena of Autler-Townes splitting
treated in Exc. 10.6.3.11, of the light-shift treated in Exc. 10.6.3.12, the dark resonances treated
in Exc. 10.6.5.2, the STIRAP method treated in Exc. 10.6.5.3, adiabatic sweeps treated in
Exc. 10.6.5.4, the dispersive interaction between atoms and light treated in Exc. 10.6.5.5, Fano
resonance-type line profiles of dark resonances treated in Exc. 10.6.5.6, and the quantum jumps,
which will be studied in later chapters. In Excs. 10.6.5.7 and 10.6.5.8 we will show, that an atomic
gas may have negative permittivity and negative permeability and, consequently, properties
usually only found in artificial metamaterials, as for example, a negative refractive index.
10.5.3 Numerical treatment of Bloch equations

Since the differential Bloch equations are linear, they can be easily solved. For example, the
prescription
~(t) = eLt ρ
ρ ~(0) (10.134)
propagates the Bloch vector to later times.
The matrix L is not invertible, but by applying the condition Tr ρ = 1, a component of the
density matrix can be eliminated, for example by letting,
X
ρ11 = 1 − ρkk . (10.135)
k
~red , has the length n2 −1, and from L we obtain the (trace-)reduced,
The resulting state vector, ρ
now invertible matrix Lred and the inhomogeneity vector b. The differential equation is now,
d
~red = Lred ρ
ρ ~red + b , (10.136)
dt
with the stationary and time-dependent solutions,
~red (∞) = −L−1

ρ red b , ~red (t) = eLred t ρ
ρ ~red (0) + (1 − eLred t )~
ρred (∞) . (10.137)
Once the matrix L or the matrix Lred and the inhomogeneity vector b are determined for
a system, the state of the atom can be calculated at any time, as well as the populations
and coherences. The system’s free parameters are the natural transition linewidths and the
detunings, as well as the intensities and emission bandwidths of the incident light fields.
10.5.3.1 Simulation of the Schrödinger and Bloch equation

Once we have written the solution of the Schrödinger equation in the form (5.74) with a time-
independent Hamiltonian Ĥ, or of the Bloch equations in the form (10.134) or (10.137) with a
time-independent Liouvillian L, we can easily simulate temporal evolutions of quantum systems.
If the Hamiltonian or Liouvillian depend on time, for example, when the Rabi frequencies are
pulsed or the detunings are ramped, we must solve the equations iteratively. That is, we chose
time intervals ∆t sufficiently short, so that the Hamiltonian (or the Liouvillian) can be considered
constant during this interval, and we propagate the wavefunction (or the Bloch vector) to later
times via:
|ψ(t + ∆t)i = eiĤ(t)∆t |ψ(t)i or ~(t + ∆t) = eL(t)∆t ρ
ρ ~(t) , (10.138)
and insert the solution obtained again into equations (10.138) with the Hamiltonian Ĥ(t + ∆t)
(or the Liouvillian L(t + ∆t)) adjusted to the new time.
Example 42 (Electromagnetically induced transparency ): In some special cases, the
three-level Bloch equations can be solved analytically. The system in Λ-configuration schema-
tized in Fig. 10.4(a), where the two lasers satisfy the condition ∆12 = ∆23 can exhibit a dark
resonance leading to the phenomena of electromagnetically induced transparency (EIT) and
electromagnetically induced absorption. In these resonances a dramatic change of the refrac-
tive index is observed despite the fact that the atom becomes transparent, Re χ 0 and
|Im χ| Re χ: p
Re n = 1 + Re χ 0 ,
resulting in a high group velocity,
c
vg = dn
.
n + ω dω
EIT is usually studied in Λ-type systems, but similar phenomena can be found in cascade-
type systems [262, 259], which will be studied here. Disregarding the decay rate Γ13 , the
Bloch equations (10.130) and (10.133) give the coherences,
iΩ12 iΩ23
ρ̇12 = −Λ12 ρ12 + 2 (ρ11 − ρ22 ) − 2 ρ13
ρ̇13 = −Λ∗13 ρ13 − iΩ12 iΩ23
2 ρ23 − 2 ρ12
iΩ23 iΩ12
ρ̇23 = −Λ23 ρ23 + 2 (ρ22 − ρ33 ) − 2 ρ13 .
Assuming stationarity and negligible depletion of the ground state, ρ11 = 1,
iΩ12 iΩ23
0 = −Λ12 ρ12 + 2 − 2 ρ13
0 = −Λ∗13 ρ13 − iΩ12 iΩ23
2 ρ23 − 2 ρ12
iΩ12
0 = −Λ23 ρ23 − 2 ρ13 .
Substituting the third into the first equation,

iΩ12 iΩ23
0 = −Λ12 ρ12 + 2 − 2 ρ13
Ω212
0 = −Λ∗13 ρ13 − iΩ23
4Λ23 ρ13 − 2 ρ12 .
and finally,
iΩ12 4Λ∗13 Λ23 + Ω212
ρ12 = .
2 Λ12 (4Λ∗13 Λ23 + Ω212 ) + Ω223 Λ23
The macroscopic polarization is now P = N V d12 ρ21 , with the number of atoms N . In the
limit of weak probes, the dressed susceptibility follows from P = ε0 χE12 = NV d12 ρ21 ,
N d12 N |d12 |2
χ= ρ21 = ρ21 .
V ε0 E12 V ε0 ~Ω12
1
For a resonant probe laser, ∆23 = 0 and with Γ13 ' 0, we have Λ13 = 2 Γ23 + i∆12 and
1 Ω2
Λ23 = 2 (Γ23 + Γ12 ). The susceptibility in the probe transition is now, using Θ ≡ Γ23 + 2Λ1223 ,
2
N |d12 |2 Ω
Γ23 + 2Λ12 −2i∆12
χ= iΩ12 23
V ε0 ~Ω12
2

Ω
Γ23 + 2Λ12 −2i∆12 (Γ12 +2i∆12 )+Ω223
23
2
N |d12 |
= iΩ12 (Θ−2i∆12Θ−2i∆ 12
)(Γ12 +2i∆12 )+Ω223
= χ0 + iχ00 .
V ε0 ~Ω12
We consider, for example, the intercombination line of atomic strontium 1 S0 -3 P1 (λ12 =

689 nm and Γ12 = (2π) 7.6 kHz) be the ’dressing’ transition 3 P1 -(5s4d)3 D1 (λ23 = 2700 nm
and Γ23 = (2π) 90.3 kHz), be the ’dressing’ transition 3 P1 -(5s5d)3 D1 (λ23 = 497 nm and
Γ23 = (2π) 2.3 MHz), both characterized by Γ23 Ω12 , Γ12 , |∆12 |, such that Θ ' Γ23 .
Hence,
N |d12 |2 2∆12 + iΓ23
χ0 + iχ00 = .
V ε0 ~ Ω23
The refraction index follows with,
p 1
n= 1+χ'1+ χ .
2
Its imaginary part originates from the decay term of the atom: it is here responsible for the
absorbing nature of the cloud. EIT is characterized by a pronounced dispersion and a small
concomitant absorption.
−3
x 10
2 0.5
1.5
ρee
1 Re χ
0.5 0
0
−20 −10 0 10 20
0.2
−0.5
0.15 −20 −15 −10 −5 0 5 10 15 20
mm
0.1
ρ
0.05
1
0
−20 −10 0 10 20 0.8
1
0.6
Im χ
0.95
0.4
gg
0.9
ρ
0.2
0.85
0.8 0
−20 −10 0 10 20 −20 −15 −10 −5 0 5 10 15 20
Δge/Γge Δge/Γge
Figure 10.5: (Code: LM Bloch EitT hreeLevelCascadeStat.m) EIT signal for the cascade sys-
tem of strontium with the transitions at 689 nm and 497 nm with Ω12 = Γ12 , Ω23 = Γ23 and
∆23 = 0. The red lines are calculated by numerical integration of the Bloch equations. The
dotted lines are obtained from analytical formulas based on the assumptions of weak ground
state depletion (which is not really correct in the chosen parameter regime.
10.5.4 General rules for setting up multilevel Bloch equations

The canonical way of deriving multi-level Bloch equations starts from a von Neumann equation
for the total density operator for the atom embedded in the electromagnetic mode structure of
the environment including incident laser beams. After tracing over the degrees of freedom of
the electromagnetic vacuum and using the Markov and the Born approximations, one arrives at
a master equation of the form [91],
i
ρ̂˙ = − [H, ρ̂] + LΓ ρ̂ + Lβ ρ̂ . (10.139)
~
Here H is the Hamiltonian of coupled atom-laser system. The Liouville superoperators,

X h i h i
† †
LΓ ρ̂ = Γkl σkl , ρσkl + σkl ρ, σkl (10.140)
k,l
X h i h i
† † † †
Lβ ρ̂ = Γkl σkl σkl , ρ̂σkl σkl + σkl σkl ρ̂, σkl σkl ,
k,l
describe dissipation through spontaneous emission from the various excited states and decoher-
ence due to phase fluctuations of the laser fields, respectively. The operators σkl = |kihl| describe
the transition between two states. Γkl is the spontaneous decay of level |ki in to |li, and βkl
is the emission bandwidth of the laser coupling the two levels. Simple but tedious algebraic
transformations of the master equation lead, in the rotating wave approximation, to a set of
linear first-order differential equations in the populations of the atomic excitation levels and the
coherences between them. The equations are called the optical Bloch equations.
Alternatively, the Bloch equations may be found by breaking down the multi-level scheme
into a set of three-level systems. Respecting a few symmetry considerations, the multi-level
Bloch equations can then be reassembled from the three-level Bloch equations corresponding
to every possible combination of three levels. Based on such considerations, we provide in the
following a simple recipe for setting up Bloch matrices for arbitrary level schemes. Let us regard
a n-level atom. Its internal state is fully described by the populations ρkk and the (complex)
coherences ρkl , with k, l = 1, .., n. In this work we describe the coherences by their real and
imaginary parts. The labeling is such that the levels are sorted according to their excitation
energy, Ek < El for k < l. We define the Bloch vector,
~ ≡ (ρ11 ...ρnn Im ρ12 Re ρ12 Im ρ13 Re ρ13 ...
ρ (10.141)
...Im ρ1n Re ρ1n Im ρ23 Re ρ23 ...
...Im ρn−1,n Re ρn−1,n ) .
The Bloch equations then formally read,
~˙ = M ρ
ρ ~, (10.142)
where in the given Bloch vector basis the matrix M has the following structure,
 
(A) (B1 )
M = (C) (D)  . (10.143)
(B2 )
(D) (C)
The different blocks of the matrix have the following significations. Block A handles the
transfer of populations by spontaneous decay. Its rank corresponds to the number of levels n.
The diagonal elements of this block are the decay rates Γ of the excited states. The off-diagonal
elements Γkl denote the gain of level k from a decaying level l. Conservation of energy P thus
requires that the sum of the transition rates cancels for every column of matrix A, Γ = k Γkl ,
as it is the case for the two-level Bloch matrix. If the levels are sublevels of a Zeeman and/or
hyperfine split multiplet, the rates have to be weighted with Wigner’s (3j) and {6j} symbols,
Γkl = ΓSkl . The relative oscillator strengths Skl are given in Sec. 10.5.4.2.
The blocks Bk treat the interdependence of the populations and the coherences. B1 describes
how the coherence between any pair of states driven by a light field generating a Rabi frequency
Ωkl influences the populations. The block consists of convoluted 2 × 2 matrices of the form,

ρ̇kk −Ωkl 0 Im ρkl
∼ . (10.144)
ρ̇ll Ωkl 0 Re ρkl
B2 describes how the populations in turn influence the coherences,

1
Im ρ̇kl Ω − 12 Ωkl ρkk
∼ 2 kl . (10.145)
Re ρ̇kl 0 0 ρll
The Rabi frequencies have to be weighted not only with the relative oscillator strength Skl , but
also with the projection Hkl of the laser polarization onto the orientation of the magnetic field
and the laser polarization, Ωkl = Ωx Skl Hkl . Here Ωx is the Rabi frequency generated by a laser
on a transition, whose oscillator strength is 1. The projection is calculated in Sec. 10.5.4.3 for
the three possible laser polarizations, i.e. for σ ± and for π light.
The matrix C rules the influence of the decays of the coherences, of the detunings ∆kl =
ωx − ωatom , and the laser linewidths βkl . Note that the detuning of the laser frequency ωx is
negative for red-detuned light. In the chosen basis it breaks down into an array 2 × 2 matrices
aligned along the diagonal of M . Their shape is,
γkl
Im ρ̇kl − 2 −∆kl Im ρ̇kl
∼ . (10.146)
Re ρkl ∆kl − γ2kl Re ρkl
P
where γkl = m,Em <Ek ,El (Γkm + Γlm ) + 2βkl . Often the levels are sublevels of a Zeeman and/or
hyperfine split multiplets. In this case the frequency shift Zkl of the level is added to the detuning
∆kl . The shift is calculated in Sec. 10.5.4.4 for the example of the 6 Li D2 line.
The block D governs the interdependences of all laser-driven coherences of the atom. The
block contains 2 × 2 submatrices at any place of the matrix M , where the row index pair (mn)
and the column index pair (kl) have one index in common provided the two different indices
correspond to the Rabi frequency of an incident laser:,

Im ρ̇mn 0 ± 12 Ωkl Im ρpq
∼ . (10.147)
Re ρ̇mn ± 21 Ωkl 0 Re ρpq
The submatrix elements indexed by column (pq) and row (mn) are non-zero if one of the indices
p or q is equal to one of the indices m or n and the unequal indices correspond to a laser-driven
transition. In order to find the correct signs of the submatrix elements, we distinguish four
cases: 1. For m = p, n = k, and q = l the signs are: (− + ); 2. for n = q, m = k, and p = l the
signs are: (+ − ); 3. for m = q, n = k, and p = l the signs are: (+ + ); and 4. for n = p, m = k,
and q = l the signs are: (− − ). A proper parametrization is proposed in the next section.
10.5.4.1 Recipe for D transitions in alkalines

In order to give a simple algorithm we parametrize the particular choice of sorting the com-
ponents of the vector, we define a new index µ running from 1 to n2 by setting (%µ ) ≡ (ρkl ),
where,
µ(k, k) = k (10.148)
2
µ(k, l) = 2nk − n − k − k + 2l − 1 ,
so that,
%µ(k,k) = ρkk (10.149)

%µ(k,l) = Im ρkl
%µ(k,l)+1 = Re ρkl .
The Bloch equations then formally read,

%̇µ = Mµν %ν . (10.150)
We illustrate the procedure by considering the case of the 6 Li D2 line with 6 ground states
k ∈ G ≡ {1, ..., 6} belonging to the 2 S1/2 hyperfine levels F = 21 , 32 and 12 excited states
k ∈ E ≡ {7, ..., 18} belonging to the 2 P3/2 hyperfine levels F = 12 , 32 , 52 .
According to the parametrization (10.142) the block A of the matrix M is filled with the
following components,
Mµ(kk),µ(kk) = −Γ for k ∈ E (10.151)
Mµ(kk),ν(ll) = ΓSkl for k ∈ G and l ∈ E ,
where the relative oscillator strength Skl is given by Eq. (10.158).
The blocks Bk of the matrix M are filled with the components,
Mµ(k,k),ν(k,l) = −Mµ(l,l),ν(k,l) (10.152)
= −2Mµ(k,l),ν(k,k) = 2Mµ(k,l),ν(l,l)
= −Ωkl Skl Hkl ,

kl
ˆ
for k ∈ G and l ∈ E. The projection onto the quantization axis Hkl = Hkl kl | , B, ml
|ˆ − mk is
given by Eq. (10.161).
The block C contains the components,
Mµ(k,l)µ(k,l) = Mµ(k,l)+1µ(k,l)+1 (10.153)
= − Γ2 − Γ2 δk≥7 − βpb δ2<k<7 − βrp δk≤2
for k ∈ G ∪ E and l ∈ E and ,
Mµ(k,l),µ(k,l)+1) = −Mµ(k,l)+1,µ(k,l)) (10.154)
= −∆pb δk>2 − ∆rp δk<3 + Zkl (B)
for k ∈ G and l ∈ E,
Mµ(k,f ),µ(k,f )+1) = −Mµ(k,f )+1,µ(k,f )) (10.155)
= −∆pb + ∆rp + Zk,10 (B) − Zf,10 (B)
for k ∈ {1, 2} and f ∈ {3, .., 6}. The Zeeman shift Zkl is given by Eq. (10.162).
Finally, the block D is filled with the components,
Mµ(k,f ),µ(f,l)+1 = (δf <k − 12 )Ωkl Skl Hkl (10.156)
Mµ(k,f )+1,µ(f,l) = (δl<f − 21 )Ωkl Skl Hkl
Mµ(l,f ),µ(f,k)+1 = (δl<f − 12 )Ωkl Skl Hkl
Mµ(l,f )+1,µ(f,k) = (δf <k − 12 )Ωkl Skl Hkl ,
where,
Ωkl ≡ (Ωrp δk≤2 + Ωpb δk≥3 )δl≥7 (10.157)
kl ≡ (ˆ
ˆ rp δk≤2 + ˆ
pb δk≥3 )δl≥7 .
for k ∈ G ∪E and l ∈ E andf ∈ G ∪ E but f 6= k, l. The projection onto the quantization axis
Hkl = Hkl |ˆˆkl
kl
| , B, ml − mk is given by Eq. (10.161).
The Eqs. (10.151)-(10.156) form together an algorithm to generate the matrix allowing one
to numerically solve the Bloch equations (10.150), as has been done in the main text.
10.5.4.2 Relative forces of oscillators
Spontaneous transitions between hyperfine- and Zeeman split levels have to be weighted ac-
cording to the Wigner-Eckardt theorem using Clebsch-Gordan (3j) and Wigner {6j} symbols.
Consider the transition |(Jk , I)Fk , mk i ↔ |(Jl , I)Fl , ml i. The relative oscillator strength is,
2
Fk κ Fl
Skl = (10.158)
mk sign (ml − mk ) −ml
2
Jl Jk κ (2Fk + 1)(2Jl + 1)(2κ + 1)
.
Fk Fl I 2I + 1
10.5.4.3 Elliptical laser polarization
The transition rates additionally depend on the relative orientation of the laser polarizations and
the magnetic field direction. This dependence is accounted for by decomposing the polarization
vector into the,
B ê3 × ĝ ê2 × ê3
ê3 = , ê2 = , ê1 = , (10.159)
B |ê3 × ĝ| |ê2 × ê3 |
where ĝ is an arbitrarily chosen direction, e.g. gravity. The relative amplitude of the transitions
∆mJ = 0 is proportional to the projection of the polarization vector on the magnetic field axis
ζ0 = (ε̂ · ê3 )2 for π-polarized light. To estimate the amplitude of the transitions ∆mJ = ±1, we
must project onto the coordinates,
ê± = √1 (ê1 ∓ iê2 ) , (10.160)

2
and we obtain ζ±1 = (ε̂ · ê± )2 for σ ± -polarized light. Hence,
Hkl = ζ∆mJ = ζml −mk . (10.161)
With this generalization the Bloch equations can e.g. be employed to calculate Hanle res-
onances quantum mechanically. The Hanle effect occurs when a magnetic and an optical field
compete for the quantization axis.
10.5.4.4 Hyperfine and Zeeman splitting
The nuclear spin of the 6 Li atom is I = 1, its electron spin is S = 21 . The excitation states
are characterized by quantum numbers Jk , Fk , mk . The electron angular orbital momentum is
Lk = δk≥7 , and the electron angular orbital momentum is Jk = 21 δk≤6 + 32 δk≥7 . The hyperfine
structure of the excited state 2 P3/2 can be written as νhf 1 = −2.8 MHz, νhf 2 = 0 MHz, and
νhf 3 = 1.7 MHz. Hence, the hyperfine splitting is inferior to the natural decay rate Γ =
(2π) 6 MHz,
µB |B|
Zkl = 2π~ (gFk mk − gFl ml ) (10.162)
+ νhf 1 δ7≤l≤8 + νhf 2 δ8≤l≤13 + νhf 3 δ13≤l≤16 ,
where gFk is the Landé factor of hyperfine level Fk .

10.6. EXERCISES 271
10.6 Exercises
10.6.1 Density matrix
10.6.1.1 Ex: Trace of an operator
P
The trace of an operator Â is defined by Tr Â = n hn|Â|ni.
a. Show that the trace is independent of the chosen basis!
b. Show that Tr ÂB̂ = Tr B̂ Â!
10.6.1.2 Ex: Pure states and mixtures

Consider a system of two levels coupled by a light mode. The Hamiltonian can be written
(~ ≡ 1),
0 Ω
Ĥ = .
Ω ω0
Calculate ρ̂, ρ̂2 and hĤi for the following two cases:
a. The atom is in a superposition state, |ψi = α|1i + β|2i e
b. The atom is a statistical mixture of eigenstates, ρ̂ = µ|1ih1| + ν|2ih2|.
10.6.1.3 Ex: Mixture of states

A two-level atom is initially in a superposition of two states |ψi = √12 |1i + √12 |2i. An apparatus
measures the populations of the states, but the experimenter forgot to read the indicated result.
a. Describes the state the atom by the density operator.
b. Now the experimenter returns to the device. Calculate with which probability he reads the
state |1i.
10.6.1.4 Ex: Thermal population of a harmonic oscillator

In thermal equilibrium the energy states of a system are populated following Boltzmann’s law,
e−nβ~ω 1
Pn = P −mβ~ω
with β≡ .
me kB T
Consider a one-dimensional harmonic oscillator characterized by the secular frequency ω and,

using the density operator, calculate the mean quantum number of the population and the mean
energy.
10.6.1.5 Ex: Thermal mixture

We consider a thermal non-interacting atomic gas in one dimension. Instead of describing the
state of the atomic ensemble, we can consider a single atom with a distributed probability of
having a given velocity v. The density operator of the continuous degree of freedom can be
written, Z r
m 2
ρ̂ = dv e−mv /2kB T |vihv| ,
2πkB T
and the trace of an arbitrary observable Â,
Z
hÂi = Tr ρ̂A = duhu|ρ̂Â|ui .
Now imagine a device capable of measuring the speed of a single atom randomly chosen within
the cloud.
a. Express the probability of measuring a specific velocity v 0 for this atom using the density
operator.
b. Express the expectation value of the average velocity by the density operator.
10.6.2 Bloch equations for two-level atoms

10.6.2.1 Ex: Derivation of Bloch equations
Derive the Bloch equations explicitly based on the temporal evolutions of the coefficients a1,2
(5.63) knowing that ρij = a∗i aj .
10.6.2.2 Ex: General solution of Bloch equations
Derive the solution (10.53) of the Bloch equations (10.50).
10.6.2.3 Ex: Expansion in Pauli matrices
Show explicitly Tr ρ̂σ̂ − σ̂ + = ρ11 .
10.6.2.4 Ex: Bloch vector and Bloch equations
Show that Eq. (10.64) is equivalent to the Bloch equations (10.50).
10.6.2.5 Ex: Normalization of the Bloch vector
Verify k~
ρk = 1.
10.6.2.6 Ex: Bloch equations, the Rabi method
Free atoms be illuminated by light pulses characterized by the Rabi frequency Ω, whose pulse
Rt
area is (i) Ω dt = π and (ii) = 2π. For which frequency tuning ∆ = ω − ω0 the excited state
0
population is maximum? Draw the spectral profile of the population in the range −5 < ∆/Ω < 5.
10.6.2.7 Ex: Sequence of Ramsey pulses
Many atomic clocks work according to the Ramsey spectroscopy method: The two-level atom
is resonantly excited by a microwave π/2-pulse. Then, the phase of atomic coherence precesses
freely over a period of time T accumulating an angle φ. Finally, a second π/2-pulse is applied
and the population of the upper-level is measured. Calculate this population as a function of
the angle φ. Neglect spontaneous emission.
10.6.2.8 Ex: Analytical treatment of the Ramsey experiment
Derive the analytic formula for the final population ρ22 for the Rabi and Ramsey experiments.
Derive and compare the line widths of the ’interference fringes’ in these two experiments.
10.6. EXERCISES 273
10.6.2.9 Ex: Atomic clocks by the Ramsey method with spontaneous emission
In this exercise we study the Ramsey method used in atomic clocks. For this, we will consider
a two-level system |1i and |2i excited by a microwave radiation field characterized by the Rabi
frequency Ω12 , and we will compare two cases: without and with spontaneous emission:
a. Write down the Hamiltonian of the system, propose a sequence of pulses allowing the observa-
tion of the Ramsey fringes, do a numerical simulation of the Schrödinger equation (based on the
prescription (10.138)), and prepare a graph of the type 10.1 illustrating the temporal evolution
of the Bloch vector during the sequence.
b. Calculate numerically from the Schrödinger equation the population ρ22 immediately after
the pulse sequence as a function of the detuning ∆12 of the radiation field, and prepare a graph
of the spectrum. Also, assuming a decay rate of Γ12 = 0.1Ω12 , calculate the population ρ22 as a
function of detuning ∆12 from the Bloch equations (making sequences of type (10.138)), prepare
a new graph, and compare it with the previous graph obtained by the Schrödinger equation.
c. What happens to the width of the fringes, when the free precession time τ between the Ram-
sey pulses is increased? Prepare a graph of the inversion 2ρ22 − 1 as a function of ∆12 and τ
and interprete the results.
10.6.2.10 Ex: Photon echo
’Photon echo’ is a powerful spectroscopic technique that allows circumvention of certain dephas-
ing processes, for example, the Doppler shift due to the atomic motion in a thermal sample of
atoms. The technique resembles the Ramsey method with the difference, that between the two
Ramsey π/2-pulses, that is, during the free precession time, we apply an additional π-pulse,
which inverts the imaginary part of the coherence. We will study this method by numerical
simulation of the Schrödinger equation and the Bloch equations for a two-level system with and
without spontaneous emission:
a. Write down the Hamiltonian of the system and do a numerical simulation of the Schrödinger
equation (concatenating the pulses as explained in Eq. (10.138)) for the following temporal pulse
sequence:
(i) resonant π/2-pulse (∆12 = 0) choosing Ω12 = 2,
(ii) evolution for a time T without radiation (Ω12 = 0),
(ii) resonant π/2-pulse using the same parameters as in (i),
(iv) evolution for a time T without radiation, and
(v) resonant π/2-pulse identical to the first pulse.
Prepare a graph of type 10.1 illustrating the temporal evolution of the Bloch vector during the
sequence. Now, repeat the sequence taking into account a possible Doppler shift leading to
∆12 6= 0.
b. Repeat the calculation of (a), now numerically solving the Bloch equations, which allow the
occurrence of spontaneous emission (Γ12 = 0.03Ω12 ). Interpret the results.
10.6.3 Bloch equations with spontaneous emission
10.6.3.1 Ex: Stationary solution of the Bloch equations
Derive the stationary solution of the Bloch equations including spontaneous emission.
10.6.3.2 Ex: Solution of the Bloch equations for resonant excitation
Derive the solution (10.78) of the Bloch equations with spontaneous emission for resonant exci-
tation.
10.6.3.3 Ex: Determinant of the Bloch matrix
In Sec. 10.1.1 we already saw that det ρ̂ = 0 for conservative systems. Now, show explicitly for
the Bloch matrix of a two-level system, that det ρ = 0 only holds in the absence of spontaneous
emission.
10.6.3.4 Ex: Density operator with dissipation
Discusses the phenomenon of dissipation in the example of

a. a thermal sample of two-level systems |ii = |1i, |2i characterized by the density operator
ρ = |iihi| ⊗ |vihv|, where |vi is the velocity state of the atom and
b. a two-level atom coupled to a radiation field, ρ = |iihi| ⊗ |nihn|, where |ni is the number of
photons inside the mode.
10.6.3.5 Ex: Bloch vector
A two-level atom with decay rate Γ = 2π × 6 MHz be excited by a light field detuned by ∆ = 2Γ
and whose intensity is a quarter of the saturation intensity. Write down the Bloch vector for
t → ∞.
10.6.3.6 Ex: Purity of two-level atoms with spontaneous emission
Calculate for a driven two-level atom in the stationary limit Tr ρ e Tr ρ2 .
10.6.3.7 Ex: Bloch sphere
~ ≡ (2Reσ+ , 2Imσ− , σz ),
Check the temporal evolution of the norm of the Bloch vector defined by ρ
where the σk are the Pauli matrices, for a resonantly excited two-level system with and without
spontaneous emission.
10.6.3.8 Ex: Atomic beam
An atomic beam is illuminated perpendicular to its propagation direction by (quasi-)monochromatic,

collimated laser pulses having the intensity I = 1 W/cm2 , the wavelength λ = 780 nm, and the
duration 200 ns. The laser is tuned to the center of an atomic resonance line (Γ/2π = 6 MHz).
a. How does the population of the upper atomic state develop?
b. How does the dynamics change, when the light is detuned by 100 MHz?
10.6.3.9 Ex: General form of the master equation
Show that the general form of the master equation: ρ̇ = − ~i [Ĥ, ρ] − Γ2 (2σρσ + − σ + σρ − ρσ + σ),
reproduces the Bloch equations including spontaneous emission.
10.6. EXERCISES 275
10.6.3.10 Ex: Quantum Zeno effect and saturation broadening

In this exercise we study saturation broadening effect in a three-level system |1i, |2i, and |3i in
V -configuration, as shown in Fig. 10.4(b), excited by two resonant lasers with the Rabi frequen-
cies Ω12 and Ω23 .
a. From the eigenvalues E1,2 of the effective Hamiltonian (10.88) of the system, describe the
behavior of the real part (energy shift) and the imaginary part (linewidth) as a function of the
Rabi frequency. Prepare diagrams Ω12 versus Re E1,2 and versus Im E1,2 and discuss the limits
Ω12 > 12 Γ12 e Ω12 < 21 Γ12 .
Saturation broadening can be measured experimentally in a three-level system in V -configuration.
To reproduce the experiment by numerical simulations of the Bloch equations (10.130),
b. write down the Liouville matrix L of the system and calculate the time evolution of the
Bloch vector via equation (10.134) varying the Rabi frequency Ω23 . Choosing the parameters
Γ23 = Γ12 , Γ13 = 0.001Γ12 , Ω12 = 0.2Γ12 , and ∆12 = 0 = ∆23 , prepare a 3D curve (similar to
Fig. 10.2(a)) of the population ρ33 (t).
c. Interpret the results in terms of broadening by saturation. The broadening can also be un-
derstood in terms of the quantum Zeno effect, where the transition |1i-|2i plays the role of
the ’observed system’ and the transition |2i-|3i the role of the measuring device or ’meter’ (for
example, we can observe the light scattered on the ’meter transition’ to infer the evolution of
the ’system transition’).
10.6.3.11 Ex: Saturation broadening and Autler-Townes splitting

In this exercise we study the Autler-Townes effect in a two-level system |1i and |2i resonantly
excited (∆12 = 0) by a laser with the Rabi frequency Ω12 :
a. From the eigenvalues E1,2 of the effective Hamiltonian (10.88) of the system, describe the
behavior of the real part (energy shift) and the imaginary part (linewidth) as a function of the
Rabi frequency. Prepare diagrams Ω12 versus Re E1,2 and versus Im E1,2 and discuss the limits
Ω12 > 12 Γ12 and Ω12 < 21 Γ12 .
The Autler-Townes effect can be measured experimentally by probing the population of level
|2i via excitation of a third (higher) level by a second laser with the Rabi frequency Ω23 . Thus,
we obtain a three-level system in cascade configuration, as shown in Fig. 10.4(c). In order to
reproduce the experiment by numerical simulations of the Bloch equations (10.130),
b. write down the Liouville matrix Lred reduced by the trace condition (10.135) and
c. compute the stationary Bloch vector from equation (10.137) varying the detuning of the
probe laser ∆23 and the Rabi frequency Ω12 of the system under study (|1i and |2i). Choosing
the parameters Γ23 = 0.5Γ12 , Γ13 = 0.01Γ12 , Ω23 = 0.1Γ12 , prepare a 3D curve (similar to
Fig. 10.2(a)) of the stationary population ρ22 (∞). Interpret the results.
10.6.3.12 Ex: Light-shift

In this exercise we study the effect of the dynamic Stark shift (or light shift) of the energy levels
of a two-level system |1i and |2i excited by a laser with the Rabi frequency Ω12 and the detuning
∆12 :
a. From the eigenvalues E1,2 of the effective Hamiltonian (10.88) system, find approximations
for weak coupling (Ω12 Γ12 ) and strong coupling (Ω12 Γ12 ). Prepare a graph showing the
eigenvalue spectrum (separating the parts Re E1,2 and Im E1,2 ) as a function of detuning ∆12 for
various values of Ω12 . Also search for approximations valid for large detunings ∆12 Γ12 , Ω12
and add them to the graph.
The light shift can be experimentally measured in a three-level system in Λ-configuration, as

illustrated in Fig. 10.4(a). To reproduce the experiment by numerical simulations of the Bloch
equations (10.130),
b. write the Liouville matrix Lred reduced by the condition to the trace (10.135) and calculate
the stationary Bloch vector from equation (10.137) varying the detunings of the two lasers ∆12
and ∆23 . Choosing the parameters Γ23 = Γ12 , Γ13 = 0.01Γ12 , Ω12 = 2Γ12 , and Ω23 = 0.2Γ12 ,
prepare a 3D curve (similar to Fig. 10.2(a)) of the stationary population ρ22 (∞). Interpret the
results.
10.6.4 Line broadening mechanisms

10.6.4.1 Ex: Optical density of a cold cloud
The cross section of an atom with the resonant frequency ω0 moving with velocity v and irradi-
ated by a laser beam of frequency ω is,
6π Γ2
σ(v) = .
k 2 4(ω − ω0 − kv)2 + Γ2
The normalized one-dimensional Maxwell distribution,
r
m 2
ρ(v)dv = e−mv /2kB T dv .
2πkB T
a. Calculate the absorption profile of the resonance line at 461 nm (Γ461 = (2π) 30.5 MHz) of a
strontium gas cooled to the Doppler limit (kB TD = ~Γ) of this transition.
b. Calculate the absorption profile of the resonance line at 689 nm (Γ689 = (2π) 7.6 kHz) of a
strontium gas cooled to the Doppler limit of the transition at 461 nm.
c. Compare the optical densities in case of resonance.
Help: To evaluate the convolution integral approximate the narrower distribution by a δ-
function maintaining the integral over the distribution normalized.
10.6.4.2 Ex: Saturated absorption spectroscopy

Saturated absorption spectroscopy is a technique to avoid Doppler enlargement. The diagram,
shown in Fig. 10.6, consists of a cell filled with a rubidium gas (resonance frequency ω0 = ck =
2πc/780 nm, decay rate Γ = (2π) 6 MHz) and two laser beams with the same frequency ω but
counterpropagating, one called saturation and another called proof. The one-dimensional and
normalized Maxwell velocity distribution is,
r
m 2
ρ(v)dv = e−mv /2kB T dv .
2πkB T
The gas is at T = 300 K, where the partial pressure of rubidium is around P = 10−1 mbar.
The length of the cell is L = 10 cm. The laser has an intensity below the saturation limit, such
that the cross section of an atom moving at velocity v is,
6π Γ2
σ(v) = .
k 2 4(ω − ω0 − kv)2 + Γ2
The saturation laser has high intensity. We suppose here, Ω ≡ 10Γ, where Ω is the frequency of
Rabi caused by the saturation beam. In this way it creates a population Na nd of atoms in the
10.6. EXERCISES 277
excited state. As this population lacks in the ground state, Ng = N − Ne , the absorption of the
proof beam is decreased by the factor,
Ne Ω2
= .
N 4(ω − ω0 + kv)2 + 2Ω2 + Γ2
R∞ Ng −Ne
Calculate for laser proof spectrum of optical density, OD(ω) = Ln −∞ N σ(v)ρ(v)dv, and
the intensity of light transmitted through the cell, II0 = e−OD .
Figure 10.6: Scheme of saturation spectroscopy.
Paulo Cesar Ventura da Silva

10.6.4.3 Ex: The Zeeman slower
A necessidade
Consider a tube do desacelerador
through which passes Zeeman (2,5 pontos)
a collimated beam of atoms, all having the same ini-
tial velocity v = v0 . In the opposite direction to the atomic motion travels a collimated and
Considere um tubo por onde passa um feixe colimado de átomos, todos inicialmente com velocidade
monochromatic light beam with frequency ω = kc. The absorption rate for photons by an atom
𝑣 = 𝑣 . No sentido contrário ao do movimento dos átomos, incide um feixe luminoso, colimado e
has a 0Lorentzian profile, which can be written as:
monocromático, com frequência 𝜔 = 𝑘𝑐. Conforme foi estudado, a taxa de absorção de fótons por um
átomo é uma lorentziana, podendo ser escritaW0 como: Γ2
W (v) = ,
2π (ω − ω0 + kv)2 + (Γ/2)2
2
𝑊0 Γ
where Γ is the natural width 𝑊(𝑣) of the=spectral line at
2𝜋 (𝜔 − 𝜔0 + 𝑘𝑣) ω 0 , and W⁄0 2is)2a constant. The frequency of
2 + (Γ
the light is tuned in order to compensate for the Doppler effect at the beginning of the tube,
δOnde Γ é 0a =
= ω −ω −kv0natural
largura (the light is tuned
da linha to the
espectral emred𝜔0of 𝑊0 éresonance).
, ethe uma constante.As the atoms area frequência
Sintoniza-se decelerated,
they cease to be resonant with the light beam and fail to absorb photons.
da luz de modo a compensar o efeito Doppler no início do tubo, ou seja, 𝛿 = 𝜔 − 𝜔0 = −𝑘𝑣 This can be 0avoided
(a luz
by
é dessintonizada para o vermelho da ressonância). Conforme os átomos são desacelerados, eles using
employing the so-called Zeeman-slowing technique, which compensates for the effect
the Zeeman-shift
deixam de estar eminduced by magnetic
ressonância fields.
com o feixe In thisdeixando
luminoso, exercise,de weabsorver
will study what
fótons. happens
Isso poderiaifser
this
technique is not used.
evitado com a técnica de resfriamento Zeeman, que compensa o efeito com campos magnéticos. Aqui,
a. For an atom with velocity v, write an expression for the mean travel distance ∆s(v) before
vamos ver o que acontece caso essa técnica não seja utilizada.
a) Para um átomo com velocidade 𝑣, escreva

Figure uma expressão
10.7: Zeeman slowerpara a distância média Δ𝑠(𝑣) que ele
scheme.
percorre até absorver um fóton, em função dos parâmetros Γ, 𝑣0 , 𝑘 e 𝑊0. (O tempo médio que ele leva
itpara
absorbs a photon
absorver um fóton ⁄𝑊(𝑣)). of
as éa1function (0,5the parameters Γ, v0 , k, and W0 . (The mean time it takes
pontos)
to absorb a photon is W (v)−1 ).
b.
b) The velocity do
A velocidade of the
átomoatom as a function
em função do número 𝑛 de
of the number of absorbed
fótons absorvidos é 𝑣photons
𝑛 = 𝑣0 −is𝑛Δ𝑣, v0 − n ~k
vn =onde m,
Δ𝑣 =
the ℏ𝑘⁄𝑚term
second beingdevido
é o recuo the recoil due to
à absorção de the absorption
um fóton. of a single
A distância photon.
total média The average
percorrida por um total
distance
átomo apóstraveled by Nanfótons
absorver atomé after absorbing
estimada por: N photons is estimated by:
XN 𝑁 Z N
𝑁
S= ∆s(vn ) ' ∆s(v )dn .
𝑆 = ∑ Δ𝑠(𝑣𝑛 ) ≈ 0∫ Δ𝑠(𝑣𝑛n) d𝑛
n=0
𝑛=0 0
Calcule a distância média necessária para que os átomos sejam freados até 𝑣 = 0 (ignore o limite
Doppler). Deixe em função de Γ, 𝑣0 , Δ𝑣, 𝑘 e 𝑊0. Dica: faça a mudança de variável 𝑛 → 𝑣 na integral,
isso pode economizar alguns cálculos. (1,5 pontos)
c) Tipicamente, a dessintonia da luz |𝛿| = 𝑘𝑣0 é bem maior do que a largura natural Γ da transição. O
que acontece com 𝑆 no limite em que 𝑘𝑣0 ≫ Γ? Interprete esse resultado, justificando a necessidade
Calculate the average distance required for the atoms to be slowed down to v = 0 (ignoring the
Doppler limit). Write the expression as a function of Γ, v0 , k, and W0 . Help: Do the following
change of variables to simplify the evaluation of the integral: n → v.
c. Typically, the detuning of the light, |δ| = kv0 , is much larger than the natural width Γ of the
transition. What happens to S in the limit when kv0 Γ? Interpret this result, justifying the
need for the Zeeman-slowing technique.
10.6.4.4 Ex: Saturation intensity

Calculate the saturation intensity for the sodium transition 3s 2 S1/2 , F = 2 ←→ 3p 2 P3/2 , F 0 = 3.
The natural width of the transition is Γ/2π = 9.89 MHz and the wavelength λ = 590 nm.
10.6.4.5 Ex: Pressure broadening

At what pressure the collision broadening [given by the expression (10.96)] between sodium
atoms in the ground state dominates the width of the D2-transition at ambient temperature.
The natural width of the D2-line is Γ/2π = 6 MHz.
10.6.4.6 Ex: Collision broadening

A simple collision broadening model for a resonance line describes the impact of collisions (av-
erage collision rate τ0 ) on the light wave emitted through random phase jumps happening at
irregular time steps. Calculate the spectral distribution of the light intensity.
Help: Calculate the Fourier transform of the wave trains and weight the spectral distribution
of the intensity with the probability density of encountering the collision time τ in the atomic
sample.
10.6.4.7 Ex: Optical density of a hot cloud

Calculate and draw the effective Lorentz profile, Gauss profile and Voigt profile for the resonance
line at 461 nm (Γ = (2π) 32 MHz) of a strontium gas heated to the temperature 400 C and the
pressure P = 10−4 mbar inside a 15 cm long cell.
10.6.4.8 Ex: Rate equations as a limiting case of Bloch equations

We show in this exercise that, in the limit Γ Ω, we can derive, from the Bloch equations, the
Einstein rate equations. Proceed as follows:
a. Apply the condition ρ̇12 = 0 to the Bloch equations for a two-level system (10.73), determine
ρ12 (∞), and replace this stationary value in the equations for the populations ρkk (t) using, as
an abbreviation, the transition rate R ≡ γs, where s is the saturation parameter (10.75).
b. Integrate the rate equations over the entire spectrum, i.e. ∆ ∈ [−∞, ∞], and derive Einstein’s
equations using the relations (??), (??), and (??).
10.6.5 Multi-level systems

10.6.5.1 Ex: Bloch equations for three levels
An excited Λ-shaped atom consists of two ground states |1i and |3i, which are coupled by two
lasers with Rabi frequencies and detunings Ω12 and ∆12 respectively Ω23 and ∆23 through an
excited level |2i. Derive the Bloch equations from this system from the general master equation.
10.6. EXERCISES 279
10.6.5.2 Ex: EIT & dark resonances

In this exercise we study so-called dark resonances, which are responsible for the phenomenon
of electromagnetically induced transparency (EIT). Such resonances are observed in three-level
systems |1i-|2i-|3i in Λ-configuration, as shown in Fig. 10.4(a), when the laser detunings are
chosen so as to satisfy ∆12 = ∆23 .
a. From the Bloch equations (10.130) show analytically that, in a stationary situation, the
population of the excited state is ρ22 (∞) = 0 in the center of the dark resonance.
Dark resonances can be observed experimentally. To reproduce the experiment by numerical
simulations of the Bloch equations (10.130), write down the Liouville matrix Lred reduced by the
trace condition (10.134) and calculate the stationary Bloch vector from equation (10.135) varying
the detunings of the two lasers ∆12 and ∆23 . Choosing the parameters such that Γ23 = Γ12 ,
Γ13 = 0.01Γ12 , Ω12 = 2Γ12 , and Ω23 = 0.2Γ12 , prepare a 3D curve [similar to Fig. 10.2(a)] of the
population ρ22 (∞). Interpret the results.
10.6.5.3 Ex: STIRAP

In experiments with cold atoms it is often necessary to transfer populations between ground
states, for example, between specific levels of a hyperfine structure. One possible procedure is the
method of optical pumping, from the initial ground state to an excited state, which subsequently
decays to the final state by spontaneous emission. The problem with this incoherent procedure
is, that one can control into which ground state level the atom will decay, and that it heats atoms
due to the photonic recoil associated with the scattering of light. In this exercise we studied an
alternative method, called Stimulated Raman Adiabatic Passage (STIRAP), which allows the
coherent transfer of population between two states by counter-intuitive pulse sequences:
a. Consider a three-level system in Λ-configuration, as shown in Fig. 10.4(a), initially being in
the state |1i. Write the system’s Hamiltonian in the interaction picture. Now, choose ∆12 =
0 = ∆23 , and a temporal variation of the Rabi frequencies described by Ω12 (t) = Γ12 ( 12 +
1 1 1
π arctan Γ12 t) and Ω23 (t) = Γ12 ( 2 − π arctan Γ12 t). With this, solve the Schrödinger equation
(10.138) iteratively within the time interval t ∈ [−40/Γ12 , 40/Γ12 ], while continuously adjusting
the Rabi frequencies.
b. The dynamics can also be calculated via a numerical simulations of the Bloch equations
(10.130). Write down the Liouville matrix and prepare a simulation using the same parameters
as in (b) and additionally Γ23 = Γ12 /2, Γ13 = Γ12 /500.
c. Interpret the results.
10.6.5.4 Ex: Adiabatic sweeps

In experiments with cold atoms it is often necessary to transfer populations between ground
states, for example, between specific levels of a Zeeman structure. One possible procedure is the
method of optical pumping, from the initial ground state to an excited state, which subsequently
decays to the final state by spontaneous emission. The problem with this incoherent procedure
is, that one can control into which ground state level the atom will decay, and that it heats atoms
due to the photonic recoil associated with the scattering of light. In this exercise we study an
alternative method, called adiabatic sweep, which allows the coherent transfer of population
between the two outer states of a degenerate multiplet, as shown in Fig. 10.8, via an adiabatic
ramp of the frequency of the incident radiation:
a. Write down the Hamiltonian of the system in the interaction picture. Now, choose Ω/2π =
8 kHz and apply a linear ramp of the radiation detuning between −50 kHz < ∆(t)/2π < 50 kHz
during a time interval of 2 ms. With this, solve the Schrödinger equation (10.130) iteratively
varying the detuning.
b. Write down the Liouville matrix of the system and do a numerical simulation of the Bloch
equations (10.130) using the same parameters as in (a). Interpret the results. What you observe
when you introduce a decay rate between adjacent levels of Γ/2π = 200 Hz?
Figure 10.8: Energy levels of an atom in the ground state with Zeeman structure (for example,
|J = 1, mJ = −1, 0, +1i) as a function of the applied magnetic field.
10.6.5.5 Ex: Dispersive interaction between an atom and light

Radiation which is tuned far from a resonance can change the phase of an atomic dipole moment
without changing the populations 8 . We study this effect in a three-level system in cascade config-
uration excited by two radiation fields, as illustrated in Fig. 10.4(c), simulating the Schrödinger
equation and the Bloch equations.
a. Write down the Hamiltonian Ĥ for this system letting ∆12 = 0.
b. Now, consider the subsystem |2i-|3i, write down its Hamiltonian Ĥ23 , determine the eigenval-
ues, and assume that this transition be excited very far-off resonance. That is, for ∆23 Ω23 , Γ23
expand the eigenvalues of Ĥ23 up to second order in Ω23 . Finally, replace the submatrix Ĥ23 in
the complete Hamiltonian Ĥ by the matrix of the expanded eigenvalues. This procedure corre-
sponds to treating the transition |2i-|3i as a perturbation of the transition |1i-|2i until second
order.
c. Assume that the atom is initially in the ground state and compute the time evolution of the
state via the Schrödinger equation (10.138) using (a) the perturbed Hamiltonian and (b) the
exact Hamiltonian for the following sequence of pulses:
(i) a π/2-pulse on the transition |1i-|2i,
(ii) a pulse with a variable duration between 0 and ∆t = Ω223 /4π∆23 applied to the transition
|2i-|3i,
(iii) a π/2-pulse on the transition |1i-|2i. What you observe?
d. Establish the Liouville matrix L for the same system and calculate the time evolution of the
Bloch vector during the sequence by the Bloch equations (10.138) choosing the same parameters
as in (c) and additionally Γ23 = 1, Γ13 = Γ23 , Γ12 = 0.01Γ23 , and Ω12 ≫ ∆23 , Γ23 . Prepare a
3D curve [similar to Fig. 10.2(b)] of the population ρ22 (t). Interpret the results.
10.6.5.6 Ex: Fano profile of in dark resonance

The dark resonance studied in Exc. 10.6.5.2 may in some circumstances adopt an asymmetric
profile. Calculate, for a three-level system in Λ-configuration, as shown in Fig. 10.4(a), starting
8
This type of interaction is used in the implementation of quantum gates in quantum computing.
10.6. EXERCISES 281
from the Bloch equations (10.130) with the Liouville matrix Lred reduced by the trace condi-
tion(10.134), the spectrum ρ22 (∆2 3) for the following set of parameters: Γ12 = 2, Γ23 = Γ12 /2,
Γ23 = 0.1Γ12 , Ω12 = 10Γ12 , Ω23 = 5Γ23 , ∆12 = −5Γ12 e ∆23 = [−1 : .01 : 1]Γ23 . Interpret the
spectrum in terms of a Fano resonance.
10.6.5.7 Ex: Gas with negative permittivity
10.6.5.8 Ex: Gas with negative permeability

Theoretically, under certain conditions, gases may exhibit negative permittivity and permeabil-
ity, and therefore refraction [229]DOI , [227]DOI , [228]DOI , [197]DOI , [156]. To study this phe-
nomenon we consider a three-level system in Λ-configuration with an electric dipole transition
and another magnetic dipole transition. The objective is to balance the electrical dipole moment
excited by a probe laser and the magnetic dipole moment excited via a Raman transition by
both, the probe laser and a control laser. The Raman transition simulates an effective magnetic
field. Since the magnetic moment is smaller by a factor of α2 , the electric moment must be
reduced by detuning the probe laser, as shown in Fig. 10.9.
Figure 10.9
a. Consider a three-level system in Λ-configuration. The transitions |1i-|2i and |2i-|3i are as-
sumed to be electric dipoles and |1i-|3i a magnetic dipole, such that, Γ12 , Γ23 Γ13 . Extract
from the Bloch equation (10.130) the equations for the coherences ρ12 , ρ13 , and ρ23 .
b. Suppose, that the excitation on the probe transition be so weak, Ω12 Γ12 , that it does not
succeed to empty the ground state. In this approximation eliminate the dynamics of ρ23 and
deduce the stationary solution for ρ12 and ρ13 .
c. Calculate the magnetic susceptibility χm = [228] with the following parameters Γ12 =
7 · 107 s−1 , Γ23 = 3 · 107 s−1 , Γ13 = 2 · 107 s−1 , Ω12 = 0.1Γ12 , Ω23 = 2 · 108 s−1 , ∆23 = 0
in the regime ∆12 = [−15Γ23 , 15Γ23 ].
d. Simulate the Bloch equations (10.130) and compare with the numerical solution.
Chapter 11
Atoms in quantized radiation fields

Até agora só tratamos o campo óptico como uma onda clássica estacionário ou propagante,
enquanto o nosso átomo de dois nı́veis fui considerado como uma entidade sujeita às leis da
mecânica quântica, sujeito à perturbação induzida pelas ondas oscilatórias. Esse procedimento
conduz naturalmente para oscilações das populações e coerências entre os estados do átomo.
No entanto, para problemas de campos fortes envolvendo um espectro de energia atômica sig-
nificativamente modificado, uma abordagem não-perturbativa, independente do tempo pode ser
frutı́fera. Uma solução independente do tempo para a equação de Schrödinger com átomos e
campos acoplados é chamada estado vestido. Ela fui usada a primeira vez para interpretar o
’desdobramento’ dos espectros de rotação molecular na presença de campos intensos clássicos de
radiofrequência. O tratamento semiclássico é adequado para uma ampla variedade de fenômenos
e tem a virtude de simplicidade matemática e familiaridade. No entanto, às vezes vale a pena
considerar o campo como uma entidade quântica também. A interação átomo-campo torna-se
então um intercâmbio de quantas do campo, os fótons, com o átomo. Esta abordagem nos per-
mite expressar o estado de número de fótons, também chamado de estado de Fock, e o estado
discreto do átomo em pés iguais e de escrever as funções de estado do sistema átomo-campo em
uma base de produto de estados fotônicos e atômicos. Diagonalização dos termos de acoplamento
dipolar no hamiltoniano do sistema também dá origem a soluções independentes do tempo, de
estados vestidos da equação de Schrödinger completamente quântica.
Começamos este capı́tulo com a quantização do campo de luz e depois exprimimos a interação
átomo-campo numa forma totalmente quantificada. Examinaremos alguns exemplos ilustrando
como a imagem de estados vestidos pode fornecer informações úteis sobre interações luz-matéria.
Finalmente, mostraremos como estados vestidos semiclássicos também podem ser obtidos.
11.1 Quantization of the electromagnetic field

Já vimos que a energia de um campo de luz monocromático de frequência ω é quantizado em
pequenas porções iguais, tal que a energia total é n~ω, onde n é um número interno. O espectro
de energia é o mesmo como aquele do oscilador harmônico. Por isso, podemos identificar um
modo luminoso com um oscilador e adotar todo o formalismo desenvolvido para o oscilador
harmônico. O formalismo será assumido como conhecido no seguinte.
11.1.1 Field operators

A ideia essencial por trás da quantização do campo é a substituição dos osciladores harmônicos
clássicos discutido na Seç. 3.4 por osciladores quânticos. A fim de realizar esta quantização
da forma mais simples, introduzimos o potencial escalar Φ e potencial vetor A como feito em
283
284 CHAPTER 11. ATOMS IN QUANTIZED RADIATION FIELDS
teoria eletrodinâmica 1 . No espaço livre, onde não há cargas nem correntes, e dentro do calibre
de Coulomb temos a solução da equação de onda (??) generalizada para uma distribuição de
vetores de onda k 2 ,
X
A(r, t) = ~k [A+
0k e
i(k·r−ωk t)
+ A−
0k e
−i(k·r−ωk t)
], (11.1)
k
onde já isolamos o caráter vetorial devido à polarização ~k do modo k da luz. Obviamente,
A− + ∗
0k = (A0k ) . Como cada amplitude e a polarização da onda do potencial vetorial A0k e A0k
∗
devem satisfazer a equação de onda separadamente, chegamos à relação de dispersão,

ωk = ck . (11.2)
Com os resultados (??) e (??) sabemos, que a energia em cada modo radiativo é,
Ek = ~ωk Nk = uk V = 2ε0 V ωk2 |A0k |2 = 2ε0 V ωk2 (A− + + −
0k A0k + A0k A0k ) . (11.3)
A segunda quantização agora consiste em interpretar o modo como oscilador harmônico quântico,
isto é, entendemos as observáveis como operadores e permitimos flutuações quânticas:
Ĥk = ~ωk (N̂k + 21 ) = 2ε0 V ωk2 (Â− + + −
0k Â0k + Â0k Â0k ) . (11.4)
Introduzindo operadores de campo normalizados,
q q
âk 4ε0 V~ ωk ≡ Â+
0k e â †
k
~ −
4ε0 V ωk ≡ Â0k , (11.5)
tal que,
Ĥk = ~ωk (â†k âk + 12 ) . (11.6)
A analogia nós permite interpretar eles como operadores de criação e de aniquilação. Finalmente,
podemos reescrever (11.1) como,
q h i
Âk (r, t) = 4ε0 V~ ωk ~k âk ei(k·r−ωk t) + â†k e−i(k·r−ωk t) . (11.7)
Tais combinações de operadores e seus conjugadas complexas já conhecemos do OH (3.66).

Os operadores do campo elétrico e magnético para os modos da cavidade podem ser con-
struı́dos à partir de,
q
~ωk i(k·r−ωk t) − â† e−i(k·r−ωk t) ~
Êk = − ∂∂tÂk
= i 2 0V
âk e k k
q . (11.8)
~ωk i(k·r−ωk t) − â† e−i(k·r−ωk t) k × ~
B̂k = ∇ × Ak = i 2ε0V
âk e k k
Podemos calcular o médio sobre um perı́odo da energia do modo k-th na cavidade por uma
versão quantizada da Eq. (11.9),
Z
ε0
Ēk = 2 hnk |Êk · Êk |nk idV . (11.9)
O resultado (11.6) é exatamente o que Planck havia sugerido (embora estritamente falando,
sua sugestão foi a quantização dos osciladores nas paredes condutantes da cavidade, e não
do campo), para explicar a distribuição da intensidade espectral radiada por um corpo negro.
Vemos agora que segue naturalmente da quantização dos modos de campo na cavidade. Resolva
o Exc. 11.8.1.1.
1
Vide a apostila Eletrodinâmica do mesmo autor [55].
2
A interação átomo-luz pode depender da polarização da luz à respeito do eixo de quantização do átomo
definido, e.g. por um campo magnético. Nestes casos precisamos estender o ı́ndice k para incluir o estado de
polarização (k, λ).
11.1. QUANTIZATION OF THE ELECTROMAGNETIC FIELD 285
11.1.2 Interaction of quantized fields with atoms

Agora que temos uma imagem clara do campo quantizado com as energias dos modos dadas
pela Eq. (11.9) e estados de número de fótons dada pelos auto-estados do oscilador harmônico
quantizado, |ni, estamos em posição de considerar o nosso átomo de dois nı́veis interagindo com
este campo de radiação quantizado. Se excluirmos emissão espontânea e processos estimulados
para o momento, o hamiltoniano do sistema combinado átomo e campo é,
Ĥ = Ĥatom + Ĥf ield + Ĥatom−f ield . (11.10)
Descrevemos o átomo por um sistema de dois nı́veis,

−~ω0 ~ω0
Ĥatom = 2 |gihg| + 2 |eihe| , (11.11)
Ĥf ield é o hamiltoniano do campo quantizado, expresso pela Eq. (11.6), e Ĥatom−f ield o a in-
teração átomo-campo. Para o hamiltoniano sem interação, Ĥ = Ĥatom + Ĥf ield , os auto-estados
são simplesmente estados de produto dos estados atômicos e dos estados de número de fótons,
|g, ni = |gi|ni e |e, ni = |ei|ni . (11.12)
O lado esquerdo da Fig. 11.1 mostra como as auto-energias dos estados produtos consistem de
duas escadas deslocadas pela energia de dessintonização ~∆. Escrevemos o operador hamilto-
niano do átomo Eq. (11.11) como soma de projetores sobre os auto-estados não perturbados
utilizando a relação de completeza e a ortogonalidade dos auto-estados. Com a mesma ideia
podemos reescrever o operador dipolar definido na Eq. (??),
X X X
d̂ = |iihi|d| |jihj| = dij |iihj| . (11.13)
i j i,j
Note-se que d só tem elementos não-diagonais.
Figure 11.1: Esquerda: Estados de número de fótons e os dois estados estacionários do átomo de
dois nı́veis. Meio: Dobre escada mostrando a base de estados produtos de estados de número
e estados atômicos. Direita: Estados vestidos construı́dos por diagonalização do hamiltoniano
completo na base de estados produtos.
Agora vamos usar as Eqs. (11.8) juntos para descrever a interação átomo-campo através do
hamiltoniano Ĥatom−f ield = −d̂ · Ê,
r h i
XX ~ωk
Ĥatom−f ield = i dij · ~k âk ei(k·r−ωk t) − â†k e−i(k·r−ωk t) |iihj| . (11.14)
2ε0 V
k i,j
Para o nosso átomo de dois nı́veis interagindo com um único modo do campo apenas temos,
r h i
~ωk
Ĥatom−f ield = i dge · ~k âk ei(k·r−ωk t) − â†k e−i(k·r−ωk t) (|gihe| + |eihg|) . (11.15)
2ε0 V
11.1.2.1 Rotating wave approximation for dressed states

Esse hamiltoniano contem quatro termos descrevendo os seguintes processos,
|e, ni −→ |g, n − 1i o átomo e deexcitado com a absorção de um fóton;
|g, ni −→ |e, n − 1i o átomo e excitado com a absorção de um fóton;
|e, ni −→ |g, n + 1i o átomo e deexcitado com a emissão de um fóton;
|g, ni −→ |e, n + 1i o átomo e excitado com a emissão de um fóton.

Obviamente, só o segundo e o terceiro termo respeitam conservação de energia (em processos
de primeira ordem) e podem servir como estados iniciais e finais num processo fı́sico real. A
Fig. 11.2 mostra esquemas desses quatro termos. Centrando-nós no segundo e terceiro termos,
podemos simplificar a notação. Identificando σ̂ + = |eihg| e σ̂ − = |gihe| das Eqs. (10.59) e
introduzindo a abreviação da frequência de Rabi,
r
~ωk
1
2 ~Ω(r) ≡ dge · ~k eik·r , (11.16)
2ε0 V
o hamiltoniano de interação fica,
Ĥatom−f ield = 2i ~Ω(r)σ̂ + âk e−iωk t − 2i ~Ω∗ (r)σ̂ − â†k eiωk t . (11.17)
Avaliando os elementos da matriz Ĥatom−f ield para o estado átomo-luz geral,
|ψi = N1 |g, n − 1ie−iωg t + N2 |g, nie−iωg t + N3 |e, n − 1ie−iωe t + N4 |e, nie−iωe t , (11.18)
P √ P √
utilizando as notações â† = n n + 1|n + 1ihn| e â = n n|n − 1ihn| e a ortogonalidade
hj, m|i, ni = δij δmn , vemos que (considerando por simplicidade um átomo localizado na origem),
hψ|Ĥatom−f ield |ψi = 2i ~Ω(0)× (11.19)

h i
× N1∗ N4 e−i(ωk +ω0 )t + N3∗ N2 e−i(ωk −ω0 )t − N2∗ N3 ei(ωk −ω0 )t − N4∗ N1 ei(ωk +ω0 )t ,
onde ω0 ≡ ωe − ωg . Vemos, que negligenciar o primeiro e quarto processo (ou seja, os ter-
mos ∝ σ̂ ± â± do hamiltoniano) é equivalente a fazer a aproximação da onda rotatória (RWA),
onde desprezamos os termos girando com a frequência ωk + ω0 , e que realmente só precisamos
considerar o acoplamento entre os dois estados vestidos |g, ni e |e, n − 1i.
É importante ressaltar, que o primeiro e quarto termo podem ser usados para alguns estados
intermediários em processos de ordem superior, tais como absorção multifotônica ou processos de
espalhamento Raman. De fato, quando a frequência de Rabi é muito grande, Ω ' ω, os processos
de excitação e deexcitação se sigam tão rapidamente, que a conservação de energia pode ser
violada para tempos curtos. Deslocamentos energéticos devido às contribuições desprezadas na
RWA se chamam deslocamento de Bloch-Siegert 3 .
3
O deslocamento não é observado, quando os termos não-rotativos σ ± a± são proibidos por outras regras de
conservação ou de seleção. Por exemplo, quando uma ressonância é excitada por luz σ ± , a RWA fica exata.
11.2. QUANTUM CORRELATIONS IN THE JAYNES-CUMMINGS MODEL 287
11.1.3 Dressed states

Dentro da nova base dos estados vestidos, o problema se reduz a diagonalização do hamiltoniano
de um átomo de dois nı́veis quase-degenerados (∆ ω0 ), no qual a amplitude dos elementos
não diagonais é o dada por 21 ~Ω. As auto-energias do hamiltoniano completo Ĥ são,
E± = ~2 (ωg,n + ωe,n−1 ) ± ~2 G . (11.20)
onde ~ωg,n e ~ωe,n−1 são as energias dos estados produto ~ωg + n~ωk e ~ωe + (n − 1)~ωk .
Figure 11.2: Quatro termos na interação átomo-campo. Os termos (b) e (c) conservam a energia
em processos de primeira ordem, enquanto (a) e (d) não conservam.
Os estados de produto átomo-campo oferecem um conjunto natural de uma base para o

hamiltoniano da Eq. (11.10). Os estados resultantes da diagonalização do hamiltoniano nessa
base são chamados estados vestidos. Como indicado na Fig. 11.1, o dubletos vizinhos da dupla
escada se ’repelem’ sob a influência da interação Ĥatom−f ield na Eq. (11.10). Os coeficientes
misturados formam o problema familiar de dois nı́veis, agora chamados de |ai e |bi. Da Fig. 11.1
vemos,
|a, N i = cos θ|g, ni + sin θ|e, n − 1i

. (11.21)
|b, N i = cos θ|e, n − 1i − sin θ|g, ni
com
Ω
tan 2θ = ∆ , (11.22)
√
onde a separação entre membros do mesmo estado vestido é G = Ω2 + ∆2 . Os números n
denotam a quantidade de fótons dentro do feixe laser, os números N a quantidades de pacotes
de energia dentro do sistema, isto é, os fótons mais a excitação possı́vel do átomo.
11.2 Quantum correlations in the Jaynes-Cummings model

O modelo de Jaynes-Cummings descreve a dinâmica de um único átomo de dois nı́veis vestido
de um único modo de laser monocromático na ausência de processos de emissão espontânea. O
modelo se tornou um modelo paradigmático da mecânica quântica com aplicações em informação
quântica, pois ele permite protocolos de emaranhamento de estados atômicos e a implementação
de portas quânticas, que serão discutidas nas Secs. ??. Estudaremos no seguinte primeiramente
a interação de um átomo com um modo óptico desprezando efeitos de dissipação. Depois,
discutiremos o impacto de processos de dissipação.
A evolução dinâmica de estados puros então segue a equação de Schrödinger. O hamiltoniano
deste sistema é dado por (11.17). Deixando ~ = 1 e assumindo, que o átomo seja na origem (tal
que Ω(r)eik·r = Ω(0)), podemos escrever o hamiltoniano como,
Ĥ = ωâ† â + ω0 (σ̂ + σ̂ − − 21 ) + 12 Ω(âσ̂ + + â† σ̂ − ) , (11.23)

onde ω é frequência da luz, ω0 a frequência da transição atômica e Ω a frequência de Rabi gerada

por um único fóton. Escolhemos a representação de Fock para o modo luminoso, representa-
mos as transições atômicas pelas matrizes de Pauli e abrimos o espaço produto ρf ield ⊗ ρatom
generalizando os operadores â± y â± ⊗ I e σ ± y I ⊗ σ̂ ± . Explicitamente, temos,

†
P √ 1 0 +
P 0 0
â = n n + 1|n + 1i 0 1 hn| e σ̂ = n |ni 1 0 hn|
(11.24)
P √ 1 0 P 0 1
â = n n|n − 1i 0 1 hn| e σ̂ − = n |ni 0 0 hn| .
O hamiltoniano Ĥ pode ser decomposto em sub-hiper-espaços de n fótons Ĥn :

X (n − 1)ω − ω0 0

0 0
√
0 Ω2 n + 1

Ĥ = |ni 2 + |n − 1i Ω √ + |n + 1i hn|
0 nω + ω20 2 n 0 0 0
n
 ω0 
−2
 ω0 Ω 
 2 2 
 Ω
ω− 2 ω0 
 2 √ 
 ω0
ω +√ 2 Ω 
= 2 2  (11.25)
 Ω
2 2ω − ω0 
 2 2 
 2ω + ω20 · · ·
 
.. ..
. .

M (n − 1)ω + 0 ω Ω √ M
2 n
Ω√
= 2 = Ĥn .
ω0
2 n nω − 2
n n
O operador densidade do subespaço é,

|n − 1i|2ih2|hn − 1| |n − 1i|2ih1|hn|
ρ̂n = . (11.26)
|ni|1ih2|hn − 1| |ni|1ih1|hn|
Os autovalores podem ser facilmente calculados por 4 ,

X X
det Ĥn = det Ĥn , (11.27)
n n
√ a luz e a transição atômica ∆ ≡ ω − ω0 e a frequência de Rabi

definindo a dessintonização entre
generalizada de n fótons $n ≡ ∆2 + nΩ2 = |$n |eik·R , que contem a função espacial do modo
luminoso. Isso dá a matriz diagonal de autovalores,

(n − 21 )ω + $2n 0
Ên = . (11.28)
0 (n − 12 )ω − $2n
À partir da transformação Ĥn Un = Un Ên , sob a condição que Un é unitário e hermitiano, Un† Un ,
√
e usando a abreviação tan 2φn ≡ nΩ/∆, obtemos:

cos φn sin φn
Un = . (11.29)
− sin φn cos φn
4
Valem as seguintes regras para determinantes
det(AB) = det A det B e (det A)−1 = det A−1 .
Os autovalores correspondentes aos autovalores Ên são obtidos por Ĥn |xi = enx |xi, ou seja,
Ên Un−1 |xi = enx Un−1 |xi. Conhecendo Ên |yi = eny |yi, obtemos |xi = Un |yi. A evolução temporal
do estado de Jaynes-Cummings, |ψ(t)i = e−iĤt |ψ(0)i, é descrito pela transformação,
e−iĤn t = Un e−iÊn t Un†

cos2 φn e−i$n t/2 + sin2 φn ei$n t/2 cos φn sin φn (ei$n t/2 − e−i$n t/2 )
= e−i(n−1/2)ωt
cos φn sin φn (ei$n t/2 − e−i$n t/2 ) sin2 φn e−i$n t/2 + cos2 φn ei$n t/2
(11.30)
A probabilidade de transição entre estados vestidos é,
4nΩ2 ∆2 $n t
|h2, n − 1|e−iĤn t |1, ni|2 = 2
sin2 . (11.31)
$n 2
A evolução temporal segue com [141],
ρ̂(t) = e−iĤn t ρ̂(0)eiĤn t ≡ L(t)ρ̂(0) . (11.32)
Resolva o Exc. 11.8.2.1.
11.2.1 The classical and quantum limits

11.2.1.1 The limit of high laser intensities and resonant interaction
O limite clássico é encontrado no limite n → ∞, onde um único fóton não faz diferencia, ou
seja, podemos identificar os estados |ni e |n + 1i. Então podemos aproximar o hamiltoniano do
sistema (11.25) pelo traço do próprio hamiltoniano sobre o número de fótons,
X Ω√

(n − 1)ω + ω20 n
Ω√
Ĥsemi = lim Trf ield Ĥ = hm|ni 2 hn|mi . (11.33)
n→∞
2 n nω − ω20
n
Essa situação, como ilustrada na PFig.n 11.3, descreve bem os estado de um laser aproximado
2
por um estado coerente, |αi = n √αn! |nie−|α| /2 . Para n → ∞, a incerteza da distribuição
√
poissoniana é pequena, ∆n/n̄ = 1/ n → 0, tal que o modo luminoso é caracterizado pelo
número médio de fótons e flutuações são desprezı́veis. Isso nós permite substituir a distribuição
poissoniana, Pn = δnn̄ ,
ω0
(n̄ − 1)ω 0 2 0 0 $2n̄
Ĥsemi = Ĥlaser + Ĥatom + Ĥint = + + $n̄∗ . (11.34)
0 n̄ω 0 − ω20 2 0
Agora, no caso de uma interação ressonante, ∆ = 0, a evolução de Jaynes-Cummings é,

−iĤn̄ t √1 e−i(n̄−1/2)ωt
cos 12 $n̄ t i sin 12 $n̄ t
e = , (11.35)
2 i sin 21 $n̄ t cos 12 $n̄ t
que é um resultado já obtido no Exc. 2.6.4.1.
Example √ 43 (Resonant π/2 pulse): Neste exemplo, consideramos pulsos π/2 ressonantes,
isto é, n̄Ωt = 12 π. A evolução de Jaynes-Cummings agora simplifica para,

1 i
e−iĤn̄ t = 21 e−i(n̄−1/2)ωt . (11.36)
i 1
Para grandes n̄, um pulso ressonante π/2 faz (suprimindo fases dinâmicas irrelevantes),

|2i|n̄ − 1i π/2 (|2i|n̄ − 1i + i|1i|n̄i)
y , (11.37)
|1i|n̄i (i|2i|n̄ − 1i + |1i|n̄i)
ou seja, para um campo coerente,

|2i|αi π/2 (|2i + i|1i)|αi
y . (11.38)
|1i|αi (i|2i + |1i)|αi
Obviamente, a estrutura do campo |αi não é afetada e recuperamos a dinâmica de um átomo
de dois nı́veis excitado por uma radiação clássica ressonante descrita pelas equações de Bloch
(10.50).
Figure 11.3: Esquema dos nı́veis atômicos para implementação de interações ressonantes com
campos clássicos (na transição inferior) e dispersivas com campos quânticos (na transição supe-
rior).
11.2.1.2 Dispersive interaction, the limit of large detunings

A dinâmica dispersiva de Jaynes-Cummings pode ser implementada por irradiação de um campo
de luz suficientemente dessintonizado para evitar processos de espalhamento de Rayleigh, como
ilustrado na Fig. 11.3. Esta interação resulta em um deslocamento de fase dos nı́veis atômicos.
√
Para |∆| nΩ o hamiltoniano simplifica para:
Ω√

(0) (1) (n − 21 )ω − ∆ 0 0 n
Ĥn = Ĥn + Ĥn = 2 + Ω√ 2 . (11.39)
0 (n − 12 )ω + ∆
2 2 n 0
(0)
No caso não-perturbado temos: Ĥn |ψn i0 = En |ψn i0 , onde o subespaço de n fótons é aberto
pela base |ni = (1 0) e (0 1). Em segunda ordem de perturbação:
(1) (2)
0 X hn|Ĥn(1) |mihm|Ĥn(1) |ni nΩ2
hψn |Ĥn |ψn i = hn|Ĥn + Ĥn |ni + (0) (0)
=∓ , (11.40)
En − Em 4∆
n6=m
que é um resultado já obtido nos Excs. 5.5.1.4 e 10.6.3.12. Em notação matricial,
2
(1) nΩ /4∆ 0
Ĥn ' . (11.41)
0 −nΩ2 /4∆
O operador de propagação temporal (11.30) então simplifica para,
!
(1) e inΩ2 t/4∆ 0
e−iĤn t = 2 . (11.42)
0 e−inΩ t/4∆
O fato, que os estados atômicos fundamental e excitado evolvem com diferentes fatores de fase
é importante, como mostraremos no seguinte exemplo.
Example 44 (Pulso π dispersivo): Como no caso prévio, consideramos um átomo de

dois nı́veis sujeito à um campo coerente, mas agora dessintonizado fora de ressonância.
Introduzindo a abreviação ϕ ≡ Ω2 t/4∆, a evolução de Jaynes-Cummings fica,
inϕ
(1) e 0
e−iĤn t = . (11.43)
0 e−inϕ
O fato, que o deslocamento de fase nϕ depende do número de fótons, e que ele vai em
sentidos opostos para os estados fundamental e excitado é interessante. Já estudamos no
Exc. 10.6.5.5, que a interação dispersiva do átomo com um campo de luz pode causar um
deslocamento de fase para o vetor de Bloch. Agora, percebemos que além disso, provoca um
deslocamento de fase da amplitude de probabilidade de ter n fótons no campo da luz por
um valor proporcional à n, ou seja (suprimindo fase dinâmicas irrelevantes),
inϕ
|2i|n − 1i nϕ e |2i|n − 1i
y . (11.44)
|1i|ni e−inϕ |1i|ni
Aplicando este resultado em estados de Glauber,

P n !
|2i|αi nϕ |2i n √αn! einϕ |ni |2i|αeiϕ i
y P αn −inϕ = . (11.45)
|1i|αi |1i n √n! e |ni |1i|αe−iϕ i
Obviamente, a fase do campo luminoso é deslocada por um valor ±ϕ, que depende da
população do átomo.
Notamos aqui, que essa dinâmica estudada no último exemplo fornece um método de trans-
ferir coerência de uma superposição atômica para um correlação quântica de um campo de luz.
Basta colocar o átomo inicialmente numa superposição de estados |1i+|2i, e o campo vai evoluir
para um estado de gato de Schrödinger |αeiϕ i+|αe−iϕ i. A transferência de correlações quânticas
entre graus de liberdade acoplados é uma das caracterı́sticas do modelo de Jaynes-Cummings.
Como exemplos estudaremos o fenômeno do colapso e renascimento quântico no Exc. 11.8.2.2 e
o desdobramento de Rabi no vácuo no 11.8.2.3.
11.2.1.3 Temporal evolution of the Bloch vector and the Q-function

No limite de baixas intensidades laser devemos considerar distribuições fotônicas não necessaria-
mente coerentes. A solução estacionária da equação de Schrödinger consiste dos estados vestidos
|1, ni e |2, n − 1i. Se, agora, expandimos um estado geral de Jaynes-Cummings nas amplitudes
cjn (t),
X
|ψi = (c1,n |1, ni + c2,n−1 |2, n − 1i) , (11.46)
n
eles seguem a equação de Schrödinger:

d c2,n−1 c2,n−1
= Ĥn . (11.47)
dt c1,n c1,n
A evolução dos coeficientes cjn descreve completamente a dinâmica de Jaynes-Cummings do

sistema através da formula (11.30). Obviamente, o estado de Jaynes-Cummings é normalizado,
pois,
X
hψ|ψi = Tr |ψihψ| = (|c1,n |2 + |c2,n |2 ) = 1 . (11.48)
n
O valor esperado para observáveis do campo Â|ni = An |ni é,

X X
hψ|Â|ψi = Tr ρ̂Â hi|hn|ψihψ|Â|ni|ii = An (|c1,n |2 + |c2,n |2 ) . (11.49)
i,n n
O valor esperado para o operador de aniquilação â|ai = an |ai é,

X√
hψ|â|ψi = n(c∗1,n−1 c1,n + c∗2,n−1 c2,n ) . (11.50)
n
Para determinar o estado interno do átomo, devemos tracejar sobre o campo luminoso. Como
processos de dissipação são desprezados, temos um estado puro descrito por, ρ̂ = |ψihψ|. As
populações e coerências são, portanto,
X X
ρij = hi|Trf ield ρ̂|ji = hi| hn|ψihψ|ni|ji = ci,n c∗j,n . (11.51)
n n
A projeção sobre o estado atômico é feita por,

P
|jihj|ψi cj,n |j, ni
= Pm 2
. (11.52)
hψ|jihj|ψi m |cj,m |
Com isso, podemos calcular o vetor de Bloch atômico, cuja norma interessantemente NÃO é
preservada, pois,
√ 
2Re ρ 12
√
|~
ρ| =  2Im ρ12  2
= 2|ρ12 | − 2ρ11 ρ22 = −2 det ρ̂ (11.53)
ρ22 − ρ11
X X X X
=2 c1,n c∗2,n c∗1,n c2,n − 2 |c2,n |2 |c1,n |2 6= 1 .
n n n n
Para determinar o estado do campo luminoso, devemos tracejar sobre o estado atômico. Por
exemplo, a amplitude de probabilidade de encontrar o estado |ψi em |ni é,
hn|ψi = c1,n |1i + c2,n |2i , (11.54)
tal que,
X
pn = hn|Tratom ρ̂|ni = hn| hi|ψihψ|ii|ni = |hn|ψi|2 = |c1,n |2 + |c2,n |2 . (11.55)
i=1,2
Para caracterizar o campo óptico separadamente do estado atômico podemos tentar, por
um cálculo similar à (11.49), projetar o estado de Jaynes-Cummings sobre uma base de estados
coerentes. Assim, a amplitude de probabilidade de encontrar o estado |ψi em |αi é,
X α∗n
2 /2
hα|ψi = e−|α| √ (c1,n |1i + c2,n |2i) (11.56)
n n!
2
X ∗n
α αm
|hα|ψi|2 = e−|α| √ √ (c∗1,m c1,n + c∗2,m c2,n ) ,
n n! m!
tal que,
 2 2 
X n
α X α 
n
−|α|2 
πQ(α) ≡ hα|Tratom ρ̂|αi = e c1,n √ + c2,n √ . (11.57)
n! n!
n n
Derivaremos este resultado no Exc. 11.8.2.4. Essa grandeza, que se chama função Q, permite a
ilustração do estado num plano Re α-Im α [31]. Ela geralmente é fácil de calcular, mas não exibe
muita informação, e.g. sobre fenômenos de interferências produzidos por correlações quânticas.
Na seção seguinte, vamos definir outras funções de distribuição, como a função de Wigner, que
também pode ser avaliada à partir dos coeficientes de Jaynes-Cummings [81].
(a) (b) (c) 0.15 (d)

0.6
0.4
0.1
1 1
12
0.2
pn
Reρ
11
ρ -ρ
0 0.5
22
0
1 0.05 10
-1 0
-0.2
-1 0 -10 0
0 Imρ -0.4 0 Imα
12 0
Reρ 1 -1 10 20 30 40 0 20 40 Reα 10 -10
12
t n
Figure 11.4: (Code: LM Quantumf ields JcummingsDispersive.m) Evolução do estado du-

rante uma interação tipo Jaynes-Cummings: (a) Vetor de Bloch, (b) evolução temporal da
coerência ρ12 mostrando o fenômeno do colapso e renascimento, (c) distribuição de fótons e
(d) função Q(α).
O Jaynes-Cummings, obviamente, pode transferir coerências entre um átomo e um campo

de luz. No Exc. 11.8.2.5 estudamos um caso, como uma sequência de pulsos de Ramsey cria um
estado de gato de Schrödinger em um campo luminoso.
11.2.2 Quantum correlations in light modes

Correlações quânticas resultando em distribuições não-poissonianas do número de fótons são
possı́veis, por exemplo, através da interação de Jaynes-Cummings de um átomo com um modo
de luz, como discutido na Sec. 11.2. As vezes, pode ser vantajoso representar estas correlações
numa base de estados coerentes. Existem várias representações, as mais conhecidas sendo as
funções de distribuição coerente Q e P e a função de Wigner W . Antes de mostrar quais são as
correlações quânticas produzida por uma interação tipo Jaynes-Cummings, vamos discutir nesta
seção as funções de distribuição coerente dos estados mais comuns, isto é, os estados de Fock e
de Glauber, o estado térmico e o gato de Schrödinger.
Para definir as funções de distribuição coerente introduzimos primeiramente as funções car-
acterı́sticas χN,S,A e as funções de distribuições correspondentes,
† ∗
χN (λ) = Tr ρ̂eλâ eλ â , P = F −1 χN
∗ †
χS (λ) = Tr ρ̂eλ â−λâ , W = F −1 χS .
∗ †
χA (λ) = Tr ρ̂e−λ â e−λâ , Q = F −1 χA
Aqui, N denota a ordem normal, S a ordem simétrica e A a ordem antissimétrica. As funções
de distribuições são relacionadas. Usando a formula de Baker-Haussdorff (3.90) é fácil mostrar,
2 /2 2 /2
e−|λ| χN (λ) = χS (λ) = e|λ| χA (λ) . (11.58)
Assim, a função Q corresponde à uma função de Wigner suavizada que, por sua vez, corresponde
à uma função P suavizada. A transformação de Fourier complexa inversa dá imediatamente 5 ,

2 2 2 2 2 −2|λ|2 1 2
Q = W ? e−2|λ| = P ? e−2|λ| ? e = P ? e−|λ| . (11.59)
π π π π
As funções de distribuição podem ser interpretadas da maneira seguinte: P (α) é a probabilidade

de encontrar o estado coerente |αi dentro da mistura estatı́stica,
Z
ρ̂ = P (α)|αihα|d2 α , (11.60)
e Q(α) é o valor esperado do operador densidade:
Q(α) = π1 hα|ρ̂|αi . (11.61)
Os estados são expressos por funções de peso bidimensionais P, Q, W . A seguir, damos alguns
exemplos para essas representações.
Figure 11.5: (Code: LM Quantumf ields V ariousW igners.m) Representações Q (primeira

linha) e Wigner (segunda linha) de um estado de Glauber (primeira coluna), de Fock (segunda
coluna) e de um gato de Schrödinger (terceira coluna).
5 1
exp(−a|λ|2 + bλ + cλ∗ )d2 λ = 1
exp( bc
R
Usando a formula integral, π a a
).
11.2.2.1 Fock state in the coherent representation

Na base de estados de número |ni, o estado de Fock é caracterizado por,
|ni = (â† )n |0i
ρ̂ = |nihn| . (11.62)
Pk = δnk
Este estado pode ser expandido em estado coerente |αi pelo seguinte procedimento. Para grandes
n calculamos primeiramente a função de distribuição coerente P ,
√
P|ni (α) = δ (1) (|α| − n) , (11.63)
pois ela nós permite derivar a matriz densidade pela formula (11.60),
Z
√
ρ̂|ni = δ (1) (|α| − n)|αihα|d2 α (11.64)
Z ∞ Z 2π Z 2π
√
= δ (1) (|α| − n)|αihα| |α|2 d|α|dϕα = n |αihα|dϕα .
0 0 0
A função de distribuição coerente Q é, inserindo o operador densidade obtida em (11.64),
Q|ni (α) = π1 hα|ρ̂|ni |αi (11.65)

Z 2π Z 2π iϕ 2
|α|2n −|α|2

1 2 1 −α−|β|e β
= πn |hα|βi| dϕβ = n e dϕβ , e ,
0 π 0 πn!
e, finalmente, a função de Wigner,

n0
X
2 −2|α|2 n0 n0 (−4|α|2 )n
W|ni (α) = e (−1) . (11.66)
π n n!
n=0
11.2.2.2 Representations of Glauber states

Vice versa, os estados coerentes podem ser expandidos por estados de Fock por,
X αn
|αi = e−|α|/2 √ |ni
n n!
X |α|n e−|α|
ρ̂ = |nihn| . (11.67)
n
n!
2 |α|2n
Pn = |hn|αi|2 = e−|α|
n!
A função de distribuição coerente P é,
P|βi (α) = δ (2) (α − β) , (11.68)

a matriz densidade, Z
ρ̂|βi = δ (2) (α − β)|αihα|d2 α = |βihβ| . (11.69)
a função de distribuição coerente Q,
1 −|α−β|2
Q|βi (α) = e , (11.70)
π
e a função de Wigner,
2 −2|α−β|2
W|βi (α) = e . (11.71)
π
Example 45 (State of a laser ): Seguinte P. Saldanha, o estado correto de um feixe laser

não é simplesmente um estado coerente, mas:
Z
dϕ
ρ̂ = |αeiϕ ihαeiϕ | . (11.72)
2π
Fazendo a média, este estado pode ser escrito como uma superposição de estados de Fock,
X
ρ̂ = Pn |nihn| , (11.73)
n
mas sem fase especı́fica,

X
ρ̂ 6= c∗m cn |mihn| . (11.74)
m,n
11.2.2.3 Thermal states

Um modo de luz em uma mistura térmica não pode ser representado por um estado puro, mas
precisa da matriz densidade,
X n̄n
ρ̂therm = |nihn| . (11.75)
n
(1 + n̄)1+n
1 −|α|2 /n̄
Ptherm (α) = e , (11.76)
πn̄
a matriz densidade, Z
1 2 /n̄
ρ̂ = e−|α| |αihα|d2 α , (11.77)
πn̄
1 2
Qtherm (α) = e−|α| /(n̄+1) . (11.78)
π(n̄ + 1)
1 2
Wtherm (α) = e−|α| /(n̄+1/2) . (11.79)
π(n̄ + 1/2)
11.2.2.4 Schrödinger cat states

Estados de gato de Schrödinger são estados correlacionados de muitas partı́culas (ou quase-
partı́culas). A expansão do estado de Schrödinger por estados de Fock é,
X 1 ± (−1)n |α|n e−|α|
|ψ± i = √ |ni . (11.80)
n 2 n!
P|β0 i|β1 i (α) = δ (2) (α − β0 ) + δ (2) (α − β1 ) , (11.81)
a matriz densidade,
ρ̂ = , (11.82)
2 2
Q|β0 i|β1 i (α) = π1 e−|α−β0 | + π1 e−|α−β1 | . (11.83)
A função caracterı́stica com ordem normal é,

+ ∗ ∗ ∗β
χN (λ) = hβ0 |eλa e−λ a |β1 i = eλβ0 −λ 1
, (11.84)
Z
∗ α−λa∗
W (α) = 1
π χS (λ)eλ d2 λ (11.85)
Z
2 /2 ∗ +(−α+β )λ ∗ −β ∗ )(α−β )
= 1
π e−|λ| e(α−β1 )λ 0
d2 λ = π2 e(α 1 0
.
Para β0 = β1 recuperamos o estado de Glauber,

+ ∗
χN (λ) = (hβ0 | ± hβ1 |) eλα e−λ α (|β1 i ± |β0 i) , (11.86)
e 2 2 ∗ ∗ −β ∗ )(α−β
e−2|α−β0 | + e−2|α−β1 | ± 2Re e2β1 β0 e−2(α 1 0
W (α) = ∗ . (11.87)
π 1 ± Re e2β1 β0
No nı́vel microscópico, temos o exemplo dos graus de liberdade internos de átomos, que
podem estar em estados de superposição |hi|ji| = δij . Por outro lado, os estados de gato
de Schrödinger são estados em campos contı́nuos de Schrödinger com superposições quânticas
mesoscópicas |hα|βi| = e−|α−β| . Também temos que enfatizar a diferença fundamental entre os
gatos de Schrödinger e os estados de superposição de modos,
|ψi = |αi + |βi =6 |αi|βi
ρ = |αihα| + |βihβ| + |αihβ| + |βihα| 6 |αβihαβ|
= . (11.88)
2 2 2
W (α) = π1 (e−2|α−α0 | + e−2|β−β0 | + termos de interferência 6= π1 e−2|αβ−α0 β0 |
Os gatos de Schrödinger exibem interferências no espaço de fase, enquanto que para super-
posições de modos, interferências aparecem apenas quando se varia um parâmetro (por exemplo,
o comprimento de um braço do interferômetro).
Interferências quânticas macroscópicas (isto é, interferências que são detectáveis com apar-
elhos macroscópicos, por exemplo, em esquemas heteródinos) são nomeadas gatos (fuzzy) de
Schrödinger, se os estados de interferentes estiverem visivelmente separados no espaço de fase.
Eles são muito sensı́veis à dissipação e são facilmente convertidos em misturas estatı́sticas.
Por exemplo, |αi ± | − αi contém apenas números de fótons ı́mpares (par) na função de dis-
tribuição Pn . Depois de algum tempo ∼ τcav /N , a distribuição é convertida em uma distribuição
poissoniana. Quanto maior N , mais rápida será a decoerência e, portanto, os estados do gato
de verdade nunca serão observados.
11.2.2.5 The Q-, P - and W -function for Jaynes-Cummings cat states

Para estado |ψi = 2−1/2 (|αi + i|βi), temos,
2 /2 αn 2 /2 βn
hψ|â|ψi = 12 (α + β) tal que c1,n = e−|α| √ e c2,n = ie−|β| √ .
2n! 2n!
E para o estado |ψi = 2−1/2 (|ai + |βi),
2 2 ∗ ∗

hψ|â|ψi = 12 α + β + e−|α| /2−|β| /2 (βeγ β + γeγβ ) .
Seja,
|ψi = √1 (|1i|βi + i|2i|γi) . (11.89)
2
Comparando com os coeficientes de JC segue,
X√ X√
β = 21 nc∗1,n−1 c1,n e 1
2 nc∗2,n−1 c2,n . (11.90)
n n
As funções Q, P e W para gatos de Schrödinger usuais são conhecidas. Agora, só inserir β e γ.
Example 46 (Função de Wigner para dinâmica de Jaynes-Cummings): Outras
funções de distribuição costumam ser usadas para caracterizar o estado de um campo de
Heisenberg. A função de Wigner, por exemplo, é definida como [81],
Z
∗ ∗
W (α) = π12 χS (λ)eαλ −α λ d2 λ onde d2 λ = dRe λ dIm λ .
Aqui, precisamos introduzir as função caracterı́sticas,

2 † ∗
χS (λ) = e−|λ| /2 χn (λ) onde χN (λ) = hψ|eλâ e−λ â |ψi ,
P
onde expandimos |ψi = n c1,n |1i|ni + c2,n−1 |2i|ni. Agora, calculamos,
s
k n!
â |ni = |n − ki
(n − k)!
e
∗ X (−λ∗ )k X (−λ∗ )k n
e−λ â |ni = âk |ni = √ |n − ki ,
k! k! k
k k
e,
s
Xn
λâ† −λ∗ â (−λ∗ )k λm−n+k n m
hn|e e |ni = p se m≥n
k!(m − n + k)! k m−n+k
k=0
r n
n! m−n X m (−|λ|2 )k
= λ
m! k+m−n k!
k=0
r
n! m−n m−n
= λ Ln (|λ|2 ) polinômio de Laguerre
m!
2
= e−|λ| /2
unm (λ) se m≥n senão pegue umn (−λ∗ ) .
Função caracterı́stica,
X † ∗ † ∗
χn (λ) = c∗1,m c1n hm|eλâ e−λ â |ni + c∗2,m−1 c2,n−1 hm − 1|eλâ e−λ â |n − 1i
m,n
r
X n! m−n m−n
= (c∗1,m c1n + c∗2,m c2,n ) λ Ln (|λ|2 ) .
m,n
m!
Função caracterı́stica,
X
χS (λ) = (c∗1,m c1n + c∗2,m c2,n )unm (λ) .
m,n
Devemos distinguir os casos, porque unm (λ) = umn (−λ∗ ) = (−1)m−n umn (λ∗ ),
X X
χS (λ) = (c∗1,m c1n + c∗2,m c2,n )unm (λ) + (c∗1,m c1n + c∗2,m c2,n )unm (−λ∗ )
m≥n m<n
r
X n! −|λ|2 /2 m−n
2 −|λ|2 /2
= (|c10 | + |c20 |)e + e Ln (|λ|2 )
m>n
m!
∗
(c1,m c1n + c∗2,m c2,n )λm−n + (c1,m c∗1n + c2,m c∗2,n )(−λ∗ )m−n .
Transformada de Fourier de unm (|λ|) em vez de unm (λ),

Z Z
αλ∗ −α∗ λ 2
unm (|λ|)e d λ= unm ( 12 |xλ + ipλ |)eipα xλ −ixα pλ dxλ dpλ
Z Z
1
= e+ipα xλ unm ( 12 xλ )dxλ − e−ixα pλ unm ( 21 pλ |)dpλ ,
ipα − xα
e
Z r Z
n! 2 x2λ ipα xλ
e +ipα xλ
unm ( x2λ )dxλ = e−xλ /2 xm−n
λ Lm−n
n ( 2 )e dxλ
m!
r r
n! π 1 m−n 2
= (−1)int 2 e−pα /2 Hen (pα )Hem (pα ) .
m! 2 n!
Função de Wigner,
r r
∗ π −1
1 m−n
W (α) = 1
π 2 (c1,m c1n
+ c∗2,m c2,n ) (−1)int 2
α 2 n!m!
h i
−p2α /2 −x2α /2
e Hen (pα )Hem (pα ) + e Hen (xα )Hem (xα ) .
11.2.3 The Jaynes-Cummings model with dissipation

Aplicando o método da simulação quântica de Monte-Carlo da função de onda para o modelo
de Jaynes-Cummings, escrevemos o hamiltoniano efetivo do campo luminoso como [92],

Ĥef f = ωâ† â + ω0 σ̂ + σ̂ − 1
2 + Ω
2 âσ̂ + + â† σ̂ − + iΓ + −
2 σ̂ σ̂ . (11.91)
Ou em notação matricial,
 1 
− 2 ω0 0
 0 1 i 1 
 2 0 − 2Γ
ω 2Ω 
 1
ω − 12 ω0 
 2Ω 0 √ 1 
 1 i 
Hef f =  0 ω + 2 0 − 2Γ
√ ω 22Ω  .
 1
22Ω 2ω − 12 ω0 0 
 
 1 i 
 0 2ω + 2 0 − 2Γ
ω 
..
.
(11.92)
O fluxograma da simulação é,
|1i(h1|+h2|)|ψi |ψi
|ψi → |ψi →
P |(h1|+h2|)|ψi| P ||ψi|
|1i n (c1n |ni+c2n−1 |n−1i) c1n |1i|ni+c2n−1 |2i|n−1i
= √P 2
= n√
P 2 2
n |c1n +c2n | | n (|c1n | +|c2n | )
projeção & evolução dinâmica . renormalização
|ψ(t + dt)i = e−iHef f dt |ψi
- ↓ %
?
sim ζ > hψ(t + dt). |ψ(t + dt)i no
variável aleatória
Absorção ou espalhamento de luz causa o decaimento do campo óptico como α(t) ∝ e−κt/2 .
A projeção (em notação por componentes) é implementada por,
1
c0jn ≡ pP cjn δj1 , (11.93)
2
n n(|c1n + c2n | )
a renormalização por,
1
c0jn ≡ pP cjn , (11.94)
2 2
n (|c1n | + |c2n | )
e a evolução dinâmica por c0jn ≡ e−κnt/2 cjn . Notamos, que a dissipação por perdas da cavidade
também pode ser tomada em conta pela equação mestre. Faz o Exc. 13.7.2.1.
11.3 Spontaneous emission and light scattering

11.3.1 Interaction of atoms with vacuum modes
O hamiltoniano (11.10) descreve uma dinâmica puramente coerente. Este sistema paradigmático
de um único átomo de dois nı́veis imóvel e interagindo com um único modo de cavidade é
o chamado modelo de Jaynes-Cummings discutido na Sec. 11.2. O modelo é suficientemente
simples para permitir a obtenção de resultados analı́ticos. No entanto, o hamiltoniano (11.10)
não inclui processos de emissão espontânea, que podem ser entendidos como interação do átomo
com os modos luminosos do vácuo. Ou seja, devemos estender o hamiltoniano,
Ĥ = Ĥatom + Ĥf ield + Ĥatom−f ield + Ĥvacuum + Ĥatom−vacuum . (11.95)
A evolução do sistema representado pelo hamiltoniano (11.113) é descrita por um operador
densidade total, ρ̂total (t), obedecendo a equação de von Neumann,
dρ̂total i
= − [Ĥ, ρ̂total ] , (11.96)
dt ~
11.3. SPONTANEOUS EMISSION AND LIGHT SCATTERING 301
que tem a solução,
ρ̂total (t) = e−iĤt/~ ρ̂total (0)eiĤt/~ ≡ e−iLt ρ̂total (0) . (11.97)
Frequentemente, estamos somente interessado na evolução ou do campo de luz ou do estado

interno do átomo. Nestes casos, calculamos o traço sobre todos os graus de liberdade, que não
são observados,
ρ̂atom = Trlight ρ̂total e ρ̂light = Tratom ρ̂total . (11.98)
Escolhendo o estado inicial do vácuo eletromagnético (para emissão espontânea) como o
vácuo fotônico ρ̂f ield = |{0}ih{0}| e definindo um operador de projeção sobre este estado, P̂ ... ≡
ρ̂f ield (0)Trf ield ... = P̂ 2 ..., e, além disso, aplicando as aproximações da onda rotativa, de Markov
e de Born e considerando apenas os graus internos de liberdade de um átomo de dois nı́veis,
segue-se depois de alguns cálculos a equação de Bloch-Lindbladt ou equação mestre [212]. Para
uma discussão sobre a validade da aproximação de Born-Markov [188]. Para a relação entre a
aproximação de Markov e a regra de ouro de Fermi [3].
Na seção seguinte vamos dar uma derivação simplificada concentrando-nós na situação de
um único átomo parado no espaço, excitado por um laser e espalhando fótons para o vácuo
eletromagnético.
11.3.1.1 The spontaneous emission

A emissão espontânea pode ser entendida como um processo de difusão de energia de um sistema
com um número de graus de liberdade restrito para um grande banho térmico. O sistema de um
átomo excitado por um laser pode ser descrito dentro de um espaço de Hilbert dois-dimensional.
No entanto, fótons podem ser espalhadas por emissão espontânea em todas as direções do espaço.
Para tomar em conta esse fato incluir no hamiltoniano não só a interação do átomo com o laser
incidente (vetor de onda k0 , frequência ωk0 , força do acoplamento gk0 ), más também com os
modos do vácuo eletromagnético (vetor de onda k, frequência ωk , força do acoplamento gk ).
Figure 11.6: Espalhamento de um laser por um átomo.
Vamos ver, que com esse hamiltoniano podemos derivar, num cálculo é conhecido como teoria
de Weisskopf-Wigner, a equação de Schrödinger para as amplitudes dos nı́veis atômicos (10.69)
inclusive emissão espontânea.
Seja a frequência do laser ω0 , a frequência da ressonância atômica ωa e a frequência da luz

espalhada ω. O hamiltoniano é,

Ĥ = ~gk0 σ̂ − e−iωa t + σ̂ + eiωa t â†k0 eiω0 t−ik0 ·r + âk0 e−iω0 t+ik0 ·r
. (11.99)
P
+ k ~gk σ̂ − e−iωa t + σ̂ + eiωa t â†k eiωk t−ik·r + âk e−iωk t+ik·r
Aqui, Ω0 é a frequência de Rabi da interação entre o átomo e o modo da bomba (que é tratado
como campo clássico), σ̂ − é o operador
p de abaixamento da excitação atômica, âk é o operador de
aniquilação de um fóton, e gk = d ω/(~ε0 V ) descreve o acoplamento entre o átomo e um modo
do vácuo com o volume V . O átomo tem dois estados, o estado fundamental |gi e estado excitado
|ei. Como estamos considerando apenas um átomo fixo no espaço 6 , podemos colocar-lhe na
posição r = 0. Além disso, considerando um laser incidente com alta potência,
√
âk0 |n0 ik0 = n|n0 − 1ik0 ' |n0 ik0 , (11.100)
âk0 é aproximadamente uma observável proporcional à raiz da intensidade. Como [âk0 , â†k0 ] ' 0,
√
podemos desconsiderar a natureza quântica e substituir, Ω0 ≡ 2 n0 gk0 . Com a aproximação da
onda rotatória (rotating wave approximation, RWA) o hamitoniano fica
h i Xh i
Ĥ = ~2 Ω0 σ̂ − â†k0 ei∆0 t + h.c. + ~ gk σ̂â†k ei∆k t + h.c. , (11.101)
k
introduzindo as abreviações
∆0 ≡ ω0 − ωa e ∆k ≡ ωk − ωa . (11.102)
O estado do sistema é dado por

X
|Ψ(t)i = α(t)|0ia |0ik + β(t)|1ia |0ik + γk (t)|0ia |1ik . (11.103)
k
A evolução temporal das amplitudes é obtida inserindo o hamiltoniano e o ansatz acima

dentro da equação de Schrödinger,
∂ i
|Ψ(t)i = − Ĥ|Ψ(t)i . (11.104)
∂t ~
Obtemos,
α̇(t) = −i Ω20 ei∆0 t β(t) (11.105)

X
β̇(t) = −i Ω20 α(t)e−i∆0 t − igk γk (t)e−i∆k t
k
γ̇k (t) = −igk ei∆k t β(t) .
Escolhemos as condições iniciais,
α(0) = 1 e β(0) = 0 e γk (0) = 0 . (11.106)
Integramos a terceira equação,

Z t
0
γk (t) = −igk ei∆k t β(t0 )dt0 (11.107)
0
e substituimos dentro da segunda equação,

X Z t
Ω0 0
β̇(t) = −i α(t)e−i∆0 t − gk2 ei∆k (t −t) β(t0 )dt0 . (11.108)
2 0
k
6
Não deixamos o átomo ser acelerado pelo recuo fotônico.
11.3.1.2 The Markov approximation

Para sistemas pequenas (o que é o caso de um átomo só) podemos aplicar a aproximação de
Markov 7 dizendo que a variação das amplitudes, β(t0 ) ' β(t), é mais lenta do que a evolução
do sistema, ei(ωk −ωa )t , na equação integro-diferencial. Com essa aproximação a equação integro-
diferencial, que e equivalente a uma equação de ordem arbitrariamente alta, se reduz á uma
equação diferencial simples de primeira ordem. Definindo t00 ≡ t − t0 ,
d Ω0 X Z t 0
β(t) = −i α(t) − gk2 ei(ωk −ωa )(t −t) β(t0 )dt0 (11.109)
dt 2 0
k
Ω0 X Z t 00
= −i α(t) − gk2 e−i(ωk −ωa )t β(t − t00 )dt00 .
2 0
k
Rt 0
Usando a aproximação de Markov β(t − t00 ) ' β(t), com lim 0 e−i(ωk −ωa )t dt0 = πδ(ωk − ωa ), e
P R 3 t→∞
V
substituindo k −→ (2π) 3 d k, chegamos a,
d Ω0 X
β(t) ' −i α(t) − gk2 β(t)πδ(ωk − ωa ) (11.110)
dt 2
k
Z
Ω0 V
= −i α(t) − β(t) gk2 πδ(ωk − ωa )d3 k
2 (2π)3
Ω0 V 1 Ω0 Γ
= −i α(t) − 3
β(t)4πgk2 a πka2 = −i α(t) − β(t) .
2 (2π) c 2 2
No último passo introduzimos, como abreviação, a taxa de emissão espontânea,
V 2 2
Γ≡ k g , (11.111)
πc a ka
Finalmente,
d Ω0 d Ω0 Γ
α(t) = −i β(t) e β(t) = −i α(t) − β(t) . (11.112)
dt 2 dt 2 2
Esses são exatamente as equações das amplitudes de probabilidade (10.69) derivadas da
equação de Schrödinger, mas agora com o termo de emissão espontânea sendo explicitamente
derivado.
11.3.1.3 Complete hamiltonian of atom-photon interaction

No nı́vel mais fundamental, a interação da luz com um único átomo envolve vários graus de
liberdade: 1. o movimento do centro de massa do átomo, 2. o campo da luz incidente, 3. a
excitação eletrônica do átomo e 4. os modos do vácuo eletromagnético 8 . Cada grau de liberdade
carrega sua energia que, em mecânica quântica, é representada por um hamiltoniano. Além
disso, as interações 5. entre a luz e o átomo e 6. entre o átomo e o vácuo também contribuem
componentes para o hamiltoniano completo através das energias associadas aos processos de
absorção e emissão,

P2
Ĥ = Ĥelectron + + Ĥf ield + Ĥvacuum + Ĥatom−f ield + Ĥatom−vacuum . (11.113)
2M
7
A aproximação não vale necessariamente para nuvens grandes de átomos.
8
Não consideramos efeitos de muitos corpos devido à estatı́stica quântica ou interação interatômicas.
Em muitas situações na óptica quântica NÃO podemos desprezar o impacto do recuo fotônico
e omitir o movimento do centro de massa atômico [primeiro termo do hamiltoniano (11.113)].
Por isso, estenderemos no Cap. 12.3.2 a discussão para incluir o movimento atômico e as forças
da luz sobre átomos.
11.3.2 Resonance fluorescence and coherent and incoherent light scattering
A situação tı́pica para um experimento de espectroscopia é ilustrada na Fig. 11.7: Um feixe

de luz entendido com onda plana incide num átomo (ou numa amostra de muitos átomos),
uma parte de luz é absorvida e remitida em uma direção descrita por um ângulo sólido dΩ. O
espalhamento de luz é, obviamente, um processo de segunda ordem envolvendo duas transições,
uma absorção e uma emissão.
Figure 11.7: (a) Geometria de um experimento de espalhamento. (b) Contribuições espectrais

elásticas e inelásticas da luz espalhada para luz espalhada por um átomo de três nı́veis.
Radiação pode ser absorvida ou espalhada por um átomo em diferente maneiras, dependendo
se a interação for espalhamento elástico ou espalhamento inelástico, um processo coerente ou in-
coerente, espontâneo or (bosonicamente) estimulado. Estas propriedades são caracterı́sticas
para muitos processes, em particular, a fluorescência ressonante (isto é, absorção e emissão),
espalhamento de Rayleigh ou espalhamento de Raman. No seguinte, vamos esclarecer esta clas-
sificação.
Cada processo de espalhamento é espontâneo ou estimulado 9 . Processos espontâneos po-
dem ser entendidos como sendo estimulados flutuações do vácuo. Espalhamento de Rayleigh é
elástico, isto é, a energia cinética do átomo espalhador é a mesma antes e depois do processo
de espalhamento. Em contraste, o espalhamento de Raman é inelástico. Emissão espontânea
é devido ao decaimento da população de um estado excitado, o espalhamento espontâneo de
Rayleigh é devido ao decaimento de um momento dipolar induzido. Nos dois casos, o processo é
estimulado por flutuação do vácuo: Na emissão espontânea um autoestado de excitação interna
decai devido ao acoplamento à um reservatório eletromagnético. No espalhamento espontâneo
de Rayleigh a superposição coerente decai.
9
Teorias clássicas do espalhamento de luz através de excitação do movimento eletrônico dentro dos modelos
de Lorentz ou de Drude podem ser consultada nas Secs. apostila Eletrodinâmica do mesmo autor [55]. . Nestes
modelos entendemos muitos aspectos do espalhamento de Compton, de Thomson e de Rayleigh.
11.3.2.1 Resonance fluorescence

Desde que fizemos a segunda quantização (11.8) sabemos, que o campo de luz emitida por um
radiador na zona de radiação (λ r) é, tomando em conta a retardação 10 , dado por,
† −
hÊ+
s (r, t)i ∝ hâs σ i ∝ ρ̃21 e hÊ− + † − +
s (r, t)Ês (r, t)i ∝ hâs σ âs σ i ∝ ρ22 . (11.114)
Por isso, o campo elétrico emitido por um átomo e a intensidade da luz espalhada são dados
por,
eω02 ˆ · r12
hÊ+
s (r, t)i = − ρ̃21 (t − rc )e−iω(t−r/c)
4πε0 c2 r
, (11.115)
4
α~ω0 |ˆ · r12 |2
I¯s = cε0 hÊ− +
s (r, t)Ês (r, t)i = ρ22 (t − rc )
4πc2 r2
com a definição da constante de Sommerfeld α = e2 /4πε0 ~c. Calculamos o fluxo total de fótons
emitidos,
Z ¯ 2 Z
(sp) Is r 1 α~ω04 |r12 |2 cos2 θ
Wf i = dΩ = ρ22 (t − rc ) sin θdθdφ (11.116)
~ω0 ~ω0 4πc2
8π α~ω04 |r12 |2 2α
= 2
ρ22 (t − rc ) = 2 ω03 |r12 |2 ρ22 (t − rc ) .
3~ω0 4πc 3c
O resultado coincide com taxa de emissão espontânea Γ calculada em (??).
A seção eficaz diferencial de espalhamento pode ser definida por,
dσ ω I¯s r2
≡ . (11.117)
dΩ ωs I¯0
11.3.2.2 Coherently scattered light and saturation

A intensidade total da luz espalhada sendo I¯s , a fração da luz coerentemente espalhada é,
I¯scoh hÊ− +
s (r, t)ihÊs (r, t)i
= . (11.118)
I¯s hÊ− +
s (r, t)Ês (r, t)i
Inserindo as expressões (11.115) e a solução estacionária das equações de Bloch (10.73),
I¯scoh |ρ̃21 (∞)|2 1 I¯sincoh

= = = 1 − . (11.119)
I¯s ρ22 (∞) 1+s I¯s
Ou seja, desde que a fluorescência ressonante é proporcional à população do estado excitado,
Stot ≡ ρ22 (∞), as partes coerentes e incoerentes da fluorescência são,
s/2 s2 /2
Scoh = |ρ21 (∞)|2 = e Sincoh = ρ22 (∞) − |ρ21 (∞)|2 = . (11.120)
(1 + s)2 (1 + s)2
Portanto, Sincoh = sScoh .

Notamos, que a integral do espectro de fluorescência é,
Z ∞
Ω2 π Ω2 ΩΓ πΩ
2
2 2 2
d∆ = √ −→ . (11.121)
−∞ 4∆ + 2Ω + Γ 2 2Ω2 + Γ2 2Γ
10
A versão clássica desta fórmula é pode ser consultada na apostila Eletrodinâmica do mesmo autor [55],
Sec. 9.1.3, Eq. (8.41) .
Figure 11.8: (Code: LM Quantumf ields InElastic.m) Espalhamento elástico versus inelástico.
11.3.3 Correlation functions

As propriedades da coerência de um campo de luz (ou de um condensado de Bose-Einstein) são
medidas por funções de correlação de nèsima ordem função de correlação,
hÊ − (r1 , t1 )..Ê − (rn , tn )Ê + (rn+1 , tn+1 )..Ê + (r2n , t2n )i
g (n) (r1 , t1 , .., r2n , t2n ) ≡ q . (11.122)
− + − +
hÊ (r1 , t1 )Ê (r1 , t1 )i..hÊ (r2n , t2n )Ê (r2n , t2n )i
Particularmente importantes são as funções de correlação de 1a e 2a ordem g (1) e g (2) . Con-

siderando somente coerências no domı́nio temporal, i.e. k1 k k2 e (t2 − zc2 ) − (t1 − zc1 ) = τ , como
Ê(r, t) = Ê(ωt − k · r), podemos definir (sem perda de generalidade),
hÊ − (t)Ê + (t + τ )i hÊ − (t)Ê − (t + τ )Ê + (t + τ )Ê + (t)i

g (1) (τ ) ≡ e g (2) (τ ) ≡ , (11.123)
hÊ − (t)Ê + (t)i hÊ − (t)Ê + (t)i2
onde, Z t
1
h· · ·it = lim · · · dt (11.124)
t→∞ t 0
denota a média temporal. Definindo a intensidade como Iˆ = 2ε0 cÊ + Ê − , as coerências ficam,
2
g (1) (τ ) ≡ 2εI0 c hT N Ê − (t)Ê + (t + τ )i e ˆ I(t
g (2) (τ ) ≡ 2εI0 c hT N I(t) ˆ + τ )i . (11.125)
As coerências devem ser calculadas à partir dos operadores de campo respeitando a ordem tem-
poral e a ordem normal. Elas são grandezas úteis para descrever fenômenos como o bunching
de fótons ou entender os espectros de fluorescência ou o espalhamento de luz por átomos cor-
relacionados. g (1) mede a coerência de um campo de luz (quanto ele se assemelha à uma onda
senoidal). g (2) mede, para um dado grau de coerência, o desvio do campo de luz do estado
quântico que mais se aproxima à luz clássica (quanto ele se assemelha à um laser).
As funções de correlação g (1) e g (2) são experimentalmente medidas em experimentos de
Young e Hanbury-Brown-Twiss. Eles são ilustrados nas Figs. 11.9.
Coerência e caos são propriedades contrárias da luz, que podem ser quantificadas pelo es-
pectro da luz ou a função de autocorrelação. A função de autocorrelação de primeira ordem
descreve flutuações de fase e amplitude, a função de autocorrelação de segunda ordem descreve
flutuações de intensidade. O perfil de emissão de uma fonte de luz emerge geralmente como
uma combinação de vários efeitos fı́sicos: O acoplamento a um banho térmico dá origem a
P −~ω(n+1/2)/k
uma distribuição térmica da energia de radiação, Pn = e −~ω(n+1/2)/k B T / ne BT .
Numa fonte de luz, o meio ativo exibe ressonâncias e alargamentos, e o ressonador imprime uma
estrutura modal.
Figure 11.9: (a) Esquema do experimento de Young. (b) Esquema do experimento de Hanbury,
Brown e Twiss. O experimento de Young revela a coerência de um campo, isto é, sua capacidade
de interferir. Em contraste, o experimento de Hanbury-Brown-Twist indica as correlações entre
as (quase-)partı́culas, que compõem o campo, isto é, efeitos devido às estatı́sticas quânticas ou
interações.
11.3.3.1 Examples of correlation functions

Luz coerente em primeira ordem satisfaz |g (1) (τ )| = 1, luz incoerente |g (1) (τ )| = 0 e para luz
parcialmente temos valores intermediários. Luz coerente em segunda ordem satisfaz g (2) (−τ ) =
τ τ
g (2) (τ ) e 1 ≤ g (2) (0) ≤ ∞ e 0 ≤ g (2) (τ ) ≤ ∞ e g (2) (0) ≥ g (2) (τ ) → c 1. Para luz caótica temos
g (2) (τ ) = 1 + |g (1) (τ )|. Vamos agora olhar para alguns casos especı́ficos para os quais as funções
de correlação podem facilmente ser calculadas.
Com as definições (11.123) é fácil calcular a função de autocorrelação e o espectro de um
campo de luz laser,
laser (11.126)
E = eRiωt
e−iωτ dt
=⇒ g (1) (τ ) = R
dt
= e−iωτ
(1) (11.127)
=⇒ F g (τ ) = δ(∆)
g (2) (τ ) = 1
Vemos, que os valores absolutos das coerências de primeira e segunda ordem são constantes, e
que o espectro é fino como uma função δ. Par um laser sujeito à ruı́do branco de fase (ζ seja
um is a número aleatório com distribuição normal) temos,
laser ruidoso (11.128)

E = eRi[ωt+ζ(t)]
e−i[ωτ +ζ(t+τ )−ζ(t)] dt
=⇒ g (1) (τ ) = R
dt
= e(iω−γ)τ
γ/π
(11.129)
=⇒ F[g (1) (τ )] = ∆2 +γ 2
g (2) (τ ) = 1
Agora, a coerência de primeira ordem decai exponencialmente, |g (1) (τ )| = e−γτ , tal que o espec-
tro tem um perfil lorentziano. Este resultado já foi derivado na Sec. ?? para a largura natural
de uma transição sujeito à emissão espontânea. Entendemos a conexão pela interpretação da
emissão espontânea como sendo induzida por flutuações do vácuo, que tem um espectro de ruı́do
branco.
Na Sec. 10.4.2 já vimos, que o alargamento por colisões pode ser tratado assumindo, que a
luz é emitida como uma superposição de ondas coerentes tendo a mesma frequência, En (t) =
eiωt+iφn (t) , mas sendo interrompidas aleatoriamente por saltos de fase,

Z X X
hE ∗ (t)E(t + τ )i = e−iωt−iφn (t) eiωt+iωτ +iφm (t+τ ) dt (11.130)
n m
XZ
= eiωτ eiφm (t+τ )−iφn (t) dt = N hEn∗ (t)Em (t + τ )iδnm .
n,m
Termos cruzados zeram. O alargamento por pressão é homogêneo, mas o fato que os pacotes
de onda são espalhados por átomos diferentes se traduz numa coerência de segunda ordem
modificada,
XZ Z ∞
∗ iωτ iφn (t+τ )−iφn (t) iωτ
hEn (t)En (t + τ )i = e e dt = e p(τ )dτ . (11.131)
n τ
A densidade de probabilidade de encontrar um intervalo coerente de duração τ é p(τ )dτ =

γc e−γc τ dτ , o que finalmente dá,
alargamento por pressão (11.132)
P iωt+iφn (t)
E(t) = ne
=⇒ g (τ ) = e −γc |τ |
(1) iωτ
(1) (11.133)
=⇒ F g (τ ) = ∆γ/π 2 +γ 2
c
g (2) (τ ) = 1 + e−γc |τ |
O espectro é um lorentziano com a largura total γ 0 = γ + γc .
2 2
1.5 1.5
g (1)
g (2)
1 1
0.5 0.5 non-classical regime
0 0
-10 0 10 -10 0 10
τ τ
Figure 11.10: (Code: LM Quantumf ields Coherences.m) Funções de correlação de primeira e

segunda ordem para (vermelho) o laser, (ciano) ruı́do de fase, (azul) alargamento por colisões,
(magenta) luz caótica e (verde) luz térmica filtrada.
Para um conjunto átomos em movimento térmico devemos permitir frequências diferentes,

En (t) = eiωn t+iφn , mas fases independentes do tempo,
Z X X
hE ∗ (t)E(t + τ )i = e−iωn t−iφn eiωm t+iωm τ +iφm dt (11.134)
n m
XZ X
−iωn t−iφn +iωm t+iωm τ +iφm
= e dt = eiωn τ .
n,m n
Termos cruzados zeram. O alargamento Doppler é inomogêneo. A densidade de probabilidade

2 2
para frequências emitidas por átomos térmicos é uma gaussiana, p(ω)dω = (2πδ)−1/2 e−(ωn −ω0 ) /2δ dω,
tal que,
amostra térmica (11.135)

R 2 2
hE ∗ (t)E(t + τ )i = N (2πδ)−1/2 eiωn τ e−(ωn −ω0 ) /2δ dωn
=⇒ iωτ −δ 2 τ 2 /2
g (1) (τ ) = eq
ln 2 − ln 2·ω 2 /δ 2 (11.136)
=⇒ F[g (1) (τ )] = πδ 2
e
−δ 2 2
g (2) (τ ) = 1 + e τ /2
Para uma linha totalmente caótica,
linha caótica (11.137)

g (1) (τ ) = δ(0)
(11.138)
F[g (1) (τ )] = 1
Luz mono-modo caótica pode ser vista como luz multi-modo incoerente, onde todos os modos
exceto um único são filtrados por um etalon. Essa luz é caracterizada por |g (1) (τ )| = 1 e
g (2) (τ ) = 2, apesar do comprimento de coerência sendo τ → ∞.
Dois modos sem flutuações interferindo satisfazem |g (1) (τ )| = 1, enquanto dois modos sem
flutuações satisfazem |g (1) (τ )| = cos 21 (ω1 − ω2 )τ .
11.3.4 The spectrum of resonance fluorescence

As funções de correlação definidas em (11.123) representam um conceito interessante para de-
scrição da fluorescência ressonante e de fenômenos como o antibunching observados na fluo-
rescência.
11.3.4.1 The Wiener-Khintchine theorem

Se a dependência de tempo de uma onda é dada por E(t), chamamos de espectro a densidade
de potência Se (ω), que depende da amplitude do campo da seguinte maneira:
Re (τ ) = hÊ ∗ (t)Ê(t + τ )it (11.139)
é a função de autocorrelação e
Se (ω) = (FRe ) (ω) = |(FE) (ω)|2 (11.140)
é o espectro. Esta relação é chamada teorema de Wiener-Khintchine. Também,

R ∞ podemos consid-
erar as quantidades normalizadas, dividindo-as por Re (0) = hÊ ∗ (t)Ê(t)it = −∞ |(FE) (ω)|2 dω =
R∞
−∞ Se (ω)dω. Obtemos,
Re (τ ) Se (ω)
g (1) (τ ) = e F (ω) = Fg (1) (ω) = . (11.141)
Re (0) Re (0)
A última grandeza é o perfil da linha. Podemos ver que, para calcular o espectro da fluorescência
ressonante, só precisamos calcular a função de correlação g (1) , isto é, as amplitudes do campo
Ê(t) que, por sua vez são relacionados aos operadores de campo (11.8). Os operadores de campo
seguem as soluções das equação de Bloch, que sendo linear, têm a seguinte forma geral,
X
ρij (t + τ ) = αijkl (τ )ρij (t) + βij (τ ) . (11.142)
k,l
A condição de traço é satisfeita, quando (i, j), (k, l) 6= (1, 1).
11.3.4.2 The quantum regression theorem

Para sermos capazes de explorar essa relação para calcular as coerências, precisamos do chamado
teorema da regressão quântica,
X X
hÂ(t + τ )i = ξi (τ )hÂi (t)i =⇒ hB̂(t)Â(t + τ )Ĉ(t)i = ξi (τ )hB̂(t)Âi (t)Ĉ(t)i .
i i
(11.143)
Example 47 (Regressão quântica aplicada à equação de Langevin): Temos,
Ȧµ = Dµ (t) + Fµ (t) (11.144)

0
hFµ (t)Fν (t)i = 2hDµν iδ(t − t ) .
Sabemos,
hAµ (t)Fν (t)i = hDµν i e hFµ (t)Aν (t)i = hDµν i (11.145)
e o teorema da regressão quântica dá,
d
hAµ (t)Aν (t0 )i = hDµ (t)Aν (t0 )i , (11.146)
dt
pois se t0 < t o termo hFµ (t)Aν (t0 )i zera para um processo Markoviano.
11.3.4.3 Bloch equation for a two-level system

A transformada de Fourier da coerência de primeira ordem, g (1) (τ ) = e−iωτ G(τ ), dá,
F (ν) = (Fg (1) )(ν) = F[e−iωτ ] ? F[G(τ )] = δ(ν − ω) ? F[G(τ )] = (FG)(ν − ω) . (11.147)
Portanto, podemos olhar para o espectro não deslocado, (FG)(ν). Como o espectro de flu-
orescência é determinado pela coerência de primeira ordem, que depende dos operadores de
campo que, em torno, dependem das populações e coerências atômicas, temos que resolver a
equação de Bloch.
Para um átomo de dois nı́veis as equações de Bloch reduzidas pela condição de normalização
(10.135) são,
    
−2γ − 2i Ω i
2Ω ρ22 0
~˙ = A~
ρ ρ + b = −iΩ −γ − i∆ 0  ρ12  +  i Ω 
2 (11.148)
i
iΩ 0 −γ + i∆ ρ21 −2Ω
com a solução (10.137), ou seja, ρ(t + τ ) = eLτ ρ(t) + (1 − eLt )L−1 b. Essa solução pode ser
colocada na forma seguinte,
X
ρkl (t + τ ) = α(kl)(mn) (τ )ρmn (t) + β(kl) (τ ) , (11.149)
(mn)
onde (mn), (kl) = (22), (12), (21) identificando,

   Lτ 
α22,22 α12,22 α21,22 (e )11 . .
α(kl)(mn) (τ ) = α22,12 α12,12 α21,12  ≡  . . . = eLτ (11.150)
α22,21 α12,21 α21,21 . . .
   
β22 ((1 − eLτ )L−1 b)1
β(kl) (τ ) = β12  ≡  .  = (1 − eLτ )L−1 b .
β21 .
Usando operadores quântica na imagem de interação, |kihl| = σ̂kl , temos,

†
hσ̂12 (t)i = hσ̂21 (t)i = ρ12 (t)eiω0 t e hσ̂22 (t)i = hσ̂12 (t)σ̂21 (t)i = ρ22 (t) . (11.151)
dando,
X
hσ̂kl (t + τ )i = e(l−k)iω0 (t+τ ) e(m−n)iω0 t α(kl)(mn) (τ )hσ̂mn (t)i + e(l−k)iω0 (t+τ ) β(mn) (τ )h1i .
(mn)
(11.152)
11.3.4.4 Correlation functions

Olhamos agora para o campo de radiação, que é relacionado ao operador do momento dipolar,
Ê − = γ σ̂21 , (11.153)
onde γ é simplesmente uma constante. Substituindo essa relação nas funções de correlação
(11.123) obtemos,
hσ̂21 (t)σ̂12 (t + τ )i hσ̂21 (t)σ̂12 (t + τ )i
g (1) (τ ) = = (11.154)
hσ̂21 (t)σ̂12 (t)i hσ̂22 (t)i
hσ̂21 (t)σ̂21 (t + τ )σ̂12 (t + τ )σ̂12 (t)i hσ̂22 (t)σ̂22 (t + τ )i
g (2) (τ ) = = .
hσ̂21 (t)σ̂12 (t)i2 hσ̂22 (t)i2
Agora, podemos calcular, deixando ξi (τ ) ≡ α(12)(mn) , Ĉ(t) ≡ 1, e B̂(t) ≡ σ̂21 (t),

P iω0 (t+τ ) e(m−n)iω0 t α iω0 (t+τ ) β
(1) (mn) e (12)(mn) (τ )hσ̂21 (t)σ̂mn (t)i + e (12) (τ )hσ̂21 (t)i
g (τ ) =
hσ̂22 (t)i
(11.155)
P (m−n)iω t
(mn) e
0 α
(22)(mn) (τ )hσ̂22 (t)σ̂mn (t)i + β(22) (τ )hσ̂22 (t)i
g (2) (τ ) = .
hσ̂22 (t)i2
Usando σ̂21 σ̂mn = σ̂2n δm1 ,
α(12)(12) (τ )hσ̂21 (t)σ̂12 (t)i + β(12) (τ )heiω0 t σ̂21 (t)i
g (1) (τ ) = eiω0 τ (11.156)
hσ̂22 (t)i
α(22)(21) (τ )heiω0 t σ̂22 (t)σ̂21 (t)i + α(22)(22) (τ )hσ̂22 (t)σ̂22 (t)i + β(22) (τ )hσ̂22 (t)i
g (2) (τ ) = .
hσ̂22 (t)i2
Voltando para o operador densidade e deixando t → ∞,

(1) iω0 τ ρ21 (∞) β(22) (τ )
g (τ ) = e α(12)(12) (τ ) + β(12) (τ ) e g (2) (τ ) = eiω0 τ , (11.157)
ρ22 (∞) ρ22 (∞)
ou seja,

[L−1 b](21)
g (1) (τ ) = eiω0 τ eLτ (12)(12)
+ (1 − eLτ )L−1 b
(12) [L−1 b] (22)
. (11.158)
[(1−eLτ )L−1 b](22)
g (2) (τ ) = eiω0 τ [L−1 b] (22)
Essas funções de correlação podem facilmente ser calculadas pela resolução numérica das equações
de Bloch (11.148). A Fig. 11.11 mostra as funções de correlação e o espectro de fluorescência
derivado por transformação de Fourier (11.141).
(a) 1 (b) 0.05 (c) 2 (d) 100
0.8 80
1.5
0
0.6 60
g (τ)
g (τ)
kk
I
1
(1)
(2)
ρ
0.4 40
-0.05
0.5
0.2 20
0 -0.1 0 0
0 5 10 0 5 10 0 5 10 -50 0 50
Γt Γt Γt Δ
Figure 11.11: (Code: LM Quantumf ields ResF luM ollowSpectrum.m) (a) Evolução temporal
da população do estado excitação. (b) Função de correlação g (1) (τ ) e (c) g (1) (τ ). (d) Espectro
de Mollow.
O espectro exibe três linhas no espectro conhecidas como o tripleto de Mollow. Note, que
o tripleto não é observado em emissão espontânea (precisamos da excitação por um laser).
A posição das linhas é facilmente entendido na imagem dos átomos vestidos: Enquanto, na
aproximação semi-clássica os autovalores são calculados com o hamiltoniano Ĥatom + Ĥlaser ,
Ĥint sendo tratado com perturbação, na imagem do átomo vestido eles são calculados por
Ĥatom + Ĥlaser + Ĥint . O que, na imagem semi-clássica é interpretado com transição de dois
fótons aparece, na imagem do átomo vestido como uma transição simples entre estados vestidos.
11.3.4.5 Weak excitation

No caso de um átomo de dois nı́veis fracamente excitado, |Ω| Γ, temos soluções analı́ticas
(10.79) das equação de Bloch. Podemos, então, pegar os coeficientes αijkl e βij e inserir dentro
das funções de correlação, dando,
g (1) (τ ) = e−iωτ , (11.159)
(2) −2γτ
g (τ ) = 1 + e − 2 cos ∆τ ,
F (ωs ) = (Fg (1) )(ωs ) = δ(ωs − ω) .
Estas funções mostram, que o espectro fica essencialmente composto pelo espalhamento de
Rayleigh na frequência da luz incidente. A luz é ’antibunched’ e decai com uma oscilação em
torno do valor 1. A contribuição da luz elasticamente espalhada domina em baixo da saturação.
11.3.4.6 Resonance fluorescence with collisions

Tomando em conta alargamento da pressão, encontramos, em fortes intensidades, um espectro
contı́nuo sobreposto ao espectro de fluorecência devido aespalhamento inelástico. As equações
11.4. BEAM SPLITTING AND QUANTUM AMPLIFICATION 313
de Bloch de dois nı́veis na presença de colisões são dadas por (10.73), onde γ 0 = γ + γcoll é a
largura da linha alargada por colisões. Dentro deste modelo e no limite Ω Γ, o espectro de
fluorescência ressonante, incluindo espalhamento elástico e inelástico, é dado por [164],
γ0 − Γ γ 0 /π Γ
F (ωs ) = + δ(ωs − ω) .
γ 0 (ω0 − ωs )2 + γ 02 γ 0
Example 48 (Four-level system): As linhas de estrôncio combinam três transições para

diferentes estados de Zeeman. Assim, a luz emitida é,
Ê − (t) = Êσ−− (t) + Êπ− (t) + Êσ−+ (t)
e as funções de correlações,
h ih i
hÊ − (t)Ê + (t + τ )i h Êσ−− (t) + Êπ− (t) + Êσ−+ (t) Êσ+− (t + τ ) + Êπ+ (t + τ ) + Êσ++ (t + τ ) i
g (1) (τ ) ≡ = h ih i .
hÊ − (t)Ê + (t)i h Êσ−− (t) + Êπ− (t) + Êσ−+ (t) Êσ+− (t) + Êπ+ (t) + Êσ++ (t) i
Rotulando os estados,
Êσ−− = γ σ̂21 , Êπ− = γ σ̂31 , Êσ−+ = γ σ̂41 ,
chegamos a,
h[γ σ̂21 (t) + γ σ̂31 (t) + γ σ̂41 (t)] [γ σ̂12 (t + τ ) + γ σ̂13 (t + τ ) + γ σ̂14 (t + τ )]i
g (1) (τ ) =
h[γ σ̂21 (t) + γ σ̂31 (t) + γ σ̂41 (t)] [γ σ̂12 (t) + γ σ̂13 (t) + γ σ̂14 (t)]i
P
α (τ )h[σ̂21 (t) + σ̂31 (t) + σ̂41 (t)] σ̂12 (t)i + β(12) (τ )hσ̂21 (t)i
P(mn) (12)(mn)
+ (mn) α(13)(mn) (τ )h[σ̂21 (t) + σ̂31 (t) + σ̂41 (t)] σ̂13 (t)i + β(13) (τ )hσ̂31 (t)i
P
+ (mn) α(14)(mn) (τ )h[σ̂21 (t) + σ̂31 (t) + σ̂41 (t)] σ̂14 (t)i + β(14) (τ )hσ̂41 (t)i
=
h[σ̂21 (t) + σ̂31 (t) + σ̂41 (t)] [σ̂12 (t) + σ̂13 (t) + σ̂14 (t)]i
α(12)(12) (τ )hσ̂21 (t)σ̂12 (t)i + β(12) (τ )hσ̂21 (t)i
+α(13)(13) (τ )hσ̂31 (t)σ̂13 (t)i + β(13) (τ )hσ̂31 (t)i
+α(14)(14) (τ )hσ̂41 (t)σ̂14 (t)i + β(14) (τ )hσ̂41 (t)i
=
h[σ̂21 (t) + σ̂31 (t) + σ̂41 (t)] [σ̂12 (t) + σ̂13 (t) + σ̂14 (t)]i
α(12)(12) (τ )ρ22 (∞)i + β(12) (τ )ρ21 (∞) + α(13)(13) (τ )ρ33 (∞)
+β(13) (τ )ρ31 (∞) + α(14)(14) (τ )ρ44 (∞) + β(14) (τ )ρ41 (∞)
= .
ρ22 (∞) + ρ33 (∞) + ρ44 (∞) + ρ23 (∞) + ρ32 (∞) + ρ24 (∞) + ρ42 (∞) + ρ34 (∞) + ρ43 (∞)
11.4 Beam splitting and quantum amplification

O divisor de feixe é um dispositivo quântico geral que mistura dois modos representados por
operadores de campo â e b̂ de acordo com o hamiltoniano,
Ĥ = ~2 Ω(âb̂† + â† b̂) . (11.160)
Ele pode ser descrito na imagem de Schrödinger ou de Heisenberg.

11.4.1 The beam splitter in various representations

11.4.1.1 Schrödinger picture
Na imagem de Schrödinger, se |ψ0 i = |α0 i|β0 i é o estado dos modos antes do divisor de feixe, a
equação de Schrödinger, i~∂t |ψ(t)i = Ĥ|ψ(t)i, nós dá o estado depois do divisor pela solução,
† +â† b̂)
|ψ(t)i = eiΩt/2(âb̂ |ψ0 i . (11.161)
Um divisor de 50% corresponde à um pulso Ωt = π/2. .
Figure 11.12: Beam splitter
11.4.1.2 Heisenberg picture

Também podemos descrever o divisor de feixe na imagem de Heisenberg. Com as regras de
comutação,
[â, â† ] = 1 = [b̂, b̂† ] , [â, b̂] = 0 , (11.162)
e as equações de Heisenberg,
â˙ = ~i [Ĥ, â] = 2i Ω[(âb̂† + â† b̂), â] = − 2i Ωb̂ (11.163)

˙
b̂ = ~i [Ĥ, b̂] = 2i Ω[(âb̂† + â† b̂), b̂] = − 2i Ωâ .
Com isso calculamos,

¨ = − 1 Ω2 â ¨
â 4 e b̂ = − 14 Ω2 b̂ , (11.164)
com a solução,

â(t) cos 12 Ωt −i sin 21 Ωt â0
= . (11.165)
b̂(t) −i sin 21 Ωt cos 12 Ωt b̂0
p
Introduzindo a abreviação η ≡ cos(Ωt/2), podemos descrever a evolução como,
√ √
â(t) η −i 1 − η â0
= √ √ . (11.166)
b̂(t) −i 1 − η η b̂0
Percebemos 11 , que o feixe refletido na superfı́cie de um meio opticamente mais denso sofre
um deslocamento de fase de π/2.
11
Este fato é uma consequência da time-reversal invariance no divisor de feixe.
11.4.1.3 Glauber representation

Na imagem de Heisenberg, as funções de onda dos estados quânticos (e portanto o operador
densidade e a função de Wigner) permanecem inalteradas durante a evolução, i.e. ,
|ψi ≡ |αi|βi = |α0 i|β0 i ≡ |ψ0 i , (11.167)
ρ|ψi = ρ|ψ0 i ,
W|ψi (γ) = W|ψ0 i (γ) ,
P|ψi (γ) = P|ψ0 i (γ) ,
Q|ψi (γ) = Q|ψ0 i (γ) ,
onde os |ψi são estados quânticos arbitrários e |γi são estados de Glauber. Isto significa, que dois
modos de campo misturados num divisor de feixe não interferem no espaço de fase, i.e. não de-
senvolvem correlações quânticas. De fato, isso seria fácil demais; veremos em breve na Sec .11.6,
que precisamos trabalhar um pouco mais para produzir correlações quânticas.
11.4.1.4 Fock representation

Alternativamente, podemos descrever o divisor de feixe na representação de Fock 12 . O hamilto-
niano do divisor de feixe acopla dois modos de osciladores harmônicos. Expandindo em uma base
de Fock dois-dimensional por |ψi ≡ (...|0i|na i....|nb i|0i...) podemos ganhar mais introspecção:
Ĥ = ~2 Ω(âb̂† + â† b̂) (11.168)
X p p
~
= 2Ω na (nb + 1)|na − 1, nb + 1ihna , nb | + (na + 1)nb |na + 1, nb − 1ihna , nb |
na ,nb
X
= Ĥa+b ,
na +nb
onde,  √ 
0 1nb p
√
 1nb 
 0 2(nb − 1) 
 p .. 
Ĥa+b = ~2 Ω 
 2(nb − 1) .  .
 (11.169)
 .. √ 
 . na 1
√
na 1 0
Os sub-espaços com na +nb +1 fótons são completamente degenerados, pois det(λIa+b −Ha+b ) =
λna +nb +1 = 0. A degenerescência é removida, quando introduzimos mecanismos de perda para
um dos modos. Assim, o hamiltoniano pode ser entendido como um sistema de Dicke com
multiplicidade 1 (n\+ n ) = n + n + 1.
2 a b a b
Example 49 (Divisor de feixe com 0 ou 1 fóton): Pegamos, por exemplo na , nb = 0, 1.

t
Então, na base |0, 0i |0, 1i |1, 0i |0, 2i |1, 1i |2, 0i · · · a matriz do hamiltoniano
fica,  
0
 0 1 
 
 1 0 
1 
Ĥ = 2 Ω  √  .
0 2 0 
 √ √ 
 2 √0 2
0 2 0
12
Uma discussão mais profunda se encontra na Ref. [160].
É facilmente verificado, que a matrix dos autovetores e a matriz dos autovalores,

   
1 0 0 0 0 0 0 0 0 0 0 0
0 −1 1 0 0 0  0 −1 0 0 0 0
   
0 1 1 0√ 0 0  ~ 0
 0 1 0 0 0
U =
0
√ 
 respetivamente E = Ω 0

 0 0 − 12 2 −1 1
2 2
2
 0 0 −2 0 0

0 0 0 1√ 0 1  0 0 0 0 0 0
√
0 0 0 − 12 2 1 1
2 2
0 0 0 0 0 2
satisfazem U −1 HU = E. Portanto,
|ψi = U −1 eiEt U |ψ0 i ,

T
e achamos, que o estado |1, 1i = 0 0 0 0 1 0 é transformado numa superposição,
T i
Ωt=π/2
0 0 0 √1 i sin Ωt cos Ωt √1 i sin Ωt −→ √ (|0, 2i + |2, 0i) .
2 2
2
T
Similarmente, achamos que o estado de superposição √1 (|1, 0i+|0, 1i) = 0 1 −1 0 0 0
2
é transformado em
1 T Ωt=π/2 1
e− 2 iΩt 0 1 −1 0 0 0 −→ e−iπ/4 √ (|1, 0i + |0, 1i) .
2
11.4.2 Fock and Glauber states at a beam splitter

O divisor de feixe divide um estado de Fock contendo N fótons em dois estados de Glauber,
XN
1 N 1
|ψi = √ + (â†1= â†2 )N |0, 0i (â†1 )n (â†2 )N −n |0, 0i
√ (11.170)
2 N/2 N! 2 N/2 N ! n=0 n
s s
N N
1 X N X N
= N/2 |n, N − ni = 0.5n 0.5N −n |n, N − ni
2 n n
n=0 n=0
N
r n
X n
(N/2) −N/2 X αn
' e |n, N − ni = e−|α|/2 √ |n, N − ni ,
n! n!
n=0 n=0
usando a aproximação da distribuição binomial pela distribuição de Poisson,

n
N n N →∞ (pN ) −pN
p (1 − p)N −n −→ e . (11.171)
n n!
e definindo α ≡ N/2. A normalização é hψ|ψi = 1. A população de um modo individual é,

s s
N
X N
1 N † N 1 X N N
hn̂1 i = hψ|â†1 â1 |ψi = hm, N − m| â1 â1 |n, N − ni = N n= .
2N m n 2 n 2
n,m=0 n=0
(11.172)
O resultado (11.170) mostra que, ignorando um dos modos, o outro modo fica automaticamente
num estado de Glauber.
Além disso,
n N −1
1 X N 2 N X N N (N + 1)
hn̂21 i = n n = n (n + 1) = (11.173)
2 n 2 n 4
n=0 n=0
s
n
1 X N N (N − 1)
hn̂1 n̂2 i = n n(N − n) = N hn̂1 i − hn̂21 i =
2 n 4
n=0
O parâmetro de ’squeezing’ é,

σ 2 (n̂1 − n̂2 ) hn̂2 − 2n̂1 n̂2 + n̂22 i 2hn̂21 i − 2hn̂1 n̂2 i
ξ12 = = 1 = =1. (11.174)
hn̂1 i + hn̂2 i N N
As funções de correlação para tempos iguais são,
D E
â†1 â†1 â1 â1 n
1 1 X N hn̂21 i − hn̂1 i
g11 = = n(n − 1) = =1 (11.175)
hn̂1 i2 hn̂1 i2 2n n hn̂1 i2
n=2
D E
â†1 â†2 â2 â1 N
1 1 X N N hn̂1 i − hn̂21 i N −1
g12 = = 2 n
n(N − n) = 2
=
hn̂1 ihn̂2 i hn̂1 i 2 n hn̂1 i N
n=0
As inigualdades de Cauchy-Schwarz e a inigualdade quântica ambas são satisfeitas,

√
g12 ≤ g11 g22 (11.176)
s
1 1
g12 ≤ g11 + g22 +
hn̂1 i hn̂2 i
Em comparação, um estado de Glauber é dividido normalmente,

2 /2
X αn 2
X αm
|ψi = |α1 i|α2 i = e−|α1 | √ 1 |nie−|α2 | /2 √ 2 |mi (11.177)
n n! m m!
2 2
X n
α α m
= e−|α1 | /2−|α2 | /2 √ 1 √2 |ni|mi .
n,m n! m!
11.4.2.1 Density matrix representation

A matriz densidade para um estado puro é,

|hψ|1i|2 h1|ψihψ|2i
ρ̂ = |ψihψ| = . (11.178)
h2|ψihψ|1i |hψ|2i|2
A evolução de um tal estado é descrito pela equação de von Neumann:
i~∂t ρ̂(t) = [Ĥ, ρ̂(t)] . (11.179)
Para o divisor de feixe obtemos,

 
|hψ|0, 0i|2 h0, 0|ψihψ|0, 1i h0, 0|ψihψ|1, 0i
h0, 1|ψihψ|0, 0i |hψ|0, 1i|2 h0, 1|ψihψ|1, 0i 
 
ρ̂ = |ψihψ| = h1, 0|ψihψ|0, 0i h1, 0|ψihψ|0, 1i |hψ|1, 0i|2  . (11.180)
 
..
.
Example 50 (Matriz densidade para o divisor de feixe com 0 ou 1 fóton): E.g. para
estados de superposição, |ψi = √12 (|0, 1i ± |1, 0i),
 
0 0 0
0 1
± 12 
 2 
ρ̂ = 0 ± 12 1  .
 2 
..
.
Obviamente, ρ̂ = ρ̂2 . Para o estado de superposição acima, ∂t ρ̂(t) = 0.
11.4.3 Shot-noise
O limite de Heisenberg nas fases de quadratura de um campo de luz determina o ruı́do de shot-
noise na intensidade do feixe de luz. Para medir este ruı́do dividimos o feixe por uma separatriz
em dois feixes â e b̂ e recombinamos estes feixes numa segunda separatriz,
ĉ = √1 (â
2
+ b̂) , dˆ = √1 (â
2
− b̂) . (11.181)
Detectados por fotodetectores com o coeficiente de ganho g estes dois feixes produzem correntes,
Iˆc = gĉ† ĉ , Iˆd = g dˆ† dˆ . (11.182)
Adicionamos e subtraı́mos estes foro correntes,
Iˆ+ ≡ Iˆc + Iˆd = g(â† â + b̂† b̂) , Iˆ− ≡ Iˆc − Iˆd = g(â† b̂ + b̂† â) . (11.183)
Iˆ+
2
= g 2 [n̂2a + n̂2b + 2n̂a n̂b ] , Iˆ−
2
= g 2 [(â† b̂)2 + (b̂† â)2 + â† â + b̂b̂† + ââ† + b̂† b̂] . (11.184)
Os valores esperados são,
hIˆ+ i = ghn̂a ia , hIˆ+ i = 0 (11.185)

hIˆ+
2
i = gh(â† â)2 ia , hIˆ−
2
i = g 2 hn̂a ia . (11.186)
ˆ 2 i ≡ hIˆ2 i − hIi
Agora, com a definição, h(∆I) ˆ 2 , obtemos o ruı́do de intensidade do campo,
h∆Iˆ+
2
i = h(∆Iˆ+ )2 i = g 2 h(∆n̂a )2 i , (11.187)
e o ruı́do de shot noise

h∆Iˆ−
2
i = g 2 Ia . (11.188)
Vide [tèse de PhD do Hans Marı́n Flores]. .
11.4.4 Quantum amplifier

Um amplificador quântico é um amplificador baseado em métodos de mecânica quântica para
amplificar um sinal quântico; exemplos incluem os elementos ativos de lasers e amplificadores
ópticos de pulsos coerentes. O amplificador quântico é caracterizado por seu ganho e seu ruı́do
quântico. Esses parâmetros são interdependentes; quanto maior o coeficiente de amplificação,
maior a incerteza (ruı́do). No caso dos lasers, a incerteza corresponde à emissão espontânea
amplificada do meio ativo.
Figure 11.13: Shot noise
O sinal quântico é tipicamente o modo de um campo elétrico (representado por um operador

de aniquilação â) tendo o valor médio hâiin . A amplificação quântica é uma transformação
unitária Û , atuando em um estado inicial |ini e produzindo (na figura de Schrödinger) o estado
amplificado,
|outi = Û |ini . (11.189)
The amplification depends on the mean value hâi of the anihilation operator and its dispersion
hâ† âi − hâ† ihâi. A coherent state is a state with minimal uncertainty; when the state is trans-
formed, the uncertainty may increase. This increase can be interpreted as noise in the amplifier.
The gain G can be defined as follows:
hâiout
G= . (11.190)
hâiin
The quantum amplifier can also be described in the Heisenberg picture; the changes are
attributed to the amplification of the field operator. Thus, the evolution of the operator âout is
given by
âout = Û † âÛ , (11.191)
while the state vector remains unchanged. The gain is then given by,
hâout iin
G= . (11.192)
hâiin
In general, the gain G may be complex, and it may depend on the initial state. For laser
applications, the amplification of coherent states is important. Therefore, it is usually assumed
that the initial state is a coherent state characterized by a complex-valued initial parameter α,
such that |ini = |αi. Even with such a restriction, the gain may depend on the amplitude or
phase of the initial field.
In the following, the Heisenberg representation is used; all brackets are assumed to be eval-
uated with respect to the initial coherent state,
noise = hâ†out âout i − hâ†out ihâout i − (hâ† âi − hâ† ihâi) . (11.193)
The expectation values are assumed to be evaluated with respect to the initial coherent state.
This quantity characterizes the increase of the uncertainty of the field due to amplification. As
the uncertainty of the field operator does not depend on its parameter, the quantity above shows
how much output field differs from a coherent state.
11.4.4.1 Linear phase-invariant amplifier

Linear phase-invariant amplifiers may be described as follows. Assume that the unitary operator
U amplifies in such a way that the input â and the output âout = Û † âÛ , are related by a linear
equation,
âout = câ + sb̂† , (11.194)
where c and s are c-numbers and b̂† is a creation operator characterizing the amplifier. Without
loss of generality, it may be assumed that c and s are real. The commutator of the field operators
is invariant under unitary transformation U :
[âout , â†out ] = 1 = [â, â† ] , [ain , a†out ] = 0 = [aout , a†in ] . (11.195)
From the unitarity of U , it follows that b̂ satisfies the same commutation relations. The c-
numbers are then c2 − s2 = 1. Hence, the phase-invariant amplifier acts by introducing an
additional mode to the field, with a large amount of stored energy, behaving as a boson. Calcu-
lating the gain and the noise of this amplifier, one finds G = c, and
noise = c2 − 1 . (11.196)
The coefficient g = |G|2 is sometimes called the intensity amplification coefficient. The noise
of the linear phase-invariant amplifier is given by g − 1. The gain can be dropped by splitting
the beam; the estimate above gives the minimal possible noise of the linear phase-invariant
amplifier. The linear amplifier has an advantage over the multi-mode amplifier: if several modes
of a linear amplifier are amplified by the same factor, the noise in each mode is determined
independently;that is, modes in a linear quantum amplifier are independent.
To obtain a large amplification coefficient with minimal noise, one may use homodyne de-
tection, constructing a field state with known amplitude and phase, corresponding to the linear
phase-invariant amplifier. The uncertainty principle sets the lower bound of quantum noise in
an amplifier. In particular, the output of a laser system and the output of an optical generator
are not coherent states.
The multiplicative amplifier D also adds additive noise F . We have DD† = 1,
† †
aout a F1
= D in + . (11.197)
aout ain F2
11.4.5 Homodyne detection and inverse Radon transform

In the method of homodyne detection or phase-sensitive detection the signal is obtained by
superposing the field mode of interest with a local oscillator with a relative phase θ at a beam
splitter and a subtraction of the photo currents in the two ports of the interferometer:
∆Jˆ = â†t ât − b̂†t b̂t (11.198)

p η→1/2
= (2η − 1)(â†0 â0 + b̂†0 b̂0 ) + 2 (1 − η)η(â†0 â0 + b̂†0 b̂0 ) −→ â†0 â0 + b̂†0 b̂0 . (11.199)
If the local oscillator is a classical light field b̂0 = αLO e−iθ ,
∆Jˆ = |αLO |(â0 e−iθ + â†0 eiθ ) (11.200)

√
= 2|αLO |x̂θ , (11.201)
11.5. THE LASER 321
where the field mode is expressed by the Hermitian quadrature components â0 = 2−1/2 ·(x̂θ +iŷθ ).
The expectation value of ∆Jˆ is afflicted with the Heisenberg uncertainty and can be expressed
as the first moment of the Wigner function W (α):
Z
ˆ
hψ|∆J|ψi = W|ψi (α)∆Jd ˆ 2α (11.202)
√ Z
= 2|αLO | W|ψi (xθ , pθ )xθ dxθ dpθ (11.203)
√ Z ∞
= 2|αLO | wθ (xθ )xθ dxθ .
−∞
Here, the distribution function integrated over a rotated quadrature component pθ is given by,
Z ∞
wθ (xθ ) ≡ Wθ (xθ , pθ )dpθ . (11.204)
−∞
This is called the radon transform. The distribution function wθ (pθ ) as well as the Wigner
function are normalized to 1. Multiple measurements of the expectation value R xθ = hψ|x̂θ |ψi now
yields a histogram H|ψi (xθ ) reflecting, if normalized, w|ψi (xθ ) = H|ψi (xθ )/ H|ψi (xθ ) exactly the
distribution function.
Considering the finite detector efficiency [160] wθ (xθ ) must be generalized to a convolution
with an apparatus function ζ(x):
√ 1 2
wθreal ( ηxθ ) = (wθideal ? ζ) onde ζ(x) = p e−ηα /(1−η) . (11.205)
π(1 − η)
A finite detector efficiency degrades the contrast of the quantum interference structures.
With the procedure of optical homodyne tomography or quantum state endoscopy the Wigner
function for
R e.g. a Schrödinger cat state can be reconstructed from a set of distribution functions
wθ (xθ ) = W (αeiθ )dpθ measured for various phases θ [160]. To do this the data set is exposed
to an inverse radon transform:
Z ∞ Z πZ ∞
1
wθ (xθ )|ζ|eiζ [Re (2 θ]
−1/2 αe−iθ )−x
W (α) = 2 dxdθdζ . (11.206)
4π −∞ 0 −∞
In contrast to the conventional homodyne detection, where the phase dependency of amplitude
noise is recorded, the homodyne tomography allows the complete reconstruction of a quantum
state through measurement of the distribution of the amplitude noise power,
Z
ωs ωs /2π
Ps = dt|hI(t)i|2 (11.207)
2π 0
for various phases.
Alternatively, to the homodyne method, one may reconstruct the photon distribution in field
modes from their temporal evolution [250]. Another method could be to use atoms as sensors
for the quantum state of a light field in a Jaynes-Cummings type dynamics.
11.5 The laser

O laser é um dispositivo emitindo luz com propriedades insuperáveis, que permitiram avanços
técnicos e cientı́ficos incomparáveis (ou só comparáveis com aqueles devido a invenção do tran-
sistor).
O laser produz a luz através de um processo de amplificação óptica baseado na emissão

estimulada de radiação electromagnética. O termo ’laser’ originou-se como um acrônimo para
’Amplificação de Luz por Emissão eStimulada de Radiação’. Um laser difere de outras fontes
de luz em que emite luz de forma coerente. A coerência espacial permite a luz a ser focalizada
em um ponto, a concentração de energia neste ponto permite aplicativos como o corte a laser e
litografia. Coerência espacial também permite que um feixe de laser fique colimado em grandes
distâncias, isto é, a luz emitida forma um feixe concentrado numa linha que se propaga em
direção reta. Lasers também podem ter alta coerência temporal, o que lhes permite emitir luz
com um espectro muito estreito, isto é, eles podem emitir uma única cor de luz. Coerência
temporal pode ser usada para produzir pulsos de luz tão curto quanto um femtosegundo.
Em 1917, Albert Einstein estabeleceu os fundamentos teóricos do laser num artigo ’Zur
Quantentheorie der Strahlung’ (’Sobre a Teoria Quântica da Radiação’) através de uma re-
derivação da lei da radiação de Max Planck. Ele propus um mecanismo explicando como a luz é
absorvida e emitida de átomos. O ingrediente fundamental erá, que o fóton pode ser emitido de
dois maneiras diferentes, por emissão espontânea, que é um processo que ocorre sem causa fı́sica,
ou por emissão estimulada. A emissão estimulada ocorre devido à luz já presente no sistema e
representa o mecanismo fundamental do laser. Na década subsequente pesquisadores alemães e
americanos confirmaram experimentalmente os fenômenos de emissão estimulada e da absorção
negativa, isto é, do ganho. Em 1950, Alfred Kastler (fı́sico francês e Prêmio Nobel de Fı́sica
de 1966) propus o método de bombeamento óptico, confirmado experimentalmente dois anos
depois por outros fı́sicos franceses.
Em 1953, Charles Townes produziu o primeiro amplificador de micro-ondas chamado de
maser, um dispositivo que opera de maneira semelhante ao laser, mas amplificando radiação
de microondas em vez de radiação infravermelha ou visı́vel. No entanto, o maser do Townes
erá incapaz emitir luz de maneira contı́nua. Em 1955, na União Soviética, Nikolay Basov e
Aleksandr Prokhorov resolveram o problema da operação contı́nua usando átomos com mais de
dois nı́veis de energia. Estes sistemas de nı́veis permitiram realizar uma inversão de população
sustentada de um nı́vel energético podendo decair num sistema menos energético liberando luz
por emissões estimuladas. Townes relata que vários fı́sicos eminentes, entre eles Niels Bohr,
John von Neumann e Isidor Rabi, argumentaram que o maser viola o princı́pio da incerteza
de Heisenberg e, portanto, não poderia funcionar. Em 1964, Charles Townes, Nikolay Basov
e Aleksandr Prokhorov dividiram o Prêmio Nobel de Fı́sica, ’para o trabalho fundamental na
área da eletrônica quântica que levou à construção de osciladores e amplificadores baseados no
princı́pio do maser a laser’.
Em 1957, Charles Townes e Arthur Schawlow, dos laboratórios da Bell Labs, começaram
seriamente um estudo do laser. O conceito foi originalmente chamado de ’laser ótico’. Em 1958,
a Bell Labs submetiou uma patente para a sua radiação óptica propostas, e Schawlow e Townes
apresentaram um artigo cientı́fico. Simultaneamente, na Universidade de Columbia, o estudante
de pós-graduação Gordon Gould estava trabalhando em uma tese de doutorado sobre os nı́veis
de energia do tálio excitado. Em 1957-8, Gould e independentemente Prokhorov, Schawlow
e Townes propuseram o uso de um ressonador aberto (que virou mais tarde uma componente
essencial do laser). O termo laser foi inventado por Gould. Também, ele propus várias possı́veis
aplicações para um laser como a espectrometria, interferometria, radar e a fusão nuclear. Ele
continuou a desenvolver a ideia, e entrou com um pedido de patente em abril de 1959. O
Escritório de Patentes dos Estados Unidos indeferiu o seu pedido, e concedeu uma patente para
a Bell Labs, em 1960. Gould ganhou sua primeira patente menor em 1977 após uma briga de
28 anos, e ainda não foi até 1987 que ele ganhou a primeira vitória significativa processo de
patente, quando um juiz federal ordenou que o Escritório de Patentes dos Estados Unidos a
11.5. THE LASER 323
emitir patentes de Gould para o bombeamento óptico e o laser baseado no princı́pio de descarga
elétrica num gás. A questão de como atribuir crédito para inventar o laser continua por resolver
pelos historiadores.
Em 16 de maio de 1960, Theodore Maiman operou o primeiro laser funcionando, no Hughes
Research Laboratories, Malibu, Califórnia, à frente de várias equipas de pesquisa, incluindo as
do Townes na Universidade de Columbia, do Schawlow na Bell Labs e do Gould na empresa
TRG (Grupo Técnico de Investigação). O laser operacional do Maiman utilizou um cristal de
estado sólido de rubi sintético bombeado por uma lâmpada de relâmpago para produzir luz laser
vermelho a 694 nm de comprimento de onda; no entanto, o dispositivo só era capaz de forma
pulsada, devido ao seu sistema de criação de bombeamento de três nı́veis. Mais tarde em 1960,
foi construı́do o primeiro laser de gás, usando hélio e neon que foi capaz de operação contı́nua no
infravermelho. Basov e Javan propuseram o conceito de diodo laser semicondutor. Em 1962, foi
realizado o primeiro dispositivo de dı́odo laser, feito de arseneto de gálio emitindo luz perto do
infravermelho. Hoje em dia existem diodos laser em vários regimes até o ultravioleta ’blue-ray’.
Interessantemente, apesar de muitas tentativas, ainda não foi possı́vel fabricar diodos laser
amarelos ou verdes.
11.5.1 Features and operation of lasers

Geralmente o laser tem uma frequência (uma cor) extremamente bem definida. Além, disso a
luz de um laser é direcionada, polarizada e coerente.
Para entender o funcionamento de um laser consideramos o processo de absorção e de emissão

de luz por um átomo. Seguinte o modelo de Bohr um fóton absorvido levanta um elétron de uma
órbita inferior para uma órbita superior, e quando o elétron volta para o estado fundamental,
ele reemite um fóton em uma direção arbitrária.
Quando irradiamos uma amostra de N átomos N1 átomos no estado fundamental por um
campo de radiação com a intensidade I(ν), a taxa de depende através de uma constante B12 ,
que é caracterı́stica para a transição, do produto
Rabs ∝ B12 I(ν)N1 . (11.208)
Figure 11.14
A taxa de emissão depende do número de átomos N2 no estado excitado, tal que
Rsp ∝ A21 N2 . (11.209)

Como o estado excitado tem mais energia, ele pode obviamente decair sozinho (isto é, espon-
taneamente) para um estados de energia inferior. A ideia brilhante do Einstein agora foi de
postular um terceiro processo, que ele chamou de emissão estimulada,
Rst ∝ B21 I(ν)N2 . (11.210)
Neste processo, um fóton incidente estimula um átomo excitado de transferir o elétron para
uma órbita inferior. A energia liberada é então usada para formar um segundo fóton em todos
os respeitos idêntico ao primeiro. Este processo é necessário para garantir que em equilı́brio
térmico a população dos estados segue a lei de Boltzmann.
Figure 11.15
Obviamente, absorção diminuı́ a intensidade de um feixe de luz atravessando a amostra

atômica, enquanto emissão estimulada amplifica a luz. Para amplificar a luz incidente, os ganhos
de intensidade devem superar as perdas. Portanto, precisamos que os processos de absorção
sejam menos frequente do que os processos de emissão estimulada, ou seja o número de átomos
no estado excitado N2 deve superar o número de átomos no estado fundamental N1 < N2 .
Podemos facilmente, escrever a equação de taxas,
dN2 dN1
= −A21 N2 − B21 I(ν)N2 + B21 I(ν)N1 = − , (11.211)
dt dt
com N = N1 + N2 . É fácil resolver está equação. O resultado é,
I(ν)B21 N
N2 = [1 − e−(A21 +2B21 I(ν))t ] < N1 . (11.212)
A + 2B21 I(ν)
A representação gráfica mostra o comportamento temporal das populações N1 (em verde na

figura) e N2 (em azul), que alcance um estado de equilı́brio após um certo tempo. Aumentando
a intensidade da luz incidente, observamos que as curvas se aproximam, mas nunca se cruzam.
Isto é, em um sistema de dois nı́veis, sempre vale N1 > N2 e as populações nunca são invertidas.
Portanto, amplificação da luz como no laser não acontece.
Felizmente, podemos recorrer a um truque inserindo um terceiro nı́vel. Garantindo que a
taxa de decaimento do estado E3 é muito mais lenta do que o bombeamento óptica para este
estado através da transição excitada E1 −→ E2 e depois de decaimento rápido E2 −→ E3 ,
podemos alcançar a situação N3 > N1 . Agora, é possı́vel amplificar uma luz ressonante com a
transição E2 −→ E3 .
Quais são as condições mı́nimas que precisamos satisfazer para realizar um laser? A primeira
condição é, que o ciclo de bombeamento seja irreversı́vel para garantir que os processos de
emissão estimulada e absorção não se compensam. A emissão espontânea é irreversı́vel é pode
ser incorporada num sistema de três nı́veis.
A segunda condição é a existência do processo de emissão estimulada, porque queremos que
o fóton amplificado seja uma cópia exata do fóton incidente.
11.5. THE LASER 325
1000
800 aumentando I(ν)
600
N1 , N2
400
200
aumentando I(ν)
0
0 1 2 3 4 5
tempo
Figure 11.16
A terceira condição é um mecanismo de retroação que sincroniza os processos de amplificação

pelos átomos dentro de um meio desordenado, como por exemplo um gás. Espelhos são ideais,
pois aumentam a distância dentro da qual átomos invertidos podem amplificar a luz. Também,
os espelhos definem a fase da onda de luz, pois a onda estacionária formada pelo campos de luz
contrapropagantes deve ter nós nas superfı́cies dos espelhos.
Comparando as propriedades de uma luz incandescente e de um laser ...
11.5.2 Applications of lasers

Entre suas muitas aplicações, lasers são usados em leitores ópticos de disco, impressoras a laser e
scanners de código de barras; fibra óptica e comunicação óptica em espaço livre; cirurgia a laser
e tratamentos de pele; soldadura e corte de materiais; dispositivos militares para a marcação de
alvos; medição de distância e de velocidade; e projetores, laser pointers, em CDs e DVDs e para
usinagem de materiais.
Na pesquisa fundamental (em particular, com gases atômicos, metamateriais, etc..) o laser
tem importância elementar nas áreas de metrologia, relógios atômicos, 1000 vezes mais estáveis
do que os melhores relógios atuais, pentes de frequência, computadores quânticos ou fotônicos.
11.5.2.1 Use of lasers in cold atoms laboratories

A luz é uma onda. Com o laser encontramos um processo e um dispositivo para render esta luz
coerente. Do outro lado, desde de Broglie sabemos que a matéria também é uma onda. Será
que podemos fazer um laser de matéria.
De fato, isto é possı́vel. A primeira onda de matéria coerente foi observada em 1995. Este
estado da matéria, também se chama condensado de Bose-Einstein, e foi prevista por Bose e
Einstein em 1924. Para criar um condensado de Bose-Einstein, precisamos similarmente ao
laser, que as ondas interferem construtivamente de maneira que elas se amplificam. Para isso, é
necessário que o comprimento de onda de Broglie das partı́culas seja maior do que a distância
entre elas. Assumindo uma distancia média é da ordem de µm, isto corresponde a uma velocidade
dos átomos de mm/s ou seja uma temperatura de alguns 100 nK.
Em nossos laboratórios, estudamos o comportamento de matéria ultrafria e particularmente a
maneira de como a luz interage com esta matéria. Apesar dessa interação ser muito fundamental
e estudada deste muito tempo, ainda tem muitas questões abertas. Por exemplo, ainda não é
Figure 11.17
muito claro como os átomos, que são os constituintes elementares, de qualquer objeto de matéria
devem cooperar para produzir os fenômenos bem conhecidos de espalhamento de luz, como a
refração, a difração ou o espalhamento de Bragg ou de Mie.
A nossa ideia é, que o melhor sistema para estudar essas questões são nuvens de átomos
resfriados para temperaturas extremamente baixas. Uma temperatura baixa significa, que o
movimento dos átomos fica muito lento, tão lento que o movimento não perturba a interação e
atrapalha as medidas. Assim, podemos medir a interação coletiva entre a luz e os átomos sob
condições puras, autênticas.
Nós jogamos agora um pulso curto de um feixe laser e observamos a reação dos átomos que
nos estudamos em circunstâncias variadas.
11.6 Squeezed states

11.6.1 Squeezing operator
Em analogia com o operador de deslocamento (3.91), podemos analisar outro operador dado
por,
†2 /2−ξ ∗ â2 /2)
Q̂(ξ) ≡ e(ξâ , (11.213)
e que chamaremos de operador de compressão ou operador de squeezing. Em analogia com o
cálculo (3.92), usando as regras de comutação, é possı́vel verificar a unitariedade deste operador
[165] (vide Exc. 11.8.6.1).
∗
In particular, usando a relação (2.162) e a abreviação Â ≡ 2ξ â†2 − ξ2 â2 , podemos mostrar
[237] (vide Exc. 11.8.6.2),
h i h h ii
Q̂† (ξ)âQ̂(ξ) = eÂ âe−Â = â + [Â, â] + 2!1 Â, [Â, â] + 3!1 Â, Â, [Â, â] ... (11.214)
|ξ| †
= â + ξâ† + 1 ∗
2! ξξ â + 1 ∗ †
3! ξξ ξâ + ... = â cosh |ξ| + â sinh |ξ| ,
ξ∗
11.6. SQUEEZED STATES 327
e similarmente para â† , tal que com ξ ≡ reiϕ ,
Q̂† (ξ)âQ̂(ξ) = â cosh r + e−iϕ â† sinh rQ̂† (ξ)â† Q̂(ξ) = â† cosh r + eiϕ â sinh r .
e
(11.215)
As formulas (11.215) descrevem uma transformação de Bogolubov. Resolve o Exc. 11.8.6.3.
11.6.1.1 Single-mode Bogolubov transform

Um par de operadores de aniquilação e criação satisfazendo a relação de comutação [â, â† ] = 1
pode ser convertido em um outro par de operadores por uma transformação de Bogolubov,
b̂ ≡ uâ + vâ† , b̂† ≡ u∗ â† + v ∗ â , (11.216)

para números complexos u e v. Postulando a mesma relação de comutação para os novos
operadores, [b̂, b̂† ] = 1, obtemos imediatamente a condição,
|u|2 − |v|2 = 1 . (11.217)
2 2
Comparando com a identidade hiperbólica cosh r − sinh r = 1, podemos parametrizar as
constantes como,
u = cosh r e v = eiϕ sinh r . (11.218)
Isto é interpretado como transformação linear simpléctica do espaço de fase.
11.6.1.2 Squeezing of the uncertainty relation

As partes real e imaginária dos operadores transformados, definidas por b̂ = x̂b + iŷb , são
hermitianas e satisfazem a relação de incerteza de Heisenberg (2.85),
i 1 1
[x̂b , ŷb ] = 2 e ∆x̂b ∆ŷb ≥ 2i h[x̂b , ŷb ]i = 4 . (11.219)
No entanto, vamos dar uma olhada às incertezas separadamente. Eles se relacionam ao modo
de Glauber por,
x̂b = 12 (b̂ + b̂† ) = 12 (â cosh r + â† eiϕ sinh r) + 12 (â† cosh r + âe−iϕ sinh r) (11.220)
ϕ=0
= x̂a cosh r + 21 sinh r(â† eiϕ + âe−iϕ ) −→ x̂a er
ŷb = 1
2i (b̂ − b̂† ) = 1
2i (â cosh r + â† eiϕ sinh r) − 1 †
2i (â cosh r + âe−iϕ sinh r)
ϕ=0
= ŷa cosh r + 1
2i sinh r(â† eiϕ − âe−iϕ ) −→ ŷa e−r .
As flutuações individuais são (assumindo ϕ = 0),

∆x̂2b = hx̂2b i − hx̂b i2 = e2r hx̂2a i − hx̂a i2 (11.221)

= 14 e2r 1 + hâ2 i + hâ† i2 + 2hâ† âi − hâi2 − hâ† i2 − 2hâ† ihâi .
Com hâi = 0,

∆x̂2b = e2r 1
2 Re hâ2 i + 1
4 + 12 hâ† âi (11.222)

∆ŷb2 = e−2r − 12 Re hâ2 i + 1
4 + 12 hâ† âi .
Olhando para o vácuo coerente, â|αi = 0, então,

∆x̂b = 21 er e ∆ŷb = 12 e−r , (11.223)
e o estado comprimido fica no mı́nimo de incerteza.
11.6.2 Coherence of squeezed states

O estado comprimido ou squeezing contem correlações quânticas não-clássicas, como mostraremos
calculando g (1) e g (2) para estados comprimidos,
hb̂† b̂b̂† b̂i cosh 2r

g (2) (0) = =1+ . (11.224)
hb̂† i2 hb̂i2 sinh2 r
The coherent representation distribution functions are,
P|βi (α) = (11.225)

ρ=
and,
sech r −(|α|2 +|β|2 )+(α∗ β+αβ ∗ ) sech r− 21 [eiθ (α∗2 −β ∗2 )+e−iθ (α2 −β 2 )] tanh r
Q|βi (α) = e (11.226)
π
and,
∗ 2 (α−α∗ )2
W|βi (α) = 1
2π exp(− (α+α
2e−2r
)
+ 2e2r
) . (11.227)
11.6.3 Homodyne signature

Let us mix squeezed light b̂ with a local oscillator â at a beamsplitter,

x̂ 1 1 −i b̂
= √ . (11.228)
ŷ 2 −i 1 â
The homodyne signal is,
Phody ∝ x̂† x̂ − ŷ † ŷ (11.229)

† †
= iâ b̂ − iâb̂
= i(â† − â)b̂r − (â† + â)b̂p

= 2|α| b̂r sin θ − b̂p cos θ ,
if the local oscillator can be considered classical, α = |α|eiθ . I.e. the phase of the local oscillator
permits us to select either one of the quadrature components.
11.6.4 Hamiltonian for squeezing

The particular form of the squeezing operator (11.213) already suggests that squeezing my by
obtain by an interaction following the Hamiltonian,
Ĥ = ~ωâ† â + ~gâ†2 + i~g ∗ â2 . (11.230)
The Hamiltonian leads to,

ȧ = −iωâ + 2gâ† . (11.231)
I.e. the interaction should include correlation pair production as it is the case for parametric
processes or four-wave mixing.
11.6. SQUEEZED STATES 329
Cavities are good for producing squeezing. However, the unused ports of the cavity let
vacuum fluctuations come in, which partially overrule squeezing.
The numbers of photons in the squeezed state,
†2 /2−ξ ∗ â2 /2 ∗2 /2−ξ ∗ α2 /2 |α|n e−|α| ξα∗2 /2−ξ∗ α2 /2
hn|α, ξi = hn|eξâ |αi = eξα hn|αi = e (11.232)
n! !
(eiθ tanh r)n/2 − 1 (|β|2 −e−iθ β 2 tanh r) βe−iθ/2
= e 2 Hn √ .
2n/2 (n! cosh r)1/2 2 cosh r sinh r
In the photon representation we can easily see that the squeezed vacuum is (unlike the coherent
and the Fock vacuum) not empty,
α→0
hα, ξ|n̂|α, ξi = |α|2 + sinh2 |ξ| −→ sinh2 |ξ| (11.233)
α→0
∆α,ξ n̂ = |α| + 2 cosh2 |ξ| sinh2 |ξ| −→ 2 cosh2 |ξ| sinh2 |ξ| .
Squeezed vacuum contains contributions from many |ni.
11.6.5 Multimode squeezing

Define,
b̂ = µâ + νĉ† . (11.234)
Again using µ2 − ν 2 = 1, the standard commutation rules for â and ĉ give,
[b̂, b̂† ] = 1 (11.235)

i
[b̂r , b̂p ] = 2 .
The individual variances read,
∆b̂2r = hb̂2r i − hb̂r i2 (11.236)

= 14 h(b̂ + b̂† )2 i − 14 hb̂ + b̂† i2
= 41 h(µâ + µâ† + νĉ + νĉ† )2 i

= 14 µ2 + 41 ν 2 + 12 µ2 hâ† âi + 21 ν 2 hĉ† ĉi + 12 µν hâĉi + hâ† ĉ† i .
using hâi = hĉi = hâ† ĉi = 0.

Two-mode squeezing can exist even if the individual modes are not squeezed,
X
|r, φi = cosh−1 r tanhn reinφ |r, φia |r, φib . (11.237)
n
A two-mode squeezed vacuum state can be generated by the squeezing operator

∗
Q̂(ξ) ≡ exp ξ2 âb̂ − 2ξ â† b† . (11.238)
Remember that the single-mode squeezing is obtained if â = b̂. In a number state base
1 X
|r, φi = (tanh r)n einφ |nia |nib . (11.239)
cosh r n
Two-mode relative number squeezing parameter

σ 2 (ni − hnj i)
ξi,j = . (11.240)
hni i + hnj i
11.7 Photon counting statistics

11.7.1 Photon counting with classica fields
There is a nice introduction by [164], p.229. Irradiate a photon counter with efficiency η with
a classical light beam of cycle-averaged intensity I(t). The probability density of registering a
count is p(t) = ηI(t). Now consider a time interval [t, t + T ]. Be Pm (t, T ) the probability to
encounter m counts. We now separate a time period dt which is too short to yield more than
one photon, [t, t + t0 + dt]. Then the probability of finding m photons in this interval is,
Pm (t, t0 + dt) = Pm (t, t0 )[1 − p(t)]dt + Pm−1 (t, t0 )p(t)dt . (11.241)
From this one derives a differential equation, whose solution is,
R m
t+T
p(t 0 )dt0 Z t+T
t 0 0
Pm (t, T ) = exp − p(t )dt . (11.242)
m! t
R
¯ T ) ≡ T −1 t+T I(t0 )dt0 and averaging over a number of initial start times t,
Defining I(t, t
* +
¯ T) m
ηT I(t, ¯
Pm (T ) ≡ hPm (t, T )it = e−ηT I(t,T ) . (11.243)
m!
t
The probability follows as an approximation to a binomial distribution. One can show that
for noise-free light as well as for chaotic light with short coherence length the distribution is
Poissonian. For chaotic light with long coherence length it is thermal.
11.7.2 Photon counting with quantum fields

Quantummechanically the probability is obtained by replacing the intensity by an operator
and the starting time averaging by a trace over the density operator and the normal-ordered
probability Pm [96, 95, 243, 224],
h im
ˆ
ηT I(T ) ˆ
Pm (T ) = Tr ρ̂N e−ηT I(T ) , (11.244)
m!
R
ˆ ) ≡ T −1 T 2ε0 cÊ − (r, t)Ê + (r, t)dt. The field operators can be expanded, Ê + (r, t) =
where I(T
p 0
~ω/2ε0 V e−iωt+ik·r b̂, only considering one mode. We obtain,
iˆ
(ξ b̂† b̂)m −ξb̂† b̂
Pm (T ) = Tr ρ̂N e . (11.245)
m!
where ξ ≡ ηc~ωT /V is the quantum efficiency,
1 X X X (−ξ b̂† b̂)l
Pm (T ) = hk| ρ̂ |nihn|N (ξ b̂† b̂)m |ki (11.246)
m! n
l!
k l
∞
X k−m
X ξ m (−ξ)l
= Pk hk|(b̂† )m+l (b̂)m+l |ki
m!l!
k=m l=0
X∞ k−m
X
ξ m (−ξ)l k!
= Pk
l! m!(k − m − l)!
k=m l=0
X∞ k−m ∞
k m X k−m l
X k m
= Pk ξ (−ξ) = Pk ξ (1 − ξ)k−m .
m l m
k=m l=0 k=m
11.7. PHOTON COUNTING STATISTICS 331
Second-order correlation (for a single mode there is no time-dependence,

∞
1 X 1 X X n m
m(m − 1)Pm (T ) = m(m − 1) Pn ξ (1 − ξ)n−m (11.247)
m̄2 m (ξn̄)2 m n=m
m
1 X X n − 2
= 2 n(n − 1)Pn ξ m−2 (1 − ξ)n−m
n̄ n m<n
m − 2
1 X hn(n − 1)i
= 2 n(n − 1)Pn = ≡ g (2) (τ ) .
n̄ n hn̄i2
For a photon number state ρ = |nihn| and Pk = δnk ,

n m
Pm (T ) = ξ (1 − ξ)n−m (11.248)
m
1 X 1 1
g (2) (τ ) = 2 k(k − 1)δnk = 2 n(n − 1) = 1 − .
n̄ n n
k
P |α|n e−|α| n̄k −n̄
For a coherent state ρ = n
√
n!
|nihn| and Pk = k! e ,
X n̄k
−n̄ k m
Pm (T ) = e ξ (1 − ξ)k−m (11.249)
k! m
k
∞
(ξn̄)m −n̄ X n̄k (1 − ξ)k
0 0
(ξn̄)m −ξn̄
= e 0
= e
m! 0
k! m!
k =0
1 X n̄k X n̄k0
g (2) (τ ) 2 k(k − 1) e−n̄ = 0!
e−n̄ = 1 .
n̄ k! 0
k
k k
Similarly we get for a thermal state,

1 X X 0
(ξn̄)m (2) n̄k 1 n̄k
Pm (T ) = g (τ ) = k(k − 1) = (k + 2)(k + 1)
(1 + ξn̄)1+m n̄2 (1 + n̄)k (1 + n̄)2 0 (1 + n̄)k0
k k
(11.250)
1 2
= = 2(1 + n̄) .
(1 + n̄) [1 − n̄/(1 + n̄)]3
2
11.7.2.1 Generalization for time-dependent fields and finite detector bandwidth

The measured count rate is then replaced by,
Z T Z T
†
ξ b̂ b̂ → dτ dτ 0 ξ(τ − τ 0 )b̂† (τ )b̂(τ 0 ) (11.251)
0 0
with the spectral sensitivity,

Z ∞
1 0
ξ(τ − τ 0 ) ≡ ξ(ω)e−iω(τ −τ ) dω . (11.252)
2π −∞
RT
For large detector bandwidth ξ(ω) = ξ, so that we may set ξ b̂† b̂ → ξ 0 dτ b̂† (τ )b̂(τ ). For this
case the probability becomes,
1
Pm (T ) = Tr ρ̂T N B̂(T )m e−B̂(T ) , (11.253)
m!
RT
where B̂(T ) ≡ ξ 0 dτ b̂† (τ )b̂(τ ).
Carrying out the ordering we get now,
*∞ Z Z Z +
X (−ξ)k+l T T T
† † †
Pk (T ) = TN dτ1 dτ2 .. dτk+l b̂ (τ1 )b̂(τ1 )b̂ (τ2 )b̂(τ2 )..b̂ (τk+l )b̂(τk+l )
k!l! 0 0 0
l=0
(11.254)
*∞ +
X (−ξ)k+l Z T Z τ1 Z τk+l−1
= dτ1 dτ2 .. dτk+l b̂† (τ1 )b̂† (τ2 )..b̂† (τk+l )b̂(τk+l )..b̂(τ2 )b̂(τ1 ) ,
k!l! 0 0 0
l=0
when we assume dτ1 < dτ2 < .. < dτn+l .

The expectation value for registering n counts is,
X ?
hni(T ) = kPk (T ) = hT N B̂(T )i (11.255)
k
X D E Z T D E
1
= T N B̂(T )k e−B̂(T ) = ξ dτ b̂† (τ )b̂(τ ) .
(k − 1)! 0
k
The variance is often described by Mandel’s Q-function,
h∆n2 i(T ) − hni(T ) hn(n − 1)i(T ) − hni2 (T )

Q(T ) ≡ = . (11.256)
hni(T ) hni(T )
We have,
X ?
hn(n − 1)i(T ) = k(k − 1)Pk (T ) = hT N Â2 (T )i (11.257)
k
Z T Z T D E
?
=η 2
dτ dτ 0 T N b̂† (τ )b̂† (τ 0 )b̂(τ 0 )b̂(τ ) . (11.258)
0 0
The conditional probability to detect a photon at time t = τ after a successful detection at

time t = 0 is,
D E
T N b̂† (τ )b̂† (τ 0 )b̂(τ 0 )b̂(τ )
g (2) (τ ) ≡ D E2 , (11.259)
b̂† (τ )b̂(τ )
hence,
 
R t+T R t+T
 t dτ t − dτ 0 S 2 (τ τ 0 ) hT N E − (τ )E − (τ 0 )E + (τ 0 )E + (τ )i 
Q(T ) = hni(T )  R 2 − 1 (11.260)
t+T
S t dτ hE − (τ )E + (τ )i
Z T h i
hni(T ) (2)
=2 dτ (T − τ ) g (τ ) − 1 .
T2 0
I.e. we can follow the evolution of the photon number variance as time goes on [243, 233, 238].
11.8. EXERCISES 333
11.7.3 Waiting time distribution

The conditional probability P̃ (τ ) that after a detection at time 0 any other photon (not neces-
sarily the next one) is detected at time τ is the sum of the probabilities to detect the first, the
second, etc. photon, X
P̃ (τ ) = P (k) (τ ) . (11.261)
k
Rτ
Now P (2) (τ ) = 0 P (1) (t)P (1) (τ − t)dt, so that
g (2) (τ ) ∝ P̃ (τ ) . (11.262)
11.8 Exercises
11.8.1 Quantization of the electromagnetic field
11.8.1.1 Ex: Estatı́stica fotônica
Um ressonador óptico contem 10 fótons no modo T EM00q . Qual é a probabilidade de encontrar,
num qualquer tempo, 1 fóton resp. 10 fótons, quando a luz é (a) térmica, (b) coerente? Qual é
no caso (a) a temperatura da luz?
11.8.2 Quantum correlations in the Jaynes-Cummings model

11.8.2.1 Ex: Modelo de Jaynes-Cummings
Considere o hamiltoniano de Jaynes-Cummings.
a. Determine com a equação de Schrödinger o sistema de equações diferenciais dando a evolução
temporal dos coeficientes c2,n (t) e c1,n+1 (t) nessa representação na imagem de interação dentro
da aproximação da onda rotativa (RWA).
b. Calcule a evolução temporal para a condição de inı́cio c2,n (0) = 1 e c1,n+1 (0) = 0 para o caso
particular ω = ω0 .
c. Generaliza o resultado da tarefa a. para um campo de muitos modos, para o qual inicialmente
(i) todos os modos do campo k são no estado vazio |0i e (ii) o átomo está no estado excitado
|ai. Utilize o ansatz
X
|ψ(t)i = c2 (t)e−iE2 t/~ |2, 0i + c1,k (t)e−i[E1 /~+ωk ]t |1, 1k i ,
k
e determine as equações de movimento para as amplitudes c2 e c1,k .
11.8.2.2 Ex: Colapso e renascimento quântico no modelo de Jaynes-Cummings

Considere o hamiltoniano de Jaynes-Cummings e mostre que a coerência quântica entre os dois
nı́veis atômicos pode desaparecer totalmente por longos tempos e ressurgir depois. Explique,
como isso é possı́vel.
11.8.2.3 Ex: Desdobramento de Rabi no vácuo

Calculo o desdobramento Autler-Townes para um átomo excitado interagindo com uma cavidade
vazia.
11.8.2.4 Ex: A função Q no um estado de Jaynes-Cummings
Calcule a função Q para um estado de Jaynes-Cummings à partir da sua definição (11.57).
11.8.2.5 Ex: Criação de correlações quânticas num modo óptico
Mostraremos neste exercı́cio como, por operações coerentes num sistema de três nı́veis, podemos
criar correlações quânticas tipo gato de Schrödinger num modo óptico. No sistema ilustrado
na Fig. 11.18 imaginamos a transição inferior excitada por pulsos π/2 de radiação clássica
ressonante microonda (como descrito pela operação (11.35)). A transição superior é excitada
por pulsos π de uma luz laser quântica e sintonizada muito fora de ressonância, assim criando
uma dinâmica dispersiva (como descrito pela operação (11.35)). No tempo t = 0 o átomo seja
localizado
√ no estado |1i. Agora, aplica a seguinte sequência de pulsos: (i) pulso microonda
com n̄Ω12 t = π/2, (ii) pulso óptico com Ω223 t/4∆23 = π, (iii) outro pulso microonda π/2 e
finalmente (iv) um pulso óptico ressonante com a transição |2i-|3i projetando a população do
átomo em um dos estados da transição microonda. Descreve a evolução do estado do sistema
ao longo da sequência. Determine o estado final do modo óptico?
Figure 11.18: (a) Esquema de nı́veis e (b) sequência de pulsos.
11.8.3 Spontaneous emission and light scattering

11.8.3.1 Ex: Derivação das equações de taxas para átomo de dois nı́veis
Inserindo o ansatz (11.103) na equação de Schrödinger derive as equações de movimento (11.105)

para as amplitudes da função de onda.
11.8.4 Beam splitting and quantum amplification

11.8.5 The laser
11.8.6 Squeezed states
11.8.6.1 Ex: Unitariedade do operador de compressão
Calcule hÊi e ∆Ê.
11.8.6.2 Ex: Transformação pelo do operador de compressão
Demonstre a relação (11.214).

11.8. EXERCISES 335
11.8.6.3 Ex: Estados comprimidos

ξ∗ 2 ξ † 2
â − 2 â
Estados comprimidos podem ser introduzidos por aplicação do operador S(ξ) ≡ e 2
sobre um estado de Glauber |αi, onde ξ é o parâmetro de compressão.

Calcule hα, ξ|n̂|α, ξi e mostre com α → 0, que o vácuo comprimido não é vazio.
11.8.7 Photon counting statistics

Chapter 12
Atomic motion in electromagnetic

fields
Até agora –e principalmente na Seç. 3.5.4– analisamos o movimento de partı́culas quânticas
em paisagens de potenciais sem especificar a origem fı́sica dos potenciais. Conhecemos a força
gravitacional, que pode ser derivada do potencial de atração homogêneo da Terra,
F = −∇Vgrv = −∇(mgz) = −gmêz . (12.1)
Uma outra força fundamental vem do electromagnetismo. Já estudamos –principalmente no na
Seç. 8.4.3– a reação da camada eletrônica de átomos à campos eletromagnéticos aplicados.
Em contraste, este capı́tulo será dedicado ao movimento do centro de massa atômico sujeito à
forças resultando de interações com campos eletromagnéticos. Começaremos, na primeira seção,
com forças eletromagnéticas do tipo Coulomb-Lorentz agindo sobre cargas (e.g. ı́ons), dipolos
elétricos permanentes (e.g. moléculas polares) e dipolos magnéticos permanentes (e.g. átomos
paramagnéticos). Também, serão discutidas situações mais complexas, como o espalhamento
de luz por átomos confinados e átomos interagindo com cavidades ópticas, e os potenciais
adiabáticos.
A segunda seção será inteiramente dedicada às forças exercidas por feixes luminosos, em
particular a pressão radiativa e a força óptica dipolar atualmente muito usadas em experimentos
de resfriamento e aprisionamento atômico. Deixaremos a questão da aplicação para o Cap. 14.7.6
e concentraremos aqui na derivação (semiclássica ou quântica) e a interpretação das forças. De
fato, o entendimento das forças ópticas sobre átomos necessita a consideração dos graus de
liberdade internos.
Muitos objetos quânticos são dotados, além dos graus de liberdade relacionados ao movi-
mento do centro-de-massa (energia cinética ou potencial), de graus de liberdade internos, por
exemplo, o movimento de elétrons em átomos ou moléculas. No caso mais simples, o hamiltoni-
ano de um tal sistema se compõe de uma parte externa juntando a energia cinética e a energia
potencial e uma parte contando para energia de excitação ~ω0 de um estado interno |ei,
p2
Ĥatm = + V (r) + ~ω0 |eihe| . (12.2)
2m
A escala temporal do movimento eletrônico geralmente é muito rápida em comparação com
o movimento do núcleo, onde se concentra (quase) toda a massa de um átomo. Por isso, as
dinâmica externa e interna (eletrônica) se desacoplam, o que permite a separação do movimento
do centro de massa atômica em duas partes,
|ψi = |ψiext |ψiele , (12.3)
onde para um simples átomo de dois nı́veis, |ψ(t)iele = cg (t)|gi + ce (t)|ei, com o estado atômico
fundamental |gi e o estado excitado |ei. Os estados externos são autoestados do momento no
337
338 CHAPTER 12. ATOMIC MOTION IN ELECTROMAGNETIC FIELDS
caso de uma partı́cula livre, |ψiext = |pi. Para partı́culas num confinadas num potencial os
estados externos são os autoestados vibracionais, |ψiext = |ni. As evoluções temporais dos
graus de liberdade in- e externos são governadas por equações de Schrödinger independentes.
Para nuvens atômicas frias a energia cinética é bem menor do que a energia de excitação, o que
permite a separação das escalas de energia. Isto é, os graus de liberdade internos são gelados
no estado fundamental. Muitos fenômenos, por exemplo, a condensação de Bose-Einstein e a
dinâmica dos condensados são descritos neste regime.
Figure 12.1: Os graus de liberdade internos de átomos frios são termicamente gelados.
Não obstante, o fato de ser termicamente gelado não impede a excitação intencional do
grau de liberdade interno por irradiação de campos eletromagnéticos sintonizados perto de res-
sonâncias acoplando nı́veis eletrônicos de energia. Em caso de acoplamento, os graus de liberdade
externos e internos devem ambos ser considerados.
12.1 Electromagnetic forces

O impacto de campos eletromagnéticos sobre a dinâmica interna de átomos já foi estudado no
Cap. 8.4.3. Aqui vamos concentrar sobre a força sobre o centro de massa exercida pelo gradiente
de potenciais eletromagnéticos:
F = −∇Ĥint , (12.4)
onde o hamiltoniano de uma carga interagindo com campos eletromagnéticos é obtido no acopla-
mento mı́nimo (8.11) por,
1 −~2 2 i~q
Ĥint = (−i~∇ − qA)2 + qΦ ' ∇ + A · ∇ + qΦ . (12.5)
2m 2m m
À partir desta formula podemos calcular todas as forças eletromagnéticas.

O acoplamento dos graus de liberdade externos e internos é devido ao recuo fotônico trans-
ferido ao átomo durante processos de absorção e emissão. Ou seja, o fato que a interação com
luz no mesmo tempo excita o átomo e exerce uma força acopla os graus de liberdade. Isso
se manifesta no hamiltoniano do átomo interagindo com um campo luminoso pela aparição de
termos juntando operadores agindo sobre diferentes graus de liberdade,
Ĥ = ~ωâ† â + Ĥint + Ĥatm onde Ĥint = ~Ω(r̂)eik·r̂ â† σ̂ + c.c. , (12.6)

P
onde σ̂ ≡ |gihe| e â ≡ n |nihn + 1| e ~Ω(r̂) ≡ d12 · E é a constante de acoplamento ou
frequência de Rabi. O hamiltoniano é aquele do modelo de Jaynes-Cummings, só que surge,
além dos operadores de campo â e do átomo σ̂, um operador para a posição do átomo r̂, cuja
12.1. ELECTROMAGNETIC FORCES 339
caracterı́stica quântica não levamos muito à sério até agora. Ele aparece na frequência de Rabi
e também no termo eik·r̂ . Agora, devemos nós lembrar, que
Umomentum = e−ik·r̂ = |p + ~kihp| (12.7)
e o operador unitário do recuo fotônico na absorção. Mostraremos em breve, que é justamente
este termo no hamiltoniano que dá origem à todos fenômenos de força da luz sobre átomos.
A presença do operador de posição no hamiltoniano de Jaynes-Cummings introduz um novo
grau de liberdade. Sem potencial externo (o sistema é invariante à translações espaciais), este
grau de liberdade é simplesmente o momento do centro de massa atômica, tal que o novo conjunto
de números quânticos é |j, n, pi. Estritamente falando deverı́amos abrir o espaço Hilbert total
pelo produto externo, Ĥele ⊗ Ĥrad ⊗ Ĥext .
12.1.1 Forces on charges and electric dipole moments

Como mostrado na Eq. (8.8), as equações (12.4) e (12.5) levam (obviamente) à força de Coulomb-
Lorentz sobre cargas e correntes.
Em óptica atômica, a força de Coulomb-Lorentz é utilizada, por exemplo, para acelerar ou
aprisionar ı́ons (vide Sec. 14.5) e outras partı́culas eletricamente carregadas.
Átomos naturalmente não exibem momento elétrico dipolar permanente, quando não são su-
jeitos à campos elétricos externos. Em contraste, moléculas polares, como por exemplo, dı́meres
heteronucleares, têm. Isso permite, influenciar seu movimento por campos elétricos inomogêneos
(vide Sec. 14.5.3).
12.1.2 Forces on magnetic dipole moments

Átomos neutros são insensı́veis à campos elétricos. Mas como já vimos no Cap. 8.4.3, o movi-
mento orbital dos elétrons corresponde à uma corrente circular gerando um momento magnético
dipolar permanente µ ~ , que pode interagir com campos magnéticos externos. Já mostramos no
cálculo (8.14) e (8.18) que a energia de interação (12.5) pode ser escrita como,
g J µB gJ µB gJ µB
U = −~ µJ · B = − J · B −→ − |J| · |B| = − mJ B , (12.8)
~ ~ ~
onde o fator de Landé é dado pela formula (8.20),
J(J + 1) + S(S + 1) − L(L + 1)
gJ = 1 + . (12.9)
2J(J + 1)
Aqui J = L+S é o momento angular total seguindo do acoplamento do momento angular orbital
total e do spin total de todos os elétrons. Se o átomo tem um spin nuclear I diferente de zero,
então F = J + I se substituı́ para J na Eq. (12.8), o fator g generaliza para (8.34) 1 ,
F (F + 1) + J(J + 1) − I(I + 1)
gF ' gJ . (12.10)
2F (F + 1)
Na Sec. 8.2 usamos esta formula para calcular o deslocamento Zeeman dos nı́veis de energia
internos. Mas, de acordo com a formula (12.4), a interação também gera uma força agindo sobre
o centro-de-massa de átomo,
g F µB
f =− mF ∇B . (12.11)
~
1
Note, que a formula só vale par a campos fracos. Para campos fortes o desdobramento Zeeman muda para o
desdobramento Paschen-Back da estrutura hiperfina.
Em caso de ausência de estrutura hiperfina simplesmente substituı́mos f por J.

Obviamente, a força é condicionada à existência de um gradiente do valor absoluto do campo
magnético. Ela foi usada pela primeira vez no famoso experimento de Stern-Gerlach, que con-
duziu à descoberta do elétron (vide Sec. 4.3.3). Em óptica atômica (vide Sec. 14.4), esta força é
muito utilizada para realização de armadilhas magnéticas para átomos frios. Faz os Excs. 8.4.2.2,
12.3.1.1 e 12.3.1.2.
12.1.3 Adiabatic potentials

Um potencial adiabático pode ser usado para realização de potenciais de aprisionamento mais
complicados [52]. Para estudar potenciais adiabáticos consideramos o sistema acoplado | 12 , 12 i ↔
| 12 , − 12 i. Uma generalização para sistemas multinı́veis F > 21 é simples. O hamiltoniano de
estados vestidos do nosso sistema de dois nı́veis é uma matriz 2 × 2,
1
2 µB gF B(z) − 12 ~ω 1
2 ~Ω
Ĥ(z) = 1 . (12.12)
2 ~Ω − 21 µB gF B(z) + ~ω
Para simplificar, assumimos uma geometria unidimensional, B = B(z), mas podemos facilmente
generalizar para três dimensões. Os autovalores de H são
q
1
E± (z) = ± ~2 Ω2 + [µB gF B(z) − ~ω]2 . (12.13)
2
Suficientemente longe da ressonância, ~Ω |µB gF B(z) − ~ω|, obtemos
1 ~2 Ω2
E± (z) ≈ ± [µB gF B(z) − ~ω] ± , (12.14)
2 4[µB gF B(z) − ~ω]
onde o segundo termo pode ser interpretado como o deslocamento Stark dinâmico dos nı́veis de
energia.
Para ilustrar a ação da radiofrequência, calculamos a energia potencial e os estados vestidos
para átomos 6 Li. Por simplicidade, assumimos um gradiente de campo magnético 1D linear
B(z) ≡ zb. Fig. 12.2(a) mostra o acoplamento de radiofrequência e Fig. 12.2(b) os estados
vestidos para dois subestados magnéticos acoplados por uma radiofrequência. A mı́nimo apare-
cendo na curva superior da Fig. 12.2(a) pode servir como potenciais de confinamento. Usando
uma radiação rf com várias frequências, mı́nimos de potenciais podem ser gerados em várias
distâncias z. No Exc. 12.3.1.3 calculamos um exemplo.
A força na base dos estados vestidos é calculada a partir de,
d
F= p = i~−1 [H, p] , (12.15)
dt
com o hamiltoniano (12.12),
X
F(r) = hF̂(r)i = −Tr atom,laser ρ̂Or Ĥ = − hn, j|ρ̂∇r Ĥ|j, ni .
n,j
Consideramos apenas uma dimensão e desconsidere os graus de liberdade do campo de radiação,
F (r) = −Tr atom,laser ρ̂∂z Ĥ (12.16)

X
=− hj|ρ̂∂z 12 µB gF B|1ih1| − 21 ~ω|1ih1| − 12 µB gF B|2ih2| + 12 ~ω|2ih2| + 12 ~Ωeikz σ̂ † + c.c. |ji .
j
12.2. OPTICAL FORCES ON FREE AND CONFINED ATOMS 341
(a) 4 (b) 1.5
1
2
0.5
Δ/2π (kHz)
Δ/2π (kHz)
0 ω 0 Ω Δ
-0.5
-2
-1
-4 -1.5
0 0.1 0.2 0.3 0.4 0 0.1 0.2 0.3 0.4
z (μm) z (μm)
Figure 12.2: (Code: LM F orces P otencialAdiabatico.m) (a) Energias potenciais para um es-
quema de nı́veis F = 12 com um fator g de g = − 23 (como no caso do estado fundamental 2 S1/2
do 6 Li). Uma radiofrequência (seta) acopla os subestados mF = ± 12 . Aqui b = 200 G/cm e
ω = 2π × 5 kHz. (b) Estados vestidos não-acoplados (linha pontuada), estados vestidos acopla-
dos (linha sólida) e deslocamentos dinâmicos de Stark (dash-dotted) calculados na aproximação
longe de ressonância. A frequência de Rabi é Ω = 2π × 700 Hz.
Aqui nós negligenciamos qualquer possı́vel dependência de posição de Ω,

X
F (r) = − 12 µB gF ∂z B hj|ρ̂(|1ih1| − |2ih2|)|ji
j
1
= − 2 µB gF ∂z B(ρ11 − ρ22 ) . (12.17)
Se os átomos entrarem na área de acoplamento de maneira adiabática, as populações dos

potenciais adiabáticos dependerão apenas de z. Isso é análogo à transferência adiabática por
varreduras adiabáticas ou sequências de pulsos STIRAP. Caso contrário, eles também dependem
da história (trajetória do átomos) podendo resultar em transições de Landau-Zener.
12.2 Optical forces on free and confined atoms

A luz carrega momento, e o espalhamento de luz por um objeto produz uma força sobre aquele
objeto. Embora uma consequência direta da teoria clássica de Maxwell, foi somente em 1933, que
esta propriedade da luz foi verificada por Frisch através da observação de um desvio transversal
muito pequeno (3 · 10−5 rad) em um raio de sódio atômico exposto à luz de uma lâmpada
ressonante. Com a invenção do laser ficou mais fácil observar efeitos mecânicos la luz, pois a
luz mais intensa e altamente direcional exerça forças muito maiores. Embora estes resultados
acenderam o interesse em usar as forças da luz para controlar o movimento de átomos neutros, as
bases fundamentais para a compreensão das forças da luz sobre átomos não foram desenvolvidas
antes do final da década 1970. Demonstrações experimentais inequı́vocas de resfriamento e
aprisionamento de átomos não foram realizadas antes de meados dos anos 80. Nesta seção,
discutiremos alguns aspectos fundamentais das forças da luz. Esquemas utilizados para resfriar
e aprisionar átomos neutros serão mostrados nas Secs. 14.2 e 14.3.
A força da luz exercida sobre um átomo pode ser de dois tipos: uma força dissipativa
espontânea e uma força conservativa dipolar. A força espontânea surge do impulso experimen-
tado por um átomo quando absorve ou emite uma quanta de impulso fotônico. Como vimos na
Seç. ??, quando um átomo espalha luz, a seção transversal ressonante de espalhamento pode
πλ2
ser escrita pela Eq. (??), σ0a = gg12 2 0 , onde λ0 é o comprimento de onda ressonante. Na região
óptica do espectro eletromagnético os comprimentos de onda da luz são da ordem de várias cen-
tenas de nanômetros, e as seções transversais ressonantes para espalhamento se tornam muito
grande, (∼ 10−9 cm2 ). Cada fóton absorvido transfere uma quanta de impulso ~k ao átomo na
direção de propagação. A emissão espontânea após a absorção ocorre em direções aleatórias, e ao
longo de muitos ciclos de absorção-emissão, ela cancela para zero em médio. Por consequência,
a força espontânea total age sobre o átomo na direção da propagação da luz, como mostrado
esquematicamente no diagrama da Fig. 12.3. A taxa saturada de espalhamento de fótons por
emissão espontânea (o valor recı́proco do tempo de vida do estado excitado) fixa o limite superior
para a magnitude da força. Esta força é às vezes chamada de força de pressão radiativa.
A força de gradiente dipolar pode ser facilmente entendida considerando a luz como uma
onda clássica. É simplesmente a força no médio temporal decorrente da interação do dipolo
da transição, induzida pela oscilação do campo elétrico da luz, com o gradiente da amplitude
do campo elétrico. Focalização do feixe luminoso controla a magnitude deste gradiente, e a
sintonização da frequência óptica abaixo ou acima da transição atômica controla o signo da
força que age sobre o átomo. Ajustando a luz abaixo da ressonância atrai o átomo para o centro
do feixe de luz, enquanto ajustando a luz acima da ressonância repele-lo. A força dipolar é
um processo estimulado em que não há troca de energia entre o campo e o átomo. Fótons
são absorvidos de um modo e reaparecem por emissão estimulada em outro. Conservação de
momento exige que a mudança de direção de propagação de fótons desde um modo inicial
para um modo final deixe o átomo com um recuo. Contrário a força espontânea, não há, em
princı́pio, nenhum limite superior para a magnitude da força dipolar, pois é uma função apenas
do gradiente de campo e dessintonização.
Dentro da teoria do electromagnetismo calculamos forças radiativas sobre cargas pelo tensor
de estresse de Maxwell 2 . A interação da radiação com átomos dotados de graus de liberdade
internas exibindo ressonâncias pode ser tratado qualitativamente pelo modelo de Lorentz 3 .
No seguinte, mostraremos cálculos quantitativos semiclássicos e quânticos: A força de um
feixe luminoso sobre um átomo pode ser calculada de muitas maneiras diferentes, cada uma
enfatizando um aspecto ligeiramente diferente: À partir da força de Lorentz clássica exercida
sobre um átomo por campos eletromagnéticos pode ser derivada uma equação semiclássica de
Fokker-Planck [236]. Na Sec. 12.2.1 derivaremos as duas contribuições (força dipolar e pressão
radiativa) em uma teoria semiclássica [97]. Wineland et al. [256] escolha como ponto de partida
a seção cruzada do processo elementar de espalhamento (Sec. 12.2.3). Dalibard et al. [60]
desenvolve uma teoria quântica usando a representação dos estados vestidos (Sec. 12.2.2). Cirac
et al. [46] mostram uma abordagem baseada na equação mestre (Sec. 12.2.3).
12.2.1 The dipolar gradient force and the radiation pressure force
Para calcular as forças da luz sobre um átomo, descrevemos o átomo como um sistema de dois
nı́veis: Um nı́vel fundamental |1i e um nı́vel excitado |2i decaindo para o nı́vel fundamental com
a taxa Γ. A diferencia de energia dos nı́veis é ω0 ≡ E2 − E1 . A luz seja um feixe laser com a
frequência ω, que pode ser dessintonizada da transição atômica, ∆ ≡ ω − ω0 . Para descrever
a interação, consideramos a parte (12.6) do hamiltoniano total descrevendo a interação [60].
2
Vide a apostila Eletrodinâmica do mesmo autor [55], Sec. 6.2.3.
3
Vide a apostila Eletrodinâmica do mesmo autor [55], Sec. 7.2.4.
Utilizando o operador de densidade ρ̂ 4 , podemos calcular a força que o campo da luz exerce
sobre o átomo,
F(r) = hF̂(r)i = −Trat ρ̂∇r Hint (12.18)

X
= − 21 ~ hj|ρ̂|∇r Ω(r)eik·r−i∆t |2ih1| + Ω(r)e−ikr+i∆t |1ih2| |ji
j

= − 12 ~∇r Ω(r) h1|ρ̂eik·r−i∆t |2i + h2|ρ̂e−ik·r+i∆t |1i

− 2i ~kΩ(r) h1|ρ̂eik·r−i∆t |2i − h2|ρ̂e−ik·r+i∆t |1i .
Agora, colocamos o átomo no lugar r = 0,
F(0) = − 12 ~∇r Ω(0)(ρ12 e−i∆t + ρ21 ei∆t ) − 2i ~kΩ(0)(ρ12 e−i∆t − ρ21 ei∆t ) . (12.19)
As grandezas ρ12 ≡ h1|ρ̂|2i = ρ∗21 são as coerências, que se desenvolvem num sistema de dois
nı́veis excitado por um feixe laser. Inserindo as soluções estacionárias das equações de Bloch,
Ω2 (2∆ − iΓ)Ω
ρ22 = e ρ12 = e−i∆t . (12.20)
4∆2 + 2Ω2 + Γ2 4∆2 + 2Ω2 + Γ2
obtemos
4∆Ω ΓΩ2
F(0) = − 12 ~ ∇ r Ω + ~k . (12.21)
4∆2 + 2Ω2 + Γ2 4∆2 + 2Ω2 + Γ2
Γ 2
Com a definição da seção cruzada dessintonizada, σa (∆) = σa (0) 4∆2 +2Ω2 +Γ2 ,

2Ω2 Ω2 σa (∆)
F(0) = − 12 ~∆∇r ln 1+ + ~k . (12.22)
4∆2 + Γ2 Γ σa (0)
A seção cruzada ressonante para uma transição ’clássica’ é σa (0) = 3λ2 /2π.
Aparentemente, a força contem duas contribuições. A força de gradiente dipolar pode ser
derivada de um potencial. Ela é proporcional ao gradiente de intensidade e pode ser interpre-
tada como resultando de processos de absorção seguido por emissão auto-estimulada. Perto da
ressonância ela é dispersiva. Longe da ressonância pode ser aproximada por
−~∆Ω2 |∆|Γ ~Ω2

Fdp = ∇r −→ −∇ r . (12.23)
4∆2 + Γ2 4∆
A força de pressão radiativa é dissipativa. Perto da ressonância ela é absortiva. Ela é propor-
cional ao gradiente da fase e a única força ativa em ondas planas. Ela pode ser interpretada como
resultando de processos de absorção seguido por emissão espontânea. Com Ω2 = σa (0)ΓI/~ω
obtemos uma formula,
I
Frp = ~k σa (∆) = ~kγsct , (12.24)
~ω
que descreve a força como produto do número de fótons no feixe incidente, I/~ω, a seção
transversal para absorção, σa (∆) e o recuo por fóton, ~k. γsct é a taxa de espalhamento. A força
de gradiente dipolar (e o potencial associado) é frequentemente usado para aprisionar átomos, e a
força de pressão radiativa é frequentemente usada para resfriar-lhes. Note que ainda precisamos
4
O operador denside consiste de uma parte interna (excitação do átomo) e uma parte luminosa ρ̂ = ρ̂atom ⊗
ρ̂laser , onde a parte do movimento ρ̂motion foi desprezada, apesar de ter um papel importante em óptica atômica.
Figure 12.3: Esquerdo: Um átomo com a massa m e a velocidade vA se move para o direito e
absorve um fóton propagante para esquerdo com o momento ~kL . Centro: Um átomo excitado
sofre uma mudança de momento pA = mvA − kL . Direito: A reemissão isotrópica de um fóton
resulta, no médio de muitos ciclos absorção-emissão, em uma mudança de momento para o
átomo de hpA i = mvA − kL .
corrigir as Eqs. (12.23) e (12.24) para tomar em conta o quadrado do médio sobre as orientações
do elemento da matriz do momento de transição, d12 /3.
O parâmetro de saturação,
1 2
Ω
s = 22 1 2 , (12.25)
∆ + 4Γ
permite escrever a força de gradiente dipolar e a força de pressão radiativa como,
~∆ 1 ~∆ ~kΓ 1
Fdp = − ∇s = ln[1 + s] e Frp = . (12.26)
6 1+s 6 6 1+s
Eq. (12.26) mostra que ’satura’ a força de pressão de radiação quanto s aumenta, e é, portanto,
limitada pela taxa de emissão espontânea. O parâmetro de saturação descreve essencialmente a
importância relativa dos termos que aparecem no denominador da função de perfil de linha para
as forças da luz sobre o átomo. A taxa de emissão espontânea é uma propriedade intrı́nseca do
átomo, proporcional ao quadrado do momento da transição atômica, enquanto que o quadrado
da frequência de Rabi é uma função da intensidade do laser incidente. Se s 1, a emissão
espontânea é rápida comparado a qualquer processo estimulado, e o campo de luz é dito ser
fraco. Se s 1, a oscilação de Rabi é rápida comparado a emissão espontânea e o campo é
considerado forte.√ O fator de perfil da linha indica um ’alargamento de potência’ por saturação
de um fator de 2. Note, que a força e o potencial de gradiente dipolar, Eqs. (12.26), não
saturam quando a intensidade do campo da luz é aumentado. Geralmente Fdp e Udp são usados
para manipular e aprisionar átomos com um laser sintonizado longe de ressonância para evitar
absorção.
Muitas vezes, o momento de transição pode ser orientado usando luz circularmente polar-
izada. Nesse caso, todas as expressões anteriores para Fdp , Frp e Udp devem ser multiplicados
por 3. A partir de agora vamos abandonar a mediação sobre orientações e usar só d212 para o
quadrado do momento de transição. Resolva os Excs. 12.3.2.1 e 12.3.2.2.
12.2.2 Semiclassic calculation of dipole force and radiative pressure

Em mecânica quântica calculamos a força pela equação de Heisenberg [97],
d i
F̂ = p̂ = [Ĥ, p̂] = −∇r Ĥint . (12.27)
dt ~
Assim, a força é dada pelo gradiente da energia de interação entre o átomo e o campo luminoso.
Dentro da aproximação dipolar a energia de interação é dada por d · E(r). A força agora é,
F(r) = hF̂(r)i = h∇r [d · E(r)]i = h(d · ∇r )E(r)i − hd × (∇r × E(r))i ≡ FC (r) + FL (r) . (12.28)
A primeira contribuição pode ser interpretada como a força de Coulomb sobre o elétron
performando oscilações rápidas na posição r(t) = r0 + e−1 P(r0 , t). O segundo termo é a força
de Lorentz mediada no tempo agindo sobre o momento dipolar elétrico oscilante [119, 121, 120],
FC = ehEi e FL = −hd × ∂t Bi = h∂t d × Bi . (12.29)
A relação entre o momento elétrico
√ dipolar induzido pela luz e a polarizabilidade d = α(E)E
onde ανν ≡ αν + iβν e Eν ≡ Iν eiψν fica,
3
X 3
X
F= αν ∇Iν + 2 βν Iν ∇ψν . (12.30)
ν=1 ν=1
12.2.3 Force exerted by a quantized radiation field

Fótons carregam uma unidade de momento p = ~k, que eles transmitem ao átomo durante
o processo de absorção ou de emissão. Ou seja, a lux exerce um recuo sobre os átomos.
Emissão
P espontânea acopla à todos os modos radiativos do vácuo eletromagnético, Ĥcm−vacum =
j Ĥcm−laser (kj ). Podemos tracejar sobre estas variáveis somente mantendo aquelas do átomo
e do laser. Seguinte Cirac et al. [256, 46], a aleatoriedade do recuo por emissão espontânea é
contabilizada por, Z
ρ̂ → S(r)eik·r ρ̂e−ik·r dΩ , (12.31)
4πR2
tal que o operador de Lindbladt fica,
Z
†
Latom ρ̂ = −Γ{σ̂ σ̂ ρ̂(t) − 3
4π S(r)eik·r σ̂ ρ̂(t)σ̂ † e−ik·r dΩ + ρ̂(t)σ̂ † σ̂} (12.32)
Lcavity ρ̂ = −κ{â† âρ̂(t) − 2âρ(t)â† + ρ̂(t)â† â} .

P
Onde e±ik·r = p |p ∓ ik · rihp| e S(r) = 21 1 + ( k·r kr )
2 e dΩ = dϕd cos ϑ. Calculamos a força
e estabelecemos uma equação de Fokker-Planck para a função de Wigner.
12.2.4 Refraction of atoms by light and of light by atoms

Luz não-ressonante age sobre os graus
R de liberdade
externos de átomos por um deslocamento
de fase da onda de Broglie, exp i~−1 U (r, t)dt , simultaneamente sobre os graus de liberdade
internos por um deslocamento de Stark dinâmico ou light shift dos nı́veis de energia pelo valor
U (r). O vetor de Bloch definido por,
 1 
√ cg c∗
2 e
 
ρ ≡  √12 c∗g ce  (12.33)
2
|ce | − |cg |2
descreve, sob a influencia da interação dispersiva uma precessão em torno do eixo polar. Isso
foi discutido no Exc. 10.6.5.5. O deslocamento de Stark provoca uma rotação de ~−1 U (r)t.
Simultaneamente, o átomo é sujeito à uma força, que corresponde ao gradiente do potencial
−∇U (r), como ilustrado na Fig. 12.4(a). Vemos, que os deslocamentos de fase da onda de
Broglie e do vetor de Bloch são iguais. Finalmente, a fase do modo luminoso é deslocada pelo
mesmo montante num efeito chamado de refração. Ou seja, os graus de liberdade internos,
externos e luminosos são emaranhados.
Figure 12.4: O diagrama esquerda mostre estados produtos e estados vestidos para sintonização
azul. Note que a população está no nı́vel superior e o átomo está sujeito à força (repulsiva)
procurando campos fracas quando entra no feixe laser. O diagrama direito é similar, más para
sintonização vermelha. A população está no nı́vel inferior e o átomo está sujeito à força (atrativa)
procurando campos fortes.
Este fato tem, tem um uso prático em interferômetros atômicos, porque muitas vezes é mais
fácil detectar uma interferência dos estados de excitação interna, do que das ondas de Broglie.
Com o emaranhamento, basta medir uma interferência para conhecer a outra.
Por meio de variações locais do potencial U (r), e.g. por um laser focalizado, é possı́vel ma-
nipular uma frente de onda de Broglie do mesmo jeito que, em óptica clássica, manipulamos um
frente de onda de luz por lentes ou outros componentes, e.g. pelo ı́ndice de refração representado
por uma nuvem atômica perto de ressonância, como ilustrado na Fig. 12.4(b).
A orientação da força depende a frequência da luz em relação à frequência da ressonância.

Ela atrai o átomo para regiões onde o campo é forte, quando a frequência é sintonizada abaixo de
ω0 , e atrai o átomo para regiões de campos fracos quando sintonizada acima de ω0 . A integração
sobre as coordenadas espaciais relevantes resulta em um potencial efetivo ou uma barreira para
o átomo. O comportamento qualitativo do potencial de gradiente dipolar e seu efeito sobre
o movimento do átomo é facilmente visualizado na imagem dos estados vestidos. A Fig. 12.5
mostra o que acontece quando um átomo entra numa região bem definida de um campo óptico,
por exemplo, um feixe de laser focalizado.
Fora da zona de acoplamento átomo-dipolo a expressão ~Ω é desprezı́vel e os ’estados vestidos’
são apenas estados produtos átomo-campo. Quando o átomo entra no campo, essa expressão
torna-se diferente de zero e os estados átomo-campo se combinam para produzir um conjunto
de estados vestidos. Os nı́veis energéticos dos estados produtos se ’repelem’ e se aproximam dos
nı́veis dos estados vestidos. Assumindo que o laser é suficientemente dessintonizado para manter
a taxa de absorção insignificante, a população permanece no estado fundamental. Podemos ver
que a dessintonização azul (vermelha) leva a um potencial repulsivo (atraente) para o átomo
ficando no estado fundamental. Além disso, como ~Ω é diretamente proporcional à raiz da
Figure 12.5: Analogia entre óptica de luz e óptica atômica.
intensidade do laser, é obvio que um aumento dessa intensidade (potência óptica por unidade
de área) amplifica a força sobre o átomo (F ' ∇R Ω).
12.2.5 Inelastic scattering from free thermal atoms

Estritamente falando, não existe espalhamento elástico, pois a lei da conservação do momento
requer a transferência de momento fotônico para o átomo espalhador que, por consequência,
muda sua energia cinética. Para compensar esta mudança de energia cinética a frequência da
luz reespalhada deve ser diferente de maneira a preservar a energia total.
Figure 12.6: Scheme of the scattering.
Calcularemos no seguinte a distribuição de frequência da luz espalhado por um átomo em

função da sua velocidade inicial p1 , da frequência ω1 da luz incidente e do ângulo de espal-
hamento, isto é, o ângulo entre os modos k1 e k2 . Começamos escrevendo as leis de conservação
de energia e de momento,
~k1 + p1 = ~k2 + p2 , (12.34)

p21 p22
~ω1 + = ~ω2 + .
2m 2m
Eliminando p2 destas equações, obtemos,
~2 k12 ~k1 · p1 ~2 k22 (~k1 + p1 ) · ~k2

~ω1 − − = ~ω2 + − . (12.35)
2m m 2m m
Supomos ~2 k12 /2m ' ~2 k22 /2m, tal que obtemos,
1 − ~2 ω1 /mc2 − p1 /mc cos ^(k1 , p1 )

ω2 = ω1 , (12.36)
1 − ~ω1 /mc2 cos ^(k1 , k2 ) − p1 /mc cos ^(p1 , k2 )
ou ainda,
~2 ω1 /mc2 [−1 + cos(ϑin + ϑout )] + p1 /mc(cos ϑout + cos ϑin )

ω2 − ω1 = ω1 (12.37)
1 − ~ω1 /mc2 cos(ϑin + ϑout ) − p1 /mc cos ϑout
onde chamamos os ângulos ϑin = ^(k1 , p1 ) e ϑout = ^(k2 , p1 ).

Agora, consideramos um ângulo de espalhamento particular dado por ϑout = ϑin . Isto é
relevante para espalhamento de Bragg. Definimos ε1 = ~k12 /m,
ε1 (−1 + cos 2ϑ) + 2ω1 p1 /mc cos ϑ

ω2 − ω1 = (12.38)
1 − ε1 /ω1 cos 2ϑ − p1 /mc cos ϑ
' ε1 (−1 + cos 2ϑ) + 2ω1 p1 /mc cos ϑ .
O segundo termo é a condição de Bragg:
∆ω ' 2k1 v1 cos ϑ . (12.39)
Ela é explorada na espectroscopia de RIR, pois a distribuição de momento em p1 se torno em

uma distribuição de frequência ∆ω, que pode ser medida por batimento com um modo ’idler’
irradiado k3 , que pode ser escolhido k3 ≡ k2 . Para átomos parados, p1 = 0, a formula (12.38))
fica,
ω2 − ω1 ' ε1 (−1 + cos ϑ) . (12.40)
Para p1 assumimos uma distribuição de Maxwell-Boltzmann. O que nós queremos saber é o
espectro ω2 . Isso significa uma mudança de frequência independente da velocidade devido ao
recuo. Átomos em movimento podem induzir a decoerência.
Para átomos em repouso o deslocamento é tipicamente da ordem de ωs − ωi ≈ (2π) 10 kHz, o
que, em muitas situações é desprezı́vel tal que podemos considerar o espalhamento como elástico.
Mas deslocamento de frequência também depende da velocidade inicial através do deslocamento
Doppler v. Num gás térmico as velocidades são distribuı́das seguinte a distribuição de Maxwell-
Boltzmann. Por isso, a luz espalhado por espalhamento de Rayleigh por nuvens de átomos livres
é sujeito ao alargamento Doppler.
12.2.6 Optical forces on confined atoms

12.2.6.1 Transitions between vibrational states of a harmonic oscillator
Em presencia de um potencial externo a energia do movimento do centro-de-massa (CM) é
quantizada. Consideramos agora um átomo confinado num potencial exterior harmônico,
!
p̂2 X 1 1
†
Ĥcm = + V (r̂) = ~ζ |nihn| + = ~ζ b̂ b̂ + , (12.41)
2m n
2 2
onde ζ é a frequência secular do OH.

Se o tempo de vida de uma transição atômica é muito mais curto do que um perı́odo de
oscilação, Γ z, a posição dos átomos não muda perante o processo de transição, é temos uma
transição de Franck-Condon. O potencial de confinamento pode depender ou não do estado
interno do átomo. Para ı́ons numa armadilha de Paul não depende, más para átomos em
armadilhas magnéticas geralmente depende,
Ĥatom = |giĤatom−cm,g hg| + |ei(Ĥatom−cm,e + ω0 )he| (12.42)

p̂2
Ĥatom−cm,j = + Vj (r̂) ,
2m
onde ω0 ζ é a frequência da transição atômica.
No seguinte vamos supor potenciais independentes do estado atômico interno. Como as-
sumimos o nosso potencial harmônico, podemos tratar as dimensões espaciais
p separadamente e
quantizar canonicamente como no caso de modos ópticos, definindo aho ≡ ~/mζ:

1 ẑ iaho † 1 ẑ iaho
b̂ ≡ √ + p̂ e b̂ ≡ √ − p̂ . (12.43)
2 aho ~ 2 aho ~
Obtemos:
m 2 2 1
V (z) = ζ ẑ = ~ζ(b̂ + b̂† )2 . (12.44)
2 4
12.2.6.2 Photonic recoil in the harmonic oscillator

Uma maneira de especı́fica de excitar uma oscilação harmônica de um átomo aprisionado é pelo
recuo fotônico recebido mediante um processo de espalhamento de luz. E.g. considerando a onda
plana de radiação E(r, t) = E0~y eikz z−iωt , o impulso recebido durante um processo de absorção
pode ser incluı́do no hamiltoniano de interação por,

Ĥint = ~Ω
2 e ikẑ
|eihg| + +e −ikẑ
|gihe| − |eiωhe| , (12.45)
onde ~Ω ≡ he|y|giE0 é a frequência de Rabi da transição interna. O último termo vem da trans-
formação na imagem de interação. O operador e−ikẑ corresponde à uma transformação do tipo
(2.184), ou seja, uma transição para um sistema se movendo com uma velocidade correspondente
à um recuo fotônico ~k. Obviamente, à partir do estado fundamental, o hamiltoniano somente
acopla estados vibracionais excitados,
hn, g|Ĥint |0, gi = 0 e hn, e|Ĥint |0, gi = 12 ~Ωhn, e|eikẑ |0, gi . (12.46)
Calculamos a função de onda do oscilador acelerado por,

r
ikẑ ikz mζ −mζz 2 /2~ ikz
hz|e |0i = e hz|0i = e e , (12.47)
π~
com o momento médio depois da absorção,
hp̂i = h0|e−ikẑ p̂eikẑ |0i = ~k . (12.48)
12.2.6.3 Decoupling of internal and external motion

P
Expandimos o estado do sistema em |ψ(t)i = n = 0∞ (cn,g |n, gi + cn,e |n, ei, inserimos ele junto
com o hamiltoniano (12.42) e a interação (12.45) na equação de Schrödinger,

d|ψi 1
i~ = ~ζ b̂† b̂ + + |ei(ω0 − ω)he| + ~Ω2 e ikẑ
|eihg| + +e −ikẑ
|gihe| |ψi , (12.49)
dt 2
projetando sobre os estados hn, g| e hn, e| derivamos facilmente as seguintes equações de movi-
mento,
∞
X
dcn,g
= −iζ(n + 12 )cn,g − iΩ
2 cm,e hn|e−ikẑ |mi (12.50)
dt
m=0
∞
X
dcn,e
= −iζ(n + 12 )cn,e − i(ω0 − ω)cn,e − iΩ
2 cm,g hn|eikẑ |mi .
dt
m=0
O fato que, obviamente, somente termos proporcionais à hn|eikẑ |mi contribuem, é devido ao fator
de overlap deFranck-Condon e permite de analisar transição diretas entre nı́veis vibracionais do
mesmo oscilador.
Os elementos da matriz de transição são, com ẑ ∼ b̂ + b̂† :
X X
|hm|eikẑ |ni|2 = |hm|1 + ikẑ|ni|2 = 1 − ik 2 z 2 n2 + n+1
2 , (12.51)
m m
e,
1 −m~k2 /mζ
hm|eikẑ |0i = e . (12.52)
m!
Isso vale no regime de Lamb-Dicke.
12.2.6.4 Absorption on sidebands

12.2.6.5 Lamb-Dicke regime
Consideramos uma partı́cula aprisionada num potencial harmônico com a energia,
1 1 1
Ekin = mv02 = mΩ2 z02 = ~Ω(n + ) . (12.53)
2 2 2
Como discutido na Sec. 10.4.5, a luz incidente é absorvida num espectro de bandas laterais
discretas separadas por Ω com amplitudes dadas por Jn (kv0 /Ω). O ı́ndice de modulação kv0 /Ω =
kx0 = 2πx0 /λ também é chamado de parâmetro de Lamb-Dicke. Quando a amplitude de
modulação é dentro do chamado regime de Lamb-Dicke, kv0 Ω, as primeiras bandas laterais
ficam menores do que a portadora, J1 (kv0 /Ω) < J0 (kv0 /Ω) e, portanto, não contribuem à largura
da distribuição Doppler da frequência. Ou seja, o efeito Doppler linear zera. Do outro lado, a
localização do átomo melhor que um comprimento de onda, 2πx0 λ, implica
(2n + 1)~2 k/m Ω . (12.54)
Isto é, átomos frios em estados vibracionais n baixos não podem ajustar o deslocamento do recuo
fotônico ~2 k/m dentro do espectro vibracional da armadilha. Por consequência, o recuo deve
ser absorvido pela armadilha inteira. Este é o regime de aprisionamento forte (strong binding
regime) em armadilhas de ı́ons, que é análogo ao efeito de Mößbauer discutido embaixo.
12.2.6.6 The Mößbauer effect

Comparando de radiação γ e o regime óptico,
γ para 57 F e óptico para 172 Y b+
energia do fóton ≈ 14 keV ≈ 4 × 1018 Hz ≈ 3 eV ≈ 640 THz
recuo ≈ 2 meV ≈ 500 MHz ≈ 10−10 ≈ 20 peV ≈ 5 kHz ≈ 10−11
largura da linha ≈ 5 neV ≈ 1 MHz ≈ 2 × 10−13 ≈ 5 feV ≈ 1 Hz ≈ 2 × 10−15
vibrações da rede ≈ 10 meV ≈ 3 THz ≈ 3 neV ≈ 1 MHz
Fótons γ emitidos por núcleos atômicos transferem um recuo tão grande para o núcleo, que
recebem um deslocamento Doppler tão grande, que ele não pode ser reabsorvido por outros
átomos que são em repouso. Se o núcleo estava em repouso antes da emissão, então pnucl = ~k
e, portanto, kv = ε e também ε Γ.
Se o núcleo está incorporado num cristal, aquelas frequências vibracionais que são maiores
do que a frequência de recuo fotônico, ε Γ, isto é, estamos no regime de Lamb-Dicke, então o
12.3. EXERCISES 351
fóton γ deve carregar sua própria energia de recuo, enquanto o momento do recuo é recebido pela
rede inteira. Isto é, a frequência não fica alterada, e a largura da transição é a largura natural.
Por isso, no regime de Lamb-Dicke, o efeito Doppler de primeira ordem cancela, e o átomo fica
no mesmo estado vibracional do seu potencial de localização. Isto é o efeito de Mößbauer. Além
do mais, ξ Γ. Um efeito similar pode ser observado em armadilha magneto-ópticas operadas
em transições muito finas, como discutido embaixo.
12.3 Exercises
12.3.1 Electromagnetic forces
12.3.1.1 Ex: O efeito Stern-Gerlach
Considere um condensado de Bose-Einstein de 87 Rb em uma superposição de dois estados Zee-
man. De repente um gradiente magnético de ∂z B = 100 G/cm é aplicado por 2 ms. Calcule a
separação dos centros de massa das duas partes do condensado depois de 10 ms de tempo de
voo.
12.3.1.2 Ex: Potencial de aprisionamento magnético

Invente um potencial de aprisionamento magnético.
12.3.1.3 Ex: Potenciais adiabáticos

Um potencial adiabático pode ser usado para realização de potenciais de aprisionamento mais
complicados [52]. Para estudar estes potenciais consideramos o sistema de dois estados Zeeman
m = ± 12 acoplados por uma radiação de radiofrequência ~ω. O hamiltoniano de estados vestidos
do nosso sistema de dois nı́veis é uma matriz 2 × 2,
1
µ B − 12 ~ω 1
2 ~Ω
Ĥ = 2 B 1 ,
2 ~Ω − 21 µB B + 12 ~ω
colocando o zero energético no meio entre os estados. Supõe, que o campo magnético cresce
linearmente ao longo do eixo z, B(z) = z∂z B, onde ∂z B é o gradiente, e que a radiofrequência
é sintonizada em ressonância com a diferencia das energias dos estados Zeeman numa certa
distância z0 tal que, ~ω = µB z0 ∂z B.
a. Calcule as autoenergias do sistema acoplado em função de z.
b. Expande as autoenergias em torno da posição z0 .
c. Qual seria a frequência de oscilação de átomos aprisionados dentro do potencial adiabático?
d. Expande as autoenergias em ~Ω para posições longe de ressonância.
12.3.2 Optical forces on free and confined atoms

12.3.2.1 Ex: Força dipolar para grandes dessintonias
Ω2
Verifique que no limite de grandes dessintonias o potencial dipolar Eq. (12.22) tende para −→ 4∆ .
12.3.2.2 Ex: Pressão radiativa

Calcule a força de pressão radiativa exercida sobre um átomo de estrôncio por um feixe laser
em geometria de onda plana (I = 10 mW/cm2 ) sintonizado 50 MHz embaixo da ressonância em
461 nm (Γ/2π = 30.5 MHz).
Chapter 13
Quantum measurement
Desde a fundação da mecânica quântica, a questão da relação entre o mundo e o que pode-
mos aprender sobre ele, isto é, entre a realidade e o observador, erá em primeiro plano das
discussões. Cientistas como Bohr, Heisenberg, Schrödinger e Einstein ocuparam posições con-
troversas e brigaram sobre a interpretação correta da mecânica quântica. O processo da medição
é suposto fornecer informação sobre o mundo lá fora, mas não erá claro se esta informação pode
ser completa e exata ou se existem limitações ou variáveis escondidas. Também, não erá claro se
a medida pode ser não-invasiva ou se o medidor sempre deve atrapalhar o fenômeno sob inves-
tigação. A etapa mais importante nesta questão foi a interpretação de Copenhague formulada
por Bohr, Heisenberg e Born em 1927 e elaborada por von Neumann e Dirac depois. Contestada
muitas vezes no passado, ela ainda permanece válida hoje em dia.
Neste capı́tulo vamos estudar o processo da medição visto pela mecânica quântica e discutir
alguns efeitos aparentemente paradoxos, que permitem aprofundar nosso entendimento. Entre
eles são o salto quântico, o gato de Schrödinger, o efeito Zeno quântico e paradoxo de Einstein-
Podolski-Rosen.
13.1 The observer and the reality

Seguinte a interpretação de Copenhague, previsões teóricas tem um caráter probabilı́stico. No
entanto, isto não é uma expressão da imperfeição da teoria, mas do caráter intrinsecamente
indeterminı́stico de processos quânticos 1 . Além disso, a interpretação de Copenhague renuncia
a atribuir aos objetos do formalismo quântico, como a função de onda e os operadores, uma
realidade no sentido direto. Em vez disso, os objetos do formalismo somente representam um
meio de previsão probabilı́stica dos resultados de uma medição. Estes resultados são os únicos
elementos verdadeiramente reais.
A teoria quântica e as suas interpretações são, portanto, de importância fundamental para
a visão de mundo cientı́fico e o conceito de natureza.
13.1.1 Schrödinger’s cat

No mundo microscópico, a relação entre a amostra e o observador é muito delicada. É dessa
delicadeza que surgem efeitos quânticos que parecem paradoxos através da nossa visão clássica
do mundo. Assim, não é surpreendente que uma das áreas de investigações mais fascinantes
é a interface entre os mundos clássicos e quânticos, macroscópicos e microscópicos. Para os
pioneiros da mecânica quântica a questão mais importante era do tipo: ”Como é possı́vel que
uma partı́cula microscópica voa simultaneamente através de duas aberturas?”. Hoje em dia,
1
Notamos que é problemático identificar imprevisibilidade e indeterminismo. É possı́vel que não podemos
prever eventos especı́ficos, sem ter que assumir que estes eventos ocorrem de maneira aleatória.
353
354 CHAPTER 13. QUANTUM MEASUREMENT
nós acostumamos com este fato simplesmente aceitando de considerar partı́culas como ondas.
Mas ainda não entendemos muito bem ”Porque o mundo clássico é tão diferente do mundo
quântico?”, ”Porque a mecânica quântica permite superposições quânticas de estados classica-
mente proibidas?”, ”Porque as leis fundamentais da mecânica quântica são invariáveis à respeito
da flecha do tempo, enquanto o mundo macroscópico sempre vai do passado para a futuro?”,
”Como pode ser que na mecânica quântica permite efeitos sem causa como a emissão espontânea,
enquanto o mundo cotidiano parece ser determinado?”
Figure 13.1: Fenda dupla.
Claramente, em algum limite, a mecânica quântica deve abranger a fı́sica clássica. Más
apesar do teorema de correspondência de Ehrenfest, esse fato é longe de ser trivial. Algumas
previsões da fı́sica clássica e da fı́sica quântica são fundamentalmente diferentes e, em alguns
casos, contraditórias. Os estados do gato de Schrödinger são o epı́tome desse fato: Em uma
versão do famoso paradoxo, uma partı́cula atravessa uma fenda dupla. Por trás de uma das
fendas é um detector. Se ele registra uma partı́cula, um aparelho matando um gato está acionado.
Sabemos que, na realidade quântica a partı́cula atravessa as duas fendas em um estado de
superposição, tal que o gato deveria ser num estado de superposição também. Na mecânica
quântica os gatos podem estar em uma superposição de ”morto” e ”vivo”.
Figure 13.2: A medida de um sistema quântica pressupõe a interação do sistema com um reser-
vatório perturbando a sua dinâmica.
Acreditamos hoje que as respostas das perguntas acima são, de alguma maneira escondidas
nos processos que destroem as superposições dos gatos de Schrödinger na transição do mundo
13.1. THE OBSERVER AND THE REALITY 355
microscópico até o mundo macroscópico. No entanto, os detalhes deste processo de destruição das
coerências quânticas, chamado de decoerência, são muito complicadas e representam um assunto
de pesquisa contemporânea. Na pesquisa moderna, isso é uma das motivações para tentar criar
em laboratórios os maiores (quase macroscópicos) sistemas quânticos possı́veis, colocar-lhes em
estado de gato de Schrödinger e estudar a sua decoerência 2 .
13.1.1.1 Quantum measurement

Cada sistema não perturbado segue a equação de Schrödinger. Uma vez que o hamiltoniano foi
determinado, a solução formal,
|ψi = e−iHt/~ |ψ0 i , (13.1)
permite calcular a evolução temporal, isto é, a trajetória da função de onda. A evolução é
coerente e reversı́vel no tempo.
Agora, o processo da medida de uma estado quântico puro acontece —seguinte a inter-
pretação de Copenhague e como já discutido na Sec. 2.2.7— em dois passos consecutivos: No
primeiro passo, a interação da amostra quântica com o medidor destrói todas as coerências e
projeta o estado puro numa mistura estatı́stica de estados próprios do dispositivo de medida 3 .
Seguinte von Neumann, o impacto do aparelho de medição sobre o sistema quântico é tão forte,
que a evolução coerente é interrompida e o sistema quântico se projeta sobre o grau de liberdade,
que o aparelho quer medir, e.g., a posição ou o momento de uma partı́cula, mas não os dois
no mesmo tempo. A projeção transforma um estado quântico puro |ψi dentro de uma mistura
estatı́stica de auto-estados,ρ,
X
ρamostra = |ψ(t)ihψ(t)| y ρproj = |hψ|ki|2 |kihk| . (13.2)
k
Este processo é irreversı́vel, isto é, separa o passado do futuro. Esta evolução não é descrita pela
equação de Schrödinger. Em vez disso, a redução súbita do estado deve ser postulada, como
feito no famoso axioma de von Neumann.
Num segundo passo, o observador olha para o dispositivo de medição confirmando um dos
resultados possı́veis. Assim, ele transforma o estado em um auto-estado do dispositivo:
ρproj y ρmedidor = |kihk| . (13.3)
A partir deste momento, podemos de novo deixar o sistema quântico sozinho até a próxima
medida.
Visto dentro do sistema da amostra, a evolução do processo de medida é descontinua, pois
destrói todas coerências possı́veis entre seus estados. De fato, o problema vem do comportamento
não-ideal do dispositivo de medida (simbolizado por | ↑i antes da medida). Uma medida ideal
não-invasiva 4 deixaria o estado quântico |ψi inalterado:
H
|ψi| ↑i −→ |ψi| %i , (13.4)
enquanto o dispositivo de medida muda para um estado (| %i depois da medida) indicando o
estado atual da amostra. No entanto, isto é normalmente, impossı́vel sem correlação previamente
estabelecida entre |ψi e | ↑i. Num dispositivo de medida real, o acoplamento entre |ψi e | ↑i
requer que estes sistemas sejam não ortogonais.
2
Existem tentativas de introduzir o conceito da seta temporal também no mundo microscópico: ”In an isolated
system, spontaneous processes occur in the direction of increasing entropy.” [180].
3
Notamos, que somente se todas as observáveis comutandas do sistema são medidas e registradas, torna-se
ρmedidor um estado puro. Senão ρmedidor continua sendo uma mistura parcial.
4
Vide a discussão da medida de não-demolição quântica.
13.1.1.2 Measurement-induced decoherence

Uma vista mais moderna da medida quântica é a seguinte: O mundo exterior (chamado de
reservatório) lê o sistema quântico causando, devido a esta transferência de informação, uma
destruição não-reversı́vel de coerência. Por consequência, o operador densidade condensa so-
bre a sua diagonal. Do outro lado, o sistema inteiro (incluindo o reservatório) sempre evolve
coerentemente seguindo a equação de von Neumann com o hamiltoniano de tudo Htudo :
i
ρ̇ = [ρ, Htudo ] . (13.5)
~
Sendo Hamostra o pequeno sistema quântico sob investigação, uma descrição completa do pro-
cesso de medida necessita a inclusão do observador, isto é, o hamiltoniano total é

amostra 0
H = Hamostra ⊗ Hmedidor = . (13.6)
0 medidor
Idealmente, o sistema evolui independentemente sem perturbação do medidor que representa

um reservatório. Infelizmente, isso também significa que o medidor evolui independentemente,
isto é, ele não esta influenciado pelo sistema e, portanto, não fornece informação sobre o sistema.
Para permitir a transferência de informação, precisamos acoplar os espaços respetivos por uma
interação Ω, tal que,
amostra Ω
H= . (13.7)
Ω medidor
Traçando sobre os graus de liberdade do universo menos aqueles do sistema quântico, obtemos
uma equação mestre,
i
ρ̇amostra = [ρ, Hamostra ] + Lreserv ρ . (13.8)
~
Figure 13.3: A emissão espontânea pode ser visto como acoplamento do sistema sob investigação
a um reservatório externo.
Example 51 (Medida quântica num sistema de dois qubits): Para discutir isso num
exemplo, consideramos o sistema mais simples imaginável: Dois átomos de dois nı́veis, onde o
primeiro representa o sistema que nós queremos medir e o segundo o medidor. Introduzimos
a seguinte base:
       
1 0 0 0
0 1 0 0
|1i ≡ | ↓i| ↓i =        
0 , |2i ≡ | ↑i| ↓i = 0 , |3i ≡ | ↓i| ↑i = 1 , |4i ≡ | ↑i| ↑i = 0 .
0 0 0 1
O hamiltoniano dos átomos independentes é,
H = | ↓ih↓ | ⊗ | ↑ih↑ | .
A discussão sobre a interpretação correta do processo de medida ainda está em andamento.

Teorias modernas descrevem a redução do estado em termos de decoerência quântica devido a
interações do sistema com o meio ambiente. Outras interpretações envolvem histórias decoerentes
ou assumem mundos múltiplos [202]. Na prática, o interesse atual na decoerência quântica
é motivado pelo fato que este fenômeno pode se revelar o fator limitante fundamental para
utilidade de computadores quânticos. Outra área interessante onde a mecânica quântica encontra
a fı́sica clássica é a ocorrência do caos quântico.
13.1.2 The quantum jump

Obviamente, o processo da medida quântica inclusive a descontinuidade na projeção do estado
poderia ser completamente entendido dentro de um tratamento completo do sistema que incluiria
o dispositivo de medida. Na prática, isto é ilusório devido ao número excessivo de graus de
liberdade do dispositivo clássico de medida (p.ex., um gato de Schrödinger).
Do outro lado, muitas caracterı́sticas da medição quântica podem ser ilustradas num simples
átomo de três nı́veis com uma transição fraca, representando a amostra quântica, e uma transição
forte, representando o medidor. A tese defendida no seguinte é, que este sistema de três nı́veis,
chamado de amplificador quântico, dá uma visão profunda do que acontece no processo da
redução do estado e, por isso, pode ser considerado paradigmático para a medida quântica.
Para discutir melhor a dinâmica deste sistema, vamos primeiramente introduzir o método
da simulação de Monte-Carlo da função de onda.
13.1.2.1 Quantum Monte-Carlo wavefunction simulation of the two-level system

A ocorrência possı́vel de emissão espontânea produz uma dinâmica chamada de trajetória quântica,
que pode ser descrita por um hamiltoniano efetivo não-hermitiano,

+ i 0 Ω
Ĥef f = ~∆σz + ~Ωσ + c.c. − Γσz = , (13.9)
2 Ω ∆ − i Γ2
no âmbito de incluir processos de dissipação de energia. O problema deste hamiltoniano é, ele
†
† −iĤef f t 6= eiĤef f t .
também não é comutável, [Ĥef f , Ĥef f ] 6= 0, e a sua dinâmica não é unitária, e
Isso significa, que já a possibilidade de emissão espontânea impede a reversibilidade da dinâmica.
Observamos um decrescimento temporal da norma hψ(t)|ψ(t)i indicando uma perda de energia.
que a norma está diminuindo
hψ|ψi = hψ0 |e−iĤt eiĤt |ψ0 i −→ e−Γt . (13.10)
A perda da normalização durante a evolução até a ocorrência do próximo salto quântico é devido
a dissipação de energia dentro do reservatório,

ρamostra 0
Tr ρamostra → 0 enquanto Tr =1, (13.11)
0 ρreserv
e representa uma medida da probabilidade que um processo irreversı́vel tem sido ocorrido.
Processos dissipativos podem ser simulados jogando dados com números aleatórios ζ. Dividi-
mos o tempo em pequenos intervalos dt e propagamos a função de onda de ψ(t) para ψ(t + dt).
Depois de cada intervalo perguntamos, qual fui a probabilidade acumulada no tempo do inter-
valo para ocorrência de uma transição, 1 − hψ(t)|ψ(t)i. A transição é abrupta e se chama de
salto quântico. Agora, geramos um número aleatório ζ, uniformemente distribuı́do entre 0 e 1,
que nós comparamos á probabilidade. No caso, ζ > 1 − hψ(t)|ψ(t)i, concluı́mos que não houve
salto quântico, e deixamos o sistema prosseguir em paz, só renormalizando a função de onda
para compensar as perdas [178, 59]. No caso contrário, ζ < 1 − hψ(t)|ψ(t)i, concluı́mos que
houve salto quântico, e o sistema está projetado no auto-estado ψ0 . Agora, a evolução começa
de zero. A simulação pode ser feita assim,
!
(1−iĤdt)|ψ(t+dt)i
se ζ > 1 − hψ(t)|ψ(t)i
|ψ(t)i y |ψ(t + dt)i ≡ hψ(t)|ψ(t)i . (13.12)
|ψ0 i se ζ < 1 − hψ(t)|ψ(t)i
Essa trajetória simula só uma de muitas trajetórias possı́veis do sistema.

Cada salto quântico projeta sobre o estado fundamental e constitui uma medição, pois nos
fala que fluorescência tem sido detectada. O nosso sistema de três nı́veis, portanto, pode ser
considerado o protótipo de um aparelho de medida quase ideal, o objeto medido sendo a pop-
ulação do estado metaestável e o dispositivo de medição sendo a transição dipolar. Medições
demasiamento frequentes perturbam a evolução coerente do sistema e inibem a sua dinâmica:
O estado metaestável não pode mais ser excitado: Quanto mais um observador tenta extrair
informação do sistema, tanto mais este para de evoluir. Este efeito é conhecido como efeito de
Zeno quântico.
A evolução da matriz densidade segue na média sobre todas as trajetórias possı́veis, ρ(t) =
|ψ(t)ihψ(t)|. A emissão espontânea é incluı́da pela possibilidade do sistema de subir uma redução
do estado ou projeção sobre um auto-estado com a probabilidade ζ que corresponde a probabil-
idade acumulada para emissão espontânea. ζ é simulado por um número aleatório distribuı́do
uniformemente. A modificação de |ψ(t)i por não-observação de emissão espontânea reduza a
população do estado excitado por 1− 12 Γdt, enquanto a população do estado fundamental fica in-
alterada. Isso significa que |ψ(t)i deve ser renormalizado. Isso é o método de simulação quântica
de Monte-Carlo da função de onda.
Saltos quânticos foram observados em sistemas de três nı́veis [191, 221, 222, 27].
13.1.2.2 Three-level systems: The epitome of quantum measurement

A respeito da interação misteriosa entre a amostra e o medidor, muito pode ser entendido por
uma comparação entre dois procedimentos possı́veis: 1. tratar a amostra e o medidor separada-
mente e explicar a extração de informação seguinte o postulado de von Neumann; 2. tratar a
amostra e o medidor por uma teoria unificada.
Um dos sistemas mais simples imagináveis permitindo esta comparação é o sistema de três
nı́veis com duas transições excitadas por campos de radiação e conectadas por um estado fun-
damental comum. Como ilustrado na Fig. 13.5(a) este sistema de três nı́veis pode ser um átomo
com uma transição forte (p.ex., a transição S1/2 − P1/2 num ı́on de bário Ba+ ) e uma transição
fraca (p.ex., a transição quadrupolar proibida S1/2 − D5/2 no mesmo átomo). Chamamos a
transição forte de objeto sob medida e a transição fraca de dispositivo de medida. Claramente,
este sistema permite estudar o processo de medida de von Neumann inclusive a observação di-
reta de saltos quânticos. No mesmo tempo, ele é suficientemente simples para se prestar a uma
descrição teórica completa. Assim, o sistema de três nı́veis se torna o epitome de um dispositivo
de medida quântica.
11
0.5
ρ
0
0 50 100 150 200
1
11
0.5
ρ
0
0 50 100 150 200
t (ns)
Figure 13.4: (Code: LM M easurement T wolevelM onteCarlo.m) Simulação de Monte-Carlo

quântica da função de onda.
A pergunta agora é:
1. se existem processos de redução súbita de estados na realidade e se eles podem ser obser-
vados em saltos quânticos entre estados 5 ;
2. ou se eles devem desaparecer dentro de uma teoria de tudo.
Tornamos nossa atenção para o átomo de três nı́veis: Obviamente, o átomo vai preferencial-
mente espalhar fótons na transição forte. No entanto, quando o elétron de valência é ’arquivado’
ou ’engavetado’ no estado excitado, nada de fluorescência pode ser observada na transição forte
dipolar. Entendemos a transição fraca como o objeto sendo medido pelo medidor, que neste
caso, é a transição forte.
Figure 13.5: (a) Medida quântica no exemplo do átomo de três nı́veis incorporando uma transição
fraca observada e uma transição forte observando. (b) Signal Random Telegraph na fluorescência
ressonante devido a saltos quânticos.
5
’If we have to go on with these damned quantum jumps, then I’m sorry that I ever got involved.’
(E. Schrödinger, 1952).
13.1.2.3 Quantum Monte-Carlo wavefunction simulation of the quantum amplifier

Para evitar o impedimento da evolução do sistema (transição fraca) pela presencia do laser
medindo a transição forte podemos implementar so chamado amplificador quântico de Dehmelt,
que consiste em irradiar de maneira alternante o laser de sistema (na fase S − D) e o laser de
medida na fase (S − P ). A ausência do laser S − P permite evitar alargamento de saturação 6 ,
desdobramento Stark dinâmico, e deslocamento de luz (light shift).
A irradiação alternada dos lasers S − D e S − P também pode ser tratada pelo método de
simulação quântica de Monte-Carlo da função de onda (13.12) usando o hamiltoniano
 1 1

0 2 Ωsp 2 Ωsd
Ĥef f =  21 Ωsp −∆sp − 2i Γsp 0  , (13.13)
1
2 Ωsd 0 −∆sd
onde as frequência de Rabi Ωsd se Ωsd são ligadas alternativamente.

Na simulação 13.6 os saltos quânticos para estados engavetados aparecem com perı́odos
longos sem população no P1/2 (primeiro perı́odo S − P , onde a população do S1/2 ilustrada
pela curva vermelha tende para 0 para tempos longos). Isto é, está redução do estado por não-
observação precisa de um tempo de evolução finito, que simplesmente vem da incerteza se a
não observação é realmente devido a um engavetamento ou a ausência acidental de processos de
espalhamento na transição S − P : Afinal não é previsı́vel, quando o próximo fóton da emissão
espontânea será emitido, mesmo o tempo de vida do estado excitado sendo curto. Só que
para tempos de observação mais longos é cada vez mais improvável, que a ausência de fótons
não é devido a um engavetamento. É está improbabilidade que faz convergir a população
rapidamente para o estado metaestável. No segundo perı́odo observamos transições rápida para
o P1/2 seguidas de decaimentos súbitos para o estado fundamental. Estes correspondem a fótons
espalhados pela transição forte.
13.1.2.4 Comparison with Bloch equations

Alternativamente a simulação quântica da equação de Schrödinger efetiva, a evolução pode
ser descrita por uma matriz densidade, cuja evolução é governada pela equação mestre, neste
contexto chamada de equação de Bloch traços verdes na Fig. 13.6. Obviamente, como esta
equação trata médias de muitas trajetórias, não pode produzir saltos quânticos. No entanto, a
equação mestre descreve o conjunto objeto-medidor, mostrando que também podemos considerar
o conjunto, isto é, os dois sistemas de dois nı́veis como partes do mesmo sistema de três nı́veis.
Neste sistema global o postulado de von Neumann da redução do estado é substituı́do por
uma separação da escala temporal dentro da qual a medida é feita e da escala dentro da qual o
processo observado acontece.
13.1.2.5 Interpretation of quantum jumps

Toda informação sobre átomos individuais é recolhido a partir de fótons espontâneos. Essa
observação projeta o estado de excitação interna, mas esse conceito não está incluı́do na descrição
das equações de Bloch. No entanto, eles podem ser generalizadas da seguinte maneira [183] ou
[263]. Projetamos o operador de densidade total ρAF do átomo mais o campo (ou seja, o
6
Isso é equivalente ao efeito quântico de Zeno, pois o alargamento de saturação (ou efeito Stark dinâmico) do
nı́vel fundamental, que é comum as duas transições forte e fraca, reduz o overlap espectral entre transição e laser
e portanto a probabilidade de excitação do nı́vel metaestável.
S-D S-P S-D S-P S-D

1
D5/2
0.5
0
1
S1/2 0.5
0
1
P1/2
0.5
0
0 200 400 600 800 1000
t (ns)
Figure 13.6: (Code: LM M easurement QJM onteBloch.m) Amplificador quântico comparando

simulações quânticas de Monte-Carlo (linhas vermelhas) e as equações de Bloch (linhas pretas),
que coincidem bem com as médias sobre 100 trajetórias de Monte-Carlo (linhas verdes). Os
perı́odos S − D (fundo branco) representam a evolução livre do sistema quântico, os perı́odos
S − P (fundo amarelo) representam os perı́odos de medição.
reservatório dos modos do vácuo) para o subespaço de estados tendo exatamente n de fótons no
reservatório,
ρ(n) = TraceF (P (n) ρAF ) , (13.14)
e derivamos, à partir da equação de von Neumann, a equação mestre para o estado atômico
ρ(n) sob a restrição de um número fixo de fótons n no campo. A partir da equação mestre,
derivamos agora as equações de movimento para o vetor de Bloch %(n) . Eles diferem apenas em
(n)
um termo das equações de Bloch: A expressão Γ12 ρ22 , que descreve o decaimento espontâneo
da população do nı́vel excitado da ressonância de monitoramento, é substituı́da pela expressão
(n−1)
Γ12 ρ22 :
d (n)
% = (L − |1iΓ12 h2|) %(n) + |1iΓ12 h2|%(n−1) , (13.15)
dt
P (n)
A condição do traço para o subestado com o número de fótons n é violada: j ρjj 6= 1. A
explicação fı́sica para isso é a seguinte: Enquanto a emissão induzida e a absorção mantêm o
número de fótons no sistema combinado luz-átomo, a emissão espontânea diminuı́ o número de
fótons. À redução do estado de superposição atômica nas equações de Bloch corresponde, nas
equações modificadas (13.14), o colapso do subespaço descrito por %(n) com a constante de tempo
Γ12 e a transição do sistema para o outro subespaço %(n−1) , cuja evolução é guiada por outra
equação diferencial correspondente, agora, para n − 1 fótons. Cada detecção de fluorescência
no momento t = 0 determina a condição inicial para o desenvolvimento futuro do sistema para
ρ(n) (0) = 0 e ρ(n−1) (0) = |1ih1|. A densidade de probabilidade c(t) para uma nova observação
de emissão espontânea no tempo t com a eficiência de detecção η, ou em outras palavras, o
histograma da durações dos perı́odos de escuridão no sinal de fluorescência relacionada com a
solução %(n) da parte homogênea da equação (13.14):

4
X
(n) d (n)
c(t) = ηΓ12 ρ̃22 (t) = η ρ̃ (t) , (13.16)
dt jj
j=1
O segundo passo segue imediatamente à partir da parte homogênea da equação (13.14).
13.1.3 Weak measurements

Medições fortes deixam o sistema quântico medido num auto-estado sem incerteza. Mas é
possı́vel imaginar, que o dispositivo de medido não interage fortemente com sistema quântico,
tal que o sistema não é fortemente perturbado. O preço a pagar será, que o resultado da medida
fica incerto (teorema no free lunch).
Consideramos o uso de uma ancilla (donzela, grau de liberdade adjunto), por exemplo, um
campo ou uma corrente, para sondar um sistema quântico. A interação entre o sistema e a
sonda correlaciona os dois sistemas.
13.1.3.1 Weak interaction and measurement by coupling to an ancilla

Consideramos um sistema inicialmente no estado quântico |ψi e uma ancilla inicialmente no
estado |φi, tal que o estado combinado é, |Ψi = |ψi ⊗ |φi. Os dois sistemas interagem através
do hamiltoniano H = A ⊗ B, que gera a evolução temporal U (t) = e−ixtH (em unidades onde
~ = 1), onde x é a força da interação (em unidade de frequência angular). Assumimos um
tempo de interação fixo t = ∆t tal, que λ = x∆t é muito pequeno, isto é, λ3 ≈ 0. A expansão
de U em λ dá,
λ2 2 λ2 2
U ≈ 1 ⊗ 1 − iλH − 2 H + O(λ3 ) = 1 ⊗ 1 − iλA ⊗ B − 2 A ⊗ B2 . (13.17)
Quando é suficiente expandir a transformação unitária em baixas ordens de teoria de per-
turbação, falamos de interação fraca. Como λ e λ2 são pequeno, o estado combinado depois da
interação não será muito diferente do estado inicial,
λ2 2
|Ψ0 i = (1 ⊗ 1 − iλA ⊗ B − 2 A ⊗ B 2 )|Ψi . (13.18)
Agora fazemos uma medida sobre a ancilla para extrair informação do sistema. Isto é
chamado de medida P mediada por ancilla. Consideramos medidas na base |qi (do sistema da
ancilla), tal que q |qihq| = 1. A ação da medida sobre o sistema total é descrita pela ação do
projetor Πq = 1 ⊗ |qihq| sobre |Ψ0 i. Seguinte a teoria da medida quântica, o estado condicional
depois da medida é,
P
λ2 2 2
0
Πq |Ψ i 1 ⊗ |qihq| 1 ⊗ k |kihk| − iλA ⊗ B − 2 A ⊗ B
|Ψq i = p = |ψi ⊗ |φi (13.19)
0 0
hΨ |Πq |Ψ i N
P 2
1 ⊗ k |qihq|kihk| − iλA ⊗ |qihq|B − λ2 A2 ⊗ |qihq|B 2
= |ψi ⊗ |φi
N
2
1hq|φi − iλAhq|B|φi − λ2 A2 hq|B 2 |φi
= |ψi ⊗ |qi ,
N
p
onde N = hΨ0 |Πq |Ψ0 i é o fator de normalização para a função de onda. Note, que o estado da
ancilla recorde o resultado da medida. O objeto
λ2 2
Mq ≡ hq|e−iA⊗B |φi ' 1hq|φi − iλAhq|B|φi − 2
2 A hq|B |φi (13.20)
é um operador sobre o espaço Hilbert total e chamado de operador de Kraus. A respeito dos
operadores de Kraus o estado do sistema combinado depois da medida é,
Mq |ψi
|Ψq i = q ⊗ |qi . (13.21)
hψ|Mq† Mq |ψi
Os objetos Eq = Mq† Mq são elementosP da chamada POVM ou Positive Operator Valued

(Probability) Measure e devePobedecer q E Pq = 1, tal que as probabilidades correspondentes
se adicionam para unidade: q Pr(q|ψ) = q hψ|Eq |ψi = 1. O sistema da ancilla não é mais
correlacionado com o sistema primário. Ele simplesmente recorde o resultado da medida, tal que
podemos calcular o traço sobre ele. Fazendo isso, chegamos ao estado condicional do sistema
primário sozinho,
Mq |ψi
|ψq i = q , (13.22)
hψ|Mq† Mq |ψi
que ainda etiquetamos com o resultado da medida q. De fato, estas consideração permitem
derivara uma trajetória quântica.
13.1.3.2 Kraus operator for position measurement
Como exemplo canônico de um operador de Kraus usamos [18, 42] pegamos H = x ⊗ p, onde a
posição e o momento satisfazem a relação de comutação, [x, p] = i. A função inicial da ancilla
seja uma distribuição gaussiana,
Z
1 02 /4σ 2
|φi = dq 0 e−q |q 0 i (13.23)
(2πσ 2 )1/4
A função de onda de posição da ancilla é,
1 2 2
φ(q) = hq|φi = 2 1/4
e−q /4σ . (13.24)
(2πσ )
Os operadores de Kraus são (em comparação com a discussão precedente colocamos λ = 1)
1 2 2
M (q) = hq|e−ix⊗p |φi = 2 1/4
e−(q−x) /4σ , (13.25)
(2πσ )
pois o operador e−ix⊗p faz uma translação espacial quando aplicado no grau de liberdade da
posição. Os elementos POVM correspondentes são,
1 2 /2σ 2
E(q) = Mq† Mq = √ e−(q−x) , (13.26)
2πσ 2
R
que obedecem dqE(q) = 1.
Calcular hψq |ψq i = hψ 0 |M (q)† M (q)|ψ 0 i.
Note que limσ→0 E(q) = |x = qihx = q|. Isto é, num limite particular, estes operadores
convergem para uma medida forte da posição. Para σ → ∞, falamos de medida fraca.
Outro exemplo seria o átomo de três nı́veis do exemplo do amplificador quântico de Dehmelt.
13.2 Open systems and the master equation

Vamos derivar a equação mestra de um sistema quântico aberto. Como pressuposto comum,
assumimos que o meio ambiente (chamado de banho) e o sistema sob consideração são sistemas
quânticos no sentido de que (1) os graus de liberdade relevantes são completamente caracteriza-
dos por vetores de estado (ou matrizes densidade), e (2) a evolução temporal do sistema total é
unitária U (t) = e−iĤt . O hamiltoniano total, Ĥ = Ĥsys + Ĥres + V é assumido ser independente
do tempo e consiste de três partes, a saber hamiltoniano do sistema Ĥsys , o hamiltoniano do
banho (ou reservatório) Ĥres e a interação V entre o sistema e o banho. O objetivo da equação
mestre é encontrar a dinâmica do sistema traçando sobre todos os graus de liberdade do banho.
Isso nem sempre é possı́vel, e vamos supor que a interação V é suficientemente fraca de modo
que a teoria de perturbação é aplicável.
Na representação de interação a evolução da matriz densidade total ρ̂tot fica
dρ̃tot
i~ = [Ṽ (t), ρ̃tot ] . (13.27)
dt
onde ρ̃tot (t) ≡ U0† ρtot U0 , e Ṽ (t) ≡ U0† V U0 e U0 = e−i(Ĥsys +Ĥres )t/~ . Essa evolução é, por
enquanto, muito geral, e a solução pode ser escrita formalmente
Z t
1
ρ̃tot (t) = ρ̃tot (0) + dt1 [Ṽ (t1 ), ρ̃tot (t1 )] . (13.28)
i~ 0
Iterando mais uma vez:

Z t Z t Z t1 h i
1 1
ρ̃tot (t) = ρ̃tot (0) + dt1 [Ṽ (t1 ), ρ̃tot (0)] + dt1 dt2 Ṽ (t1 ), [Ṽ (t2 ), ρ̃tot (t2 ] .
i~ 0 (i~)2 0 0
(13.29)
No seguinte, vamos chamar várias aproximações para simplificar os cálculos, especialmente a
aproximação de Born, a suposição de um estado inicial de produto e, mais tarde, a aproximação
de Markov.
13.2.1 Born approximation for weak coupling

Aqui assumimos a interação Ṽ é fraca. Então, podemos esperar que, repetindo o processo
iterativo, a série convergirá e escrever a solução geral como,
X Z t Z tn−1 h i
1
ρ̃tot (t) = ρ̃tot (0) + dt1 ... dtn Ṽ (t1 ), ..., [Ṽ (tn ), ρ̃tot (0)] . (13.30)
(i~)n 0 0
n≥1
Esta maneira de terminar uma equação iterativa por ρtot (0) é geralmente conhecida como a
aproximação de Born. Aqui, vamos apenas até a segunda ordem em Ṽ . Tomando o traço sobre
o banho,
ρ̃sys (t) = Trres ρ̃tot (t) , (13.31)
extraı́mos a matriz densidade só do sistema, dando,
Z t Z t Z t1 h i
1 1
ρ̃sys (t) = ρ̃sys (0)+ dt1 Trres [Ṽ (t1 ), ρ̃tot (0)]+ dt1 dt2 Trres Ṽ (t1 ), [Ṽ (t2 ), ρ̃tot (0)] .
i~ 0 (i~)2 0 0
(13.32)
13.2. OPEN SYSTEMS AND THE MASTER EQUATION 365
13.2.2 Assumption of an initial product state

Em seguida, precisamos fazer a suposição bastante importante, que o estado inicial entre o
sistema e o meio ambiente não sejam correlacionados, ou matematicamente falando, eles podem
ser escritos como estados produtos,
ρ̃tot (0) = ρ̃sys (0) ⊗ ρres (0) . (13.33)
Um outro pressuposto não é essencial, mas muitas vezes válido é, que Trres [Ṽ (t1 ), ρ̃tot (0)] = 0.
Neste caso, o termo de primeira ordem zera. Em segunda ordem, podemos escrever,
Z t Z t1 h i
M(t) 1
ρ̃sys (t) = e ρ̃sys (0) onde M(t)χ ≡ dt1 dt 2 Trres Ṽ (t1 ), [Ṽ (t2 ), χ ⊗ ρ res ] ,
(i~)2 0 0
(13.34)
é um superoperator agindo sobre o operador densidade do sistema. Pegando a derivada temporal,
temos a equação mestre explicita,
Z t h i
dρ̃sys (t) d 1
= Lρsys (t) = (M(t)ρ̃sys (t)) = dτ Trres Ṽ (t), [Ṽ (τ ), ρ̃sys (t) ⊗ ρres ] .
dt dt (i~)2 0
(13.35)
O superoperador L se chama operador de Lindbladt.
13.2.3 Markov approximation for short memory

Aqui, temos de avaliar os termos envolvendo considerando a média em relação ao banho ter-
mal, que se supõe ter memória curta, no sentido de que o tempo de correlação é muito curto.
Matematicamente,
Z t Z t
dτ Trres Ṽ (t)Ṽ (τ )ρres = dτ Trres Ṽ (t − τ )Ṽ (0)ρres (13.36)
0
Z0 ∞
' dτ Trres Ṽ (t − τ )Ṽ (0)ρres .
0
Em outras palavras, a função de correlação de dois pontos é significativo somente quando t ' τ ,
e é válido estender o limite superior para o infinito. Esta é a aproximação de Markov.
13.2.4 Example: Two level system with spontaneous emission

Como exemplo consideramos agora a equação mestre do movimento browniano de um oscilador
harmônico quântico. Esta pode ser escrita,
Z t
dρ̃sys 1 Ṽ (t)Ṽ (τ )ρ̃(t) ⊗ ρres − Ṽ (t)ρ̃sys (t) ⊗ ρres Ṽ (τ )
= dτ Trres . (13.37)
dt (i~)2 0 −Ṽ (τ )ρ̃sys (t) ⊗ ρres Ṽ (t) + ρ̃sys (t) ⊗ ρres Ṽ (τ )Ṽ (t)
O acoplamento do sistema ao banho é assumido como sendo da forma,

Ṽ = ~ â† Γ̂(t)eiΩt + âΓ̂† (t)e−iΩt , (13.38)
P
onde Γ̂(t) = k gk b̂k e−iωk t , os operadores bosônicos â e b̂k agem respectivamente sobre o sis-
tema (com a frequência Ω) e o banho (com a frequência ωk ). Aqui gk caracteriza a força do
acoplamento entre os osciladores do sistema e do banho. Portanto,
13.2.5 Example: Damped harmonic quantum oscillator

Como exemplo consideramos agora a equação mestre do movimento browniano de um oscilador
harmônico quântico. Esta pode ser escrita,
Z t
dρ̃sys 1 Ṽ (t)Ṽ (τ )ρ̃(t) ⊗ ρres − Ṽ (t)ρ̃sys (t) ⊗ ρres Ṽ (τ )
= dτ Trres . (13.39)
dt (i~)2 0 −Ṽ (τ )ρ̃sys (t) ⊗ ρres Ṽ (t) + ρ̃sys (t) ⊗ ρres Ṽ (τ )Ṽ (t)
O acoplamento do sistema ao banho é assumido como sendo da forma,

† iΩt † −iΩt
Ṽ = ~ â Γ̂(t)e + âΓ̂ (t)e , (13.40)
P
onde Γ̂(t) = k gk b̂k e−iωk t , os operadores bosônicos â e b̂k agem respectivamente sobre o sis-
tema (com a frequência Ω) e o banho (com a frequência ωk ). Aqui gk caracteriza a força do
acoplamento entre os osciladores do sistema e do banho. Portanto,
 † Γ(t)eiΩt + aΓ† (t)e−iΩt
†
iΩτ + aΓ† (τ )e−iΩτ ρ̃(t) ⊗ ρ


 a a Γ(τ )e res 

Z t  † Γ(τ )eiΩτ + aΓ† (τ )e−iΩτ 
dρ̃sys − a† Γ(t)eiΩt + aΓ† (t)e−iΩt ρ̃(t) ⊗ ρ res a
=− dτ Trres .
dt 0 
 − a† Γ(τ )eiΩτ + aΓ† (τ )e−iΩτ ρ̃(t) ⊗ ρres a† Γ(t)eiΩt + aΓ† (t)e−iΩt  
 
+ρ̃(t) ⊗ ρres a† Γ(τ )eiΩτ + aΓ† (τ )e−iΩτ a† Γ(t)eiΩt + aΓ† (t)e−iΩt
(13.41)
Vamos ter um olhar mais atento sobre um dos termos,
Z t n o Z t D E
† † −iΩτ †
T ≡− iΩt
dτ Trres a Γ(t)e aΓ (τ )e ρ̃(t) ⊗ ρres = −a aρ̃(t) dτ Γ(t)Γ† (τ ) eiΩt e−iΩτ .
0 0 res
(13.42)
Teremos de avaliar quantidades, como
Trres (V (t)V (s)ρres ) = ~2 a† ahΓ(t)Γ† (t)ires eiΩ(t−s) + ~2 aa† hΓ† (t)Γ(t)ires e−iΩ(t−s) , (13.43)

onde hΓ(t)Γ† (t)ires ≡ Trres Γ(t)Γ† (t)ρres , e para o banho termico, hb†j bk i = δjk nk and hbj b†k i =
−1
δjk (1 + nk ) and nk = eβ~ωk − 1 . Portanto,
X D E Z t X Z ∞
†
T = −a aρ̃(t) gj gk bj b†k dτ e i(ωj t−ωk τ ) iΩ(t−τ )
e †
' −a aρ̃(t) gk2 nk dτ ei(ωk −Ω)(t−τ ) .
res 0 0
j,k k
(13.44)
Em seguida, teremos de usar a relação
Z ∞
dτ e±iετ = πδ(ε) ± iP V , (13.45)
0
onde P V denota a Cauchy parte do valor principal. Estas correspondem ao ’deslocamento de

Lamb’ e ’shift Stark’ na frequência, e é considerado pequeno em comparação com Ω e deve ser
negligenciado aqui.
X −1 Z ∞ X −1
†
T = −a aρ̃(t) 2
gk e β~ωk
−1 dτ ei(ωk −Ω)(t−τ ) = −a† aρ̃(t) gk2 eβ~ωk − 1 πδ(ωk − Ω)
k 0 k
(13.46)
X −1 X γ
= −a† aρ̃(t) gk2 eβ~Ω − 1 π = −πN a† aρ̃(t) gk2 δ(ωk − Ω) = N aa† ρ̃(t)π . (13.47)
2
k k
13.3. REPEATED MEASUREMENTS 367
−1 P 2
onde N ≡ eβ~Ω − 1 . Define γ2 ≡ k gk δ(ωk − Ω). O procedimento pode ser repetido
por todos os termos da equação mestre. Temos, então, a equação mestra para um oscilador
harmônico amortecido,
dρ̃ γ γ
= (N + 1) 2aρ̃a† − a† aρ̃ − ρ̃a† a − N 2a† ρ̃a − aa† ρ̃ − ρ̃aa† . (13.48)
dt 2 2
13.2.6 Thermalization
Para completar a discussão, vamos considerar o tempo de desenvolvimento do número de fótons
significa ha† ai. Note que Trres aa† ρ̃ = Trres aa† ρ, e pode ser útil para usar (com n̂ = a† a)
n̂a = an̂ − a e n̂a = an̂ − a para simplificar o lado direito da equação mestre. temos
dha† ai
= −γha† ai + γN , (13.49)
dt
ea solução para esta equação é
hn(t)i = hn(0)ie−γt + N (1 − e−γt ) , (13.50)

−1
o que sugere que hn(t → ∞)i → N = eβ~Ω − 1 , como esperado para a razão de termalização
[212]. Para uma discussão sobre a validade sobre a aproximação Born-Markov, ver [188]. Relação
entre a aproximação de Markov e Regra de Ouro de Fermi, ver [3].
13.3 Repeated measurements

13.3.1 The quantum Zeno effect
O famoso problema inventado pelo filosofo grego Zeno (490-430 AC) vai assim: Aquiles e uma
tartaruga organizam uma corrida. O Aquiles arrogante deixe uma vantagem de 100 metros para
a tartaruga, mas ele acha que nunca vai conseguir ultrapassar a tartaruga. Pois no instante
que ele cobriu os 100 metros a tartaruga avançou de 10 metros, e assim para frente. Na versão
quântica podemos imaginar um Gedankenexperiment onde um feixe laser passa através de uma
π
serie infinitamente densa de n polarizadores, cada um sendo rotado por um ângulo 2n à respeito
do precedente. Cada polarizador representa uma medida instantânea da polarização do feixe.
O resultado deste arranjo é que a medição contı́nua do sistema governa a sua evolução e roda a
polarização por um ângulo de π/2. Um experimento similar pode ser imaginado por uma serie
de medidas de Stern-Gerlach do spin de um átomo.
Em cada versão do efeito Zeno, o sistema está inibido de se desenvolver livremente por causa
de medidas frequentes do seu estado atual. Aquiles certamente seria capaz de ultrapassar a
tartaruga se ele não sempre olhasse para ela para avaliar a distância restante [137] 7 .
Consideramos um sistema descrito pelo hamiltoniano Ĥ. A equação de Schrödinger descreve
a evolução temporal da função de onda |ψ(t)i = e−iHt/~ |ψ0 i. Podemos então calcular a amplitude
e a probabilidade do estado permanecer no estado inicial,
hψ0 |ψ(t)i = hψ0 |e−iĤt/~ |ψ0 i e P (t) = |hψ0 |ψ(t)i|2 . (13.51)

7
O efeito de Zeno quântico foi frequentemente usado para justificar a relevância fı́sica do postulado de redução
do estado. Foi mostrado, no entanto, que este postulado não é essencial para o entendimento do efeito quântico
de Zeno [22]. O efeito já segue diretamente da equação de Schrödinger e portanto tem uma natureza puramente
dinâmica. Isso mostra que a projeção é um construto puramente matemático sem realidade fı́sica. Isso pode ser
visto na discussão do amplificador quântico proposto pelo Hans Dehmelt que considerou um sistema atômico de
três nı́veis excitado numa transição fraca e medido por saltos quânticos numa transição forte (vide Sec. ??).
A equação de Schrödinger nos leva inevitavelmente a termos em tempos curtos, esse padrão é
predominado por uma função quadrática que foi batizada por Misra e Sudarshan em 1977 por região
“Zeno”, em alusão ao famoso filósofo Zenão de Eleia que propôs o paradoxo da flecha.
Zenão propôs vários paradoxos, no caso específico da flecha, ele diz que se uma flecha em voo
instantaneamente ocupa sempre o seu espaço, e que algo parado também ocupa sempre o seu mesmo
368 espaço, então uma flecha em vooCHAPTER 13. QUANTUM
em qualquer instante MEASUREMENT
também está em repouso.
Figure 13.7: Decaimento quadrático da população de um estado sujeito à decaimento.

Fig.1- Gráfico dos distintos tempos de um sistema instável.
Para tempos curtos podemos expandir
i 1
hψ0 |ψ(δt)i = |ψ0 i − Ĥδt/~|ψ0 i − 2 Ĥδt2 |ψ0 i + ... = |ψ0 i + |δψi , (13.52)
~ 2~
tal que
i 1 1
hψ0 |ψ(t)i ' 1 − Ĥδt − 2 Ĥδt2 e P (t) ' 1 − (hĤ 2 i0 − hĤi20 )δt2 . (13.53)
~ 2~ ~2
q
Dessa forma podemos explicitar o tempo de Zeno a partir das equações acima, τZ = ~2 / hĤ 2 i0 − hĤi20 .
Realizamos agora medidas de von Neumann sucessivas. Realizando ao total N medidas ao longo
de um tempo t, isso nos leva a uma frequência de medida τ −1 . Iremos fazer essas medidas a fim
de verificar se o sistema esta ainda no seu estado inicial, porém a cada medida nosso sistema é
projetado de volta ao estado inicial seguindo novamente seu processo de transição. A população
do estado inicial então será dada por:
P (N ) (τ ) = P (N ) (t/N )N . (13.54)
A Fig. 13.8 mostra a trajetória para cinco medidas com intervalos de tempo τ . A linha tracejada
mostra a população caso não fosse feita medida alguma, notamos uma diferença muito grande,
pois a cada medida o sistema retorna ao estado de evolução que está em regime de tempos
curtos. Há ainda de se notar que caso extrapolássemos o numero de medidas para infinito nossa
probabilidade iria cada vez mais se aproximar da unidade,
" 2 #N
t N grande 2 /N τ 2 N →∞
1− −→ e−t Z −→ 1 . (13.55)
N τZ
Vide Exc. 13.7.3.1 e 13.7.3.2 8

A supressão da evolução de um sistema quântico foi observada experimentalmente em al-
guns sistemas microscópicos. Em 1989, Itano, Heinzen, Bollinger e Wineland [137] conseguiram
observar o efeito Zeno quântico. Utilizando ions de berı́lio e um pulso ressonante eles contaram
o número de fótons emitidos devido a relaxação do sistema. Depois disso, usando um pulso
8
A referencia [PRA50, 4582 (1994)] [22] que discute Itano et al. [137] mostra que o postulado da redução do
estado não é essencial para o entendimento do efeito Zeno quântico. O efeito segue diretamente da equação de
Schrödinger e é de natureza puramente dinâmica. Enquanto o efeito de Zeno quântico foi frequentemente usado
para justificar a relevância fı́sica do postulado da redução do estado, agora parece que a projeção é um construto
puramente matemático sem realidade fı́sica (vide a discussão dos saltos quânticos no sistema de três nı́veis [?].
13.3. REPEATED
PodemosMEASUREMENTS
ilustrar esse efeito com um gráfico. 369
Figure 13.8: Inibição

Notemosdo
quedecaimento damedidas
foram feitas cinco população de um
com intervalos estado
de tempo sujeito ver
τ e podemos à decaimento
pela linha por me-
didas repetidas
tracejada(aqui paradeNsobrevivência
a expectativa = 5). Adolinhaestadotracejada
caso não fosse(sólida) representa
feita medida a probabilidade
alguma, notamos uma de
diferença muito grande, pois a cada medida o sistema retorna ao estado de evolução
sobrevivência com (sem) as medidas. A linha cinza representa uma função exponencial de in- que está em regime
de tempos curtos. Há ainda de se notar que caso extrapolássemos o numero de medidas para infinito
terpolação.
nossa probabilidade iria cada vez mais se aproximar da unidade, e esse resultado é absurdamente
surpreendente!
ultravioleta (para simular uma medição), perceberam que um número menor de fótons foi emi-
tido, indicando que a ”observação” fez com que os ı́ons demorassem mais para se excitarem (e
decairem) do que em um sistema livre de observação.
Entretanto, ainda hoje se é discutido no meio cientı́fico a respeito da existência do efeito Zeno.
Alguns trabalhos chegaram a propor até a possibilidade de um efeito anti-Zeno [?, 4, 247], onde
a observação aceleraria a evolução do sistema. Atualmente o efeito Zeno ainda é estudado,
tanto de um ponto de vista básico quanto para possı́veis aplicações na metrologia, computação
Sistema de dois estados quântico
e informação quântica [147].
Um dos sistemas mais simples para ilustrar esse fenômeno é o sistema de dois estados oscilando
pela frequência de Rabi. Podendo ser ilustrado pictoricamente por um átomo sendo incidido por um laser
13.3.1.1 Philosophical implications
que possui frequências de ressonância com estados de transição desse mesmo átomo.
Quando proposto pela
Nesse primeira
caso temos vez, o efeito
a hamiltoniana foi considerado um paradoxo: como poderia uma
da interação:
partı́cula instável nunca decair, apenas por ser observada continuamente? Uma vez que o ques-
tionamento remete ao famoso paradoxo da flecha que voa, proposto por Zenão, o efeito recebeu
seu nome.
O paradoxo clássico consiste em uma flecha que voa. Em qualquer instante de tempo ela
ocupa apenas um Ondeespaço
os estadosigual
+ e – são
aodescritos pelas matrizesOu
seu tamanho. de Pauli
seja,e σ1,2,3
em são auto estados.individuais ela está
momentos
parada, em um espaço que não se move. Se ela está parada em cada e em todos os instantes do
seu voo, conclui-se que a flecha não se move. Por um argumento similar Zenão concluiu que não
existe movimento. Atualmente, sabemos que esse paradoxo é falso pois o tempo e o movimento
não são discretos. Mas, essa questão só pode ser resolvida após o desenvolvimento do cálculo,
que permitiu a definição da velocidade instantânea [182].
Quando proposto, o efeito quântico também causou estranheza. Os autores propuseram em
seu artigo original o que aconteceria com o gato de Schrödinger, caso o efeito fosse provado. O
gato está numa caixa com um veneno, que está associado a um átomo instável. Se o átomo
decair o veneno será aberto e o gato morrerá. Como há uma superposição entre o estado que
decaiu e o que não decaiu, o gato está numa superposição de vivo e morto. Porém, o efeito Zeno
nos mostra que se medirmos o estado do átomo de maneira contı́nua, o átomo nunca irá decair.
Poderı́amos assim salvar o gato do seu destino cruel?
Essa pergunta só pode ser respondida alguns anos mais tarde, quando o efeito foi verificado
experimentalmente.
Um questão interessante é a respeito da ”quanticidade” do efeito! Em qual aspecto ele é não
clássico? O efeito de Zeno quântico supõe a redução completa do estado para um autoestado.
No entanto, podemos imaginar medições clássicas que também reduzem o estado (como p.ex. a
medida da polarização de um feixe de luz).
13.3.2 Quantum non-demolition measurement
13.4 Welcher Weg information

13.4.1 The Elitzur and Vaidman bomb testing problem
Misturando os conceitos de partı́culas e ondas chegamos as vezes em conclusões aparamente
paradoxas. Um exemplo, é o problema de testar bombas de Elitzur e Vaidman. Eles imaginaram
um interferômetro de Mach-Zehnder com a particularidade, que o espelho refletor em um dos
braços seja conectado à um dispositivo medindo o recuo fotônico. Isto é, quando um fóton passa
por este braço, o espelho sofre uma pequena aceleração, que seja suficiente para ativar uma
bomba explosiva.
Agora, distinguimos dois casos: 1. O detetor de recuo não funciona, isto é, a bomba não
é armada. 2. A bomba é armada. Para o caso da bomba falsa, ajustamos o interferômetro
de maneira a produzir interferência destrutiva em uma das saı́das do interferômetro. Isto é,
depois de ter enviado muitos fótons através do interferômetro, nunca observando nenhum fóton
no saı́da escura, podemos ter quase certeza de que a bomba não é operacional.
No caso em que a bomba é operacional, a observação de um recuo fotônico destrói o padrão
de interferência nas saı́das do interferômetro, pois a explosão da bomba nos informa em qual
braço o fóton passou. No entanto, neste caso o padrão de interferência também é destruı́do,
quando o fóton passa pelo outro braço. Então, no caso (2) fótons podem ser detectados nas
duas saı́das com igual probabilidade. Em particular, a detecção de um fóton na ”saı́da escura”
do caso (1).
Portanto, é possı́vel acontecer que um fóton atravessa o interferômetro pelo braço que não
contem a bomba e sai pela ”saı́da escura”. A probabilidade para isso acontecer é só 25%, mas
não obstante a observação de um fóton na ”saı́da escura” nos informe que a bomba é operacional
sem ter interagido com ela 9 ..
13.5 Noisy measures

13.5.1 Quantum projection noise
O indeterminismo intrı́nseco à mecânica quântica ocasiona consequências graves na metrologia.
Para mostrar isso, consideramos o exemplo de um sistema de dois nı́veis |1i e |2i. Este sistema
pode ser num estado de superposição |ψi. A probabilidade 10 de encontrar o sistema em um dos
dois estados |ii é pi = hP̂i i = |hψ|ii|2 , onde P̂i é o operador de projeção. O resultado de uma
medida da população é aflita por uma incerteza inerente expressa pela variância (∆pi )2 = pi (1 −
pi ). Em outras palavras, a projeção aleatória do sistema sobre a base dos auto-estados induz um
ruı́do chamado ruı́do quântico de projeção 11 Este ruı́do inibe determinação das probabilidades
9
Vide https://www.thorlabs.com/newgrouppage9.cfm?objectgroup id=6635
10
Adotamos aqui o ponto de vista da interpretação de Copenhague da redução do estado quântico, mas notamos
que a interpretação como mistura estatı́stica baseada na matriz de densidade dá os mesmos resultados.
11
O ruı́do de projeção pode ser interpretado como shot noise. Enquanto, o shot noise óptico em fotodetectores
é gerado pela repartição discreta da energia do campo em fótons, o ruı́do de projeção é a consequência da
discretização dos nı́veis de excitação eletrônica.
13.6. TOPOLOGICAL PHASES 371
pi numa única medida. No entanto, medindo as populações com uma amostra de n átomos ou
repetindo a medida n vezes com um único átomo em condições idênticas, podemos reduzir a
incerteza. A probabilidade de encontrar um átomo r vezes no estado |2i é,

n r
Pn,r,2 = p (1 − p2 )n−r . (13.56)
r 2
O valor esperado e a variância desta distribuição binomial são [136],

n
X n
X
2
r̄ = rPn,r,2 = np2 , (∆r) = (r − np2 )2 Pn,r,2 = np2 (1 − p2 ) . (13.57)
r=0 r=0
√
Note que o desvio padrão diminuı́ com o número de átomos, σ = ∆r/r ∼ 1/ n.
Sob a influencia de um campo de radiação, a população do sistema de dois nı́veis (su-
Ω2 2 Gt
posto livre de emissão espontânea) executa oscilações de Rabi, ρ22 (t) = G 2 sin 2 , onde G =
√
2 2
∆ + Ω . A probabilidade de encontrar o sistema no estado |2i, portanto, varia no tempo,
p2 (t) = ρ22 (t), e a distribuição binomial (13.56) fica,

n Ω 2n 2r 1 2n−2r 1

Pn,r,2 (t) = G sin 2 Gt cos 2 Gt . (13.58)
r
Aumentando o número de pmedidas, n → ∞, essa função condensa em torno de um pico estreito

−1/2n
na posição Gt = 2 arcsin r/n. A largura do pico evolui como 2 arccos(2 ).
Resumindo, mesmo supondo uma sensitividade perfeita para medida de populações (e.g.,
usando o método de dupla ressonância microonda-óptica) é impossı́vel medir a probabilidade p2
num único átomo em um único experimento. Como uma tal observação só admite dois resultados
possı́veis, ”fluorescência observada” ou ”não observada”, isto é, ρ22 = 1 ou ρ22 = 0, uma gama
inteira de populações possı́veis entre 0 e 1 é excluı́da. Portanto, uma única observação só fornece
informação parcial, que pode ser amelhorada gradualmente com cada observação consecutiva.
13.5.1.1 Ramsey experiments

O experimento de Ramsey é basicamente equivalente ao experimento de Rabi descrito acima
exceto por uma rotação adicional no espaço de configuração permitindo a medida da fase de
precessão via espectroscopia de população. Os franges de Ramsey são aproximadamente dados
por p2 = 0.5(1 + cos[(ω − ω0 )T ]). A grandeza interessante é a incerteza de frequência,
p
∆r np2 (1 − p2 ) 1
= = √ . (13.59)
(∂r/∂ω)|(ω−ω0 )T =π/2 n(∂p2 /∂ω) T n
13.6 Topological phases

Consideramos um hamiltoniano Ĥ(r(t)), que somente depende implicitamente do tempo, isto é,
via algum parâmetro dependente do tempo RI (t). Então o hamiltoniano evolui desenvolvendo
uma fase dinâmica não mensurável e adicionalmente acumula uma fasegeométrica (também
chamada de fase topológica ou Berry fase de). Está fase geométrica, que depende da trajetória,
é adiabaticamente seguido no espaço de parâmetros,
H|ψ(t)i = i~∂t |ψ(t)i . (13.60)

Assumimos que em qualquer instante de tempo o sistema fica num autoestado |n(r)i,
H|n(r)i = En (r)|n(r)i . (13.61)
Quando Ĥ segue a trajetória C : t → r(t), então vimos de

h R i
t
|ψ(t)i = exp − ~i 0 En (r(t0 ))dt0 exp(iγn (C))|n(r(t))i (13.62)
que a fase γn : C → γn (C) não é integrável. Berry mostra que o sistema acumulou, depois de
um ciclo completo no espaço dos parâmetros r(T ) = r(0) e |n(r(0))i = |ψ(0)i a fase,
ZZ
γn (C) = Vn (r)dS , (13.63)
C
onde
X hn(r)|∇r H(r)|m(r)i × hm(r)|∇r H(r)|n(r)i
Vn (r) = Im . (13.64)
(Em (r) − En (r))2
m6=n
Consideramos o exemplo de um sistema de dois nı́veis sem decaimento descrito pelo vetor
de Bloch,   
Re Ω σx
1
H = Rσ ≡ Im Ω σy  . (13.65)
2
∆/2 σz
Os parâmetros r mudam adiabaticamente. É fácil ver que,
E± = ±R/2 e ∇r H = σ/2 (13.66)

hn± (r)|∇r H(r)|n∓ (r)i × hn∓ (r)|∇r H(r)|n± (r)i r
Vn (r) = Im 2
=± 3 .
(E+ (r) − E− (r)) 2R
A fase geométrica resultante,

ZZ
dS Ω(C)
γn (C) = ∓ 2
=∓ (13.67)
C 2R 2
depende do ângulo sólido encerrado.
Example 52 (Fase topológica no sistema de dois nı́veis): Consideramos o seguinte

estado [214],
n± (r) = cos θ|gi ± e±iφ sin θ|ei .
Agora queremos calcular a fase topológica,
I
γ± = ihn± (r)|∇r |n± (r)idr .
C
Aplicando o gradiente em coordenadas esféricas,

1 ∂ 1 ∂ ∂
∇r = êθ + êφ + êr
r ∂θ r sin θ ∂φ ∂r
na função |n+ i, achamos,
sin θ e±iφ cos θ ieiφ

∇r |n+ (r)i = −êθ |gi + êθ |ei + êφ |ei ,
r r r
e
i sin θ
ihn± (r)|∇r |n± (r)i = iêφ .
r
Finalmente,
I I I
− sin θ − sin θ
γ+ = êφ dR = r sin θdφ = sin2 θφ̇dt .
C r C r C
A condição de adiabaticidade é essencial para emergência de fases topológicas. O sistema

sempre permanece num autoestado (números quânticos fixos) quando variamos parâmetros do
ambiente mais lentamente do que todas constantes caracterı́sticas do sistema, mesmo quando o
hamiltoniano é dependente do tempo (autovalores variáveis).
13.6.0.1 Generalization of the Berry following Aharonov

Aqui não impomos condições ao hamiltoniano a respeito do seu comportamento adiabático, e o
estado são precisa ser um aautoestado:
Ĥ|ψ(t)i = i~|ψ̇(t)i . (13.68)
Um processo é cı́clico, quando existe um τ , tal que,
|ψ(τ i = ei[f (τ )−f (0)] |ψ(0)i . (13.69)
Definindo o espaço dos raios por |ψ̃(t)i = e−if (t) |ψ(t)i, obtemos,
|ψ̃(τ )i = |ψ̃(0)i , (13.70)
e a partir da equação de Schrödinger obtemos,

Z t Z t
f (t) − f (0) = − ~1 hψ(t)|Ĥ(t)|ψ(t)idt + d
hψ̃(t)|i dt |ψ̃(t)i ≡ δ + β . (13.71)
0 0
Portanto, no espaço dos raios temos uma curva fechada:
C : [0, τ ] −→ ψ(t) ∈ H (13.72)

↓ e−if (t)
C 0 : [0, τ ] −→ ψ̃(t) ∈ P .
A fase dinâmica δ pode ser zerada por uma escolha apropriada de Ĥ(t), mas não a fase topológica
β. β não depende de Ĥ(t), mas é uma propriedade geométrica da curva projetando H sobre P.
Em contraste com eiβ , a fase β é somente determinada modulo 2πn.
13.6.0.2 Berry phase for two-level systems

Seguinte [2] consideramos o hamiltoniano,

∆/2 Ω
Ĥ = ~ . (13.73)
Ω/2 −∆/2
A solução da equação de Schrödinger é |ψ(t)ie−iĤt/~ |ψ(0)i. A matriz dos autovalores é,

p
~ 2 2 1 0
Ê = Gσ̂z com G≡ ∆ +Ω e σ̂z = . (13.74)
2 0 −1
O deslocamento total depois de um ciclo de precessão é,
Ĥ
φ= ~t = 21 Gt = π . (13.75)
A fase dinâmica é
Z 2π/G Z 2π/G
δ = ~1 hψ(t)|Ĥ|ψ(t)idt = 1
~ heiGt/2 ψ(0)| ~2 G|e−iGt/2 ψ(0)idt . (13.76)
0 0

cos θ
Partindo do estado inicial ψ = ,
i sin θ
Z 2π/G
1~
δ= ~2 (cos2 θ − sin2 θ) > dt = −π cos θ . (13.77)
0
A fase geométrica corresponde ao ângulo sólido encerrado,
β = φ − δ = π(1 − cos θ) = 12 Ω(C) . (13.78)
Apesar do átomo não estar num autoestado e de não ter variação adiabática dos parâmetros.
13.6.1 Aharonov-Bohm effect

Um caso particular para fase topológica é o efeito de Aharonov-Bohm. Se trata do seguinte
fenômeno: Na teoria eletrodinâmica, os efeitos do electromagnetismo sempre se exprimem pelas
forças de Coulomb e de Lorentz, que podem ser descritas por campos elétricos e magnéticos.
Os potenciais eletromagnéticos podem ser introduzidos para simplificar os cálculos, mas eles
são observáveis com realidade fı́sica. Em contraste, em mecânica quântica os potenciais eletro-
magnéticos são mais fundamentais do que os campos eletromagnéticos. Isso é demonstrado no
efeito de Aharonov-Bohm.
Figure 13.9: (a) Esquema para medir o efeito de Aharonov-Bohm. (b) Efeito de Aharonov-
Casher: os eletrodos φ não produzem campos elétricos dentro dos condutores; mesmo assim,
observa-se na saı́da do interferômetro interferência construtiva ou destrutiva dependendo do
potencial aplicado.
A ideia deste efeito é esquematizado na Fig. 13.9: Um feixe de elétrons é coerentemente

divido em dois braços (e.g. por uma dupla fenda) passando pelos dois lados de um solenoide
infinitamente extenso é perfeitamente blindado. Desta maneira o campo magnético B zera na
região exterior ao solenoide, mas existe não-obstante um potencial vetor A. Observamos na
tela um padrão de interferência dos dois braços do interferômetro eletrônico. Quando ligamos o
solenoide observamos um deslocamento do padrão de interferência.
13.6.1.1 Aharonov-Bohm effect and gauge transformation

Sejam R e S duas regiões espaciais desconectadas. Suponha que os campos elétrico e magnético
são mantidos nulos na região R. Desse modo, classicamente seria impossı́vel medirmos alguma
alteração na dinâmica de um corpo confinado à região R decorrente da alteração do campo
magnético confinado à região S. O efeito de Aharonov-Bohm mostra o contrário, isto é, elétrons
na região R sem campo magnético sentem fluxos de campo magnético numa região S, sendo que
R e S não tem interseção!
Na teoria clássica do electromagnetismo, em uma região espacial vazia (a menos de fontes de
carga elétrica e de corrente elétrica) os campos elétrico E(r, t) e magnético B(r, t) se relacionam
com as densidades de carga elétrica ρ(r, t) e de corrente elétrica j(r, t) segundo as equações
de Maxwell. Para uma situação fı́sica numa região espacial, conhecidas as fontes ρ e j e as
condições de contorno que os campos E e B devem satisfazer, é possı́vel determinar os campos
como solução desse sistema de equações diferenciais parciais.
Na eletrodinâmica clássica, observado de um referencial inercial, força eletromagnética Fem
sobre um corpo pontual de carga q, que no instante t esteja na posição r com velocidade v, é
dada pela lei de força de Lorentz:
Fem (r(t), t) = qE(r(t), t) + qv(t) × B(r(t), t) . (13.79)
A teoria eletrodinâmica afirma a existência de duas funções Φ(r, t) e A(r, t), tais que:
∂A(r, t)
B(r, t) = ∇ × A(r, t) e E(r, t) = −∇Φ(r, t) − . (13.80)
∂t
Dessa forma, pode-se usar as equações (13.80) para reescrever as equações de Maxwell.
Os potenciais Φ e A não são unicamente obtidos, mas quaisquer Φ e A que levam aos mesmos
campos E e B, e assim à mesma fı́sica, são equivalentes. Iremos determinar Φ e A adotando
uma condição adicional que deverá ser obedecida. Isso significa adotar um calibre. Adotaremos
o calibre de Lorentz:
1 ∂Φ(r, t)
∇ · A(r, t) + 2 =0, (13.81)
c ∂t
onde c é a velocidade de propagação da luz no vácuo.
13.6.1.2 Equation for quantum particle exposes to a vector potential A

Suponha uma partı́cula (sem spin), de massa m e carga q, cuja função de onda é confinada
a uma região R (conexa por caminhos). Exigimos Φ = 0 e E = 0 = B, mas A 6= 0, isto é,
∇ × A(r, t) = 0. Note que juntamente com (13.80) isso obriga que A seja estacionário. De
acordo com a mecânica quântica a função de onda Ψ da partı́cula deve obedecer a seguinte
equação de Schrödinger:
2
1 ~ ∂Ψ(r, t)
∇ − qA(r) Ψ(r, t) + V (r)Ψ(r, t) = i~ . (13.82)
2m i ∂t
Em (13.82) o potencial vetor A está presente, mesmo que φ, E e B sejam mantidos nulos em
toda a região R.
Como o rotacional de A zera em R, já que a integral pode ser calculada por qualquer
caminho contido em R que seja deformável a um ponto O ∈ R (escolhido arbitrariamente),
podemos definir o seguinte campo escalar:
Z
q r
g(r) ≡ A(x) · dx . (13.83)
~ 0
De (13.83) temos:
q
∇g(r) =A(r) . (13.84)
~
Agora, já mostramos na Sec. 2.5.3, que a função de onda
ψ(r, t) ≡ e−ig(r) Ψ(r, t) (13.85)
corresponde à transformação de calibre (2.192) e, dado a condição (2.191), satisfaz a mesma

equação de Schrödinger como Ψ(r, t). Mostramos isso explicitamente no Exc. 13.7.6.1. Assim, a
presencia de um potencial vetor na região R, mesmo na ausência de campos (E, B e Φ são nulos,
e portanto A é estacionário) se traduz por um deslocamento de fase eig(r) da função de onda.
Uma questão interessante estudada no Exc. 13.7.6.2 é, se isso implica a perda da liberdade da
escolha do campo de calibre.
Example 53 (Observation of the Aharonov-Bohm effect): Suponha a situação onde

dentro de uma caixa cúbica realizamos, no vácuo, o experimento da dupla fenda, usando
elétrons. Temos uma fonte que incide um fluxo de elétrons sobre um primeiro anteparo, com
duas fendas. A função de onda eletrônica difrata por ambas as fendas, propaga interferindo
onde se superpõe, e então incide sobre um segundo anteparo, que mede (marca) onde o elétron
incidiu. Entretanto, logo após o primeiro anteparo, na posição média das duas fendas, temos
uma região cilı́ndrica inacessı́vel aos elétrons, dentro da qual existe um solenoide ideal de raio
a que mantém um fluxo de campo magnético constante. Pelos equipamentos do laboratório
esse fluxo pode ser ajustado, como um parâmetro, mas entre os ajustes ele é constante. O
campo magnético (e também elétrico) do solenoide está confinado numa região S, dentro da
qual a função de onda eletrônica é asseguradamente nula. Esse confinamento dos campos
é seguro, feito com camadas de materiais, inclusive supercondutores. Na região R em que
a função de onda do elétron pode ser não nula, os campos são mantidos nulos. R e S tem
inserção vazia, tanto R quanto S é conexa por caminhos.
Mostraremos que o fluxo do campo magnético na região S pode ser medido pela dinâmica
eletrônica na região R, ainda que que o elétron nunca esteja na região S, e esteja confinado
a uma região R livre de campos E e B. Esse é o efeito Aharonov-Bohm (magnético).
O campo B na região interna do solenoide é dado por (I é a corrente elétrica no fio, N é a
densidade de voltas do fio),
B(r, t) = µ0 INêz .
Fora do solenoide, isto é, para ρ > a, temos,
ΦB
A(r, t) = êφ ,
2πρ
onde ΦB = πa2 B(0, t) é o fluxo do vetor magnético B pela seção transversal do solenoide.
Num ponto rsim do anteparo, localizado no plano de simetria do sistema, calculemos g(rsim )
por dois caminhos diferentes: ambos começam na fonte e terminam no anteparo final, mas
um atravessa pela fenda da esquerda e outro pela fenda da direita:
Z Z
q rsim qΦB 1 qΦB
g(rsim ) = A(x) · dx = φ̂ · (ρφ̂dφ) = ± .
~ 0 2π~ ρ 2~
O sinal + significa que a integração foi feita no sentido de A, e assim no sentido de I no

solenoide. O sinal - é o contrário. A diferença de fase, no ponto rsim , entre esses dois
caminhos será:
qΦB
δ= .
~
Isto é, a diferença de fase (observável no experimento, por exemplo, pelo padrão de inter-
ferência) é diretamente proporcional ao fluxo de campo magnético B, mesmo que a função
13.7. EXERCISES 377
de onda seja nula na região S, dentro da qual o campo B está confinado.

Aproveitando a descrição que fizemos, imagina agora uma outra situação. A fonte de elétrons
está desligada. Confinamos uma função de onda de um elétron a um trilho unidimensional
fechado, uma circunferência de raio b que pode ser desenhada seguindo-se uma linha de
campo de êφ numa região de R, e não de S. Pode-se mostrar que o fluxo ΦB quebra a
degenerescência de nı́veis de energia desse elétron:
2
~2 qΦB
En = n− ,
2mb 2π~
com inteiro, isto é, n = 0, ±1, ±2, ... [103].
13.6.1.3 Generalizations of the Aharonov-Bohm effect

O efeito de Aharonov-Bohm pode ser generalizada para os graus de de liberdade internos de um
único átomo, isto é, do espaço real para o espaço de configuração. Imaginamos um interferômetro
de Mach-Zehnder, onde um dos braços atravessaR uma região de campo constante e homogêneo.
A força de Lorentz correspondente FR = d3 r ρ(r)E(r) + j(r) × B(r) zera, mas a onda de Broglie
sofre um deslocamento de fase χ = Hint dt:
potenciais
R escalares
χ= R − eφdt ∇φ = 0 for e−
− R d · Edt ∇ × E = ∇ · E = 0 Mg, Yb+
− µ ~ · Bdt ∇ × B = ∇ · B = 0 n, Yb+
potenciais
H vetoriais
− H eAdr ∇×A=0 e− , (ABE)
− H d × Bdr ?
− µ × Edr n, Ca, (ACE)
13.6.1.4 Topological phase in configuration space

Fazemos um experimento de Ramsey temporal com um único ı́on aprisionado excitando uma
transição hiperfina. Entre os pulsos aplicamos um campo magnético por um tempo t. A ase
acumulada será φ = (~ µ · B/~)t. Essa fase corresponde à precessão do momento dipolar excitado
pelo primeiro pulso de Ramsey. A fase pode ser interpretada pelo efeito de Aharonov-Bohm
considerando 1. o campo magnético é homogêneo e 2. mesmo assim age sobre o spin, não por
uma força mas por um deslocamento de fase.
13.7 Exercises
13.7.1 The observer and the reality
13.7.2 Open systems and the master equation
13.7.2.1 Ex: Equação mestre para cavidades
Seja κ a taxa de perda de fótons por transmissão através dos espelhos da cavidade. Mostre que
a equação mestre dá
dρ̂ κ κ
= (n̄ + 1) 2âρ̂â† − â† âρ̂ − ρ̂â† â − n̄ 2â† ρ̂â − ââ† ρ̂ − ρ̂ââ† ,
dt 2 2
P
onde o operador de densidade é definido por ρ ≡ m,n Pm,n |mihn|. e n̄ é o número de fótons
térmicos.
13.7.3 Repeated Measurements

13.7.3.1 Ex: Efeito de Zeno
Discute o efeito de Zeno quântico no exemplo de um feixe laser passando por um meio bire-
fringente. Compare as situações sem polarizadores e com um número infinito de polarizadores
verticais.
13.7.3.2 Ex: Efeito de Zeno

Um átomo de dois nı́veis ressonantemente excitado por um laser pode ser descrito pelo hamil-
toniano:
0 12 Ω
H= 1 .
2Ω 0
A solução da equação de Schrödinger dá

−itĤ/~ cos 12 Ωt i sin 12 Ωt 1
|ψ(t)i = e |ψ0 i =
i sin 12 Ωt cos 12 Ωt 0

se o átomo inicialmente está no estado fundamental hψ0 | = 1 0 . A medição da população do
estado fundamental só pode ser feita através uma projeção da função de onda, isto é, o resultado
da medição é descrita por k|1ih1|ψ(t)ik2 . Qual é o estado final do átomo quando a população
do estado fundamental
a. é medida uma vez após um tempo de evolução t = π/Ω;
b. é medida uma vez após n intervalos de tempo tn = π/nΩ;
c. quando é medida n vezes após tempos de evolução tn = π/nΩ;
d. quando n → ∞.
13.7.3.3 Ex: Efeito Zeno Quântico por Tiago S. do Espirito Santo

No efeito Zeno quântico temos que a evolução temporal de um sistema é inibida por causa de
sucessivas medidas, do estado atual, em um curto intervalo de tempo. Considere um sistema
descrito por um hamiltoniano independente do tempo Ĥ e que a evolução temporal do sistema
seja dado pelo operador unitário Û = e−iĤt/~ .
a. Calcule a probabilidade P (t) do sistema permanecer em um estado inicial |Ψ0 i com a aprox-
imação de tempo curto, isto é, considere até o termo de segunda ordem da expansão da expressão
para probabilidade. Utilize a simplificação:
~
τz = q ,
hĤ 2 i − hĤi2
onde o termo τz é chamado de tempo Zeno.

b. Se forem realizadas N medidas no tempo t, temos T = t/N o intervalo entre as medidas.
Quando uma medida é realizada, o sistema é projetado no estado inicial e recomeça a evolução
temporal. Assim, após N medidas, a probabilidade do sistema permanecer no estado inicial
é dada por [P (T )]N . Mostre que para um número infinito de medidas, N → ∞, o sistema
permanece no estado inicial sem perda de probabilidade: [P (T )]N = 1. Interprete o resultado.
c. Um dos sistemas mais simples que se pode imaginar é um sistema de dois nı́veis com oscilação
Rabi. Tem-se o hamiltoniano:
0 Ω
Ĥ = .
Ω 0
13.7. EXERCISES 379
Encontre a expressão para τz em função de Ω para o estado inicial (1 0)† .

d. Se escolhermos um tempo de evolução t = 0.01τz τz e realizarmos N = 5 medidas nesse
intervalo, qual a probabilidade do sistema ter permanecido no estado inicial?
e. Considerando, agora, um canal de decaimento para o estado (0 1)† com Γ = 4γ de forma
que simulamos um sistema com medição contı́nua. Isto é, o sistema é inicialmente preparado no
estado (1 0)† e, se observamos emissão pelo decaimento, significa que o sistema saiu o estado
inicial. Temos agora o hamiltoniano:

0 Ω
Ĥ = .
Ω −2iγ
Para esse sistema, a amplitude de probabilidade para o estado inicial vale:

1 γ −(γ−∆)t/~ 1 γ −(γ+∆)t/~
hΨ0 |Ψ(t)i = 1+ e + 1− e
2 ∆ 2 ∆
p
com ∆ = γ 2 − Ω2 . Para uma taxa de decaimento γ Ω, encontre a probabilidade do sistema
permanecer no estado inicial. Interprete o resultado.
Formulário:
x2
ex = 1 + x + + O(x3 ) , (1 − x)N = 1 − N x + O(x2 )
2
2 2 3 0 Ω −iĤt/~ cos Ωt
~ −i sin Ωt
~
cos (x) = 1 − x + O(x ) , Ĥ = →e = .
Ω 0 −i sin Ωt
~ cos Ωt
~
13.7.4 Welcher Weg information

13.7.5 Noisy measurements
13.7.6 Topological phases
13.7.6.1 Ex: Efeito de Aharonov-Bohm como transformação de calibre
Mostre explicitamente, que a função de onda transformada por uma transformação de calibre
(13.85) satisfaz a equação de Schrödinger.
13.7.6.2 Ex: O efeito de Aharonov-Bohm e a transformação de calibre

A fase do padrão de interferêcia no efeito de Aharonov-Bohm é fixa pelo fluxo magnético através
do solenoide. Isso significa que perdemos a liberdade de escolher um potencial de calibre ar-
bitrário?
Part IV
Collective Scattering of Light
381
Part V
Atom Optics
383
Chapter 14
Manipulation of atomic gases

The field of atom optics is about the movement of atoms and their control by technical tools.
At high velocities, with no external forces, the atoms follow straight paths, in the same way as
light beams in the classical optical. At low speeds, they propagate as waves, similarly as light
waves in the optics within the Maxwell theory. The term atom optics emphasizes analogy and
duality in the behavior of microscopic particles.
The duality principle is one of the fundamental ideas of quantum mechanics. The appearance
of an object as a wave or as a particle depends on the situation in which it is observed. The wave
nature of light is well known in classical physics. However, Louis de Broglie was the first in 1924
to apply the principle of duality also to massive particles and to predict that particles, under
certain conditions, behave like waves whose wavelengths increase as their velocity decreases.
Each particle (or set of particles) is delocalized along a distance corresponding to the de Broglie
wavelength. This feature of the material was soon discovered experimentally in electron beams
and is still used today in commercial devices, for example in electron microscopes.
The laser was discovered in 1956. In a laser, light particles are forced to oscillate syn-
chronously, that is, coherently. By analogy, we can raise the question whether a similar phe-
nomenon can occur with massive particles, and whether it is possible to construct an atom laser.
Such a device would emit coherent matter waves just as the laser emits a coherent light. When a
gas is cooled to very low temperatures, the Broglie waves of the atoms become very long and, if
the gas is sufficiently dense, eventually overlap. If the gas consists of a single species of bosonic
particles all atoms being in the same quantum state, its Broglie waves constructively interfere
thus forming a huge wave of coherent matter. This wave of matter is described by a single wave
function exhibiting long range order and having a single phase. If this wave function is formed
within a trap, all atoms accumulate in their ground state. Thus, we obtain a pure quantum
state of many bodies in the degree of kinetic freedom 1 . The transition of a gas from individual
atoms to a degenerate mesoscopic body-many quantum state occurs as a phase transition and is
named Bose-Einstein condensation (BEC) to the homage of Bose and Einstein that calculated
the effect already in 1924 [32, 80].
The experimental prediction check of Bose and Einstein was a long cherished dream of many
physicists. On the one hand, several phenomena have been related to BEC in the past, for
example, the phenomenon of superfluidity in liquid helium and superconductivity. On the other
hand, these strongly interacting systems are not pure enough to clearly identify the role of BEC.
In 1995, however, Bose-Einstein condensation of weakly interacting confined atomic gases was
achieved in several laboratories [6, 65, 36, 113]. This success gave rise to a glowing revolution
in atomic optics documented in an enormous amount of theoretical and experimental work.
While the initial work focused on the condensation equilibrium thermodynamics near the phase
1
In particular, for very cold atoms whose internal excitation occurs on a very different energy scale, the degree
freedom is frozen and does not influence kinetics.
385
386 CHAPTER 14. MANIPULATION OF ATOMIC GASES
transition, soon after the dynamic response of condensates to the perturbations was the subject
of in-depth investigations, followed by the study of superfluid characteristics, quantum transport
phenomena, the interaction of condensed with light, of condensed gas mixtures [190, 235] and
the behavior of condensates in periodic potentials. To name only a few landmarks, we mention
the creation of vortices [171, 167] and quantum turbulence [122], the realization of various
types of atom lasers [174, 5, 30, 110]and atom interferometers with condensates [112, 152],
the coherent amplification of waves of matter [134, 153, 135, 72], the creation of the Mott
state in optical lattices [99], the study of condensates in reduced dimensions [186], Anderson’s
location of waves of atomic matter [43, 215], the observation of resonances of Feshbach type
collisions [53, 133, 248] and Efimov states of [154, 20], the creation of homonuclear molecular
condensates [100, 143, 56, 264, 260, 123] and heteronuclear [198] and degenerate Fermi gases
[69], to the observation of the BCS pairing [128, 101], the matter wave superradiance [134] and
the interaction of condensates with optical cavities [231, 40] and with surfaces [26].
It is clearly unthinkable to discuss all matters in this course. Let us, however, give a basic
and practical introduction to atomic optics with condensates. The course begins in this chap-
ter with a presentation of the most important experimental techniques for cooling, confining,
manipulating and detecting atomic gases. Knowledge of these techniques will allow a better
understanding of how it is possible to generate and analyze all the effects mentioned above. The
chapter ?? introduces the Bose-Einstein condensation phenomenon and the following chapters
focus on the thermodynamic, superfluid, coherent and dielectric properties of condensates.
The comparable success of atomic optics can be documented by Nobel laureates who have
been assigned to this area of physics:
• Dehmelt, Paul, Ramsey (1989)
• Philips, Chu, Cohen-Tannoudji (1995)
• Cornell, Wieman, Ketterle (2000)
• Hänsch, Glauber, Hall (2005)
• de Gennes (1991)
• Leggett (2003)
• Wineland, Haroche (2014)
• Ashkin (2018)
For review articles on BEC see [200], [58], [50], [93], [217], [234], [146] and check in internet
sites http://amo.phy.gasou.edu/bec.html and http://jila.edu/bec.html.
In this chapter we review the basic techniques of Atomic Optics, emphasizing the cooling,
entrapment and measurement of cold atomic gases.
14.1 The atomic motion

14.1.1 The atom as a matter wave
We have already emphasized that atomic optics deals with the movement of atoms within a gas,
that is, we are interested only in the degree of outer freedom of atoms. To describe the motion
14.1. THE ATOMIC MOTION 387
Figure 14.1: De Broglie wave at different temperatures.
of a massive particle, we write the Hamiltonian of a free particle for a one-dimensional motion,
~2 d2
Ĥ = − . (14.1)
2m dx2
Therefore, the general solution of the Schrödinger stationary equation,
Ĥψ(x) = Eψ(x) , (14.2)
is r
ikx −ikx 2mE
ψ(x) = Ae + Be com k= . (14.3)
~2
Note well, that the functions eikx are not quadratically integrable. On the other side, they do
not represent actual physical systems. In practice, we need to consider wave packets or specify
a finite volume for the particle.
Note also that the spectrum of eigenvalues is continuous. To R guarantee the interpretation
2 3
as density of probability we will require quadratic integrability, |ψ| d r = 1. This means that
the wave function can not be infinite in a finite volume. But it can be infinite in an infinitely
small volume.
The description of the atomic motion by a wave equation emphasizes the fact that microscopic
particles have wave properties. Each atom corresponds to an de Broglie wave
h
λdB = , (14.4)
p
which describes the coherence length of the atom.
14.1.1.1 Characteristic velocities

The behavior of an atom described by the Schrödinger equation depends very much on its kinetic
energy. At high speeds (or short waves of Broglie), it will behave with a classical particle with
a well defined trajectory. At low speeds (or long waves of Broglie), its propagation will be that
of a wave and will show phenomena of diffraction and interference. Therefore, it is important
to highlight some characteristic speed regimes.
In optical cooling techniques we remove kinetic energy by light scattering in electronic tran-
sitions. Therefore, we can compare the kinetic energy (or temperature) of a set of atoms with
the width of the Γ transition. The Doppler limit is given by (see Exc. 14.7.1.1),
~
kB TD = Γ. (14.5)
2
We can also compare the kinetic energy with the energy transferred to an atom by the absorption
of a single photon. The photonic recoil is given by,
~2 k 2
kB Trec = . (14.6)
2m
Atomic clouds with temperatures around TD = 100..1000 µK they call it cold. Clouds with
temperatures around Trec = 100..1000 nK they are called ultracold.
In most experiments of atomic optics we do not work with individual atoms (or ions), but
with a set of atoms, called clouds. In general, sets can not be described by a single wave function.
Or we describe each atom by a separate wave function (which only works when atoms do not
interact), or we describe the cloud by probability distributions (such as a density matrix). We
now consider a thermal cloud. The Maxwell-Boltzmann distribution of velocities is,
r 3
m 2
g(v) = e−mv /2kB T . (14.7)
2πkB T
R R∞
This distribution is normalized, g(v)d3 v = 0 4πv 2 g(v)dv = 1. Average velocity is now
Z r
kB T
v̄ = vg(v)dv = . (14.8)
m
0.5
N
0
-5 0 5
v
Figure 14.2: (Code: AO T echniques M axwellBoltzmann.m) Maxwell-Boltzmann distribution.
We define the thermal de Broglie wavelength of an atomic ensemble as,

s
h 2π~2
λtherm ≡ = . (14.9)
mv̄ mkB T
This represents the average Broglie wave of the atoms in the set. When in a dense gas this
quantity exceeds the average distance between atoms,
ρ ≡ nλ3therm > 1 , (14.10)
where n is the density of atoms, we enter into a new scheme, where Maxwell-Boltzmann law is to
be valid. As λtherm ∝ T −1/2, this regime is at low temperatures. The amount ρ is called phase
14.1. THE ATOMIC MOTION 389
space density. When the density in the phase space approaches 1, it means that it increases the
probability of finding more than one atom per elementary cell of the phase space. Here we enter
the regime of quantum degeneration, where the Boltzmann statistic must be replaced by the
Bose-Einstein statistic in the case of bosons or Fermi-Dirac in the case of fermions. This will be
dealt with in the next chapter. Of condition nλ3therm ' 1 we obtain
2
1 2π~ (2π~)2 n2/3
kB Tc = = . (14.11)
m λtherm m
14.1.2 Localized atoms

To avoid the influence of the environment on the atoms, we often trap them in a potential which
suspends them in a volume away from massive walls. The Hamiltonian of trapped atoms is,
~2 d2
Ĥ = − + U (x) . (14.12)
2m dx2
As the wavefunction is now localized, the spectrum of possible energies organizes itself into
discrete levels. The atoms are allocated in populations of these levels.
Often, a 3D potential can be written in the form,
U (x, y, z) = Ux (x) + Uy (y) + Uz (z) . (14.13)
This is the case, for example, for a rectangular well characterized by Ux (x) = Uy (y) = Uz (z) =
U0 /3 inside the well. The relationship (14.13) also holds for a harmonic potential,
m 2 2
U (r) = ωx x + ωy2 y 2 + ωz2 z 2 . (14.14)
2
In these cases, it is generally useful to do the following ansatz for the wavefunction,
ψ(r) = ψx (x)ψy (y)ψz (z) , (14.15)
since insertion into the Schrödinger equation,

2
~2 d d2 d2
− + + + Ux (x) + Uy (y) + Uz (z) ψx (x)ψy (y)ψz (z) = Eψx (x)ψy (y)ψz (z) ,
2m dx2 dy 2 dz 2
(14.16)
separates it into three independent one-dimensional equations,
~2 ψx00 (x)
− + Ux (x) = const. ≡ Ex , (14.17)
2m ψx (x)
and similarly for y and z. Since E = Ex + Ey + Ez , we can have the same energy for different
combinations of Ex , Ey and Ez . That is, multidimensional systems are often degenerate.
14.1.3 Density-of-states of a trapping potential

The way an atomic cloud accommodates itself within a trapping potential is governed by the
density of available states. To calculate this density, we consider the Hamiltonian H(r, p) =
~2 k2
2m + U (r). For a cylindrical harmonic oscillator we write,
m 2 2 m 2 2
U (r) = ω r + ωz z where r 2 = x2 + y 2 , (14.18)
2 r 2
Figure 14.3: (Code: AO T echniques Iof f eP otential.m) (a) The figure shows two dimensions of
a Ioffe-Pritchard type magnetic trapping potential (characterized by being approximately linear
at large distances from the center and harmonic near the center). (b) Harmonic approximation
(most experimentally feasible potentials are approximately harmonic near the center). (c) One-
dimensional cut through the potential of (a,b). (d) Density of states for a harmonic (dotted
line) and a Ioffe-Pritchard type potential (solid line).
or U (x) = m 2 2 2 2 2 2 2
2 ωr ρ , where ρ = x +y +λ z with λ = ωz /ωr . We define ω̄ = (ωr ωz )
2 1/3 = λ1/3 ω .
r
The individual particle levels of this Hamiltonian are, nx ny nz = ~ωx nx + ~ωy ny + ~ωz nz , where
nj are integer numbers.
We now introduce the density of states η() for nan arbitrary potential via,
Z Z Z Z p
1 3 3 (2m)3/2 3
η()d ≡ d rd k = d r d − U (r) . (14.19)
(2π) 3 (2π) ~
2 3
For the cylindrical harmonic trap defined in (14.18), we find with a little help from Dr. Bronstein
[38],
Z r
(2m)3/2 3 m
η() = d r − ωr2 ρ2 (14.20)
(2π) ~
2 3 2
Z 1 Z √1−x̃2 Z √1−x̃2 −ỹ2 p
1 82
= dx̃ √ dỹ √ dz̃ 1 − x̃2 − ỹ 2 − z̃ 2 .
(2π)2 (~ω̄)3 −1 − 1−x̃2 − 1−x̃2 −ỹ 2
The resolution of the integral gives,

2
η() = (harmonic potential) . (14.21)
2(~ω̄)3
Another example is the box potential. In this case we can simply obtain,
Z
(2m)3/2 3 √ (2m)3/2 √
η() = d r = V (box potential) . (14.22)
(2π)2 ~3 V (2π)2 ~3
14.2. OPTICAL COOLING 391
A generalization is discussed in the Exc. 14.7.1.2.
14.2 Optical cooling

The force of light exerted on an atom can be of two types: a spontaneous dissipative force and
a dipole conservative force. Spontaneous force arises from the impulse experienced by an atom
when it absorbs or emits a quantum of photonic impulse. Each absorbed photon transfers an
impulse quantum ~k to the atom in the direction of propagation. Spontaneous emission after
absorption occurs in random directions, and over many absorption-emission cycles, it cancels to
zero in the medium. Consequently, the total spontaneous force acts on the atom in the direction
of the propagation of light. The saturated rate of scattering of photons by spontaneous emission
(the reciprocal value of the excited state lifetime) sets the upper limit for the magnitude of the
force. This force is sometimes called radiation pressure force.
The dipolar gradient force can be easily understood by considering light as a classical wave.
It is simply the force in the temporal medium resulting from the interaction of the transition
dipole, induced by the electric field of light oscillation, with the electric field amplitude gradient.
Focusing of the light beam controls the magnitude of this gradient, and the tuning of the optical
frequency below or above the atomic transition controls the sign of the force acting on the atom.
Adjusting the light below the resonance attracts the atom to the center of the light beam,
adjusting the light above the resonance repels it. The dipole force is a stimulated process in
which there is no energy exchange between the field and the atom. Photons are absorbed in one
way and reappear by stimulated emission in another. Preservation of momentum requires that
the change of direction of propagation of photons from an initial mode to an end mode leaves
the atom with a retreat. Contrary to spontaneous force, there is, in principle, no upper limit
for the magnitude of the dipole force, since it is a function of the field gradient only and the
detonation.
We can treat these observations quantitatively by considering the amplitude, phase and
frequency of a classical field interacting with the dipole of an atomic transition in a two-level
atom. What follows now is sometimes called the Doppler cooling model. It turns out that
atoms with a hyperfine structure in the ground state can be cooled below the limit predicted by
this Doppler model. To explain this unexpected sub-Doppler cooling, models involving atoms
moving slowly in a polarization gradient of a standing wave have been invoked. Let’s briefly
outline the Sec. 14.2.1 the physics of these polarization gradient cooling mechanisms.
14.2.1 Optical molasses

À partir da Eq. (12.24) e das definições anteriores de Ω e Ωsat , e com a intensidade I ∝ E02 ,
podemos escrever o parâmetro de saturação,
I Ω2 Ω2
s= = 2 = 2 , (14.23)
Isat Ωsat Γ /2
e
~kΓ s
Frp = . (14.24)
2 (2∆/Γ)2 + 1 + s
Agora, se consideramos um átomo propagando na direção +z com a velocidade vz contrapropa-
gante á onda de luz dessintonizada da ressonância por ∆, a dessintonia total será
∆ −→ ∆ + kvz . (14.25)
onde o termo kvz é o deslocamento de Doppler. A força F− agindo sobre o átomo será na direção
contrária ao movimento. Em geral,
~kΓ s
F± = ± . (14.26)
2 (2(∆ ∓ kvz )/Γ)2 + 1 + s
Supõe que temos dois campos propagando nas direções ±z. Pegamos a força total F = F+ + F− .
Se kvz é pequeno em comparação com Γ e ∆, achamos através de uma expansão de Taylor
kvz (2∆/Γ)
Fz ' 4~ks . (14.27)
1 + s + (2∆/Γ)2
Esta expressão mostra, que se a dessintonia ∆ é negativa (isto é, para o lado vermelho da
ressonância), então a força de resfriamento vai se opor ao movimento e ser proporcional à
velocidade atômica. A Fig. 14.4 mostra essa força dissipativa restauradora em função de vz
numa dessintonia ∆ = −Γ e I = Isat /2. O movimento unidimensional do átomo, sujeito a
uma força de oposição proporcional à sua velocidade, é descrito por um oscilador harmônico
amortecido. O ’amortecimento Doppler’ ou ’coeficiente de atrito’ é o fator de proporcionalidade,
−4k 2 (2∆/Γ)
αd = s . (14.28)
1 + s + (2∆/Γ)2
No entanto, o átomo não vai resfriar indefinitivamente. Num certo ponto a taxa de resfria-
mento Doppler será equilibrada pela taxa de aquecimento vindo das flutuações de momento do
átomo absorvendo e reemitindo fótons. O limite de resfriamento Doppler é dado por,
Γ
kB T = ~ . (14.29)
2
This limit is generally, by alkaline atoms, in the order of hundreds of microkelvin. In the early
years of cooling and imprisonment, before 1988, the Doppler threshold was thought to be a
true physical barrier. But this year, several groups have shown that, in fact, atoms could be
cooled well below
114 the Doppler
CHAPTERthreshold. The FROM
6. FORCES effect is caused by the
ATOM-LIGHT hyperfine structure of the
INTERACTION
ground state. Let us describe simplified one-dimensional models in the next section. Resolve
û4ÿ û

!"#%$&
û4ÿ ý '" #(
!)$+*(," -. /#0

Γ
û4ÿ ûû

∆ω
4 132 Γ
ü4
û ÿý 4 5687
139
ü û4ÿ û
üªý û ü|þ û û þ û ý û

Figure 14.4: One-dimensional radiative pressure

Figure 6.2: One dimensional Doppler
Doppler radiation force as
pressure a function
force of the atomic velocity
vs. atom velocity
along the axis along
z forz-axis red−Γ
a redfor∆a = and aof light
detuning intensity
one natural of I and
line width = 2I sat . intensity
a light The solidof line shows the
2Isat
exact expression for. The
thefull line plots the
restorative exact
force [Eq expression
(14.26)].forThe
the restoring
broken force
line (Eq.6.13).
shows the approximate,
The dashed line plots the approximate expression (linear in velocity dependence)
linear expression of 6.14.
of Eq. the velocity dependence of the Eq. (14.27).
the Excs. 14.7.2.1, 14.7.2.2, and 14.7.2.3.

This expression shows that T is a function of the laser detuning, and the mini-
mum temperature is obtained when ∆ω = − Γ2 . At the this detuning,
Γ
kB T = ~ (6.18)
2
which is called the Doppler-cooling limit. This limit is typically, for alkali atoms,
on the order of a few hundred microkelvin. For example the Doppler cooling
limit for Na is T = 240 microkelvin. In the early years of cooling and trapping,
prior to 1988, the Doppler limit was thought to be a real physical barrier, but
in that year several groups showed that in fact Na atoms could be cooled well
14.2.2 Sub-Doppler cooling

Two major mechanisms cooling atoms to temperatures below the Doppler boundary are based
on spatial polarization gradients of the light field through which the atoms move. These two
mechanisms, however, invoke very different physical mechanisms, and are distinguished by the
spatial dependence of the polarization of light. A key point is that these under-Doppler mecha-
nisms only work on atoms with multiple levels. In particular, it is essential to have several levels
in the ground state. Two parameters, the coefficient of friction and the velocity of capture,
determine the importance of these cooling processes. In this section we compare the expres-
sions for these quantities in the sub-Doppler regime to those found in the conventional model of
one-dimensional Doppler cooling in optical melasses.
Figure 14.5: The upper line shows the change in polarization as a function of distance (in units
of the wavelength) for the ’lin-perp-lin’ stationary waveform configuration. The figure below
shows a simplified image of the Sisyphus cooling mechanism for a two-level atom Jg ↔ Je .
14.2.2.1 Lin ⊥ lin molasses

No primeiro caso, duas ondas de luz contrapropagantes com polarizações ortogonais formam
uma onda estacionária. Essa configuração é familiarmente chamada ’lin-perp-lin’. Fig. 14.5
mostra o que acontece [61]. Vemos na figura que, se tomarmos como ponto de partida uma
posição onde a polarização da luz é linear (1 ), depois de uma distância de λ/8 a polarização
evolue de linear para circular com orientação σ− . Em seguida, durante os próximos λ/8, a
polarização novamente altera para linear, mas em uma direção ortogonal à primeira (2 ). Em
seguida, entre λ/4 e 3λ/8 a polarização se torna novamente circular, mas no sentido oposto
(σ+ ), e finalmente, depois de uma distância de λ/2 a polarização é novamente linear, mas fora
de fase com relação a (1 ). Ao longo da mesma distância, o acoplamento luz-átomo produz
um deslocamento periódico da energia (light-shift) dos nı́veis Zeeman do estado fundamental.
Para ilustrar o mecanismo de resfriamento, assumimos o caso mais simples, de um transição
Jg = 21 −→ Je = 23 . Como mostrado na Fig. 14.5 o átomo movendo-se através da região
de z em torno λ/8, onde a polarização é primariamente σ− , terá sua população bombeada
para Jg = − 21 . Além disso, os coeficientes de Clebsch-Gordan que controlam o acoplamento
dipolar da transição para Je = 23 impõem que o nı́vel Jg = − 12 acopla-se para σ− com uma
força três vezes maior do que o Jg = + 12 . A diferença das forças de acoplamento leva para o
desdobramento do light shift entre os dois estados fundamentais mostrado na Fig. 14.5. Como
o átomo continua a mover-se para +z, as forças relativas do acoplamento são revertidos perto
de 3λ/8, onde a polarização é essencialmente σ+ . Assim, os nı́veis relativas de energia dos dois
estados fundamentais hiperfinos oscilam ’fora de fase’, quando o átomo se move através da onda
estacionária. A idéia fundamental é, que a taxa de bombeamento óptico, sempre redistribuindo
população ao nı́vel hiperfino inferior, retarda o light shift do átomo se movendo através do
campo. O resultado é um ’efeito de Sı́sifo’, onde o átomo fica a maioria do tempo em subnı́veis
escalando uma colina de potencial, assim convertendo energia cinética em energia potencial. Essa
energia potencial acumulada é posteriormente dissipada por emissão espontânea para os modos
eletromagnéticos do vácuo. Simultaneamente, a emissão espontânea transfere a população de
volta para a os dois nı́veis mais baixos do estado fundamental. O diagrama inferior da Fig. 14.5
ilustra o atraso de fase do bombeamento óptico. Para que esse mecanismo de resfriamento
funciona, o tempo de bombeamento óptico, que é controlado pela intensidade da luz, deve ser
menor do que o tempo do light shift, essencialmente dado pela velocidade do átomo. Como o
átomo está se movendo lentamente, tendo sido previamente resfriado pelo mecanismo Doppler,
o campo de luz deve ser fraco, a fim de diminuir a taxa de bombeamento óptico e de atrasar
o bombeamento com respeito á modulação do light-shift. Essa imagem fı́sica combina a força
óptica dipolar conservadora, cujo integral espacial dá origem aos montes e vales do potencial,
sobre os quais o átomo se move, e a dissipação de energia irreversı́vel por emissão espontânea,
que é necessária para obter um resfriamento. Podemos tornar a discussão mais precisa e obter
expressões simples para o coeficiente de atrito e a velocidade de captura através de algumas
definições. Como no modelo de resfriamento Doppler nós definimos o coeficiente de atrito αlpl
como constante de proporcionalidade entre a força F e a velocidade atômica v.
F = −αlpl v . (14.30)
Assumimos que o campo de luz e dessintonizado para o vermelho da transição Jg - Je ,
∆ = ω − ω0 , (14.31)
e denotamos os light-shifts dos nı́veis Jg = ± 12 como ∆± , respectivamente. Na posição z = λ/8,

∆− = 3∆+ e em z = 3λ/8, ∆+ = 3∆− . Como o campo aplicado é sintonizado pelo vermelho,
todos os ∆± têm valores negativos. Agora, para o mecanismo de resfriamento ser efetivo, o
tempo do bombeamento óptico τp deve ser semelhante ao tempo requisito para um átomo com
a velocidade v de se mover de baixo para cima do monte de potencial, λ/4 v ,
λ/4
τp = (14.32)
v
ou
Γ0 ' kv , (14.33)
onde Γ0 = 1/τp e λ/4 ' 1/k. Agora, a energia W dissipada durante um ciclo de escalada e
emissão espontânea é, essencialmente, a diferencia de energia média dos dois estados fundamen-
tais deslocados pelo light-shift, ∆ls ≡ ∆+ + ∆− , por exemplo, entre os pontos z = λ/8 e 3λ/8
ou W ' −~∆ls . Portanto, a taxa de dissipação de energia é
dW
= Γ0 ~∆ls . (14.34)
dt
No mesmo tempo, cada mudança de energia temporal de um sistema sempre pode ser exprimida
como dW
dt = F · v. Portanto, nesse modelo unidimensional, considerando a Eq. (14.31), podemos
escrever,
dW
= −αlpl v 2 = −Γ0 ~∆ls , (14.35)
dt
tal que com (14.33),
Γ0 ~∆ls ~∆ls ~k 2 ∆ls
αlpl = − ' −kv ' − . (14.36)
v2 v2 Γ0
Note que desde que ∆ < 0, αlpl é uma grandeza positiva. Também note, que em grandes
dessintonias, (∆ Γ) a Eq. (12.23) dá,
U ∆ls Ω2
= = . (14.37)
~ 4 4∆
Também é verdade que para grandes light shifts em comparação com a largura natural
do estado fundamental (∆ls Γ0 ), e para grandes dessintonias no vermelho da ressonância
(∆ & 4Γ),
Γ ∆2
' (14.38)
Γ0 4Ω2 .
Portanto, o coeficiente de atrito sub-Doppler também pode ser escrito
~k 2 ∆
αlpl = − (14.39)
4Γ
Eq. (14.39) faz duas predições remarcáveis: Primeiro, o coeficiente de fricção sob-Doppler ’lin-
perp-lin’ pode ser um grande número em comparação com αd . Note que da Eq. (14.28), com
I . Isat e ∆ Γ,
3
2 Γ
αd ' ~k , (14.40)
∆
e
4
αlpl ∆
' . (14.41)
αd Γ
Segundo, αlpl é independente da intensidade do campo aplicado. Esse último resultado é difer-
ente do coeficiente de fricção, que é proporcional a intensidade do campo até saturação (vide
Eq. (14.28)). No entanto, mesmo quando αlpl parece impressionante, a gama de velocidades
atômicas onde pode operar é restrita pela condição,
Γ0 ' kv . (14.42)
A razão das velocidades de captura para resfriamento Doppler versus sob-Doppler é, portanto,
somente
vlpl 4∆ls
' . (14.43)
vd ∆
Fig. 14.6 ilustra graficamente a comparação entre os mecanismos de resfriamento Doppler e sub-
Doppler ’lin-perp-lin’. A diferencia dramática na gama de captura é evidente da figure. Note
também que as pentes nas curvas dão os coeficientes de fricção para os dois regimes e que, dentro
da gama de captura, a pente fica bem mais rı́gida para o mecanismo sub-Doppler.
118 CHAPTER 6. FORCES FROM ATOM-LIGHT INTERACTION
NPO Q
l
k NPOR
Γ
ij
h_
g bc NPO N
d ef
a bc S NPOR
^ _`
S NPO Q
S RUT S R#N ST N T RUN R#T
VW
XY[Z\]
Figure 6.4: Comparison of slope, amplitude, and ’capture range’ of Doppler

Figure 14.6: Comparação de cooling
encosta, amplitude
and Sisyphus cooling. e intervalo de captura dos resfriamentos Doppler
e Sı́sifo.
and µ ¶4
αlpl 1 ∆ωL
'
αd 2 Γ
14.2.2.2 σ + -σ − molasses
and second, that α is independent of the applied field intensity. This last
lpl
result differs from the Doppler friction coefficient which is proportional to field
intensity up to saturation (cf. Eq. 6.15). However, even though αlpl looks
O segundo mecanismo opera comthedois
impressive, range offeixes
by the condition that,
contrapropagantes
atomic velocities circularmente polarizados em
over which it can operate is restricted
sentidos opostos. Quando os dois feixes contrapropagantes Γ ' kv têm a mesma amplitude, a polar-
0
The ratio of the capture velocities for Doppler vs. sub-Doppler cooling it there-
ização resultante é sempre fore
linear
only.
e ortogonal ao eixo de propagação, mas a ponta do vetor de
4∆
polarização traça uma hélice com um passo de v /vλ' (vide Fig. 14.7). A fı́sica do mecanismo sob-
∆ω
lpl d
L
Doppler não implica escaladas Figurede colinas

6.4 illustrates e emissão
graphically espontânea,
the comparison mas
between the Doppler and um
the desequilı́brio das taxas
“lin-perp-lin” sub-Doppler cooling mechanism.The dramatic difference in cap-
de espalhamento de fótons turedasrange duas ondas
is evident from thede luzNotecontrapropagantes
figure. also that the slopes of the curvesquando o átomo se move
give the friction coefficients for the two regimes and that the within the narrow
ao longo do eixo z. Este desequilı́brio
velocity capture range ofleva a theuma
it action, slope offorça dependente
the sub-Doppler mechanism is da velocidade tentando
markedly steeper.
restaurar a posição do átomo.TheOsecond fator essencial
mechanism operates with levando à taxa debeams
the two counterpropagating espalhamento
cir- diferencial é a
cularly polarized in opposite senses. When the two counterpropagating beams
criação de uma orientação da população ao longo do eixo z entre os
have the same amplitude, the resulting polarization is always linear and orthogo-subnı́veis do estado atômico
fundamental. Os subnı́veis with mais populados espalham mais fótons. Agora, considerando o dia-
nal to the propagation axis, but the tip of the polarization axis traces out a helix
a pitch of λ. Figure 6.5 illustrates this case. The physics of the sub-Doppler
grama de nı́veis energéticosmechanism
(vide does Fig. not 14.5) e a simetria
rely on hill-climbing dosemission,
and spontaneous coeficientes
but on an de Clebsch-Gordan, é
1 3
evidente que transições Jg = 2 ↔ Je = 2 acopladas por luz linearmente polarizada não podem
produzir uma orientação da população no estado fundamental. Na verdade, o sistema mais sim-
ples mostrando este efeito é Jg = 1 ↔ Je = 2. Uma medida para essa orientação é a magnitude
do elemento da matriz hJz i entre os subnı́veis Jgz = ±1. Se o átomo permaneceu estacionário
em z = 0, interagindo com a luz polarizada ao longo da direção y; os light shifts ∆0 , ∆± dos
três subnı́veis do estado fundamental seriam
∆+1 = ∆−1 = 34 ∆0 , (14.44)
e as populações estacionárias 4/17, 4/17 e 9/17, respectivamente. Obviamente, luz polarizada

linearmente não produz orientação estacionária, hJz is = 0. Más quando o átomo começa de
se mover ao longo do eixo z com velocidade v, ele vê uma polarização linear precessando em
torno do eixo da propagação com o ângulo ϕ = kz = −kvt. Essa precessão dá origem a um
novo termo no hamiltoniano, V = kvJz . Além disso, quando transformamos para um sistema
rodando de coordenadas, as auto-funções do hamiltoniano do átomo em movimento nesse novo
sistema ’inercial’ tornam-se combinações lineares das funções de base do átomo em repouso. A
avaliação do operador de orientação estacionário, Jz , no sistema inercial agora não é zero [61],
40 ~kv
hJz i = = ~(Π+ − Π− ) . (14.45)
17 ∆0
Note que a medida da orientação somente é diferente de zero quando o átomo se move. Na
Eq. (14.45) denotamos as populações dos subnı́veis |± > como Π± , e interpretamos os elementos
da matriz que não são zero como uma medida direta da diferencia de população entre os nı́veis
14.2. OPTICAL6.3.
COOLING
SUB-DOPPLER COOLING 119 397
σ+ σ−
o
n λprq λps
Figure 14.7: Polarização como função da distancia (em unidades do comprimento de onda) para
Figure 6.5: Polarisation as a function of distance (in wavelength units) for the
a configuração deσonda
+ estacionária σ + -σ − .
, σ − standing wave configuration.
|± > do estado fundamental. Note que, desde que ∆0 étheuma

imbalance in the photon scattering rate from
quantidade negativa (sintonização
two counter-propagating light
vermelha), a Eq. waves
(14.45)as the atom moves along the z axis. This− imbalance do
nós explica, que a população Π é maior leadsque
to aΠvelocity-
+ . Agora, quando o
átomo viajando na direção +z fica exposto à duas ondas luminosas, uma com a to
dependent restoring force acting on the atom. The essential factor leading polarização σ−
(σ+ ) propagandothe nadifferential
direção −zscattering
(+z),rate is the creation of population
a preponderância orientation
de população no along the |−i resultará
estado
z axis among the sublevels of the atom ground state. Those sublevels with more
numa taxa de espalhamento mais alta pela onda propagante na direção −z. Portanto, o átomo
population scatter more photons. Now it is evident from a consideration of the
será sujeito à umaenergy
forçalevel
total oposto
diagram a seu
(v.s. movimento
Fig. 6.3) e proporcional
and the symmetry a sua velocidade. A taxa
of the Clebsch-Gordan
de espalhamento coefficients
diferencialthat é Jg = 21 ↔ Je = 32 transitions coupled by linearly polarized
light cannot produce a population 40 kvorientation
0 in the ground state. In fact the
simplest system to exhibit this effect isΓJg.= 1 ↔ Je = 2, and a measure of the (14.46)
17 ∆0
orientation is the magnitude of the hJz i matrix element between the Jgz = ±1
Com o momento sublevels.
quantizado ~katom
If the transferido
remained em cada evento
stationary at z = 0,de espalhamento,
interacting with the alight
força total é
polarized along y, the light shifts ∆0 , ∆±1 of the three ground state sublevels
would be 40 ~k 2 vΓ0
F = . (14.47)
17 ∆0 3
∆+1 = ∆−1 = ∆0 (6.27)
4
O coeficiente fricção αcp é
40 4/17, 0
and the steady-state populations 4/17, 2 Γ and 9/17 respectively. Evidently
α = − ~k , (14.48)
linearly polarized light will cp
not produce
17 net
a ∆0steady-state orientation, hJz i. As
the atom begins to move along z with velocity v, however, it sees a linear polar-
o que é uma quantidade positiva,
ization precessing desde
around queof ∆
its axis 0 é negativo
propagation with anpara sintonização
angle ϕ = −kz = −kvt. vermelha. Con-
trastando αcp com αlpl
This vemos gives
precession que αrise deve
cp to serterm
a new muito menor
in the desde que
Hamiltonian, V =sempre assumimos que
kvJz . Fur-
os light shifts ∆ thermore,
sejam muitoif we maiores
transform do to aque
rotating coordinatedeframe,
as larguras linhatheΓ0 eigenfunctions
. No entanto, a taxa de
belonging to the Hamiltonian of the moving atom in this new ’inertial’ frame
aquecimento devido ás flutuações de recuo também é muito menor. Assim, as temperaturas
become linear combinations of the basis functions with the atom at rest. Eval-
mı́nimas que podemuation ofalcançadas
ser a partir
the steady-state dos dois
orientation mecanismos
operator, sob-Doppler
Jz , in the inertial frame is são
now comparáveis.
Embora o mecanismo do resfriamento Doppler também depende de um desequilı́brio de es-
palhamento de ondas luminosas contrapropagantes, esse desequilı́brio vem do fato que, devido ao
deslocamento Doppler sentido pelos átomos em movimento, as probabilidades para espalhamento
por fóton são diferentes. No mecanismo sub-Doppler as probabilidades de espalhamento das duas
ondas de luz são iguais, mas as populações do estado fundamental não são. O estado com maior
população experimenta a maior taxa.
14.2.3 Raman cooling

14.2.3.1 Optical cooling of confined particles
It is also possible to cool ions confined in a trap [255]. The direction of their motion and
their velocity change periodically with the secular frequencies ζr und ζz . For optical cooling
it is sufficient to irradiate a single red-detuned running-wave light field: In a real ion trap the
cylindric symmetry cannot be realized with absolute precision so that we get different secular
frequencies ζx 6= ζy 6= ζz and a coupling of the degrees of freedom for all directions of space.
The cooling of the ionic motion in a single direction results in a cooling of the motion in the
other directions.
In the rest system of an ion oscillating in a harmonic trap the Doppler-shift of the laser
frequency changes periodically: v(t) = v0 cos ζr,z t. The ion absorbs therefore in its rest system
the light on a comb of sidebands whose distance and strength depend on the oscillation frequency
and amplitude. The absorption profile of a transition in such a harmonically vibrating ion follows
as a convolution of the Lorentz profile LΓ , describing the naturally broadened resonance, with
a function S, describing the splitting of the absorption profile into sidebands [138]:
X
A(∆) = (LΓ ? S)(∆) , S(∆) = Jn (k · v0 /ζr,z )2 δ(∆ − nζr,z ) . (14.49)
n
Jn denotes the Bessel function of nth order. In essence, the system is governed by three time-
constants: The natural decay width of the cooling transition Γ is a measure for the inneratomic
time scale, since it determines the average duration of absorption-emissions cycles. The secular
frequencies ζr,z determine the time scales for changements in the external degrees of freedom,
i.e. for changements of the ion’s location and velocity. The Doppler-shift kv0 of the resonance
frequency in the return point of the ion motion finally, is a measure for the kinetic energy of the
ion.
The relative importance of these three characteristic frequencies reveal the state of the ion
in the trap. The modulation index kv0 /ζr,z decides on the hight and the number of sidebands in
the excitation spectrum. The better the ion has been cooled, the smaller the modulation index
and the smaller the hight and number of sidebands. The kinetic energy of the ion is,
1 1 2 2
Ekin = mv02 = mζr,z x0 . (14.50)
2 2
The modulation index kv0 /ζr,z = kx0 = 2πx0 /λ is also called Lamb-Dicke parameter. By
cooling the der Lamb-Dicke parameter is so much reduced and the ion is so well localized that
its motional sidebands are smaller than the wavelength of the exciting light. It then is in the
so-called Lamb-Dicke regime x0 λ [?] and has so small motional sidebands that they do not
contribute to the line shape and do not influence the line width. Therefore the linear Doppler
effect vanishes.
The quantity ζr,z /Γ defines the resolution of the sidebands. If the resolution is poor, we talk
about weak confinement, else about strong confinement. Therefore the same ion can be weakly
confined with respect to an allowed transition and strongly confined with respect to a forbidden
transition. The cooling processes in the two cases of strong and weak trapping must be described
differently. At weak confinement the oscillation frequency ζr,z is so slow that many absorption-
emission cycles with the time constant Γ−1 can occur during one oscillation period. Cooling
process and cooling limit are approximately the same as for free particles and are described by
Doppler cooling.
14.2.3.2 Raman sideband cooling

In the case of strong confinement for the description of the cooling process we must consider
the quantization of the motional energy in the harmonic potential. The two levels coupled
by the narrow transition split into vibrational sublevels |nr,z i, which are populated in thermal
equilibrium according to the Bose-Einstein distribution and have the kinetic energies Ekin ,
1
nr,z = and Ekin = ~ζr,z (nr,z + 21 ) . (14.51)
e~ζr,z /kB T −1
To perform the so-called optical cooling sideband cooling [255] the laser is tuned to the first
lower sideband. The laser light is then scattered in a Raman-Anti-Stokes process at the excited
electronic state with a vibrational quantumnumber lower by 1 |e, nr,z − 1i. The subsequent
spontaneous decay occurs most probably to the same vibrational substate of the ground state
|g, nr,z − 1i. The net effect of such a scattering process therefore is a transition to the next lower
vibrational quantum number. The zero point energy of the ion in the trapping potential cannot
be underscored by cooling, Ekin > 12 ~ζr,z (for the Yb+ ion it is Ekin > 2 neV). However, the
uncertainty of the kinetic energy, and the temperature T given by (14.51) have no lower limit
[75].
At every absorption process, free particles carry away the momentum of the photons ~k.
The recoil of a free Yb+ ion corresponds to the frequency shift ε/2π = 5.3 kHz. On a narrow
transition, it yields a resonance at the frequency ε. For trapped ions, this is not the case, because
the momentum is absorbed by the whole trap (s. analogy to the efeito Mößbauer).
14.2.4 Stimulated Raman sideband cooling

We may use two lasers detuned far from resonance to couple two vibrational states. However,
additional dissipation by optical pumping is still required.
Numerous schemes have been tested to cool neutral atoms in optical lattices. For the schemes
to work, the ion should be already in the Lamb-Dicke regime. Otherwise, transitions with
transfer of higher vibrational quantum numbers nr,z are possible during spontaneous emisssion.
The Lamb-Dicke limit is set by kr < 1, or,
mωtrap
hni = . (14.52)
2~k 2
This means that higher trap frequencies ease the required temperature at which sideband cooling
can start to work.
14.2.5 Adiabatic decompression

A condição para descompressão adiabática de um potencial de aprisionamento é,
|ω̇trap |
ωtrap . (14.53)
ωtrap
A população dos nı́veis quantizados não deve mudar sob descompressão adiabática, e~ωi /kB Ti =
e~ωf /kB Tf , e a densidade do espaço de fase fica inalterada, ni λdB,i = nf λdB,f . Se isto é verdade,
então a temperatura e a densidade mudam como,
3/2
ωf Tf nf
= = . (14.54)
ωi Ti ni
14.3 Optical and magneto-optical traps

14.3.1 The magneto-optical trap
Um obstáculo aparentemente temı́vel para a ideia de confinar partı́culas por forças ópticas é o
teorema de Earnshaw óptico. Este teorema afirma que, se uma força é proporcional à intensidade
da luz, sua divergência deve ser nula, porque a divergência do vetor de Poynting, que expressa
o fluxo direcional de intensidade, é nula num volume sem fontes ou sumidouros de radiação. A
ausência de divergência impede a possibilidade de uma força restauradora para o interior em
todos os lugares de uma superfı́cie fechada [9]. No entanto, o teorema de Earnshaw óptico pode
ser contornado por um truque esperto. Levando em conta os graus de liberdade internos do
átomo (ou seja, os nı́veis de energia), eles podem mudar a proporcionalidade entre a força e o
vetor de Poynting de uma forma dependente da posição, tal que o teorema de Earnshaw óptico
não se aplica. Confinamento espacial é, então, possı́vel utilizando a força de pressão radia-
tiva produzida por feixes ópticos contrapropagantes. A configuração de armadilha atualmente
mais utilizada se base num gradiente radial de um campo magnético produzido por um campo
quadrupolar e três pares de feixes ópticos contrapropagantes, circularmente polarizados, dessin-
tonizados para o vermelho da transição atômica que se cruzam em ângulo reto no ponto onde
o campo magnético é zero. Essa armadilha magneto-óptica (MOT) usa o deslocamento Zeeman
dependente da posição dos nı́veis eletrônicos quando o átomo se move no campo magnético ra-
dialmente aumentando. O uso de luz circularmente polarizada, dessintonizada para o vermelho
por cerca de um Γ resulta em uma probabilidade de transição variando espacialmente, cujo
efeito é produzir 122
uma forçaCHAPTER 6. FORCES
restauradora FROM ATOM-LIGHT
que empurra o átomo paraINTERACTION
a origem. Para esclarecer
|G}~>
| }~ {

y ∆
|G}~#
vwx
tu
σ+ σ−
| }~ {
z
{
Figure 14.8: Esquerda:

Figure Diagrama do deslocamento
6.6: Left: Diagram dos nı́veis
of the Zeeman shift of de energia
energy levelsnuma MOT quando um
in a MOT
átomo se move fora do atom
as an centro da armadilha.
moves to away fromAthe força
traprestauradora forceforce
center. Restoring fica becomes
localizada em torno
localized around resonance positions as indicated. Right:
das posições de ressonância indicadas. Direito: Esquema de uma montagem Schematic of a typical
tipica da MOT
MOT setup showing six laser beams and antihelmholtz configuration producing
mostrando seis feixes laser e bobinas
quadrupole magnetic field. em configuração anti-Helmholtz produzindo um campo
magnético quadrupolar.
The atom will therefore experience a net restoring force pushing it back to
como es esquemathe de origin.
aprisionamento funciona,
If the light beams consideramos
are red-detuned ∼ Γ, um
then átomo de dois
the Doppler shiftnı́veis
of com uma
transição J = 0 →
theJatomic
= 1 movendo-se ao longo
motion will introduce da direção z. Aplicamos
a velocity-dependent term to theum campo magnético
restoring
force such that,
B(z) crescendo linearmente for small
com displacements
a distância and velocities,
da origem. the total restoringZeeman
Os deslocamentos force dos nı́veis
can be expressed as the sum of a term linear in velocity and a term linear in
eletrônicos dependem da posição,
displacement1 ,
.
FM OT = F1z + F2z = −αz − Kz. (6.33)
µB gF mF dB
∆B = z ≡ ∂ωzeem z , (14.55)
~ the equation
From Eq. (6.33) we can derive dz of motion of a damped harmonic
oscillator with mass m,
.. 2α . K
z+ z + z = 0. (6.34)
m m
The damping constant α and the spring constant K can be written compactly
in terms of the atomic and field parameters as
2
16 |∆0 | (Ω0 ) (k/Γ)
α = ~kΓ h i2 h i2 (6.35)
2 4(∆0 )2
1 + 2 (Ω0 ) 1 + 1+2(Ω 0 )2
14.3. OPTICAL AND MAGNETO-OPTICAL TRAPS 401
onde µB é a constante Zeeman para o deslocamento de frequência da transição dentro do campo

magnético (vide Fig. 14.8). Também aplicamos campos ópticas contrapropagantes ao longo das
direções ±z tendo polarizações circulares de sentidos opostos e sendo dessintonizados para o
vermelho da transição atômica. Fica claro com a Fig. 14.8, que um átomo movendo para ±z vai
espalhar fótons do tipo σ∓ com uma taxa mais rápida do que fótons do tipo σ± , pois o efeito
Zeeman vai puxar a transição ∆mJ = ∓1 mais perto da frequência da luz. A expressão para a
força de pressão radiativa, que estende a Eq. (14.26) para incluir o efeito Doppler kvz e o efeito
Zeeman, fica
~k 2Ω2
F±z = − Γ . (14.56)
2 4(∆ ± kvz ± ∂ωzeem z)2 + 2Ω2 + Γ2
O átomo vai, portanto sentir uma força restauradora empurrando-lhe de volta para a origem.
Se os feixes de luz são sintonizados para o vermelho por ∆ = −Γ, o deslocamento Doppler do
movimento atômico introduz um termo dependendo da velocidade na força restauradora, tal
que, para pequenas deslocações e velocidades, a força restauradora total pode ser exprimida
pela soma de um termo linear na velocidade e um termo linear na deslocação,
FM OT = F1z + F2z = −αż − κz . (14.57)
Da Eq. (14.57) podemos derivar a equação do movimento de um oscilador harmônico amortecido

com a massa m,
2α κ
z̈ + ż + z = 0 . (14.58)
m m
A constante de amortecimento α e a constante da mola κ podem ser escritos de maneira compacta
em termos de parâmetros atômicos e do campo como
16~kΓΩ2 ∆∂z ωzeem

κ= . (14.59)
4∆2 + 2 · 6Ω2 + Γ2
e
k
α=κ . (14.60)
∂z ωzeem
Condições tı́picas para MOTs são Ω = Γ/2, ∆ = −Γ. Para MOTs tı́picas,
α ' 2 · 10−22 Ns/m e κ ' 3.7 · 10−19 N/m . (14.61)
Também podemos estimar a curvatura da MOT,

r
κ
ω= ' (2π) · 200 Hz . (14.62)
m
MOTs são realizadas com bobinas em configuração anti-Helmholtz gerando potenciais de
geometria quadrupolar. Perto do centro, o campo magnético e seu valor absoluto são bem
aproximados por,  
x p
B = q y  e |B| = qB r2 + 4z 2 , (14.63)
−2z
com r2 = x2 + y 2 e o gradiente q ≡ ∂r B é uma constante, que só depende da geometria das
bobinas e a corrente nelas. Assim, a extensão dos resultados de cima a três dimensões é simples,
se consideramos o fato que o gradiente do campo quadrupolar na direção z é o dobre do gradiente
nas direções x, y, tal que κz = 2κx = 2κy . O termo de amortecimento proporcional à velocidade
implica que a energia cinética E se dissipa do átomo (ou de um conjunto de átomos) como
E/E0 = e−2αt/m , (14.64)
onde m é a massa atômica e E0 a energia cinética no começa do processo de resfriamento.

Portanto, o termo da força dissipativa resfria o conjunto atômico e no mesmo tempo combina
com o termo de deslocação para confinar-lhe. A constante temporal de amortecimento,
m
τ= (14.65)
2α
é tipicamente dezenas de microssegundos. É importante ter em mente que uma MOT é anisotrópica,
desde a força de restauração ao longo do eixo z é o dobro da força de restauração na plano x-y.
Além disso, uma MOT fornece uma armadilha dissipativa, não conservadora. Por isso é mais
correto caracterizar-a pela velocidade de captura máxima, em vez de dar a ’profundidade’ da
armadilha.
Os primeiros experimentos com átomos presos em MOTs foram realizadas desacelerando
um feixe atômico para carregar a armadilha. Mais tarde, uma fonte contı́nua não resfriada
foi usada para isso, mostrando que a armadilha pode ser carregada com átomos lentos de um
vapor em temperatura ambiente. Assim fui possı́vel utilizar uma célula de vapor para capturar
átomos frios da extremidade de baixa velocidade da distribuição de Maxwell-Boltzmann, e as
montagens experimentais tornaram-se mais simples. Agora muitos grupos no mundo utilizam
esses montagens para aplicações que vão desde a espectroscopia de precisão para o controle
óptico de colisões reativas. Resolva os Excs. 14.7.3.1 e 14.7.3.2.
14.3.1.1 Density in a MOT

A MOT normalmente captura em torno de um milhão de átomos em um volume menos de 1 mm3 ,
resultando em densidades de ∼ 1010 cm−3 . Dois processos limitam a densidade alcançável numa
MOT: (1) perdas de átomos aprisionados por colisão e (2) forças repulsivas entre os átomos
causadas por reabsorção de fótons espalhados no interior da armadilha. Perdas por colisões,
decorrem de duas fontes: Átomos quentes do gás residual dentro da câmara que batem em
átomos frios fora da MOT por impacto elástico, e encontros binárias entre os átomos frios. A
’repulsão induzida por fótons’ ou aprisionamento de radiação surge quando um átomo perto
do centro da MOT espontaneamente emite um fóton que é então reabsorvido por um outro
átomo antes de poder sair do volume da MOT. Essa absorção resulta num aumento de 2~k
no impulso relativo dos dois átomos e produz uma força repulsiva proporcional ao produto da
seção transversal de absorção para o feixe de luz incidente. Quando esta força externa repulsiva
equilibra a força confinante, um aumento do número de átomos aprisionados leva a maior nuvens
atômicas, mas não para densidades mais altas.
14.3.1.2 Dark SPOT

A fim de superar o efeito de ’aprisionamento de radiação’, os átomos podem ser opticamente
bombeados para um nı́vel hiperfino ’obscuro’ do estado fundamental que não interage com a
luz da armadilha. Em uma MOT convencional geralmente emprega-se um feixe de luz auxiliar
chamado ’rebombeio’, copropagante com os feixes da armadilha, mas sintonizada para uma
transição vizinha entre nı́veis hiperfinos dos estados fundamental e excitado. O rebombeio
recupera a população esvazianda fora da transição de cı́clica entre os dois nı́veis usados para a
MOT. Como um exemplo, a Fig. 14.9 mostra as transições da armadilha e do rebombeamento

normalmente empregado numa MOT de Na. O esquema, conhecido como armadilha óptica
obscura de força espontânea (dark SPOT), passa o feixe de rebombeio através de uma placa de
vidro com um pequeno ponto preto. A sombra desse ponto é projetado no centro da armadilha de
tal maneira, que os átomos no centro não são rebombeados para a transição cı́clica, mas gastam
a maior parte do tempo (∼ 99%) em nı́veis hiperfinos ’obscuros’. No entanto, o resfriamento
e confinamento continuam a funcionar na periferia da MOT, mas o núcleo central não sente
pressão para fora. A dark 6.4.
SPOT aumenta a densidade
THE MAGNETO-OPTICAL TRAP (MOT) por quase duas
125 ordens de magnitude.

%)P

+ ¢¡ £¤ ¥¦
§©¨«ª¬®ª %
+ ¢¡ £¤ ¥& 3

Figure 6.7: Hyperfine structure in sodium atom showing the usual cooling,
Figure 14.9: Estrutura hiperfina em átomos de sódio mostrando as transições usuais de resfria-
pumping, and repumping transitions.
mento, de bombeamento e rebombeamento.
red of resonance. The obvious advantage of large detunings is the suppression
of photon absorption. Note from Eq. 6.2 that the spontaneous force, involving
absorption and reemission, falls off as the square of the detuning while Eq. 6.8
shows that the potential derived from dipole force falls off only as the detuning
itself. At large detunings and high field gradients (tight focus) Eq. 6.8 becomes
14.3.2 Optical dipole traps ~ |Ω0 |
2
U' , (6.39)
4∆ω
Para aplicar variações temporais
which shows thatnothepotencial
potential becomesdedirectly
confinamento,
proportional to light o uso de campos magnéticos é
intensity
and inversely proportional to detuning. Therefore at far detuning but high
um procedimento lento e desajeitado. Do outro lado, temos feixes
intensity the depth of the FORT can be maintained but most of the atoms lasers que podem ser variados
rapidamente e de maneira MOTs bemare:localizada. A força dipolar exercida
will not absorb photons. The important advantages of FORTs compared to
12 −3
(1) high density (∼ 10 cm ) and (2) a well-defined polarization
por um feixe laser é igual ao
gradiente da frequência Rabi axis F = −∇(d · E), isto é, ela dá origem a um potencial óptico além do
along which atoms can be aligned or oriented (spin polarized).
disadvantage is the small number of trapped atoms due to small FORT volume.
The main
potencial de aprisionamento. The bestAnumber

força pode
achieved ser10atraente
is about atoms. 4
(na direção do máximo de intensidade)
ou repulsiva e ter uma grande variedade
6.4.5 Magnetic trapsde geometrias possı́veis, por exemplo, com um feixe
de laser focado, um pode modificar
Pure magnetic traps a have
densidade
also been used de local
to study e agitar
cold collisions, por
and they are exemplo um condensado.
critical for the study of dilute gas-phase Bose-Einstein condensates (BECs) in
A interação dipolar é sempre acompanhada
which collisions figure importantly. We deanticipate
umatherefore
manipulação
that magnetic trapsda fase de Broglie de que
esculturas a frente de onda questão de uma forma controlável. Assim, é interessante imaginar
configurações especı́ficas para o campo de luz, como por exemplo ondas estacionárias formando
redes ópticas (vide Sec. ??).
Embora a MOT funciona como uma ’célula de reação’ versátil e robusta para o estudo de
colisões frias, as frequências dos feixes de luz devem ser sintonizados perto de transições atômicas,
e um fração considerável dos átomos permanece em estados excitados. Perdas da armadilha por
colisões em estados excitados e repulsão induzida por fótons limitem as densidades realizáveis.
Uma armadilha óptica sintonizada longe de ressonância (far off-resonance optical trap, FORT),
em contraste, utiliza a força dipolar em vez da força espontânea para confinar os átomos e pode,
portanto, operar longe de ressonância com uma população insignificante em estados excitados.
A FORT consiste em um único feixe Gaussiano, linearmente polarizado, bem focalizado e muito
dessintonizado no vermelho da ressonância. A vantagem óbvia de grandes dessintonias é a su-
pressão da absorção de fótons. Note da Eq. (12.21), que a força espontânea, envolvendo absorção
e reemissão, cai com o quadrado da dessintonia enquanto o potencial derivado da força dipo-
lar cai apenas linearmente com a dessintonia. Para grandes dessintonias e fortes gradientes do
campo as Eqs. (12.23) e Eqs. (12.24) tornam-se [104] 2 ,

~Ω(r)2 3πc2 Γ I(r) 3πc2 Γ 2
U (r) ' = I(r) e ~γsct (r) ' σa (∆) = I(r) , (14.66)
4∆ 2ω03 ∆ ω 2ω03 ∆
p
utilizando a frequência de Rabi ~Ω = d12 E, o momento dipolar d = 3πε0 ~Γ/k 3 e a intensidade
I = ε20 c|E|2 . Isso mostra que o potencial torna-se diretamente proporcional à intensidade de
luz e inversamente proporcional à dessintonia. Portanto, em grande dessintonia mas intensidade
muito alta, a profundidade da FORT pode ser mantida, enquanto os átomos não absorvem fótons.
As vantagens importantes das FORTs em comparação com MOTs são: (1) altas densidades
(∼ 1012 cm−3 ) e (2) um eixo de polarização bem definido ao longo do qual os átomos podem
ficar alinhados ou orientados (polarização dos spins).
Feixes de lasers são facilmente manipuláveis em posição, intensidade e frequência. Por isso, a
forca dipolar de feixes sintonizados longe de ressonâncias são frequentemente usados para trans-
porte ou manipulação de objetos microscópicos por optical tweezers. Além disso, sobreposições
de vários feixes permitem realizar geometrias de potenciais interessantes. Uma rede óptica, por
exemplo, construı́da por três pares ortogonais de feixes de lasers, gera um cristal cúbico de
potenciais. Resolva os Excs. 14.7.3.4, 14.7.3.6 e 14.7.3.5.
14.3.2.1 Spin relaxation

Quando o estado fundamental tem uma estrutura hiperfina, existe um outro mecanismo de
relaxação: Perto de ressonância espalhamento Raman pode induzir transições entre estados
hiperfinos causando uma redistribuição das populações entre subestados Zeeman chamada de
relaxação do spin. Em armadilhas magnéticas, isso pode levar à perdas, pois não todos subesta-
dos Zeeman são aprisionados.
A taxa de um processo de espalhamento arbitrário à partir de um estado |F, mi mediante
vários estados excitados possı́veis |Fj0 , m0j i até um estado final |F 00 m00 i é seguinte a formula de
Kramers-Heisenberg [179],
2
(Fj0 m0j )
X αF m→F 0 m0
γF m→F m ∝
0 0 . (14.67)
j ∆Fj0 m0j
Longe de ressonância o espalhamento diminuı́ como ∆2 para espalhamento Rayleigh, F m =

F 0 m0 . Espalhamento Raman, F m 6= F 0 m0 , é adicionalmente suprimido por interferência destru-
tiva dos diferentes caminhos de espalhamento.
No caso do rubı́dio, calculamos,
2
3c2 ω 4 70 2 1 3 1 1 3 1 I0
γspin = Γ − . (14.68)
8π 81 ωD1 ∆D1 ωD2 ∆D2 ~ω
14.3.2.2 Potential of a gaussian beam

A far-off resonance optical trap (FORT) é um exemplo de uma armadilha óptica baseada nas
forças dipolares [104]. A distribuição de intensidade de um feixe gaussiano com o diâmetro na
cintura w0 é 3 ,
2P (−2x2 −2y2 )/w02 −z 2 /z 2
I(r) = e e R , (14.69)
πw02
2
Jürgen Bosse claims that the dipole trap formula given in [104] is wrong!
3
Vide a apostila Eletrodinâmica do mesmo autor [55].
onde Pcav é a potência total do feixe e zR ≡ πw02 /λdip o comprimento de Rayleigh num dado
comprimento de onda λdip . O potencial dipolar é dado por (14.66). Usando a profundidade o
potencial,
3πc2 Γ 2P
U0 ≡ <0 , (14.70)
2ω03 ∆ πw02
podemos aproximar, perto do centro do potencial, isto é, próximo do eixo óptico, r 12 w0 , e
dentro do comprimento de Rayleigh, z πw02 /λ, o potencial por potencial harmônico 4 ,

(−2x2 −2y 2 )/w02 −z 2 /zR
2 2x2 + 2y 2 z2
U (r) ' U0 e e ' U0 1 − − 2 (14.71)
w02 zR

m 2 2 m 2 2 U0 r2 z2
≡ U0 + ωr r + ωz z ≡ kB T + 2+ 2 .
2 2 kB T 2r̄ 2z̄
Isso leva às equivalências,
q √ q
2 U0 2U0
ωr = w0 q m e ωz = zR q m
. (14.72)
w0 kB T zR kB T
r̄ = 2 U0 e z̄ = √
2 U0
Example 54 (Armadilha dipolar para rubı́dio): As formulas (14.66) valem para um

sistema de dois nı́veis. No caso das linhas D1 e D2 do rubı́dio devemos considerar todas
contribuições ponderadas pelas dessintonias respectivas,

~Γ 1 gD2 /gD1 I0 3~πc2 Γ I0
U0 ≡ σ0 + ' ,
4 ∆D1 ∆D2 ~ω 2ω 2 ∆ ~ω
onde gD2 /gD1 = 2.
Similarmente, a taxa de emissão espontânea é,

πc2 Γ2 1 gD2 /gD1 I0
γsct ' + .
2ω 2 ∆2D1 ∆2D2 ~ω
A taxa de emissão espontânea decai mais rapidamente com a dessintonização que a pro-
fundidade do potencial. Assim, para evitar aquecimento, é uma vantagem de trabalhar em
dessintonias elevadas e providenciar intensidades laser mais altas. Definindo a temperatura
de recuo por,
~2 k 2
Trec = ,
kB m
a taxa de aquecimento é [104],
1 ~2 k 2
Ṫ = Trec γsct = γsct .
3 3mkB
Potenciais fracos podem ser deslocados ou distorcidos por gravidade. Por exemplo, numa
armadilha harmônica U = m 2 2 m 2 2 m 2 2 m 2 2 2 m 2 2
2 ωr r + 2 ωz z − mgz = 2 ωr r + 2 ωz (z − g/ωz ) − 2 g /ωz , os
átomos cedem uma altura g/ωz2 . Em armadilhas dependendo do tempo a gravidade causa um
comportamento mais complexo [111]. Trabalhos importantes têm sido feitos por [45, 207, 84,
106, 1, 62, 158, 159, 73].
4
O diâmetro de um feixe gaussiano pose ser caracterizado de várias maneiras,
r̄1/e2 -radius √ r̄1/2-radius
r̄1/√e-radius = √ = 2 r̄1/e2 -radius = ,
2 2 ln 2
e r̄-rms ≡ r̄1/√e-diam e r̄-hwhm ≡ r̄1/2-diam e r̄-diam = 2r̄-radius .
Figure 14.10: (Code: AO T echniques Dipoletraps.m) (esquerda) Potencial dipolar criado por
um feixe gaussiano. (direita) Potencial dipolar criado por uma onda de luz estacionária.
14.3.2.3 Trapping in standing light waves

Se ambas os modos contrapropagantes são bombeados com potências diferentes, P± , a dis-
tribuição de intensidade é,
2 (−2x2 −2y2 )/w02 −z 2 /z 2 p ikz
p 2
−ikz
I(r) = e e R P + e + P − e . (14.73)
πw02
A profundidade do potencial é,

√ √
3πc2 Γ 2( P+ + P− )2
U0 = <0 . (14.74)
2ω03 ∆ πw02
Portanto, dentro do comprimento de Rayleigh, o potencial é,

√
(−2x2 −2y 2 )/w02 P+ + P− + 2 P+ P− cos kz
U (r) ' U0 e √ . (14.75)
P+ + P− + 2 P+ P−
Deixando as potências ser iguais,

P+ =P− (−2x2 −2y 2 )/w02 2x2 + 2y 2 k 2 z 2
U (r) −→ U0 e sin2 kz
2 ' U0 1− − . (14.76)
w02 4
Isso leva às equivalências,

q q
2 U0 U0
ωr = w0 q m e ωz = k 2m
√ . (14.77)
w0 kB T 2
r̄ = 2 U0 e z̄ = k
14.4 Magnetic traps

Armadilhas magnéticas puras também têm sido utilizados para estudar colisões frios, e eles são
fundamentais para o estudo de condensados de Bose-Einstein (BEC) de gases diluı́dos, em que
as colisões jogam um papel importante.
A caracterı́stica mais importante que distingue as armadilhas magnéticas é que eles não pre-
cisam de luz para confinar átomos. Armadilhas sem luz são livre de aquecimento dos átomos por
absorção de fótons, uma condição necessária para a realização de BEC. Armadilhas magnéticas
dependem da interação do spin atômico com campos e gradientes magnéticos para conter átomos.
Dependendo do signo de U e F, átomos em estados cuja energia aumenta ou diminui com o
campo magnético são chamados de ’buscadores de campo fraco’ ou ’buscadores de campo forte’,
respectivamente. Em princı́pio, seria possı́vel aprisionar átomos em qualquer um desses estados,
14.4. MAGNETIC TRAPS 407
necessitando apenas de produzir um mı́nimo ou um máximo de campo magnético. Infelizmente

só buscadores de campo fraco podem ser presos em campos magnéticos estáticos, porque em
espaço livre só pode ter mı́nimos. Mesmo quando os buscadores de campo fraco não estão nos
nı́veis hiperfinos mais baixos, eles ainda podem ser armadilhados, porque a taxa de transição por
emissão espontânea através do dipolo magnético é de ∼ 1010 s−1 . No entanto, colisões mudando
o spin podem limitar a densidade máxima. Resolva o Exc. 14.7.4.1.
A armadilha magnética estática mais básica para átomos neutros baseá-se numa configuração
anti-Helmholtz, semelhante a uma MOT, produzindo um campo magnético quadripolar axial-
mente simétrico. Como essa configuração de campo sempre tem um ponto central onde o campo
magnético desaparece, transições não-adiabáticas de Majorana podem acontecer quando o átomo
passa pelo ponto zero, transferindo a população de um estado de buscador de campo fraco para
um de buscador de campo forte o que efetivamente expulsa o átomo da armadilha. Este problema
pode ser superado usando uma garrafa magnética sem nenhum ponto do campo zero. A garrafa
magnética, também chamada de armadilha de Ioffe-Pritchard, foi utilizada para alcançar BEC
em uma amostra de átomos de Na pré-resfriada em uma MOT. Outros métodos para eliminar
o ponto de campo zero são potenciais orbitando temporalmente mediados (TOP) ou um ’plug
óptico’ que consiste num feixe óptico intenso sintonizado pelo azul alinhado ao longo do eixo de
simetria da armadilha magnética e produzindo um potencial repulsivo para impedir os átomos
de entrar na região de campo nulo. A tecnologia das armadilha continua a se desenvolver e a
recente conquista de BEC vai estimular armadilhas mais robustas contendo um maior número
de átomos. Atualmente ∼ 106 átomos podem ser presos num condensado de Bose-Einstein
carregado por uma MOT contendo ∼ 109 átomos.
Resolva os Excs. 14.7.4.2, 14.7.4.3, 14.7.4.4 e 14.7.4.5.
14.4.1 Quadrupolar traps and Majorana spin-flips

14.4.1.1 Maximum density in quadrupolar traps
Como primeiro exemplo consideramos o campo magnético quadripolar,

 
x
B= y  . (14.78)
−2z
Verificamos facilmente, que,

 
2x
∇·B=0 mas ∇|B| = 2y  . (14.79)
4z
Assim, o potencial magnético quadripolar é linear nas coordenadas espaciais,

p
U (r) = −~
µ · B = −|~
µ||B| = µB gF mF ∂r B r2 + 4z 2 , (14.80)
onde para a armadilha quadripolar, 2∂r B = ∂z B. Para calcular o raio rms de uma nuvem de
temperatura T aprisionada neste potencial colocamos U (r̄) ≡ kB T e obtemos a distribuição de
densidade,
√
r2 +4z 2 /r̄
n(r) = n0 e−U (r)/kB T = n0 e− . (14.81)
A normalização requer,
Z Z ∞Z ∞ √
2 2
N= n(r) d3 r = n0 e− r +4z /r̄ 2πr drdz (14.82)
R 3 −∞ 0
Z ∞Z ∞ Z ∞ 2|z|
2|z|
= n0 2πr̄ 2 −ξ
ξe dξdz = n0 2πr̄ 2
e− r̄ 1 + r̄ dz
−∞ 2|z|/r̄ −∞
Z ∞
= n0 2πr̄3 e−ζ (1 + ζ) dζ = n0 4πr̄3 .
0
Portanto, o volume efetivo é, Vef f = 4πr̄3 .
14.4.1.2 Landau-Zener transitions and Majorana spin flips

A armadilha quadripolar é a mais simples de realizar tecnicamente. Infelizmente, essa armadilha
não é estável por causa de um fenômeno chamado de transições de spin-flip de Majorana ex-
pelindo átomos da nuvem aprisionada. Esse problema é particularmente grave para hidrogênio,
onde pode induzir uma explosão de relaxação [126]. O ponto crı́tico da armadilha é o centro,
onde o campo magnético desaparece deixando os átomos desorientados, isto é, prontos para
reorientar os seus spins.
A frequência de Larmor no campo de uma armadilha quadrupolar é dada por,
µN
ωLarmor = ~ r∂r B . (14.83)
Através da sua velocidade v o átomo veja uma mudança na orientação do campo magnético,
v · ∇|B| v
' . (14.84)
|B| r
Portanto, as transições de Majorana, que acontecem quando essa mudança está mais rápida do
que a frequência Larmor [205], ocorrem dentro de um raio rsf definido por,
v
ωLarmor (rsf ) < . (14.85)
rsf
Eles ocorrem com uma taxa correspondendo ao fluxo de átomos através do elipsoide,
1 N 3 v
' 3 rsf = ... . (14.86)
τsf Rcloud rsf
Em 1995 foram propostas várias técnicas para superar este problema. Foram propostas
geometrias de armadilhas magnéticas, como a armadilha cloverleaf ou a armadilha TOP (time-
orbiting potential), tendo no centro um campo magnético finito, ou armadilhas hı́bridas uti-
lizando um plugue óptico para tapar o vazamento de átomos. Estes avanços técnicos foram os
últimos necessários para atingir a condensação de Bose-Einstein. Resolva o Exc. 14.7.4.6.
14.4.2 Magnetic Ioffe-type traps

Foram desenvolvidos outros potenciais de aprisionamento evitando o problema dos Majoranas
spin-flips. Um exemplo é armadilha magnética tipo potencial de Ioffe,
q
U (r) = µB gF mF B02 + (r∂r B)2 + (z∂z B)2 . (14.87)
Este potencial de aprisionamento magnético pode ser harmonicamente aproximado por,

(r∂r B)2 (z∂z B)2
U (r) ' µB gF mF B0 + + (14.88)
2B0 2B0

m m r2 z2
≡ const + ωr2 r2 + ωz2 z 2 ≡ kB T const + 2 + 2 ,
2 2 2r̄ 2z̄
p
onde os raios rms r̄ = ωr−1 kB T /m seguem da normalização da densidade n(r) = n0 e−U (r)/kB T
ao número de átomos,
Z Z ∞ Z ∞
−r2 /2r̄2 2 2
3
N = n(r)d r = n0 e 2πdr e−z /2z̄ dz = n0 (2π)3/2 r̄2 z̄ ≡ n0 Vef f . (14.89)
0 −∞
As frequências da armadilha podem ser estimadas como,
s
µB (∂r Br,z )2
ωr,z = . (14.90)
mB0
O campo gravitacional da Terra deforma o potencial de aprisionamento e, no caso de um
potencial harmônico, causa um deslocamento gravitacional sem alterar as frequências seculares
do potencial. Estudamos o impacto da gravitação no Exc. 14.7.4.4.
14.4.2.1 Characterization of Ioffe-type traps

A distribuição de densidade ’time-of-flight’ é,
r s r
q
2 2 2 k B T 1 2 kB T
rT oF = r̄ + v̄ tT oF = 2
+ tT oF ' tT oF . (14.91)
m ωr m
A densidade do espaço de fase é,
3/2 3 3
3 N 2π~2 2 ~ Tc
ρ = n0 λdB = 3/2 2
= N ωr ωz = ζ(3) . (14.92)
(2π) r̄ z̄ mkB T kB T T
onde ζ(3) = 1.202 e,
1/3
N ωr2 ωz
kB Tc = ~ (14.93)
ζ(3)
é a temperatura crı́tica. A taxa de colisão máxima é,
r
kB T
γcoll = n0 σv̄ = n0 4πa2s . (14.94)
m
A taxa média de colisão pode ser obtida à partir de,
Z R
1 3 σv̄n2 (r)d3 r
γ̄coll N = γcoll (r)n(r)d r = R . (14.95)
N n(r)d3 r
14.4.3 Radiative coupling and evaporative cooling

Como vimos na última seção, o resfriamento óptico se torna ineficaz quando a densidade do gás
é elevada. Precisamos de um outro mecanismo de dissipação para resfriar átomos confinados
em armadilhas. Um outro método chamado de evaporação tem sido proposto por Hess [125]
para o hidrogênio spin-polarizado (H↑) e foi observado por Masuhara et al. [170]. Mais tarde,
evaporação foi utilizada para como os metais alcalinos [1, 205, 64]. Uma revisão detalhada do
assunto foi publicada por Ketterle e van Druten [145].
14.4.3.1 Evaporative cooling

Evaporação sempre ocorre quando partı́culas energéticas deixam um sistema com energia de
ligação finita, removendo mais do que sua parte na energia média por partı́cula. Consideramos
aqui o caso de um potencial de aprisionamento com extensão finita, isto é, o potencial tem
uma borda ou um bico através do qual átomos quentes com energia cinética suficiente para
atingir essa região podem deixar a armadilha. No caso ideal, isto levará a um trancamento
completa da parte quente da distribuição de velocidade de Maxwell-Boltzmann em equilı́brio.
Se o sistema restante acha de novo um equilı́brio térmico, será em uma temperatura mais baixa.
A redistribuição de energia cinética entre os átomos levando à termalização acontece através
de colisões elásticas. Precisa de mais ou menos três colisões por átomo para retermalizar uma
nuvem [185, 258]. A taxa máxima de colisões elásticas entre átomos aprisionados é
√
γcoll = n0 σv̄ 2 ∝ ρ3 N 2/3 , (14.96)
onde n0 é a densidade de pico,

σel = 8πa2s , (14.97)
é a secção transversal para colisões elásticas e, v̄ sendo a velocidade térmica média da nuvem,
√
2v̄ é a velocidade média relativa entre dois átomos de [144]. Essa formula dá taxa média de
colisões no centro da nuvem, onde a densidade é mais alta. Para calcular a taxa total de colisões,
precisamos integrar sobre o volume inteiro da nuvem,
Z R
1 3 σv̄n2 (r)d3 r
γ̄coll = N γcoll (r)n(r)d r = R . (14.98)
n(r)d3 r
√
Para potenciais harmônicos achamos uma taxa média reduzida por 2 2, para potenciais lineares
por 8. Finalmente, a taxa de eventos colisionais fica duas vezes menor, pois envolve cada vez
dois átomos.
Obviamente, o processo de evaporação fica mais lento, quando a nuvem resfria mais, a
menos que a borda do potencial está reduzida de modo que os átomos mais quente da nuvem
mais fria são evaporados. Baixando continuamente a borda do potencial enquanto a nuvem
atômica está retermalizando (este procedimento é chamado evaporação forçada) temperaturas
muito baixas, no regime de nano-Kelvin, podem ser alcançados, e a densidade do espaço de fase
pode ser aumentada por seis ordens de grandeza até o limiar da condensação de Bose-Einstein.
Naturalmente, isso só é possı́vel sacrificando muitos átomos quentes. Mesmo com uma rampa
de evaporação (isto é, a baixa controlada da borda de trancamento) bem otimizada, apenas 1%
atingem a fase de condensação após cerca 500 colisões por átomo.
Dois aspectos devem ser apontados sobre a otimização da rampa de evaporação. O primeiro
aspecto é, que colisões elásticas com átomos do vapor residual na câmara de vácuo limitam
o tempo de vida da armadilha. Portanto, a evaporação precisa ser rápido, o que requer uma
alta taxa de colisões elásticas ou um bom vácuo. É preciso encontrar um compromisso entre um
resfriamento evaporativo lente e eficiente e a minimização de perdas devido a um tempo de evap-
oração muito longo. O segundo aspecto é, que a dimensionalidade da superfı́cie de evaporação
determina a eficácia do resfriamento. Na primeira demonstração de evaporação átomos H↑ de
uma nuvem quente foram ejetados por um ponto de sela. A sela fui uma pequena região afastada
do origem da armadilha, e apenas átomos com energia cinética suficiente em uma certa direção,
Ez > Uedge , pudiam deixar a armadilha. Neste caso, a evaporação é chamada de 1-dimensional.
Mesmo se redistribuição ergódica devido à anarmonicidades do potencial conduzem, mais cedo
ou mais tarde, todo os átomos por esta região, este efeito se torna menos pronunciado quando
a nuvem esfria, uma vez que os átomos se acumulam no fundo aproximadamente harmônico (e
portanto separável) do potencial. Este fato tem inibida uma evaporação eficiente em H↑ abaixo
de 120 µK [89].
Uma segunda técnica de evaporação foi demonstrada em armadilhas chamadas time-orbiting
potential (TOP) [205]. É uma caracterı́stica das armadilhas TOP de exibir uma região espacial
chamada ”cı́rculo fatal” (death-circle) onde átomos de passagem são ejetados. Este cı́rculo fatal
pode agir como uma superfı́cie de evaporação 2-dimensional se o raio do cı́rculo é suficientemente
grande [113]. Sob a influência da gravidade a dimensionalidade é ainda reduzida a 1D [144].
A técnica de evaporação de maior sucesso implementado até agora é baseada num acopla-
mento radiativo de estados confinados e livres. Discutimos essa técnica nas seções seguintes.
Publicações sobre resfriamento evaporativo são [166], [177], [205], [211], [?], [175], [185], [83],
[85], [118], [193], [125], [170], [64], [145], [?], [114], [168], [78], [126]. See ([50], Sec. 3.1.4) para
uma visão geral.
14.4.3.2 Evaporation ramp

A rampa de evaporação ideal é obtida na maneira seguinte. O intervalo de tempo que a ”faca
microonda” precisa estar numa certa frequência para expelir os átomos quentes é,
1
dt ∝ . (14.99)
γcoll
A próxima frequência de evaporação é colocada num valor inferior correspondente à uma fração
da distribuição de Boltzmann,
dνrf ∝ −T . (14.100)
Por consequência,
dνrf
∝ −γcoll T . (14.101)
dt
Para uma armadilha quadripolar, γcoll = σel v̄n0 ∝ T −5/2 , enquanto para uma armadilha
harmônica, γcoll ∝ T −1 . Assumindo sempre equilı́brio térmico, T ∝ νrf , chegamos á ,
dνrf −3/2 dνrf

∝ −νrf resp. ∝ −1 , (14.102)
dt dt
o que finalmente dá,
νrf ∝ αq (tf − t)2/5 resp. νrf = αh (tf − t) . (14.103)
14.4.3.3 Radiative coupling of internal state

A técnica origina-se numa ideia proposta por Pritchard e colaboradores [114], que já tiveram al-
guma experiência com espectroscopia de radiofrequência em átomos neutros magneticamente
presos [168, 118]. A dependência espacial do desdobramento Zeeman é uma caracterı́stica
intrı́nseca de armadilhas magnéticas. A irradiação de uma onda de radiofrequência com uma
determinada frequência acopla subestados Zeeman confinados e livres em uma distância bem
definida do centro da armadilha. Isto dá origem a uma superfı́cie de evaporação 3D, onde os
átomos de passagem podem fazer transições de tipo Landau-Zener e ser expelidos da armadilha.
As vantagens técnicas dessa técnica são substanciais: O potencial de aprisionamento magnético
não precisa ser manipulado, por exemplo, pela criação de um bico, e as margens do potencial
podem ser facilmente controladas pela radiofrequência. Se a evaporação está forçada por uma
redução contı́nua da radiofrequência e se a rampa de evaporação é otimizada, a densidade irá

aumentar assim como a taxa de colisões. A retermalização acelerará e iniciará uma evaporação
auto-acelerada (run-away evaporation). A evaporação por rf foi demonstrado pela primeira vez
por Ketterle e colegas [64].
Outro mecanismo de resfriamento baseado em colisões é o resfriamento simpático. A técnica
foi originalmente usada em armadilhas de ı́ons. Mais tarde ela foi aplicada em átomos neutros
confinados em armadilhas magnéticas. A ideia consiste em levar o gás em contacto térmico
com um gás tampão frio. Em alguns casos, o gás tampão pode ser opticamente ou evaporati-
vamente resfriado. Resfriamento simpático tem sido utilizado em armadilhas magnéticas para
criar condensados duplas [190] e para resfriar férmions até o regime de degeneração quântica
[69].
14.4.3.4 Adiabatic and diabatico limits of rf-induced evaporation

Evaporação induzida por rf pode ser descrita no formalismo do átomo vestido [49], onde os
diferentes estados mF de um átomo com spin F são acoplados à um campo de rf 5 que presumimos
ser linearmente polarizado:
B(t) = Bêrf cos ωt . (14.104)
O elemento da matriz de acoplamento entre os nı́veis, |F, mF i e |F, mF ± 1i é
µB g p
Ω= |Brf × êB | F (F + 1) − mF (mF + 1) , (14.105)
4~
onde g é a fator g atômico e êB é a orientação do campo estático magnético local.
Os potenciais adiabáticos U (r) são obtidos através dos autovalores dos estados vestidos para
o campo magnético local B(r). Na imagem do átomo vestido, consideramos a energia total do
átomo mais o campo de N fótons de radiofrequência. Sem acoplamento, isto significa simples-
mente que N ~ω é adicionado à energias atômicas Zeeman, resultando em um padrão Zeeman
para cada N sendo verticalmente deslocado por N ~ω. Nas posições em que o campo rf está em
ressonância, curvas com ∆N = 1 se cruzam. O acoplamento transforma um cruzamentos em
cruzamento evitado. Isto determina o padrão dos nı́veis de energia adiabáticos [ver Fig. 14.11(b)].
Um átomo em movimento lento permanece na curva do potencial adiabático. Como exem-

plo, vamos supor um átomo no estado hiperfino |F, F i se movendo se afastando do centro da
armadilha. Quando ele chega perto de ressonância, o campo rf mistura este estado com outros
estados mF , o que leva à curva do potencial mais plana. Além do ponto de ressonância, o
estado atômico tem sido transformado adiabaticamente em um estado não-confinado, buscador
de campo forte, e o átomo é repelido da armadilha. Assim, ao atravessar o cruzamento evitado,
o átomo tem emitido 2F fotões rf de maneira estimulada e inverteu a orientação de ambos o
spin do elétron e do núcleo.
Deste jeito a radiofrequência induz uma superfı́cie de potencial adiabático com uma profun-
didade de aproximadamente |mF |~(ω − ω0 ), onde ω0 é a frequência rf ressonante no centro da
armadilha. O processo de evaporação corresponde, então, ao extravasamento dos átomos mais
energéticos fora da armadilha.
Para esta imagem adiabática ser válida, uma condição de adiabaticidade deve ser cumprida.
Esta condição requer que a diferença de energia devido ao cruzamento evitado ser maior do que
a incerteza de energia relacionada com o tempo limitado que um átomo com a velocidade v
passa na região de ressonância. Para um sistema de dois nı́veis acoplados por um elemento de
5
Alternativamente, uma frequência de microondas pode ser utilizada para acoplar nı́veis hiperfinos diferentes.
Figure 14.11: (Code: AO T echniques AdiabaticP otentials.m) (Esquerda) Potenciais devido à

estrutura Zeeman de um átomo no estado fundamental com F = 1. (Direita) Potenciais re-
sultantes de um acoplamento dos potenciais por uma radiação radiofrequente ressonante com a
diferencia dos potenciais na posição 0.7.
matriz V12 e um átomo movendo-se com a velocidade v ao longo do eixo z, a probabilidade de

transição P entre as curvas adiabáticas é dada pela fórmula de Landau-Zener [216],
P = 1 − e−ξ , (14.106)
com ξ = 2π|V12 |2 /~gµB ∂z Bv. O teoria de Landau-Zener é estritamente válida apenas para um
sistema de dois nı́veis, que nós usamos aqui só para uma discussão qualitativa dos dois casos
limitantes.
Para um campo de rf fraco, ξ 1, P é muito menor do que 1, isto é, os átomos permanecem
predominantemente na superfı́cie diabática mostrado na Fig. 14.11(a). A probabilidade para
uma transição de spin (spin flip) é, P ≈ t, o que descreve o limite diabático do resfriamento
evaporativo induzido por rf: Os nı́veis atômicos de energia são quase não-perturbados, os átomos
de esparrinham muitas vezes através da superfı́cie de ressonância, e só depois de 1/P oscilações,
eles fazem um spin-flip do estado hiperfino |F, F i para o |F, F − 1i.
O limite adiabático é claramente a situação ideal para o resfriamento evaporativo. No en-
tanto, o processo de evaporação numa armadilha (com o tempo de oscilação Tosc ) satura em
uma potência rf menor. A condição para saturação é P ≈ Tosc /τel , onde τel é o tempo médio
entre duas colisões. Isso significa, que um átomo energético está evaporado antes que colidir
novamente.
Apenas a componente do campo magnético da radiação rf perpendicular ao campo magnético
de aprisionamento induz spin-flips. Em certas geometrias do potencial de confinamento, por
exemplo a armadilha quadrupolar, o campo magnético cobre o ângulo sólido inteiro. Por con-
sequência, há dois pontos onde o campo de aprisionamento e do campo rf são paralelos e os
elementos da matriz de transição são zero. Numa área em torno destes pontos, o caso de acopla-
mento diabático é realizado, mas na prática, a transição rf pode ser suficientemente saturada de
modo a que esta área é pequeno e não afeta fortemente a eficiência da evaporação.
Note também, que gravitação deforma a superfı́cie do potencial de confinamento e pode
reduzir a eficiência da evaporação [144]. Resolva o Exc. 14.7.4.8.
Figure 14.12: Potencial efetivo devido a modulação espacial rápida.
Figure 14.13: Criação de um buraco repulsivo por luz sintonizada pelo azul de uma transição
atômica.
14.4.4 Sympathetic cooling

O resfriamento evaporativo depende da taxa de colisões elásticas interatômicas. No entanto, ex-
istem gases com comprimento de espalhamento desfavorável, isto é, pequeno ou mesmo negativo.
Enquanto em temperaturas baixas só ocorrem colisões de onda s, estes colisões são proibidos
para gases fermiônicos. Estes espécies não podem ser resfriadas por evaporação, mas existe
uma outra técnica chamada de resfriamento simpatético. Aqui está utilizada uma mistura de
espécies, a primeira sendo resfriada ativamente (por exemplo por evaporação) e a segunda passi-
vamente por colisões elásticas com átomos da primeira espécie. É claro, que só funciona quando
o comprimento de espalhamento interespécies e a relação das massas são adequados.
Seguinte [189] o fator de redução devido a diferencia de massas dos parceiras de uma colisão
é,
4m1 m2
ξ= . (14.107)
(m1 + m2 )2
Em torno de 3/ξ colisões por átomo são requisitas para termalização completa de um gás. Por
exemplo, para a combinação Rb-Li temos 3/ξ = 12.4. A taxa de colisões é,
Z
Γcoll = σ12 v̄ n1 (r)n2 (r)d3 r , (14.108)
onde a velocidade térmica média é,

s
8kB T1 T2
v̄ = + . (14.109)
π m1 m2
14.5. OTHER TRAPS 415
A temperatura instantânea é calculada por,

1 d∆T
γtherm = − , (14.110)
∆T dt
ou para simulações: ∆T (t+dt) = ∆T (t)−∆T (t)γtherm dt. Seguinte [68] a taxa de retermalização
é conectada à taxa de colisão via,

ξ ∆E1→2 ∆E2→1 ξ Γcoll Γcoll
γtherm = + = + . (14.111)
3 N1 kB ∆T N2 kB ∆T 3 N1 N2
Para uma armadilha harmônica isotrópica podemos
derivar soluções analı́ticas. Com os
1 2 2 1 2 2 2 2 2 2
potenciais Vj (r) = 2 mj ωrj r + 2 mj ωzj r = kB Tj r /2r̄j + z /2z̄j as densidades ficam nj (r) =
2 /2r̄ 2 −z 2 /2z̄ 2
n0j e−r j j , e o integral fica [187],
Z
2 2 2 2 2 2 2 2
Γcoll = σ12 v̄n01 n02 e−r /2r̄1 −z /2z̄1 −r /2r̄2 −z /2z̄2 d3 r (14.112)
σ12 v̄n01 n02 (2π)3/2

=q
(x̄−2 −2 −2 −2 −2 −2
1 + x̄2 )(ȳ1 + ȳ2 )(z̄1 + z̄2 )
σ12 v̄N1 N2
= p ,
(2π)3/2 (x̄1 + x̄22 )(ȳ12 + ȳ22 )(z̄12 + z̄22 )
2
colocando N = n0 (2π)3/2 x̄ȳz̄. Definindo m2 ω22 = β 2 m1 ω12 ,
(m1 ω12 )3/2 σ12 v̄N1 N2

Γcoll = . (14.113)
(2πkB )3/2 (T1 + T2 β −2 )3/2
Mais uma vez seguinte [68] a taxa de retermalização fica,
p
ξ(N1 + N2 )(m1 ω12 )3/2 σ12 T1 /m1 + T2 /m2
γtherm = . (14.114)
3π 2 kB (T1 + T2 β −2 )3/2
√
Para uma armadilha anisotrópica pegue, ω1 = ω1x ω1y ω1z a média geométrica das frequências
da armadilha. O comprimento de espalhamento então segue de,
p
|a12 | = σ12 /4π . (14.115)
14.5 Other traps

14.5.1 Ion traps
A carga elétrica de ı́ons permite seu controle eficiente por campos elétricos e magnéticos via forças
de Coulomb e de Lorentz. O controle é tão bom, que foi possı́vel armazenar ı́ons individuais
em dois tipos de armadilhas diferentes. Na chamada armadilha de Penning [204], partı́culas
eletricamente carregadas são sujeitos a um campo eletrostático quadrupolar radialmente atraente
sobreposto por um campo magnetostático axial forçando os ı́ons em orbitais circulares fechados 6
6
Note, que campos puramente eletrostáticos não se prestam para aprisionamento, pois a condição necessária
para a existência de mı́nimos em um potencial, ∂i ∂j φ < 0, não obedece à equação de Laplace. Partı́culas
carregadas em campos elétricos podem, portanto, só por confinado dinamicamente.
Na chamada armadilha de radiofrequência ou armadilha de Paul —o Wolfgang Paul recebeu o

premio Nobel em 1989 junto com o Hans Dehmelt e o Norman Ramsey— os ı́ons são colocados
em um campo elétrico alternante com simetria quadrupolar. Configurações hiperboloides de
eléctrodos produzem potenciais em forma de sela, como mostra a Fig. 14.15. Isto é, em qualquer
instante de tempo, o potencial e parabolicamente repulsivo na direção axial e parabolicamente
atraente em direção radial. Na média temporal o potencial é parabólico, isto é, os ı́ons realizam
oscilações harmônicas com frequências próprias independentes da amplitude de oscilação [201]:
φ(r, z) = φ0 (t)(r2 − 2z 2 ) , r 2 = x2 + y 2 . (14.116)
A polaridade é alternada com radiofrequência:
φ0 = φdc + φac cos(Ωa t) . (14.117)
Ωa /2π denota a frequência do campo aplicado, Vdc a amplitude da parte dc da voltagem, Vac
Figure 14.14: Geometria da armadilha de Paul.
a amplitude da parte ac, e r0 e z0 são os raios equatorial e polar da armadilha. O potencial

φ(r, z, t) exerce, na média temporal, uma força central sobre o ı́on, se o campo de radiofrequência
satisfaz condições especificas.
14.5.1.1 Evaluation of the stability diagram

A armadilha de Paul não necessariamente tem geometria quadrupolar. Para determinar as
frequências seculares do pseudo-potencial, procedemos da seguinte maneira: A posição r0 do
mı́nimo do potencial φ depende da geometria dos eletrodos e das tensões aplicadas. Quando
expandimos o potencial em torno do mı́nimo,
φ(r) = φa + (r − r0 )∇φ(r0 ) + 21 [(r − r0 )∇]2 φ(r0 ) + ... (14.118)

2 2
≡ φa [1 + br (r − r0 ) + bz (z − z0 ) ] .
No último passo, assumimos que o potencial tem forma quase cilı́ndrica. Para uma dada geome-
tria As curvaturas bz,r podem ser extraı́das de simulações numéricas. A partir da equação de
continuidade, encontramos bz = −2br . A polaridade dos electrodos é modulada com a frequência
Ω,
φ(r, t) = φ(r)(ζ − cos Ωt) . (14.119)
As equações de movimento são derivadas de mr̈ = −e∇φ(r, t),
mr̈j + 2eφa bj (ζ − cos Ωt)rj = 0 . (14.120)
Introduzindo os parâmetros a e q,
8eφa bz ζ 4eφa bz
az = = −2ar e qz = = −2qr , (14.121)
mΩ2 mΩ2
chegamos à chamada equação de Mathieu [173, 86],
r̈j + 41 Ω2 (aj − 2qj ζ cos Ωt)rj = 0 . (14.122)
Essas equações predizem órbitas estáveis, desde que os parâmetros a e q estejam dentro do
chamado diagrama de estabilidade mostrado em Fig. 14.16.
φ(r,z) (u.a.)
0
−2
10
10
0 0
z (mm) −10 −10 r (mm)
Figure 14.15: (Code: AO T echniques IonsSaddleP ot.m) Ilustração dois-dimensional do poten-

cial dependente do tempo: em cada instante temporal o potencial tem a forma de uma sela. O
potencial gira em torno do eixo vertical com uma velocidade apropriada.
De acordo com essas equações, o ı́on passa por movimentos oscilatórios que são definidos
pelos parâmetros da armadilha ai e qi . Para que o movimento do ı́on seja finito, a sua amplitude
de oscilação pode não ultrapassar o limite definido pelos elétrodos. Esta condição impõe um
regimes possı́vel para os parâmetros da armadilha chamado diagrama de estabilidade [173].
0.2
-0.2
ar
-0.4
-0.6
0 0.5 1
az
Figure 14.16: (Code: AO T echniques IonsStabilityDiagram.m) Diagrama de estabilidade.
No limite |ai |, qi 1 o ı́on se desloca só de uma pequena distância s r0 durante um perı́odo
de modulação Ωa . Então o ı́on sofre um movimento lento periódico chamado de macromovimento
dentro do potencial de aprisionamento com frequência secular ζi . Este movimento é modulado
por uma oscilação rápida chamada de micromovimento excitada pelo campo de modulação Ωa .
Sem voltagem dc aplicada entre o anel é as tampas, ai = 0, o movimento do ı́on é descrito pela
seguinte equação simples:

ri (t) = ri0 1 − 21 qi cos Ωa t cos ζi t , ζi = √1 qi Ωa
8
, i = r, z . (14.123)
A órbita do ı́on está confinada à região interna da armadilha, se sua energia cinética é inferior a
mζr2 r02 + M ζz2 z02 . Uma vez que a armadilha é, a qualquer momento, focando em algumas direções
e desfocando em outra, não se trata de um potencial conservador. O movimento oscilatório
(desconsiderando o micromovimento) do ı́on, no entanto sugere um modelo, onde a armadilha é

descrita por um pseudo-potencial [86, ?] cuja profundidade é,
qz
Dz = 8 eVac = 2Dr se ai = 0 . (14.124)
motion of the ion red: x green: y blue: z

2
5
y (µm)
r (µm)
0 0
−5
−2
−4 −2 0 2 0.2 0.4 0.6
x (µm)
1
Figure 14.17: (Code: AO T echniques IonsP aulT rapSimul.m) Movimento micro e macro do
v (cm/s)
ı́on. 0
−1
Outras geometrias para os eléctrodos, desviando do quadrupolo perfeito, também0.2 são
0.4 possı́veis.
0.6
Estas armadilhas também são bem descritas pela equação (14.116), quando o ı́on fica perto do
40
centro. Para nuvens de ı́ons, por exemplo, armadilhas multipolares de ordem superior [252],
eF (eV/cm)
20
armadilhas Paul-Straubl [225] e anéis de armazenamento [249] têm sido utilizados. Particu-
0
larmente importante para o armazenamento de muitos ı́ons resfriados −20
com aplicações na com-
putação quântica é a chamada armadilha de Paul linear [209, 88,−40 208]. Aqui os ı́ons imóveis
estão alinhadas numa cadeia linear. A vantagem da armadilha linear, em comparação0.2 0.4com out-0.6
t + 0 µs
ras armadilhas de muitas partı́culas, é o acesso óptico mais fácil de ı́ons individuais por lasers e
a possibilidade de zerar o micromovimento.
14.5.2 Micromotion
O movimento do ı́on na armadilha de Paul é uma superposição de duas vibrações com as
frequências de oscilação respectivas Ωa (frequência da modulação) e ζr,z (frequências seculares
para as vibrações em direção radial e axial). Para o ı́on em equilı́brio térmico, isto é não resfriado,
as energias cinéticas médias do micro- e do macromovimento são iguais [29].
O macromovimento pode ser reduzido por resfriamento ao contrário do micromovimento é
que constantemente excitado pelo modulação do campos elétrico aplicado [47]. Do outro lado,
a amplitude do micromovimento diminui com a distância do ı́on do centro da armadilha e, no
mı́nimo do pseudo-potencial desaparece totalmente. Portanto, para suprimir o micromovimento,
é impreterı́vel resfriar o macromovimento e pular o átomo para o centro, se necessário, usando
campos elétricos adicionais. Como a frequência do micromovimento é muito mais elevada do
que aquela do macromovimento, as bandas laterais dinâmicas podem ser facilmente resolvidas.
Quando a frequência de modulação Ωa é muito alta, as frequências seculares do macromovimento
também são altas. Neste caso, até as bandas laterais do macromovimento podem ser resolvidas
por transições atômicas dipolares largas. Este é chamado regime de acoplamento forte.
Devido a repulsão coulombiana só um único átomo pode ficar no centro da armadilha de Paul,
tal que é difı́cil zerar o micromovimento. Uma solução consiste em usar armadilhas lineares,
onde o centro, onde o micromovimento zera é estirado numa linha.
Outras geometrias para os eletrodos também são possı́veis. Para nuvens iônicas e.g. potenci-
ais multipolar de ordem superior [252], armadilhas de Paul-Straubl [225], e potenciais toroidais
[249] tem sido usados. Particularmente importante para o aprisionamento de vários ı́ons resfri-
ados é a chamada armadilha de Paul linear [209, 88, 208]. Aqui os ı́ons imóveis estão alinhados
em linha reta. A vantagem da armadilha linear, comparada a outras armadilhas de muitas

partı́culas, consiste em sua melhor geometria de coleta de fluorescência e menor micromovi-
mento. Resolva o Exc. 14.7.5.1.
Example 55 (Cálculo numérico do campo elétrico criado por uma superfı́cie car-
regada): Para calcular o potencial de aprisionamento de uma partı́cula carregada acima
de uma estrutura de microtrap planar, procedemos da seguinte maneira. A energia de uma
carga num campo elétrico é H = −eφ. O potencial eletrostático é dado pela lei de Coulomb,
Z Z Z
1 X ρ(r0 ) 0 1 X r − r0 0 1 X E(r0 ) 0
φ(r) = dV − φ n df + df ,
4π0 n Vn |r − r0 | 4π n Sn |r − r0 |3 4π n Sn |r − r0 |
where φn is the voltage applied to the n-th boundary. In practice, electric fields are generated
by electrodes set to specific voltages. Using the Dirichlet boundary conditions, we only retain
the second term. Furthermore, to account for the planar geometry of the chip electrodes, we
only consider surface boundaries in the y 0 = 0 plane,
Z
1 X ydx0 dz 0
φ(r) = − φn p 3 .
4π n Sn (x − x0 )2 + y 2 + (z − z 0 )2
This implies that the field lines cross the chip surface orthogonally, which in reality is only
true if the chip electrodes cover the whole area. Therefore, we only consider small gaps
f1
10 mm
f2
f3
f4
frf
Figure 14.18: Possible design for a microchip ion trap. φn are static potentials except for φ0 ,
which is alternates sign with radio frequency.
between the electrodes. We digitize the integral by dividing every electrode φn into a number
of identical surface elements ∆fm ,
1 X y∆fm
φ(r) = − φn p 3 .
4π n,m (x − xm )2 + y 2 + (z − zm )2
This formula can easily be evaluated numerically. A concrete example for a microchip ion
trap is shown in Fig. 14.18.
Example 56 (Cálculo numérico do campo magnético criado por um fio de cor-

rente): Current-carrying wires may exert Lorentz forces on the ions. The magnetostatic
field is given by the Biot-Savart law,
Z
µ0 (r − r0 ) × j 0
B(r) = dV .
4π V |r − r0 |3
In practice magnetic fields are created by current-carrying wires. Those can be parametrized
Figure 14.19: (Code: AO T echniques IonsBecCalcs.m) Two-dimensional cuts through the elec-
tric potential generated by the microchip shown in Fig. 14.18 for φrf = 100 V and φj = 0.
by one-dimensional currents, j =Iδ 2 ds, so that,

Z
µ0 I ds0 × (r − r0 )
B(r) =
4π C |r − r0 |3
q
2 2 2 2 2 2
µ0 I X dsy,n (z − zn ) + dsz,n (x − xn ) + dsx,n (y − yn )
|B(r)| = p 3 .
4π n (x − xn )2 + (y − yn )2 + (z − zn )2
can immediately be numerically solved.
14.5.2.1 Electronic detection of ions

A presença de ı́ons na armadilha pode ser sondada através do amortecimento que induzem num
circuito de ressonância eletrônica acoplado [251, 253].
14.5.3 QUEST
Átomos e dı́meros homonucleares não possuem momento dipolar elétrico permanente, mas po-
dem ter momento dipolar magnético permanente. Por isso, dı́meros homonucleares devem ser
confinados por gradientes de campo magnético ou um momento de dipolo elétrico deve ser in-
duzido por um campo eletromagnético oscilante. No regime óptico, isso foi demonstrado com a
armadilha quase eletrostática (QUEST).
Em contraste, dı́meros heteronucleares são moléculas polares com um momento dipolar
elétrico permanente, que pode ser bastante grande se eles estiverem profundamente ligados.
De acordo com o teorema de Earnshaw, não tem máxima de campo magnético ou de campo
magnético estático em espaço livre. Assim, nenhum estado ’high-field seeking’ pode ficar preso.
Em princı́pio, armadilhas ópticas dipolares tipo QUEST também podem ser usadas para dı́meros
heteronucleares. Mas o problema é que, em contraste com as moléculas homonucleares, são
possı́veis transições entre os estados fundamentais da vibração. Assim, a luz criando a QUEST
induz transições redistribuindo os ı́ons sobre todos os estados vibracionais.
14.6. ANALYSING TECHNIQUES 421
rf
Figure 14.20: Circuito de ressonância para detecção eletrônica do ı́ons. A armadilha de Paul
é operada por uma radiofrequência enquanto uma voltagem contı́nua é escaneada através do
diagrama de estabilidade. Simultaneamente, um oscilador é sintonizado perto de uma das
frequências seculares da armadilha. Quando o ponto de estabilidade é tal que a frequência sec-
ular coincide com a frequência do oscilador, o movimento dos ı́ons é parametricamente excitado
e o circuito de ressonância é amortecido. Este amortecimento é detectado por um amplificador
de banda estreita.
Muito longe de ressonância,

I(r)
Udip (r) = −αstat . (14.125)
2ε0 c
Moléculas homonucleares fracamente ligadas são sujeitos à soma das forças restauradoras de
armadilhas magnéticas ou dipolo agindo sobre os átomos individuais, µm = 2µa e dm = 2da .
Isso vale para moléculas heteronucleares, desde que o potencial de aprisionamento seja muito
mais fraco que a energia de ligação.
Example 57 (Momento dipolar elétrico permanente do LiRb): A energia de interação

dois dipolos é,
1 p1 · p2 − 3(p1 · r̂)(p2 · r̂)
Ĥint = .
4πε0 r3
Assim, dois dipolos idênticos com 1 Debye = 10−27 /2.998 Cm= 10−19 /c Cm2 /s= 39.36 eaB
paralelamente orientados numa distancia r = 1 µm tem a energia,
1 p2
Ĥint = ≈ h × 1.5 MHz ≈ kB × 73 µK .
4πε0 r3
Por exemplo, LiRb tem o momento dipolar elétrico -2 to-4.2 Debye.
14.6 Analysing techniques

Para analisar o estado cinético de um gás e, por exemplo, identificar a presencia de um con-
densado de Bose-Einstein, é preciso medir as distribuições espaciais ou de momento. Para isso,
devemos jogar partı́culas para o gás e medir o espalhamento delas. A partı́cula mais adequada
para penetrar uma câmara de vácuo ultra-alto é o fóton. Por isso, com poucas exceções onde
são utilizados elétrons, todas informações sobre gases ultrafrios temos através de suas reações
com feixes de lasers [129, 41, 7, 140, 98].
14.6.1 Time-of-flight imaging

As técnicas de imagem mais comuns medem a absorção de um feixe laser pela nuvem atômica
depois de um tempo de voo ou a dispersão dele pela in-situ. A amplitude E0 de uma onda de
luz atravessando uma nuvem atômica de diâmetro L com o ı́ndice de refração η é modificada
por um fator eiωLη/c . Para uma nuvem inomogênea, temos
Z ∞
iωL/c ω
E = E0 e exp i (η(r) − 1) dz . (14.126)
c −∞
Podemos aproximar o ı́ndice de refração pela susceptibilidade atômica,

p 1
η= 1+χ≈1+ χ . (14.127)
2
A parte imaginária da susceptibilidade é relacionada ao coeficiente de absorção α e a parte real
ao coeficiente de dispersão δ,
α 2δ
Im χ = e Re χ = . (14.128)
ω/c ω/c
Agora, os coeficientes de absorção e de dispersão podem ser relacionados com a secção transversal
óptica σ(∆) [164], onde ∆ é a dessintonia entre a frequência da luz e uma ressonância atômica,
cuja largura de linha é Γ. Esse resultado se chama teorema óptico,
∆
α = nσ(∆) e δ = nσ(∆) , (14.129)
Γ
onde n e a densidade da nuvem. Finalmente, obtém-se a lei de Lambert-Beer,
Z ∞
i ∆
E = E0 e iωL/c
exp iσ(∆) − n(r)dz ≡ E0 eiωL/c e−α/2 eiδ , (14.130)
2 Γ −∞
com a distribuição de densidade n(r). Considerando a intensidade, I ∝ |E|2 , temos,

Z ∞
I 2i∆
= exp −σ(∆) 1 + n(r)dz ≡ e−α e2iδ . (14.131)
I0 Γ −∞
A absorção α descreve as perdas de intensidade para o feixe laser por espalhamento por
átomos desordenados. Ela é forte perto de ressonância, mas diminuı́ quadraticamente com a
dessintonia ∆. Ela dá origem à força de pressão radiativa. A dispersão δ descreve a refração
do feixe por uma distribuição de densidade continua [51, 12]. Ela desaparece em ressonância é
diminuı́ lentamente (∝ ∆) com a dessintonia. Ela está ligada à força dipolar. O coeficiente δ
descreve o deslocamento de fase da onda eletromagnética transmitida através da nuvem atômica.
14.6.2 Absorption imaging

Agora resumimos o processo experimental de imagem por absorção (vide Fig. 14.21): A ar-
madilha é subitamente desligada, assim deixando a nuvem, acelerada pela gravitação terrestre,
cair por um tempo de vôo de alguns ms. Em seguida, um pulso de um feixe laser ressonante, cujo
diâmetro é muito maior do que o tamanho da nuvem, é irradiado. A atenuação da intensidade
Figure 14.21: Sequência temporal da expansão balı́stica. 18 ms após desligar a armadilha, a

nuvem atômica é iluminada por uma pulso laser ressonante. A sombra imprimida pela nuvem
sobre o feixe é fotografada por uma câmera CCD.
do feixe I ∼ |E|2 pode ser relacionada através do coeficiente de absorção α (também chamado
de densidade óptica) à densidade atômica:
Z
I(x, y)
− ln = α(x, y) = σ(∆) n(r)dz . (14.132)
I0
A sombra impressa pela nuvem atômica sobre o perfil transversal do feixe laser é registrada
numa câmera CCD.
Já notamos, que a absorção está acompanhada de pressão radiativa. Após alguns eventos
de espalhamento, os átomos tem acumulado momento por recuo fotônico suficiente para ficar
fora de ressonância devido ao efeito Doppler. Fótons chegando aos átomos subsequentemente
não são mais espalhados e só aumentam a iluminação da CCD sem trazer informação sobre a
distribuição dos átomos. Consequentemente, é vantajoso usar pulsos de laser bem curtos. Além
disso, a intensidade do feixe laser não deve saturar a transição a fim de garantir uma secção
transversal óptica independente da intensidade e garantir a validade da lei de Lambert-Beer.
Finalmente, a frequência do laser deve ser sintonizada perfeitamente para ressonância, ∆ = 0.
De outro modo, acontece dispersão levando à focalização ou defocalização do feixe por refração.
Isso distorce a imagem e vira impossı́vel a estimação do tamanho da nuvem.
A Fig. 14.22 mostra exemplos de imagens de absorção de uma nuvem atômica tomados
em fases diferentes do processo de evaporação. A Fig. 14.22(a,b) fui tomada na temper-
atura 320 nK; a nuvem é grande e isótropo, e portanto, puramente térmica. Em 250 nK [ver
Fig. 14.22(c,d)] uma parte de forma elı́ptica aparece no centro da nuvem térmica. E em 180 nK
[ver Fig. 14.22(e,f)] a nuvem térmica desapareceu quase completamente para o benefı́cio do con-
densado. Uma avaliação quantitativa da fração condensada é dada na Sec. 15.1.4. Resolva o
Exc. 14.7.6.1.
Figure 14.22: Imagens de absorção após tempo de vôo permitem identificar a presencia de um
condensado de Bose através da sua distribuição de momento caracterı́stica. Mostrado são uma
foto acima (a), ligeiramente abaixo (b), e muito abaixo da transição de fase (c) (figuras de [113]).
14.6.3 Dispersive imaging

Por causa da expansão balı́stica, e também devido à pressão radiativa, que leva uma aceleração
e aquecimento da nuvem, a técnica de imagem por absorção é destrutiva. Isto é, o processo
de medida transforma as distribuições da nuvem tal, que uma segunda imagem tomada depois
da primeira dá resultados diferentes. No entanto, existe uma técnica de imagem não-destrutiva
chamada de dispersiva ou de imagem por contraste de fase. Nesta técnica, a luz laser está sin-
tonizada suficientemente longe de ressonância, |∆| Γ, para emissão espontânea e aquecimento
induzido por recuo fotônico aleatório são desprezı́veis [7]. Isto permite, por exemplo, tomar
uma série de fotografias consecutivas e fazer um filme da evolução temporal da nuvem. Outra
vantagem desta técnica está devida à pequena densidade óptica fora de ressonância, pois isso
permite tomar fotografias de nuvens muito densas, em particular, enquanto eles estão confinadas
dentro de uma armadilha, isto é in situ.
A quantidade medida neste método é o deslocamento da fase local da frente da onda laser
de prova. Para transformar o perfil de fase em um perfil de intensidade, utiliza-se um método
clássico baseado na interferência do feixe prova com o perfil de fase distorcida e uma onda plana
de referência. Na prática, existem várias possibilidades. Para imagens de fundo escuro (dark-
ground imaging), uma parte do feixe incidente (isto é, do feixe antes da sua interação com os
Figure 14.23: Esquema para imagens dispersivas.
átomos) é bloqueada por trás da zona de interação (vide Fig. 14.23)
2 ∆2
α→0
I¯dg = 12 |E − E0 |2 = I0 e−α/2+iδ − 1 −→ I0 δ 2 = I0 α 2 . (14.133)
Γ
O sinal de intensidade I¯dg é quadrático na densidade óptica α.

Para imagens de contraste de fase, o feixe original recebe um deslocamento de fase de λ/4 à
respeito da parte do feixe tendo sido interagido com os átomos:
2
¯ 1 ±iπ/2 2 −α/2+iδ ±iπ/2 α→0 2 ∆
Ipc = 2 |E − E0 + E0 e | = I0 e −1+e −→ I0 (±1 + δ) ' I0 1 ± α .
Γ
(14.134)
A intensidade I¯pc é linear em α, e desde então mais sensı́vel a sinais fracos. A terceira técnica,
chamada imagens de contraste de polarização, detecta a birefringência local da nuvem atômica
[34].
As técnicas de imagem mostrada até agora só permite visualizar a distribuição da densidade
instantânea da nuvem atômica n(r). Se estiver interessado em outras propriedades, temos a
concepção do experimento, de tal maneira que a informação desejada aparece na distribuição de
densidade. Por exemplo, para medir as frequências de excitação de um condensado, que pode
perturbar a sua forma e observar a evolução temporal, oscilatório n(r, t) por imagem dispersiva
[142, 176, 8, 146].
14.6.4 Reconstrução de column-integrated absorption images

p
Assume cylindrical symmetry n(r) = n(r, z), with r = x2 + y 2 . Absorption images are column-
integrated, i.e. they are taken by integration along the x-axis,
I(y, z) R
= e−σ n(r,z)dx = e−σf (y,z) . (14.135)
I0 (y, z)
The radial density can be recovered by tomography [67, 79, 199],

Z
1
n(r, z) = (Fy f )(κy , z)J0 (κy r)dκy . (14.136)
(2π)2
This is called image reconstruction or Fourier reconstruction or inverse Abel transform.

14.6.4.1 Programas on inverse Abel transformation

• MW AbelTrafo1: Inverse Abel transform using Bessel of a real Fermi sphere.
• MW AbelTrafo2: Inverse Abel transform using Bessel of an arbitrary function in 2D.
• MW AbelTrafo3: Inverse Abel transform .
• MW AbelTrafo4: Forward Abel transform, comparison with column integration and ana-
lytical formula.
• MW AbelTrafo5: Forward Abel transform, comparison with column integration and ana-
lytical formula in 2D.
14.6.5 Condensable atomic species

Early work on BEC comes from [223, 19, 28, 44, 162]. Proposals for atomic gases come from
[117, 240, 239]. An appropriate BEC candidate must fulfill a few conditions: The transition
wavelengths must be accessible by laser light, the level scheme should exhibit a closed cycling
transition for laser cooling and have a reasonable pressure in gas phase. Furthermore, it is
desirable to have a large HFS, metastable electronic state, no trapping state, large positive
scattering length, Feshbach resonances. For sympathetic cooling it may be nice to have several
isotopes of the same element.
The most common elements are alkalis, alkali earths and noble gases. The following gases
have already been condensed 1 H, 1 He∗ , 7 Li, 23 Na, 85 Rb, and 87 Rb [6], [63], [36], [35], [218], [54],
[113], [116]. Investigations in 1 Ne∗ , 39 K, 133 Cs, x Sr, x Cr and 40 Ca are underway [232], [107],
[150], [210], [230], [23], [89].
14.7 Exercises
14.7.1 The atomic motion
14.7.1.1 Ex: Limite de recuo
Calcule o limite de Doppler, o limite de recuo e o limiar da degenerescência quântica para uma
densidade de n = 1014 cm−3 para a transição de sódio D2 (λ = 590 nm, Γ/2π = 10 MHz) e a
transição de rubı́dio D2 (λ = 780 nm, Γ/2π = 6 MHz).
14.7.1.2 Ex: Densidade de estados para potenciais não-harmônicos

x p y l z q
Calcule a densidade de estados para potenciais não-harmônicos, H = ~2m
2 k2
+ 2x̄ + +
2ȳ 2z̄
usando a Ref. [13]. Aplica o resultado para um potencial quadrupolar.
14.7.2 Optical cooling

14.7.2.1 Ex: Potencial de confinamento
Considere um laser mono-modo (TEM00 ) focalizado em uma nuvem de átomos frios de Na à
uma temperatura de 450 µK. Para uma dessintonia de 600 MHz e um diâmetro de ponto focal
de 10 µm calcule a intensidade do laser (W/cm2 ) necessária para produzir um poço de potencial
suficiente para conter os átomos. O momento de transição (a.u.) de Na é 2.55.
14.7. EXERCISES 427
14.7.2.2 Ex: Melasses ópticas

Melasses ópticas são criadas (numa dimensão) por dois feixes lasers contrapropagantes sin-
tonizadas para o vermelho. Cada um dos feixes de laser exerce sobre os átomos a força de
pressão radiativa F± = ~k Γ2 [2(∆±kv)/Γ]
s
2
+1+s
. ∆ é a sintonização do laser, ν é a velocidade de
um átomo.
a. Mostre que para pequenas velocidades (|kv| Γ e ∆ ≤ Γ) a melasse óptica pode ser enten-
dida como força de fricção e calcule o coeficiente de fricção.
b. Processos de aquecimento por emissão espontânea limitem a temperatura mı́nima que pode
ser alcançada em melasses ópticas. Calcule a sintonização do laser onde a temperatura alcance
o valor mı́nimo e especifica o limite de resfriamento.
Ajuda: Supõe uma melasse umdimensional e assume, que a emissão espontânea só acontece ao
longo dessa dimensão. A taxa de aquecimento segue da taxa de espalhamento R através de
dE
d hp2 i ~2 k2 dE

dt heat = dt 2m = 2m 2R, a taxa de resfriamento segue de dt cool = F v.
14.7.2.3 Ex: Chafariz atômico

Em chafarizes atômicos átomos são acelerados fora de uma melasse óptica (λ = 780 nm) para
cima por dois dos feixes laser da melasse orientados por cima num ângulo de 45◦ . Esses feixes
são sintonizados pelo azul e os feixes contrapropagantes com a mesma dessintonização pelo lado
vermelho da ressonância. Qual deve ser a sintonia para os átomos acelerados dentro do resson-
ador de microondas da fonte dentro do campo de gravitação num tempo de 1 s?
Ajuda: Suponha que o ressonador seja perto da posição da melasse e com comprimento negli-
genciável.
14.7.3 Optical and magneto-optical traps

14.7.3.1 Ex: Potencial quadripolar
Mostre que para uma armadilha quadripolar sempre vale 2∂r Bqua = ∂z Bqua .
14.7.3.2 Ex: Linearização da MOT

Derive o coeficiente de atrito e a constante de mola para uma MOT.
14.7.3.3 Ex: Armadilha dipolar perto de uma linha de intercombinação

a. O estrôncio tem uma transição forte (Γ461 = (2π) 30.5 kHz) em 461 nm e uma ressonância fraca
da transição de intercombinação (Γ689 = (2π) 7.6 kHz) em 689 nm. Um feixe laser gaussiano
com a potência P = 10 mW focalizado para uma cintura de w0 = 100 µm é sintonizado
∆689 = −(2π) 10 GHz abaixo da transição de intercombinação. Calcule a profundidade do
potencial e as frequências de vibração para átomos aprisionados pelo feixe laser considerando
ambas as ressonâncias. Qual é a taxa de espalhamento pelas duas transições.
b. Supõe que a nuvem atômica aprisionada consiste em N = 108 átomos na temperatura T =
10 µK. Calcule a densidade atômica n0 no centro da nuvem.
14.7.3.4 Ex: Armadilha dipolar com um feixe focalizado

a. Calcule as frequências de vibração para átomos aprisionados de 87 Rb para uma armadilha
óptica que consiste de um feixe laser focalizado com a potência P = 10 W e o diâmetro do feixe
w0 = 100 µm. O feixe de laser seja dessintonizado 5 nm ao lado vermelho da ressonância de

rubı́dio D1 localizada em λ = 795 nm.
b. Supõe que a nuvem atômica aprisionada consiste em N = 108 átomos na temperatura
T = 100 µK. Calcule para os dois tipos de armadilhas a densidade atômica n0 no centro da
nuvem.
c. A seção cruzada para colisões elásticas seja σ = 10−12 cm2 . Quantas vezes os átomos se
encontram em médio dentro dos dois tipos de armadilhas?
14.7.3.5 Ex: Armadilha dipolar com um feixe focalizado para Na
Considere um laser mono-modo (TEM00 ) focalizado em uma nuvem de átomos frios de Na à

uma temperatura de 450 µK. Para uma dessintonia de 600 MHz e um diâmetro de ponto focal
de 10 µm calcule a intensidade do laser (W/cm2 ) necessária para produzir um poço de potencial
suficiente para conter os átomos. O momento de transição (a.u.) de Na é 2.55.
14.7.3.6 Ex: Rede óptica
Um feixe laser com λdip = 1064 nm, P = 2 W, e w0 = 50 µm e subdividido em três feixes

retrorefletidos se cruzando com ângulos retos formando uma rede óptica cúbica para estrôncio
cuja transição relevante fica em λSr = 461 nm com ΓSr = (2π) 32 MHz. Calcule a profundidade
do potencial e as frequências seculares.
14.7.4 Magnetic traps

14.7.4.1 Ex: Inexistência de armadilha para buscador de campo forte
Mostre que não é possı́vel criar máxima de intensidade com um potencial magnético.
14.7.4.2 Ex: Armadilha harmônica
Calcule as frequências de vibração para átomos aprisionados de 87 Rb para uma armadilha

harmônica com os átomos no estado |F = 1, mF = −1i com o fator gF = 1/2.
14.7.4.3 Ex: Armadilha magnética quadrupolar
a. Considere átomos de 87 Rb aprisionados em uma armadilha magnética com B(x, y, z) =

(x, y, −2z) × 200 G/cm. Os átomos são no estado |F = 1, mF = −1i com o fator gF = 1/2.
Verifica se é razoável assumir frequências de vibração constantes para tais armadilhas.
b. Supõe que a nuvem atômica aprisionada consiste em N = 108 átomos na temperatura
T = 100 µK. Calcule a densidade atômica n0 no centro da nuvem.
c. A seção cruzada para colisões elásticas seja σ = 10−12 cm2 . Quantas vezes os átomos se
encontram em médio dentro da armadilha?
14.7.4.4 Ex: Inclinação por gravitação
Considere (a) uma armadilha quadrupolar, (b) uma armadilha harmônica isotrôpica. Qual é o
gradiente, respectivamente curvatura necessário para levantar uma nuvem de rubı́dio sujeito à
gravitação? Qual é a inclinação da nuvem dentro do potencial devido à gravitação?
14.7. EXERCISES 429
14.7.4.5 Ex: Compressão adiabática
Como varia a temperatura quando está comprimida adiabaticamente (a) uma armadilha quadrupo-
lar, (b) uma armadilha harmônica. Como variam a densidade, a densidade do espaço de fase,
a taxa de colisões elásticas. Ajuda: Define a compressão para armadilha quadrupolar como
η ≡ ∂r Br,f inal /∂r Br,initial e para armadilha harmônica como η ≡ ωr,f inal /ωr,initial .
14.7.4.6 Ex: Armadilha TOP
A armadilha TOP (time-orbiting potential trap) fui a primeira sucedida para criação de conden-
sados de Bose. Ela consiste da superposição de um campo magnético quadrupolar (gradiente
2∂r Bqua = ∂z Bqua ) e um campo magnético homogêneo rotatório no plano de simetria do campo
quadrupolar, Btop . Átomos oscilando com uma amplitude além de um dado raio rd , chamado
do ’circulo da morte’, fazem transições de Majorana e são expulsados da armadilha. Calcule o
raio da morte.
14.7.4.7 Ex: Potenciais ópticos anulares
An interesting system is the 1D array of annular optical potentials realized in a standing wave
formed by red-detuned Gaussian beam and a counterpropagating blue-detuned doonat-mode. In
general, the tight longitudinal confinement freezes out the axial motion by quantum confinement.
It can be readily shown [257] that in the far-off resonance case and if the potential is approximate
by a harmonic potential around its minimum the eigenenergy spectrum is given by,
~2 p2
Epq = U0 + ~ω(q + 12 ) + .
2mR02
It thus reproduces the ro-vibrational spectrum of a 2D artificial molecule and gives rise to two
normal motions. In its ground state, we have the atom optical analog of a 2D rigid rotator.
Gravity plays formally the same role as static electric fields for molecules. Such systems might
be interesting for investigating the selection rules for transitions between ro-vibrational states
involving conservation of total angular momentum of light and atoms and yield insight into the
concept of orbital angular momentum of light fields.
14.7.4.8 Ex: Transições Landau-Zener
Considere uma nuvem de rubı́dio no estado fundamental 2 S1/2 , F = 1, mF = −1 confinado

num potencial isotrópico quadrupolar com o gradiente 200 G/cm. Para iniciar uma evaporação
eficiente por radiofrequência, você quer, que átomos atravessando o limite onde a radiofrequência
acopla os estados Zeeman fazem uma transição com a probabilidade de 95%. Qual é a amplitude
do campo magnético necessário.
14.7.4.9 Ex: Amortecimento em misturas de espécies
Como descrever o amortecimento em misturas de espécies e como utilizar uma medida do tempo
de amortecimento para determinação do comprimento de espalhamento interspecies?
14.7.5 Others traps

14.7.5.1 Ex: Repulsão coulombiana na armadilha de Paul linear
A repulsão coulombiana na armadilha de Paul linear impede, que dois ı́ons ficam simultanea-
mente no estado fundamental. Determine a extensão espacial do estado fundamental e a pro-
fundidade do potencial na aproximação do pseudo-potencial. Qual é a distância de equilı́brio
dos ı́ons?
14.7.6 Analysing techniques

14.7.6.1 Ex: Densidade óptica
Uma nuvem de N = 106 átomos de 87 Rb está preparada em uma armadilha harmônica cilı́ndrica
caracterizada pelas frequências de vibração axial ωz = (2π) 50 Hz é radial ωr = (2π) 200 Hz. O
experimentador faz a imagem de absorção após 18 ms tempo de vôo mostrada em Fig. 14.22(a).
Um pixel da câmera CCD corresponde à 5 µm em espaço real.
a. Em qual temperatura espera-se a transição de fase para condensado de Bose-Einstein?
b. Determine a temperatura da amostra.
c. Avalia densidade.
d. Avalia densidade óptica ressonante para a transição D2 em 780 nm ao longo do eixo de
simetria da nuvem aprisionada.
Chapter 15
Thermodynamics of quantum gases

The journey of the quest for Bose-Einstein condensation (BEC) begins with its prediction by
Bose and Einstein in 1926. The first hint, that the condensation was more than just a theoretical
fantasy came from London [163], who linked the newly discovered phenomenon of superfluidity
in 4 He to BEC. However, the interpretation of the λ-point in terms of BEC was not obvious,
because strong interactions between particles concealed the role of quantum statistics, and the
thermodynamic potentials exhibited divergences at the critical temperature instead of discon-
tinuities, as expected for an ideal gas BEC. These uncertainties triggered an intense search for
other systems. In 1954, Schafroth pointed out that electron pairs can be seen as composite
bosons and may form Bose-Einstein condensates at low temperatures [223]. In 1957, Bardeen,
Cooper and Schrieffer developed the microscopic theory of superconductivity [19], after other
researchers, including Blutt, Schaffrot, Fröhlich and Bogolubov, had suggested a relationship of
this phenomenon to Bose condensation of electron pairs (nowadays called Cooper pairs).
Figure 15.1: Illustration of atomic Broglie waves. From the top to the bottom the temperature
of the atoms is decreasing.
Motivated by the need to test the concept of condensation of composite particles in weakly
interacting systems, in 1962 Blatt et al. proposed the investigation of the BEC in gases of
excitons [28]. Excitons are bound electron-hole pairs that can form a weakly interacting gas
in certain non-metallic crystals. They are interesting because their small mass allows BEC at
high temperatures and gas density can be controlled over a wide range, by only modifying the
intensity of the optical excitation. Being quasi-particles, excitons can be created and annihilated,
that is their number is not conserved. Excitons were discovered in 1968, and the first evidence
for Bose-Einstein of biexciton molecules in a CuCl crystal dates back to 1979 [44].
The laser as coherence phenomenon between photons shares many analogies with conden-
431
432 CHAPTER 15. THERMODYNAMICS OF QUANTUM GASES
sates. However, photons are quasi-particles as well, and again their number is not conserved 1 .
Hence, there is no phase transition: When an optical cavity containing photonic modes is cooled,
the photons prefer to disappear in the walls of the cavity instead of condensing.
Hecht [117] suggested in 1959, followed by Stwalley and Nosanow [240] in 1976, that an
atomic hydrogen gas with polarized spins would be an appropriate candidate for BEC. The
advantage of this system is that interactions between atoms are weak and only give rise to a
negligible quantum depletion below 1%. In 1978 Greytak and Kleppner started at the MIT
intensive efforts to generate BECs in dilute hydrogen gases. In the 1990s, important advances in
the cooling of atoms using laser light allowed to reach very low temperatures, and the invention
of the magneto-optical trap (MOT) for neutral atoms permitted their spatial confinement and
the compression of their density. These successes boosted efforts to try to create BEC in alkaline
gases, which have electronic level schemes that lend themselves to optical cooling. Later, it was
discovered that the phase space density in MOTs is limited by radiation trapping effects. As
a solution to this problem, scientists had to learn how to trap atoms without the use of light
in conservative traps, e.g. by their magnetic dipole moment, and to replace optical cooling
with evaporative cooling. This was the crucial step that finally permitted to reach BEC in
alkaline gases in 1995. Later, the hydrogen experiment, which initially stimulated the alkaline
experiments, now taking advantage of their success, has been taken to BEC as well [89].
Why did it take so long to reach Bose-Einstein condensation, seven decades after its pre-
diction by Bose and Einstein? How can we see when we have a condensate? What are the
characteristics of a BEC accessible to observation and how to measure them? These are the
answers that we will answer in the following sections. Solve Exc. 15.3.1.1.
Figure 15.2: Scheme of the Broglie wave of cold atoms. From top to bottom the temperature
decreases.
15.1 Quantum statistics of an ideal Bose gas

The canonical approach to statistical mechanics begins with the probabilistic analysis of Boltz-
mann’s velocity distribution of an ideal gas. For a gas consisting of particles of mass m at
temperature T , the velocity distribution is given by the well-known Maxwell-Boltzmann law
(MB) [131]
r 3
m 2
g(v) = e−mv /2kB T , (15.1)
2πkB T
where kB is the Boltzmann constant. Maxwell-Boltzmann’s law was experimentally proven by
Otto Stern in 1920, using a primitive atomic beam and a simple time-of-flight technique based
on a rotating drum for selecting atomic velocities. With the advent of laser spectroscopy, the
MB law and its limitations can be tested with highly improved accuracy. This law describes
1
The chemical potential of photons is µ = 0.
15.1. QUANTUM STATISTICS OF AN IDEAL BOSE GAS 433
well the behavior of weakly interacting hot atoms. Deviations from this law are insignificant
until, at low temperatures, quantum effects come into play. For this to happen the temperature
must be so low that the atomic Broglie wavelengths become comparable to the average distance
between particles. For a gas in thermal equilibrium the characteristic wavelength, called thermal
de Broglie wavelength, is,
s
2π~2
λth = , (15.2)
mkB T
where ~ = h/2π is Planck’s constant. In a gas of density n, the mean distance between particles
is n−1/3 . So, quantum effects are expected to emerge when n−1/3 ∼ λdB (T ), such that the limit
for this regime is defined by,
2π~2 2/3
kB T (n) = n . (15.3)
m
For example, an atomic gas with density n ∼ 1016 cm−3 and temperature 900 K is certainly
in the classical regime, since n−1/3 ∼ 106 cm λdB = 10−9 cm. To observe quantum effects,
we need relatively dense and cold clouds of atoms. In most gases, lowering the temperature or
increasing the density promotes the system to liquidity before the quantum regime is reached.
Well-known exceptions are spin-polarized hydrogen (H↑), which does not become liquid and
helium, which exhibits quantum degeneracy effects in the liquid phase, although these effects
are quite complex due to strong interparticle forces.
We have already seen that all particles in the quantum world are either bosons with integer
spin or fermions with semi-integer spin. Fermions do not share a quantum state, because they
must follow the Pauli’s exclusion principle. They obey a quantum statistical distribution called
Fermi-Dirac distribution (FD). In contrast, bosons enjoy to share a quantum state and even
encourage other bosons to join them in a process called bosonic stimulation. Bosons obey a
quantum statistical distribution called Bose-Einstein distribution (BE). The basic difference
between the MB-statistics on one hand and the BE- or FD-quantum statistics on the other is
that the former applies to identical particles which, however, are distinguishable from each other,
while the second describes identical indistinguishable particles. For the BE/FD statistics one
can derive [149] the occupancy number for a non-degenerate quantum state having the energy
ε when the system is kept at temperature T ,
1 1
wT,µ = 3 β(ε−µ)
, (15.4)
(2π) e ∓1
where we used the abbreviation β ≡ 1/kB T . The upper sign refers to the BE statistics, the
lower sign to the FD statistics. The chemical potential µ is an important system parameter,
which helps to normalize the distribution (15.4) to the total number of particles,
X
N= wT,µ (ε) . (15.5)
ε
Similarly, the total energy of the system is given by,

X
E= εwT,µ (ε) . (15.6)
ε
A very remarkable effect occurs in a bosonic gas at a certain characteristic critical temperature
Tc : below this temperature a substantial fraction of the total number of particles occupies the
lowest energy state, while all other states are occupied by a negligible number of particles.
Above the transition temperature the macroscopic observables of the gas, such as pressure, heat
capacity, etc., receive contributions of all states with a certain statistical weight, but without
favoring the state of lower energy. Below the transition temperature, the observables are altered
by a macroscopic occupation of the ground state, which results in dramatic changes of the
thermodynamic properties. The phase transition is named after Shandrasekar Bose [32] and
Albert Einstein [80] Bose-Einstein condensation (BEC).
15.1.1 Condensation of a free gas confined in a box potential

One of the keys to understanding BEC is the behavior of the chemical potential µ at very low
temperatures. The chemical potential is responsible for the concentration of a large number
of atoms in the ground state N0 . A system with a large number of non-interacting bosons
condenses to the ground state when the temperature approaches zero, N0 → N . The Bose-
Einstein distribution function (15.4) gives the population of the ground state, ε = 0, in the zero
temperature limit, N = limT →0 (e−βµ − 1)−1 = −1/βµ, or in terms of the fugacity defined by,
Z ≡ eβµ , (15.7)
we may write, Z ' 1 − 1/N . It should be noted that the chemical potential in a bosonic system
must always be less than the ground state energy in order to guarantee non-negative occupation
wT,µ (ε) of any state. Z ' 1 denotes macroscopic occupation of the ground state. We define
the critical temperature for Bose-Einstein condensation via the occupation of the ground state.
Above this temperature the occupancy of ground state is not macroscopic, below it is.
For a Bose gas of N non-interacting particles with mass m confined inside a box potential of
volume V = L3 the critical temperature for BEC can be calculated from equation (15.3). The
boundary conditions require that the momenta satisfy pj = 2π~lj /L, where j = x, y or z and j
are integers. Each state is labeled by a set of three integers (lx , ly , lz ). In the thermodynamic
limit, the sum over all quantum states can be converted into an integral over a continuum of
states,
X N →∞ 1 Z Z
−→ 3 d3 rd3 p = d3 rd3 k . (15.8)
~
r,k
For a free gas with energy ε = p2 /2m we can derive, simplifying the calculation using
the density of states η defined by (14.19), which basically depends on the geometry of our
system 2 . Using the occupation number wT,µ (ε) for the Bose-Einstein distribution (15.4) in the
thermodynamic limit and the density-of-states in a box potential (14.22), let us calculate the
total number of particles,
ZZ Z ∞
3 3 3
N = N0 + wT,µ (ε(r, k))d rd k = N0 + (2π) wT,µ (ε)η(ε)dε (15.9)
0
√ 3 Z ∞
2m ε1/2 dε
= N0 + V ,
(2π)2 ~3 0 eβ(ε−µ) − 1
where the ground state population N0 is maintained explicitly. In the process of converting the
sum to an integral (15.8) the density of states disappears when we approach the ground state.
This error is corrected by adding a contribution of an explicit term N0 to the integral.
2
We must, however, keep in mind that the state density approach is an approximation not valid for experiments
with a limited number of atoms.
15.1.1.1 Riemann’s zeta function

At this point, to help simplifying the notation, we introduce the Bose function and its integral
representation,
X∞ Z ∞
Zt 1 xξ−1 dx
gξ (Z) = = . (15.10)
tξ Γ(ξ) 0 Z −1 ex − 1
t=1
where Γ(η) denotes the Gamma function. Analogically, we can define the Fermi function via,
∞
X Z ∞
(−Z)t 1 xξ−1 dx
fξ (Z) = − ξ = . (15.11)
t Γ(ξ) 0 Z −1 ex + 1
t=1
For classical particles, Z ∞

1 xξ−1 dx
cξ (Z) = =Z . (15.12)
Γ(ξ) 0 Z −1 ex + 0
A particular value is the Riemann zeta-function,
gξ (1) = ζ(ξ) . (15.13)
(a) 3 (b) 6
c2
g3 5
2 f3
g9 (z)
4
1(9)
g 3/2
f 3/2 3
1
2
0 1
0 0.2 0.4 0.6 0.8 1 1 2 3 4
z 9
Figure 15.3: (Code: AO T hermodynamics BoseF ermiF unction.m) (a) Bose and Fermi func-
tions for box potentials (g3 and f3 ) and for harmonic potentials (g3/2 and f3/2 ). Also shown is
the Boltzmann limit (15.12). (b) Riemann function.
With this definition and the definition of the thermal Broglie wavelength (15.2), Eq. (15.9)
becomes,
V
N = N0 + 3 g3/2 (eβµ ) . (15.14)
λth (T )
(3/2)
We can use Eq. (15.14) to calculate the critical temperature3 Tc defined for N0 % 0 and
(3/2)
µ → 0. Above the phase transition, T > Tc , the population is distributed over all states,
(3/2)
each individual state being weakly populated. Below Tc the chemical potential is fixed by
µ = 0 and the number of the particles occupying the excited states is,
!3/2
V T
Nth = 3 g3/2 (1) = N (3/2)
, (15.15)
λth (T ) Tc
3
The superscript (3/2) denotes the box potential shape of the trapping potential. See 15.3.1.3 for an explanation
of the notation.
with g3/2 (1) = 2.612. Since N0 + Nth = N , the number of particles in the ground state is given
by,
!3/2 2/3
(3/2)
N0 min(T, Tc ) ~2 N
=1− (3/2)
with Tc(3/2) = , (15.16)
N Tc 2πm V g3/2 (1)
N0 /N is the fraction of the atomic cloud which is condensed in the ground state. The abrupt
(3/2)
occurrence of a finite occupation in a single quantum state at temperature below Tc indicates
a spontaneous change in the system and a thermodynamic phase transition. Solve Exc. 15.3.1.2.
15.1.2 Condensation of a harmonically confined gas

The critical temperature Tc0 can be significantly altered, when the atoms are confined to a
spatially inhomogeneous potential. The critical temperature depends on the general shape and
the tightness of the potential. Let us consider N particles of an ideal Bose gas distributed over
several quantum states of an arbitrary potential. The occupation number wT,µ (ε) of particles
at an energy level ε is still given by (15.4), the ground state energy is defined as zero. In
the thermodynamic limit, the relation between the chemical potential and the total number of
particles is still given by Eq. (15.9), with an adequate density of states η(ε). The state density
for an arbitrary confinement potential U (r) can be found by generalizing the calculation to the
free gas. The phase space volume between the energy surfaces ε and ε + dε is proportional
to the number of states in this energy range. However, the external potential limits the space
available for the gas. For a harmonic potential (14.18) with the mean secular frequency ω̄ the
density-of-states η(ε) has already been calculated in Eq. (14.21). With this, we can analogically
to (15.9) calculate,
ZZ Z ∞
3 3 3
N = N0 + wT,µ (ε(r, k))d rd k = N0 + (2π) wT,µ (ε)η(ε)dε (15.17)
0
Z ∞
1 ε2 dε kB T 3
= N0 + = N0 + g3 (Z) .
2(~ω̄)3 0 eβ(ε−µ) − 1 ~ω̄
In the same way as for a potential well we find for a harmonic potential,
!3
kB T 3 T
Nth = g3 (1) = N , (15.18)
~ω̄ (3)
Tc
with g3 (1) = 1.202. Since N0 + Nth = N , the number of particles in the ground state is,
!3 1/3
(3)
N0 min(T, Tc ) N
=1− (3)
with kB Tc(3) = ~ω̄ . (15.19)
N Tc g3 (1)
The superscript (3) indicates the harmonic shape of the trap.

Fig. 15.4(left) traces the condensed fraction Nc /N measured as a function of the reduced tem-
(3)
perature T /Tc . Experiments [113, 82] confirm Bose’s ideal gas theory in the thermodynamic
limit.
We note that smaller trapping volumes (or tighter potentials) increase the critical tempera-
ture Tc , thus allowing for quantum degeneracy at higher temperatures, which can be advanta-
geous in experimentation. Also, at a given temperature, a strongly confining potential reduces
the total minimum number of atoms required to reach condensation.
2.0
1.5
E / NkBTo
0.0
1.0
-0.2
∆
0.5 -0.4
0.4 0.8 1.2 1.6
0.0
0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
T / To(N)
(3)
Figure 15.4: Left: Condensed fraction Nc /N as a function of reduced temperature T /Tc . The
circles represent measured values. The upper solid line shows the theory for an ideal gas,
the intermediate solid line shows a curve fitted to the measurements, and the lower part is
a theoretical curve taking into account finite size effects and interatomic interactions. Right:
Measurement of the release energy [82].
15.1.3 Energy and heat capacity

When the number of atoms is limited, N < ∞, we expect a slightly reduced critical tempera-
ture [105]. In addition, the interatomic interaction reduces the critical temperature [13]. The
technical resolution of most experiments today is not sufficient to permit the study of these
effects. However, measurements of other thermodynamic quantities such as energy and heat
capacity [69, 82] showed significant deviations from the ideal gas behavior due to interaction
effects. Therefore, although temperature being the most basic variable of the thermodynamic
state, the system needs to be characterized by other quantities.
Heat is not a state variable because the amount of heat needed to raise the temperature of the
system depends on how the heat is transferred. The heat capacity quantifies the system’s ability
to secure its energy. In conventional systems, the heat capacity is typically given either specified
at constant volume or at constant pressure. Together with this specification, heat capacities are
extensive state variables. When crossing a phase transition, the temperature-dependent heat
capacity measures the degree of change in the system above and below the critical temperature
and provides valuable information about the general type of phase transition.
R
The total energy E/N ≡ N −1 wd3 xd3 k per particle is given by,
R
E wT,µ (x, k)d3 xd3 k
= R (15.20)
N wT,µ (x, k)d3 xd3 k
R
η()(eβ(−µ) − 1)−1 dε g4 (Z)
= R = 3kB T .
η()(eβ(−µ) − 1)−1 dε g3 (Z)
For a confined gas, volume and temperature are interdependent, and the concept of pressure
is somewhat vague. In this case, we can not refer to the heat capacity at constant volume or
pressure. However, one can define the heat capacity for a fixed number of particles,
∂E(T )
C(T ) = . (15.21)
∂T
×10 5
6 2
(μK)
0
4
Nc
-2
μ/kB
2
-4
0 -6
0 0.5 1 1.5 0 0.5 1 1.5
T/Tc T/Tc
10 15
(μK)
10
C/N kB
5
E/kB
0 0
0 0.5 1 1.5 0 0.5 1 1.5
T/Tc T/Tc
Figure 15.5: (Code: AO T hermodynamics BoseT hermP otentialsBox.m) Numerical calcula-

tion of thermodynamic potentials for a Bose gas as a function of temperature for a given trap-
ping potential. (a) Chemical potential, (b) energy, (c) heat capacity per particle, and (d) total
heat capacity.
Recalling the implicit dependencies of the thermodynamic variables on temperature, we can

evaluate (15.19):
Z ∞
2 0 ε − µ β(ε−µ)
C(T ) = β εf (ε) ρ(ε) µ (T ) + e dε , (15.22)
0 T
where the derivative of the chemical potential approaching the phase transition from above,
T & Tc0 , is
R∞
0 + 1 0 εf (ε)2 ρ(ε)eβε dε
µ (Tc ) = − R ∞ . (15.23)
T 0 f (ε)2 ρ(ε)eβε dε
Calculating the second moments of the distributions obtained for the same density by time-
of-flight of absorption images, we obtain the kinetic energy,
Z
p2
U= n(p)d3 p . (15.24)
2m
For confined ideal gases, the virial theorem ensures Ekin + Epot = 2Ekin . For real gases, the
repulsive energy of the mean field adds to this energy, E = Ekin + Epot + Eself . The sudden
extinction of the trapping potential before time-of-flight takes away the potential energy Epot
non-adiabatically. The kinetic energy and the self-energy of the condensate are fully converted
into kinetic energy during ballistic expansion. It is this energy, p2 /2m = Ekin + Eself , which is
sometimes called release energy, which is measured after ballistic expansion 4 . Fig. 15.4(right)
shows a measurement of the release energy. Solve the Exc. 15.3.1.3.
4
It is interesting to measure the heat capacity of a partially condensed cloud near the critical point and analyze
the discontinuity, because it contains important information about interatomic interactions and finite-size effects
([50], Seç. 3.4). In addition, the classification of Bose-Einstein condensation as a phase transition depends very
much on the behavior of the thermodynamic potential near the critical point [157, 131].
15.1.4 Distribution functions for a Bose gas

Bose-Einstein condensates consist of atoms sharing a single quantum state. In inhomogeneous
potentials, the condensate and the thermal fraction form spatially separated clouds, concen-
trated around the center of the potential and therefore very dense. For this reason, interatomic
interaction effects generally dominate the density and momentum distribution of the condensed
fraction. However, the non-condensed (or normal, or thermal) fraction is also subject to mod-
ifications due to the bosonic nature of the atoms. Since the density of the normal fraction is
generally much smaller, these modifications are weak. In this section, we will only discuss these
effects briefly, but we note that the calculations are analogous to the calculations for fermionic
gases presented in Sec. 15.2.4.
For an ideal Bose gas the density and momentum distributions are expressed by Bose func-
tions g3/2 (z) [50]. For example, as will be derived in Exc. 15.3.1.4(a), the density and momentum
distributions are,
1
n(x) = g3/2 (e−β[U (x)−µ] )
λ3dB
bosonic distribution functions . (15.25)
a6trap 2
n(k) = 3 g3/2 (eβ(µ−p /2m) )
λdB
In the classical limit, we can calibrate the chemical potential by Eq. (15.14) for a box potential
or by (15.17) for a harmonic potential,
( N 3
βµ βµ βµ V λth 3
for a box potential
g3/2 (e ) → c3/2 (e ) = e = ~ω̄ . (15.26)
c3 (eβµ ) =N kB T for a harmonic potential
Hence, we obtain for the classical density distribution,
1 −β[U (x)−µ] eβµ −βU (x)

n(x) = c3/2 (e ) = e (15.27)
λ3th λ3th
( N −βU (x)
V qe for a box potential
= mω̄ 2
3 2 2 .
N 2πk BT
e−βmω̄ x /2 for a harmonic potential
Similarly, the momentum density distribution is given by,
a6trp β(µ−p2 /2m)

a6trp eβµ −βp2 /2m
n(k) = c3/2 (e ) = e (15.28)
λ3th λ3th
( N 6 −βU (x)
V atrp
q
e for a box potential
= 1
3 2 .
N ~3 2πmk BT
e−βp /2m for a harmonic potential
We see that we recover the Maxwell-Boltzmann velocity distribution, as seen in Fig. 15.6,
r 3
m3 m 2
n(v) = n(k) 3 = N e−βmv /2 . (15.29)
~ 2πkB T
(a) #10 14 (b) #10 -17

8 2
(cm!3 )
(cm!3 ) 6 1.5
4 1
n(p)
n(x)
2 0.5
0 0
-50 0 50 -50 0 50
x=atrp atrp p=7h
Figure 15.6: (a) Density and (b) momentum distribution of a Bose gas (red) and a Boltzmann
gas (green) at T = 1.1Tc (solid line) and at T = 2Tc (dotted line).
15.1.4.1 Ballistic expansion

To describe the density distribution of an ultracold Bose-gas after a time-of-flight we replace in
the second Eq. (15.25): k = mr/~ttof . We obtain the density distribution,

m 3 m 3 a6trp 2 2
ntof (r, ttof ) = n(k = mr/~ttof ) = g3/2 (e(µ−mr /2ttof )/kB T ) (15.30)
~ttof ~ttof 3
λth
r 3
T →∞ m 3 1 2 2 N 2 2
−→ N~3
e−mr /2ttof kB T = e−r /2rrms ,
~ttof 2πmkB T 3/2
(2π) rrms 3
where we defined,
kB T ttof
rrms ≡ . (15.31)
m
This distribution does not directly depend on the potential U (r), that is, the expansion is
isotropic. However, the chemical potential does depend on the potential. For very long flight
times (usually several 10 ms) the density resembles a Gaussian distribution [50]. In Exc. 15.3.1.4(b)
we determine the time-of-flight density distribution of an ultracold Bose gas.
In a time-of-flight experiment, any deviation observed between the results (15.30) and (15.31)
points towards an impact of quantum statistics. However, absorption images only record pro-
jections of the time-of-flight distribution on a plane.
15.1.4.2 Temperature and excitations

The temperature of a Bose condensate is given by the ratio of the numbers of condensed and
thermal atoms. What about collective excitations in pure condensate? Can they be cooled?
In fact, an oscillating BEC is not in thermal equilibrium. However, in the presence of some
dissipation mechanism the excitations may thermalize evolving towards a steady state. Once
the collective excitations have become thermal excitations, they simply increase the thermal
fraction.
15.2 Quantum statistics of an ideal Fermi gas

Atoms are fermions or bosons, or depending on their spin is integer or semi-integer. For example,
87 Rb atoms with their total integer spin of F are bosons, while 40 K atoms having a half-
15.2. QUANTUM STATISTICS OF AN IDEAL FERMI GAS 441
integer spin are fermions. At high phase space densities, atoms have to figure out how they
will organize their coexistence. Bosons encourage each other to occupy the same phase space
cell, in contrast to the reluctant fermions, which prefer to follow Pauli’s exclusion principle. The
different behavior is described by different quantum statistics that determine how the phase space
(i.e., the available energy levels) has to be filled by the atoms. The Bose-Einstein distribution is
valid for bosons, the distribution of Fermi-Dirac for fermions and both asymptotically approach
the Boltzmann distribution at high temperatures. We have seen that bosons undergo a phase
transition and condense in the ground state when the temperature is reduced below a critical
threshold. On the other hand, the fermions organize their phase space, so that their energy levels
are arranged like a ladder. The impact of fermionic quantum statistics on a cold cloud of atoms
were observed experimentally by DeMarco and Jin [69, 195]. They cooled a two-components
Fermi gas of 7 × 105 potassium atoms down to 300 nK, which corresponded to 60% of the atoms
populating energy levels below the Fermi energy. The measured density distribution was found
to deviate from the one expected for an ideal Boltzmann gas 5 .
15.2.1 Chemical potential and Fermi radius for a harmonic trap

The phase space density for a degenerate Fermi gas has already been given in (15.4). In the
same way as for a Bose gas, the chemical potential must satisfy the normalization condition,
ZZ Z
N= wT,µ (x, k)d3 xd3 k = (2π)3 wT,µ (ε)η(ε)dε (15.32)
Z ∞
1 ε2 dε kB T 3
= = f3 (Z) .
2(~ω̄)3 0 eβ(ε−µ) + 1 ~ω̄
In the last line, we inserted the density of the states into a harmonic potential (14.21) and used
the definition of the Fermi functions (15.11).
For low temperatures, x ≡ µ 1, we can use the Sommerfeld expansion which in first order
gives fη (ex ) ' xη /Γ(η + 1). From this we immediately obtain the energy, the Fermi radius and
the momentum of free particles,
EF = µ(T = 0) = ~ω̄(6N )1/3 (15.33)

s s
2EF 2EF
rF = 2
and zF =
mωr mωz2
r
2mEF
KF = .
~2

Using the second order of the Sommerfeld expansion, fξ (ex ) ≈ xξ /Γ(ξ+1) 1 + π 2 ξ(ξ − 1)/6x2 + ... ,
we obtain for the chemical potential the equation, 0 = µ3 + (πkB T )2 µ − EF3 . The approximate
solution of this equation, neglecting terms such as 4π 6 kB
6 T 6 27E 6 , is
F
" 2 #
π2 kB T
µ = EF 1− . (15.34)
3 EF
5
We note that meanwhile ultracold two-components Fermi gas have been demonstrated to form bosonic Cooper-
pairs, similarly to the phenomena known as superconductivity in some metals and as superfluidity of the fermionic
3
He.
Z→1
In the limit of high temperatures, fη (Z) −→ Z
3 3
kB T βµ kB T
N= e = (1 + βµ + ...) , (15.35)
~ω̄ ~ω̄
3 3
~ω̄ 1 EF
µ ≈ kB T ln N = kB T ln .
kB T 6 kB T
This means that we recover the Boltzmann gas, which satisfies an ideal gas equation similar to
that of particles in a box potential,
3
kB T
N= (Boltzmann) . (15.36)
~ω̄
For comparison, for bosons we have,

3
N0 T g3 (Z)
=1− (Bose) . (15.37)
N Tc g3 (1)
The chemical potentials are calculated in Fig. 15.7(a).
15.2.2 Energy
R
The total energy E1 ≡ E/N ≡ N −1 wd3 xd3 k per particle is given by,
R
wT,µ (x, k)d3 xd3 k
E1 = R (15.38)
wT,µ (x, k)d3 xd3 k
R −1
η() eβ(−µ) + 1 dε f4 (Z)
= R −1 = 3kB T .
η() eβ(−µ) + 1 dε f3 (Z)
Again using the Sommerfeld approximation, we see that for T → 0 the energy is limited by
3
E1 = f4 (eβµ ) (15.39)
N β(β~ω̄)3

3µ4 2π 2 T →0 3
= 3 1+ 2
+ ... −→ EF (Fermi) .
4EF (βµ) 4
because the atoms are forced to adopt a state of larger dynamics in the outermost regions of
T →0
the trap. This implies, E1 /EF −→ 3/4.
In the limit of high temperatures, T → ∞, a classical gas has the energy per particle,
3

3 −1 (βEF )
E1 = f4 f3 ≈ 3kB T (Boltzmann) , (15.40)
N β(β~ω̄)3 6
Z→0
which is seen by taking the high temperature limit fη (Z) −→ Z and extrapolating to all Z.
T →∞
This implies, E1 /EF −→ 3kB T /EF .
Comparing with bosons,
g4 (1)
E1 = 3kB T ≈ 2.7 (Bose) . (15.41)
g3 (1)
15.2.3 Entropy and heat capacity

R
The entropy S = −kB g() [w ln w + (1 − w) ln(1 − w)] d per particle is,
f4 (Z) µ 4E1 µ
− =
S1 = 4kB − . (15.42)
f3 (Z) T 3T T

1
The heat capacity per particle C1 = ∂E 0
∂T is easily calculated using Zfη (Z) = fη−1 (Z),
N

f4 (Z) 3µ f4 (Z)f2 (Z) E1 3µ f4 (Z)f2 (Z)
C1 = 3kB − 1− = − 1− . (15.43)
f3 (Z) T f3 (Z)2 T T f3 (Z)2
For fermions well below the Fermi temperature, T → 0, using the Sommerfeld approximation,
we calculate,
2
T →0 3π kB T
C1 −→ (Fermi) . (15.44)
2 TF
For high temperature T
C1 ≈ 3kB (Boltzmann) . (15.45)
(a) (b) 6
2
5
(7K)
(7K)
4
0
3
E=kB
7=kB
-2
2
1
-4
0
0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8
T=TF T=TF
#10 6
(c) 10 (d) 10
8 8
C=N kB
6 6
C=kB
4 4
2 2
0 0
0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8
T=TF T=TF
Figure 15.7: (Code: AO T hermodynamics F ermiT hermP otentials.m) Numerical calculation of

thermodynamic potentials for Bose (red) and Fermi (green) gases as a function of temperature
for a given harmonic trapping potential. (a) Chemical potential, (b) energy, (c) heat capacity
per particle, and (d) total heat capacity. The dotted magenta line in (a) shows the chemical
potential calculated from the Sommerfeld approximation.
15.2.4 Distributions of a Fermi gas

15.2.4.1 Spatial distribution
The density distribution is,
Z Z
1 2k 2 dk
n(x) = wT,µ (x, k)d3 k = 2 2 2 (15.46)
(2π) eβ[~ k /2m+U (x)−µ] + 1
Z √
1 2m 3/2 εdε 1 2m 3/2
= = Γ(3/2)f3/2 (e−β[U (x)−µ] ) ,
(2π)2 ~2 e β[ε+U (x)−µ] +1 (2π) 2 β~2
such that,
n(x) = λ−3
dB f3/2 (e
−β[U (x)−µ]
) (Fermi) . (15.47)
At low temperatures, T → 0, we can apply the Sommerfeld expansion [39], which to first order
gives µ → EF ,
3/2
1 Γ(3/2) 2m
n(x) ≈ [µ − U (x)] (15.48)
(2π)2 Γ(5/2) ~2
3/2
1 2 2m 3/2 m 2 2 3/2 8λ N ρ2
= EF − ωr ρ = 2 3 1− 2 .
(2π)2 3 ~2 2 π RF RF
At high temperatures, T → ∞, we should recover the Boltzmann gas situation,
n(x) = λ−3
dB f3/2 (e
−β[U (x)−µ]
) (15.49)

2 3/2
3 −βU (x) mβ ω̄ 2 2 +ω 2 y 2 +ω 2 z 2 )/2
≈ λ−3
dB N (β~ω̄) e = N e−βm(ωx x y z .
2π
R
It’s easy to check, n(x)d3 x = N . Introducing the peak density n0 , we obtain,
2 ρ2 /2k
n(x) = n0 e−mω BT
(Boltzmann) . (15.50)
q
The rms-radius of the distribution is σj = kB T /mωj2 , which seems contrary to the above
D E
results, m ω
2 j
2 x2 = k T . In comparison,
j B
h i
n(x) = λ−3 g
dB 3/2 e β(µ−U (x))
(Bose gas above Tc ) . (15.51)
p p
where λdB = 2π~2 /mkB T e atr = ~/mω̄.
15.2.4.2 Momentum distribution

The momentum distribution is,
Z Z
3 1 rdrdz
ñ(k) = wT,µ (x, k)d x = 2 β[ε(k)+mω 2 2 (15.52)
(2π) e r ρ /2−µ] + 1
Z
1 4πρ2 dρ
=
(2π)3 eβ[ε+mωr2 ρ2 /2−µ] + 1
3/2 Z √ 3/2
1 2 tdt 1 2
= = Γ(3/2)f3/2 (eβ(µ−ε) ) , (15.53)
(2π)2 βmωr2 eβ[ε+t−µ] + 1 (2π)2 βmωr2
such that,
ñ(k) = λ−3 6
dB atr f3/2 (e
β(µ−ε)
) (Fermi) . (15.54)
At low temperatures, T → 0,
3/2
1 2 Γ(3/2)
ñ(k) ≈ (β [µ − ε])3/2 (15.55)
(2π)2 βmωr2 Γ(5/2)
3/2 3/2 3/2
1 2 2 ~2 k 2 8 N k2
≈ EF − = 1 − .
(2π)2 mωr2 3 2m π 2 KF3 KF2
This can easily be integrated by dimensions,
Z ∞ Z ∞ Z Z 3/2
8 N k2
ñT →0 (kz ) = ñcl (k)dkx dky = 2 3 1− 2 dkx dky (15.56)
−∞ −∞ π KF |k|≤KF KF
Z 2π Z √K 2 −kz2 !3/2 5/2
8 N F kz2 kρ2 16 N kz2
= 2 3 1− 2 − 2 kρ dkρ dφ = 1− 2 .
π KF 0 0 KF KF 5π KF KF
R∞
It is easy to check −∞ ñT →0 dkz = N , with Maple.
At high temperatures, T → ∞, we should recover the Boltzmann gas situation,
3/2
~2 ω̄ 2
ñ(k) ≈ N e−βε (Boltzmann) . (15.57)
2πmωr2
√
Since ε is the kinetic energy, the rms-radius k 2 of this distribution is β~2 hk 2 i = m. In
comparison,
h i
β (µ−p2 /2m)
ñB (k) = λ−3
dB a6
tr g3/2 e (Bose gas above Tc ) . (15.58)
Example 58 (Integrated momentum distribution of a Fermi gas): To integrate the

momentum distribution of finite temperature Fermi gas by dimensions,
3/2 Z ∞ Z ∞ Z ∞
1 2 4πr̃2 dr̃
ñ(kz ) = 2 dky dkx (15.59)
(2π)3 βmω̃tr −∞ −∞ 0 eβε−βµ+r̃2 + 1
3/2 Z ∞ Z ∞
1 2 4πr̃2 dr̃
= 3 2 2π β~ 2 k 2 /2m+β~2 k 2 /2m−βµ+r̃ 2 kρ dkρ
(2π) βmω̃tr 0 0 e z ρ +1
3/2 Z Z
1 2 2m ∞ ∞ k̃ρ dk̃ρ
= 2 2
r̃2 dr̃
π βmω̃tr β~ 0 0 e β~ 2 k 2 /2m−βµ+r̃ 2 +k̃ 2
z ρ + 1
3/2 Z ∞
1 2 2m 1 ∞ 2 1
dr̃
= 2 2
r̃ ln −β~ 2 k 2 /2m+βµ−r̃ 2 −k 2
π βmω̃tr β~ 2 0 1+e z ρ
0
1/2 Z ∞
2 2 2 2 2
= 2 2 r̃2 ln 1 + eβµ−β~ kz /2m−r̃ dr̃ .
π (β~ω̃tr ) βmω̃ tr 0
15.2.4.3 Time-of-flight distribution

To describe time-of-flight images we substitute k = mr/~t. We obtain the density distribution
from a convolution,
Z 3
1 3 3 δ (x − x0 − pt/m)
nT oF (x, t) = d x 0 d k (15.60)
(2π)3 eβ(H(x0 ,p)−µ) + 1
Z
1 d3 k
=
(2π)3 eβ(H(x+pt/m,p)−µ) + 1
Z
1 dkx dky dkz
= .
(2π)3 e βΣj [~2 kj2 /2m+ 12 mωj2 (xj +~kj t/m)2 ]
/Z + 1
We rewrite the exponent,
~2 kj2 /2m + 21 mωj2 (xj + ~kj t/m)2 = ~2 kj2 /2m(1 + ωj2 t2 ) + ωj2 txj ~kj + 21 mωj2 x2j (15.61)
s √  2
~2 kj2 ωj2 txj 2m mωj2 x2j
= 2
(1 + ωj t2 ) + q  +
2m 2 1 + ωj2 t2 2(1 + ωj2 t2 )
m
= ξj + ω̌j2 x2j .
2
r
−1/2 2
where we define ω̌i ≡ ωi 1 + ωi t 2 2 . With the substitution dξj = dkj 2~ m j ξ 1 + ω 2 t2 we
j
obtain
3/2 Z
1 mkB T 1 β 3/2 (ξx ξy ξz )−1/2 dξx dξy dξz
nT oF (x, t) = Q 2 2
(15.62)
(2π)3 2~2 i 1 + ωi t eβΣj [ξj + 2 ω̌j xj ] /Z + 1
m 2 2
Z
1 1 ω̃ 3 β 3/2 ξ −3/2 4πξ 2 dξ
= ,
23 π 3/2 λ3dB ω̄ 3 eβΣj [ξ+ 2 ω̌j xj ] /Z + 1
m 2 2
where ω̄ ≡ (ωx ωy ωz )1/3 and ω̌ ≡ (ω̌x ω̌y ω̌z )1/3 .
1 ω̌ 3
βµ− 12 βmΣj ω̌j2 x2j
nT oF (x, t) = f3/2 e . (15.63)
λ3dB ω̄ 3
For long times-of-flight t ω −1 ,

1 1 m 3
β (µ−mx2 /2t2 )
nT oF (x, t) = f3/2 e = ñ(mx/t) . (15.64)
λ3dB ω̄ 2 t2 ~t
At low temperatures,
m 3 N 8 (mx/~t)2
3/2
nT oF (x, t) = 1− (15.65)
~t KF3 π 2 KF2
 !2 3/2
m 3 R 3 R mx/~t
F  F 
= 1−
~t 6π 2 λ (48N λ)1/3
# 1014
2
1.5
(cm!3 )
1
n
0.5
0
0 50 100 150 200
r (7m)
Figure 15.8: (Code: AO T hermodynamics F ermiDistributionT of.m) Time-of-flight velocity

distribution of (red) a Li Fermi gas at T = 0 with vanishing initial spatial distribution [39] and
(green) a thermal gas at T = TF .
At high temperatures,
1 1 2 2
nT oF (x, t) = f3/2 (eβ(µ−mx /2t ) ) (15.66)
λ3dB ω̄ 2 t2
1 1 β(µ−mx2 /2t2 )
≈ 3 e
λdB ω̄ 2 t2
3/2
mkB T 3/2 1 ~ω̄ 3 −βmx2 /2t2 ω̄ m 2 2
≈ 2 2 2
N e ≈N 2 e−βmx /2t .
2π~ ω̄ t kB T t 2πkB T
A rms-width is,
Z

2
rT oF = r2 nT oF (x, t)d3 x (15.67)
Z
1 ω̌ 3 2 βµ− 12 βmΣj ω̌j2 x2j
= 3 3
r f3/2 e d3 x
λdB ω̄
Z
2 εg(ε)dε kB T g4 (Z)
= 2
= .
mω̌r N e β(ε−µ) +1 mω̌r2 g3 (Z)
This shows that the width of the flight-of-time
√ distribution can simply be obtained from the spa-
tial distribution by substituting ω → ω/ 1 + ω 2 t2 . Of course this does not hold for condensed
gases Bose.
15.2.5 Equipartition theorem

We find for harmonic traps,
R Z
U (x)wT,µ (x, k)d3 xd3 k 1 mω 2 r2 d3 xd3 k
Epot,1 = R = (15.68)
wT,µ (x, k)d3 xd3 k (2π)3 N 2 eβ[~ k /2m+mω2 r2 /2−µ] + 1
2 2
Z
16 u4 v 2 dudv
=
πN β 4 (~ω)3 eu2 +v2 /Z + 1
Z R 2 2
1 ~2 k 2 d3 xd3 k ~ k wT,µ (x, k)d3 xd3 k
= 2 2 2 2 = R = Ekin,1 .
(2π)3 N 2m eβ[~ k /2m+mω r /2−µ] + 1 2m wT,µ (x, k)d3 xd3 k
This confirms the equipartition theorem for confined particles, which postulates,
E = Ekin + Epot = 2Ekin . (15.69)
In flight time, however, Epot suddenly vanishes.
15.2.5.1 Calibrating the number of atoms

Experimentally, to calibrate N , we can use either the measured value of hk 2 i at T = 0, which
gives µ = EF = 4E/3 and consequently,
3
32 ~2 hk 2 i
N= . (15.70)
3 6m~ω̄
Or we determine the temperature Tg where the Boltzmann gas turns into a Fermi gas 3µ/4 =
3kB Tg ,

32 kB Tg 3
N= . (15.71)
3 ~ω̄
15.2.6 Density and momentum distribution for anharmonic potentials

15.2.6.1 Width of momentum distribution for anharmonic potentials
If the potential is non-harmonic, the widths of Fermi distributions must in general bep calculated
numerically. I.e. first g() is determined by integrating for every value of the root − U (x)
over the entire volume, where U (x) < , i.e. in the case of cylindrical symmetry,
Z
(2m)3/2 p
g() = − U (r, z)rdrdz . (15.72)
2π~3
R −1
Second the chemical potential must also be calculated numerically from N = g() eβ(−µ) + 1 d
by minimizing the function, Z
g(x/β)dx

o(Z) = βN − . (15.73)
ex /Z + 1
Finally, the rms-momentum width of a degenerate Fermi-gas is calculated from,
Z
hk 2 i E1 1 g()d
2 = = β(−µ)
. (15.74)
kF EF N EF e +1

It is important to note that the temperature cannot be obtained from ~2 k 2 /2m = 3N kB T
any more. Rather for a given hk 2 i the parameter β in the integral ((??)) must be fitted to satisfy
the equation.
Alternatively, we may assume a polynomial potential for which the density of states can be
described by g() ∝ n . Then,
R −1
hk 2 i 1 g() eβ(−µ) + 1 d T (n + 1)fn+2 (Z)
2 = R −1 = , (15.75)
kF EF g() e β(−µ) +1 d TF fn+1 (Z)
For a harmonic potential we recover the energy formula,

hk 2 i 3T f4 (Z)
2 = , (15.76)
kF TF f3 (Z)
and for hot clouds the classical limit holds,

hk 2 i n+1
2 = . (15.77)
kF βEF
Must for a single dimension the value be divided by three? ~2 hkj2 i = 2mkB T f4 (Z)/f3 (Z) setting
= ~2 k 2 /m.
For a harmonic potential g() ∝ 2 and for a linear potential g() ∝ 7/2 . Intermediate
values are possible for non isotropic traps, which are linear in some directions and harmonic in
others, e.g. for a radially quadrupolar and axially harmonic trap, we expect g() ∝ 3 and thus
E = 4N kB T . In general, we may have more complicated situations, where the trap becomes
non-harmonic beyond a certain distance from the origin. In those cases, the density of states
may be approximated by series,
g() ∝ 2 + η3 , (15.78)
where η is a small parameter, so that,
R
hk 2 i 1 (3 + η4 )(eβ(−µ) + 1)−1 d T 3f4 (Z) + 12ηf5 (Z)
2 = R = , (15.79)
kF 2 3
EF ( + η )(e β(−µ) −1
+ 1) d TF f3 (Z) + 3ηf4 (Z)
which in the classical limit gives rise to energies E = 3..4N kB T depending on the value of η.
Such effects must be considered when the time-of-flight method is used for temperatures
measurements. For example, if we underestimate g(ε) by assuming a harmonic potential at all
ε, although the potential is quadrupolar at large ε kB T , we get a wrong estimate for the
temperature Twrng = E/3N kB instead of Tcorr = E/4N kB .
15.2.6.2 Width of the density distribution for anharmonic potentials

The result also permits to calculate the rms spatial width,
X3 m f4 (Z)
ω 2 hx2j i = 3kB T . (15.80)
j=1 2 j f3 (Z)
Let us for simplicity assume ωi = ωj . So in the classical limit,
hx2j i hx2 i E1 1..1.3T
= = = . (15.81)
RF2 3RF2 3EF TF
If the potential is non-harmonic, the widths of Fermi distributions must in general be calculated
numerically. We may use the same results for the density of states and the chemical potential
as for the momentum width calculations. Then,
R −1
hx2j i E1 1 g() eβ(−µ) + 1 d
2 = = R −1 . (15.82)
RF 3EF 3EF g() eβ(−µ) + 1 d
15.2.7 Classical gas

For a harmonic potential [64] the fraction of particles with energy smaller than ηkB T is,
Z ηkB T
1 d 2 + 2η + η 2
N (η) = ≈ 1 − (15.83)
2N (~ω̄)3 0 eβ /Z − 0 2eη
Z ηkB T
1 2 d 6 + 6η + 3η 2 + η 3
E(η) = ≈ 3 − ,
2N (~ω̄)3 0 eβ /Z − 0 6eη
while for a quadrupole potential,

√ 105 + 70η + 28η 2 + 8η 3 √
N (η) ≈ −2 z √ η
+ erf( η) (15.84)
π105e
2 √ 945 + 630η + 252η 2 + 72η 3 + 16η 4 9 √
E(η) ≈ − z √ η
+ erf( η) .
9 π105e 2
15.2.7.1 Momentum distribution for a classical gas

For high temperatures, T → ∞, we should recover the ideal Boltzmann gas situation, f3/2 → id,
Z Z
1 4πρ2 dρ 1 −β(ε−µ) 2 2
ñT →∞ (k) = 3
= 2e e−βmωtr ρ /2 ρ2 dρ (15.85)
eβ [ε+mωtr ρ /2−µ]
2
(2π) 2
2π
3/2
1
= 2 e−β(ε−µ) = λ−3 6
dB atr e
β(µ−ε)
.
2πβmωtr
R
Since the chemical potential satisfies the normalization, ñT →∞ (k)d3 k = 1,
3/2 3 s 3
1 ~ωtr ~2 2 2
ñT →∞ (k) = 2 N e−βε = N e−~ k /2mkB T . (15.86)
2πβmωtr kB T 2πmkB T
This is easy to integrate by dimensions, so that,

Z Z s
∞ ∞
~2 2 2
ñT →∞ (kz ) = ñT →∞ (k)dkx dky = N e−~ kz /2mkB T . (15.87)
−∞ −∞ 2πmkB T
The rms-width of this distribution is,

√
mkB T
∆kz = . (15.88)
~
15.2.7.2 Correlating fluctuations in the number of atoms

Experimentally the atom number will fluctuate. To minimize this problem we can renormalize
it to a norm number Nnorm by multiplying the measured time-of-flight variance hx2tof i with,
hx̃2tof i
N f4 (Z̃)
CT (N ) = = (15.89)
hx2tof i
Nnorm f4 (Z)

N f4 f3−1 (Nnorm (β~ω̄)3 )
= .
Nnorm f4 f3−1 (N (β~ω̄)3 )
15.2.8 Intensive and extensive parameters

The distinction between an intensive parameter and an extensive parameter is based on the
concept that the system under investigation can be subdivided into smaller, identical and non-
interconnected entities within which the parameter in question has the same properties, such
that this parameter does not change, when the system is divided or subentities are combined 6
6
How to derive the thermodynamic potentials from the macrocanonical partition function? How to calculate
pressure and volume? [check thermodynamic derivations from Romero-Rochin [see Freddy’s thesis].
An intensive property is a global property, meaning that it is a physical property of a system

that does not depend on the system size or the amount of material in the system. Examples of
intensive properties are the temperature and the hardness of an object. No matter how small a
diamond is cut, it maintains its intrinsic hardness.
By contrast, an extensive property is one that is additive for independent, non-interacting
subsystems. The property is proportional to the amount of material in the system. For example,
both the mass and the volume of a diamond are directly proportional to the amount that is left
after cutting it from the raw mineral. Mass and volume are extensive properties, but hardness
is intensive.
The ratio of two extensive properties, such as mass and volume, is scale-invariant, and this
ratio, the density, is hence an intensive property.
Intensive parameter are: chemical potential, concentration, density (or specific gravity),
ductility, elasticity, electrical resistivity, hardness, magnetic field, magnetization, malleability,
melting point and boiling point, molar absorptivity, pressure, specific energy.
Extensive parameter are: energy, entropy, Gibbs energy, length, mass, particle number,
momentum, number of moles, volume, magnetic moment, electrical charge, weight.
15.2.9 Signatures for quantum degeneracy of a Fermi gas
Whether an atom is a fermion or a boson uniquely depends on its total spin. Halfinteger spin
particles are fermions, integer spin particles are bosons. E.g. Rb atoms have in the ground state
J = 1/2, I = 7/2, integer F , and are therefore bosons. Ca+ ions have J = 1/2 and no hyperfine
structure so that F is halfinteger, and are therefore fermions. 6 Li has half-integer F and is a
boson.
For a composite particle the quantum statistical nature may depend on the interaction
strength of the partners. For weak interaction, e.g. Feshbach the total spins of the partners
will couple to a total total spin, which determines the nature of the composite particle. A
fermion pairing with a fermion or a boson pairing with a boson will be bosons. A fermion
pairing with a boson will be a fermion. Composite trimers may be either bosonic or fermionic
depending on the coupling scheme. Can the quantum nature change with the tightness of the
binding? What is the total spin of a deeply bound molecule? [213, 33, 94], [?, 127, 206, ?]
15.2.9.1 Optical density of a Fermi gas
With the local density of a Fermi gas,
kF3
nloc = (15.90)
3π 2
the optical density is at T = 0,
Z Z RF 3/2
8σ N x2 + y 2 z2
σndy = 2 3 1− − 2 dy (15.91)
π RF −RF RF2 ZF
3/2 Z RF 3/2
8σ N x2 z2 y2
= 2 3 1− 2 − 2 1− 2 dy .
π RF RF ZF −RF RF − x2 − RF2 z 2 /ZF2
q
Writing a = RF / RF2 − x2 − RF2 z 2 /ZF2 ,
Z Z a
8σ N
σndy = 2 2 4 (1 − ỹ 2 )3/2 dỹ (15.92)
π RF a −a
2σ N p p
= 2 2 4 9a 1 − a2 − 2a3 1 − a2 + 3 arcsin a .
π RF a
In the center, a = 1,
Z
3N σ 9mω 2 N
σndy = 2 = 2 r , (15.93)
πRF kL EF
such that for EF ' 1µK we expect nloc ' 4 × 1012 cm−3 .
15.2.9.2 ’Pauli blocking’ of sympathetic cooling

For a harmonic trap U = µB = mω 2 r2 the rms-radius of a thermal cloud,
s s
2kB T kB T
rrms = = , (15.94)
mωr2 µ∂r2 B
is independent on the atomic mass. This means that a Li and a Rb cloud in the same harmonic
trap at the same temperature have the same radius. This ensures good overlap. E.g. at T =
10 µK assuming the Rb secular frequencies ωr ' 2π × 300 Hz and ωz ' 2π × 30 Hz, we expect
rrms = 16 µm and zrms = 160 µm. However below the temperature 0.5TF , which is TF ' 1 µK
for NF = 104 , the quantum pressure stops the reduction of the fermion cloud while cooling.
This evtl. reduces the overlap with the boson cloud, disconnects the two clouds and stops the
evaporative cooling. On the other hand, the interaction energy of the boson cloud also increases
its size, when the Rb cloud approaches the critical temperature Tc ' 0.6 µK for NB = 106 .
The Pauli blocking of sympathetic cooling is a signature for the advent of quantum statistics
[70, 102, 196]. It is due to a reduced mobility (or better reduced available phase space at
collisions) of the atoms and not to be confused with the prohibition of s-wave collisions due to
the Pauli exlusion principle. Furthermore, elastic collisions are suppressed [69], because atoms
cannot be scattered into occupied trap levels [130, 244, 108, 109].
• FermiBlocking1: Density distribution of a Li Fermi gas.
• FermiBlocking2: Momentum distribution of a Li Fermi gas.
• FermiBlocking3: Time-of-flight distribution of a Li Fermi gas.
• FermiBlocking4: Temperature dependence of the rms-size of a Li Fermi gas [39].
• FermiBlocking4a: Temperature dependence of the rms-size of a Li Fermi gas [39], in the

Sommerfeld approximation.
• FermiBlocking4b: Temperature dependence of the rms-size of a Li Fermi gas [39], numerical

integration of density of states.
• FermiBlocking5: Temperature dependence of ToF sizes for various atom numbers.

15.2.9.3 Superfluid suppression of sympathetic cooling

The fermions inside the bosonic cloud can be regarded as impurities. If they travel too slow,
v < c, and if the condensed fraction is too large, the motion will be frictionless and thermalization
stops. If they travel fast, quasiparticles are excited,
√ which can be removed bymevaporation. With
the typical velocity of sound in the BEC c = ~ 16πna/2mB ≈ 2 mm/s, or 2 c ≈ kB × 20 nK,
2
we see that this is no real danger.
15.2.9.4 Component separation

If the interspecies interaction h is stronger than the inter-bosonic interaction, the components
may separate [194]. Otherwise a small fermionic cloud stays inside the BEC.
15.2.9.5 Excess energy modifies 2nd moment

Independent on any model, just look deviation from Gaussian
R (interaction energy plays no role
for the fermions). Also calculate the 2nd moment U = Ekin (k)n(k)dk, where n(k) is measured
in time-of-flight and Ekin = ~2 k 2 /2m.
15.2.9.6 Modification of light scattering

The unavailability of final momentum states inhibits scattering in a similar way as the Lamb-
Dicke effect. Forward scattering is suppressed, because all small momentum states are occupied.
Furthermore, spontaneous emission is suppressed like in photonic band gaps. However, here it
is rather an atomic momentum band gap. Could it be that because scattering is suppressed,
in-situ images of fermions are hampered?
A condition for this effect to play a role is krec kF . For Li the temperature must be
kB TF = ~2 kF2 /2m = ~ω̄(6N )1/3 ~2 kL2 /2m ≈ kB × 3 µK. I.e. we need quite large Fermi gases.
15.2.9.7 Hole heating

Loss processes that remove particles from an atom trap leave holes behind in the single particle
distribution if the trapped gas is a degenerate fermion system. The appearance of holes increases
the temperature, because of an increase in the energy share per particle if cold particles are
removed. Heating is significant if the initial temperature is well below the Fermi temperature.
Heating increases the temperature to T > TF /4 after half of the systems lifetime, regardless
of the initial temperature. The hole heating has important consequences for the prospect of
observing Cooper pairing in atom traps.
15.2.10 Fermi gas in reduced dimensions

In n dimensions with the energy ε = aps + brt [161] we have to generalize the results of the last
chapter,
Γ ns + 1 Γ nt + 1 n/s+n/t
N =g 2 (kB T ) fn/s+n/t (z) . (15.95)
(2~)n an/s bn/t Γ n2 + 1
1 2
This gives for a harmonic trap where ε = 2m p +m 2 2
2 ω r and with the spin degeneracy factor
g = 1,
kB T n
N= fn (z) . (15.96)
~ω
The Fermi energy again follows from Sommerfeld’s expansion,
EF = (n!N )1/n ~ω . (15.97)
We now assume a 1D potential V = m 2 2

2 ωz r embedded in a 3D trap. A true 1D situation
arises when the atoms occupy all low-lying axial levels with the lowest radial vibrational quantum
number, i.e. EF ~ωr which gives,
ωr
N . (15.98)
ωz
Such quantum degenerate 1D fermion gases realize the so-called Luttinger liquid. One of the
hallmarks of Luttinger liquids is spin-charge separation.
Example 59 (Estimations
q for 1D): Let qus consider a Fermi gas in a very elongated
87 87 5
microtrap: ωr = 7 2π × 1.4 kHz and ωz = 7 2π × 15 Hz for Rb. With NLi = 10 the
Fermi temperature is as high as TF ' 5 µK. However we need N 100 to see 1D features.
2
1 2
Assume ε = 2m p + m4 b4 r4 ,
n

1 Γ +1
N= 4
n
(kB T )3n/4 f3n/4 (z)
(~b)n Γ 2 +1
!4/3n
Γ n2 + 1 Γ( 3n4 + 1)
EF ≈ (~b)4/3 N .
Γ n4 + 1
In 1D,
1.02
N= (kB T )3/4 f3/4 (z)
~b
EF ≈ 0.87(N ~b)4/3 .
15.3 Exercises
15.3.1 Quantum statistics of Bose-condensates
15.3.1.1 Ex: Boson or fermion?
Whether an atom is a fermion or boson solely depends on its total spin. Half-integer spin
particles are fermions, integer spin particles are bosons. For example, Rb atoms have in the
ground state J = 1/2, I = 7/2 and F integer, and therefore are bosons. Ca+ ions have J = 1/2
and no hyperfine structure, and thus are fermions. 6 Li has the half-integer F and is a boson.
Decide on the bosonic or fermionic nature of the following atoms/molecules:
85 Rb with I = 3/2 in the state 2 S
1/2
88 Sr with I = 0 in the state 1 S
0
88 Sr with I = 0 in the state 3 P
2
87 Sr with I = 9/2 in the state 1 S
0
172 Yb+ with I = 0 in the state 2 S
1/2
171 Yb+ with I = 1/2 in the state 2 S
1/2
15.3. EXERCISES 455
15.3.1.2 Ex: Monoatomic gas

Consider a classical monoatomic gas made up of N non-interacting atoms of mass m confined
in a container of volume V , at temperature T . The Hamiltonian corresponding to an atom is
given by Ĥ = (p2x + p2y + p2z )/2m.
√
a. Show that the atomic canonical partition function is ζ = V /λ3 , where λ = h/ 2πmkB T is
the thermal de Broglie wavelength.
b. Using ζ of the previous item, obtain the system’s partition function Z and the Helmholtz free
energy F . Also obtain the free energy per atom f = F/N in the thermodynamic limit N −→ ∞,
V −→ ∞, v = N/V fixed.
c. Obtain internal energy U and the gas pressure p.
d. Calculate the chemical potential and entropy per atom in the thermodynamic limit.
15.3.1.3 Ex: Generalization for arbitrary potentials in reduced dimensions

The calculation of the thermodynamic potentials can be generalized to arbitrary trapping po-
tentials and dimensions [32, 80, 66, 14, 261, 16, 15, 105, 124, 148, 145, 192, 157, 163, 82]. To do
so, we consider a generic power law potential confining an ideal Bose gas in α dimensions,
Xα xi ti
U (r) = ,
i=1 ai
and define a parameter describing the confinement power of the potential,

α Xα 1
ξ= + .
2 i=1 ti
For example, for a three-dimensional potential, α = 3. Now, for a 3D harmonic potential, ξ = 3,

and for 3D box potential, ξ = 3/2.
a. Calculate the density of states η using the equation (14.19) employing Bose functions (15.10).
b. Prove the following expressions:
bosonic potentials

N0 min(T, Tc ) ξ
= 1−
N T
c
E gξ+1 (Z) min(T, Tc ) ξ
= ξ
N kB T gξ (Z) Tc
S gξ+1 (Z) 2µ
= 4 −
N kB gξ (Z) kB T
.
C gξ+1 (Z) min(T, Tc ) ξ gξ (Z) max(T − Tc , 0)
= ξ(ξ + 1) − ξ2
N kB gξ (Z) Tc gη−1 (Z) T − Tc
CT >Tc gξ+1 (Z) g ξ (Z) C T <Tc gξ+1 (1)
= ξ(ξ + 1) − ξ2 , = ξ(η + 1)
N kB gξ (Z) gξ−1 (Z) N kB gξ (1)
∆CTc CTc− − CTc+ gξ (1)
= = ξ2
N kB N kB gξ−1 (1)
15.3.1.4 Ex: Time-of-flight distribution of a Bose-gas

a. Derive the formulae (15.25) describing the density and momentum distribution of an ultracold
Bose-gas.
b. Calculate the time-of-flight distribution of a Bose-gas as a function of temperature (i) ana-

lytically for a harmonic potential and (ii) numerically for an arbitrary potential.
15.3.2 Quantum statistics of a Fermi gas

Bibliography
[1] C. S. Adams, H. J. Lee, N. Davidson, M. Kasevich, and S. Chu, Evaporative cooling in a
crossed dipole trap, Phys. Rev. Lett. 74 (1995), 3577, .
[2] Y. Aharonov and J. Anandan, Phase change during cyclic evolution, Phys. Rev. Lett. 58
(1987), 1593, .
[3] R. Alicki, The markov master equations and the fermi golden rule, International Journal
of Theoretical Physics (1977).
[4] T. P. Altenmüller and A. Schenzle, Phys. Rev. A 49 (1994), 2016.
[5] B. P. Anderson and M. A. Kasevich, Macroscopic quantum interference from atomic tunnel
arrays, Science 282 (1998), 1686, .
[6] M. H. Anderson, J. R. Ensher, M. R. Matthews, C. E. Wieman, and E. A. Cornell,

Observation of bose-einstein condensation in a dilute atomic vapor, Science 269 (1995),
198, .
[7] M. R. Andrews, M. O. Mewes, N. J. Van Druten, D. S. Durfee, D. M. Kurn, and W. Ket-

terle, Direct, non-destructive observation of a bose condensate, Science 273 (1996), 84.
[8] M. R. Andrews, D. M. Stamper-Kurn, H. J. Miesner, D. S. Durfee, C. G. Townsend,

S. Inouye, and W. Ketterle, Erratum: Propagation of sound in a bose-einstein condensate,
Phys. Rev. Lett. 80 (1997), 2967.
[9] A. Ashkin and J. P. Gordon, Stability of radiation-pressure particle traps: an optical

earnshaw theorem, Phys. Rev. Lett. 40 (1978).
[10] N. W. Ashkroft and D. N. Mermin, Solid state physics, Hartcourt College Publishers, 1976.
[11] P. W. Atkins and R. S. Friedman, Molecular quantum mechanics, Oxford University Press,
1997.
[12] R. Bachelard, N. Piovella, and Ph. W. Courteille, Cooperative scattering and radiation
pressure force in dense atomic clouds, Phys. Rev. A 84 (2011), 013821.
[13] V. S. Bagnato, G. P. Lafyatis, A. G. Martin, E. L. Raab, R. N. Ahmad-Bitar, and D. E.

Pritchard, Continuous stopping and trapping of neutral atoms, Phys. Rev. Lett. 58 (1987),
2194.
[14] V. S. Bagnato, D. E. Pritchard, and D. Kleppner, Bose-einstein condensation in an ex-

ternal potential, Phys. Rev. A 35 (1987), 4354, .
[15] Vanderlei S. Bagnato, Simple calculation of the spatial evolution of atoms in a trap during
occurance of bose-einstein condensation, Phys. Rev. A 54 (1996), 1726.
[16] Vanderlei S. Bagnato and Daniel Kleppner, Bose-einstein condensation in low-dimensional

traps, Phys. Rev. A 44 (1991), 7439, .
457
458 BIBLIOGRAPHY
[17] A. Bambini and S. Geltman, Feshbach resonances in cold collisions of potassium atoms,
Phys. Rev. A 65 (2002), 062704, .
[18] A. Barchielli, L. Lanz, and G. M. Prosperi, A model for the macroscopic description and
continual observations in quantum mechanics, Il Nuovo Cimento 72 (1982), 79.
[19] J. Bardeen, L. N. Cooper, and J. R. Schrieffer, Microscopic theory of superconductivity,

Phys. Rev. 106 (1957), L162, .
[20] G. Barontini, C. Weber, F. Rabatti, J. Catani, G. Thalhammer, M. Inguscio, and F. Mi-

nardi, Observation of heteronuclear atomic efimov resonances, ePrints 09014584 (2009),
.
[21] G. R. Barwood and et al., Frequency measurements on optically narrowed rb-stabilized

laser diodes at 780 nm and 795 nm, Appl. Phys. B 53 (1991), 142, .
[22] A. Beige and G. C. Hegerfeldt, Projection postulate and atomic quantum zeno effect, Phys.
Rev. A 53 (1996), 53.
[23] H. C. W. Beijerinck, E. D. J. Vredenbregt, R. J. W. Stas, M. R. Doery, and J. G. C.

Tempelaars, Prospects for bose-einstein condensation of metastable neon atoms, Phys.
Rev. A 61 (2000).
[24] C. M. Bender, S. Boettcher, and P. N. Meisinger, Pt-symmetric quantum mechanics, J.

Math. Phys. 40 (1999), 2201, .
[25] C. M. Bender, Brody, and P. N. Jones, Complex extension of quantum mechanics, Phys.
Rev. Lett. 89 (2002), 270401, .
[26] H. Bender, Ph.W. Courteille, C. Marzok, C. Zimmermann, and S. Slama, Direct mea-
surement of intermediate-range casimir-polder potentials, Phys. Rev. Lett. 104 (2010),
083201.
[27] J. Bergquist, R. G. Hulet, W. M. Itano, and D. J. Wineland, Observation of quantum

jumps in a single atom, Phys. Rev. Lett. 57 (1986), 1699.
[28] M. Blatt, K.W. Boër, and W. Brandt, Bose-einstein condensation of excitons, Phys. Rev.
126 (1962), 1691.
[29] R. Blatt, P. Zoller, G. Holzmüller, and I. Siemers, Brownian motion of a parametric

oscillator: A model for ion confinement in radio frequency traps, Z. Phys. D 4 (1986),
121.
[30] I. Bloch, T. W. Hänsch, and T. Esslinger, Atom laser with a cw output coupler, Phys. Rev.
Lett. 82 (1999), 3008, .
[31] C. A. Blockley, D. F. Walls, and H. Risken, Quantum collapses and revivals in a quantized
trap, Europhys. Lett. 17 (1992), 509.
[32] Satyendra N. Bose, Plancks gesetz und lichtquantenhypothese, Z. Phys. 26 (1924), 178.
[33] T. Bourdel, J. Cubizolles, L. Khaykovich, K. M. F. Magalhaes, S. J. J. M. F. Kokkelmans,

G. V. Shlyapnikov, and C. Salomon, Measurement of the interaction energy near a feshbach
resonance in a 6 li fermi gas, Phys. Rev. Lett. 91 (2003), 020402, .
BIBLIOGRAPHY 459
[34] C. C. Bradley, C. A. Sackett, and R. G. Hulet, Analysis of in situ images of bose-einstein

condensates of lithium, Phys. Rev. A 55 (1997), 3951, .
[35] , Bose-einstein condensation of lithium: Observation of limited condensate number,

Phys. Rev. Lett. 78 (1997), 985, .
[36] C. C. Bradley, C. A. Sackett, J. J. Tolett, and R. G. Hulet, Evidence of bose-einstein

condensation in an atomic gas with attractive interactions, Phys. Rev. Lett. 75 (1995),
1687, .
[37] Léon Brillouin, La mécanique ondulatoire de schrödinger: une méthode générale de reso-
lution par approximations successives, Comptes Rendus de l’Academie des Sciences 183
(1926), 24.
[38] I. N. Bronstein and K. A. Semandjajew, Taschenbuch der mathematik, Thun und Franfurt,
Main, 1978.
[39] D. A. Butts and D. S. Rokhsar, Trapped fermi gases, Phys. Rev. A 55 (1997), 4346, .
[40] S. Bux, Ch. Gnahm, R. A. Maier, C. Zimmermann, and Ph. W. Courteille, Cavity-
controlled matter wave superradiance at the recoil limit, Phys. Rev. Lett. 106 (2011),
203601.
[41] Y. Castin and R. Dum, Bose-einstein condensates in time dependent traps, Phys. Rev.
Lett. 77 (1996), 5315, .
[42] C. M. Caves and G. J. Milburn, Quantum mechanics of measurements distributed in time

ii, Phys. Rev. A 36 (1987), 5543.
[43] J. Chabé, G. Lemarie, B. Grémaud, D. Delande, P. Szriftgiser, and J. C. Garreau, Exper-

imental observation of the anderson metal-insulator transition with atomic matterwaves,
Phys. Rev. Lett. 101 (2008), 255702, .
[44] L. L. Chase, N. Peyghambarian, G. Grynberg, and A. Mysyrowicz, Evidence for bose-

einstein condensation of biexcitons in CuCl, Phys. Rev. Lett. 42 (1979), 1231.
[45] Steve Chu, J. E. Bjorkholm, A. Ashkin, and A. Cable, Experimental observation of opti-
cally trapped atoms, Phys. Rev. Lett. 57 (1986), 314.
[46] J. I. Cirac, R. Blatt, P. Zoller, and W. D. Phillips, Laser cooling of trapped ions in a
standing wave, Phys. Rev. A 46 (1992), 2668, .
[47] J. I. Cirac, M. Lewenstein, and P. Zoller, Quantum dynamics of a laser-cooled ideal gas,
Phys. Rev. A 50 (1994), 3409, .
[48] C. Cohen-Tannoudji, B. Diu, and F. Lao, Quantum mechanics, 1, vol. 1 & 2, 1977, Dedalus:
530.12 C678q v.1 e.1.
[49] C. Cohen-Tannoudji, J. Dupont-Roc, and G. Grynberg, Atom-photon interactions, John

Wiley and Sons New York, 1992.
[50] Ph. W. Courteille, V. S. Bagnato, and V. I. Yukalov, Bose-einstein condensation of trapped

atomic gases, Laser Phys. 11 (2001), 659.
460 BIBLIOGRAPHY
[51] Ph. W. Courteille, S. Bux, E. Lucioni, K. Lauber, T. Bienaimé, R. Kaiser, and N. Piovella,
Modification of radiation pressure due to cooperative scattering of light, Euro. Phys. J. D
58 (2010), 69.
[52] Ph. W. Courteille, B. Deh, J. Fortágh, A. Günther, S. Kraft, C. Marzok, S. Slama, and
C. Zimmermann, Highly versatile atomic micro traps generated by multifrequency magnetic
field modulation, J. Phys. B 39 (2006), 1055.
[53] Ph. W. Courteille, R. S. Freeland, D. J. Heinzen, F. A. van Abeelen, and B. J. Verhaar,

Observation of a feshbach resonance in cold atom scattering, Phys. Rev. Lett. 81 (1998),
69.
[54] Ph. W. Courteille, D. J. Han, R. H. Wynar, and D. J. Heinzen, New observation of bose-
einstein condensation of 87 rb atoms in a magnetic top trap, Proc. of SPIE 3270 (1998),
116.
[55] Philippe W. Courteille, Eletrodinâmica, http://www.ifsc.usp.br/ stron-

tium/Publication/Scripts/EletroMagnetismoScript.pdf.
[56] J. Cubizolles, T. Bourdel, S. J. J. M. F. Kokkelmans, G. V. Shlyapnikov, and C. Salomon,

Production of long-lived ultracold li2 molecules from a fermi gas, Phys. Rev. Lett. 91
(2003), 240401, .
[57] M. Ben Dahan, E. Peik, Y. Castin, and C. Salomon, Bloch oscillations of atoms in an
optical potential, Phys. Rev. Lett. 76 (1996), 4508, .
[58] F. Dalfovo, S. Giorgini, L. P. Pitaevskii, and S. Stringari, Theory of bose-condensation in

trapped gases, Rev. Mod. Phys. 71 (1999), 463, cond-mat/9806038.
[59] J. Dalibard, Y. Castin, and K. Molmer, Wave-function approach to dissipative processes

in quantum optics, Phys. Rev. Lett. 68 (1992), 580.
[60] J. Dalibard and C. Cohen-Tannoudji, Dressed-atom approach to atomic motion in laser

light: The dipole force revisited, J. Opt. Soc. Am. B 2 (1985), 1707.
[61] , Laser cooling below the doppler limit by polarization gradients: Simple theoretical
studies, J. Opt. Soc. Am. B 6 (1989), 2023.
[62] N. Davidson, H. J. Lee, C. S. Adams, M. Kasevich, and S. Chu, Long atomic coherence
times in an optical dipole trap, Phys. Rev. Lett. 74 (1995).
[63] K. B. Davis, M.-O. Mewes, M. R. Andrews, N. J. van Druten, D. S. Durfee, D.M. Kurn,
and W. Ketterle, Bose-einstein condensation in an gas of sodium atoms, Phys. Rev. Lett.
75 (1995), 3969, .
[64] K. B. Davis, M.-O. Mewes, M. A. Joffe, M. R. Andrews, and W. Ketterle, Evaporative

cooling of sodium atoms, Phys. Rev. Lett. 74 (1995), 5202, .
[65] K. B. Davis, M.-O. Mewes, and W. Ketterle, An analytical model for evaporative cooling
of atoms, Appl. Phys. B 60 (1995), 155, .
[66] S. R. de Groot, G. J. Hooyman, and C. A. Ten Deldam, Proc. R. Soc. Lond. A 203 (1950),
266.
BIBLIOGRAPHY 461
[67] M. Defrise, D. W. Townsend, and R. Clack, Three-dimensional image reconstruction from

complete projections, Phys. Med. Biol. 34 (1989), 573, .
[68] G. Delannoy, S. G. Murdoch, V. Boyer, V. Josse, P. Bouyer, and A. Aspect, Understanding

the production of dual bose-einstein condensation with sympathetic cooling, Phys. Rev. A
63 (2001), 051602, .
[69] B. DeMarco and D. S. Jin, Onset of fermi-degeneracy in a trapped atomic gas, Science
285 (1999), 1703, .
[70] B. DeMarco, S. B. Papp, and D. S. Jin, Pauli blocking of collisions in a quantum degenerate
atomic fermi gas, Phys. Rev. Lett. 86 (2001), 5409, .
[71] W. Demtröder, Atoms, molecules and photons: An introduction to atomic- molecular- and
quantum physics, Springer, 2006.
[72] L. Deng, E. W. Hagley, J. Wen, M. Trippenbach, Y. Band, P. S. Julienne, J. E. Simsarian,

K. Helmerson, S. L. Rolston, and W. D. Phillips, Four-wave mixing with matter waves,
Nature 398 (1999), 218, .
[73] M. T. DePue, C. McCormick, S. L. Winoto, S. Oliver, and D. S. Weiss, Unity occupation

of sites in a 3d optical lattice, Phys. Rev. Lett. 82 (1999), 2262, .
[74] R. H. Dicke, Coherence in spontaneous radiation processes, Phys. Rev. 93 (1954), 99.
[75] F. Diedrich, J.C. Bergquist, W.I. Itano, and D.J. Wineland, Laser cooling to the zero-point
energy of motion, Phys. Rev. Lett. 62 (1989), 403, .
[76] P. A. M. Dirac, Quantized singularities in the electromagnetic field, Proc. Roy. Soc. A 133
(1931), 60.
[77] , The cosmological constants, Nature 139 (1937), 323.
[78] J. M. Doyle, B. Friederich, J. Kim, and D. Patterson, Buffer-gas loading of atoms and
molecules into a magnetic trap, Phys. Rev. A 52 (1995), .
[79] V. Dribinski, A. Ossadtchi, V. A. Mandelshtam, and H. Reisler, Reconstruction of abel-

transformable images: The gaussian basis-set expansion abel transform method, Rev. Sci.
Instr. 73 (2002), 2634, .
[80] A. Einstein, Quantentheorie des einatomigen idealen gases, S. B. Kgl. Preuss. Akad. Wiss.
35 (1924).
[81] J. Eiselt and H. Risken, Phys. Rev. A 43 (1991), 346.
[82] J. R. Ensher, D. S. Jin, M. R. Matthews, C. E. Wieman, and E. A. Cornell, Bose-einstein

condensation in a dilute gas: Measurement of energy and ground-state occupation, Phys.
Rev. Lett. 77 (1996), 4984, .
[83] U. Ernst, A. Marte, F. Schreck, J. Schuster, and G. Rempe, Bose-einstein condensation

in a pure ioffe-pritchard field configuration, Europhys. Lett. 41 (1997), 1, .
[84] W. Ertmer, R. Blatt, J. L. Hall, and M. Zhu, Laser manipulation of atomic beam velocities:
Demonstration of stopped atoms and velocity reversal, Phys. Rev. Lett. 54 (1985), 996.
462 BIBLIOGRAPHY
[85] T. Esslinger, I. Bloch, and T. W. Hänsch, Bose-einstein condensation in a quadrupople-

ioffe configuration trap, Phys. Rev. A 58 (1998), R2664.
[86] E. Fischer, Dreidimensionale stabilisierung von ladungsträger in einem vierpolfeld, Z. Phys.

156 (1959), 1.
[87] W. Fischer and I. Lieb, Funktionentheorie, Vieweg Studium, Aufbaukurs Mathematik,

1983.
[88] P. T. Fisk, M. A. Lawn, and C. Coles, Laser cooling of 171yb+ ions a linear paul trap,
Appl. Phys. B 57 (1993), 287.
[89] D. G. Fried, T. C. Killian, L. Willmann, D. Landhuis, S. C. Moss, D. Kleppner, and T. J.

Greytak, Bose-einstein condensation of atomic hydrogen, Phys. Rev. Lett. 81 (1998), 3811,
.
[90] H. Friedrich, Theoretische atomphysik, Springer, 1990.
[91] C. W. Gardiner, Quantum noise, Springer-Verlag, Berlin, 1991.
[92] B. M. Garraway, P. L. Knight, and J. Steinbach, Dissipation of quantum superpositions,

Appl. Phys. B 60 (1995), 63.
[93] S. Giorgini, L.P. Pitaevskii, and S. Stringari, Thermodynamics of a trapped bose-condensed

gas, J. Low Temp. Phys. 109 (1997), 309, .
[94] S. Giovanazzi, A. Görlitz, and T. Pfau, Tuning the dipolar interaction in quantum gases,
Phys. Rev. Lett. 89 (2002), 130401, .
[95] R. J. Glauber, Coherent and incoherent states of the radiation field, Phys. Rev. 131 (1963),
2766, .
[96] , The quantum theory of optical coherence, Phys. Rev. 130 (1963), 2529.
[97] J. P. Gordon and A. Ashkin, Motion of atoms in a radiation trap, Phys. Rev. A 21 (1980),
1606, .
[98] R. Graham and D. Walls, Spectrum of light scattered from a weakly interacting bose-
einstein condensed gas, Phys. Rev. Lett. 76 (1996), 1774.
[99] M. Greiner, O. Mandel, T. Esslinger, T. W. Hänsch, and I. Bloch, Quantum phase transi-
tion from a superfluid to a mott insulator in a gas of ultracold atoms, Nature 415 (2002),
39, .
[100] M. Greiner, C. A. Regal, and D. S. Jin, Emergence of a molecular bose-einstein condensate

from a fermi gas, Nature 426 (2003), 537, .
[101] , Probing the excitation spectrum of a fermi gas in the BCS-BEC crossover regime,
Phys. Rev. Lett. 94 (2005), 070403, .
[102] M. Greiner, C. A. Regal, C. Ticknor, J. L. Bohn, and D. S. Jin, Detection of spatial

correlations in an ultracold gas of fermions, Phys. Rev. Lett. 92 (2004), 150405, .
[103] David J. Griffiths, Introduction to quantum mechanics, Pearson Prentice Hall, New Jersey,
2005.
BIBLIOGRAPHY 463
[104] R. Grimm, M. Weidemüller, and Y. B. Ovchinnikov, Optical dipole traps for neutral atoms,
Adv. At. Mol. Opt. Phys. 42 (2000), 95, .
[105] S. Grossmann and M. Holthaus, On bose-einstein condensation in harmonic traps, Phys.
Lett. A 208 (1995), 188, .
[106] G. Grynberg, B. Lounis, P. Verkerk, J.-Y. Courtois, and C. Salomon, Quantized motion
of cold cesium atoms in two- and three-dimensional optical potentials, Phys. Rev. Lett. 70
(1993), 2249, .
[107] D. Guéry-Odelin, J. Söding, P. Desbiolles, and J. Dalibard, Is bose-einstein condensatiopn
of atomic cesium possible?, Europhys. Lett. 44 (1998), 25.
[108] S. Gupta, Z. Hadzibabic, M. W. Zwierlein, B. J. Verhaar, and W. Ketterle, Radio-frequency
spectroscopy of ultracold fermions, Sciencexpress (2003), 1, .
[109] Z. Hadzibabic, S. Gupta, C. A. Stan, C. H. Schunck, M. W. Zwierlein, K. Dieckmann,
and W. Ketterle, Fiftyfold improvement in the number of quantum degenerate fermionic
atoms, Phys. Rev. Lett. 91 (2003), 160401, .
[110] E. W. Hagley, L. Deng, M. Kozuma, J. Wen, K. Helmerson, S. L. Rolston, and W. D.
Phillips, A well-collimated quasi-continuous atom laser, Science 283 (1999), 1706, .
[111] D. S. Hall, J. R. Ensher, D. S. Jin, and et al., Recent experiments with bose-condensed
gases at JILA, Proc. SPIE 3270 (1998), 98, cond-mat/9903459.
[112] D. S. Hall, M. R. Matthews, J. R. Ensher, C. E. Wieman, and E. A. Cornell, Dynamics of
component separation in a binary mixture of bose-einstein condensates, Phys. Rev. Lett.
81 (1998), 1539, .
[113] D. J. Han, R. H. Wynar, Ph. W. Courteille, and D. J. Heinzen, Bose-einstein condensation
of large numbers of atoms in a magnetic time-averaged orbiting potential trap, Phys. Rev.
A 57 (1998), R4114.
[114] S. Haroche, J. C. Gay, and G. Grynberg (eds.), Atom traps, World Scientific, 1989.
[115] T. Hartmann, F. Keck, H. J. Korsch, and S. Mossmann, Dynamics of bloch oscillations,
New J. Phys. 6 (2004), 2.
[116] L. V. Hau, B. D. Busch, Ch. Liu, Z. Dutton, M. M. Burns, and J. A. Golovchenko, Near-
resonant spatial images of a confined bose-einstein condensates in a 4-dee magnetic bottle,
Phys. Rev. A 58 (1998), R54, .
[117] C. E. Hecht, The possible superfluid behaviour of hydrogen atom gases and liquids, Physica
25 (1959), 1159.
[118] K. Helmerson, A. Martin, and D. E. Pritchard, Laser and rf spectroscopy of magnetically
trapped neutral atoms, J. Opt. Soc. Am. B 9 (1992), 483.
[119] A. Hemmerich, Phys. Rev. Lett. 68 (1992).
[120] , Phys. Rev. Lett. 72 (1994).
[121] A. Hemmerich and T. W. Hänsch, Two-dimensional atomic crystal bound by light, Phys.
Rev. Lett. 70 (1993).
464 BIBLIOGRAPHY
[122] E. A. L. Henn, J. A. Seman, G. Roati, K. M. F. Magalh aes, and V. S. Bagnato, Emergence

of turbulence in an oscillating bose-einstein condensate, Phys. Rev. Lett. 103 (2009),
045301, .
[123] J. Herbig, T. Kraemer, M. Mark, T. Weber C. Chin, H.-C. Nägerl, and R. Grimm, Prepa-
ration of a pure molecular quantum gas, Science 301 (2003), 1510, .
[124] C. Herzog and M. Olshanii, Trapped bose gas: The canonical versus grand canonical statis-
tics, Phys. Rev. A 55 (1997), 3254, .
[125] H. F. Hess, Evaporative cooling of magnetically trapped spin-polarized hydrogen, Phys. Rev.
B 34 (1986), 3476.
[126] T. W. Hijmans, Yu. Kagan, and G. V. Shlyapnikov, Bose condensation and relaxation
explosion in magnetically trapped atomic hydrogen, Phys. Rev. B 48 (1993), 12886, .
[127] Tin-Lun Ho and E. J. Mueller, High temperature expansion applied to fermions near fes-
hbach resonance, Phys. Rev. Lett. 92 (2004), 160404, .
[128] E. Hodby, S. T. Thompson, C. A. Regal, M. Greiner, A. C. Wilson, D. S. Jin, E. A. Cornell,

and C. E. Wieman, Production efficiency of ultracold feshbach molecules in bosonic and
fermionic systems, Phys. Rev. Lett. 94 (2005), 120402, .
[129] M. J. Holland, D. S. Jin, M. L. Ciafalo, and J. Cooper, Emergence of interaction effects

in bose-einstein condensation, Phys. Rev. Lett. 78 (1997), 3801, .
[130] M. Houbiers, H. T. C. Stoof, W. I. McAlexander, and R. G. Hulet, Elastic and inelastic

collisions of li-6 atoms in magnetic and optical traps, Phys. Rev. A 57 (1998), R1497, .
[131] K. Huang, Statistical mechanics, John Wiley and Sons, 1987.
[132] R. Huesmann, Ch. Balzer, Ph. W. Courteille, W. Neuhauser, and P. E. Toschek, Single-
atom interferometry, Phys. Rev. Lett. 82 (1999), 1611.
[133] S. Inouye, M. R. Andrews, J. Stenger, H.-J. Miesner, D. M. Stamper-Kurn, and W. Ket-

terle, Observation of feshbach resonances in a bose-einstein condensate, Nature 392 (1998),
151, .
[134] S. Inouye, A. P. Chikkatur, D. M. Stamper-Kurn, J. Stenger, D. E. Pritchard, and W. Ket-

terle, Superradiant rayleigh scattering from a bose-einstein condensate, Science 285 (1999),
571, .
[135] S. Inouye, T. Pfau, S. Gupta, A. P. Chikkatur, A. Görlitz, D. E. Pritchard, and W. Ket-

terle, Phase-coherent amplification of atomic matter-waves, Nature 402 (1999), 641, .
[136] W. M. Itano, J. C. Bergquist, J. J. Bollinger, J. M. Gilli-Gans, D. J. Heinzen, F. L. Moore,

M. G. Raizen, and D. J. Wineland, Phys. Rev. A 47 (1993), 3554, .
[137] W. M. Itano, J. J. Bollinger, and D. J. Wineland, Quantum zeno effect, Phys. Rev. A 41
(1990), 2295, .
[138] W. M. Itano and D. J. Wineland, Laser cooling of ions stored in harmonic and penning
traps, Phys. Rev. A 25 (1982), 35, .
BIBLIOGRAPHY 465
[139] J. D. Jackson, Classical electrodynamics, John Wiley and Sons, 1999.
[140] J. Javanainen and J. Ruostekoski, Off-resonance light scattering from low-temperature bose
and fermi gases, Phys. Rev. A 52 (1995), 3033, .
[141] E. T. Jaynes and F. W. Cummings, Proc. IEEE 51 (1963), 89.
[142] D. S. Jin, J. R. Ensher, M. R. Matthews, C. E. Wieman, and E. A. Cornell, Collective

excitations of a bose-einstein condensate in a dilute gas, Phys. Rev. Lett. 77 (1996), 420,
.
[143] S. Jochim, M. Bartenstein, A. Altmeyer, G. Hendl, C. Chin, J. Hecker Denschlag, and

R. Grimm, Pure gas of optically trapped molecules created from fermionic atoms, Phys.
Rev. Lett. 91 (2003), 240402, .
[144] W. Ketterle and N. J. Van Druten, Bose-einstein condensation of a finite number of

particles trapped in one or three dimensions, Phys. Rev. A 54 (1996), 656.
[145] , Evaporative cooling of trapped atoms, Adv. At. Mol. Opt. Phys. 37 (1996), 181.
[146] W. Ketterle, D. S. Durfee, and D. M. Stamper-Kurn, Making, probing and understanding

bose-einstein condensates, Proc. Int. School of Phys. Enrico Fermi CXL (1999), 67, .
[147] A. H. Kiilerich and et al., Phys. Rev. A 92 (2015), 032124.
[148] K. Kirsten and D. J. Toms, Bose-einstein condensation of atomic gases in a general

harmonic-oscillator confining potential trap, Phys. Rev. A 54 (1996), 4188.
[149] C. Kittel, Elementary statistical physics, (1976).
[150] S. J. J. M. F. Kokkelmans, B. J. Verhaar, and K. Gibble, Prospects for bose-einstein

condensation in cesium, Phys. Rev. Lett. 81 (1998), 951, .
[151] K. Kopitzky, Einführung in die festkörperphysik, Teubner, 1986.
[152] M. Kozuma, L. Deng, E. W. Hagley, J. Wen, R. Lutwak, K. Helmerson, S. L. Rolston, and

W. D. Phillips, Coherent splitting of bose-einstein condensed atoms with optically induced
bragg diffraction, Phys. Rev. Lett. 82 (1999), 871.
[153] M. Kozuma, Y. Suzuki, Y. Torii, T. Saguira, T. Kaga, E. W. Hagley, and L. Deng, Phase-
coherent amplification of matter-waves, Science 286 (1999), 2309, .
[154] T. Kraemer, M. Mark, P. Waldburger, J. G. Danzl, C. Chin, B. Engeser, A. D. Lange,

K. Pilch, A. Jaakkola, H.-C. Nägerl, and R. Grimm, Evidence for efimov quantum states
in an ultracold gas of cesium atoms, Nature 428 (2005), 155.
[155] Hendrik A. Kramers, Wellenmechanik und halbzählige quantisierung, Zeitschrift für Physik
39 (1926), 828.
[156] A. Lakhtakia, Positive and negative goos-hänchen shifts and negative phase-velocity medi-
ums (alias left-handed materials), ePrints:Physics0305133 (2003).
[157] L. D. Landau, Butterworth-Heinemann, 1937.

466 BIBLIOGRAPHY
[158] J. Lawall, S. Kulin, B. Saubamea, N. Bigelow, M. Leduc, and C. Cohen-Tannoudji, Three-

dimensional laser cooling of helium beyond the single-photon recoil limit, Phys. Rev. Lett.
75 (1995), 4194.
[159] H. J. Lee, C. S. Adams, M. Kasevich, and S. Chu, Raman cooling of atoms in an optical
dipole trap, Phys. Rev. Lett. 76 (1996), 2658, .
[160] U. Leonhardt and H. Paul, Measuring the quantum state of light, Prog. Quant. Electr. 19
(1995), 89.
[161] Mingzhe Li, Zijun Yan, Jincan Chen, Lixuan Chen, and Chuanhong Chen, Thermodynamic
properties of an ideal fermi gas in an external potential with U = brt in any dimensional
space, Phys. Rev. A 58 (1998), 1445, .
[162] Jia Ling Lin and J. P. Wolfe, Bose-einstein condensation of paraexcitons in stressed Cu2O,
Phys. Rev. Lett. 71 (1993), 1222, .
[163] F. London, On the bose-einstein condensation, Nature 54 (1938), 947.
[164] R. Loudon, The quantum theory of light, Clarendon Press Oxford, 1982.
[165] R. Loudon and P. L. Knight, J. Mod. Opt. 34 (1987), 709.
[166] R. V. E. Lovelace, C. Mehanian, T. J. Tommila, and D. M. Lee, Magnetic confinement of

a neutral gas, Nature 318 (1985), 30.
[167] K. M. Madison, F. Chevy, W. Wohlleben, and J. Dalibard, Vortex formation in a stirred

bose-einstein condensate, Phys. Rev. Lett. 84 (1999), 806, .
[168] A. G. Martin, K. Helmerson, V. S. Bagnato, G. P. Lafyatis, and D. E. Pritchard, Rf

spectroscopy of trapped neutral atoms, Phys. Rev. Lett. 61 (1988), 2431, .
[169] N. Marzari, A. A. Mostofi, J. R. Yates, I. Souza, and D. Vanderbilt, Maximally localized

wannier functions: Theory and applications, Rev. Mod. Phys. 84 (2012), 1419.
[170] N. Masuhara, J. M. Doyle, J. C. Sandberg, D. Kleppner, T. J. Greytak, H. F. Hess, and

G. P. Kochanski, Evaporative cooling of spin-polarized atomic hydrogen, Phys. Rev. Lett.
61 (1988).
[171] M. R. Matthews, B. P. Anderson, P. C. Haljan, D. S. Hall, M. J. Holland, J. E. Williams,

C. E. Wieman, and E. A. Cornell, Watching a superfluid untwisting itself, Phys. Rev. Lett.
83 (1999), 3358, .
[172] A. R. McGurn, K. T. Christensen, F. M. Mueller, and A. A. Maradudin, Anderson localiza-

tion in one-dimensional randomly disordered optical systems that are periodic on average,
Phys. Rev. B 47 (1993), 13120, .
[173] J. Meixner and F.W. Schäfke, Mathieu’sche funktionen und sphäroidfunktionen, Springer-
Verlag Berlin (1954).
[174] M.-O. Mewes, M. R. Andrews, D. M. Kurn, D. S. Durfee, C. G. Townsend, and W. Ketterle,

Output coupler for bose-einstein condensed atoms, Phys. Rev. Lett. 78 (1997), 582, .
BIBLIOGRAPHY 467
[175] M.-O. Mewes, M. R. Andrews, N. J. van Druten, D. M. Kurn, D. S. Durfee, and W. Ket-
terle, Bose-einstein condensation in a tightly confining dc magnetic trap, Phys. Rev. Lett.
77 (1996), 416, .
[176] M.-O. Mewes, M. R. Andrews, N. J. van Druten, D. M. Kurn, D. S. Durfee, C. G.
Townsend, and W. Ketterle, Collective excitations of a bose-einstein condensation in a
magnetic trap, Phys. Rev. Lett. 77 (1996), 988, .
[177] A. L. Migdall, J. V. Prodan, W. D. Phillips, T. H. Bergman, and H. J. Metcalf, First
observation of magnetically trapped neutral atoms, Phys. Rev. Lett. 54 (1985), 2596.
[178] G. Milburn, Intrinsic decoherence in quantum mechanics, Phys. Rev. A 39 (1991), 5401,
.
[179] J. D. Miller, R. A. Cline, and D. J. Heinzen, Far-off-resonance optical trapping of atoms,
Phys. Rev. A 47 (1993), R4567, .
[180] Randell L. Mills, The grand unified theory of classical quantum mechanics.
[181] P. W. Milonni and R. W. Boyd, Momentum of light in a dielectric medium, Adv. Opt.
Phot. 2 (2010), 519, .
[182] B. Misra and E. C. G. Sudarshan, The zeno paradox in quantum theory, J. Math. Phys.
18 (1977), 756.
[183] B. R. Mollow, Pure-state analysis of resonant light scattering: Radiative damping, satura-
tion, and multiphoton effects, Phys. Rev. A 12 (1975), 1919.
[184] K. Mølmer, Y. Castin, and J. Dalibard, Monte-carlo wave-function method in quantum
optics, J. Opt. Soc. Am. B 10 (1993), 524, .
[185] C. Monroe, E. A. Cornell, C. A. Sackett, C. J. Myatt, and C. E. Wieman, Measurement
of cs-cs elastic scattering at t=30uk, Phys. Rev. Lett. 70 (1993), 414, .
[186] H. Moritz, T. Stöferle, K. Günter, M. Köhl, and T. Esslinger, Confinement induced
molecules in a 1d fermi gas, Phys. Rev. Lett. 94 (2005), 210401, .
[187] A. Mosk, S. Jochim, H. Moritz, Th. Elsässer, and M. Weidemüller, Resonator-enhanced
optical dipol trap for fermionic lithium atoms, Opt. Lett. 26 (2001), 1837.
[188] G. M. Moy, J. J. Hope, and C. M. Savage, Born and markov approximations for atom
lasers, Phys. Rev. A 59 (1999), 667.
[189] M. Mudrich, S. Kraft, K. Singer, R. Grimm, A. Mosk, and M. Weidemüller, Sympathetic
cooling with two atomic species in an optical trap, Phys. Rev. Lett. 88 (2002), 253001.
[190] C. J. Myatt, E. A. Burt, R. W. Ghrist, E. A. Cornell, and C. E. Wieman, Production
of two overlapping bose-einstein condensates by sympathetic cooling, Phys. Rev. Lett. 78
(1997), 586, .
[191] W. Nagourney, J. Sandberg, and H. G. Dehmelt, Shelved optical electron amplifier: Ob-
servation of quantum jumps, Phys. Rev. Lett. 56 (1986), 2797.
[192] R. Napolitano, J. De Luca, V. S. Bagnato, and G. C. Marques, Effect of finite numbers in
the bose-einstein condensation of a trapped gas, Phys. Rev. A 55 (1997), R3954, .
468 BIBLIOGRAPHY
[193] N. R. Newbury, C. J. Myatt, E. A. Cornell, and C. E. Wieman, Gravitational sisyphus

cooling of 87rb in a magnetic trap, Phys. Rev. Lett. 74 (1995), 2196, .
[194] N. Nygaard and K. Molmer, Component separation in harmonically trapped boson-fermion
mixtures, Phys. Rev. A 59 (1999), 2974, .
[195] K. M. OHara, S. R. Granade, M. E. Gehm, T. A. Savard, S. Bali, C. Freed, and J. E.
Thomas, Ultrastable co2 laser trapping of lithium fermions, Phys. Rev. Lett. 82 (1999),
4204.
[196] K. M. OHara, S. L. Hemmer, M. E. Gehm, S. R. Granade, and J. E. Thomas, Observation
of a strongly interacting degenerate fermi gas of atoms, Science 298 (2002), 2179, .
[197] M. Ö. Oktel and Ö. E. Müstecaplýoglu, Electromagnetically induced left-handedness in a
dense gas of three-level atoms, Phys. Rev. A 70 (2004), 053806, .
[198] S. Ospelkaus, A. Pe’er, K.-K. Ni, J. J. Zirbel, and B. Neyenhuis, Ultracold dense gas of
deeply bound heteronuclear molecules, ePrints (2003), 0802.1093, .
[199] R. Ozeri, J. Steinhauer, N. Katz, and N. Davidson, Direct observation of the phonon energy
in a bose-einstein condensate by tomographic imaging, Phys. Rev. Lett. 88 (2002), 220401,
.
[200] A. S. Parkins and D. F. Walls, The physics of trapped dilute gas bose-einstein condensates,
Phys. Rep. 303 (1998), 1.
[201] W. Paul, O. Osberghaus, and E. Fischer, Ein ionenkäfig, Forschungsberichte des
Wirtschafts- und Verkehrsministerium Nordrhein-Westfalen, Westdeutscher Verlag Köln-
Opladen (1958).
[202] J. P. Paz and W. H. Zurek, Environnement-induced decoherence and the transition from
quantum to classical, Lect. Notes in Phys. 587 (2002), 77, .
[203] E. Peik, M. Ben Dahan, I. Bouchoule, Y. Castin, and C. Salomon, Bloch oscillations of
atoms, adiabatic rapid passage, and monokinetic atomic beams, Phys. Rev. A 55 (1997),
2989, .
[204] F. M. Penning, Die glimmentladung bei niedrigem druck zwischen koaxialen zylindern in
einem axialen magnetfeld, Physica 3 (1936), 873.
[205] W. Petrich, M. H. Anderson, J. R. Ensher, and E. A. Cornell, Stable, tightly confining
trap for evaporative cooling of neutral atoms, Phys. Rev. Lett. 74 (1995), 3352, .
[206] D. S. Petrov, Three-boson problem near a narrow feshbach resonance, Phys. Rev. Lett. 93
(2004), 143201, .
[207] W. D. Phillips and H. Metcalf, Laser deceleration of an atomic beam, Phys. Rev. Lett. 48
(1982).
[208] J. D. Prestage, G. J. Dick, and L. Maleki, New ion trap for frequency standard applications,
J. Appl. Phys. 66 (1989), 1013.
[209] J. D. Prestage, R. L. Tjoelker, R. T. Wang, G. J. Dick, and L. Maleki, Hg+ trapped ion
standard with the superconducting cavity maser oscillator, IEEE Trans. Instr. and Meas.
42 (1993), 200.
BIBLIOGRAPHY 469
[210] M. Prevedelli, F. S. Cataliotti, E. A. Cornell, J. R. Ensher, C. Fort, L. Ricci, G. M. Tino,

and M. Inguscio, Trapping and cooling of potassium isotopes in a double-magneto-optical-
trap apparatus, Phys. Rev. A 59 (1999), 886, .
[211] D. E. Pritchard, Cooling neutral atoms in a magnetic trap for precision spectroscopy, Phys.
Rev. Lett. 51 (1983), 1336, .
[212] Ravinder Puri, Mathematical methods of quantum optics, Springer Series in Optical Sci-
ences (2000).
[213] C. A. Regal, C. Ticknor, J. L. Bohn, and D. S. Jin, Tuning p-wave interactions in an

ultracold fermi gas of atoms, Phys. Rev. Lett. 90 (2003), 053201, .
[214] M. Reich, U. Sterr, and W. Ertmer, Scheme for measuring a berry-phase in an atom
interferometer, Phys. Rev. A 47 (1993), 2518, .
[215] G. Roati, C. D’Errico, L. Fallani, M. Fattori, C. Fort, M. Zaccanti, G. Modugno, M. Mod-

ugno, and M. Inguscio, Anderson localization of a non-interacting bose-einstein conden-
sate, Nature 453 (2008), 0895, .
[216] J. R. Rubbmark, M. M. Kash, M. G. Littman, and D. Kleppner, Dynamical effects at

avoided level crossings: A study of the landau-zehner effect using rydberg atoms, Phys.
Rev. A 23 (1981), 3107.
[217] P. A. Ruprecht, M. J. Holland, K. Burnett, and M. Edwards, Time-dependent solutions of

the nonlinear schrödinger equation for bose-condensed trapped neutral atoms, Phys. Rev.
A 51 (1995), 4704.
[218] C. A. Sackett, J. M. Gerton, M. Welling, and R. G. Hulet, Measurements of collective

collapse in a bose-einstein condensate with attractive interactions, Phys. Rev. Lett. 82
(1999), 876, .
[219] M. Samoylova, N. Piovella, M. Holynski, Ph.W. Courteille, and R. Bachelard, One-

dimensional photonic band gaps in optical lattices, Annual Review of Cold Atoms and
Molecules 2 (2014), 193.
[220] M. Samoylova, N. Piovella, G. Robb, R. Bachelard, and Ph. W. Courteille, Synchronisation

of bloch oscillations by a ring cavity, Opt. Exp 23 (2015), 14823.
[221] Th. Sauter, R. Blatt, W. Neuhauser, and P.E. Toschek, Observation of quantum jumps,
Phys. Rev. Lett. 57 (1986), 1696, .
[222] , Quantum jumps observed in the fluorescence of a single ion, Opt. Comm. 60
(1986), 287, .
[223] M. R. Schafroth, Theory of superconductivity, Phys. Rev. 96 (1954), 1442.
[224] A. Schenzle, R. G. DeVoe, and R. G. Brewer, Possibility of quantum jumps, Phys. Rev. A
33 (1986), 2127, .
[225] C. A. Schrama, E. Peik, W. W. Smith, and H. Walther, Novel miniature ion traps, Opt.
Comm. 101 (1993), 32.
470 BIBLIOGRAPHY
[226] M. Schubert, I. Siemers, and R. Blatt, Line shape of three-level ions in paul traps, Phys.
Rev. A 39 (1989), 5098.
[227] Jian Qi Shen, Negatively refracting atomic vapour, J. Mod. Opt. 53 (2006), 2195, .
[228] Jian Qi Shen, J. Almlöf, and S. He, Negative permeability in a λ-type three level atomic
vapor, Appl. Phys. A 87 (2006), 291, DOI.
[229] Jian-Qi Shen, Zhi-Chao Ruan, and Sailing He, How to realize a negative refractive index
material at the atomic level in an optical frequency range, Journal of Zhejiang University
SCIENCE 5 (2004), 1322, .
[230] G. V. Shlyapnikov, J. T. M. Walraven, U. M. Rahmanov, and M. W. Reynolds, Decay-

kinetics and bose-condensation in a gas of spin-polarized triplett helium, Phys. Rev. Lett.
73 (1994), 3247.
[231] S. Slama, S. Bux, G. Krenz, C. Zimmermann, and Ph. W. Courteille, Superradiant rayleigh
scattering and collective atomic recoil lasing in a ring cavity, Phys. Rev. Lett. 98 (2007),
053603.
[232] J. Söding, D. Guéry-Odelin, P. Desbiolles, F. Chevy, H. Inamori, and J. Dalibard, Three-

body decay of a rubidium bose-einstein condensate, Appl. Phys. B 69 (1999), 257, .
[233] M. D. Srinivas and E. B. Davies, Photon counting probabilities in quantum optics, Optica
Acta 28 (1981), 981.
[234] D. M. Stamper-Kurn and W. Ketterle, Spinor condensates and light scattering from bose-
einstein condensates, Proc. Les Houches Summer School, Session LXXII (2000), .
[235] J. Stenger, S. Inouye, D. M. Stamper-Kurn, H.-J. Miesner, A. P. Chikkatur, and W. Ket-

terle, Spin domains in ground state spinor bose-einstein condensates, Nature 396 (1998),
345.
[236] S. Stenholm, The semiclassical theory of laser cooling, Rev. Mod. Phys. 58 (1986), 699,
.
[237] D. Stoler, Equivalence class of minimum uncertainty wavepackets, Phys. Rev. D 1 (2006),
3217, .
[238] , Photon antibunching and possible ways to observe it, Phys. Rev. Lett. 33 (2006),
1397, .
[239] H. T. C. Stoof, Atomic bose gas with a negative scattering length, Phys. Rev. A 49 (1995),
3824, .
[240] William C. Stwalley and L. H. Nosanow, Possible new quantum systems, Phys. Rev. Lett.
36 (1976), 910.
[241] W. R. Theis, Grundzüge der quantentheorie, Teubner Studienbücher, 1985.
[242] L. H. Thomas, Nature 117 (1926), 514, .
[243] U. M. Titulaer and R. J. Glauber, Correlation functions for coherent fields, Phys. Rev.
140 (1999).
BIBLIOGRAPHY 471
[244] A. G. Truscott, K. E. Strecker, W., I. McAlexander, Guthrie, B. Partridge, and R. G.

Hulet, Observation of fermi pressure in a gas of trapped atoms, Science 291 (2001), 2570,
.
[245] D. V. van Coevorden, R. Sprik, A. Tip, and A. Lagendijk, Photonic band gap structure of
atomic lattices, Phys. Rev. Lett. 77 (1996), 2412, .
[246] J. Vanier and C. Audoin, The quantum physics of atomic frequency standards, Adam
Hilger, Bristol and Philadelphia, 1989, Dedalus: 539.7 V258q v.1.
[247] J. von Neumann, Mathematical foundations of quantum mechanics, Princeton University
Press, 1955.
[248] V. Vuletic, A. J. Kerman, Cheng Chin, and S. Chu, Observation of low-field feshbach
resonances in collisions of cesium atoms, Phys. Rev. Lett. 82 (1999), 1406, .
[249] I. Waki, S. Kassner, G. Birkl, and H. Walther, Observation of ordered structures of laser-
cooled ions in a quadrupole storage ring, Phys. Rev. Lett. 68 (1993), 2007.
[250] S. Wallentowitz and W. Vogel, Reconstruction of the quantum mechanical state of a trapped
ion, Phys. Rev. Lett. 75 (1995), 2932.
[251] J. Walz, (1991).
[252] J. Walz, I. Siemers, M. Schubert, W. Neuhauser, and R. Blatt, Motional stability of a
nonlinear parametric oscillator: Ion storage in the rf octupole trap, Europhys. Lett. 21
(1993), 183.
[253] C. J. Weiner, J. J. Bollinger, F. L. Moore, and D. J. Wineland, Electrostatic modes as a
diagnostic, Phys, Rev. A 49 (1994), 3842.
[254] Gregor Wentzel, Eine verallgemeinerung der quantenbedingungen für die zwecke der
wellenmechanik, Zeitschrift für Physik 38 (1926), 518.
[255] D. J. Wineland and H. G. Dehmelt, Proposed 1014 δn < n laser fluorescence spectroscopy
on tl+ mono-ion oscillator, Bull. Am. Phys. Soc. 20 (1975), 637.
[256] D. J. Wineland and W. M. Itano, Laser cooling of atoms, Phys. Rev. A 20 (1979), 1521,
.
[257] E. M. Wright and D. F. Walls, Collapses and revivals of bose-einstein condensates formed
in small atomic samples, Phys. Rev. Lett. 77 (1996), 2158, .
[258] Huang Wu and Ch. J. Foot, Direct simulation of evaporative cooling, J. Phys. B 29 (1996),
L321, .
[259] Jin-Hui Wu, M. Artoni, and G. C. La Rocca, Controlling the photonic band structure of
optically driven cold atoms, J. Opt. Soc. Am. B 25 (2008), 1840, .
[260] K. Xu, T. Mukaiyama, J. R. Abo-Shaeer, J. K. Chin, D. E. Miller, and W. Ketterle,
Formation of quantum-degenerate sodium molecules, Phys. Rev. Lett. 91 (2003), 210402,
.
[261] Ziyun Yan, Bose-einstein condensation of a trapped gas in n dimensions, Phys. Rev. A 59
(1999), 4657, .
472 BIBLIOGRAPHY
[262] Deshui Yu, Photonic band structure of the three-dimensional 88 sr atomic lattice, Phys.
Rev. A 84 (2011), 043833, .
[263] P. Zoller, M. Marte, and D. F. Walls, Quantum jumps in atomic systems, Phys. Rev. A
35 (1987), 198, .
[264] M. W. Zwierlein, C. A. Stan, C. H. Schunck, S. M. F. Raupach, S. Gupta, Z. Hadzibabic,

and W. Ketterle, Observation of bose-einstein condensation of molecules, Phys. Rev. Lett.
91 (2003), 250401, .
Index
Q atomic units, 7
função, 293 Autler-Townes splitting, 255, 263
LS-coupling, 122 avoided crossing, 255
jj-coupling, 122 azimuthal equation, 104
θ transform, 69
{3j}-symbols, 8 Baker-Hausdorff
{6j}-symbol, 8 formula de, 93
{9j}-symbol, 8 Baker-Haussdorff
g-factor, 176 formula de, 293
., 63 Bargmann state, 94
baryons, 188
permeability basis, 48
negative, 264 Biot-Savart law, 179
absorção black-body radiation, 24
coeficiente de, 423 Bloch
imagem por, 423 Felix, 45
absorption, 148 Bloch equation
absorption coefficient, 30 optical, 247
absorption rate, 28 Bloch function, 154
acoplamento forte, 418 Bloch oscillation, 159
adiabática Bloch state, 158
transferência, 341 Bloch theorem, 153
adiabático Bloch vector, 45, 250
potencial, 340, 351 generalized, 261
adiabatic sweeps, 264 Bloch-Lindbladt
Aharonov-Bohm equação de, 301
efeito de, 374 Bloch-Siegert
alargamento Doppler, 309 deslocamento de, 286
alargamento por pressão, 308 Bloch-Siegert shift, 248
amplificador quântico, 318, 357, 360 Bogolubov
angular momentum quantum number, 105 transformação de, 327
aniquilação Bohr
operador de, 284 magneton, 20
ansatz, 37 Niels, 12, 20
anti-particle, 170 postulates, 20
antibunching, 309 Bohr magneton, 176
aprisionamento de radiação, 402 Bohr radius, 107
aprisionamento forte Boltzmann
regime de, 350 Ludwig, 24
aquecimento Boltzmann distribution law, 28
taxa de, 405 Boltzmann factor, 25
Aristotle, 12 Boltzmann gas, 442
armadilha de Paul linear, 418 bombas
atom laser, 385 problema de testar, 370
atom optics, 385 Born
473
474 INDEX
aproximação de, 364 contraste de fase

Max, 39, 42 imagem por, 423
Born approximation, 17 contraste de polarização
Bose function, 435 imagens de, 424
Bose-Einstein condensation, 385, 434 cooperation number, 118
Bose-Einstein distribution, 433 Copenhague
boson, 210, 433, 451 interpretação de, 353
bosonic stimulation, 433 correlação
bra, 43 função de, 306, 309
Bremsstrahlung, 26, 227 correspondence principle, 62
Brillouin Coulomb
Léon, 135 calibre de, 284
Brillouin zone, 154 Coulomb integral, 213
bunching de fótons, 306 Coulomb operator, 224
Coulomb potential, 179
caótica criação
linha, 309 operador de, 284
centrifugal potential, 106 critical temperature, 433
charge conjugation, 69 crossing
charge conservation, 69 avoided, 191
chemical potential, 221 crossover
classical optical, 385 real, 191
Clebsch-Gordan coefficient, 8, 121 cruzamento evitado, 412
coherence, 247 CSCO, 51
coherent state, 94 cyclotron frequency, 202
colapso e renascimento quântico, 291
collision, 97 dark resonance, 265
collision radius, 257 dark resonances, 263
collision rate, 257 dark-ground imaging, 423
column-integrated, 425 Darwin
commutator, 41, 47 Sir Charles Galton, 181
complete, 48 de Broglie
complete set of commuting operators, 51 Louis, 37
completeness relation, 242 de Broglie wave, 387
composite particle, 451 de Broglie wavelength
compressão thermal, 388, 433
operador de, 326 Debye law, 34
comprimido Debye model, 33
estado, 328 Debye temperature, 34
Compton decoerência, 355
espalhamento de, 304 decoerência quântica, 357
Compton effect, 27 degeneracy, 48
Compton scattering, 27 degenerate Fermi gas, 441
Compton wavelength, 168, 181 degrees of freedom, 51
conservation law, 67 Dehmelt
constant of motion, 67 Hans, 367
continuity equation, 39, 69 Democritus, 11, 12
contrast, 81 densidade óptica, 423
INDEX 475
density functional, 218 electric susceptibility, 29

density functional theory, 218 electromagnetically induced absorption, 265
density of states, 219, 390, 434 electromagnetically induced transparency, 265
density operator, 240 emission rate
density-of-states, 146 spontaneous, 28
descompressão adiabática, 399 stimulated, 28
diamagnetic term, 196 energy density, 22
Dicke cooperatividade, 116 entropy, 241
dielectric constant, 29 equipartition theorem, 448
dipolar Erwin
momento elétrico, 339 Schrödinger, 58
dipolar gradient force, 391 espalhamento interespécies
dipole moment, 114 comprimento de, 414
Dirac espectro, 309
Paul, 43, 114 estabilidade
Dirac equation, 177 diagrama de, 417
dispersion relation, 160 estresse de Maxwell
displacement operator, 93 tensor de, 342
divisor de feixe, 313 evaporação, 409
Doppler evaporação forçada, 410
efeito, 351 evaporation
limite de resfriamento, 392 run-away, 412
Doppler broadening, 258 evolution operator, 63
Doppler cooling, 391 exchange degeneracy, 210
Doppler effect, 256 exchange energy, 214, 215
Doppler limit, 388 exchange integral, 215
Doppler shift, 258 exchange operator, 224
Doppler-free spectroscopy, 256 exchange symmetry, 210
duality principle, 385 exciton, 431
Dulong-Petit law, 33 exotic atom, 187
dynamic Stark shift, 255 explosão de relaxação, 408
dynamic variable, 44 extensive parameter, 450
Dyson series, 143 external tensorial product, 55
extinction coefficient, 29
efetivo
hamiltoniano, 357 Fano resonance, 264
Ehrenfest far off-resonance optical trap, 403
Paul, 42 fase, 371
teorema de, 354 fase de
Ehrenfest theorem, 62 Berry, 371
Ehrenfest’s theorem, 42 fator de overlap de, 350
eigenfunction, 48 Fermi
eigenvalue, 41, 48 Enrico, 147
eigenvector, 48 Fermi contact term, 184
EIT, 265 Fermi energy, 217
elástico Fermi function, 435
espalhamento, 304, 347 Fermi gas model, 217
elastic collisions, 257 Fermi’s Golden Rule, 147
476 INDEX
Fermi-Dirac distribution, 433 deslocamento, 409

fermion, 210, 433, 451 Gross-Pitaevskii equation, 145
fine structure, 167, 174 Grotian diagram, 249
fine structure constant, 167 gyromagnetic ratio, 176
fluorescência ressonante, 309
flutuação do vácuo, 304 Haas-van Alphen effect, 203
flux of electromagnetic energy, 22 hadron, 188
hadronic atom, 188
flux quantum, 203
Hall effect
Fock
quantum, 203
estado de, 283
Hamilton operator, 41
Vladimir Aleksandrovich, 88, 111
Hamiltonian, 41
Fock state, 88
Hanbury-Brown-Twiss
Fokker-Planck
experimento de, 306
equação de, 345
hard core approximation, 258
força de pressão radiativa, 343
harmonic oscillator, 87
força eletromagnética, 337
Hartree
força gravitacional, 337
Douglas Rayner, 218
FORT, 404
Hartree method, 222
Fourier reconstruction, 425
Hartree-Fock equation, 224
fraca
Hartree-Fock method, 222
interação, 362
Hartree-Fock-Roothaan equation, 225
Franck-Condon
heat capacity, 437
transição de, 348
Heisenberg equation, 59, 252
Franck-Hertz experiment, 26
Heisenberg picture, 59
Fresnel’s formula, 81
helium, 212
Frisch
Hermitian operator, 44, 47
Otto Robert, 341
Hilbert space, 43, 47, 48
fugacity, 434
histórias decoerentes, 357
função P , 294 hole heating, 453
função Q, 294 homodyne detection, 320
homodyne tomography, 321
Galilei boost, 65, 69
homogeneity
Galilei transform, 65
spatial, 68
gauge field, 66
temporal, 67
gauge transform, 69
homogeneous broadening, 256
gauge transformation, 66
hyperfine splitting, 183
Gedankenexperiment, 67
hyperfine structure, 183
Gerlach
Paschen-Back effect of the, 200
Walter, 20
Zeeman effect of the, 199
Walther, 114
Glauber image reconstruction, 425
Roy, 93 impedance of free space, 23
Glauber formula, 93 induced emission, 148
Glauber state, 94 inelástico
gradiente dipolar espalhamento, 304, 312
força de, 342, 343 information entropy, 241
gravidade, 405 Inglis-Teller limit, 191
gravitacional inhomogeneous broadening, 256, 259
INDEX 477
intensity, 23 Landau-Zener
intensive parameter, 450 fórmula de, 413
interaction picture, 60, 143, 246, 247 Langevin equation, 252
intermediate coupling, 231 large component, 169
interval factor, 185 Larmor
interval rule, 186 frequência de, 408
inverse Abel transform, 425 Larmor frequency, 19
Ioffe laser, 307
potencial de, 408 Legendre
irreducible matrix element, 204 Adrien-Marie, 104
isotropy Legendre operator, 103
spatial, 68 Legendre polynomials, 104
lepton, 188
Jaynes-Cummings Leucippus, 11
modelo de, 287, 300 Lie algebra, 47
Josephson junction, 159 light shift, 345
kernel, 52 light-shift, 255, 263
ket, 43 Lindbladt
Klein-Gordon equation, 38, 168 operador de, 345, 365
Koopman’s theorem, 225 linear momentum space, 52
Kramers Liouville equation, 245
Hendrik Anthony, 135 Liouville operator, 245, 262
Kramers-Heisenberg formula, 145 localization energy, 76
Kraus Lorentz
operador de, 363 calibre de, 375
Kronig-Penney model, 162 Hendrik Antoon, 177
modelo de, 342
Laguerre Lorentz boost, 66
Edmond, 109 Lorentz transform, 66
Laguerre polynomials, 109 lowering operador, 113
Laguerre’s associated differential equation, LS-coupling, 231
109 Luttinger liquid, 454
Lamb
Willis Eugene, Jr., 183 Mößbauer
Lamb shift, 183 efeito, 399
Lamb-Dicke efeito de, 350, 351
parâmetro de, 350 macromovimento, 417
regime de, 350 magnética
Lamb-Dicke parameter, 398 armadilha, 340
Lamb-dip, 256 magnetic quantum number, 104
Lambert-Beer magneto-óptica
lei de, 422, 423 armadilha, 400
Lambert-Beer law, 33 main quantum number, 108
Landé Majorana
fator de, 339 spin-flip de, 408
Landé factor, 197, 199 Markov
Landau gauge, 202 aproximação de, 303
Landau level, 203 Markoviano
478 INDEX
processo, 310 nuclear model, 15

maser, 322 number state, 88
master equation, 255 nutation, 250
Mathieu
equação de, 416 observable, 44
Maxwell operator, 43
equações de, 375 optical density, 451
optical tweezer, 404
Maxwell theory, 385
orbital angular momentum, 65, 112
Maxwell-Boltzmann distribution, 388
orbital magnetic moment, 19
Maxwell-Boltzmann law, 388, 432
ordem antissimétrica, 293
measurement, 244
ordem normal, 293, 306
mechanics
ordem simétrica, 293
wave, 45
ordem temporal, 306
mechanics of matrices, 45
ortho-helium, 216
medição
orthogonal, 47
processo da, 353
orthogonalization by Schmidt, 49
medida mediada por ancilla, 362
meson, 188 para-helium, 216
mestre parity, 51
equação, 300, 301, 360 parity conservation, 68
micromovimento, 417 parity inversion, 69
Mie scattering, 123 Paschen-Back effect, 198
minimal coupling, 173, 196 Paschen-Goudsmith effect, 200
minimum coupling, 67 Paul
mode density, 24 armadilha de, 415
Mollow Wolfgang, 415
tripleto de, 312 Pauli
moment Wolfgang, 45
first, 40 Pauli blocking, 452
momentum of inertia, 106 Pauli equation, 175
momentum space, 40 Pauli exlusion principle, 452
Monte Carlo simulation, 145 Pauli spin matrices, 169, 249
Monte-Carlo da função de onda Pauli spin matrix, 45, 114
simulação quântica de, 299, 358 Pauli’s exclusion principle, 433
MOT, 400 Pauli’s strong exclusion principle, 211
mundos múltiplos, 357 Pauli’s weak exclusion principle, 211
muonic hydrogen, 189 Penning
armadilha de, 415
número de fótons periodic system of the elements, 226
estado de, 283 permittivity
não-observação, 358 negative, 264
Newton law, 42 perturbation theory
Newton method, 144 time-dependent, 143
no free lunch time-independent, 129
teorema, 362 phase space density, 389
Noether’s theorem, 67 phase-sensitive detection, 320
normalization, 39, 43 phonon, 91
nuclear magneton, 183 phonons, 33
INDEX 479
photoelectric effect, 26, 147 quantum jump, 264

photon, 91 quantum measurement, 45
photon echo method, 252 quantum number, 76
photonic density of states, 163 good, 74
photonic recoil, 388 quantum reflection, 81
Planck quantum state endoscopy, 321
Max, 12 QUEST, 421
Planck’s constant, 25
planetary model, 18 Rabi
Poisson distribution, 94 frequência de, 286, 338
polar equation, 104 Rabi formula, 143
population, 247 Rabi frequency, 142
population inversion, 250 generalized, 142, 248
position space, 52 Rabi method, 251
Positive Operator Valued (Probability) Mea- Rabi no vácuo
sure, 363 desdobramento de, 291
potencial adiabático, 413 radiation collapse, 18
POVM, 363 radiation pressure, 27
power broadening, 253 radiation pressure force, 391
Poynting vector, 22 radiofrequência
precession, 250, 251 armadilha de, 415
pressão radiativa radon transform, 321
força de, 342 Raman
princı́pio de espalhamento de, 304
correspondência, 354 Ramsey method, 251
probability charge, 39 Rayleigh
probability current, 39 comprimento de, 404
probability density, 39 espalhamento de, 304, 312
probability distribution, 39 John William Strutt, 3. Baron, 134
probability flux, 82 Rayleigh fraction, 134
probability wave, 39 Rayleigh-Jeans law, 24
projeção Rayleigh-Ritz method, 135
ruı́do quântico de, 370 reciprocal lattice, 154
projection of the wavefunction, 46 rede óptica, 404
projector, 44, 53 redução do estado, 358
pure state, 240 reflection, 82
purity, 241 refração, 346
refraction index, 29
quadrupolar electron-nucleus interaction refractive index
constant of the, 186 negative, 264
quadrupolar interaction, 186 regressão quântica
quantization teorema da, 310
first, 21 relaxação do spin, 404
second, 95 release energy, 438
quantization primeira Renyi entropy, 241
first, 95 representation, 42
quantum confinement, 428 reservoir, 45
quantum defect, 191, 229 resfriamento evaporativo, 411
480 INDEX
resfriamento simpático, 412 Slater

Riemann zeta-function, 11, 435 John Clarke, 211
rigid rotator, 106 Slater determinant, 211
rigid rotor, 250 small component, 169
ring Sommerfeld
non-commutative, 47 Arnold, 21
rising operator, 113 Arnold Johannes Wilhelm, 182
rotating wave approximation, 146, 247, 302 Sommerfeld expansion, 11
rotation operator, 64 specific heat, 33
ruı́do branco de fase, 307 spectral density of modes, 24
Runge-Kutta method, 144 spectral energy density, 24
Russel-Saunders coupling, 231 spherical harmonics, 105
Rutherford spin, 114, 172
Ernest, 14 spin flip, 413
Rutherford scattering, 15 spin-charge separation, 454
Rydberg atom, 189 spin-orbit interaction, 178
Rydberg series, 189 spinning top, 250
squeezed state, see squeezing
salto quântico, 358 squeezing, 328
saturação multimode, 329
parâmetro de, 391 operador de, 326
saturation broadening, 253, 255 standard deviation, 259
saturation intensity, 256 standard model, 168
saturation parameter, 253, 256 Stark
scalar product, 47 Johannes Nikolaus, 203
scattering cross section, 14 Stark dinâmico
scattering matrix, 80 deslocamento de, 345
Schrödinger Stark effect, 203
estado de, 297 linear, 133, 204
gato de, 354 quadratic, 133, 204
Schrödinger cat state, 95 Stark shift, 191
Schrödinger equation, 37 state function, 42
Schrödinger picture, 58, 246–248 state reduction, 46
Schrieffer-Wolff transformation, 60 statistical mixture, 241
Schwartz inequality, 51 statistical operator, 240
screening, 17 steepest descent method, 144
seção eficaz diferencial de espalhamento, 305 Stefan-Boltzmann law, 36
secular determinant, 132 Stern
secular equation, 132 Otto, 20, 114
selection rule, 216 Stern-Gerlach
self-consistency, 218 experimento de, 340
shot noise, 370 Stern-Gerlach experiment, 114
shot-noise STIRAP, 264
ruı́do de, 318 subspace, 53
Shubnikov-de Haas effect, 203 superoperator, 245, see Louville operator262,
sideband cooling, 398 365
simpatético superposition principle, 42
resfriamento, 414 susceptibility, 254
INDEX 481
symmetrized wavefunction, 210 von Neumann entropy, 241

symmetry transformation, 67 von Neumann equation, 245, 262
von Neumann postulate, 46
tempo de voo, 421
teorema óptico, 422 wave equation, 37
termalização, 410 wave vector, 22
thermal equilibrium, 148 wavefunction, 39, 42
thermodynamic limit, 434 Weisskopf-Wigner
Thomas teoria de, 301
Llewellyn, 177 Wentzel
Thomas factor, 177 Gregor, 135
Thomas Precession, 177 Werner
Thomas-Fermi energy, 220 Heisenberg, 59
Thomas-Fermi equation, 221 Wien’s displacement law, 36
Thomas-Fermi model, 218 Wiener-Khintchine
Thomson teorema de, 309
espalhamento de, 304 Wigner
Joseph John, 14 Eugene Paul, 204
time reversal, 68 função de, 294, 298, 345
time-orbiting potential, 411 Wigner-Eckart theorem, 204
time-reversal invariance, 314 Wirtinger derivative, 94
TOP, 411 WKB approximation, 135
topológica
fase, 371 Young
três nı́veis experimento de, 306
sistema de, 358 Yukawa potencial, 168
trace, 57, 243
Zeeman
trajetória quântica, 357, 363
Pieter, 196
transformation matrix, 50
Zeeman effect, 19
transit time broadening, 255
anomalous, 197
translation operator, 63
normal, 197
transmission, 82
Zeeman slower, 256
two-body problem, 101
Zeeman splitting, 196
ultraviolet catastrophe, 24 Zeno
uncertainty principle, 51, 76 de Elea, 367
unitary, 47 efeito, 367
unitary transformation, 63 zero point energy, 76
vacuum fluctuation, 89
variational method, 134
vector space, 43, 47
vestido
estado, 283, 287
virial theorem, 111
Voigt profile, 259
von Neumann
equação de, 300
John, 46

MecanicaQuanticaScript PDF

Загружено:

Сведения о документе

Оригинальное название

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

MecanicaQuanticaScript PDF

Загружено:

Авторское право:

Доступные форматы

.

Graduate courses in Physics

1 Antecedents of quantum mechanics 7

2 Foundations of quantum mechanics 37

3 Linear motion / Separable potentials 75

3.3.4 Continuity of probability flow . . . . . . . . . . . . . . . . . . . . . . . . . 82

4 Rotations / Central potentials 101

5 Approximation methods 129

6 Periodic systems 153

II Atomic and Molecular Physics 165

7 The electron spin and the atomic substructure 167

7.4.3 Muonic hydrogen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189

8 Atoms with spin in external fields 195

9 Atoms with many electrons 209

10 The Bloch equations 239

11 Atoms in quantized radiation fields 283

11.3.4 The spectrum of resonance fluorescence . . . . . . . . . . . . . . . . . . . 309

12 Atomic motion in electromagnetic fields 337

IV Collective Scattering of Light 381

V Atom Optics 383

14 Manipulation of atomic gases 385

14.3.2 Optical dipole traps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 403

15 Thermodynamics of quantum gases 431

0.1 Possible topics of partial courses

The courses in Quantum Mechanics A (QM/SFI5774), of Quantum Mechanics B (QO/SFI5708),

0.2 Organization of the course

0.3 Recommended bibliography

Antecedents of quantum mechanics

1.1.1 Atomic units

Ẽ = E/(e2cgs /aB ) , (1.2)

This notation simplifies the formulas. For example:

1.1.2 Clebsch-Gordan symbols

with the abbreviation,

(j1 + j2 − j3 )!(j1 − j2 + j3 )!(−j1 + j2 + j3 )!

The {6j}-symbol describes the recoupling of of two spins,

where f (t) = (t − j1 − j2 − j3 )!(t − j1 − J2 − J3 )!(t − J1 − j2 − J3 )!(t − J1 − J2 − j3 )!(j1 + j2 +

{9j}-symbols satisfy the following orthogonality relation,

1.1.3 Functions and polynomials

1.1.3.2 Hermite polynomials

H2n+1 (0) = 0 (1.14)

1.1.3.3 Laguerre polynomials

1.1.3.4 Legendre polynomials

1.1.3.5 The Gamma function

1.1.3.6 Riemann zeta-function

gη (1) = ζ(η) , (1.24)

1.2 The discovery of the atom

Table 1.1: Historical sketch of the quantization of matter

500 a.c. Democritus invention of the atom

Table 1.2: Historical sketch of the quantization of light

1801 Young light is diffracted like a wave

1.2.1 Democrit’s model

Figure 1.1: Democritus and dust in a sun ray.

1.2.2 Thomson’s model and Rutherford’s experiment

dN Zα Ztg e2 θ Zα Ztg e2 1 Zα2 Ztg

dΩ = 2π sin θdθ = 4π sin 2θ cos 2θ dθ . (1.39)

Thus, the number dN of particles scattered to the solid angle dΩ remains,

• The integral over the probability distribution P (θ)dθ is normalized,

Similarly, we have for the surface integrals,

Figure 1.5: (Code: QM Antecedents Rutherspectrum.m) Angular dependence of the cross-

1.2.3 Emission of radiation in the planetary model

is dissipated by radiation of the power (1.48), i.e.,

Integration of this equation gives,

32π 2 ε20 m2e c3 3

dŪ (χ) B12 g1