You are on page 1of 283

Notes for edX/MITx 8.

02x Electricity and Magnetism,


Spring 2013
Thomas Backman, exscape@gmail.com
June 18, 2013
Contents
0 Preface/disclaimer 6
1 Week 1 7
1.1 Coulombs law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.2 Electric elds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.2.1 Multiple charges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.2.2 Electric eld lines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.2.3 Dipoles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.3 Electric ux and Gausss law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.3.1 Quick facts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.3.2 Slower this time... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.3.3 Gausss law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2 Week 2 14
2.1 Electrostatic Potential Energy and Electric Potential . . . . . . . . . . . . . . . . . . . . . 14
2.1.1 Electrostatic Potential Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.1.2 Electric potential . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.1.3 More on electric potential . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.2 Electric elds inside hollow conductors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.3 High-voltage breakdown and lightning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.3.1 Electric breakdown . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3 Week 3 23
3.1 Capacitance and Field Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.1.1 Field Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.1.2 Capacitance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.2 Dielectrics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.2.1 More on capacitors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.2.2 A bit on van de Graa generators . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.3 Current, resistivity and Ohms law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
4 Week 4 34
4.1 Batteries and EMF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
4.1.1 Kirchhos rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
4.1.2 Basic circuit analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
4.2 Magnetic Field and Torques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
4.2.1 The B Field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.3 Review for Exam 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
4.3.1 Thats it . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
1
5 Week 5 40
5.1 Moving Charges in Magnetic Fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
5.1.1 Moving charges, radii and special relativity . . . . . . . . . . . . . . . . . . . . . . . 40
5.1.2 Isotope separation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
5.1.3 Particle accelerators/cyclotrons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
5.1.4 Cloud chambers and bubble chambers . . . . . . . . . . . . . . . . . . . . . . . . . 45
5.2 Biot-Savart Law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
5.3 Amperes law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
5.3.1 The magnetic eld of a solenoid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
6 Week 6 51
6.1 Electromagnetic induction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
6.1.1 Lenzs law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
6.1.2 Magnetic ux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
6.1.3 Faradays law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
6.1.4 The breakdown of intuition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
6.2 Motional EMF and dynamos . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
6.2.1 Changing the area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
6.2.2 Eddy currents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
6.3 Displacement currents and synchronous motors . . . . . . . . . . . . . . . . . . . . . . . . 61
6.3.1 The amended Amperes law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
6.3.2 Displacement current . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
6.3.3 Synchronous motors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
7 Week 7 68
7.1 How do magicians levitate women? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
7.1.1 The human heart . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
7.1.2 Aurora borealis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
7.1.3 Superconductivity and magnetic levitation . . . . . . . . . . . . . . . . . . . . . . . 71
7.2 Inductance and RL circuits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
7.2.1 Direct-current RL circuits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
7.2.2 Alternating-current RL circuits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
7.2.3 More on magnetic levitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
7.3 Magnetic materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
7.3.1 A short note on motors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
7.3.2 Magnetic dipole moment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
7.3.3 The source of magnetism in matter . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
7.3.4 Magnetization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
7.3.5 Paramagnetism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
7.3.6 Diamagnetism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
7.3.7 Ferromagnetism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
8 Week 8 90
8.1 Hysteresis and electromagnets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
8.1.1 Ferromagnetism and hysteresis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
8.1.2 Maxwells equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
8.2 Review for Exam 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
9 Week 9 97
9.1 Transformers, Car Coils and RC circuits . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
9.1.1 RC circuits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
9.1.2 Transformers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
9.1.3 Spark plugs / car coils . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
2
9.2 Driven RLC circuits and resonance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
9.3 Traveling waves and standing waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
9.3.1 Traveling waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
9.3.2 Standing waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
9.3.3 Musical instruments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
10 Week 10 109
10.1 Resonance, electromagnetic waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
10.1.1 Radar and measuring distance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
10.1.2 Radio . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
10.2 Index of refraction and Poynting vector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
10.2.1 Poynting vector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
10.2.2 Waves due to accelerating charges . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
10.2.3 Spherical waves and the Poynting vector . . . . . . . . . . . . . . . . . . . . . . . . 115
10.2.4 Photons and radiation pressure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
10.2.5 Polarization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
10.3 Snells law, refraction and total reection . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
10.3.1 Total internal reection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
10.3.2 Frequency and wavelength in refraction . . . . . . . . . . . . . . . . . . . . . . . . . 122
10.3.3 Dispersion, prisms and white light . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
10.3.4 Primary colors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
11 Week 11 126
11.1 Polarizers and and Maluss law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
11.1.1 Polarization by reection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
11.1.2 Polarization by scattering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
11.1.3 Scattering demonstration and Rayleigh scattering . . . . . . . . . . . . . . . . . . . 131
11.2 Rainbows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
11.2.1 Other atmospherical optical phenomena . . . . . . . . . . . . . . . . . . . . . . . . 138
11.2.2 Polarization of rainbows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
11.3 Review for Exam 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
12 Week 12 140
12.1 Double slit interference and interferometers . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
12.2 Gratings and resolving power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
12.3 Single-slit diraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
12.4 Doppler eect and the Big Bang . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
12.4.1 The Doppler eect and light . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
12.4.2 Big Bang cosmology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
12.5 Farewell special . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
13 Homework problems 153
13.1 Week 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
13.1.1 Problem 1: Motion of charged particle in electric eld . . . . . . . . . . . . . . . . . 153
13.1.2 Problem 2: Three plates capacitor . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
13.1.3 Problem 3: Electric eld, potential, and electrostatic potential energy . . . . . . . . 154
13.1.4 Problem 4: Electric eld of a charged ring . . . . . . . . . . . . . . . . . . . . . . . 155
13.1.5 Problem 5: Two spherical conductors . . . . . . . . . . . . . . . . . . . . . . . . . . 158
13.1.6 Problem 6: Two conducting hollow cylinders . . . . . . . . . . . . . . . . . . . . . . 159
13.1.7 Problem 7: Speed of an electron . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
13.2 Week 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
13.2.1 Problem 1: Spherical capacitor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
13.2.2 Problem 2: Coaxial cylinders . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
3
13.2.3 Problem 3: The eect of a dielectric medium on capacitance . . . . . . . . . . . . . 165
13.2.4 Problem 4: Coaxial cable with dielectric . . . . . . . . . . . . . . . . . . . . . . . . 166
13.2.5 Problem 5: Capacitor network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
13.2.6 Problem 6: Resistances of conducting wires . . . . . . . . . . . . . . . . . . . . . . 169
13.2.7 Problem 7: Resistor network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
13.3 Week 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
13.4 Week 5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
13.4.1 Problem 1: Lorentz Force . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
13.4.2 Problem 2: Motion of a charged particle in magnetic eld . . . . . . . . . . . . . . . 172
13.4.3 Problem 3: Cyclotron . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
13.4.4 Problem 4: Rectangular current loop . . . . . . . . . . . . . . . . . . . . . . . . . . 173
13.4.5 Problem 5: Resistor network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
13.4.6 Problem 6: Coaxial current loops . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
13.4.7 Problem 7: Parallel plate capacitor . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
13.5 Week 6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
13.5.1 Problem 1: Amperes law in action . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
13.5.2 Problem 2: Intuition breaks down . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
13.5.3 Problem 3: Helmholtz coils . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
13.5.4 Problem 4: Spinning loop in a magnetic eld . . . . . . . . . . . . . . . . . . . . . . 184
13.5.5 Problem 5: Loop in a magnetic eld . . . . . . . . . . . . . . . . . . . . . . . . . . 185
13.5.6 Problem 6: Electrodynamic tether . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
13.5.7 Problem 7: Motor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
13.5.8 Problem 8: Auroral zone . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
13.6 Week 7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
13.6.1 Problem 1: Magnetic energy of a solenoid . . . . . . . . . . . . . . . . . . . . . . . 188
13.6.2 Problem 2: Displacement current . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
13.6.3 Problem 3: RL circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
13.6.4 Problem 4: RL circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
13.6.5 Problem 5: Opening a switch on an RL circuit . . . . . . . . . . . . . . . . . . . . . 193
13.6.6 Problem 6: Self-inductance of a toroid . . . . . . . . . . . . . . . . . . . . . . . . . 196
13.6.7 Problem 7: RL circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
13.7 Week 8 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198
13.8 Week 9 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
13.8.1 Problem 1: RC circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
13.8.2 Problem 2: RC circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200
13.8.3 Problem 3: RLC circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
13.8.4 Problem 4: An LRC circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
13.8.5 Problem 5: Design a ute . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
13.8.6 Problem 6: Width of resonance peak . . . . . . . . . . . . . . . . . . . . . . . . . . 203
13.8.7 Problem 7: Standing wave . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
13.8.8 Problem 8: Traveling wave . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205
13.8.9 Problem 9: Lightly damped undriven circuit . . . . . . . . . . . . . . . . . . . . . . 206
13.9 Week 10 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208
13.9.1 Problem 1: Traveling electromagnetic waves . . . . . . . . . . . . . . . . . . . . . . 208
13.9.2 Problem 2: A standing electromagnetic wave . . . . . . . . . . . . . . . . . . . . . . 209
13.9.3 Problem 3: E-M waves - Maxwells equations, and the speed of light . . . . . . . . . 211
13.9.4 Problem 4: Polarized radiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
13.9.5 Problem 5: Polarization of electromagnetic radiation . . . . . . . . . . . . . . . . . 216
13.9.6 Problem 6: Poynting vector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
13.9.7 Problem 7: Intensity of the sun . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
13.9.8 Problem 8: Snells law in action: ber optics! . . . . . . . . . . . . . . . . . . . . . 219
13.10Week 11 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
4
13.11Week 12 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
13.11.1Problem 1: Primary rainbow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
13.11.2Problem 2: Polarization of primary rainbow . . . . . . . . . . . . . . . . . . . . . . 223
13.11.3Problem 3: Glassbow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
13.11.4Problem 4: Secondary rainbow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
13.11.5Problem 5: Diraction pattern . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
13.11.6Problem 6: Optical resolution of the human eye . . . . . . . . . . . . . . . . . . . . 226
14 Exam problems 227
14.1 Midterm 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
14.1.1 Problem 1: Electric eld on the surface of a conductor . . . . . . . . . . . . . . . . 227
14.1.2 Problem 2: Non-conducting charged planes . . . . . . . . . . . . . . . . . . . . . . . 228
14.1.3 Problem 3: Incandescent bulbs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
14.1.4 Problem 4: Circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
14.1.5 Problem 5: Point charge in a hollow conducting sphere . . . . . . . . . . . . . . . . 231
14.1.6 Problem 6: Charges on an equilateral triangle . . . . . . . . . . . . . . . . . . . . . 232
14.1.7 Problem 7: Capacitor network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
14.2 Midterm 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235
14.2.1 Problem 1: RL circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235
14.2.2 Problem 2: Non-conservative elds . . . . . . . . . . . . . . . . . . . . . . . . . . . 236
14.2.3 Problem 3: Bainbridge mass spectrometer . . . . . . . . . . . . . . . . . . . . . . . 236
14.2.4 Problem 4: RL circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238
14.2.5 Problem 5: Magnetic eld of a loop . . . . . . . . . . . . . . . . . . . . . . . . . . . 238
14.2.6 Problem 6: Magnetic eld of a current-carrying ribbon . . . . . . . . . . . . . . . . 239
14.2.7 Problem 7: Magnetic eld of a rotating charged sphere . . . . . . . . . . . . . . . . 240
14.3 Midterm 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243
14.3.1 Problem 1: RC Circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243
14.3.2 Problem 2: RLC circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
14.3.3 Problem 3: Energy ow of a capacitor . . . . . . . . . . . . . . . . . . . . . . . . . 247
14.3.4 Problem 4: Electromagnet with small air gap . . . . . . . . . . . . . . . . . . . . . 250
14.3.5 Problem 5: RLC Circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251
14.3.6 Problem 6: Electromagnetic wave . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252
14.3.7 Problem 7: Radiation pressure on the Earth . . . . . . . . . . . . . . . . . . . . . . 253
14.4 Final Exam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255
14.4.1 Problem 1: Loop in a Magnetic Field . . . . . . . . . . . . . . . . . . . . . . . . . . 255
14.4.2 Problem 2: Non-conducting innite sheet and innite parallel slab . . . . . . . . . . 256
14.4.3 Problem 3: RLC circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260
14.4.4 Problem 4: Plane electromagnetic wave . . . . . . . . . . . . . . . . . . . . . . . . . 262
14.4.5 Problem 5: Loop in a magnetic eld . . . . . . . . . . . . . . . . . . . . . . . . . . 263
14.4.6 Problem 6: Charged particles in a magnetic eld . . . . . . . . . . . . . . . . . . . . 265
14.4.7 Problem 7: Diraction pattern . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266
14.4.8 Problem 8: Capacitor washers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268
14.4.9 Problem 9: Circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
14.4.10Problem 10: An LC circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275
14.4.11Problem 11: Dielectric sphere . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278
14.4.12Problem 12: Parallel RLC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
5
Chapter 0
Preface/disclaimer
The most important thing to keep in mind where reading these notes is that Im just another student,
taking this course. I have almost zero previous knowledge of electromagnetism, though I previously took
the MITx 6.002x Circuits and Electronics class. That knowledge helped only in small parts of this course,
however, and Ive learned almost everything except circuit analysis on a week-to-week basis, while writing
down all this.
In other words: there may well be errors in here!
In part because I make typos and other stupid mistakes, but also because I might have misunderstood
some concept somewhere. I usually point out whenever I feel less than certain, but that doesnt guarantee
anything, of course.
These notes are made up of three parts: lecture notes, homework solutions, and exam solutions.
The lecture notes are in part from the lectures as-is (essentially a transcription), but I do add my own
comments and extra information here and there.
They will lack some information from the lectures, as I dont write down everything (for example, I rarely
write down anything during demonstrations, only their results if necessary), but also have some tiny extra
bits (from the book, things I knew to begin with, or things I look up elsewhere).
I write the homework solutions as I solve them, so the text and even the answers are usually written before
Ive even clicked check. Still, I look at the stas solutions after the fact and at least if I remember using
a dierent method, I try to check that I wasnt just lucky in getting the correct answer for myself.
The exact same thing applies to the exam solutions as well.
There are no notes/solutions for homework 1, as I started writing the solutions down the second week.
By the way: I certainly welcome all feedback, corrections, comments and whatnot! Too small a fraction
of my daily e-mail is written by actual human beings (rather than being spam, mass-emails/mailing lists,
forum notices, ...), anyway. :-)
6
Chapter 1
Week 1
1.1 Coulombs law
Two point charges q
1
and q
2
, separated by distance r, equal in magnitude and of equal polarity, will repel
one another. The force on q
2
due to q
1
, in the direction 180 degrees opposite to to the direction of q
1
, is

F
1,2
=
q
1
q
2
K
r
2
r
1,2
... where K is Coulombs constant, roughly 8.987551787 10
9
N m
2
C
2
, and

F
1,2
means the force on
charge 2 due to charge 1. r
1,2
is the unit vector in the direction from charge 1 to charge 2. And, since
each force has an equal and opposite reaction, we also have a

F
2,1
(

F
2,1
=

F
1,2
)
Note how the equation is sign-sensitive: if both changes have the same sign, the force forces the particles
apart. On the other hand, if they dier in sign, the result will be negative, and so the force vector will
point towards q
1
from q
2
(and vice versa) instead of away from it.
Also:
K =
1
4
0
where
0
is the permittivity of free space, also (more recently) known as the electric constant.

0
8.854187 10
12
farad/meter
1.2 Electric elds
Imagine a setup very similar to the one above.
We have a charge, Q
+
, and a test charge q a distance r away. As before, the force on the test charge is

F =
QqK
r
2
r
If we take that force vector, and divide out q, the test charge, we get the electric eld at the point in
space where q is located. Lets call that point p; in that case, we have the electric eld

E
p
:

E
p
=

F
q
=
QK
r
2
r
where r is again the unit vector pointing away from Q in the direction of the test charge.
By convention, we choose the electric eld to be such that the eld vector is in the direction of the force
on a positive test charge. If the test charge were negative, the electric force would attract the two charges
7
toward each other, and the electric eld would be pointing towards Q.
Put more simply: for a positive charge, we draw the eld as a set of arrows pointing from the charge and
away. This eld is spherically symmetric, and as we saw above, decays in a inverse-r-squared fashion. If
the charge is instead negative, we draw it as a set of arrows pointing towards the charge, again with more
force (longer arrows, to represent larger vector magnitudes) closer to the charge.
The electric eld can be thought of as the force per unit charge (as we have divided out the charge of the
equation).
1.2.1 Multiple charges
To calculate the net electric eld for multiple charges, we use the superposition principle: very simply, we
calculate the electric eld due to each charge alone, at a certain point p; the net electric eld a that point
is then simply the vector sum of all these individual electric elds.
For example, for a set of three charges, Q
1+
, Q
2
and Q
3
, we get three separate electric elds: one that
repels from Q
1
, and two that attract towards Q
2
and Q
3
respectively. The net electric eld is then

E
p
=

E
1
+

E
2
+

E
3
and more generally, for a set of i charges, the same sum holds:

E
p
=

E
i
So depending on how much change each of these points hold, the resulting vector may be in any direction,
with any magnitude; we simply cannot say until we do a quantitative calculation of the force.
Now, then, what is the point of calculating this electric eld? Well, now we know what the electric eld
at that point is, we can very easily calculate the force vector for any charge we place at the location, like
this:

F = q

E
So, in other words, we simply multiply the charge by the electric eld vector, and we get the force. q can
be large or small, positive or negative, and the result will still come out right.
If the charge is large, the force will be large, and vice versa.
If the charge is positive, the force vector will be in the same direction as

E. If the charge is negative, it
will be in the opposite direction of

E.
In a system of multiple charges, there may be a point in space where the electric elds magnitude is
exactly zero, because the forces from all the charges in the system perfectly cancel out.
One of the simplest such cases is when you have a set of three charges, equal in polarity and magnitude,
arranged at the vertices of an equilateral triangle. In such a case, the electric eld strength will be exactly
zero in the middle of the three charges.
Charges of diering polarity can also produce points with no net force.
Imagine a one-dimensional line, measuared in meters. At the exact middle, 0 m, we have a 1C charge.
At 4 m, we have a 0.5C charge.
For a positive test charge located in between these two, i.e. on the interval (0, 4), the forces will be
additive; the positive charge at 0 will push towards the right, and the negative charge at 4 m will pull
towards the right.
If we go past 4 m, however, there will come a point where the force cancels out exactly: the stronger, but
8
more distant positive charge repels, and the weaker, but closer negative charge attracts.
In this example, that spot is at 8 + 4

2 meters, which can be found by solving the equation relating the


two electric elds:

E
0
=
1K
r
2
x

E
4
=
0.5K
(r 4)
2
x

E
0
+

E
4
= 0
where

E
0
is the electric eld due to the particle a 0 m alone, and

E
4
the eld due to the negative particle
at 4 m alone.
If we solve the equation (on the third line above), we get 13.65 meters.
1.2.2 Electric eld lines
Theres a second common way of visually representing electric elds, besides the vector eld: electric eld
lines. Here, we lose some information about the eld strength at any particular point, but can instead see
the curvature much more clearly.
Such a representation would have a ton of lines coming from (or going to, for negative charges) each charge,
curving around depending on other, nearby charges.
The further away you get from a charge, the further apart the lines will become. This represents the
inverse-r-squared weakening of the eld.
One crucial aspect about these lines is that if you place a positive charge anywhere on a eld line, the
force vector on that charge will always be tangential to the eld line at that point. If the placed charge is
negative, the force vector will simply be the same but ipped 180 degrees around.
It is also important to note that because of this, the trajectory of a charge (say, released a 0 velocity) will
not follow any one eld line exactly; the trajectory can be quite complex. Any time you reach a eld line,
however, the force vector will again be tangential to that eld line.
1.2.3 Dipoles
As we saw above, with two charges of equal polarity, there is a point between the two where the force
cancels out, and the electric eld is zero.
If we instead have two charges, equal in magnitude but opposite in sign, we have a dipole. In this case,
there is no point where the electric eld is zero.
Two charges of unequal charge and opposite polarity will, as observed from far away, behave as a single
charge whose value is the dierence of the two. For example, a +3 change next to a -1 will look like a
+2 charge from far away, and the electric eld falls o with a inverse-r-squared relationship.
With dipoles, that is no longer the case: the eld falls o with an inverse-r-cubed relationship, i.e. it
weakens much faster as the distance increases. This does make intuitive sense: from afar, it should act
like a single point charge with zero charge!
1.3 Electric ux and Gausss law
1.3.1 Quick facts
According to Wikipedia, electric ux is the rate of ow per unit charge of the electric eld. It can be
thought of as the number of electric eld lines going through a surface.
For a uniform electric eld, the ux passing through a surface of vector area S is
9

E
=

E

A = EAcos
where is the angle between the normal of the surface (the vector perpendicular to the surface) and the
electric eld lines.
For a non-uniform electric eld, the ux can be calculated by using a surface integral:

E
=

E

dA
where

dA is a vector area innitesimal.
1.3.2 Slower this time...
OK, lets start over. Say we have an open surface, e.g. a paper (unlike a closed surface, e.g. a bag or a
balloon) placed in an electric eld. We can divide the eld into innitesimals with area dA. Each of these
will have a normal unit vector n that is perpendicular to it.
Each of these innitesimal areas will have a ux d associated with it, which can be calculated:
d
E
=

E ndA =

E

dA = E dAcos
Here, is, as above, the angle between the normal and the electric eld lines. The notation with and
without the n are two ways of writing the same thing. One considers dA as an area, with a normal unit
vector to give it direction, while the second combines the to into the vector area

dA.
The ux can be larger than, smaller than, or equal to zero.
One can make a useful parallel between electric ux and air ow. Consider a hollow rectangle position
perpendicular to a uniform air ow. In other words, the normal to the rectangle is parallel with the air
ow. How much air goes through it? Well, its simply the rate of air ow multiplied by the rectangles
area.
If we rotate the rectangle 90 degrees, such that the normal is now perpendicular to to the air ow, no air
goes through the rectangle - it is parallel to the air ow.
If the angle is anything in between the two, we simply multiply the v A we already have by the cosine of
the angle. That is, the air ow through the rectangle is given by vAcos .
The exact same way can be used to calculate the electric ux through the rectangle - multiply the electric
eld vector and the vector area using the dot product, such that you get EAcos .
Example with a point charge
Say we have a point charge
+
Q, and a spherical surface with radius R surrounding it.
For any chosen innitesimal area dA on the surface, the normal will be radially outwards - as will the
electric eld, since we have a positive point charge. In fact, the two will be parallel at all locations on the
surface, so the cosine of the angle between them is always 1.
Not only that, but since we chose to do this with a spherical surface, the electric eld has exactly the
same magnitude at all points along the surface. These facts, when combined, makes for an extremely easy
calculation of the total ux - it is simply the magnitude of the electric eld, times the surface area - EA,
or in the case of a sphere, 4R
2
E.
Well, E is
QK
R
2
r, as we have seen previously, with r being the unit vector in the direction from the charge
to the chosen surface area innitesimal. K, the Coulomb constant, has an exact denition (but we often
simply use 9 10
9
instead, which about 0.14% o from the real value):
K =
1
4
0
10
If we substitute that in, we have

E =
Q
4
0
R
2
r
... which looks more complex, until we nalize the ux calculation by multiplying the electric elds
magnitude with the surface area of 4R
2
:
E 4R
2
=
Q

0
Spectacular! Note not only how simple it is, but indeed that the radius term disappeared. And, since
0
is
a physical constant, we clearly see that the net charge is the only thing that determines the ux through
this closed surface.
Indeed, this also means that the result holds regardless of the shape of the surface used to conne the
charge. Whether it is a sphere, a cylinder, a box or just something entirely dierent, the amount of ux
is exactly the same.
This can be intuitively understood if we go back to the air ow analogy: regardless of the bags size,
assuming air could travel through it, it would have to get out of there somewhere regardless of the bags
size.
Now, since electric elds from multiple point charges are added using vector addition, it is clear that any
collection of charges inside this bag must behave in the same way. That leads us to:
1.3.3 Gausss law
Gausss law states that the electric ux through a closed surface, with nonzero net charge inside, is,
regardless of the shape and size of the surface:
=


E

dA =

Q
ins

0
where (also
E
) is the total electric ux through the surface,

Q
ins
is the total electric charge inside
the surface, and
0
as always the permittivity of free space.
The ring in the integral symbol is to indicate that the surface we are integrating over is a closed surface.
Gausss law always holds, no matter how the charge is arranged inside the bag, or how it is shaped.
It is, however, not very useful unless we have some form of symmetry. The main use of Gausss law is
to calculate electric elds, and to do so, we need to take advantage of one of three forms of symmetry:
spherical, to encapsulate a point charge or such; cylindrical, to encapsulate an innite wire of charge;
and planar, to calculate electric elds around an innite plane. In all three cases, the charge distribution
must be uniform.
Gausss law application: a charged, hollow sphere
Lets attempt an application of Gausss law, on a hollow, conductive sphere, with charge Q uniformly
distributed on the surface. The sphere is of radius R.
First, we need to choose one of the gaussian surfaces: the spherical one, the cylinder (for the innite wire)
or the pillbox, also a cylinder, used for planar surfaces.
Clearly, we want the spherical one in this case; we choose a concentric sphere (i.e. it is perfectly centered
on the charged sphere) such that the radius r of the Gauss sphere touches the point where we want to
calculate the electric eld.
Then, we note a few things: the problem is spherically symmetric, so the electric elds magnitude at any
point along the surface of our Gauss sphere must be exactly the same as for any other point on it. That
follows from the symmetry of the problem.
Also, we know that the electric eld is radially outwards (or inwards, if the charge is negative) - either
11
way, it is either parallel or antiparallel between any chosen dA on the surface.
Now that we have the same situation as before - the direction of E is always parallel with the surface
normals, and we know how the surface area of a sphere - we can calculate the (again simply EA), and
then solve for the electric eld.
Lets begin with the case of r < R, i.e. inside the hollow sphere:
E4r
2
=

Q
ins

0
E =

Q
ins
4r
2

0
However, we know from the problem denition that there is no charge inside, so the numerator is zero!
Therefore, the electric eld at every point inside the sphere is zero, since the result clearly holds for any
r inside the sphere - we didnt specify it.
Next, lets look at the case where r > R, i.e. outside the hollow sphere.
The same arguments hold here, as well: the problem is spherically symmetric, so the magnitude must
be equal at all points at the distance r from the sphere; also, we know that the electric eld is radially
outwards (or inwards). Therefore, we can use the same calculation:
E4r
2
=

Q
ins

0
E =

Q
ins
4
0
r
2
This, too, is an incredible result! The result is exactly what you get for a point charge, when using
Coulombs law (only that we use
1
4
0
= K here instead of the symbol K). In other words, this charged
hollow sphere behaves exactly as a point charge to everyone outside the sphere (while, as we saw just
above, the electric eld is exactly zero inside the sphere).
Second Gausss law application: innite charged plane
Next, consider a large plane - innite, even, for the calculation - again with a uniform charge density .
That is, we have a set number of coulombs per square meter .
As before, our rst task is to choose a Gaussian surface. For planes, we use the pillbox, which is a
cylinder. Our goal is to nd the electric eld at any point in space. We choose a point d, and place
our cylinder such that the plane is exactly centered along its height. That is, d of the cylinders height is
above the plane, and d is also below the plane.
Now, for the symmetry arguments to work, we need a few things to be true: the circular area at the top
of the cylinder must be parallel to the charged plane, and same for the bottom. Second, the vertical walls
of the cylinder must be perpendicular to the plane. Lastly, as stated previously, the two side caps must
both be exactly d away from the plane, i.e. be at the same distance from the plane.
Since the surface charge distribution is uniform, we can nd the enclosed charge in our cylinder easily: it
is simply the charge density multiplied by the area of the cylinders end caps:
Q
enc
= r
2
Keep in mind that since the charge isnt in the end caps, the fact that there are two of them is irrelevant.
We are only interested in the volume they enclose, namely the innitesimally thin plane in the center.
The total ux through the end caps, however, is doubled because of this. The ux through each cap is
E A, with E being the electric eld magnitude, and A the end cap area. Therefore, the total ux is (with
A = r
2
):
12
= 2Er
2
Now that we know the enclosed charge and the total ux, we can use Gausss law to nd the electric eld.
To do so, we set
Q
enc

0
equal to the total ux :
2Er
2
=
r
2

0
2E =

0
E =

2
0
The direction of this force is z above the plane, and z below it, so we have:

E =
_

2
0
z for z > 0


2
0
z for z < 0
13
Chapter 2
Week 2
2.1 Electrostatic Potential Energy and Electric Potential
Before we get started, lets introduce the two terms we will work with. The rst is electrostatic potential
energy, U, measured in joules. The second is electric potential, V , measured in volts (1 V = 1
J
C
).
There two are, as the dimensions suggest, not interchangable, but they are related.
2.1.1 Electrostatic Potential Energy
Now then. Lets say we have two charges, q
1
and q
2
. They are both positive, and are separated by a
distance R. Because they are both positive, they repel each other, and therefore, some external agent
must have done work to move the two together. The work we put in to moving the charges together (by
convention, we count all the way from innity to their nal position) is the electrostatic potential energy.
It can be thought of as the electrical analogue of gravitational potential energy - lift an object upwards,
against the Earths gravity, and you have increased its gravitational potential energy. If you then let it
go, that potential energy will be turned into kinetic energy as the object falls back down.
In the same manner, if you bring the positive charges together, and then let them go, the electric force will
repel them and cause them to move, thus converting their electrostatic potential energy into kinetic energy.
The work we will have to do to push the charges together is of course exactly equal to the negative work
the electric force will do - they are opposite in direction (i.e. the vectors are antiparallel), but equal in
magnitude. If we use

F
we
do denote our force, we have that

F
we
=

F
el
, where the latter is the electric
force.
The work required to bring the charge in from innity, then, is:
W
we
=
_
R

F
we


dr =
_

R

F
el


dr
The latter half is, as stated above, exactly the same as the force we provide; only the sign diers between
the two forces, so we swap the integration limits to in eect multiply the integral by 1.
We know that the force and r are in the same direction, so the angle between them is 0, and therefore the
cosine of the angle is 1. Therefore we can ignore the vectors, and instead use Coulombs law for much of
the integral. Continuing:
W
we
=
_

R

F
el


dr =
q
1
q
2
4
0
_

R
dr
r
2
_

R
dr
r
2
=
1
r

R
= 0 (
1
R
) =
1
R
So, when we combine the two, we just need to add that R to the denominator, and were done:
14
U = W
we
=
q
1
q
2
4
0
R
Since the quantity of this is work, this is a scalar, and it is as stated above in joules.
Note how the equation is sign-sensitive: if the two charges are of the same sign (whether positive or neg-
ative), the resulting number is positive (since nothing in the denominator can ever be negative), while if
they dier in sign, the result is negative. This is of course exactly right: we need to do positive work if
they repel each other, but will do negative work if they attract.
The electric force is conservative. This means that regardless of which path you take from innity to the
point in question, the work done is exactly the same, no matter how convoluted the path. This is also
true of gravity, which is another example (perhaps the most common one) of a conservative force.
As a counterpoint, an example of a non-conservative force is friction. Moving an object along some short,
straight line between points will cause some amount of work, while moving it along a very long path will
cause more work.
2.1.2 Electric potential
The electric potential is closely related to the electrostatic potential energy.
W take a similar setup as in the section above, with a positive charge Q, a test charge q at point P, a
distance R away from Q.
The electrostatic potential energy of this system we already know, as we calculated it just above:
U =
Qq
4
0
R
The electric potential is dened as the work per unit charge we have to do in order to bring the charge to
that location. That is, we divide out the test charge q, and so we have that the electric potential V and
the point P is:
V
P
=
U
q
=
Q
4
0
R
And, as stated in the introduction to this section, this is measured in joules per coulomb, also known as
volts (V).
If Q is positive, the potential everywhere in space (due to that charge alone, of course!) will be positive.
If the charge is negative, the potential will be negative everywhere.
Lets calculate an example, while adding some new information still.
Take a hollow sphere with radius R = 0.3 meters, with 10C of charge on it, evenly distributed (as it will
be automatically for a conductor).
We want to calculate the electric potential everywhere in space. We divide it into two cases: r < R (inside
the sphere) and r > R (outside). Lets begin with the calculation for the potential outside the sphere, at
a point P.
Remember how V =
U
q
, where V is the electric potential and U is the electrostatic potential energy, and
how the latter is U =
_

r

F
elec


dr. Therefore, the electric potential is the same integral, except we divide
out the test charge q:
V
P
=
_

r

F
elec
q


dr
Aha! Force divided by charge is nothing other than the denition of an electric eld! Therefore this is
equal to:
15
V
P
=
_

r

F
elec
q


dr =
_

r

E

dr =
Q
4
0
r
... which is the same result we got from a point charge. In other words, just like how an electric eld is
the same for a sphere with uniform charge density and a point charge of the same magnitude, the same
thing applies to the electric potential.
Now that we can calculate the electric potential, how do we use it? Well, we can now calculate the work
required to bring a charge to a point very easily. We know from above that U = V q (except we solved
for V last time), so we simply multiply the charge we are interested in by the potential, and we have our
answer!
As an example, lets say we calculate/measure a potential of 10 kV near a charged sphere. We want to
bring 1 millicoulomb to the point in space where the potential is 10 kV. The work we must do is
W = 10
3
C 10
4
V = 10 J
So the work required is 10 joules. We can now solve the same question for a charge of 2 C instead, without
having to solve a new integral - we simply multiply it by the 10
4
volts and get our answer of 20 kilojoules.
What happens, then, if we move this charge inside this hollow, charged sphere? Well, as we saw in the
previous week, there is no electric eld whatsoever inside, so the charges would experience no force. No
force means no work, which implies the electric potential must be the same everywhere inside. Note that
it only implies that the potential is the same everywhere, and not that it is zero! It will, in fact, be 10
kilovolts everywhere inside the sphere (same as on the surface).
Equipotentials
Now, lets look at another phenomenon outside the charge. Since the electric potential for a point charge
/ charged sphere depends only on a set of constants, plus the amount of charge and the distance r, it is
clear that all points that are r away from the charge will have the exact same potential, by symmetry.
Each such surface, where all points have the same electric potential, is known as an equipotential surface.
There will be an innite amount of such surfaces at dierent distances away from the charge.
Electric elds lines are always perpendicular to equipotential surfaces: if they were not, we could decom-
pose the electric eld vector at a point on the surface, into parallel and perpendicular dimensions. That
implies that there is an electric eld along the surface, which means the surface is not an equipotential !
Therefore, eld lines are always perpendicular to equipotential surfaces.
Now, back to the example. To make things more interesting, lets add a second charge; both charges will
now be positive, and separated by a fairly small distance.
The electric potential at any point in space will be the sum of the potential due to each charge alone, so
potentials act exactly as electric elds do in this case. (A notable dierence is that the potential is a scalar
eld, however, while the electric elds is a vector eld.)
Because both of our charges are positive, the potential due to these charges will always be either positive
or zero. There is no negative charge anywhere, and so the potential cannot be negative, either.
Lets say one charge is +4, while the other is +1.
Very close to to +4 charge, the equipotential surface will be a near-perfect sphere, concentric with the
charge. Slightly further out, but still closer to the +4 charge than the +1, the surface will be a distorted
sphere, with is being stretched out towards the second charge.
Further out yet, there will be an (innite number of) equipotential surface(s) that enclose both charges,
and is shaped sort of like the symbol, except asymmetrically: the +4 charge is much stronger, and so
its equipotential is larger.
Further away yet - far away that the +4 and the +1 are essentially a point charge of +5, the equipotentials
16
will again start looking like more and more perfect spheres.
If we change the +1 charge to a 1 charge, what happens? Well, rst o, because we now have a mix of
positive and negative charges, we will now have both positive and negative equipotentials. Also, since we
have a mix, and there will be a transition between the two, there will now be points in space where the
potential is zero.
Other than that, the situation is similar to what you would expect. Close to the 1 charge, the equipo-
tentials are again spherical, but they are of negative potential. That is, if you bring a positive test charge
from innity up to that negative equipotential, you will have done net negative work. Despite having to
work your way through the positive potential of the stronger +4 charge, the attracting force of the 1
charge wins out in the end. And, since the electric force is conservative, bringing such a test charge to a
point in space where the potential is zero means you have done a net zero amount of work.
Because we now have a mix of positive and negative charges, there will both be a point where the electric
eld is zero, and a point where the potential is zero. These point are not one and the same, however!
Positive charges will always travel from high potentials to low potentials, which is analogous to gravity:
object always fall downwards (towards lower gravitational potential), and never upwards.
In electricity, unlike gravity, we also have negative charges. Negative charges always travel from low
potentials to high potential, so they essentially go in reverse.
Electric potentials: example
Lets dene two points, A and B, with electric potentials V
A
and V
B
, respectively. They are separated by
a distance R.
We know that
V
A
=
_

A

E

dr
V
B
=
_

B

E

dr
And therefore,
V
A
V
B
=
_
B
A

E

dr
V
B
V
A
=
_
B
A

E

dr
The innitesimal vector

dr implies (by convention) that we go in a straight line between the two points.
However, as stated previously, the electric force in conservative, and so it doesnt matter which path we
take. We can replace

dr with

d to imply this:
V
A
V
B
=
_
B
A

E

d
V
B
V
A
=
_
B
A

E

d
Now, then. Lets calculate an example. Say V
A
= 150 volts, and V
B
= 50 volts. Therefore, V
B
V
A
= 100
volts. If we release a free electron at point B, it will go towards V
A
- electrons travel towards higher
potentials, while a proton would travel away from it.
In this calculation, we will use an electron. The electron has a mass and a charge, of course:
m
e
= 9.10938 10
31
kg
17
Q
e
= 1.602 10
19
C
We will need those values momentarily.
Via the Work-Energy theorem, the kinetic energy of the electron will, at the end of its journey, have used
up all of its electrostatic potential energy. We know how to calculate the latter - its U = V q, and we
know both the potential and the charge. The kinetic energy of a particle is (in classical physics)
1
2
mv
2
.
We can now set the sum of the two equal to zero, substitute in the values for the situation, and solve for
v:
1
2
mv
2
+ V q = 0
mv
2
= 2V q
v = +
_
2V q
m
v = +
_
2 (100) (1.602 10
19
)
9.109378 10
31
v 5.93 10
6
m/s
Thats roughly 2% of the speed of light, despite the relatively small potential dierence (compared to
multi-kilovolt Van de Graa generators, at least)! Also note how the distance between the two points does
not matter - the ending velocity of the particle is the same either way, as can be seen by the fact that
there is no distance specied anywhere in the equation.
2.1.3 More on electric potential
Lets say that we are at a point P in space. At that point, there is a potential V
P
, as well as an electric
eld (it may be zero, of course).
We now take a very small step in the x direction, and the x direction only - y and z are kept exactly
the same. We measure the electric potential at this new point. If the potential is the same, then the x
component of the electric eld here is, by denition, zero. If it is not, the magnitude of that component is
E
x
=
[V [
x
The same holds, of course, for the y and z directions.
Note how the fraction above has units
V
m
, while describing an electric eld. We have previously used
N
C
to describe electric eld magnitudes, but the two are in fact equivalent: they are both
kg m
A s
3
in base SI
units.
Because volts per meter are more intuitive for visualising potential dierence per unit length, we will
mostly use those units from here on.
We can calculate the electric eld from a known potential, by taking the negative gradient of the potential:

E = V = grad V =
_
V
x
x +
V
y
y +
V
z
z
_
The partial derivatives shown make up the components of the vector eld:
E
x
=
V
x
E
y
=
V
y
18
E
z
=
V
z
Lets try an example, in one dimension. We have the electric potential
V = 10
5
x
which is valid over a small range of x, from 0 to 10
2
, so a distance of 1 cm.
At x = 0, we have a large plate A, and at x = 10
2
m, we have another plate B. B has a positive charge,
while A has a negative charge.
Knowing the potential, we can calculate the electric eld using the formula above. Since the y and z
components are both zero, we just take the negative derivative, and multiply it by x, as above, and get:

E = 10
5
x
We know since before that
V
A
V
B
=
_
B
A

E

d
... but in this case, we know that the movement is strictly in the x direction, so we can use

dx instead of

d. Lets also substitute in the value for



E, so that we get:
V
A
V
B
=
_
B
A
10
5
x

dx
Well, now what? Well, rst o, 10
5
is a constant, so we can move it outside the integral. Second, x and

dx are by denition in the same direction, so the cosine of the angle between them is one. Therefore, we
can get rid of all the vector stu and reduce the integral to
V
A
V
B
= 10
5
_
B
A
dx = 10
5
(x
B
x
A
)
If we plug in x
B
= 10
2
and x
A
= 0, we get 1000 volts, which is our answer. That is, A is 1000 volts
lower than B. The electric eld between the two plates point in the x direction, i.e. the left, while
the potential grows towards the right. Thats the physical meaning of having the electric eld being the
negative of the derivative of the electric potential.
2.2 Electric elds inside hollow conductors
The electric eld is always zero everywhere inside a solid conductor. The same is not necessarily true in
a non-conductor. In a conductor, it is true because if there is an electric eld, the free electrons (that
dene a conductor) would move around until the eld is cancelled out, and they no longer experience a force.
Imagine a solid (non-spherical, to show that spherical symmetry is not required) conductor. We add a
positive charge to this conductor. The question is, now: how does this charge distribute itself?
We know that it does so in a manner that ensures there is no electric eld inside the conductor (once it
has settled down), but that doesnt answer the question of exactly where the charge is.
Lets rst imagine that it spreads itself out evenly everywhere inside the solid block. Could this be?
No, it could not. We know that the electric eld must be zero inside the conductor. Now, lets choose a
Gaussian surface, e.g. a small sphere, placed at any point inside this solid conductor. Recall Gausss law:
=


E

dA =

Q
ins

0
19
We know that [

E[ is zero, since this is inside the conductor. Since the left side (the integral) is zero, we
can only conclude that

Q
ins
must be zero. There can be no charge anywhere inside the conductor -
only on the surface - or this could not be true!
So, the charge is on the surface. It is not uniformly distributed, however! That would only be the case if
the conductor were a sphere.
Now, lets see what happens if we make it hollow. That is, there is an inner surface and an outer surface
(with a nonzero amount of the conducting material between them), but the shape remains the same (re-
member that it is non-spherical!).
We now have two surfaces - one outer and one inner. Will the situation change? That is, will some of the
charge now be on the inside surface as well?
The answer is no. We can use the same argument: imagine a Gaussian surface that is located just inside
the outer surface, such that it envelops the entire shape, minus the outer surface. The Gaussian surface
is inside the conductor at all points, which means the electric eld must be zero everywhere on it. That,
in turn (just as above) means the

E

dA integral will be zero, and so there cannot be any charge enclosed
by the surface.
The charge is still exclusively on the outer surface!
Lets take it another step. We take a similar hollow conductor - a sphere, now - except we bring a charge
+Q inside the (closed) conductor... somehow.
The same rules still apply: the electric eld inside the conductor must be zero.
Using the same Gausss law argument, we can show that there must, in fact, be negative charge on the
inside surface now!
Once again, the

E

dA integral along the Gaussian surface - again chosen to enclose the conductor minus
the outer surface - must be zero, since the electric eld is everywhere zero in the conductor.
This means also that the right-hand side of Gausss law -

Q
ins

0
- must be zero! That can only happen
when the net charge inside is zero. In other words, if we bring a positive charge to the inside of the hollow
conductor, nature will cause negative charge to accumulate on the inside surface to cancel it out.
Do note that the charge will be on the inside surface only - never in the middle of the conductor. The same
argument with the small Gaussian surface as before holds: if you enclose only a tiny bit of the conductor,
the electric eld must be zero everywhere, which means the surface integral of

E

dA must be zero, which
means there can be no charge inside the conductor itself.
Now, because this conductor was neutral when we began this experiment, and brought charge inside the
hollow cavity, it must still be overall neutral. Since negative charge has gathered on the inside surface,
the same amount of positive charge must have gathered on the outside surface now!
Whats more, in order to obey the laws of physics, this charge must be uniformly distributed, since the
conductor is spherical. This would not be the case for a non-sphere.
The net eect of this is, then, that since the charge distribution on the surface is uniform, what happens
on the inside of the sphere is completely un-observable to the outside world. We can move the charge inside
the hollow cavity around, and the external electric eld will be unaected.
This eect is known as electrostatic shielding, and the conductor will often be known as a Faraday cage.
2.3 High-voltage breakdown and lightning
We now know that charge distribution is non-uniform on all non-spherical surfaces (surfaces, since we also
know that all charge will reside at the surface for solid conductors).
More specically, the charge density will be higher at regions of higher curvature, which we will now show.
Imagine two conducting spheres, far apart, but connected with conductive wire. The two spheres are
20
together an equipotential, because of the connection between them.
Lets call one A, with radius R
A
and charge Q
A
, with the other being B with radius R
B
and charge Q
B
.
Because of the large separation between the two, the potential around A is not dependent upon B. That
is, the work required to bring a charge from innity to A is essentially the same with or without B.
Therefore, we can calculate the potential of A to be:
V
A
=
Q
A
4
0
r
... but since the two are an equipotential, that must also be the potential of B, which also must follow the
same formula itself:
V
B
=
Q
B
4
0
R
B
=
Q
A
4
0
R
A
Therefore,
Q
A
R
A
=
Q
B
R
B
This must be true regardless of the radii involved, so imagine B having a radius that is 5 times larger than
A. This implies that there is also 5 times more charge on B than there is on A. However, the surface area
of B is 25 times larger than that of A.
The surface charge density is the total charge on the sphere divided by the surface area, so
=
Q
4R
2
For these facts to be able to co-exist - that the surface area is 25 times more, but the charge is only 5
times more, we get that

B
=
1
5

A
merely because of the larger radius of B. Therefore, the smaller the radius, the higher the local charge
density. And, via Gausss law, we can show that this also implies that the local electric eld will be
stronger at such points of high curvature.
There results apply to all shapes. An otherwise spherical surface with a small, pointy outwards tip will
have a stronger electric eld at that tip, due to the higher charge density.
2.3.1 Electric breakdown
When electric elds become too strong, electric breakdown will occur. This is what happens when you
have two points with vastly dierent electric potentials, and a spark forms and transfers charge, while also
creating sound and light (the spark).
It is fairly easy to see why this happens. Imagine an electron traveling in an electric eld. It will travel in
the opposite direction of the elds lines (since it is negatively charged).
It will accelerate in the eld, and it will collide with molecules (nitrogen, oxygen etc.) in the air in travels
through.
When the kinetic energy of the electron becomes high enough to ionize molecules in the air, there will now
be multiple electrons traveling. They, too, accelerate in the electric eld, ionize further molecules, etc.
If this keeps up, as it will if the electric eld remains, a conductor is formed from the ionized molecules and
electrons (which now make up a plasma). Charge is transferred, and as the molecules recombine to form
21
neutral atoms/molecules again, they will emit photons, which is why we see a spark. They also produce
a pressure wave, which is why we can also hear the sparks.
Because of what we discussed above, this will typically occur at sharp points, where the electric eld is
the strongest.
Very roughly, this breakdown voltage - at room temperature, with dry air, at standard pressure - is about
3 MV/m. So to produce a spark 10 cm long, for example, you would need a tenth of those 3 million volts
per meter, i.e. 300 kV.
22
Chapter 3
Week 3
3.1 Capacitance and Field Energy
3.1.1 Field Energy
Say we have a system with two large plates, separated by a distance h. One has charge +Q = +A, where
A is the surface area, and the other the exact opposite charge Q = A (A is the same for both plates).
Now suppose we move one of the the plate a distance x, so that they are further apart. The plates attract
each other, so its clear we must do work in order to separate the two. The work required to separate them
is the force required times the distance the force acts,
W = [

F[ x
since the electric eld is constant.
Finding the force required is simple, yet a bit tricky. We know that the electric eld strength between the
plates is

0
but we have to keep in mind that while the plate will be very thin, it does have a nonzero thickness.
We know that charge will gather on the surface, but in reality, that too cannot be of zero thickness.
So, the electric eld below the plate is

0
, half of which is due to the bottom plate. However, the electric
eld inside the plate is zero - it is a conductor, so that must be the case.
Because of this, the electric eld we need to use in our calculation is only half of what we might expect:
F =
1
2
QE
Therefore, the work is
1
2
QEx
We know that Q = A, so lets substitute that in there:
1
2
AxE
If we multiply by

0

0
, i.e. 1, we nd another interesting result:
1
2
AxE

0

0
23
We now have another

0
= E in there, so we can simplify the expression to:
1
2
AxE
2

0
Now we note that Ax is the new volume that we have created an electric eld in. Remember that A is the
plate area, and x the distance we moved it.
Therefore, we can nd that the work required per unit volume is
W
volume
=
1
2

0
E
2
... which is known as eld energy density.
Clearly, the units for this would then be joules per cubic meter,
J
m
3
.
This result is valid for all charge congurations, not just the one system we just calculated. Therefore, if
it is more convenient for us, we can now use this instead of calculating the electric potential energy due
to each point charge alone, by integrating over all space:
U =
_
1
2

0
E
2
dV
... where V is for volume, not the electric potential!
Lets calculate the electric potential energy for the plates above - which by the way is a parallel plate
capacitor. We have the integral we require just above; lets substitute in E =

0
in there:
U =
_
1
2

0
(

0
)
2
dV
Because we know the volume of the box to be Ah, and we know the electric eld outside the plates to
be zero (the plates have exactly opposite elds, so they cancel out outside of the plates), we dont need to
integrate, but get simply
U =
1
2

0
(

0
)
2
Ah =
1
2

0
Ah
We know since before that
Q = A
V = Eh
... where V is now the electric potential again. Sadly, we use the same symbol for potential and volume.
We can use the rst equivalency to get Q in there:
U =
1
2

0
Qh
And since E =

0
, we can get rid of that and the h by using V instead:
U =
1
2
QV
3.1.2 Capacitance
Lets now introduce the concept of capacitance, or for short, C (not to be confused with the unit coulomb
- just as W means both work and watt, despite work being measured in joules (watt-seconds)).
C =
Q
V
24
The unit of capacitance is the farad, F, named after Michael Faraday.
If we charge two objects up to have the same electric potential, then the one with the most charge on
it has the greatest capacitance. Thus the name - capacitance is the capacity to hold charge for a given
electric potential.
Using the denition of capacitance above, we can easily calculate the capacitance of a conducting sphere:
C =
Q
V
=
Q
Q
4
0
R
= 4
0
R
Since 4
0
10
11
(order of magnitude-wise), clearly capacitances will be extremely low even if the radius
is very high. For example, for a sphere of radius 1 meter, the capacitance is only on the order of 10
picofarads. An electrolytic capacitor, cylinder-shaped (the outside, but thats not how it works - more on
that later) with radius 1 cm and height 4 cm can have a capacitance about ten million times more than
that!
Lets now look at the case of two spheres A and B, side by side. Sphere A has charge Q
A
and potential
V
A
, while sphere B has charge Q
B
and potential V
B
.
The spheres are close to one another in space. What is the capacitance of B?
Well, we have the denition:
C
B
=
Q
B
V
B
However, keep in note what the potential V
B
really is. It is the work per unit charge you have to do to
bring a charge from innity up to the sphere. However, since A is nearby, that will charge the situation!
Carrying a positive test charge moving towards B, you will have to do work to overcome the electric force.
However, the negative sphere A will be attracting you at the same time, so the work you have to do
becomes less! Therefore, by denition, the potential of B has gone down - and from the equation above,
clearly, the capacitance of B has gone up!
Calling it the capacitance of B is a bit of a misnomer, its really the capacitance of B in the presence of
A.
Therefore, we will change the denition of capacitance to be inversely proportial to the potential dierence
between two conductors:
C =
Q
V
... where +Q is the charge on one conductor, while Q is the charge on the other one.
Lets now calculate the capacitance of two concentric hollow spheres.
We know the electric eld due to the inner sphere:

E =
Q
4
0
r
2
r
... and also know that the contribution to the electric eld inside the larger sphere, from the larger sphere,
is zero.
Therefore, we use the integral for a potential dierence over this electric eld:
V
R1
V
R2
=
_
R
2
R
1

E

dr =
Q
4
0
_
R
2
R
1
dr
r
2
=
1
r

R
2
R
1
=
_
Q
4
0
__
1
R
2

1
R
1
_
Flip the order to get rid of the minus sign:
25
V
R1
V
R2
=
Q
4
0
_
1
R
1

1
R
2
_
We can now get nd the capacitance C by dividing Q with the potential dierence above:
C =
Q
V
= 4
0
1
_
1
R
1

1
R
2
_ = 4
0
R
1
R
2
R
2
R
1
The last step is simply a simplication to make the expression easier to read, i.e. without nested fractions.
Lets now go back to our parallel plate capacitor, and calculate its capacitance - now we know to use the
potential dierence between the two plates.
We had that
C =
Q
V
Q = A
E =

0
... and V will as usual be the integral of the electric eld between the two plates. However, the eld is
constant, and so we can multiply the eld times the distance instead of a hairy dot-product integration.
Lets call the separation between the two plates d:
C =
Q
V
=
A
Ed
Lets substitute the value for E in there:
C =
Q
V
=
A
0
d
So now, the sigmas cancel:
C =
A
0
d
The capacitance is linearly proportial to area, and inversely proportial to d. The last part goes together
nicely with the spheres - the closer they are, the more one will aect the work required by the other. Closer
together means lower potential dierence, which means higher capacitance - since capacitance is inversily
proportional to the potential dierence.
3.2 Dielectrics
Lets now go back to our good old parallel plate capacitor. We will add a new twist to it shortly, but to
begin with, its the same, only with some new namings.
We have two large plates, each with area A, separated by a distance d. We put charge +Q on the top
plate, and thus get a charge Q on the bottom plate (if we do this the easy way, by connecting a voltage
source across the plates).
We call the charge density on the plates
free
, and the electric eld between the plates

E
free
=

free

0
.
The electric eld is in the downwards direction, since the top plate is positive, and the bottom plate is
negative.
Now that we have changed the plates with a voltage source, we disconnect the voltage source, thus trap-
ping the charge on the plates. Q is now a constant, no matter what we do - except connect the plates to
26
a conductor, of course. Things that we still can modify, however, include the electric eld strength, the
capacitance, and the potential dierence between the plates.
Now, lets insert a linear dielectric material between the two plates - remember, with the voltage source
disconnected, and the charge trapped. For fairly long-winded reasons I wont reproduce here (see the book
on polarization and dielectrics), this will cause a negative induced layer of charge on the bottom on the
top plate, and a positive induced layer of charge on the top of the bottom plate. That is, the charge will
be on the inside of the plates, close to the empty space between them.
We can call this charge density
ind
, for induced; thus the top plate, with charge +
free
, will also have a
charged
ind
, while the bottom plate will have
free
and +
ind
.
This induced charge will produce an electric eld, of course. Because the induced charge on the bottom is
positive, and the induced charge on the top is negative, the electric eld E
ind
will be in opposite direction
of E
free
, which was due to the charge we put on the plates. Therefore, the net electric eld between will
be lowered, since this induced-charge electric eld cancels out part of the original eld.
The net electric eld will as usual be the vectorial sum of the two; since they are exactly opposite in
direction, and its a fact that the induced eld will be smaller in magnitude than the original one, we get
that

E
net
=

E
free


E
ind
The induced charge will be some fraction of
free
. If we call it b, we have that
b
free
=
ind
(b < 1)
Therefore, we also have that
b

E
free
=

E
ind
It follows that

E
net
= (1 b)

E
free
We call that 1 b =
1

(kappa) or
1
K
. is known as the dielectric constant. It is a dimensionless number,
and is 1 for a vacuum. It is greater than one for non-vacuums.
will be greater for materials which are themselves dipoles (even in the absence of electric elds), than
for materials which will contain induced dipoles from the electric eld.
Another way of writing the previous equation, then, is

E =

E
net
=

E
free

From here on, in this section,



E will now refer to the net electric eld, i.e. the original eld minus the
induced eld, to keep things simple (as these are mostly lecture notes, and thats how Walter Lewin did
in lecture).
All in all, we see that by inserting a dielectric material between the plates, with the power supply discon-
nected and the charge trapped, the electric eld strength has gone down by a factor of . Since V = Ed
in this case of a constant electric eld, and the distance d between the plates has remained constant, then
the potential dierence must also have decreased by a factor of .
What about Gausss law, then? Well, we have that
27


E

dA =

Q
ins

0
How does this change in our situation now? Well,

Q
ins
is the net charge inside, such that a proton and
an electron would come out to 0, as the cancel each others charge out exactly.
The charge that is relevant in this case, then, is Q
net
= Q
free
+ Q
ind
, where Q
free
is positive and Q
ind
is
negative.
Theres a very easy way to take this into account: the factor perfectly captures the amount of charge
that is cancelled out or induced, so:


E

dA =

Q
free

is all we need to calculate.


We can also write it as


E

dA =

Q
free

where =
0
is called the dielectric permittivity (compared to
0
which is the permittivity of free space,
also known as the permittivity of the vacuum).
Lastly, we can write it as


D

dA =

Q
free
where

D =
0

E is called the electric displacement vector.


We will now do four experiments, each with a notable dierence from the previous, while still being very
similar.
First, lets reproduce a few equations that will be necessary here:
C =
Q
V
V = Ed
Experiment one: disconnected power supply, no dielectric
We charge two circular plates, with an initial separation of 1 mm, to a potential dierence of 1500 volts.
Once charged, we disconnect the power supply, so that the charge is trapped on the plates.
We then increase the separation from 1 mm to 7 mm. The question is: what will happen to the electric
eld, and what will happen to the potential dierence?
The electric eld must remain unchanged, because E =

0
, and is constant since the charge is trapped
on the plates!
For the same reason, however, V must increase by a factor of seven! Note that V = Ed; with E held
constant, and d increasing, V must clearly increase!
This is also exactly what the experiment shows when the plates are separated: the potential increases, but
that is the only change.
28
Experiment two: disconnected power supply, dielectric inserted
We start this experiment where the previous one ended; at d 7 mm and V 10500 volts (7 times higher
than the initial 1500).
Now, we insert a dielectric - a glass plate, with a of roughly 5.
What happens with the free charge, the E-eld, and the potential dierence between the plates?
Back to the equations!
First o, Q
free
is still trapped and so cannot change this time around, either. However, there will be an
induced E-eld from the induced charge, and so the E-eld will decrease - by a factor of .
With V = Ed, and E decreased with d held constant, clearly V will also decrease by a factor of .
The most interesting, and useful, change is with the capacitance, however! Since C =
Q
V
, and Q is held
constant while V decreases, C must increase by a factor ! Thus, capacitance as related to the geometry
of the parallel plate capacitor must be amended by this factor:
C =

0
A
d

Experiment three: power supply remains connected, no dielectric
We start over like experiment one: 1 mm separation, 1500 volts, no dielectric; however we leave the power
supply in there the entire time.
Because the power supply is now connected, with a xed potential dierence of 1500 volts, the results of
this experiment will be very dierent from the rst one. With V xed, we expect that other parameters
will have to change than ones that did last time.
Again, V = Ed, so when V is xed and d is increased, E must clearly decrease! As for the capacitance,
the only variable that changes in the above equation is d, so the capacitance must also decrease by the
same factor. Lastly, because C =
Q
V
or equivalently Q = CV , with V xed and C decreasing, Q must
decrease! This is now possible, since the power supply is connected and allows for charge to move.
Experiment four: power supply remains connected, dielectric inserted
We now move to the fourth experiment. We start where we left o in experiment three. 7 mm separation,
1500 volts with the power supply connected, and the glass plate is inserted between the plates.
We take a nal look at the equations to predict what will happen. The capacitance will clearly increase
by a factor , as capacitance is only related to plate area, separation between plates and . The latter is
the only which changes, from roughly 1 (the of vacuum, and air is extremely near 1 as well), to roughly
5, the of glass.
What about the free charge? Well, rst, lets look at the electric eld strength. V = Ed, and so E =
V
d
.
V is held constant by the power supply, d is constant, and so E must remain constant.
However, we know that inserting a dielectric will cause induced charge to appear on the plates, which
lowers the net electric eld by a factor !
For these two facts to be able to coincide, the free electric eld must increase in order to compensate
for the induced, antiparallel electric eld, in order to keep the net eld strength exactly the same as before.
Therefore, the free charge will increase by a factor of , to balance out the reduction in eld strength that
would otherwise occur.
29
3.2.1 More on capacitors
If we want to design a capacitor with a very high capacitance, how should we proceed?
Well, clearly, from the parallel plate capacitor equation, we want to maximize the area A, minimize the
distance d between the plates and maximize the dielectric constant . Those are all the variables in the
equation, so we cant do much more!
Lets compare two dierent 100 microfarad capacitors. One is rated for 40 volts, and the other for 4000
volts(!). We use polyethylene as a dielectric; it has a of roughly 3, and can tolerate roughly 18 MV/m
before breaking down.
Because of this breakdown limit, the only way to make very high-voltage capacitors is to increase the
dielectric thickness d (which is then also the plate separation distance). However, doing so reduces the
capacitance, so you need to increase the plate area to compensate! For this reason, high-voltage, high-
capacitance capacitors are physically very large. The 4 kV, 100 F capacitor mentioned looks like a car
battery, while the 40 V one is roughly 1 cm
3
in volume!
Lets do a back-of-the-envelope calculation. Say we have a plate area of 1 m
2
(for easy numbers sake, but
since the plates can be rolled up, its not an extreme value as it sounds, either), with a separation of 0.01
mm, with air between the plates. The capacitance is, according to the capacitance equation, roughly 0.88
F, and the maximum voltage before breakdown, at 3 MV/m for air, about 30 volts. In practice it will
probably be less due to imperfections etc, but this is a back-of-the-envelope calculation!
Lets now insert a dielectric of = 3 between the plates, which can also withstand about 18 MV/m before
breakdown. First, multiplies the capacitance by three, so if we are happy with the capacitance, we can
immediately reduce the plate area by a factor of to retain our 0.88 F. Thus the area is now
1
3
m
2
vs
the old 1.
Second, note that the dielectric can take a lot more than the 30 volts at the same distance now. If were
happy with the 30 volts maximum, we can reduce d by a factor of
18
3
(new / old breakdown limit), from
0.01 mm (10 m) to 1.666 m. The new maximum voltage will still be 30 volts, since we reduced the
distance by the same factor.
However, now that d is 6 times smaller, we can reduce A by a further factor of 6, getting it down from
1 initally to just 0.055 m
2
now, without any sacrice in capacitance or voltage tolerance! All we did was
insert a dielectric and reduce the distance - which we couldnt have done without the dielectric. Clearly,
dielectrics are very important for capacitors - especially when you want higher values of capacitance,
perhaps in the millifarad range and above, without making truly enormous (car battery-sized or even
larger) capacitors.
3.2.2 A bit on van de Graa generators
Imagine a fairly big, hollow, conducting sphere. We can use a high-voltage power supply, say 3 kV, to
bring charge on the sphere: we charge a small conductor, bring the two in contact, and repeat the process
until no charge is transferred any longer.
Clearly, that point will occur when the large sphere has the same electric potential as the smaller conduc-
tor. If we want to charge the big sphere, the van de Graa, to say 200 kV, this method is useless.
What we can do instead is to charge the small conductor in the same way, but instead bring it inside the
large, hollow sphere, through an opening at the bottom.
30
The electric eld inside will always be zero - not quite, since its not a perfect sphere, but close enough.
Even when the potential at the outside of the sphere and the small conductor are equal, we can still
transfer charge to the inside of the large sphere!
We can use this method to charge the van de Graa to any voltage we want, until it goes into electric
breakdown, which we cant really prevent. Breakdown is what will ultimately limit the amount of charge
we can get on the sphere, and thus ultimately the voltage.
So, how does a real van de Graa generator work? Clearly, it is not through a person manually moving
charge!
First, there is a motor that drives a belt. The belt goes from the outside of the sphere all the way inside.
We put charge on this belt, using two sharp tips that cause corona discharge. The charged belt travels up
inside the sphere, where another two sharp points through corona discharge move the charge to the inside
of the hollow sphere. Of course, the charge will move to the outer surface almost instantly, thus charging
the sphere a bit more.
3.3 Current, resistivity and Ohms law
Current is a ow of electric charge. By convention, we say that if a current goes towards the right in a
conductor, positive charges would move towards the right. Negative charges would then move towards the
left, in the opposite direction of the current ow.
While this causes some confusion, electron ow is the most common form of current. Thus, when we
talk about a current owing towards the right in e.g. a circuit diagram, the electrons are in fact moving
towards the left. However, most of the time, it doesnt really matter in which direction the electrons ow,
as long as we are consistent about this.
This convention where the electrons ow opposite to the current is called conventional current. The other
convention, where they ow in the same direction, is called electron ow.
We will stick to convential current in these notes, however.
Electrons are the main charge carrier since there are free electrons in conductors, which will move when
we apply a potential dierence over a conductor. Such a potential dierence will cause an electric eld
to appear inside the conductor, and the electrons will attempt to move in order to neutralize the electric
eld (since a force will be applied to them).
Now, lets do some dirty math to try to derive Ohms law. This will be a quick and likely hard-to-fully-
grasp explanation, as its overly simplied and rushed through. Apparently quantum mechanics is required
for a proper derivation.
In copper, at room temperature, the average speed of the free electrons is the metal is on the order of 10
6
meters per second; this is a random, thermal motion due to the temperature.
The time between collisions between the atoms and the free elecrons is roughly = 3 10
14
( is the
Greek letter tau).
The number of free electrons is roughly n = 10
29
per m
3
; roughly one per atom.
When we apply a potential dierence to the conductor, the free electrons will experience a force of mag-
nitude F = eE, where e is the magnitude of the electrons charge (the elementary charge e), and E is the
electric elds magnitude. Via Newton, we know that the electrons will now experience an acceleration of
a =
F
m
e
=
eE
m
e
, where m
e
is the mass of the electron. Because of the accelerations, they will pick up an
average speed between the collisions, of v
d
= a, where v
d
is known as the drift velocity.
If we combine the two equations, we have
31
v
d
=
eE
m
e

If we use this equation to calculate the drift velocity in a 10 meter long copper conductor, with a potential
dierence of 10 volts (and thus an average electric eld of 1 volt/meter), we get a drift velocity of roughly
v
d
=
(1.6 10
19
)(1)
9.1 10
31
(3 10
14
) 0.0053 m/s
That is in fact not a miscalculation - the drift velocity will actually be on the order of millimeters per
second! The random thermal motion causes the electrons to move at a million meter per second, and our
electric eld of 1 volt/meter doesnt even break
1
100
m/s!
Of course, without the electric eld, the net movement of the electrons would be zero - the motion is
random, and so you would expect that on average, they return to where they were.
Lets imagine now a wire, shaped like a cylinder. It has cross-sectional area A, length , and we apply a
potential dierence V such that the electric potential is higher on the right side.
This will cause a current I (the symbol for current is I) to travel in the leftwards direction, which is also
the direction of the electric eld E =
V

. Of course, the electrons will travel towards the right, opposite


to the electric eld vector.
In one second, they will travel a distance of v
d
meters to the right, by our denition of the drift velocity.
Lets look at how we can calculate the current in terms of the drift velocity and such things.
If we take a cross-section of the conductor, with area A, the volume that moves through per second wil
be v
d
A, since v
d
is the length the electrons move per second. We multiply that by n (free electrons per
cubic meter) to get the number of free electrons that pass through per second. We then multiply that by
e, the magnitude of the electrons charge, to get the total charge that passes per second.
Therefore, the current that ows is the product of the above terms:
I = v
d
A n e
We found v
d
previously, so we can substitute that value for it in there:
I =
_
eE
m
e

_
A n e =
e
2
n
m
e
AE
The rst term there, in the fraction, depends only on the conductor (for a given temperature). We denote
that by , and call it conductivity.
=
e
2
n
m
e
we be roughly 10
8
for copper at room temperature, and the units are
A
V m
, or equivalently,
S
m
(siemens
per meter), as 1 S =
1 A
1 V
.
Because E =
V

, we can also write the current as


I =
e
2
n
m
e
AE = AE =
AV

We can solve that for V , as well:


V =

A
I
32
... which is, in fact, Ohms law, since R =

A
.
We will often use =
1

, which is then known as the resistivity of a material. ( is the Greek letter rho.)
Note from the above equations that the current (for a given potential dierence) is proportional to the
conductors area, and inversely proportional to its length. This makes a lot of sense if we use the common
analogy of current being water, electric potential being water pressure, and resistance being, well, the
resistance a pipe causes. Clearly, a longer pipe will require more pressure to maintain the same current
(current is inversely proportional to ). Also, a wider pipe will allow more current to ow (current is
proportional to A).
Lets move away a bit from all these equations and their symbols, and put some numbers in there, instead.
Say we have a chunk of copper 1 mm x 1 mm x 1 m in size. The cross-sectional area is then 10
6
m.
What is the resistance? Well, we and A. is roughly 10
8
, so the resistance R is about
1
10
8
10
6
= 10
2
.
What about insulators? A good insulator - example include glass, quartz, rubber and porcelain - can have
resistivities as high as 10
12
to 10
16
, with conductivities () of one over those numbers.
For example, if the material had = 10
14
(equivalent to = 10
14
), the resistance R would be about
10
20
! With such high resistances, clearly only extremely small currents will ow even for very high
voltages.
Ohms law has many downsides. Its only valid for a subset of circuits elements, and only works in the
ideal case.
The conductivity (and thus the resistivity ) is a strong function of temperature. If the temperature
increases, then the random, thermal motion of the electrons will increase. In turn, , the time between
collisions, will go down. If we have a look at the denition of :
=
e
2
n
m
e
its clear that this means that the conductivity will go down if the temperature goes up. This has some
very big ramications.
As a real-world example, we can take a simple light bulb. Consider a small light bulb; it has a 50 ohm
resistance - when hot. When cold, the resistance is closer to just 7 ohm. What happens now? Well, the
instant we switch the circuit on, the bulb is cold, and the current will be I =
V
7
. Say V = 12 volts. In that
case, the power dissipation (which we havent discussed yet in this course) will be 20.5 watts. Considering
the tiny size of the light bulb in the example, that is a lot of power. It cant possibly keep that up, as it
will get too hot.
Instead, what happens is that the power dissipation will heat the bulb up extremely quickly, and so the
resistance will increase to the 50 ohms in about a second, if not even less. At that point, I =
V
50
, and
for 12 V, the power dissipation is now 2.88 watts, a much more reasonable gure.
Clearly then, the voltage-current relation of this light bulb is far from linear! It will be fairly linear when
the bulb is hot, but if the voltage were to increase or decrease, then the bulb would either heat up or cool
down, and so the curve would yet again be a function of temperature and non-linear.
33
Chapter 4
Week 4
4.1 Batteries and EMF
If we draw up a very simple circuit with a voltage source connected to a resistor, we nd that the current
(conventional current) from the + side of the supply, through the resistor - in through the positive side of
the resistor and out the negative, and back into the power supply.
The electric eld always points from a higher potential to a lower potential, so in the resistor it points
along with the current.
What about in the power supply? Well, we just said that the electric eld always points from a higher
potential to a lower, so we nd that the electric eld opposes the current! The current must enter the
terminal and ow out of the + terminal, while the electric eld must point from + to !
Therefore, there must be a mechanism which forces the current to run despite the opposing electric eld;
a mechanism which does work to overcome the electric force.
Lets consider a chemical battery. We have two plates, one zinc (Zn) and one copper (Cu) in a solution of
sulfuric acid (H
2
SO
4
).
In this solution, ions will form: Zn
++
, Cu
++
and SO

4
.
We connect this battery to a resistor R, and a current ows. If we now look at the current direction
and the electric eld direction, we will nd the same result as before: the align in the resistor, but are in
opposite directions inside the battery. So, again, the charge carriers in the solution must ow in a direction
that would require work to be done.
Why does this happen? The answer is that in doing so, they participate in a chemical reaction which will
net more energy than is spent on ghting the electric force.
When measuring the potential dierence of an open circuit, such as an otherwise unconnected battery, we
call that voltage the electromotive force, or EMF (sometimes in lowercase, so emf), for which we use the
symbol E, so a curly E, a bit like a large lowercase epsilon, but not quite.
Since all materials (except superconductors, which we may encounter later on) have a nonzero resistance,
the battery will have an internal resistance. Therefore, if we short-circuit a battery - even with a su-
perconductor of 0 resistance, the current that ows will be limited by the internal resistance, such that
I
SC
=
E
R
int
where I
SC
is the short-circuit current.
When we draw a current from any power source, the voltage as measured across its terminals will drop,
due to the internal resistance. Via Ohms law,
E = IR = I(R
ext
+ R
int
)
Via Ohms law, there will be a voltage drop across the internal resistance,
34
V
Rint
= IR
int
and so the potential dierence that reaches the external resistor (note that voltage doesnt ow) is
V
Rext
= E IR
int
The higher the current, the lower the potential dierence. For this reason, if you buy a 9 volt battery, the
open circuit voltage will likely be a bit higher than that (when it is new). Under relatively heavy load, the
batterys internal resistance will get a voltage drop, and so the voltage seen by the external circuit may
well be less than 9 volts.
In a similar manner, if we short-circuit such a battery, then clearly the entire 9 volts will be across the
internal resistance, causing a large current, and lots of heat production.
The power is given by P = V I = I
2
R =
V
2
R
, so for V = 9 volts and R perhaps on the order of 2 , the
current will be large, as will the power dissipation and thus heat generation be.
Power is given by energy per unit time, and so the units are joules per second, or watts (W).
From experience, I can recommend not shorting 9 volt batteries for a long time - they get hot, the outside
starts melting, and they smell awful! Other than that, its relatively safe, however, unlike other battery
chemistries - lithium-ion batteries can cause res and explosions in the worst case scenario.
Now lets look at a car battery. Such a battery would usually be a lead-acid type; they have lead and
lead-oxide plates in a solution of sulfuric acid. They are nominally 12 volts, but usually a bit higher in
practice, especially when the car is running and the battery is being charged.
Such a battery has a very low internal resistance, perhaps on the order of 20 milliohms. Thus the maximum
current, if short-circuited, is on the order of 600 amperes, which would cause a power output of
V
2
R
7200
watts inside the battery!
Thats several times more power than a space heater, which might be 2 kilowatts or so. Needless to say,
short-circuiting a car battery may be a very bad idea.
4.1.1 Kirchhos rules
A set of very useful rules - sometimes called laws, but like Ohms law they are not always valid - are called
Kirchhos rules. The rst, KVL for Kirchos Voltage Law, states that
_

E

d = 0
Or, in plain English, that the potential dierence as measured when starting at a point, moving around a
loop, and coming back to the same point, must be zero. This is always true for a conservative eld, but
is not always true - time-varying magnetic elds are one thing that will make this untrue, as we will see
later on.
We can also state this in summation notation:
n

k=1
V
k
= 0
This rule is very useful for analysis of DC circuits, where we can pretend that it is always valid. It holds
for all loops that you can choose whereby you end up where you started.
Kirchhos Current Law, or KCL, states that the current that ows into a junction/node must ow out,
i.e. there cannot be a pile-up of charge. Thus if you have a T-junction in a circuit, the net sum of the
35
currents must be zero - either one goes in and two out, or two goes in and one out. There are no other
possibilities.
We can state this using summation notation as well:
n

k=1
I
k
= 0
4.1.2 Basic circuit analysis
Because this is such a pain to describe without circuit diagrams, and since adding circuit diagrams are an
even bigger pain, I will refrain from documenting this right now. If the course continues on with anything
more than the very basic analysis done in lecture 10, I will change this. If not, see the lecture videos.
4.2 Magnetic Field and Torques
In ancient greece, in the 5th century BC, it was already known that certain rocks attract iron. One of
these minerals is now called magnetite, named after Magnesia, a district where the rocks where plentiful.
This is of course also where the names magnet and magnetism come from.
Much later, these minerals were used to create compasses, 1100 AD, in China. Yet a century later, it was
found that magnets have two points of maximum attraction, known as poles. All magnets found have two
poles; if you split a magnet in two, you get two smaller magnets with two poles each.
We name these poles north and south; the north pole will point (approximately) towards Earths geo-
graphic north pole, while the magnets south pole points (roughly) towards Earths geographic south pole.
Opposite poles attract, and like poles repel each other, as with electric charge.
Because of this, the magnets north pole will point towards the Earth south magnetic pole, which some-
what confusingly is located near the Earths geographic north pole.
(The Earths magnetic eld shifts with time, and so it is not xed at the Earths geographic poles.)
Magnetic monopoles - magnets with only one pole - could exist, but despite much research, such a magnet
has never been observed to exist.
As with electric elds, we can visualize magnetic eld by using eld lines. Unlike electric elds, however,
all known magnets are dipoles. Therefore, all eld lines will begin at the magnets north pole and end at
its south pole.
With electric eld lines, the meaning of the eld lines are that at each point, the lines are tangent to the
force a positively chargd particle would experience.
For magnetic eld lines, every point is tangent to the direction in which a compass needle would point.
Thus we can trace out magnetic elds with compass needles, if they are large enough. At the very least
its a useful way to think about them.
In 1819, Danish physicist Oersted discovered that magnetic needles move in response to electric current,
thus linking electricity with magnetism, in the rst step towards a theory of electromagnetism.
If a current moves into this page (denoted by

, which looks like the rear end of an arrow), the magnetic


eld created will be in a clockwise circle around the wire. If a current were to move out of the page
(denoted by

, which looks like an arrow coming towards you), the magnetic eld created would then
be counter-clockwise. As we will soon see, magnetic elds are always perpendicular to the current that
creates them.
Well, if a running current causes a magnet to move by exerting a force, as Oersted showed, then by New-
tons third law, the opposite must also be the case. If we run a current through a wire that is placed in
36
an external eld, the wire will experince a force. The direction of this force is given by the cross product

F =

I

B, where

B is how we (always) denote the magnetic eld. (In this instance, we used the unit
vector

B =

B
[

B[
.)
If we run a current through a wire towards the right, like so: I
1

then the magnetic eld produced would, below the wire, be going into the page, and above, be coming out
of the page.
We can use the right-hand rule, used for determining the direction of cross products, here. Or, rather, a
related rule. If you point your right thumb in the direction of the current, with your remaining ngers
slighly curled, then the other ngers will show the direction of the magnetic eld. Try it out with your
right thumb pointing towards the right, and youll nd that indeed above the wire, your ngers point
towards you (out of the page), while the hand below the wire points into the page, and away from you.
If we add a second current I
2
in the same direction, in a wire just below I
1
, the magnetic force between
the two wires will attract the wires. We can convince ourselves of this by using the above nger curl rule
again.
4.2.1 The B Field
So how do we dene a magnetic eld? Unfortunately, since we have no magnetic monopoles, we cant use
a denition such as

F = q

B as we might like to.


Instead, the magnetic eld is described in terms of how it acts on moving charges:

F
B
= q(v

B)
where is the cross product (vector product). Due to the nature of the cross product, this result is not
only always perpendicular to te velocity of the particle, but also always perpendicular to the magnetic
eld, which is at least to me is unintuitive.
The above result is sometimes considered describing the Lorentz force, but that same term often refers to
the sum of electric and magnetic eects on a charged particle.
The equation is sign-sensitive: if any of the three terms is negative, then the force ips over 180 degrees.
(Of course, if exactly two are negative, the minus signs would cancel out.)
Via the above equation we can derive the units of the magnetic eld strength:
N = C(m/s B)
where B is a temporary name for the magnetic eld units. We can simply solve for B:
B =
N s
C m
= tesla
This interesting-looking unit is called the tesla, and has the symbol T. The tesla is a very large unit;
for example, the Earths magnetic eld is on the order of 50 T. The gauss (symbol G) is also often used,
despite being a non-SI unit. They are related such that 1 T = 10
4
G, so 1 G = 10
4
T.
Because the magnetic elds force is always perpendicular to a particles velocity vector, the magnetic
eld can never do any work on a moving particle! Work is dened as the dot product of the force and
the displacement, and with an angle of 90 degrees between them, the cosine term in the dot product will
always be zero.
This means that the magnetic force cannot change the kinetic energy of a particle - it cant increase it,
and it cant decrease it. It can and will force the particle to change directions, however. This fact is
exploited in CRT monitors and television sets - which are obsolete since about a decade. Such monitors
work by ring an electron beam towards a set of phosphors. The beam is scanned over the whole image
37
multiple times a second, and oriented by using magnetic elds to force the particles to where you want them.
The total electromagnetic force on a moving, charged particle is given by

F = q(

E +v

B)
... which is, as stated above, also known as the Lorentz force. These notes will either follow the convention
of the lectures, if there is one used often enough to be noteworthy, or it will call this total force the Lorentz
force.
Lets now return to looking at the force on a wire with a current through it.
Say that we have a charge dq moving in a wire. It moves forward with the drift velocity, v
d
(rather, the
electrons move backwards with that velocity, but we can pretend its positive charge moving forwards for
mathematical simplicity, as it all works out the same). We call this current I, and it runs through an exter-
nal magnetic eld

B, which may vary at dierent points of the wire. We call the angle between v
d
and

B .
Well, if we keep working with the dierential charge dq, we get that the magnetic force on the charge is

d
FB
= dq( v
d


B)
We also know that, by denition
I =
dq
dt
Therefore, we can substitute that into the previous equation, and get

d
FB
= Idt( v
d


B)
Whether we multiply I by dt, or do so to v
d
doesnt matter; mathematically, it is the exact same thing.
We can therefore think of this as

d
FB
= I( v
d
dt

B)
We now have velocity multiplied by time, which is simply a distance. Therefore, we can substitute the
drift velocity for a chunk of the wire,

d:

d
FB
= I(

d

B)
Of course, to get the total force on the wire, we need to sum up all of these tiny segments, which means
integrating the above:

F
B
=
_
wire
I(

d

B)
In the special case that the magnetic eld is constant, at least over a portion of the wire, we can do the
above calculation with simple multiplication of the three terms instead:
[

F
B
[ IB
4.3 Review for Exam 1
Say we have a solid cylinder, with uniform charge distribution (in 3 dimensions!) coulomb per cubic
meter, and radius R.
What is the E-eld outside the cylinder (r > R)?
Lets see. We use Gausss law, as always:
38
EA =
Q

0
A would be the surface area of the sides, so 2r. Q meanwhile we need to nd by multiplying the volume
with the charge density:
Q = R
2

If we merge the two equations, we have


2rE =
R
2

0
E =
R
2

0
2r
E =
R
2
2
0
r
Next up: what is the electric eld inside (r < R)?
Since the charge distribitions is uniform, and not just all on the surface, well treat this as a non-conductor.
Therefore the answer isnt immediately zero.
Well, well want to apply Gausss law again, of course. The left-hand side will be the same, only that
r < R this time. The right side must change, however: the amount of enclosed charge will change. The
enclosed charge will now be given by the radius of the Gaussian cylinder, rather than that of the real,
physical cylinder:
2rE =
r
2

0
E =
r
2

0
2r
E =
r
2
0
4.3.1 Thats it
Everything else covered this lecture has already been covered in these notes. While watching it again is
helpful, I dont see how reading two sets of notes, with the same content, is useful. Read the rest of the
notes twice, instead!
39
Chapter 5
Week 5
5.1 Moving Charges in Magnetic Fields
5.1.1 Moving charges, radii and special relativity
As we saw in lecture last week, the magnetic force will act to charge the direction of a moving charged
particle. The magnetic force will be in a direction perpendicular to the particles velocity, and also per-
pendicular to the magnetic eld itself.
If we repeat the calculation of the new velocity vector over and over, as the charge moves small distances,
we will nd that the particle travels in a perfect circle - given a uniform magnetic eld, and no other forces
being involved.
From classical mechanics, uniform circular motion is described by a centripetal force - a force acting on
the particle towards the center of a circle. We can equate the magnetic force on the particle, given by
q(vB), with the centripetal force.
If the magnetic eld is uniform, and we choose it such that it is perpendicular to v, the magnetic force is
given by

F
B
= qvB. Equating that with the centripetal force
mv
2
R
, we can solve for the circles radius R:
qvB =
mv
2
R
qvBR = mv
2
R =
mv
qB
The equation behaves as we would expect: the higher the mass, or the velocity, the greater the radius:
the particles inertia will make it harder for the magnetic force to chance its direction very rapidly.
Likewise, if the charge is high or the magnetic eld is strong, the magnetic force will be able to act more
strongly, and so the radius will be low.
We can also write this radius in terms of the potential dierence used to accelerate the particle.
Since electrostatic potential energy is given by qV , where V is potential dierence, and all such potential
energy will turn into kinetic energy after having accelerated it, we can equate that with the calculation of
kinetic energy:
1
2
mv
2
= qV
We can then solve for v, the velocity:
v
2
= (2qV )/m
v =
_
2qV
m
40
... and then substitute that into the equation for the radius:
R =
m
_
2qV
m
qB
Lets simplify that a bit.
Instead of dividing by q and B outside, we can move q
2
and B
2
inside; same with m in front:
R =

2m
2
qV
mq
2
B
2
After cancelling stu out, were left with:
R =

2mV
qB
2
Of course, since we got this result by using the classical equation for the kinetic energy, this will only
be valid when the particles velocity is much lower than the speed of light. When it approaches perhaps
10% of the speed of light, the result is going to be a bit o. Greater speeds mean much greater inaccuracies.
Within special relativity, a common term is , the Lorentz factor. It is used, for example, to calculate
time dilation, length contraction, and kinetic energy.
It is dened as
=
1
_
1 v
2
/c
2
... where c is the speed of light in a vacuum, approximately 3 10
8
m/s. v is as usual the particles velocity.
If 1, then classical physics and special relativity will agree. The farther it is from 1, the greater the
inaccuracy of classical physics.
We can amend the formulae for the particles radius to account for relativistic eects:
R =
mv
qB
R =

( + 1)mV
qB
2
In the limit as 1, the above are identical with what we had before, as we would expect.
In special relativity, kinetic energy as related to velocity and mass is given by
K
e
= ( 1)mc
2
=
_
1
_
1 v
2
/c
2
1
_
mc
2
In order to nd the velocity for a 1 MeV electron, we set the above equal to the kinetic energy of qV , and
solve for v, not a pretty aair:
_
1
_
1 v
2
/c
2
1
_
mc
2
= qV
1
_
1 v
2
/c
2
= 1 +
qV
mc
2
_
1 v
2
/c
2
=
1
1 +
qV
mc
2
41
1 v
2
/c
2
=
1
_
1 +
qV
mc
2
_
2
v
2
/c
2
=
1
_
1 +
qV
mc
2
_
2
1
v
2
/c
2
= 1
1
_
1 +
qV
mc
2
_
2
v
2
= c
2
_
1
1
_
1 +
qV
mc
2
_
2
_
v = c

_
_
1
1
_
1 +
qV
mc
2
_
2
_
I also solved that in Mathematica, which gave me an answer with fewer levels, but equal ugliness:
v =
c

V
_
2c
2
m + qV
c
2
m + qV
... yeah. The good news, however, is that they both produce correct results.
If we stick the numbers in for a 1 MeV electron, we nd that its speed is roughly 2.8 10
8
m/s, about 93%
of the speed of light. If we do the same calculation using classical physics, we end up with a velocity of
5.93 10
8
m/s, or about twice the speed of light! Relativity is clearly a must for these speeds.
5.1.2 Isotope separation
We can use what we now know to separate isotopes of chemical elements. Because the chemical properties
of an element is based on the element itself, all isotopes are essentially identical chemically. However, we
can use other means to separate isotopes.
An an example, we will use uranium: separating uranium isotopes was a necessary step to build the atomic
bomb during World War II. (It is still a required step for a uranium-based bomb, of course.)
In nature, about 99.7% of all uranium is uranium-238 (
238
U
92
196
), meaning it has 238 protons plus neutrons
(92 protons, which makes it uranium, and 196 neutrons). The much more rare isotope of uranium-235
(
235
U
92
193
, about 0.7% in nature) is the one used for building bombs. When the isotope purity is high
enough, we call that weapons-grade uranium.
If we ionize uranium atoms, and accelerate them over a potential dierence, and then send them through
a magnetic eld, the radius of the bend depends on each uranium ions mass. Therefore, U-235, which is
lighter, will have a slighly smaller radius than the heavier U-238! How much smaller? Well, they dier in
weight by
238.05
235.04
1.28%. When accelerated over a constant potential dierence, in the formula above we
can see that the radius is proportional to

m, so we get roughly a 0.6% dierence is radius between the
two.
A device like this is known as a mass spectrometer; it is now one of many ways used to separate isotopes.
This technique, and many others used for isotope separation, also have many peaceful uses. For example,
they are used in medicine, to separate isotopes used for radiation treatment, PET scans, etc.
42
5.1.3 Particle accelerators/cyclotrons
Lets now look at how we can accelerate protons (and other charged particles) to nearly the speed of light.
One way to do this is to use a cyclotron. A cyclotron chamber is made up of two dees, shaped like
the letter D (one of which is mirrored). Together, the dees make a circle. Seen from above, we have a
circle. We add a static magnetic eld, coming upwards (out of the page, if seen from above), and inject
a moving, charged particle into the chamber.
We then introduce a potential dierence between the two dees, of say 20 kV. There will now be an electric
eld between the two dees, and as the particle moves between them, it will get accelerated, and gain 20
keV in kinetic energy. Because of the added velocity, the radius of the particle is now higher.
When the particle reaches the end of that dee, the potential dierence is reversed, such that the electric
eld is again in the same direction as the particle, and so it is again accelerated and gains 20 keV of energy,
which again increases its radius.
We repeat this process until the proton reaches the outer edge of the dees, after having gained 40 keV per
loop, many times over.
In this process, then, the magnetic eld cause the protons to travel in a circle (spiral, rather, as they are
accelerated - but not by the magnetic force), while the electric eld accelerates the particle. Both elds
are necessary, but they perform dierent functions, and only the electric force actually does physical work.
Lets try an example:
A proton is being accelerated in a cyclotron. The radius of the cyclotron is 2 m. The potential dierence
between the gaps between the Ds is 50 kV. The uniform magnetic eld has a magnitude of 1 T. What is
the maximum energy that the proton can achieve? Ignore relativistic eects.
Since we are to ignore relativistic eects, we use this equation:
R =

2mV
qB
2
... and solve it for V:
V =
B
2
qR
2
2m
If we put the values in, we get that V 1.916 10
8
volts, or about 192 MV.
By the denition of the electronvolt, the answer is then that 192 MeV is the maximum energy the particle
can achieve in this cyclotron.
Approximately how many rotations of the proton are needed to achieve this maximum energy?
The potential dierence is 50 kV, and the proton gains energy twice per rotation, so
192 MV
2 50 kV
1920 rotations
Next question...
How much time does it take a proton with kinetic energy of 5 MeV to go around the cyclotron once? For
simplicity, you may ignore relativistic eects.
As usual, we solve
1
2
mv
2
= K
e
for v to nd the velocity. The kinetic energy is given as 5 MeV, which is
5 10
6
times the elementary charge 1.602 10
19
:
v =
_
2K
e
m
=
_
2 5 10
6
1.602 10
19
1.672 10
31
3.095 10
7
m/s
We then use our old radius equation:
43
R =
mv
qB
=
1.674 10
27
3.095 10
7
1.602 10
19
1
0.32341 m
And nally, to calculate the time it takes, we divide the circumference of the full circle with the velocity:
T =
2R
v
=
2 0.32341
3.095 10
7
6.566 10
8
s
... or about 66 nanoseconds.
How much extra time will it take a proton with energy of 5 MeV to go around the cyclotron once compared
to a proton with energy of 10 MeV? For simplicity, you may ignore relativistic eects.
If we do the above calculation again but for 10 MeV, we nd the exact same result. The reason can be
shown with a bit of algebra:
T =
2R
v
=
2
mv
qB
v
=
2m
qB
The velocity cancels. Physically, this is clearly because as the velocity of the particle increases (and as
such the time per loop would go down), the radius goes up by the same factor, so the velocity terms cancel
each other out. This only holds true as long as we ignore relativistic eects, however. With relativity
taken into account, the time taken is
T =
2m
qB

Keep in mind, however, that we need to switch the direction of the electric elds, every time the particle
moves around half the circle! Therefore, the switching needs to be done roughly every 33 nanoseconds in
this case, or at 30 MHz, since
f =
1
T
=
1
33 10
9
30.3 10
6
Hz
Also, since the time per loop depends on , at relativistic speeds we need to adjust the switching frequency
as the particle is gaining speed. Such a device is called a synchrocyclotron, so named as it synchronizes
the switching.
Modern particle accelerators are rings, i.e they have a constant radius. Since the particles need to take
essentially the same narrow path both at low and at high energies, the magnetic eld is gradually increased
as the particles are accelerated (by the electric elds).
Modern accelerators, as used for cutting-edge physics research (as opposed to e.g. medical use, which use
relatively small accelerators) are emph extremely big. The Large Hadron Collider in Geneva, Switzerland
is so large that it indeed crosses the border to France: its circumference is 27 kilometers! The LHC accel-
erates protons to energies as high as 7 TeV - 7 10
12
eV. While that is just under 1 J of energy, which
sounds tiny, keep the protons incredibly tiny size in mind. The energy is enough to make them move at
99.9999991% the speed of light, or about 3 m/s slower than the speed of light!
The LHC accelerates bunches of protons, with one set going clockwise and the other counter-clockwise;
they are then made to collide with each other - thus the name Large Hadron Collider. (A hadron is a
particle made up of quarks, such as protons.)
Via E = mc
2
, and thus m =
E
c
2
, the extreme kinetic energies of these particles is converted to other
particles, some of which are of greater (rest) mass than the protons that supplied the energy. Because of
this equivalency, particle physicists often use eV/c
2
as their preferred unit of mass, rather than the unwieldy
kilogram. In such units, an electron has a rest mass of 511
keV
c
2
. That is, if an electron is converted to
44
pure energy, the amount of energy would be 511 keV, or about 8.19 10
14
J. Protons have a rest mass of
938
MeV
c
2
, or about 1.5 10
10
J in terms of energy equivalence.
5.1.4 Cloud chambers and bubble chambers
Now that we have high energy particles, how can we see the results of their movement and their collisions?
There are many techniques these days, most of which are electronic in nature, but there are very simple
solutions to this problem, as well. Cloud chambers were the rst such solution; cloud chambers are in
fact responsible for the discovery of the position (anti-electron), muon and kaon particles, in the 1930-1940s.
We can make the tracks of charged particles visible using a cloud chamber. When a charged particle
moves through air, it creates ions in its path, until it nally loses all of its kinetic energy and comes to a
halt.
A 10 MeV electron can travel about 40 meters in air; a 10 MeV proton just one meter. A 10 MeV al-
pha particle would only travel about 10 cm. (An alpha particle is a helium-4 nuclei: 2 protons, 2 neutrons.)
A cloud chamber consists of a layer of alcohol, cooled from the underside by dry ice (solid carbon dioxide).
This causes a temperature gradient through the alcohol, and there will be a layer where the alcohol is su-
percooled/undercooled: that is, the alcohol vapor is cold enough that it ought to be liquid, but something
is needed to trigger the phase change.
The same behavior can be seen with water, going from liquid to solid: if you cool water in an extremely
smooth container, you may be able to get the water to a sub-zero temperature without turning into ice.
All thats needed for it to turn solid almost instantly (a second or so for a bottle of water) is a trigger.
Digressing further, water can also be super-heated, e.g. in a microwave, also in a smooth container: it
can then turn from a non-boiling liquid to boiling/vapor when you insert a spoon or such, and explode in
your face, causing severe burns. This can mostly be prevented by inserting a non-metal object (a wooden
spoon or such) into the liquid before microwaving.
Back to the cloud chamber. Due to this supercooled state, all thats required for the vapor to condense is
a trigger. A charged particle which creates an ion trail is enough, and so ring charged particles into this
vapor will leave a trail of condensed alcohol vapor where the particle has travelled.
When the particles move though this vapor, they will gradually lose speed. If we have a constant magnetic
eld throughout the cloud chamber, particles will move in a spiral shape: as they lose energy, the radius
of their circle due to the magnetic force will become smaller and smaller constantly, and so that makes
a spiral where they end up in the middle of the spiral.
As mentioned previously, the position was discovered using a cloud chamber. Carl D. Anderson discovered
in it 1932, by observing a particle that had an electrons mass, and the correct magnitude of charge -
but the wrong sign. The particle curved in the opposite way an electron would, given the magnetic eld
conguration he had.
Bubble chambers are similar to cloud chambers, but in a way they are the opposite. In a bubble chamber,
there is liquid hydrogen, which is hot enough that it should really be in the form of gas. The charged
particles then act as seeds for the gas bubbles, such that the particles leave a trail of bubbles where they
have moved.
Over 30 new particles were discovered between 1958 and 1968, thanks to particle accelerators, cloud
chambers and bubble chambers, in the growing eld of nuclear physics.
45
5.2 Biot-Savart Law
Lets now go back to magnetism, and more specically, magnetic elds due to moving charges (current
in a wire). If we have a current going in an upwards direction though an innite wire, we know (from
experiments) that the magnetic eld is shaped like concentric circles around the wire, with the direction
given by the right-hand rule: curl your ngers, and then point your thumb in the direction of the current.
The four remaining ngers are now curled in the direction of the magnetic eld.
Also given by experiment, is that for a current I going through a wire, the magnetic eld B is proportional
to the current, and inversely proportional to the distance from the wire:
B
I
R
Weve seen before that for an innite wire of uniform electric charge, the electric eld falls o as
1
r
. The
direction of the electric eld diers from the magnetic eld created by a current, but that is irrelevant for
making this point.
The electric eld of electric monopoles, i.e individual charges, falls of as
1
r
2
, so when we integrate them
over a wire, the result is the
1
r
fallo.
This suggests, then, that if we had a set of magnetic monopoles on the wire, that the eld due to each
one would also fall o as
1
r
2
. However, magnetic monopoles dont appear to exist - one has never been
experimentally discovered.
However, it also suggests that if we were to carve this wire, where the total magnetic eld falls o as
1
r
, into
tiny segments

d, that each element would contribute a small amount that is indeed inversely proportional
to the square of the distance.
The contribution

dB due to a such a small segment

d is, where r is the unit vector in the direction of the
point where we measure, is given by

dB =
CI
r
2
(

d r)
... where C is a constant, which we write as
C =

0
4
... where
0
is called the magnetic constant, previously known as the permeability of free space.
Weve seen before that in Coulombs law, electric elds are proportional to some constant as well:
Coulombs constant, given by
1
4
0
, where
0
is called the permittivity of free space (or, more recently, the
electric constant).
The small element

dB can also be written in terms of the displacement vector r instead of the unit vector
r. In that case, because r =
r
[r[
, we multiply both top and bottom by the magnitude, and get

dB =
CI
r
3
(

d r)
Integrating over the entire wire, and using C =

0
4
, the total magnetic eld is given by

B =

0
4
_
I

d r
r
3
46
Using this equation, we can calculate the magnetic eld exactly in the center of a circular current loop.
For simplicity, we center the loop in the coordinate system.
A current of 100 A runs through a circular loop of radius 0.1 m. What is the magnitude of the magnetic
eld right at the center of the circle?
Lets look at the integral.

d and r will always be exactly perpendicular for a circle, since the point were
interested in is in the middle of the circle. The magnitude of the cross product is
[A B[ = [A[[B[sin
... where weve just established that will always be 90 degrees, so we can get rid of the vectors, and get
a regular single integral.
[

B[ =
_

0
4
I
r
2
(

d r) =

0
4
I
_
1
r
2
d
A small element d along the circle is given by rd (via arc length), so
[

B[ =

0
4
I
_
1
r
2
d =

0
4
I
_
2
0
1
r
d =

0
4
I
2
r
Simplied, and with values, we get
[

B[ =

0
I
2r
=
10
7
4 100
2 0.1
6.28319 10
4
T
Alternatively, we could make this even easier by realizing that the integral does nothing except nd the
circumference of a circle given its radius, so we can jump ahead:
[

B[ =
_

0
4
I
r
2
(

d r) =

0
4
I
_
1
r
2
d =

0
4r
2
I2r =

0
I
2r
Power transfer and power loss
Say we have a power line - a very, very long cable - with ends A and B, A being to the left. We dene
the potential at A as V
A
, and the potential at B as V
B
; there is then also a return current wire, where we
dene V = 0 at the leftmost side. (We wont calculate losses in the return cable.)
On the right side, at B, we hook up things we want to power o the electricity grid, such as computers,
fridges, and whatnot.
The cable will have a nonzero resistance, and so there will be a voltage drop over it, given by V
A
V
B
= IR,
via Ohms law. The potential at B will then be
V
B
= V
A
IR
The power extracted at B is the voltage at that point (with respect to ground) times the current, so
P = V
B
I. This power is, via conservation of energy you could say, the power provided by the power
station, minus the resistive losses on the way there:
V
B
I = V
A
I I
2
R
Since the I
2
R loss means energy is wasted at heat in the cable, which is clearly a waste of energy (and
money), we want to minimize that term. What can we do?
Well, we could reduce R, by making the cable thick - which is expensive. Or we can make it out of
materials that are more conductive than the copper that is most often used - even more expensive. Or we
could use superconducting cable, which requires cooling by liquid nitrogen or even colder temperatures -
47
more expensive yet again.
Lets calculate an example:
Calculate the power loss in a transmission line in an aluminum cable (resistivity= = 2.8 10
8
m) of
length L = 200 km and cross sectional area A = 10cm
2
when V
B
is 100 V and when V
B
= 10
5
V. Assume
we are consuming 1 MW.
The wires resistance is given by
R =

A
where is the length in meters and A the cross-sectional area in square meters, here 10cm
2
= 0.001m
2
.
Thus R 5.6.
We are consuming 1 MW, and V
B
= 100, so I =
10
6
100
= 10
4
amperes. I
2
R is then 5.6 10
8
watts, or 560
times greater than the power we are consuming!!
If V
B
= 10
5
V instead, I = 10 amperes, so I
2
R is 5.6 10
2
= 560 watts, which is just 0.056% of the power
we are consuming.
Note that the power delivered V
B
I is exactly the same; we have only changes the current-to-voltage ratio
drastically.
Clearly, what we want to do is to make the potential as high as possible, to minimize resistive losses. What
limits us? Well, for one, corona discharge. If the potential at the surface of our cables in higher than 3
MV/m, there will be corona discharge to the air surrounding the cables, which will be a big power drain.
5.3 Amperes law
For a current I going into the page, the magnetic eld B formed will be everywhere tangential to
concentric circles around the wire. The magnitude is given by
B =

0
I
2r
Note that while this looks similar, except for the , to the result we got in a Biot-Savart law calculation
earlier, they were dierent calculations: that one was for the magnetic eld inside a circular current loop,
while this is the magnetic eld for a straight wire.
This result is also found by integrating over the closed circle:
_

B

d
Note that this

d is NOT the same as the one for the Biot-Savart law!!!
In that case,

d refers to an innitesimal wire segment, of the current-carrying wire.
In this case, for Amperes law, is a tiny movement along a magnetic eld line surrounding the wire.
Because

B and

d are always parallel, the result of the integral is simply B (which we see as a constant)
times the distance around the circle, i.e. the circumference, so
_

B

d = B2r
From the rst equation, we can see that B2r =
0
I, so the radius doesnt matter. Therefore,
_

B

d =
0
I
enclosed
48
which is known as Amperes law. (This version of the law is incomplete, and will be amended twice later
in the course; once for a displacement current term, and once to take care of magnetic permeability, as

0
only holds for the classical vacuum.)
This law not only holds irrespective of the radius, but also irrespective of the shape. Any closed shape
you can draw around a current will result in the law being true.
With Amperes law, we can use the right-hand rule yet again. If we go around a surfaces edge clockwise,
and a current is owing into the surface, we consider that current to be positive, while a current that ows
out of the surface would be negative.
Lets do an example. We have a wire with radius R, with a current I coming towards us. We assume that
the current density is uniform throughout the wire.
We choose a Amperean circle around the wire, rst with r > R so that in encloses the entire wire.
We know that
_

B

d =
0
I
Because of the symmetry of the problem, the dot product will have the same value for all

d, and the two
vectors are also always parallel, so we integrate

d and get 2r instead. We then have
B2r =
0
I
B =

0
I
2r
... which is the same result we saw earlier with Biot-Savarts law.
As for direction, we use the right-hand rule: the right thumb points along with the current, towards us,
and the remaining ngers would then curl in a counter-clockwise direction.
Lets do the same calculation for r < R, so inside the wire. The same symmetry arguments still hold; the
dierence here is in the enclosed current, which is now dependent on little r as well. Since the current
density is uniform, the ratio of the areas will give us the answer:
I
enclosed
= I
r
2
R
2
= I
r
2
R
2
Thus, our equation is
B2r =
0
I
r
2
R
2
B =

0
Ir
R
2
2
The direction has of course not changed. Also, note that as r = R, i.e. on the surface of the conductor,
the two equations give the same result.
5.3.1 The magnetic eld of a solenoid
A solenoid is a long conductor, wound up into a helix. That is, it looks like a cylinder, where the cylinders
side are made up by loops of this wire.
We can use Amperes law to calculate the magnetic eld inside a solenoid - that is, in the center of this
cylinder-ish shape.
49
When we have many loops, the magnetic eld outside the solenoid will be almost zero, while the magnetic
eld inside will be almost constant throughout the solenoid.
The shape of the magnetic eld is innitely much better described in video than in text. Even images
dont quite suce, so Ill refer to the lecture video instead. (8.02 lecture 15.)
To calculate the magnetic eld strength inside, we choose an Amperean rectangle as shown. It is (of
course) made up of four dierent lines. For line 2, the magnetic eld is approximately zero, since it is
outside the solenoid, so we discard that part.
Lines 1 and 3 are perpendicular to the magnetic eld, so

B

d will be zero for them.
Therefore, only line 4 makes a useful contribution, and since weve assumed that the magnetic eld will
be constant inside, the integral is simply B.
We then need to calculate the current that penetrates the surface dened by our rectangle, to use on the
right side of the Amperes law equation.
If we dene N as the number of loops in the solenoid, over its length L, then the number of loops through
the surface is
N
L
and so the current is I times that.
We can now set up our Amperes law equation:
B =
0
I =
0
N
L
I
The little s cancel, and we get
B =

0
IN
L
This approximation comes close to the true value, as long as L R, where R is the radius of the loops of
the solenoid.
With an exaggerated example, we can show intuitively that the magnetic eld strength is propertional not
to the number of loops total, but the number of loops per unit length.
Imagine an extremely long solenoid - 100+ meters long. Each loop has a magnetic eld that is approxi-
mately like a dipole eld - which as we know fall o quite rapidly at a distance. Therefore, the magnetic
eld near one end of the solenoid has a near-zero contribution from loops near the other side, and so only
relatively nearby loops matter.
If, on the other hand, we have hundreds of loops within an extremely small length, then the total magnetic
eld will be almost the same as if we had one loop with N times the current through it, and so with many
loops per unit length, the magnetic eld is very strong.
50
Chapter 6
Week 6
6.1 Electromagnetic induction
We have previously looked at how moving electric charges (currents) create magnetic elds. Now, well
look at how magnetic eld can create electric elds, and thus currents in conductors.
Unlike the case for electric magnetic elds, where a steady current creates a steady magnetic eld (via
Biot-Savart and Amperes laws), a steady magnetic eld does not produce a steady current. In fact, a
static magnetic eld creates no current at all; only a changing magnetic eld does.
The phenomenon where a changing magnetic eld causes - or induces - a current, is known as electromag-
netic induction. It was discovered by Michael Faraday, who did experiments to nd out whether, indeed,
constant magnetic elds cause constant currents.
One experiment he did involved a battery, a solenoid, a switch, and a separate loop with a current meter
in series.
He Usedt he rst three components to create a roughly constant magnetic eld, inside the solenoid. He
then wound a loop of wire around the solenoid, and connected that to a current meter. If the hypothesis
was correct, he would have seen a steady current in the current meter whenever the solenoid was powered.
That was not what happened; indeed, there was no current at all.
However, he later noticed a current spike in the current meter while he was ipping the switch to the
solenoid, whether he powered it on or o, the direction of the current being dependent on which. He
concluded that a changing magnetic eld induces a current.
This principle is extremely important for the modern world, as it is the basis of most of our current
power-generation: wind power, hydropower and nuclear power all depend on Faradays law, as they all
make turbines spin inside magnetic elds to produce power.
6.1.1 Lenzs law
Say we have a simple loop, i.e. a wire bent into a rectangle, a circle or such a shape. We move a bar magnet
downwards into the loop, with the magnets north pole rst. The magnetic eld of the bar magnetic will
point downwards, into the loop, and will be increasing as we move in closer.
There will be a induced current in the loop, such that the currents magnetic eld opposes the magnetic
eld of the bar magnet. Via the right-hand rule, if the currents magnetic eld is eld pointing upwards,
the current is owing counterclockwise as seen from above.
If we pull the bar magnet upwards, the external (the bar magnets) B eld will be going down in the loop,
and so the current will be in the opposite direction.
The fact that the currents eld will always oppose the external eld change is known as Lenzs law. It
states nothing more than the currents direction - to calculate the currents magnitude, we must use Fara-
days law, soon to be introduced.
51
Clearly, a current cannot just come about on its own - a current is usually driven by a source, that produces
a potential dierence over e.g. a wire. In this case, it is an induced EMF due to the changing magnetic
eld, which can be written as
E
ind
= I
ind
R
... which is of course just Ohms law. R is this case, then, is the total resistance of the entire loop where
the current ows.
In another experiment by Faraday, he had a current loop connected to a battery, that created a magnetic
eld. He switched the current, such that the magnetic eld was changing (since, as we now know, a static
magnetic eld produces no EMF).
He then had a second loop, with no battery, independent from the rst, but located nearby, such that the
changing magnetic eld aects the second loop.
What he found was that the induced EMF in loop number two was proportional to the change in the
magnetic eld from the rst,
E
2

dB
1
dt
... and also that it was proportional to the area of the second loop. This gave him the idea that perhaps
the EMF is proportional to the change in magnetic ux through the surface of the loop.
6.1.2 Magnetic ux
Magnetic ux is dened in the same way as electric ux, except of course that the dot product is between
the magnetic eld and an area vector.
That is, for an open surface A, and a magnetic eld B, we integrate the dot product of all innitesimal
area vectors

dA on the surface with the local magnetic eld at that point. This gives us the magnetic ux

B
(or
B
- the Greek letters being lowercase and uppercase phi, respectively):

B
=
_

B

dA
This integral is done over the open surface where we want to know the ux.
The SI unit of magnetic ux is the weber (Wb); 1 Wb = 1 T m
2
.
When the B-eld is constant and through a planar surface, this simplies down to

B
= BAcos
... where is the angle between the magnetic eld and the unit normal perpendicular to the surface.
For electric ux, we have Gausss law, which relates the ux through a closed surface to the amount of
charge inside:

E
=
_

E

dA =
Q
encl

0
If we enclose a single electric charge (an electric monopole), whether positive or negative, the ux is always
nonzero.
If we enclose an electric dipole, such that we enclose the entire dipole, the ux will be zero. However, if
we choose the closed surface such that in encloses only part of the dipole, the ux will again be nonzero.
The electric eld lines always begin at the positive charge(s) and end on the negative charge(s).
52
In contrast, in magnetism, we have (as stated previously) never witnessed a magnetic monopole. Magnetic
eld lines are continuous, and do not end at any of the poles. Therefore, wherever in space you choose a
closed surface, exactly the same ux that enters the surface will leave it, with no known exception. There
can be no exceptions without magnetic monopoles.
Thus, unlike in the electric dipole situation, the magnetic ux through a closed surface is ALWAYS zero.
This result is usually called Gausss law for magnetism:
_

B

dA = 0
Until the day where magnetic monopoles exist, that will remain one of the laws of physics, and is one of
Maxwells four equations. (They consist of Gausss two laws, Amperes law, and Faradays law.)
We will now introduce the last of the four equations: Faradays law of induction.
6.1.3 Faradays law
Faradays law, or Faradays law of induction, relates this changing magnetic ux to the induced EMF:
E =
d
B
dt
The minus sign signies Lenzs law, that the induced current (and thus the induced EMF) will always
oppose the change in the magnetic ux. However, if we know Lenzs law, we should never need to feel
confused by the direction of the EMF.
Since we know the ux to be given by the above surface integral, we can also state the EMF as
E =
d
B
dt
=
d
dt
_

B

dA
Since a current ows in the wire, there must be an electric eld inside the wire as well. We know, then,
that the EMF must also equal the dot product of the electric eld and the innitesimal

d integrated over
the closed loop:
E =
d
B
dt
=
d
dt
_

B

dA =
_

E

d
There is a convention regarding the direction of the area vector

dA that we choose. The open surface
attached to the loop is one chosen by us, so we follow the convention of the right-hand rule here as well.
The surface chosen does not have to be at in the plane of the conducting loop; it can be any shape, for
example like an open bag. The result will be the same. This can be intuitively explained if we consider
the ux as a ow of air or water through the loop. If it comes into the opening of the loop, it must come
out through that surface, no matter its shape or size.
If we have a circular loop (for simplicity - it does not have to be any shape whatsoever, as long as it
connects back so a current can ow) in the plane of this page, and we march clockwise around it, then

dA
will be into the page. If we march counterclockwise,

dA will point out of the page.
At this point in the lecture, the professor does an experiment, similar to the one Faraday did. He wraps
a wire, connected to nothing but an ammeter, one time around a solenoid.
If we then visualize that we attach a surface to the loop around the solenoid, the changing magnetic eld
inside the solenoid will penetrate that surface, so there is magnetic ux through our surface. Furthermore,
because we are turning the solenoid o and on, the ux is changing, and so there is an induced EMF (and
thus current) in the loop weve wound around it.
53
If we increase the size of this loop, by increasing the length of wire, nothing changes. Why not? Well, the
area is larger, but the area where the magnetic ux penetrates the surface is the same: the magnetic eld
outside the solenoid is roughly zero, as we saw last week.
Therefore, if we imagine an open surface formed between the edges of the wire, the part that is inside
the solenoid is exactly the same and is dictated by the size of the solenoid only. The parts that are out-
side the solenoid are in the region where B 0 and so they dont matter;

B

dA will be zero in that region.
However, there is one way we can increase the area that the ux penetrates: wrap the wire around the
solenoid multiple times. While its very, very hard to intuitively visualize such a surface, the wire is essen-
tially shaped like a helix, and so there will be a surface that is curled around the center point. The main
point to realize is that with three loops, the surface area of the surface will be three times greater.
Unlike before, however, the ux will now penetrate this entire surface, so the ux will be three times
graeter. Therefore the time derivative of the ux - the magnitude of the EMF (ignoring the minus sign) -
will be three times greater!
Therefore, the EMF is proportional to the number of loops N around the solenoid. We can make this N
however large we wish - a thousand loops is no problem at all. This is how transformers work: the higher
the loop count, the higher the EMF.
6.1.4 The breakdown of intuition
Now comes the hard (to accept) part of this all. Weve previously used Kirchhos voltage law (or rule),
KVL, which states that
_

E

d = 0
Or, in words, that the closed loop integral of the electric eld in a closed loop is zero. That is, when you
walk around ANY closed loop in an electric circuit - you start at one point, walk around the circuit and
add all voltage drops - the sum will always be zero.
Well, that is NOT TRUE where there are changing magnetic elds involved! The integral is not zero any
more; in fact, weve already showed above that it is equal to the (negative of the) induced EMF, which is
certainly nonzero!
Because the electric elds are now non-conservative, and Kirchos rule is only valid for conservative elds,
the path we choose now matters. Previously, the voltage sum was independent of the path.
Say we have a circuit as above. The battery has an EMF of E = 1 volt. The circuit element the right
labeled V is a voltmeter.
Via very basic circuit analysis, we see that the current through the circuit is
I =
E
R
1
+ R
2
= 10
3
A
54
The voltmeter will thus display the voltage drop across R
2
, which is IR
2
, which then is V
D
V
A
as the
points are labelled:
V
D
V
A
= IR
2
= 10
3
900 = +0.9 V
If we instead look at the left side - imagine an identical voltmeter there, also attacked to point D (positive
side) and point A (negative side). Clearly, the voltmeter will read the same voltage - they are connected
at the same points.
We can think of this voltmeter measuring the drop across R
2
as well, or we can think of it measuring the
sum of the voltage drops across R
1
and the battery. The points are still V
D
and V
A
, so:
V
D
V
A
= IR
1
+E = (10
3
100) + 1 = +0.9 V
Thus, if we subtract the two equations,
(V
D
V
A
) (V
D
V
A
) = V
D
V
D
= 0
The voltage drop between V
D
and V
D
is zero, as we expect. Kirchos law says it must be.
We now remove the battery from the circuit, and instead insert a solenoid into the middle of the current
loop.
The above is now our circuit, with the second voltmeter also visible.
The magnetic eld from the solenoid is coming out of the blackboard, with the shaded area being the area
where the magnetic ux penetrates the surface of our current loop. The rest of the surface will only be
exposed to the near-zero magnetic eld outside the solenoid.
The EMF will now be a function of time, as it is given by (the negative of) the rate of charge of the
magnetic ux.
Supposed that at one instant in time, the EMF is E = 1 volt, same as we had before.
However, last time, we had a battery! The battery had, by denition, a 1 volt drop over it. That is now
gone! We get our EMF from the solenoid, now. So lets do the circuit analysis.
What is the current? Well, the magnetic ux is coming out of the blackboard, and via Lenzs law, the
current will ow such that the currents magnetic eld opposes that, i.e. is going into the blackboard.
That means the current will be clockwise, same as before.
If we then calculate V
D
V
A
on the right, we get the same result as before:
V
D
V
A
= IR
2
= 10
3
900 = +0.9 V (right side)
55
What about the left side? Well, the battery is now gone, so the result is the same as before, minus the
plus one:
V
D
V
A
= IR
1
= (10
3
100) = 0.91 V (left side)
V
D
V
A
has two dierent values, depending on which path we choose to measure it! The two
voltmeter are connected to the same points, but show dierent values! Extremely nonintuitive.
This also means, then, that Kirchos voltage law has been broken: the sum around the loop is not zero!
V
D
V
A
= +0.9 V (right side)
V
D
V
A
= 0.1 V (left side)
(V
D
V
A
) (V
D
V
A
) = V
D
V
D
= +1 V
Note that this is all at one instant in time; the result is not because the values somehow changed during
the calculation. Its just an extremely nonintuitive result that can be very hard to accept as being correct.
6.2 Motional EMF and dynamos
We have seen now that the induced EMF is related to the change in magnetic ux through a surface.
Suppose that we have a rectangular current loop of sides x and y. We assign the area vector innitesimal

dA to it, such that



dA points upwards.
We then have a uniform magnetic eld

B pointing upwards, but at an angle towards the right.
The angle between

dA and

B is .
The ux through the surface is then given by BAcos . A = xy, so

B
= xyBcos
The induced EMF is then given by the time derivative of this expression. However, note that there are
three things which can change over time: the magnetic eld, so we have
dB
dt
; the area, so we have
dA
dt
,
and the angle between them , so we have
d
dt
. That is, if is changing, we are rotating the conducting
loop inside the uniform magnetic eld.
Consider rotating this about the y axis, centered on the loop. We rotate it with angular frequency ,
which is given by 2 divided by the period T:
=
2
T
= 2f
... which is, as above, equal to 2 times the rotational frequency in hertz.
(The unit of is given as radians per second, rad/s; however radians are dimensionless, and so it is really
equivalent to hertz, s
1
; to not confuse values of angular frequency with frequency, we consider the latter
to be in radians/second.)
The angle will then become
0
+t, where
0
is the angle at t = 0. We can choose that to be 0, so that
= t. Thus, we have

B
= ABcos t
The EMF is given by the negative of the derivative of that, so
d
B
dt
= AB(sin t)
56
The EMF is the negative of the above, so the minus sign from the cosine derivative cancels:
E(t) = AB sin t
Keep in mind that A is the total surface area the ux penetrates! If you have a loop of N > 1 windings
in a shape of area A, the EMF is given by
E(t) = NA
one
B sin t
... with A
one
being the area of one winding.
The current in the loop will then also be time dependent, and it will alternate in the sinusoidal fashion
seen above; this makes it alternating current, the same stu that comes out of your wall.
The reason AC is used there will be touched upon later in the course; one big reason is that transformers
can be used to convert AC voltages up to the high voltages in power lines, to avoid losses, and then back
down to the approximately 100 to 250 volts used in homes around the world.
Another is that power generated by generators/turbines/dynamos is sinusoidal by its nature, see below.
Lets talk about about generators, turbines or dynamos - whatever you prefer. If we have permanent
magnet, and we rotate conducting windings inside that magnetic eld, we get an EMF as seen above. This
is the process used for most of our power generation - wind power, hydropower and indeed nuclear power
all drive turbines to generate electricity.
The stronger the magnetic eld, the higher the EMF. The more windings, the higher the EMF. The larger
the area, the higher the EMF. And lastly, the faster you rotate it, the higher the EMF.
The last point is useless for the power grid, however: the power grid has a xed frequency, usually of 50
Hz (in Europe, Africa, Australia and most of the world) or 60 Hz (in the US and Americas, though not
all of it).
If we were to increase , then, not only would the EMF increase, but the power would be out of sync with
the rest of the power grid, which would cause big problems.
One of the relatively smaller such problems is that many devices - some clocks, for example - use the line
frequency to keep track of time. Thus, a clock designed for 50 Hz may go 20% too fast if the line frequency
in 60 Hz. (Or it may not, depending on how it keeps time.)
Because power is given by P = V I =
V
2
R
= I
2
R, and both V and I alone increase linearly with , the
power increases (or decreases) with
2
as we change it.
6.2.1 Changing the area
We have now looked at changing the magnetic eld and changing . What about changing the area of the
loop, such that the ux through it changes that way?
Imagine we have a rectangular loop of sides x and . The right side, of length , is a crossbar; we can move
it towards the left and right, such that x changes with time. We then have a velocity vector v towards the
right (or left, when it is negative). Say we have a uniform magnetic eld

B going upwards.
The ux through the surface is then

B
= xB
We then take the time derivative of this, where x is the variable that changes with time, while and B
are constant:
57
d
B
dt
= B
dx
dt
However,
dx
dt
is simply the speed (or velocity, if we attach a direction of + x to it) of the crossbar. The
EMF is then, via Faradays law:
[E[ = B[v[
The resistance of the conducting loop is R. What is the power dissipated by the circuit?
Power is P =
V
2
R
, where V = E for this case. We found E above, so the answer is
E
2
R
=

2
B
2
v
2
R
What is the magnitude of the force experienced by the moving part of the loop?
We found before that the force on a current-carrying wire can be found via

F
B
=
_
wire
I(

d

B)
Since

d and

B are exactly perpendicular in this problem, the integral simplies down to

F
B
= IB
... where is the length of the wire segment.
I =
E
R
=
vB
R
We substitute that into the force equation and get another and B, so the answer is

F
B
=

2
vB
2
R
The direction of this force is to the left:

d

B gives the direction, where

d is an innitesimal wire
segment the current moves through. That vector is upwards (+ y, in our 2D plane), and the B-eld is
perpendicular to that, upwards in 3D (+ z).
How much power is needed to keep the bar moving at constant velocity?
Power is work per unit time, so the time derivative of work should give us the answer. We know the
magetic force, which is to the left. So we need to exert a force towards the right, of equal magnitude. The
work we do is then that force (that we have above) times the distance. We then need to take the time
derivative of that.
However, we already have the time derivative of the displacement - the velocity. So
P =

F
B
v
Since they are in exactly opposite directions, cos = 1, so the answer is the answer for the force above,
times another v:
P =

2
v
2
B
2
R
58
... which happens to exactly the same as the power dissipated by the circuit! Of course, unless we nd
somewhere else the power could be used, this is to be expected; energy cant be created nor destroyed.
The power we put in must go somewhere.
We can also write this in terms of I instead:
P = IBv
This must be equal to the EMF times the current; the current then cancels from the equation, and we get
[E[ = Bv
... which is what we found earlier, however this time we did not use Faradays law to nd it, merely the
work per unit time we do to move the crossbar against the magnetic force.
Also, note that if instead of pulling towards the right, we push the crossbar towards the left. We still do
exactly the same amount of work - positive work in both cases. When v is ipped over, the magnetic force
is also ipped over, because the current reverses direction as the magnetic ux is now decreasing.
6.2.2 Eddy currents
Lets now move our focus to something dierent, but related.
Suppose we move a solid, conductive disk through a magnetic eld. Say we have magnetic eld pointing
upwards, and we move the disk through it sideways.
We know that the changing magnetic ux through the disk will induce a current to oppose the magnetic
ux change; that is, the currents magnetic eld will be downwards, and so via the right-hand rule, the
current will be clockwise.
We call these currents eddy currents.
They produce heat in the conductor; the energy for that must come from somewhere. In this case, energy
that would otherwise be kinetic energy is used up, and so the disk slows down. This is the principle called
magnetic braking.
Eddy current calculation
A fairly hefty question was in between the lecture videos, so Ill write down how I solved it. If it werent
for the multiple-choice answers, this would make a good homework question!
A rectangular loop of wire with mass m, width w, vertical length , and resistance R falls out of a magnetic
eld under the inuence of gravity. The magnetic eld is uniform and out of the paper (

B = B x) within
the area shown (see sketch) and zero outside of that area. At the time shown in the sketch, the loop is
exiting the magnetic eld at velocity v(t) = v(t)z, where v(t)<0 (meaning the loop is moving downward,
not upward). Suppose at time t the distance from the top of the loop to the point where the magnetic
eld goes to zero is z(t) (see sketch).
59
What is the direction of the induced current in the loop?
Right-hand rule, as always. The ux is positive through the loop, but it is decreasing, so the derivative is
negative. Via Lenzs law, if the ux is decreasing, the current will create an opposing magnetic eld. If
the eld lines are out of the page but decreasing, the currents magnetic eld will be out of the page as
well, so the current is counterclockwise.
What is the direction of the I

d

B force in the top horizontal segment of the loop?
(Note that there is no net force on the other segments, since they are outside the B-eld, except for parts
of the left/right sides, which cancel.)
Sincet he current is counterclockwise, i.e. towards the left at the top, the cross product direction is left
cross out of the paper, which is upwards. That is, the magnetic force will be breaking the fall.
What is the EMF generated in the loop?
We use Faradays law. First, what is the ux, and the ux change? The ux is

B
= wz(t)B
where wz(t) is the area, changing with time. The time derivative of the above is then
d
B
dt
= w
dz
dt
B = wvB
The EMF is the negative of that, but because v is also negative, the signs cancel and we get the answer
E = wvB
Suppose the bar reaches terminal velocity (no longer accelerating). What will its downward speed be
then?
Not being well-versed in classical mechanics, I made a bit of a guess here: that when the forces are equal,
the above is true. That gave me the correct answer, at least.
So, I calculated the force due to gravity:
F
g
= mg (since F = ma)
... and the magnetic force, which is in the opposite direction:
F
B
= IwB =
_
E
R
_
wB =
wvB
R
wB =
w
2
vB
2
R
60
Set the two equal, and solve for v:
w
2
vB
2
R
= mg
w
2
vB
2
= mgR
v =
mgR
w
2
B
2
6.3 Displacement currents and synchronous motors
Lets take another look at Amperes law. Say we have a current owing towards the right, through
a capacitor (with circular plates of radius R, apparently). We know that the electric eld inside the
capacitor is
E =

free

0
=
Q
free
R
2

0
The current is, per denition
I =
dQ
free
dt
We therefore have a changing electric eld between the plates, while they are being charged, given by
the time derivative of the electric eld above. Since Q
free
is the only thing changing with time, the time
derivative is simply given by substituting the current for Q
free
:
dE
dt
=
I
R
2

0
If we now want to calculate the magnetic eld at a distance r away from the wire, at point P1, what can
we do? We can try Amperes law, but keep in mind that there is an interruption in the current due to
the capacitor plates (no current ows in the air/dielectric between them), and Amperes law is only truly
valid for innite wires (and other special cases). Lets try anyway, just to see what happens.
Amperes law, as we know it so far, states that
_

B

d =
0
I
pen
We choose an Amperean circle of radius r such that point P1 is part of the circle. We then attach an
open surface to the circle, such that the current penetrates that surface (we choose the circles area, so to
speak, as that is the simplest possible choice).
The closed line integral around a circle is simply 2r. As for I
pen
, that is the current that penetrates the
surface, which in this case is all of it. So we nd that
61
2rB =
0
I
pen
B =

0
I
pen
2r
Fair enough. What about at point P2, above the empty space in the capacitor?
Well, we use the same formula, choose an Amperean circle with radius r as before, attach the same open
surface to it... but the current penetrating the surface is now zero, so the result we get is that the magnetic
eld is zero at point P2!
The choice of the open surface is up to us, so we change our choice of surface to a bag that encloses one
capacitor plate:
We re-apply Amperes law, to nd the magnetic eld at P1 - which we found to be B =

0
I
pen
2r
previously.
However, as we reach the point where we nd the current that penetrates the surface, we now see that it
is zero! No current penetrates our surface, it just goes inside it - never through.
We had the free choice of the surface, but dierent choices gave us dierent results!
The reason behind this is that the version of Amperes law weve learned so far is incomplete. We will
now add a second term that corrects this problem.
6.3.1 The amended Amperes law
Maxwell solved this problem. There is a changing electric eld between the platse, and so there is a
changing magnetic ux through our chosen surface.
He reasoned that since, via Faradays law, changing magnetic ux induces elecrtic elds, perhaps changing
electric ux induces magnetic elds, as well.
Electric ux works just as magnetic ux (as we mentioned when we introduced magnetic ux)!, so

E
=

E

dA
The new version of Amperes law then relates the time derivative of that with the magnetic eld:
_

B

d =
0
_
I
pen
+
0

d
dt

E

dA
_
(This version of Amperes law is still incomplete - it will be adjusted in week 8 of the course; however that
adjustment is very small, and only adds one simple term multiplying
0
, to take care of the permeability
62
of other media than the classical vacuum.)
The closed loop in the rst integral is then the Amperean circle, while the surface in the second (surface)
integral is the open surface attached to that closed loop. That is, the closed loop is the opening of the
bag, while the open surface is the bags surface.
The current I
pen
is the real current that penetrates the surface. The second term,
0

d
dt

E

dA, is
called the displacement current.
Note that whenever the electric ux is constant, the time derivative goes to zero, and the old Amperes
law pops out.
Lets try to re-calculate the magnetic eld at P1, again using both the at surface and the bag. First
out is the at surface.
The left-hand side of the equation is unchanged, so 2rB, while the right-hand side turns out like this:
2rB =
0
_
I
pen
+
0

d
dt
_

0r
2
_
_
=
0
(I
pen
+ 0)
... because there is no electric ux penetrating the surface - it is outside the capacitor, where the electric
eld due to the capacitor is zero.
Thus, we nd
B =

0
I
pen
2r
Same as before.
We re-start with:
_

B

d =
0
_
I
pen
+
0

d
dt

E

dA
_
The left-hand side is unchanged, as that is the closed Amperean loop, which we are not changing. I
pen
now changes, however: it is again zero - once again, no current penetrates the surface! What about the
second term? Lets ll it in:
2rB =
0

d
dt
_
ER
2
_
The E-eld is what is changing with time, so we put that in there (we found
dE
dt
earlier):
2rB =
0

dE
dt
R
2
=
0

I
R
2

0
R
2
Note that the I here is not I
pen
, the current penetrating the surface weve chosen, but the current in the
wire. It is not zero!
If we simplify this, we get
B =

0
I
2r
Excellent! Though we found I in a very dierent way, the answer is identical!
63
Next up, lets try to calculate the B-eld inside the capacitor (while it is charging; when it is not, we have
an electrostatic situation and it ought to be zero).
We set a new point P2 at distance r from the exact center of the capacitor (in all dimensions). Since the
choice of surface should not matter, we will make it simple an pick the at surface again.
2rB =
0
_
I
pen
+
0

d
dt

E

dA
_
I
pen
is zero, since there is no current inside - only air (or dielectric)! The ux, however, is EA with the area
being r
2
(little r - the surface is smaller than the plates, so there is no contribution from the remaining
area).
2rB =
0

d
dt
_
Er
2
_
Again, we know the time derivative of the E-eld from earlier, so we substitute that in there once again:
2rB =
0

I
R
2

0
r
2
B =

0
Ir
2R
2
(for r < R)
Interesting, it is exactly zero in the middle (r = 0), then, and greatest at the edge.
6.3.2 Displacement current
Why did Maxwell call the second term the displacement current?
If you have a dielectric between the capacitor plates, the changing electric elds will cause a current;
the induced charges will be rearranged due to the changing electric elds. In a vacuum, however, there
shouldnt be a current, so the name is perhaps somewhat poorly chosen.
We have a capacitor of area A lled with a dielectric slab with dielectric constant . If Q(t) is a function
of time, what is the polarization current in the dielectric? Express your answer in terms of the magnitude
of the electric eld E(t) between the plates. Positive polarization current ows down in the picture.
The picture is then of a simple capacitor, with +Q on the top plate and Q on the bottom, with the
opposite induced charges Q
ind
on the dielectric (minus up, plus down).
Hmm, OK. E(t) is given by
E(t) =
(t)

0
=
Q(t)

0
A
My rst thought was to solve for Q and take the time derivative of both sides. That didnt work; I got
something close to the correct answer, yet far away: it had a factor instead of ( 1), where as well
see, only the latter means no current if = 1 (for a perfect vacuum).
Instead, we can solve it like this:
First, we know that the E-eld in a capacitor in a vacuum is
E
vacuum
(t) =
(t)

0
=
Q(t)

0
A
With a dielectric, there will be an induced charge Q
ind
(t) on the dielectric. This induced charge causes an
opposing electric eld, E
ind
(t), which causes E(t) to equal the original E-eld divided by :
64
E(t) =
E
vacuum
(t)

=
Q(t)

0
A
We can also write the net eld as the vectorial sum of the original eld, plus the induced eld. The latter
is in the opposite direction, so we get a subtraction:
E(t) = E
vacuum
(t) E
ind
(t) =
Q(t)

0
A

Q
ind
(t)

0
A
We can then set the two ways of writing E equal and nally solve for Q
ind
(t):
Q
ind
(t) =
1

Q(t)
Almost there. We are looking for
dQ
ind
dt
, so we need to calculate the derivative of the above. First, we
solve the E-eld equation for Q(t), so that we can substitute it in there:
E(t) =
Q(t)

0
A
Q(t) = E(t)
0
A
Make the substitution, and simplify:
Q
ind
(t) =
1

E(t)
0
A
Q
ind
(t) = ( 1)
0
AE(t)
And nally, take the time derivative:
I
polarization
=
dQ
ind
dt
= ( 1)
0
A
dE
dt
Thanks to H_Litzroth on the 8.02x forums for posting his solution, which the above is heavily based on.
6.3.3 Synchronous motors
Say we have a rotating conducting current loop in a magnetic eld, like the rst image here shows:
We then add two more loops, each 120 degrees rotated from the rst, so that we get the result shown in
the second picture, seen from the side (from the side where the wires enter in the rst picture). So we
have three loops, electrically isolated but rotating together.
As weve seen before, when they rotate, we will get an induced EMF in each loop. Due to their 120 degrees
(1/3 period) separation, we will get three sinusoidal waves, which dier only in their phase - assuming all
loops are identical in composition, area etc.
The EMF as a function of time would look something like this:
65
... with the current from loops 2 and 3 having a phase delay of 120 and 240 degrees from the rst loop,
respectively.
The period of all three loops will be equal, as they are rotating together.
Now, lets put that aside for a while, to not cause confusion. We will use the current from that 3-phase
generator, but other than that, what follows is independent of the above setup!
We now take a look at a dierent setup, of three open solenoids.
To the left is the professors drawing, which I didnt understand at rst. The second one, on the right,
is from the same angle - from straight above - and shows a snapshot at three points in time. The rst
(leftmost) frame is when the magnetic eld due to the red solenoid peaks, and the vectorial sum of the
three magnetic elds are in the direction of the arrow. (The red one alone is in that direction, and the
vector sum of the blue + yellow are also downwards.)
In the second frame, 1/6 of a period later (60 degrees phase later), the vector sum is now as the arrow
shows strongest due to the blue solenoid; again, the vector sum of the two others sum up and to strengthen
the eld in this new direcion.
Finally, another sixth of a rotation later, the magnetic eld now points as the arrow shows.
What we have, then, is a rotating magnetic eld. The solenoids themselves are perfectly stationary, but the
net B-eld they generate rotates around in a full circle, once per rotation of the original power generator.
(Once per Hz of the power we are feeding the solenoids.)
Of course, since the current is a set of smooth sinusoids, the magnetic eld changes gradually throughout
this, and doesnt just skip between the 6 positions.
If we stick a magnet in the middle of all this, we have a synchronous motor, or a 3-phase motor. The
magnet will want to align with the magnetic eld at all times, but the magnetic eld is always rotating,
and so the magnet will rotate at the same frequency - the frequency of the supplied power.
66
If we stick a conductive object inside this eld, it will want to rotate. Eddy currents are formed, and the
rotating magnetic eld exerts a torque on the currents, causing the object to rotate. The above picture
shows a rotating, conducting egg as demonstrated in the lecture, using the solenoids drawn above.
If the object is roughly spherical (or egg-shaped), the rotational frequency will be very close to the frequency
of the current (and thus the magnetic eld).
67
Chapter 7
Week 7
7.1 How do magicians levitate women?
This lecture is mostly about various concepts - the human heart, aurora borealis, and magnetic levitation.
There is little math, but much graphical content, and so it is hard to take very good notes without taking
screenshots once a minute!
7.1.1 The human heart
The human heart has four chambers: on top, the left and right atria (singular: atrium), and below that,
the left and right ventricles.
The aorta is on the left side.
The heart pumps about 4.5 liters of blood per minute. It beats about 70 times per minute (which varies
a lot between people, of course).
If the brain is without new oxygen for about 5-10 seconds, you pass out - after as little as ve missed beats.
After a few minutes, permanent brain damage is very likely. After 5 minutes, it is virtually certain, and
death is likely.
Each heart cell is a small chemical battery. In the normal state, an individual heart cell, about 10 m in
diameter, has a potential of 80 mV on the inside, with respect to the outside.
A group of pacemaker cells, located in a small area of about 1 square millimeter on the right atrium,
start the process of a heart beat by changing their potential from -80 to +20 mV. Once they do so, the
neighboring cells do the same, and a wave propagates across the entire heart, from the atrium down to
the ventricles.
When the heart cells change their potential (by an exchange of ions), they contract, and so the heart
contracts, one cell after the other, in a wave.
68
About 0.2 seconds later, the cells return to their resting potential of -80 mV, in a similar wave, this time
going from the bottom up.
They then wait from a new incoming wave from the pacemaker cell, and the process starts over about a
second later.
In the resting state, the cell has repelled positive ions, which makes the inside negatively charged. The
electric eld outside is zero, since the net charge around the cell is zero.
Here, we see a heart cell in the middle of the depolarization phase, when it changes from -80 mV to +20
millivolts (shown as 0 to simplify), by moving positive ions back inside.
The wave starts from above, and has passed through half the cell at this instant.
In the bottom half, we have a situation with negative charge on top of positive charge, which creates
a dipole-ish electric eld. When the process is completed, the inside is at +20 mV with respect to the
outside, and there are positive ions on the inside, with the corresponding negative charge, making for a
zero electric eld.
A while later, the repolarization wave comes in, from below, and the reverse process happens. The bottom
of the cell goes to the -80 mV, gradually changing into having the negative ions inside again, which will
then again create the dipole eld for a time.
Only the cells that are part of the depolarization wave at a given time - a minority, and most of the time
indeed none of them are - contribute to the dipole eld. The same is true for the repolarization wave as
well.
Since there is a net electric eld from the heart, there will be a potential dierence between dierent parts
of your body. Measuring this potential dierence is how an EKG (or ECG - electrocardiogram) works.
Many electrodes - usually 12 - are attached to various body parts. The potential dierence is only on the
order of a few (2-3) millivolts.
There is a condition known as ventricular brillation. In the United States, it kills on the order of 400 000
people every year. The basic cause is that the ventricular cells re without a message from the pacemaker
cells, and so the re randomly, and theress a non-coordinated depolarization, so that the heart stops
pumping any blood.
Clearly, then, this is a medical emergency, since as said earlier, loss of consciousness usually happens
within 10 seconds, and there is risk of permanent brain damage extremely quickly, even after a single
minute without oxygen. After a few minutes, it is almost certain.
One treatment for ventricular brillation is the well-known debrillator - a direct-current electric shock
across the heart, to reset the cells and hopefully get back synchronization from the pacemaker cells. The
shock sends a total of roughly 200-350 joules of energy into the heart, over a short period of perhaps a
tenth of a second.
The hearts pacemaker can also be faulty, which may cause various problems. One treatment for such
a condition is to implant an articial pacemaker, which can sense the hearts condition and take over if
necessary, triggering the depolarization wave.
There are also implantable debrillators, which can apply small shocks automatically when brillation
may have caused death otherwise.
69
7.1.2 Aurora borealis
If we have a magnetic eld, and move charged particle though it, lets say a positive one, we know that the
magnetic force on that particle is given by

F
B
= q(v

B)
We can do vector decomposition to get two vectors: one parallel to

B, and one perpendicular to

B. We
get

F
B
= q( v

+ v

)

B
However, note that the parallel component v

makes no dierence for the force - the angle between that


and the magnetic eld is zero (or 180 degrees), so the cross product is zero. We end up with

F
B
= q( v


B)
Consider, then, a situation with a magnetic eld towards the right, and a velocity angled upwards (still
in the same plane). The perpendicular component will make the particles go in circles, but the parallel
velocity means it will still keep going forward at the same time, so the path is a helix, a bit like a spring.
The radius of the helix will then be given by, as usual (except that only the perpendicular component
matters know)
R =
mv

qB
Meanwhile, it continues forward at the same parallel velocity, which remains unchanged.
The Earths magnetic eld is not made up by straight eld lines, however. The eld is similar to what
youd expect from a massive bar magnet buried within the Earth, with the magnets south pole near the
geographical north pole, and vice versa. (The magnetic poles drift and even shift completely with time,
over very long time periods.)
Thus, when a particle enters, it moves along a helical path around a eld line, and enters near the one of
the magnetic poles, where the eld lines terminate. Well, they dont ever terminate, but they would enter
the Earths solid part, where the particle is unlikely to get far (if it ever reaches the surface).
The sun sometimes emits massive amounts of plasma, mostly ionized gas so protons and electrons, called
the solar wind, in coronal mass ejections, or CMEs. The scale of these ejections are absolutely mind-
boggling; each arc of plasma that breaks o the sun is many, many times the size of the Earth. I feel an
image is absolutely necessary (image courtesy of NASA):
70
In a CME, such a huge lump of plasma can break o and leave the sun, moving out a a few hundred to
a few thousand kilometers per second, meaning it can reach the Earth in a few days or so. Thus, a few
days after a CME, massive amounts of charged particles can collide with the Earths magnetic eld.
When that happens, it ionizes molecules in the Earths upper atmosphere, which releases light - the color
of which depends upon which particles are ionized. We call this aurora borealis, also northern lights in the
northern hemisphere; in the southern hemisphere, it is known as aurora australialis or southern lights.
Quite a stunning display. The second image is taken by the IMAGE satellite, a satellite designed to study
the response of the Earths magnetic eld to changes in the solar wind. (The image of the Earth is added
digitally; that too is taken by a satellite, however, so that too is real.)
The third is from the International Space Station.
The colors can vary; green, white and red are the most common, but blue is another possibility. The color
depends on at what altitude the collisions happen, and also on which molecules are being excited.
7.1.3 Superconductivity and magnetic levitation
Superconductivity is a state of exactly zero electrical resistance, encountered in certain materials at very
low tempereatures. (High-temperature superconductors are under heavy research.)
The resistivity of a superconductor has a discontinuous drop to 0 below its critical temperature T
c
.
Superconductivity was discovered by Dutch phycicist Heike Kamerlingh Onnes in 1911, and gave him the
Nobel Prize in Physics 1913.
He invented a method for producing liquid helium a few years earlier; helium remains a liquid down to
absolute zero (an unreachable temperature); indeed he got it as cold as 1 kelvin!
71
Unfortunately, quantum mechanics is required to fully understand superconductivity. Even then, high-
temperature superconductors (materials that are superconductive above 30 K, previously thought to be
impossible) are not fully understood as of this writing.
High-temperature superconductors were discovered in 1986, and led to the Nobel Prize as soon as in 1987.
Because there are now materials known that are superconducting at up to around 133 K, liquid helium
or other extreme measures are no longer required, and the much more available liquid nitrogen, with a
boiling point of about 77 K (about 196 C) is cold enough.
No electric eld can exist in a superconductor. If it did, then the electric eld times a distance in the
conductor could be nonzero, meaning a potential dierence would exist. If there was a potential dierence
V , and I =
V
R
=
V
0
, clearly things dont work out.
If we approach say a superconducting disk with a bar magnet, Faradays law tells us there is a change in
magnetic ux, and thus an induced EMF. But we just said there can be no EMF in a superconductor!
This has a very interesting consequence: we have said that Faradays law always holds, so the EMF in the
superconductor must be
E =
d
B
dt
... but since E must equal zero, so must the change in the magnetic ux!
The magnet will induce eddy currents in such a way that the magnetic ux through the superconductor is
held constant!
The net magnetic eld - the superposition of the elds due to the bar magnet and due to the eddy currents
- will look like so:
There will be a magnetic pressure between them, given by
P =
B
2
2
0
N/m
2
Because of the superconductivity, the eddy currents never dissipate - I
2
R = 0 because R = 0, and so we
can simply place a regular magnet above a superconductor, and it will levitate and stay in place.
There are other ways of magnetic levitation. In a train, we can have strong magnets and move them over a
conductive surface below the train. Since there will be a change in ux penetrating the conductive surface,
eddy currents will be induced, and we know via Lenzs law that they will oppose the external magnetic
eld, thus creating a repelling force, possibly allowing the entire train to magnetically levitate above the
tracks - as long as it has a high enough speed. If the speed is low enough, the ux change through the
conductor will not be enough to produce enough force to lift the train, and it will crash (more or less) to
the ground. Clearly, such a train will need to be built to handle this either way, or it would have no way
to get moving from a standstill if it could only move while levitating!
Now then, weve seen two ways of magnetic levitation: one using superconductors, and another using
moving magnets.
72
A third form requires neither speed nor superconductors, but instead uses alternating current in a loop.
Imagine a current loop, through which we pass an alternating current, above a conducting plate.
At one moment in time, the magnetic eld due to the coil will point downwards, i.e. have its north pole
towards the conducting plate. Lenzs law tells us the eddy currents will produce a magnetic eld that
opposes this, and so the two plates will repel each other, and there will again be a magnetic pressure
between the two.
However, slightly later in time, the magnetic eld will now be decreasing, and so the eddy currents ip
direction, such that the current loops eld still has its north pole downwards (but the ux through the
plate is decreasing), but with the eddy currents eld ipped, the two will now attract each other.
It therefore seems reasonable to conclude that they will repel and attract each other equally, on average;
this is not the case, and there will be a net repelling force, the source of which is discussed in next week,
when inductance is discussed.
(A preview: there is a lag in the eddy current versus the induced EMF, such that they repel each other
more than 50% of the time, so the net force on the current loop is upwards.)
The magnetic levitation of a woman
We then reach the section responsible for the lectures name. Magicians used to (seemingly) levitate people
- stereotypically, they were women - using various tricks. Can we do it using magnetic levitation?
We have the equation for the magnetic pressure,
P =
B
2
2
0
Using a current loop of area 0.1 m
2
, we nd that using a B-eld strength of about 0.15 tesla, the pressure
is
P 9000N/m
2
Multiplying that by the area, we get a force of roughly 900 newtons, enough to lift a hair over 90 kilograms
in the Earths gravitational eld - so that should be enough to lift a person.
However, generating such a strong B-eld proves very dicult, as you need a very high current. Thus,
the demo is about levitating a blow-up doll - which tells us that magicians most likely dont use magnetic
levitation in their tricks.
7.2 Inductance and RL circuits
Self-inductance (as opposed to mutual inductance, discussed later in the course), often called inductance
for short, is a way to quantify the ability of a circuit to ght the change in magnetic ux produced by the
circuit itself.
We know that if you run a current through a circuit, a magnetic eld is created. If the current is changing
with time, then the magnetic elds are also changing with time, which causes an induced EMF in the
circuit, ghting the change.
The magnetic ux is always proportional to the current,
B
I. The constant of proportionality is then
the self-inductance, for which we use the symbol L:

B
= LI
If we combine this with Faradays law,
E =
d
B
dt
73
We nd that
E = L
dI
dt
... assuming that L is constant, i.e. it is not a function of time. That is usually a pretty good assumption,
as L is, just like capacitance C, only a function of the geometry. We will soon nd many other similarities
between capacitors and self-inductors.
Self-inductance of a solenoid
Say we run a current I through a solenoid, with N windings, each of which has radius r, and the length
of the solenoid is .
First, the B-eld inside, given by a quick-and-dirty Amperes law calculation, is
Ba =
0
I
pen
=
0
Ia
N

B =

0
NI

... where a is the length of the Gaussian rectangle we choose. (See the section on Amperes law for a
proper derivation.)
If we attach an open surface to the current loop in the solenoid - dicult to visualize, but it will have an
area where the ux penetrates of Nr
2
, with N again being the number of windings. The ux is then
that area, times the B-eld inside, so

B
= N
2
r
2

0
I

This is then equal to LI by denition, as we said earlier, so we can now calculate the self-inductance L
for a solenoid:
LI = N
2
r
2

0
I

L =
N
2
r
2

As promised, it is only a function of the geometry, plus two other constants and
0
.
The SI unit of inductance is the henry, H.
1H = 1
V s
A
All circuits have a nonzero self-inductance, just as how they have a nonzero capacitance, and (for non-
superconductors) a nonzero resistance.
We can show this easily as follows: all current ows in loops. All loops, no matter their shape, produce a
magnetic eld, which produces a magnetic ux. Therefore all circuits have a nonzero self-inductance, no
matter how small it might be.
From here on, I will mostly refer to self-inductors simply as inductors. The only potential confusion is with
mutual inductance, which will not be discussed at all in this chapter, only mentioned like now. Mutual
inductance however requires multiple coils with physical proximity, which makes it irrelevant here.
74
7.2.1 Direct-current RL circuits
Say we have a very simple series circuit: a battery, a switch, a self-inductor, and a resistor.
The switch is open to begin with, so the current at time t = 0, the current is zero.
The inductor will try to ght the change in current, and so the current will ramp up slowly. In the end,
the current in the circuit will be simply I =
V
R
, with R being the resistance of the resistor.
(We often calculate as if inductors have zero resistance; that is of course not true, since all wire will have
some resistance. However, whenever that resistance is necessary to care about, we can often model it as
an ideal inductor in series with a resistor.)
So at t = 0, I = 0, and with t , I =
V
R
. What about in between the two extremes? Well, lets try to
set up an equation that describes the circuit.
This is a critical point of this week, and perhaps of the entire course so far. Prof. Lewin states in no
uncertain terms that the way most books do this is incorrect - the end result is correct, but the thought
behind the result is misleading and/or incorrect.
Therefore I will list both the correct and the incorrect solutions here, and talk about the dierences.
This is explained in greater detail in the supplemental lecture notes for lecture 20, and is probably better
explained there, by a physics professor, rather than by a student!
So, lets see. The circuit has an inductor. That means there will be time-changing magnetic elds. That
means that Kirchos loop rule does not apply! This is crucial. Instead, we will have to do a line integral

E

d around the loop, and set it equal to
d
B
dt
, not set it equal to zero!
To clarify: Kirchhos rule is a special case of Faradays law. It only holds when the change in ux is zero,
because they state conicting results for the same integral:
_

E

d =
d
B
dt
(Faradays law, always true!)
_

E

d = 0 (Kirchhos rule, NOT always true!)
With Faradays law being the one that is always correct, clearly then Kirchhos rule is correct ONLY
when

d
B
dt
= 0
With that in mind, lets move on, and use Faradays law as necessary.
RL circuit analysis and a long note on Kirchos second rule
We start to the right of the battery. We treat the switch as ideal, so it doesnt contribute at all (we wont
even write a 0 for it in the equation, but treat it as if it doesnt exist). After that, we reach the inductor.

E

d is ZERO for the inductor, as we treat it as having no resistance (inductors are made to have as low
a resistance as possible). No resistance means no electric eld, and so the dot product must be ZERO!
So far, our integral is still at zero.
After the inductor, we reach the resistor. The electric eld inside the resistor is in the same direction as
the current, so the dot product of

E

d is positive, and equals IR; Ohms law works here. We then reach
the voltage source/battery. Here, the electric eld is in the opposite direction of the current - the E-eld
always goes from positive to negative, but the current inside a battery does not. Therefore the voltage
source contributes with a V term.
75
Finally, the sum of all these terms must be equal not to zero, but via Faradays law, to the negative of the
rate of change of the ux, that is,
d
B
dt
. All in all, the equation is then
0 + IR V =
d
B
dt
We have said earlier that
B
= LI, so if we substitute that in there, we get
0 + IR V =
d
dt
(LI)
0 + IR V = L
dI
dt
Thus, we have used Faradays law in the proper way: the left side of the line integral is the sum of the
potential dierences over the circuit: the potential dierence across the inductor is zero, because we treat
it as having 0 resistance, and thus no E-eld can exist in it.
The right side is equal to the negative of the rate of charge of the magnetic ux, as always in Faradays
law. THIS is where the inductors term is: it is due to the ux, and is not really a potential dierence
per se, as electric potential is only applicable to conservative elds - and with changing magnetic ux, we
have non-conservative elds!
We therefore have a dierential equation that we can solve, and thus nd the value of the current in the
circuit. First, however, lets take a look at the incorrect approach used by most physics books.
The approach taken by them is to modify Kirchhos loop rule, such that the potential dierence (or the
voltage drop, dierent name for the same thing) across all inductors is L
dI
dt
, if we move in the assumed
direction of the current. Thus, when we do that, and move across the loop in the same direction, and set
it all equal to zero via Kirchhos here invalid rule. We now ignore the electric elds, and think in terms
of potential: when we go up in potential, we add (so going through a voltage source from - to + means
we add, otherwise subtract; going through a resistor where we enter from the + side means we subtract).
This means we add L
dI
dt
for the inductor - remember, the minus sign is from this rule (see above), we
are adding the term. We then subtract IR for the resistor, and nally add V for the voltage source. This
is then set equal to 0:
L
dI
dt
IR + V = 0
Note that this equation is identical to the one we found before, only that it has been rearranged. If we
take our proper equation and move IR and V to the same side as L
dI
dt
, we get the above. Thus,
mathematically, both ways yield the correct answer.
So, if they give the same result, whats all the fuss about? The problem is in what the equations represent.
In the correct equation that we found, we follow Faradays law: the integral of the electric eld around
the loop -

E

d - all added up, equals the negative of the rate of change of the magnetic ux.
While the right-hand side also has the SI unit of volts, it is not a true potential dierence, as the concept
of potential does not apply to non-conservative elds.
The incorrect equation, on the other hand, is really implying that the sum of the potential dierences
around the loop (
_

E

d) is zero, which it is not! The sum of the potential dierences,
_

E

d, is equal
to
d
B
dt
and nothing else! Faradays law!
The incorrect equation therefore also implies that the potential dierence across the inductor is L
dI
dt
,
which IS what you would measure with a voltmeter, but IS NOT correct. In the ideal inductor, the E-eld
is ZERO, so

E

d for the inductor is also zero!
76
The reason a voltmeter would measure the same value is not because there is a true potential dierence,
but that when you connect a voltmeter, you get a RL-circuit with the voltmeters internal resistance
providing (most of) the resistance. Therefore, we must apply Faradays law to that circuit as well, and we
nd the same result (L
dI
dt
); however the physical cause is again a change in magnetic ux via Faradays
law, and not a true potential dierence across the inductor!
Thus, the result (the number displayed by the voltmeter) is the same, but the physical reasons are vastly
dierent!
If this is still confusing to you (especially everything prior to the voltmeter part, which seems even more
confusing), read (or re-read) the lecture supplement, and perhaps especially the last sentence:
If you nd all this confusing, you are in good company. This is one of the most dicult and subtle topics
in this course it trips up experts all the time. Not easy!
RL circuit analysis continued
So, getting back to where we were. We want to solve the dierential equation
IR V = L
dI
dt
which governs this circuit. We add the additional constraint (boundary condition) that I(0) = 0, that
is, the current at t = 0 is 0. Without this condition, we cannot get a unique solution to the dierential
equation.
Solving it, we get
I(t) =
V
R
_
1 e

Rt
L
_
Alternatively, a bit easier to read, using e
x
= exp(x):
I(t) =
V
R
_
1 exp (
Rt
L
)
_
If we take the limit as t 0, we nd that the current is indeed zero.
Also, in the limit as t the current is
I
R
, as we expected.
In between the two, the current increases exponentially, asymptotically approaching the maximum current.
During this time, the inductor sucks up energy, and stores it in the magnetic eld it creates. This energy is
later freed again when the eld collapses, whenever the inductor ghts a current change by itself providing
a current.
After one time constant, =
L
R
(the unit of henry divided by ohm is indeed seconds!), the current has
increased to 1 e
1
63.2% of its maximum value. After two time constants, the current is 86.47% of
its maximum.
If we reverse our thinking: at what time is the current x times its maximum value, for 0 < x < 1?
The current multiplier, so to speak, is then both x and 1 exp(
R
L
t), so we can set the two equal and
solve for t:
1 exp(
R
L
t) = x
exp(
R
L
t) = x 1
exp(
R
L
t) = 1 x
77

R
L
t = ln (1 x)
t =
L
R
ln (1 x)
We now know how the circuit behaves as the power is connected. What happens when it is disconnected?
That is, we remove the voltage source and replace it by a short circuit (not an open circuit!).
Since the inductance will ght the change, we expect the current to go down gradually. It has to stop
eventually; power is burned in the resistor, so without a power source, a current cannot run forever.
Until then, the magnetic eld in the inductor collapses, and the magnetic energy will power the resistor.
To nd the quantitative answer, we need to solve the same dierential equation as before, but with V = 0.
We will nd the answer to be
I(t) =
V
R
exp(
Rt
L
)
with V being the now-removed batterys voltage, so that
V
R
was the stable current prior to the change we
just did to the voltage source.
So if we redene t = 0 to be a time where the current in
V
R
, we nd that indeed, the equation holds; the
current is its maximum. Same goes for t , where the current will asymtotically approach zero, in the
same way it asymtotically reached its maximum before.
After one time constant =
L
R
, then, we nd the current to be about 36.8% of its maximum value; after
two time constants 13.6%, and so on.
A common rule of thumb in electrical engineering is to take the process as complete after 5 time constants.
Thus, for an increasing current, after 5 time constants, it will be at 99.33% of its maximum, while for a
decreasing current, it will be down to 0.67%, which we consider essentially 100% and 0%, respectively.
We know that the power dissipated in the resistor is I
2
R at all times, so we can use that fact, and the fact
that the circuit now only contains an ideal inductor, ideal resistor and ideal wires, to calculate the total
energy stored. We do that do integrating the power dissipated in the resistor over all time from t = 0,
when I = I
max
, to t , when the current has died out:
E =
_

0
I
2
R dt =
_
I
max
exp(
R
L
t)
_
2
R = I
2
max
R
_

0
exp(2
R
L
t)dt
Note the added two inside the exponential, since we raised it to the 2nd power. The result of the integral
is
_

0
exp(2
R
L
t)dt =
L
2R
So the power is
I
2
max
R
L
2R
= I
2
max
L
2
=
1
2
LI
2
max
This relation holds for the current through the inductor at any instant in time, so we can remove the
max, and get an equation relating the current to the stored energy at any time:
E =
1
2
LI
2
78
Dual circuit elements
Notice the striking similarity between this, and the energy stored in a capacitor:
E =
1
2
CV
2
This is another of many similarities between the capacitor and the inductor, which are dual circuit elements.
Capacitors store energy in electric elds; inductors store energy in magnetic elds.
Capacitors ght a change in the voltage across them; inductors ght a change in the current through them.
The current through a capacitor is related to the rate of charge of the voltage across it:
I(t) = C
dV
dt
The voltage across an inductor is related to the rate of change on the current through it:
V (t) = L
dI
dt
If we integrate both sides of these equations, we will also nd that the voltage across a capacitor is related
to the integral of the current, and that the current through an inductor is related to the integral of the
voltage across it. This had the rather profound implication that to calculate the current through an
inductor, we must know its entire history, so to speak - we must know the voltage across it from t =
up to the time where we want to know the current. Either that, or just the voltage across it from t
1
to t
2
,
plus i(t
1
).
The same, or perhaps rather the opposite, is true for capacitors.
This makes capacitors and inductors state devices, that essentially have a memory, of a sort.
Magnetic eld energy density
Going back to the magnetic eld energy, since we know that stored energy, and we know the volume of
the solenoid, andt he fact that the magnetic eld is essentially exclusively enclosed with it (the B-eld
outside due to the solenoid is roughly zero), we can calculate the solenoids magnetic eld energy density,
by dividing the stored energy by the solenoids volume.
First, we can substitute in values for L and I that we had found previously, into the equation. Namely,
these values:
B =
0
I
N

we solve that for I, and nd


I =
B

0
N
We also had a value for L:
L = r
2
N
2


0
Substituting them into the equation, we have
1
2
LI
2
=
1
2
_
r
2
N
2


0
__
B

0
N
_
2
=
1
2
_
r
2
N
2


0
__
B
2

2
0
N
2
_
1
2
r
2

0
B
2

2
0
1
2
r
2
B
2

0
79
Note however, that because weve assumed that the magnetic eld outside the solenoid is zero, and that
r
2
is the volume where the eld exists, we can divide out those terms, and nd the magnetic eld energy
density:
U =
B
2
2
0
J/m
3
This, too has an electrical analogue. We calculated the electric eld energy density for a capacitor, which
we found te be

0
E
2
2
J/m
3
In the electrical case, this represents the work required to assemble charges in a certain conguration.
In the case of the inductor, this represents the work required to get the current running through a pure
self-inductor (one with zero electrical resistance). It takes work because the inductor ghts the change.
7.2.2 Alternating-current RL circuits
What if we have an RL circuit and power it by an alternating current? Say we have as simple series circuit
with a voltage source of voltage V
0
cos t, a self-inductor L and a resistor R.
Say the current is clockwise at one moment in time.
We walk around the circuit and add up the potential dierences
_

E d, according to Faradays law, and
then set those equal to the negative of the magnetic ux change.
First, we reach the inductor. As before, the E-eld is ZERO inside the ideal inductor, so it doesnt con-
tribute at all to the potential dierence sum.
We then reach the resistor, where the E-eld is in the same direction as the current, an so we add a
potential dierence of IR.
Finally, we reach the voltage source.
The E-eld is a power source is always in the opposite direction of the current, so it contributes with a
V
0
cos t term.
Finally, on the right-hand side, we have the change in ux
d
B
dt
= L
dI
dt
, via Faradays law, and the
equivalence via the denition of inductance, LI =
B
.
All in all, we have
0 + IR V
0
cos t = L
dI
dt
And again, we solve this dierential equation. Unfortunately, the solution to this equation is a bit complex
- and this result is the simplied version; the unsimplied answer is quite a bit worse!
We nd that
I(t) =
V
0
_
R
2
+ (L)
2
cos(t )
where
tan =
L
R
We can note two imporant facts here. First, the current will be sinusoidal (as we would expect), with a
peak of
I
peak
=
V
0
_
R
2
+ (L)
2
... which is just the term that multiplies the cosine; this comes simply from the fact that the cosine just
moves between 1 and +1.
80
The phase of means the current will appear later than the voltage; they will be out of phase. In an
RL circuit, this lag is between 0 degrees (purely resistive) and 90 degrees (purely inductive).
Another interesting - and vitally important - observation is that L plays the term of a resistance, in
two dierent places, no less. The phase lag is determined by the ratio of the inductive resistance L
(properly called the inductive reactance) to the regular resistance. The maximum current, meanwhile, is
determined by the resistance of
_
R
2
+ (L)
2
.
The sum of a resistance (symbol R) and a reactance (symbol X) is called an impedance (symbol Z).
There are more things we can deduce from this relationship. When is high, the current will be low -
this makes intuitive sense, because the inductor ghts the change. When the charge is more rapid, there
will be less time for the circuit to win the ght, and the current goes down. Likewise, when L is high,
the self-inductance has more strength to ght the change, and so the current goes down.
It is also intuitively pleasing that when is zero, i.e. we have DC, the entire equation simplies down to
just
I =
V
0
R
... which of course is just Ohms law.
Heres a table over values of the circuit, for dierent angular frequencies , for values V
0
= 10 V, R = 10 ,
L = 10
2
H and V = V
0
cos t:
f (Hz) 100 10
3
10
4
(rad/s) 628 6.3 10
3
6.3 10
4
L () 6.3 63 628
I
max
(A) 0.85 0.16 0.0016
32

81

89

At low frequencies R dominates over L, with the current coming relatively close to I =
V
R
. At a 10 times
higher frequency, L dominates, and the current largely depends upon the inductance L. After another
tenfold increase in frequency, the inductor really dominates, and the resistance is almost negligible, which
causes a phase lag very close to the maximum 90 degrees.
This formes the basis of RL lters, which can be used as low-pass or high-pass lters, depending on where
you read o the output voltage.
Say that you input a piece of music, such that the voltage source is not a simple sinusoidal wave, but a
full piece of music, constructed by tons and tons of dierent sinusoids.
If we pass them through an RL circuit, dierent frequencies will aect the current dierently. Low frequen-
cies will pass through the inductor almost unharmed, which means the voltage drop across the resistor
will be about the same as it would be without the inductor.
High frequencies - high - will be blocked by the inductor, and so the current through the resistor due to
the high frequencies will be reduced drastically.
If we then read o the resistors voltage drop, and feed that into an amplier and a speaker, we can hear
that all the high frequencies are gone from a piece of music, since the high frequencies of the current was
removed, thus causing no voltage drop across the resistor.
81
7.2.3 More on magnetic levitation
Lets return to the magnetic levitation we looked at last week, with a coil hovering about a conductive
plate. The setup was little more than that. We run 60 Hz AC through a loop (many loops, rather) of
conducting wire. This is positioned above a conductive metal sheet, and switched on. Given enough
current, the loop will levitate.
Lets begin by looking at the case where the current in the coul is in a clockwise direction. Via Lenzs
law, we know that there will be eddy currents induced in the metal plate, such that the induced B-eld is
opposite to the original one.
Therefore, the loop will have its north pole downwards, and the current loop have its north pole upwards,
so they repel each other, and the loop levitates.
Slighly later in time, the current in the loop is still clockwise, but it is now decreasing. Faradays law and
Lenzs law doesnt care about the current direction, nor the magnetic eld direction, but the sign of the
rate of charge of ux, which has now ipped - it is now decreasing!
Therefore, the current loops north pole is still downwards, but the eddy current will reverse, and have its
south pole upwards. Therefore, the two will attract each other! Say goodbye to magnetic levitation. Half
the time they will attract each other!
This is clearly not the case, as this was demonstrated and did work. The secret in inductive phase lag.
The EMF induced in the plate changes instantly
1
, but the induced current does not - because the plate
has a nonzero inductance! Therefore there will be a phase lag as given by
tan =
L
R
If this lag is zero, because L = 0 (which cannot happen - but say L is extremely, extremely small), then
on average, the force repels and attracts each with 50%, so there is no net force.
If the lag is greater than zero, however, we can get a repelling force more than 50% of the time. If the
lag is the full 90

, which can only happen as the ratio


L
R
, the force would be repelling 100% of the
time.
The rst image shows the situation with L 0, where the induced EMF (green) and eddy current (blue)
are in phase. Note that the induced EMF depends not on the coil current directly, but the derivative of
it, which is why the green and red curves appear out of phase.
The second image shows the case where L R, or rather when the ratio goes to innity, and the phase
lag goes to 90

.
1
Not faster than at the speed of light, of course!
82
The reality would be somewhere in between. Any phase lag greater than 0 degrees give a net repelling
force, thus allowing levitation; a greater lag would mean a greater net repelling force.
Of course, if the conductor is in fact a superconductor, then the phase lag will be exactly 90 degrees, and
the force will always be repelling.
7.3 Magnetic materials
7.3.1 A short note on motors
The lecture begins with some information about the MIT motor contest, which was held when the lectures
were recorded. Not a lot of physics is discussed, except one point about motors and current.
The ohmic resistance of a motors windings is often very low, so a very high current will ow when the
motor is just starting up. As it starts spinning, there is an induced EMF according to Faradays law,
which will counteract the external supply. Thus the current drops drastically, and later on settles at a
value, when the rotational speed is constant.
Therefore, if we hold a motor still while it is powered, the high current may damage it, so one should take
care to avoid this.
7.3.2 Magnetic dipole moment
As we have seen with some motor designs, a current loop in a magnetic eld will experience a torque.
Imagine a rectangular current loop, with sides a and b, with a current that is counterclockwise as seen
from above. Say the left and right sides are a, and the other two b.
Say there is a uniform magnetic eld B towards the right
The magnetic force on a straight wire will be

I

B (or I

L

B, since I is not really a vector).
For the right a side, the force will be experience a force in the downwards direction; the left a side will
experience a force upwards. The b sides will experience no force, as the cross product is zero there - they
are parallel and antiparallel with the magnetic eld, respectively.
Thus, there will be a net torque on the loop, given by [[ = IabB. If we dene A = ab to be the area of
the loop, we get [[ = IAB.
If there is an angle between the loop and the B-eld, the magnitude is given by [[ = IABsin . Thus, we
can write the torque as a cross product:
= I

A

B
where the vector area

A has the magnitude of the loops area, and the direction of the upwards surface
normal perpendicular to it.
The magnetic dipole moment is dened as this product I

A:
= I

A
then has the unit of ampere-square meters, A m
2
, or equivalently joule per tesla (J/T).
So the torque can also be written as
=

B
Because of its denition, the vector always points from the dipoles south to its north pole, which makes
it a handy tool in several places. It is sometimes known as m rather than .
83
7.3.3 The source of magnetism in matter
First o, lets state one thing clearly: what will follow here is mostly a classical interpretation of mag-
netism. For the truly correct answers, quantum physics is required. However, the results we nd here are
often in fairly good agreement with the correct ones.
We know that current loops create a magnetic eld. For a simple circular loop, going counterclockwise as
seen from above, the eld (and thus the magnetic dipole moment is upwards, via the right-hand rule.
This is the source of all magnetic elds, including those inside permanent magnets!
We also know that induced eddy currents create magnetic elds, but how does that apply to permanent
magnets, which do not rely on a change in external magnetic elds, or indeed do not rely on any external
magnetic elds?
The answer is that it doesnt apply: the mechanism is dierent.
Up until this moment in the course, I struggled to understand how a permanent magnet could possibly
work, given what wed learned prior... And then I read something similar to what follows in the next
paragraph, and it just clicked. There are simple current loops in all materials!
In the classical model of the atom, we have electrons in circular orbits, orbiting around the much more
massive nucleus. Because the denition of a current is a charge moving over time, the electron orbits
constitute tiny current loops - they are moving in a circle!
We can dene the magnetic dipole moment of an electron orbital by rst nding its current in amperes,
and then using the denion for = IA.
The current is the charge, divided by its period T. The period is in turn given by the distance travelled
(the circles circumference), divided by the velocity v:
T =
2r
v
The electrons charge is e, so we divide that charge by T, looking only at the magnitudes for now:
[I[ =
[ e[v
2r
(The current is dened by the amount of charge passing by a point, per second. Dividing the charge by
the period will get us that number!)
The magnetic dipole moment is then the current times the loop area A = r
2
:
=
ev
2r
r
2
=
evr
2
Because the electron has a negative charge, the current is then in the opposite direction of its orbit. If the
electron orbits clockwise as seen from above, we dene the current I as being counterclockwise.
The magnetic eld created follows the right-hand rule of the currents direction, so the magnetic dipole
moment is then pointing upwards, according to the right-hand rule.
So if all electron orbits produce a magnetic moment, and all materials contain orbiting electrons, why are
not all materials magnetic? The answer is simple: the magnetic moment due to one electron orbiting is
usually cancelled by one orbiting in the opposite direction, so that the net force in a material with many,
many electrons will be essentially zero.
We know that electric elds can induce electric dipoles in materials. In case the molecules or atoms them-
selves are permanent electric dipoles, an external electric eld will attempt to align them. The degree of
success depends on how strong the external electric eld is, and on the temperature: a lower temperature
means less random thermal motion, and so the dipoles are easier to align.
84
There is a similar situation with magnetic elds. If we have an external magnetic eld, it can induce mag-
netic dipoles in a material, at the atomic scale. In case the molecules/atoms themselves have a permanent
magnetic dipole moment, then the external magnetic eld will attempt to align them. Again, the lower
temperature will lead to a greater degree of success, due to less thermal motion.
When we bring a material into an external eld, then, the eld inside the material from the external eld.
The external eld will often be referred to as the vacuum eld here.
When we dealt with dielectrics, we saw how an electric eld could change inside a nonconductive material;
the eect was that the electric eld inside was lowered by a factor , the dielectric constant of the material.
Since > 1 for all natural materials, at least (and = 1 for the vacuum), the electric eld inside the
material was always lower than the external electric eld that induced the change to begin with.
That is in stark contrast to what we will now explore with magnetism. There are three possibilites, rather
than the one for electric elds above, when we deal with magnetic elds. The magnetic eld inside a
material can be
1) slightly lower than the external eld (diamagnetic materials),
2) slightly higher than the external eld (paramagnetic materials), or
3) a lot higher than the external eld (ferromagnetic materials).
Ferromagnetism is the form that most people are familiar with, where e.g. iron is strongly attracted to
magnets.
The other two forms have eects that are much weaker, to the point that they often go unnoticed. For
example, water is diamagnetic, and thus repelled by magnetic elds - only that it is repelled very weakly.
Before we talk about the three forms of magnetism, we should introduce magnetization.
7.3.4 Magnetization
Now that we know where the permanent dipoles in matter come from, lets look at a larger scale than the
atomic one, where the electron orbits are, and switch our focus to what happens when we have lots and
lots of these tiny dipoles in a material.
Suppose we have a cylindrical piece of material, with end-cap area A and height L, consisting of N individ-
ual dipoles, each with a magnetic dipole moment , spread uniformly throughout the cylinders volume.
All of the dipole moments are also aligned with the cylinders axis, i.e. they all point upwards.
In the absence of an external magnetic eld, what is the eld due to these dipoles alone?
We dene the magnetization vector

M to be the net magnetic dipole moment per unit volume:

M =
1
V

i

In the case of the cylinder, since all moments are aligned, the net moment is simply N. To get the per
unit volume part, we just divide by the volume, AL:
M =
N
AL
Clearly, the direction of

M is the same as all the individual moments , if they are all aligned.
Since is in ampere-square meters (A m
2
), and the volume in cubic meters, the magnetization is then in
amperes per meter, A/m.
The average magnetic eld due to this magnetization B
M
is then given by
85

B
M
=
0

M
This is the large-scale average throughout the material; close to each of the dipoles, the eld will be very
dierent.
This principle is similar to the E-eld due to several nearby point charges; in the vicinity of the changes,
the eld will be very complex, but from very far away, the eld will look as one due to a single point
charge, with the same charge as the sum of the charges due to the collection.
7.3.5 Paramagnetism
Lets now start looking at the dierent forms of magnetism.
In paramagnetism, the atoms or molecules in the material themselves have a magnetic dipole moment,
such that they are in a way tiny magnets.
In the absence of an external eld, these dipoles are randomly oriented, such that there is no net magnetic
eld.
Once exposed to an external eld (a vacuum eld), that eld will attempt to align them. The success
depends on the vacuum elds strength, and on the temperature, such that a lower temperature means
easier alignment, due to reduced thermal motion.
The alignment will be such that the induced magnetic eld in the material adds to the external eld, and
the eld inside the material is slighly stronger than the external eld. This is in contrast to diamagnetism,
where the eld inside was weaker.
Because of this, paramagnetic materials are attracted to magnetic elds, while diamagnetic materials are
repelled.
More quantitatively, the alignment of the atomic-scale dipoles with the external eld (which well call

B
0
)
creates a net magnetization

M that is parallel to

B
0
. Since the induced eld

B
M
is parallel to

M, then

B
M
too is parallel to

B
0
. We see, then, that the eld due to the paramagnetism indeed strengthens the
external eld.
The total magnetic eld in the material is then given by

B =

B
0
+

B
M
=

B
0
+
0

M
In most paramagnetic materials,

M is proportional to the external eld

B
0
. The proportionality constant
is known as the relative permeability, and is written as
m
:

B =
m

B
0
Because the eect of paramagnetism (and diamagnetism) is so small,
m
is usually very close to 1, and so
it gets cumbersome to talk about numbers such as the relative permeability of e.g. magnesium, which is
1.000012. Because of this, we can also dene the relation in terms of the magnetic susceptibility
m
(
being the Greek letter chi) of a material. The relation between the two is such that

m
= 1 +
m
Thus, instead of the relative permeability of magnesium being 1.000012, we can talk about its magnetic
susceptibility of 1.2 10
5
.
Both relative permeability and magnetic susceptibility are dimensionless numbers.
We can then combine the relative permeability with the permeability of free space
0
to get the magnetic
permeability
m
(sometimes only , which can easily be confused with magnetic moment) of a material:

m
=
m

0
= (1 +
m
)
0
86
For paramagnetic materials,
m
>
0
. However,
m
is often on the order of 10
6
to 10
3
, with many
materials having even lower values. In other words, the permeability of these materials is very, very close
to that of the vacuum, which also implies that the magnetic forces will be rather low.
Note that one of the meanings of this is that eld lines will prefer to go through the paramagnetic
material, versus outside it, where the permeability is lower (assuming its in a vacuum or in air, with
m
essentially 1).
Lets now look at why a paramagnetic material would be attracted to an external magnetic eld.
If we imagine each molecule as a tiny bar magnet, with a dipole moment pointing at an angle with the
vacuum eld, then there will be a torque on the molecule, such that the torque causes its poles (or its
dipole moment) to align with the vacuum eld. This torque would then by given by =

B.
Paramagnetic materials will be pulled towards the strongest point in a non-uniform external eld.
Lets look at an example.
Say we have a bar magnet as above, with the north pole downwards, creating a non-uniform magnetic
eld. Below it, we have, say just an atom, with a dipole moment that is also downwards.
We can think of this in two ways. Either we simply realize that the downwards dipole moment of the atom
means that its north pole is downwards, and thus the south pole upwards, and so north and south attract
each other.
Another way to think of it is to consider the current loop as drawn. On the left side, the current goes
into the blackboard, and the magnetic force is given by the cross product of the currents direction and
the B-eld due to the bar magnet, angled downwards, so the force will be at an angle upwards/to the left.
On the opposite side, we nd the magnetic force to be upwards/to the right. The net force will then be
upwards, so the material will be attracted to the bar magnet.
What if the atom were at an angle, so that the dipole moment was towards the right in this picture? There
would be a torque on it that tries to align it with the eld. This is yet again a quantum mechanical eect,
so we will unfortunately just take that for granted now.
7.3.6 Diamagnetism
Most of the things we have said about paramagnetism either applies to diamagnetism, or the opposite
does. The dierences will be stated below.
Diamagnetic materials have no permanent magnetic dipoles; an external eld

B
0
can induce magnetic
dipole moments in atoms or molecules, however. These induced dipoles will be anti-parallel to

B, and
thus serve to weaken the eld inside the material.
Diamagnetic materials have a magnetic permeability
m
<
0
, and therefore a relative permeability
m
< 1
(since
m
=
m

0
), meaning
m
< 0.
87
Here, the magnitude of
m
is even smaller than that of paramagnetic material, perhaps on the order of
10
5
to 10
9
.
Just as how eld lines prefer to go through paramagnetic materials, because of their (slightly) higher mag-
netic permeability than the surrounding vacuum/air, eld lines prefer to avoid/go around diamagnetic
materials, as the permeability is lower than that of the vacuum.
This eect is striking in superconductors, which are essentially perfect diamagnets, with
m
= 1 and

m
= 0, such that they repel all elds entirely, except for a very thin layer near the surface (which is
usually less than 1 micrometer deep, given by the London penetration depth, which is likely not covered
by this course).
All materials are diamagnetic; however, for those that are also paramagnetic or ferromagnetic, as we can
see on the tiny susceptibility of diamagnetic materials, those other eects will usually overshadow the
diamagnetism.
As stated above, the induced dipoles weaken the internal magnetic eld, and so they oppose the external
eld. Diamagnetic materials will therefore be repelled by external magnetic elds, as their dipole moment
will always tend to oppose external elds.
Due to the low susceptibility, the eects of diamagnetism is usually very small, and often hard to even
notice, at least in everyday life. It does however result in water, a diamagnetic material, being repelled by
very strong magnetic elds, to the point that phycisists have levitated a living frog (and also mice) by the
repelling force on the water inside it. Doing so took a magnetic eld with a strength of roughly 16 tesla!
7.3.7 Ferromagnetism
Ferromagnetism is perhaps the most interesting form of magnetism, because it forms the basis of perma-
nent magnets.
Like paramagnetic materials, ferromagnetic materials also have permanent dipole moments.
However, (yet again!) for quantum mechanical reasons beyond this course, these dipoles are grouped into
domains, which are on the order of 10
6
to 10
4
m in size (very roughly), such that even large domains are
essentially microscopic. Inside each of these domains, all the individual dipole moments will be completely
aligned in one and the same direction. However, since there are many domains in a piece of material, there
may still not be a net magnetic moment. Each domain has a separate alignment, which may cancel out
with another nearby domain, to provide no net moment.
There may be up to on the order of 10
17
to 10
21
atoms in a single domain.
As weve seen before, when we expose the material to an external magnetic eld, the dipoles (and thus the
domains) will try to align along the external magnetic eld; the degree of success depends on the strength
of the external magnetic eld, and on the temperature; the lower the temperature, the easier alignment
is, due to lower thermal motion.
The magnetic eld inside a ferromagnetic material may be thousands of times stronger than that of the
external eld, that is,
m
can be 1000 or more. Thats in fairly stark contrast to paramagnetic and dia-
magnetic materials, which had factors more on the order of 1.001 or less!
In addition, the domains in a ferromagnetic materials can stay aligned, to a certain degree, after the re-
moval of the external eld, creating permanent magnets.
This alignment can be removed either by sudden shock (such as hitting the material with a hammer), or
by heating the material; in either case, the domains will randomize again and the residual magnetism may
well go away entirely.
88
When we heat a ferromagnetic materials, at a certain temperature, the domain completely fall apart, and
disappear. The point where this happens is nown as the Curie temperature, or the Curie point. Irons
Curie temperature is 1043 degrees K, or 770 degrees Celsius.
At this point, the material becomes paramagnetic instead, in a very abrupt change.
89
Chapter 8
Week 8
8.1 Hysteresis and electromagnets
Last week, we talked about magnetic permeability, the relative permeability
M
of materials, and briey
about ferromagnetism, the only form of magnetism where
M
has a signicant value. We will now talk
about how strong a magnetic eld we can theoretically get in a material, and about hysterisis in ferromag-
netism - a phenomenon where the past state of the material changes its permeability, in a nonlinear fashion
(so the magnetization and permeability are no longer directly proportional to the external magnetic eld).
If we were to align 100% of the atomic dipoles in a material, what would the strength of that eld be?
Lets start with a slightly dierent derivation, that we will later use to answer the above question.
Say we have a hydrogen atom, consisting of a proton at the center, and an electron orbiting with a radius
R, in the classical model.
If the electron, as seen from above, orbits in a clockwise fashion, the current - we have a moving charge,
so there is a current - will by convention be in the counterclockwise direction.
The magnetic dipole moment is then given by IA, where A = R
2
is the area of the current loop. The
direction will be, via the right-hand rule, upwards.
Lets then try to nd the value for I, which takes a bit more work.
If the electron is orbiting in a circle around a center point, there must, according to classical mechanics,
be a centripetal force on it, acting radially inwards. We can use Coulombs law to nd the electric force,
and set that equal to the centripetal force
mv
2
R
:
m
e
v
2
R
=
[e[[ e[
4
0
R
2
v
2
=
e
2
4
0
Rm
e
v =
e
2

0
Rm
e
The time to go one loop around (the period T) is then the distance around, 2R, divided by the velocity:
T =
4R

0
Rm
e
e
If we stick some numbers in there, using R = 510
11
for the orbit radius (the Bohr radius, about 5.2910
11
meters), we nd v 2.3 10
6
m/s and T = 1.4 10
16
s - rather incredible numbers!
90
We can now calculate the current: current is dened as the amount of charge passing a point per second
- so if we choose a point along the orbit, the electron will pass it
1
1.4 10
16
times per second. We can
therefore calculate
I =
e
T
=
1.602 10
19
1.4 10
16
1.144 10
3
A
We nd a current of about a milliampere - due to a SINGLE electron orbiting! In a copper wire, a current
of a milliampere would take roughly 6.24 10
15
electrons passing by a point per second (a coulomb is
approximately 6.24 10
1
8 electrons, the reciprocal of the elementary charge 1.602 10
19
C) and in this
case ONE electron is doing all that, because of its incredible velocity around the tiny orbit.
The magnetic moment is then = IA, as we stated, so
= IA = IR
2
8.98 10
24
A m
2
This value is known as the Bohr magneton. However, its value isnt really what we just found, but rather
the value of the Bohr magneton
B
is

B
=
e
2m
e
9.274 10
24
J/T
... which is a result from quantum mechanics, which is where comes from (the reduced Planck constant).
The value cannot be found accurately using classical physics.
The magnetic moment due to all electrons in orbit around an atom can only ever be a multiple of the
Bohr magneton - never anything it between. This is one example of the quantization (only a certain set
of discrete values are allowed) that gives quantum physics its name.
In another quantum weirdness eect, electrons themselves produce a magnetic moment due to a quantum
eect called spin, often thought classically as a literal spin of the electron, around its own axis.
The net magnetic dipole moment of an atom is then the vectorial sum of all the orbital dipole moments,
due to the electrons orbits, and of the spin dipole moments. The result of that is that most atoms and
molecules have dipole moments of either one or two Bohr magnetons. The limit on the maximum eld
strength in a material (due to the material itself), then, will be on the order of 2 Bohr magnetons, times
the number of atoms we can manage to align all in the same direction.
When we expose a material to an external magnetic eld, the eld inside will be

B =

B
0
+

B

where

B
0
is the external eld (the vacuum eld), and

B

is the eld due to the aligned dipoles.


For paramagnetic and diamagnetic materials, we then nd that

B =
M

B
0
=

B
0
+
M

B
0
. However, as we
will soon see, that relationship does not hold so easily for ferromagnetic materials. The reason is that in
ferromagnetic materials, an external eld can align dipoles so well that essentially all of them are aligned.
At that point, clearly, increasing the external eld will have no eect on the materials dipole alignment -
they are already aligned, and so

B

is at a maximum.
When all dipoles are aligned, and an external eld cannot increase the materials induced eld further, we
call that saturation.
Say we have a material where each atom contributes with 2 Bohr magnetons, with a density of N = 10
29
atoms per cubic meter in our material.
Inside the material, a solid material, the atoms will be nicely packed together. We can imagine them as a
lot of small current loops in a grid; each column of the grid (which is really 3-dimensional, of course) then
looks a lot like a solenoid - many small windings, each with a current through them. Since weve said that
all dipole moments will be aligned, we can indeed approximate the B-eld strength using the equation for
91
a solenoids magnetic eld.
Each column will have an area A and a height ; lets = 1 meter, for simplicity. Each column is then
A = A m
3
in volume. The equation for the B-eld inside a solenoid is
B =
0
I
N

So somehow, we need to nd a way to put


B
inside there. Note that N in the above equation is the
number of windings, while N is a dierent symbol (look closely), which is the number of atoms per cubic
meter.
If we take the area (not the volume!) A times N, we get the number of atoms per meter. Since each atom
is really a winding of this solenoid, we can substitute this term in for
N

in the solenoid equation, since


that term means nothing but the number of windings per meter. Thus, we have
B =
0
IAN
A-ha! But = IA, so we have
B =
0
N =
0
(2
B
)N
... using 2 times the Bohr magneton
B
as the value for the dipole moment. We can thus calculate the
B-eld strength by putting numbers in the above equation:
B (410
7
)(2 9.3 10
24
)(10
2
9) 2.34 T
The units of Wb/Am times A m
2
times m
3
check out, and equal the unit of tesla (since 1 tesla is 1
Weber per square meter).
So for the values we have chosen, with 2 Bohr magneton dipole moment per atom, and 10
2
9 atoms per
cubic meter, about 2.3 tesla is the value we nd for the maximum B-eld strength due to the material
itself.
8.1.1 Ferromagnetism and hysteresis
Lets now look at the B-eld strength inside a ferromagnetic material. Lets say that it has a relative
permeability
M
= 1000, to start with. That will only hold for a part of this section, however!
If we try to plot this, we will be unable to do so on a linear scale. Say we plot the vacuum eld B
0
on the
x axis, and the induced eld B

on the y axis. Since


M
= 1000, for each unit we go towards the right, we
need to go 1000 units up! That is clearly not possible to draw in an easy way, on a to-scale linear plot.
Nevertheless, say we do this on a linear plot, just not to scale.
In the beginning, the slope will indeed be the
m
of 1000, but after a while, so many of the atomic dipoles
will have aligned, that the eect of further increasing the external B-eld lessens. There then comes a
point where increasing it has no eect at all on B

, when all (or close to all) dipoles are aligned; this is


what we call saturation. If we do this on the material where we calculated this above, then B

will have a
magnitude of about 2.3 tesla when we reach the saturation point.
At this stage, then, the eect increasing the external eld will have on B

is essentially zero.
The total eld B = B
0
+B

inside the material can however always be increased, by increasing the strength
of the external eld; B

is the only eld that is at its maximum strength.


92
Say we start with a nonmagnetic ferromagnetic material, and stick it inside a solenoid. B
vac
, the vacuum
eld, we can always calculate: we know how to nd the B-eld inside a solenoid, all we need to measure
is the current through it.
So we start at the origin, and increase the vacuum eld, until we reach to point in the top right (in the
gure above), where B

saturates.
We then reduce the current in the solenoid to 0, and B
vac
reaches zero (at point P, just above), as expected.
However, B is not zero, because the ferromagnetic material has now become a permanent magnet!
Note that we then see that B has two possible values for B
vac
= 0!
But wait, theres more!
If we now change the direction of the current in the solenoid, the vacuum eld will reverse direction, to
the left side of the y axis. If we increase the current (in the new direction) until we reach point Q, we
see that the net eld in the material is now zero! The domains in the material are still pointing to the
direction they were in at point P, which is now in the opposite direction of the vacuum eld, since we just
reversed that.
If we then move the current back to zero again, the vacuum eld goes away, and we end up at point S
(also above). The material is now a permanent magnet again, but this time in the opposite direction as it
was before, since we coerced the domains to reverse direction.
93
Finally, if we increase the current in the original direction again, we go from S back up to the point it the
top right.
Note that if we were to decrease it again now, we would get back to point P! We can not get back to the
origin again simply by reversing the current.
This also implies that given a material, if we expose it to an external eld, we cannot calculate what the
eld inside will be - since it depends upon the history of the material.
Looking at the relative permeability
M
, then, shows why we cannot use a xed value for ferromagnetic
materials. At the two points where the hysteresis curve meets the x axis (outside the origin) ,
M
must
be zero, because there is a vacuum eld, but the B-eld inside the material is zero! B =
M
B
vac
, so
M
must be 0 if B
vac
is greater than zero, but B is exactly zero.
Consider the situation in the upper-left quadrant. The vacuum eld is towards the left, but the B-eld
inside the material is towards the right - so
M
must be negative! The same applies to the lower-right
quadrant.
Demagnetization
This raises the question: can we get the material back to being unmagnetized? The answer is yes, we can.
If we heat it above its Curie point, that will destroy the magnetic domains, and when it cools back down,
it should be unmagnetized. We can also possibly do it by shock - hitting it with a hammer, etc.
Another nice solution is to pass a sinusoidal current through it, with decreasing current, which is called
demagnetization.
Because B
vac
is linearly proportional to the current I through the solenoid, if we pass a sinusoidal current
through it, the B-eld inside the material will move along the hysteresis curve we just saw.
However, if we gradually decrease the currents magnitude, we gradually move inwards, so that the
points where we cut the x axis move closer and closer to the origin, and when the current reaches zero,
the material has been demagnetized.
Ferromagnetic materials and eld changes
If we have a permanent magnet (top left in the illustration below), and we move a bit of nonmagnetized
ferromagnetic material near it, what will happen? I think we all know, in part even since childhood, that
the ferromagnetic material will be attracted to the magnet.
Using our new knowledge, we also know that the atomic dipoles inside the material will align to support
the external eld, which in turn causes the net eld to strenghthen in the left/right plane, and so there
will be an induced south pole near the north pole of the magnet. Due to the strengthened eld inside
the material, the external eld lines are sort of sucked in into the material (the pink lines).
94
Because of this, the external eld will weaken towards the left of the picture, where two eld lines are drawn.
This can be (and is, in the lecture) be shown experimentally, by tying a ferromagnetic nail to a thread,
with a magnet hanging down. The nail is tied such that it is held up, against gravity, by the attracting
magnetic force.
When we move another ferromagnetic object (a wrench, in the lecture) towards the other side, the eld
lines are sucked in to the wrench, and the eld weakens around the nail, and it falls down, without even
being touched.
It is also demonstrated that while this eect also happens with paramagnetic materials, such as aluminium,
it is so positively tiny that the nail stays up with no problems.
8.1.2 Maxwells equations
Now that we have learned about magnetic materials, we can nally write down the complete set of
Maxwells equations (in integral form). There is only one of the four that changes, and that one is
Amperes law.
Looking at the other three equations (below), there is no reason to suspect that magnetic permeability
would cause any change. However, Amperes law uses a
0
term, which is the permeability of the classical
vacuum. If the permeability of the medium in question is dierent, then we need to use the mediums
permeability, by adding in a
M
, so that we have
M

0
instead. All in all, the four equations are, in
integral form (the only form taught in this course):


E

dA =
Q
free

0
(Gausss law)


B

dA = 0 (Gausss law for magnetism)
_

E

d =
d
B
dt
(Faradays law of induction)
_

B

d =
M

0
_
I
pen
+
0

d
E
dt
_
(Amperes law)
95
8.2 Review for Exam 2
The list of topics covered on the exam, as stated on the transparency in lecture:
Magnetic elds & eld energy
Lorentz force
Cyclotrons
BiotSavart Law
Solenoids
Amperes law displacement current
Lenzs law
Faradays law of induction
Motional EMF
Induction eddy currents
AC dynamos
3phase current
Induction motors
Aurora
Magnetic levitation bullet trains
Selfinductance
RL circuits
Quite a lot! Of course, knowing all the material from the rst exam is most likely necessary, too, so the
full list ought to be everything learned in the course so far.
Everything covered in the review was, well, a review... So I didnt take any additional notes.
96
Chapter 9
Week 9
9.1 Transformers, Car Coils and RC circuits
9.1.1 RC circuits
We previously looked at series RL circuits - circuits with an inductor and a resistor, plus a power source
and a switch. We will now look at series RC circuits - so the only dierence is that we replace the inductor
with a capacitor.
Thus, we have a circuit of a capacitor, resistor, battery with EMF V
0
, and a switch, going around the
circuit on the blackboard clockwise (not that order matters!).
The two nodes the capacitor are connected to are labelled V
A
and V
P
, so the capacitor voltage is
V
C
= V
A
V
P
Intuition goes a long way when it comes to solving this circuit, especially for the simple cases t = 0 and
t , just after we close the switch, and when it has been closed for a very, very long time.
Capacitors take time to charge up to a voltage, just as how inductors take time to charge with a current.
The two circuit elements are duals - as mentioned previously. That means that while inductors take time
to let a current through, with a EMF that goes down to 0 with time, capacitors are the opposite: the
current goes to 0 with time, while the voltage increases to its maximum.
So at t = 0, we expect the capacitor to be unchanged (V
C
= 0), but charging (I = I
max
=
V
0
R
).
At t , the current must die down, at which point V
C
= V
0
(otherwise part of the batterys EMF must
be across the resistor, which means I =
V
R
,= 0). So at this point, V
C
= V
0
and I = 0.
What about in between? Well, to nd the correct answer, lets get serious and analyze this circuit properly.
The lecture questions asks us to use Faradays law, which is unnecessary since there is no changing magnetic
ux (Kirchhos rule works), but lets do so.
We start before the battery and go clockwise, adding whereever the electric eld is in the direction we
are moving, subtracting otherwise. We nd:
V
0
+
Q
C
+ IR =
d
B
dt
= 0
Alternatively, adding when we go up in potential, subtracting when we go down:
+V
0

Q
C
IR = 0
97
The equations are of course the same, with both sides having been multiplied by 1 in the second case.
We can write I as
dQ
dt
, which it equals by denition. We then get, rearranged,
Q
C
+ R
dQ
dt
V
0
= 0
... which is a dierential equation in one variable. The solution to it, given the initial condition V
C
(0) = 0,
is
Q(t) = V
0
C
_
1 exp
_

t
RC
__
Knowing the charge is generally not that useful, but knowing the current is; since current is the time
derivative of charge, we dierentiate this result with respect to t:
dQ
dt
= I(t) = V
0
C
_
1
RC
exp
_

t
RC
__
I(t) =
V
0
R
exp
_

t
RC
_
To nd the voltage across the capacitor, we simply use the denition of V =
Q
C
, and the result for Q we
already found. We divide out the C, and nd
V
C
(t) =
Q(t)
C
= V
0
_
1 exp
_

t
RC
__
Much like RL circuits, RC circuits also have a time constant (tau). In the case of RC circuits, its given
by RC, the unit of which is seconds:
Ohm =
Volt
Ampere
Farad =
Coulomb
Volt
Ohm Farad =
Volt Coulomb
Ampere Volt
=
Coulomb
Ampere
=
Ampere second
Ampere
= second
As with RL circuits, after one time constant , the values have either gone down to
1
e
of their initial value,
or up to (1
1
e
) of their maximum, depending on in which direction they are going.
So the potential dierence across the capacitor after one time constant = RC is 63.2% of its maximum,
while the current is down to 36.79% of its initial value.
If we now ip the switch in the circuit, such that the capacitor and resistor are in series with nothing else
in there, the capacitor will dischange and its stored energy will be burned as heat in the resistor.
The equation governing that circuit will be
Q
C
IR = 0
98
If we dene I to be positive in the opposite direction, this turns into
Q
C
+ R
dQ
dt
= 0
The solution for that dierential equation is
Q(t) = V
0
C exp
_

t
RC
_
We use the same tricks (dividing by C and dierentiating with respect to t, respectively) to nd V
C
(t) and
I(t):
V
C
(t) =
Q(t)
C
= V
0
exp
_

t
RC
_
dQ
dt
= V
0
C
_

1
RC
_
exp
_

t
RC
_
=
V
0
R
exp
_

t
RC
_
I(t) =
dQ
dt
=
V
0
R
exp
_

t
RC
_
So the current will now ow in the counterclockwise direction (remember that we redened a positive I as
counterclockwise) as opposed to clockwise back when the battery was charging the capacitor.
So to summarize, when charging a capacitor in a simple RC circuit, the following equations apply:
I(t) =
V
0
R
exp
_

t
RC
_
V
C
(t) = V
0
_
1 exp
_

t
RC
__
When discharging, the following equations apply (assuming it is charged to voltage V
0
):
I(t) =
V
0
R
exp
_

t
RC
_
V
C
(t) = V
0
exp
_

t
RC
_
... where the current is in the opposite direction when discharging.
99
9.1.2 Transformers
Transformers are used to transform AC (and only AC!) currents and voltages.
To construct one, we take a coil, which we call the primary, with N
1
turns, and self-inductance L
1
.
We then add a second winding, electrically isolated, around the same core. This winding is called the
secondary, and has N
2
turns and self-inductance L
2
.
Since they are wound together (while electrically isolated), there is magnetic ux coupling between the
two. As the current changes in the primary, via Faradays law, a current is induced in the secondary.
So, we apply Faradays law to the primary side. We begin at the top, and rst go through the primary
coil, where we know there is no electric eld (since we treat the inductor as ideal, R = 0). Thus that
doesnt contribute to the integral at all.
We then reach the voltmeter, where the current is opposing the direction we are moving around the circuit,
so it contributes with a V
1
term. According to Faradays law, that is then equal to the negative of the
rate of charge of magnetic ux, or
V
1
= L
1
dI
1
dt
... via the denition of inductance. This is a simplied result, where the proper result also has a term
related to the mutual inductance between the two coils, and the current in the secondary.
We do the same on the secondary. The current is clockwise here as well, so we get a positive contribution
at the voltmeter:
V
2
+ 0 = L
2
dI
2
dt
Again, we ignore the contribution of the mutual inductance.
The L
dI
dt
are, as usual, the induced EMF due to Faradays law. Therefore, we can also write the equations
as
[V
1
[ = E
1
= L
1
dI
1
dt
= N
1
d
B
dt
[V
2
[ = E
2
= L
2
dI
2
dt
= N
2
d
B
dt
Due to the way the coils are constructed, assuming we have perfect magnetic coupling, the ux term is
the same for both equations! Therefore, we nd that

V
2
V
1

=
N
2
N
1
The result speaks for itself: we can choose the number of windings on each coil such that we create huge
voltages at the secondary - step-up transformation - or, we can do the opposite, and get a step-down
100
transformer instead.
Because ideal transformers are lossless, we can show that as power is conserved, a transformer that steps
down voltage will step up current, and vice versa. Thus, a transformer with N
2
/N
1
= 100 will draw a
much larger current on the primary than it delivers on the secondary, since the voltage at the secondary is
larger, but the power must be the same on both sides. If we ip the primary and secondary, the opposite
is true: with the ip, the output voltage is very low, but the current is much greater than the current
drawn at the primary.
The product V
1
I
2
is always equal to V
2
I
2
in an ideal transformer; more on this below.
Keep in mind that these voltages are peak values (or RMS values); transformers only work with AC, since
there would be no change in magnetic ux for DC.
Transformation is used a lot in power distribution. As discussed earlier in the course, power transfer is
done at very high voltages, which means we can power with less I
2
R losses, as I becomes smaller for a
given power transferred.
The power is generated, upconverted to a few 100 kilovolts or so, and fed into the power lines. When
it reaches a city, it is stepped down, but is still in the kilovolts range. Then, close to homes, it is again
stepped down, to the 110 120 volts or 230 240 volts used around the world.
Lets return to talking about currents in transformers. The currents are proportional/inversely propertional
to the windings ratio if a few things are true - that by no means are always true. They are:
R L in both the primary and the secondary
No energy is lost to eddy currents, e.g. in the transformer core, which is often iron
The ux coupling is ideal, so that the ux in the primary equals the ux in the secondary
In these cases, power is conserved, so V
1
I
1
= V
2
= I
2
.
If all of these are true, then it is true that

I
2
I
1

=
N
1
N
2
9.1.3 Spark plugs / car coils
We can use our newfound knowledge on transformers to understand how the spark plug in a car works.
(A spark plug is used to ignite the fuel-air mixture in the engine.)
The car battery is a 12-volt DC battery. However, by placing a switch in series with the transformers
primary, we can still get a huge
dI
1
dt
value by letting the current built up, with the switch closed, and then
opening the switch, thus instantaneously breaking the current.
L
1
will then try to oppose this change, and generate an EMF in response. However, the magnetic ux
change responsible for the EMF, via Faradays law, is shared by the secondary.
101
The secondary is then wound such that N
2
N
1
, and the induced EMF in the secondary will be way
higher than the already high value in the primary.
Therefore, if we simply connect two wires to the secondary, and place them a short space apart, we will
get sparks between them if the electric eld between them exceeds the roughly 3 million volts per meter
that causes electric breakdown in air.
For a distance of, say 3 mm, that voltage would be 9 kilovolts. We cound generate a hundred times that
with this method, but that would not be necessary for this usage.
9.2 Driven RLC circuits and resonance
We have discussed RL and RC circuits previously, and will now discuss RLC circuits (or LCR, or LCR,
etc.), i.e. series circuits with a power source, a resistor, a capacitor and an inductor.
We will only talk about the case of a sinusoidal voltage applied to the circuit.
Walking clockwise from the top-left corner in the circuit on the blackboard, using Faradays law, we get
Q
C
+ 0 + IR V
0
cos t = L
dI
dt
The zero term comes from the self-inductor: we treat it as ideal, so the E-eld inside is zero, and thus it
has no real voltage drop. Measuring across it will show a voltage, which is rather an induced EMF than
a true potential dierence.
We want to write everything in terms of Q, plus the constants (R, L, C and V
0
, so we replace the
dI
dt
with
its other form, via I =
dQ
dt
, and move it to the other side at the same time:
Q
C
+ R
dQ
dt
+ L
d
2
Q
dt
2
= V
0
cos t
The way to solve this second-order dierential equation wasnt shown in lecture, as that is the subject of
a course not yet taught at this stage at the MIT students studies.
The solution for the current in the circuit is given as
I(t) =
V
0
_
R
2
+
_
L
1
C
_
2
cos (t )
where
tan =
L
1
C
R
As before, the expression multiplying the cosine is the maximum current possible through the circuit, at
a particular V
0
.
This solution is only valid for the steady state. When we power such a circuit up, the sine wave will be
multiplied by an exponential term, which turns into 1 once steady state is reached.
We can now dene two (or four, perhaps) new terms. The term in the numerator of the tangent is called
the reactance, symbol X, and is eectively a frequency-dependent resistance:
X = L
1
C
Combined with the resistance, the reactance gives us the impedance, symbol Z:
102
Z =

R
2
+ X
2
=

R
2
+
_
L
1
C
_
2
The two parts of the reactance are the inductive reactance X
L
= L and the capacitive reactance
X
C
=
1
C
, respectively.
Both reactance and impedence have the units of ohms, and can in many cases be treated as resistances,
as long as the input is sinusoidal.
The solution has a few quite interesting subtleties. One of them is resonance: there will be a value for
where the current is at a maximum. We can nd this by doing some simple math (no calculus optimization
required). Looking at the expression for I
max
:
I
max
=
V
0
_
R
2
+ (L
1
C
)
2
... we see that no matter what, the R term will limit the current. Since the reactance is squared, we
cannot get it negative, only zero. So it will be at its lowest when
L
1
C
= 0

2
CL 1 = 0

2
CL = 1
=
1

LC
This is known as the resonance frequency, often given the symbol
0
.
At resonance, the current is given simply by
I
max
=
V
0
R
The current will also be in phase with the driving voltage, as becomes zero.
Do note, however, that when not at resonance, can be either greater than zero (which causes a phase
lag: current lagging the voltage), or smaller than zero, in which case the current leads the voltage. The
latter is due to the capacitive reactance, while the former is due to the inductive reactance.
Lets now look at how the circuit behaves at dierent values of , the driving frequency, for xed values
of R, L and C.
As the frequency is extremely low, 0, the current will go to zero, due to the capacitor. At that point,
our current really goes to DC, and so we charge the capacitor to V
0
, and then the current stops. This can
of course be seen in the equation, from the
1
C
term.
As , the current again becomes zero. This time, the inductor is responsible. With the extremely
high dIdt, the inductor has a lot of power to say no to a getting a current through it.
We already saw that the maximum possible current is when the reactance becomes zero, at the point
where the inductive and capacitance reactances take each other out.
Between the two extremes, we got a plot that looks something like this (on a log scale):
103
As noted in the graph (in pink), below the resonance frequency, the capacitor plays the main role in de-
ciding the current, and the phase in also in favor of the capacitor, such that the current leads the voltage.
Above the resonance frequency, the inductor dominates. The phase is in favor of the inductor here, such
that the voltage leads the current.
Exactly at the resonance frequency, there is no phase dierence between the voltage and current, and the
current is at a maximum, as stated previously.
There are two points along the plot above where the current is 0.707 (
1

2
) times its maximum value.
These two points are known as the half-power points: the power I
2
R will be exactly half of its maximum
when the current is
1

2
of its maximum.
If we label the lower half-power point
1
and the upper
2
, the width
2

1
between these points is
known as the bandwidth, , which for a series RLC circuit is
=
R
L
(the derivation of this is not shown.).
The quality, or quality factor Q, not to be confused with charge, is given by

=
1

LC

L
R
=
L
R

LC
=
1
R
_
L
C
... if we ip the fraction and multiply, instead of the unreadable 4-level fraction division.
Clearly, this is inversely proportional to the bandwidth, so the narrower the bandwidth, the higher the
quality factor, and vice versa. A high-quality resonator will have a very narrow and sharp peak.
Resonant circuits can be used in radios, TVs, etc, to select a station. The antenna will recieve a wide
variety of stations all at the same time, and a nely tuned, high-Q resonant circuit with adjustable
0
(often using an adjustable capacitor) is used to lter out one of them.
In reality, it is more complex, but the principle is absolutely used as part of the full solution.
Metal detectors also use resonant circuits. They are often built using two coils, nely tuned, and at
resonance. When a piece of metal comes near, eddy currents are induced in the metal, and so the magnetic
ux coupling between the loop changes, the mutual inductance between the two coils is disturbed, and the
system goes o resonance, and sounds the alarm.
The metal detectors at airport use this principle in a very simple matter: you walk in between the two
coils, having one of them on each side of you. If the circuit is sensitive enough, a piece of metal moved in
between them can disturb the resonance enough that it is detectable.
104
9.3 Traveling waves and standing waves
9.3.1 Traveling waves
Say we have a very simple equation, such as
y =
1
3
x
and we want to make it move towards the right, at a certain velocity v = 6.
All we have to do to make this happen is to replace x with x vt, so we have
y =
1
3
(x vt) =
x
3
2t
At t = 0, this simplies down to the original equation, At t > 0, the equation will be shifted, and if
graphed, it will have moved towards the right.
Replace the minus sign in x vt with a plus sign, and it will instead move towards the left, again with
velocity v.
Lets look at a dierent function: y = 2 sin(3x). This time, we have a proper wave.
On this plot, we measure the wavelength (Greek lowercase lambda) as the length of a full cycle, so the
time between one peak and the next, or one trough and the next (or any two similar points).
In this example, the wavelength is =
2
3
.
We also introduce the symbol k for the wave number, dened as k =
2

, so k = 3 for this case. So k is


then the number multiplying x in the sine function.
Now, can we make this wave move, in the same manner as we did with the line? Indeed, we do as we did
before: we replace x with x vt. For v = 6 m/s again, we nd
y = 2 sin(3(x 6t))
There is more information in this equation: the 2 multiplying the sine gives us the amplitude. The 3, the
wave number, tells us the wavelength (via k =
2

, so =
2
k
), and the 6 tells us the velocity. The minus
sign also tells us that it is travelling towards the right. Thus the equation is of the form
y = Asin(k(x vt))
If we attach a rope to a spinning wheel, we can create a traveling wave, propagating in the rope.
Say the wheel rotates with angular frequency . The period of one wave will as always be T =
2

. One
wavelength is as always the length between two peaks of the wave, or the distance traveled during one
105
period, which is simply = vT, which also equals =
v
f
, where f is the frequency in hertz (f =
1
T
, and
= 2f). Its then also true that f =
v

.
We can write the equation in a dierent form:
y = Asin(kx t)
Via the conversion formulas we just found, this too contains all the information. k gives the spatial in-
formation, such as . gives the angular frequency/frequency/period/velocity, either alone or combined
with k: v =

k
As an example, lets look at the equation
y = 4 sin(2(3x 9t))
If we multiply it out, we get
y = 4 sin(6x 18t)
... which tells us k = 6 and = 18, traveling towards the right.
Thus the amplitude is A = 4, the velocity v =

k
= 3 m/s. The wavelength is = vT = v
2

=
2
3
18 =

3
,
and the frequency f =

2
=
9

.
9.3.2 Standing waves
Say we have a wave
y
1
= y
0
sin(kx t)
which is the travelling towards the right. However, we also have a second wave, identical except for a plus
sign, which tells us it is travelling towards the left:
y
2
= y
0
sin(kx + t)
Via trigonometric identities, the sum of these is
y = y
1
+ y
2
= 2y
0
sin(kx) cos(t)
This is a standing wave. If we plot it at various values of t, we nd that some points, called nodes, are
completely xed: their y-value is xed. These points occur at multiples of

2
.
All points along the curve that are not nodes will move up and down, but not towards the left or towards
the right - note that there is no x t anywhere any more. So the sinusoid will essentially bob up and
down, while standing still.
We can make standing waves in a similar fashion to making traveling waves, but we need to get the fre-
quency of them just right.
Say we have a length of string, and we hold it xed at the ends, so the ends become nodes.
It is held by two people, and one of them is moving their node up and down slightly (which makes it not
a node, but with small movements, we can ignore that). When the wave reaches the end of the string, it
will reect back towards the rst person. When it comes back to the rst person, it will again reect, etc.
If the conditions are right, these reections can support each other, and the amplitude will increase. This
happens at a set of frequencies, which we call the resonance frequencies, in a similar fashion to RLC
106
circuits. However, in this case, there are in theory an innite amount of such frequencies.
For the lowest possible such frequency, which we call the fundamental (or rst harmonic), there will be
nodes at the ends of the string, but nowhere else. Thus the length of the string will be half the wavelength
- or, the wavelength will be twice the strings length.
For the next possible frequency, the second harmonic, the frequency is exactly twice that of the rst har-
monic. Here, there will be a node at the exact center of the string, which will not move at all. So now we
have three nodes: one at each end, plus one at the exact center.
For the third harmonic, the frequency is exactly three times the rst harmonics frequency, and there will
be another node, splitting the string into 3 parts that move, with a total of 4 nodes (one at each end, one
at 1/3 of the strings length, and one at 2/3 the strings length).
There are always n + 1 nodes for the nth harmonic, assuming we count nodes at the string ends.
The fundamental, denoted by
1
, is then 2L, where L is the length of the string, as we stated above.
The frequency of the fundamental, f
1
, is given by f
1
=
v

in general, so in this case, f


1
=
v
2L
.
For the second harmonic,
2
= L, which follows from f
2
= 2f
1
(or from drawing it out, in which case it is
immediately obvious).
In general, then,

n
=
2L
n
f
n
=
v

n
= nf
1
=
nv
2L
9.3.3 Musical instruments
When we excite a string in an instruments whether it is a string inside a piano being hit by a hammer,
a guitar string hit by a guitar pick, or a violin string hit by a bow it will be exposed to a whole range
of frequencies. However, it will tend to oscillate only in the resonant frequencies, which is why we can get
well-tuned sound out of such instruments.
So a ring that has a resonant frequency of 400 Hz, when struck, will tend to oscillate at 400 Hz, 800 Hz,
1200 Hz etc., all at the same time.
So for designing a stringed musical instrument, a key equation is
f
1
=
v
2L
v is decided by the strings material and tension:
v =

Tension
Mass/length
(proved in the next MIT physics course, 8.03 Vibrations and Waves, not here).
A regular steel-stringed guitar usually has roughly the same tension for all strings, something on the order
of 7-10 kg-force of tension per string for a 6-string guitar in standard tuning, with standard string gauges.
The dierence between the strings is then mostly in their material - indeed, they usually all dier in thick-
ness, from 0.010" for the thinnest E string, to 0.046" for the thickest. (Dierent guitarists have dierent
tastes; 0.009"-0.0042" is about the thinnest anyone uses, and they can go up to about 0.013"-0.056", still
107
in standard tuning, which then causes much more tension).
The tension is what we adjust when we tune a stringed instruments - we wind the string either more (more
tension) or less, until it resonates at the (fundamental) frequency we want.
When we play the instrument, we change the strings length, by shortening it with our ngers.
On fretted instruments, like guitars and electric basses, the string is shortened by a metal bar (the fret)
going across the neck; there are usually 17-24 such frets on a guitar.
The instruments sound is then generated as the string vibrates, which pushes and pulls on the surrounding
air, creating pressure waves in the air. Those pressure waves eventually reach our ear drums, and causes
the ear drums to vibrate at the same frequencies. That is then interpreted as sound by the ear and our
brains.
Lets now discuss wind instruments.
Say we have a completely closed box, of length L. At the left end of the box, we put a loudspeaker, which
generates a certain sound frequency.
We can now get resonance, and a standing wave, in the air inside the box. The frequencies are goverened
by the same equation as for the stringed instruments, except that v is now xed: it is the speed of sound
in the air inside the box.
We cant do much about the temperature - but if it does change, the fundamental of the instrument will
also change. This can be compensated for by adjusting L to cancel out the error, if applicable to the
instrument.
Because of the equation that governs them (above), the smaller L is, the higher the fundamental frequency.
Therefore a trombone, which is large, has a lower frequency than a trumpet, which is smaller. A tuba has
an even lower frequency sound.
For a ute, if you cover all the drilled holes on it, the resonant chamber will be long, and the frequency
will be low. If you take your ngers o the holes, so that they are open, the frequency will go up.
108
Chapter 10
Week 10
10.1 Resonance, electromagnetic waves
The lecture begins with talk about mechanical resonance, without much math or things to take notes of.
It is demonstrated how playing a loud tone at a wine glasss resonance frequency can shatter the glass.
After that, a movie is shown about the Tacoma Narrows Bridge, that collapsed only 4 months after its
construction in 1940.
The bridge resonated with the wind, such that huge waves went along it. On the day of its collapse, it
went into a twisting motion that proved too much for it to handle, and the bridge collapsed. I would
recommend looking up footage of this on the internet, as the collapse was actually lmed.
Lets now turn to electromagnetic waves. Two of Maxwells equations are extra important here:
_

E

d =
d
B
dt
(Faradays law)
_

B

d =
0

m
_
I +
0

d
E
dt
_
(Ampere-Maxwell law)
At least to begin with, we will introduce EM waves without the rigorous dierential equation treatment.
As the above equations show, we can get an electric eld due to a changing magnetic eld (Faradays law).
Not only that, but we can also get a magnetic eld due to a changing electric eld (the Ampere-Maxwell
law).
One possible EM wave that meets all four of Maxwells equations is

E = E
0
x cos(kz t)

B = B
0
y cos(kz t)
Thus, the electric eld is solely in the x direction, the magnetic eld solely in the y direction, while the
wave propagates in the (positive) z direction, as seen by the term inside the cosine. Furthermore, we see
that they have the same frequency/wavelength, and that they are in phase with each other. That is, when
E = 0, B = 0, and when E = E
0
(the maximum value), B = B
0
, etc.
The wave propagates with v =

k
= c, with c being the symbol for the speed of light (in a vacuum), about
3 10
8
m/s (to within 0.07%!).
The equations above are for a plane wave. If we take a plane perpendicular to z, at a single moment in
time, the E and B vectors are the same everywhere in that plane. They satisfy Maxwells equations only
if
B
0
=
E
0
c
109
and
c =
1

0
Now, lets move into some math.
At t = 0, we draw a loop in the y-z plane. Due to the nature of a plane wave, the eld is constant along
one value of y, so the marked strip with width dz has a constant eld along one side. If we assume dz is
very small, we can consider the eld constant through its width.
We mark the width (in the z dimension) as /4 as it will simplify the result we get.
We attach a at surface to the loop, bounded by /4, and the y and z axes, and apply Amperes law to
it:
_

B

d =
0

0
d
E
dt
Because we are in a vacuum,
m
= = 1, and I = 0 as there can be no conduction current through a
vacuum.
First, what is the ux through one such slice dz? Flux is the dot product of E-eld strength dot area, so
the ux is
d
E
= dzE
0
cos(kz t)
if we choose dir

dA = x, i.e.

dA points upwards.
We need to nd the ux through the entire surface, though, so we integrate the above:

E
=
_
/4
0
dzE
0
cos(kz t) = E
0
_
/4
0
cos(kz t)dz
We could simply integrate this, take the time derivative, and have our answer - but doing that in math
software yields a relatively complex answer, as we need to substitute a few values to make the answer
(very) simple.
d
E
dt
= E
0
d
dt
_
_
/4
0
cos(kz t)dz
_
We can dierentiate this inside the integral, instead of rst calculating the integral and then taking the
derivative of what we nd; this will make the calculation involve fewer terms.
When we take the time derivative of cos(kz t), a term comes outside. Since is a constant, we
move it outside the integral, too. The cosine turns into sin inside the integral; we then substitute for
t = 0, so that the integral turns into
_
/4
0
sin(kz). Combining these steps, we get
110
d
E
dt
= E
0

_
/4
0
sin(kz)dz
= E
0

_
/4
0
sin(kz)dz (cancelling minus signs)
= E
0


2
We get this result after calculating the integral, and then making the substitution k =
2

(we could make


the opposite substitution as well). We can simplify it further:
d
E
dt
= E
0


2
= E
0

= E
0
f
= E
0
v
= E
0
c
The second line is simply a rewrite with no actual change. In the third, we use f =
2

. In the fourth, we
use v = f, and in the last, the velocity v of EM radiation is c, the speed of light (or perhaps rather the
speed of EM radiation!) in a vacuum.
So we nally have the change in electric ux, and we now need to do the closed loop integral of the
Ampere-Maxwell law. With the result we just found, we now have
_

B

d =
0

0
E
0
c
Keep in mind that

B is strictly in the + y direction, i.e. out of the blackboard. If we do the closed loop
integral, counterclockwise as seen from above, starting at the outer edge, we nd that for the outer edge,

B and

d are perpendicular; one is in + y and one in + z, so that contributes nothing to the dot product.
For the rightmost side, the B-eld is zero everywhere along the line, by the denition of the plane wave.
(It is zero for all y along that line.)
For the short inner portion, they are again perpendicular, so that part doesnt contribute.
The only part that does contribute is the part along the y axis, with length . The eld is constant along
111
that line - again, denition of a plane wave, so that part simply contributes B
0
. So with the integral
evaluating to that simple value, we nd
B
0
=
0

0
E
0
c
B
0
=
0

0
E
0
c
The cancels, and we nd a relation between the two eld strengths.
The following properties are always true of traveling EM waves:

E v

B v

E

B

E and

B are in phase

E

B = v

B =
E
0
c
c =
1

0
The second-last one does not contradict our nding above, because of the last one. We can solve the two
as a system of equations and will nd the value for c shown, which means that the two equations are really
equivalent.
We know that electromagnetic waves travel at (propagate at) c 3 10
8
m/s, we can then calculate some
rough times required to move certain distances.
One foot takes about a nanosecond. 30 meters, the length of the lecture hall where the lecture is recorded,
about 0.1 microsecond. In one second, it can move all the way to the moon, and about eight minutes to
the sun. That is, it takes light 8 minutes to move between the surface of the sun and the Earth.
The closest star (except for the sun, of course!), Proxima Centauri, is about 4.2 lightyears away, so light
would (by the denition of a lightyear) take light 4.2 years to travel to us from there.
The nearest large galaxy, the Andromeda galaxy, is about 2.5 million lightyears away. When we look at
it in the sky, we see it as it were 2.5 million years ago!
Since one year is (in the denition used) 365.25 days (on average), one year is 365.25 24 60 60 = 31557600
seconds, so multiplying that by the speed of light, we nd that one light year is about 9.46 10
15
meters.
10.1.1 Radar and measuring distance
We can measure distances by measuring how long it takes for light to reect o a target.
We sent a pulse, and begin measuring the time taken. When the reection returns, we stop measuring,
and calculate the distance as
2d = ct
d =
ct
2
where t is the measured time, and c is as usual the speed of light. We have to divide by two since the light
has travelled both there and back, and we only want the one-way distance.
We can measure the distance to the moon in this way, by sending focused laser pulses. Astronauts from
both the US Apollo missions and the Soviets have left corner reectors / retroreectors on the moon.
112
These devices reect light back in the exact direction it came from - as opposed to a mirror/mirror-like
surface, where the angle would be of the same magnitude, but the reection would only come back if you
looked at it exactly straight on.
10.1.2 Radio
Now to something slightly dierent: radio, as in both radio waves and radio stations. An AM radio station
might transmit somewhere in the 1000-1000 kHz range (as an order of magnitude). How is audible sound,
<20 kHz, transmitted that way?
Well, we use AM: amplitude modulation. First, we generate a carrier wave, a high-frequency sinusoid.
The frequency of the carrier wave is the frequency a listener will have to tune in to.
We then modulate that signal, by changing its amplitude, by essentially simply multiplying the two
frequencies. Thus we will end up with a signal that is of the (simplied, not truly accurate) form
y = sin(210
6
t) cos(210
3
t)
To listen to such a station, we tune an RLC circuit in our radio to the carrier frequency, such that it is at
resonance only at the frequency we want to listen to. (We need a high-Q / low-bandwidth resonator for
this, or we might pick out nearby stations as well!)
After that, we need to demodulate the signal, to get only the envelope, which is what was the signal to
begin with (at the radio station).
10.2 Index of refraction and Poynting vector
10.2.1 Poynting vector
Lets now look at the energy content of electromagnetic waves. We have seen earlier that electric elds
and magnetic elds both have an associated energy density, given by
U
E
=
1
2

0
E
2
J/m
3
U
B
=
1
2

0
B
2
=
1
2
0
E
2
c
2
J/m
3
The latter follows from the equivalence B
0
=
E
0
c
that we found earlier.
However,
c =
1

0
c
2
=
1

0
Therefore,
U
B
=
1
2
0
E
2

0
=
1
2

0
E
2
J/m
3
So the energy denssity of the B-eld is exactly the same as the energy density of the E-eld in EM waves!
The total energy density is then the sum, so we just get rid of the one-half, ane we nd
U
total
=
0
E
2
=
0
EBc
where we use that E = Bc (via B =
E
c
).
113
Now that we know that EM waves carry some energy, we can ask the question: if an EM wave passes
through a surface, how much energy passes through one square meter per second? That is, what is the
energy ux, in watts per square meter (joules per second per square meter)?
Imagine a square surface, of side one meter, perpendicular in space to the waves propagation. Thus, the
dot product works out to not have the cosine term, as that would equal one (cos = cos 0 = 1).
Since light travels a distance of c meters in one second, we can calculate the energy stored in a box that
goes back such a distance from the square we are interested in. We know that all energy inside this box
will pass through the square, so we calculate the total energy by multiplying the energy density by c 1 1
cubic meters, and nd
U
total
c =
0
EBc
2
Because c
2
=
1

0
, that is equivalent to
U
total
c =
EB

0
The unit of this is then (as noted above) watts per square meter, or
J
m
2
s
.
What we have found is called the Poynting vector, named after its inventor (though it was simultanesously
created by others, as well). The pronunciation is very close to pointing vector (many likely say it exactly
like that), perhaps with a bit more of the y sound. It is generally written as

S =

E

B

E and

B are always perpendicular in a traveling EM wave, so the cross product can be seen as being there
only to provide the direction; the magnitude is always [

S[ =
EB

0
.

E and

B are not constant in value, however! They are sinusoidal in nature, and so the Poynting vector
must be described by a sin
2
or cos
2
function.
We can think about its time averaged value, which may be more useful, since it varies constantly (and
usually with extremely high frequency - visible light is in the 10
14
Hz range, for example!).
The time average of a sin
2
or cos
2
is one half the maximum, so we nd
S =
1
2
E
0
B
0

0
=
1
2
E
0
2

0
c
10.2.2 Waves due to accelerating charges
Now that we have learned a bit about plane waves, lets look at some other types. Plane waves are useful
for modelling some situations, but far from all.
Given the equation for a plane wave, we can calculate its value for any time, past or present, at any point
in space, no matter how distant. Thats not too realistic.
Lets look at the electric eld lines of an accelerating point charge.
114
To the left, we have a few point lines of a point charge (postive or negative; we cant tell from the lines
alone, and it doesnt matter.)
To the right, inside the circle, we have one point charge, at two points it time: t = 0 and t = t.
The circle - really a sphere in three dimensions - has a radius of ct, which is the distance light can travel
in the time that has passed.
Just outside the circle/sphere, we draw the eld lines as they are at time t = 0 (in white), radially outwards
from the point charge in the center.
In pink, we have the new eld lines of the charge, at its new location, at t = t.
Because these are the same eld lines, only at a dierent location, they must be connected to the old eld
lines. Information about the change in location cannot have propagated to outside the sphere, because
light travels at a nite velocity. Therefore, the old and new eld lines must be connected, which we
draw as the green kinks.
Note that the eld line on the direction of movement is straight and without any kinks.
At this point in the lecture, there is a video demonstration of this, with moving/accelerating charges,
which really must be seen in video.
So similarly to what we have in the picture above, and the video, we can look at an antenna, with an AC
current going through it. Clearly, if the charges move back and forth, they need to start and stop, too,
which means they must accelerate (both positive and negative acceleration, i.e. deceleration).
Along the direction of the current, there will be no EM waves produced, for the reason shown above and
in the video. In the plane perpendicular to the current, the eld strength will be at a maximum, and for
angles in between, the value will be somewhere between zero and the maximum.
10.2.3 Spherical waves and the Poynting vector
Lets return to the Poynting vector.
If assume that the Sun has a symmetric, radial power output of 3 10
26
watts, then we can calculate a
rough number for the amount of received power per square meter, at the surface of the Earth, 150 million
km away:
p = 3 10
26
W
1
4(150 10
9
m)
2
1061 W/m
2
1 kW/m
2
This is modelled as a spherical wave - a plane wave has a constant value, as the energy is not spreading
out, but moving together as a plane.
We can think of spherical waves as emanating from a point source, such that the intensity falls o as
1
r
2
.
The greater the distance (the 150 10
9
m here) is, the greater the surface area of the sphere where the
energy ux density is calculated, and the lower the energy ux density is. (Of course, if we could surround
115
the entire sun, then we would capture the entire output regardless of the radius of our sphere, just that
the energy ux per square meter of the sphere would be lower for greater radii.)
If we instead think of this as a plane wave, we can calculate the equivalent electric eld:
1000 =
E
0
2

0
c
E
0
613 V/m
(the lecture says 870 V/m, but Im not sure how that number is found!)
Regardless of the exact number, what does this represent? Well, its some equivalent E-eld value, that
would produce this energy ux if it were a plane wave. Is that useful? Perhaps, perhaps not, depending
on what we are trying to do.
We can often choose a model to use - plane waves, spherical waves, individual photons - to explain some
phenomenon.
10.2.4 Photons and radiation pressure
Next, lets have a look at photons. Photons are individual light particles - quanta of light. Each photon
has a certain frequency, and a certain energy (the energy is proportional to the frequency).
Because photons have energy, they have momentum - despite having no mass!
In classical mechanics, the magnitude of momentum is given by p = mv, which clearly must be 0 for a
massless particle.
The answer comes from Einsteins relativity, and the worlds most famous equation... except the famous
form is simplied:
E = mc
2
This is only a special case of the full version:
E
2
= p
2
c
2
+ m
2
c
4
Thus, for a photon, with mass m = 0, we nd
E
2
= p
2
c
2
E = pc
p =
E
c
(E = mc
2
is the special case where p = 0, for a body it its rest frame.)
So we see that the momentum of a photon is given by the energy it carries, divided by the speed of light.
But now, something else happens: the transfer of momentum is what we call force (

F =
d p
dt
- Newtons
second law)!
So say we have a square meter, perpendicular to some incoming EM radiation (at a wavelength that would
be absorbed by the surface). Using the Poynting vector, we have
S =
energy
m
2
s
If we divide both sides by c, we nd
S
c
=
energy
c m
2
s
We nd that E/c term again, which is the momentum. So this can be rewritten as
116
S
c
=
momentum
m
2
s
The units of this is then momentum per unit time, which is force, per square meter, which is pressure.
We call this pressure radiation pressure. So what this is saying is that for a given area, with a certain
amount of incoming EM radiation perpendicular to it, there will be a force proportional to the area, and
proportional to the energy contained in each photon, pushing this surface backwards (in the same direc-
tion as the radiation wave is propagating).
Keeping in mind that the time-average value of the Poynting vector is what we are really interested in, we
can write the pressure as
p =
S
c

where
=
_

_
0, if the material is entirely transparent to the radiation,
1, if 100% of the radiation is absorbed,
2, if 100% of the radiation is reected
Of course, values in between are also possible.
The reason for these values can be seen quite easily from the conservation of momentum.
The professor used the examples of throwing tomatoes earlier in the lecture, so we will use that mechanical
analogue.
Since classical momentum is the product of mass and velocity, a (very large!) tomato of mass 1 kg, moving
at 5 m/s, will have a momentum of 5 newton-seconds. If one such tomato hits a target per second, and
falls straight down after hitting, a force of 5 newtons will be applied to the target.
If the tomato were instead somehow bouncing back, in a perfectly elastic collision, then the force would
be twice that: rst the momentum is absorbed by the target, and then it must add the same magnitude
of momentum back in the opposite direction, so the transfer of momentum is twice as large.
In between the lecture segments, we have the following question:
What is the radiation pressure on a spherical iron asteroid of approximate radius 100 m? The asteroid is
at a distance of about 400 million km from the sun. Assume that all radiation is absorbed. Again assume
that the power output of the Sun is 3 10
26
W as shown in the lecture (note that the actual measured
value is closer to 3.8 10
26
W).
We take the Suns power output to be spherically symmetric, so the energy ux density at the distance is
S = 3 10
26

1
4(400 10
9
m)
2
149.2 W/m
2
We divide that by c, and nd the radiation pressure:
p
149.2
3 10
8
5 10
7
N/m
2
The pressure will only be on half the asteroid, as only one side is facing the Sun.
While we will nd a half-decent approximation by multiplying the area facing the sun (2r
2
) with the
pressure, we really need to consider that parts of that area are almost entirely parallel to the pressure,
while the area near the asteroids center will be mostly perpendicular.
F = r
2
p
We can then calculate the force due to the radiation pressure:
117
F = (5 10
7
)100
2
0.0157 0.02 N
(Im not a fan of the rounding, but apparently this is the way to calculate it; I did it dierently to begin
with, which found exactly 0.02, but that method was incorrect!)
Thats a very small force indeed! We can compare it with the gravitational force between the asteroid and
the sun. The Suns mass is 2 10
30
kg, and the density of iron 7.87 g/cm
3
(= 7870 kg/m
3
) (as given in the
question).
Gravitational force is given by
F
g
= G
m
1
m
2
r
2
with the m variables being the masses, and G the gravitational constant, G 6.67 10
11
, and r the
distance between the objects. We nd
F
g
6.67 10
11

(2 10
30
)(7870 4/3 100
3
)
(400 10
9
)
2
2.8 10
7
N
Well, that force is about about 1.4 billion times stronger than the force due to radiation pressure! Radiation
pressure thus has a very small eect, but one that still cannot be neglected! Interplanetary spacecraft can
veer o course by hundreds or thousands of kilometers if radiation pressure is neglected.
10.2.5 Polarization
Lets look at the electric eld due to an oscillating charge, moving back and forth with frequency , at
the center of the coordinate system above.
Say we are at point P, with the position vector r to the origin, and angle from the z axis.
Knowing some rules will allow us to nd the E-eld at point P.
First o,

E is always perpendicular to r.
Second, a, r and

E are always located in one plane.
If we double the charge, or double the acceleration, the E-eld strength will double. That is
E qa
As stated earlier, however, no EM radiation will go out in the direction of accerelation ( = 0), and the
maximum will go out perpendicular to that direction ( =

2
). We can show this by adding a sin term
to the proportionaly equation.
The E-eld strength is also inversely proportional to the distance r, so we nd
118
E
qa sin
r
Because E is also proportional to B, and because the Poynting vector

S is proportional to the product of

E and

B,
S
q
2
a
2
sin
2

r
2
For the same reasons as the suns energy ux density falls of as
1
r
2
, this must do so, too, or conservation
of energy would be violated.
Here is a transparency sheet showing linear polarization:
At the center, we have our oscillating point charge, constantly accelerating up and down.
We see that there is no EM waves going out upwards, since that is in the same direction that the charge
is oscillating.
To the middle-left, bottom-left, middle-right and bottom-right of the image, we are in the plane perpen-
dicular to the oscillation, and so we see a maximum of the EM radiation here.
In the top right of the image, we see that for

4
, the magnitude of the E-vector is smaller, but still
not zero as it is at the top ( = 0).
We also see that a, r and

E are all in one plane (to see this, rst look at the leftmost or rightmost set of
r and

E, and then the angled one).
We call this linearly polarized radiation. The electric eld and the magnetic are in phase, as this is a plane
wave.
If instead we had two perpendicular plane waves of equal amplitude, but with a 90 degree dierence in
phase, we would call that circular polarization.
The ending of the lecture (lecture 28) should really be seen, as it clearly demonstrates polarization, and
explains and shows it much, much better than I could hope to do using only text and a few images here
119
and there.
(Of course, these notes are not intended to ever be a replacement for the lectures! They are incomplete, may
contain some errors, and are ridiculously unlikely to explain things as well as a world-famous professor!)
10.3 Snells law, refraction and total reection
Say we have a water surface, surrounded by air, and a light beam hits that surface, as seen here:
Some light will be reected, and turn back up into the air. Some light will be refracted, and enter the
water.
The incident light (the incoming light), the reected light and the refracted light all lie in one plane.
For the reections,
1
=
3
, where the angles are measured with respect to the normal to the surface, as
shown. That is, the reection will exit at the same angle as it entered, something we all see intuitively
from using mirrors.
Willebrord Snellius introducted the concept of index of refraction, or n.
A materials index of refraction is the ratio between the speed of light in a vacuum, to the speed of light
in the material. Thus, if light travels at v = 0.5c in a material, that materials index of refraction is 2.
n =
c
v
For this reason, n 1 for all regular materials.
The index of refraction of vacuum is, by denition, one. For air, it is extremely close to 1, and is therefore
usually treated as 1. Waters index of refraction is about 1.3, and glasss about 1.5.
Snells law states that
sin
1
sin
2
=
n
2
n
1
Or, equivalently, and perhaps easier to remember,
n
1
sin
1
= n
2
sin
2
120
Thus, the ratio of the materials indices of refractions is related to the ratio of the sines of the angles.
Alternatively, we can think of it as a ratio of the speed of light in the material, in which case we nd
sin
1
sin
2
=
v
1
v
2
Snells law appears to have been rst discovered by Muslim scholar Ibn Sahl, in 974 AD, 637 years prior
to Snellius! The western name remains Snells law, however.
Using Snells law - which remains its name - we can calculate the angle
2
given the materials refraction
indices and the angle of incidence
1
.
As an very quick example, lets look at the case above, going from air to water. Say the angle of incidence,

1
, is 60 degrees. n
1
= 1 and n
2
= 1.3, so
sin /3
sin
2
= 1.3
sin
2
=
sin /3
1.3
sin
_

3
_
=

3
2
, so

2
= arcsin

3
2 1.3
41.77

10.3.1 Total internal reection


Lets then look at the reverse situation:
Here, the light ray is going from water, to air.
The lecture question asks, for n
1
= 1.331 (n
2
= 1, as it is air), how large must
1
be for
2
to equal 90
degrees?
We set up Snells law again:
sin
1
sin /2
=
1
1.331
sin
1
=
1
1.331

1
= arcsin
_
1
1.331
_
48.70

121
If the angle is greater than that, nature cannot deal with that refraction, and we get total reection: all
the light is reected, back into the water. Total reection happens when the angle is greater than a critical
angle, that is
sin
c
r =
n
2
n
1
, if n
1
> n
2
So in this case, the critical angle is 48.7

. If
1
is greater than that, there would be no refraction, only
reection.
This principle (total internal reection) has some interesting uses. Perhaps most signicantly, it is the
principle behind optical ber, used (among other things) to create high-speed data links between comput-
ers, cities, and even countries and continents.
Such a ber is made from a continuous piece of a form of glass or plastic. Light enters one end, runs in
a straight line until it hits the outside, which is manufactured to have a lower refractive index than the
inside. Therefore, as long as the angle is greater than the critical angle, total internal reection causes the
light to keep bouncing away inside the ber.
This process is repeated over and over and over, until it reaches the end of the ber.
Materials research have made modern optical ber extremely low-loss: a single unbroken piece of ber can
be used to communicate over a distance of several hundred kilometers without any elecrtrical equipment
to strengthen the signal.
When a surface is smooth, parallel incoming light beams will reect with equal angles (
3
=
1
), and thus
exit in parallel; we call this a specular reection. Shining a laser pointer on a mirror is a great example of
a specular reection.
If the surface is rough, incoming light may be reected in essentially random directions, and the light
will not exit with the same angle, but exit in all sorts of directions. We call this a diuse reection; one
example might be a laser pointer on a wooden table - or on most non-shiny surfaces, really.
We will use the term reection to mean specular reection, as those are the kinds that can be easily
calculated, without knowing near-atomic-scale details of surfaces.
If a light ray enters a material, and then exits again, the new twice-refracted light ray will always be
parallel with the incident ray. It will not be on the straight line, however, unless n
1
= n
2
.
10.3.2 Frequency and wavelength in refraction
As light enters a medium, its frequency stays constant, but its wavelength changes. The relationship
v = f
must always hold, but because v changes upon entering a dierent material, one of the other two must
also decrease. The frequency cannot change, imagine a wave passing two observers inside two dierent
material; point A is just before the boundary between the two materials, and point B just after. If
the waves frequency diers between the two points, that would imply that energy were piling up or
disappearing around the boundary.
Thus, the wavelength has to change. The wavelengths follow the relationship
v
1
=
1
f
v
2
=
2
f

2
=
v
1
v
2
=
n
2
n
1
We saw previously that the speed of light is given by
122
c =
1

0
As this result uses only two results, the (electric) permittivity of the vaccuum
0
, and the (magnetic)
permeability of the vacuum
0
, it would make sense that if we instead used a materials permittivity and
permeability values, we would get the answer for the speed of light inside that medium. And, indeed, we
do. That is:
v =
1

M
We can also write this as
v =
c

M
... which tells us that because
v =
c
n
n =

M
Thus, we can calculate the index of refraction for any material for which we know its dielectric constant
and relative permeability.
Both of these two values are functions of frequency. Of course,
M
1 for most materials (paramagnetic
and diamagnetic materials), so that doesnt have a huge eect on the refractive index.
That they depend on frequency makes intuitive sense, when we think about what they represent.
The dielectric constant tells us how strongly dipoles in a material align to cancel out the external eld.
However, they only align due to the external eld, and so when the external eld alternates back and
forth extremely quickly, the dipoles dont quite have time to align. Therefore, decreases as frequency
increases.
The same argument can be applied to the relative permeability.
As an example, for water,
M
1 at all frequencies. However, varies: at near-zero frequencies, it is
about 79; at 10
8
Hz (100 MHz) it is almost the same, at about 78. At the frequencies of visible light,
about 5 10
8
however, it has dropped drastically to around 1.33 or so (depending on the exact wavelength,
as well see next).
10.3.3 Dispersion, prisms and white light
Due to the eect discussed above, the index of refraction depends on the wavelength of the light.
In water, the index of refraction for red light is about n
red
= 1.331, while that for blue light is about
n
blue
= 1.343; so we see that red light is about 0.9% faster than blue light in water.
We call this eect dispersion.
What we perceive as white is a combination of others colors. Therefore, we can use a prism to split white
light into its constituent colors, using dispersion. Because the index of refraction is dierent for each color
(or each wavelength/frequency), the colors will split up as soon as they enter the prism. Because the
side they exit the prism on will not be parallel with the side they enter it on, the refraction will not be
cancelled, but rather enhanced further, i.e. the dierent colors are separated further.
So if we shine white light (combining all wavelengths) into a prism, we will see the visible spectrum of
colors on the other side. In order of decreasing wavelength: red, orange, yellow, green, blue and violet.
These are the colors of the rainbow, and are also often given as red, orange, yellow, green, blue, indigo
and violet (Roy G. Biv as a mnemonic - think of it as a name).
123
Say that we are in a room lit by a regular, incandescent light bulb. The light will be comprised on various
frequencies, and will be roughly white (more yellow-reddish, but still comprising many colors).
Without this wide-spectrum light, we could not see colors properly. In a room lit exclusively by a
monochromic (single-color) red, we could not perceive green or blue, only shades of red.
Therefore, we always design general-purpose light bulbs to produce many colors of light.
(You may have seen the yellow-orange-ish light produced by sodium-vapor street bulbs, and how every-
thing looks orange-y in such light.)
10.3.4 Primary colors
The human eye has three types of cone cells, that respond to dierent wavelengths of light (they each have
a wavelength that is the most likely to cause a response). One type responds best to short wavelengths
(blue), one to medium wavelengths (green) and one to long wavelengths (green-yellow); the latter two have
signicant overlap.
Because of these three cone cell types, we can create almost any color of light (to a human observer) by
mixing red, green and blue light in the correct proportions.
Those three colors are then known as primary colors.
Computer graphics (and televisions, etc.) are based around this concept. A modern LCD display works
by mixing these three colors, as in this example:
124
(Photo of LCD screen by Mattia Luigi Nappi, licensed under CC-BY-SA 3.0)
The top picture shows a zoomed-out version of a picture of an LCD screen (displaying a picture of a man
in a tie). The bottom picture is the full-sized version, displaying the same pixels ase the area marked in
white.
Its clear here that each pixel (each dot on the screen) is made up by three subpixels: one red, one green
and one blue.
Other display technologies exist, but all combine these colors in one way or another, to create an image
with millions (16.7 million for 8 bits per color channel) to billions (1.07 billion colors for 10 bits per color
channel) of colors, from having just varying intensities of red, green and blue arranged in some pattern.
125
Chapter 11
Week 11
11.1 Polarizers and and Maluss law
This lecture will focus entirely on the polarization of visible light.
The light from the sun or light from a light bulb is not polarized.
If we think about individual photons as plane waves, each photon will have a well-dened linear polariza-
tion.
A collection of multiple photons may then look like the above, where each photon is coming straight out
of the blackboard, with the lines representing the direction of oscillation of the E-eld.
Some are horizontal, some vertical, some at angles... and there is no preferred direction if we average over
time.
In 1938, Edwin Land invented a material that can polarize such light. Consider the case of a diagonally
polarized photon entering the polarizer, when the polarizer is held such that it polarizes vertically:
The resulting polarized photon will have a magnitude that is equal to the vertical component of the incom-
ing photon, in other words, E
0
cos , where is the angle between the polarizer direction and the photon,
126
as shown.
We can nd light intensity (in watts per square meter) using the Poynting vector, which is proportional
to E
2
- or E B, where B is proportional to E.
Because of this square dependence, the intensity is reduced by cos
2
.
In order to nd the intensity change over the average of all incoming photons, we take an average of this,
over all possible angles. The average of a cosine squared function is one half the maximum, as weve seen
before. Thus, in a best-case theoretical scenario, we halve the intensity, but in return get 100% linearly
polarized radiation, when inputting entirely unpolarized radiation.
If we had such a polarizer, it would be an HN50 polarizer - meaning 50% of the light gets through and
becomes polarized.
In reality, we can only get smaller numbers than that, such as HN25, HN40, etc, for polarizers where 25%
or 40% of the incoming light gets through, respectively.
Lets try to apply that:
Linearly polarized light of intensity I
0
passes through an HN25 linear polarizer. An HN25 polarizer lets
through 1/2 as much light as an ideal HN50 polarizer which has no attenuation at all. The angle between
the E-vector of the incoming light and the E-vector of the outgoing light that emerges from the other side
of the HN25 polarizer (i.e. the polarization direction of the HN25) is 30

. What is the intensity of the


light that emerges?
Lets see. We have
I = I
0
cos
2

... but thats for the ideal case of the HN50 polarizer. We need to further halve the result, so the answer
is
I =
1
2
I
0
cos
2
=
I
0
2
3
4
= 0.375I
0
This result, that the intensity is reduced by cos
2
, is known as Malus law.
Two things should be noted here, however: one is that not 100% of the light exiting a polarizer will be
polarized. The ratio of the transmission of the unwanted components to the wanted is called the extinction
ratio, which is about 1:500 for Polaroid lm, but can be 1 : 10
6
for certain polarizers.
The second is that the way we talk about individual photons essentially losing energy through a polarizer
is not true.
Blue light has a higher frequency (shorter wavelength) than red light. Violet is the shortest wavelength
we can see; ultraviolet is shorter. On the other side of the scale, red is the longest wavelength we can see,
and longer wavelengths are infrared.
A photon either comes through a polarizer, or it doesnt. If it did come through with less energy, its
wavelength would have shifted to become longer (as wavelength and photon energy are related via c), and
the color of the photon would have become redder. This is not the case.
As usual with small things, quantum mechanics has the correct answer. Malus law does hold, however,
even through our derivation was a bit of a cheat.
11.1.1 Polarization by reection
We can produce polarized light without a polarizer per se, by relecting it o a dielectric, like glass or
water.
Unfortunately, the full derivation (using Maxwells equations) isnt shown, and instead its mentioned that
127
the derivation will be part of MITs 8.03 course (Vibrations and Waves).
Lets have a look:
In order to do this, we need to decompose the vectors into two components: one perpendicular to the
blackboard (in to/out of the blackboard), and one parallel to the blackboard. Both are perpendicular to
the rays propagation, however, which is a must for traveling EM waves.
Apart from the decomposition, there is nothing new going on here: we have a reection, where the reec-
tion angle equals the angle of incidence, and a refraction, for which we can use Snells law.
The incident light, reected light and refracted light all lie in one plane (the plane of the blackboard, in
this case): we call this the incident plane, or plane of incidence.
The incident light is not polarized in our example. That implies that that the perpendicular and parallel
components of the incident light must be equal, or else the light would have a preferred direction, and
thus not be entirely unpolarized.
Using Maxwells equations, in the derivation not shown, we can relate the parallel components of these
beams. We get two relations, and two equations: one that relates the parallel components of the incident
light and the reected light, and another that does that same for the parallell components of the incident
light and the refracted light.
We can do the same with the perpendicular components, which gives another two equations, for a total of
four.
Looking at those four (unmentioned) equations, it can be seen that while the parallel and perpendicular
components of the incident light must be equal, that will no longer be the case for the reected light, nor
for the refracted light. In other words, they will be partially polarized.
One of the four equations, for the parallel components of the incident and reected light, is:
E
0refl
= E
0inc
n
1
cos
2
n
2
cos
1
n
1
cos
2
+ n
2
cos
1
We can simplify this using Snells law (again, derivation not shown):
E
0refl
= E
0inc
tan(
1

2
)
tan(
1
+
2
)
Remember, however, that the light intensity is related to the Poynting vector, which in turn is proportional
to E
0
2
, while the above gives the value for E
0
; we need to square the above to get a relation useful for
128
light intensity.
In the equation above, we can see that as
1
+
2
goes to 90 degrees, we get the tangent of 90 degrees,
which in innite in magnitude. That means the parallel component goes to zero - and so if any light exits,
then clearly it must be 100% polarized in the penpendicular direction!
If
1
+
2
is 90 degrees, via trigonometry,
sin
2
= cos
1
So in this special case, we can replace all sin
2
s with cos
1
s. That means we can write down Snells law:
sin
1
sin
2
=
n
2
n
1
=
sin
1
cos
1
=
tan
1
So when the condition is met that
tan
1
=
n
2
n
1
we get 100% polarized reections (or possibly no reections at all). The angle for which this happens is
then

1
= arctan
_
n
2
n
1
_
This is known as the Brewster angle (or Brewsters angle), after its discoverer.
This has another interesting meaning: when 100% polarized light hits this surface - polarized in the correct
direction, of course - there will be no reection whatsoever, and the light is perfectly transmitted through
the surface.
The above relation only holds for dielectrics only, not for conductors. Light that reects o metals will
act dierently.
11.1.2 Polarization by scattering
We have now seen two ways of polarizing light: using polaroid lm - devices created to be polarizers - and
by reection.
There are many ways light can be come polarized, and by scattering is another.
When light shines through very ne particles (much smaller than 1/10th of a micron), the light exiting in
one plane will become 100% polarized; namely in the plane where all photons exit at 90 degrees compared
with the angle at which they enter.
Say a single photon comes in from the left, towards the right, polarized in the up/down direction, all in
the plane of the blackboard.
(We talk about a single photon to begin with, however the light as a whole is unpolarized, with individual
photons polarized in all possible directions.)
The photon enters some ne dust, and so charged particles in the dust (electrons, protons) will experience
an acceleration in the same direction as the photons polarization, given by
a =

F
m
=
q

E
m
... where q

E is the electric force on a charged particle.


Electrons, having a much, much lower mass than protons, will experience a much greater acceleration.
129
The charges accelerating and oscillilating up and down to the left (in the picture below), will produce an
oscillating electric eld at a point P, that is perpendicular to the position vector between them, and in
the same plane as r and a:
There is only one possible direction for those two constraints combined, and it is drawn above.
The photon enters the dust, is absorbed by the dust, and another photon is then re-emitted at exactly the
same frequency. Why?
Well, the incident photon, with angular frequency , causes an acceleration in charged particles, again
with that same frequency . Accelerating charges cause EM radiation, and the frequency of the radiation
will be the frequency of acceleration, and so the re-emitted photon will have the same frequency as the
incident one.
The direction of the new photon will have changed, however!
There will be no radiation in the direction of acceleration (i.e. in the direction of polarization for the
incident photon), as no EM waves are ever emitted in the direction of acceleration, as we have seen earlier.
There is however a high probability that it leaves in the plane perpendicular to the acceleration (that plane
has the highest probability to contain the re-emitted photon). For angles between the acceleration and
the emission direction other than 90 degrees, the probability is lower, but still nonzero. It is only zero at
= 0, i.e. in the direction of acceleration.
Light that is scattered at exactly 90 degrees will be 100% polarized.
Consider a light beam (the cylinder), consisting of many photons, each individually polarized in random
directions, so that the light beam as a whole can be considered unpolarized, drawn here as the lines in all
directions inside the cylinder.
130
Each photon individually moves into the dust, and is scattered according to its own polarization.
The ones that exit at a 90 degree angle with respect to the incident beam will become 100% polarized, i.e.
the ones going to the left or right as shown, plus out of or in to the blackboard.
So in that plane, all light will become 100% polarized. We can look at that plane from above, look at the
beam head-on, so that the photons would be coming straight out of the blackboard.
Now, we see three individual photons (drawn in white, pink and blue), each with a dierent polarization,
coming out of the blackboard (we are looking into the beam).
If we then draw a position vector r as shown, how will the light be polarized at the end of that vector,
after having been scattered?
Well, it must still follow the same two rules: perpendicular to r, and in the plane of r and a. There is only
one solution per photon - and only one solution altogether, and so, as shown, they are all polarized in the
same direction, and so the light is 100% polarized when scattered at a 90 degree angle.
At a 45 degree angle (or angle between 0 and 90 degrees), the light exiting will only be partially polarized.
At 0 degrees, it will remain unpolarized.
11.1.3 Scattering demonstration and Rayleigh scattering
In the demonstration for this 90-degree-polarization, by shining light on cigarette smoke (made up of very
ne particles), another phenomenon is obvious: most of the light that exits is blue. That is because the
probability of light scattering is way higher for blue light (short wavelengths) than it is for red light, so
the smoke appears blue.
This phenomenon is known as Rayleigh scattering. The equation involved for the scattering light intensity
is fairly complex, but its notable that it is proportional to
4
. In other words, smaller values for lambda
(bluer light) will yield much higher values; the scattering probability is about seven times higher for blue
light than it is for red.
When the particles we scatter the light o are much larger than the above less than 0.1 microns, the eect
the wavelength has is reduced, and for much larger particles, it is negligible.
Rayleigh scattering is the reason for why the sky is blue (and red, at dawn and sunset). Mid-day, the
sun is high in the sky, and light scatters o dust in the atmosphere, and even o the air particles, due to
thermal uctuations that cause density dierences in the atmosphere.
131
The light that reaches us on the ground, that is not in the direction of the sun itself, is then in big part
scattered light. Due to the dependence on wavelength, more of that scattered light will be blue, and so
the sky appears blue.
At a 90 degree angle to the sun, because of what we discussed above, the light will be linearly polarized.
At other angles, it will be partially polarized.
Clouds are white because the particle sizes involved are large enough that scattering is about equally likely
for all colors.
Rayleigh scattering also tells us why the sky is red at dawn or sunset (as mentioned above).
It those cases, the sun is very low in the sky, and the sunlight has to go through a lot of more atmosphere
in order to reach someone on Earths surface. Because blue light is scattered more than red light (and
green light, being in between blue and red in wavelength, also being scattered more than red), most of
that light will have scattered away from the rays that nally reach you.
And so, only the red light remains.
Clouds are reddish at dawn/sunset for about the same reason: the light that reaches the cloud, and
eventually reaches you, is mostly devoid of shorter wavelengths, since those have been scattered away
already.
Having only red light shine on an otherwise white object will make it appear red, and that is what we see.
11.2 Rainbows
The lecture begins with the professor asking 15 questions about rainbows: if the red is on the inside or the
outside, the radius and width of the bow in degrees, whether it is brighter on the inside or the outside,
where they can be seen in the sky (in which direction), and many more, to illustrate that while everyone
has looked at a rainbow, not many have truly seen them.
132
Above is a single raindrop, with the sun at the horizon for simplicity.
So sunlight hits the raindrop. We draw a single ray of light, which hits the raindrow with angle of incidence
i (
1
in Snells law is then i).
Some light is reected at point A, and some is refracted, at an angle r, which we can nd by using Snells
law.
The refracted light will hit point B, where some is again refracted, back into air, and some is reected to
point C.
The light at point C again hits a water-to-air boundary, where some light is refracted out and some re-
ected.
Via Snells law, the exit angle for the refraction at point C is i, same as the incidence angle, and via
geometry and the law of reection, the internal angles are all r.
I should point out the angles and as I personally found them hard to see.
is the angle between the line through the drops center (the i = 0 line) and the ray leaving at point C.
is the angle all the way around from the same line, so they are related as shown in the second equation
below:
= 180

+ 2i 4r (by adding up the deections)


= 180

= 4r 2i
... where r is given by Snells law:
sin i
sin r
=
n
2
n
1
sin i
n
= sin r
r = arcsin
_
sin i
n
_
... where n is the refractive index of the water, for the given wavelength. So the relation between i and
is
= 4r 2i = 4 arcsin
_
sin i
n
_
2i
has a minimum value of about 138 degrees, which means has a maximum of about 42 degrees (180 -
138 degrees). This can be seen if the above equation is plotted, for 0 i 90 degrees; 90 degrees being
the maximum possible incidence angle, when light hits the raindrop at the very top.
So far, weve only really talked about light entering the top half, and leaving the bottom half of the
raindrop. However, it is after all spherically symmetric, so the opposite can happen. Light can enter the
bottom half, and refract/reect its way to exiting at the top.
Because of this, and because is in the range 0 to 42 degrees, we get a cone of light that exits the raindrop
on the left side:
133
The angle between the center line and the top is 42 degrees, as is the other angle.
There is axial symmetry around the center line.
None of this light can exit behind, since the angle between the center line an behind is much greater than
the maximum of 42 degrees.
However, these values are for red light, with n 1.331. For blue light, n 1.343, for which we nd
another result! For n
blue
, we instead nd a maximum for of 40.7 degrees, instead of 42.4 for red light!
In other words, the angle of maximum intensity varies with the wavelength, which is one of the crucial
eects that produces a rainbow. As mentioned earlier, this eect is known as dispersion. What we get
then, is this:
For red light
max
is 42.4 degrees. Since that is the maximum, it can vary between 0 and 42.4, depending
on the angle of incidence. If light enters with i = 0, then it will just reect back along the center line;
thats no problem. The only problem is for the angle to be greater than 42.4 degrees (again, for red light).
For blue light, the maximum angle possible is smaller at about 40.7 degrees, and so that will create a
smaller cone, as shown above. Green is between the two, and will create a cone with an angle in between
40.7 and 42.4 degrees.
However, as emphasized above, those are maximum angles only. For angles of 0 40.7, all colors of
light are allowed (assuming the blue at 40.7 is the shortest visible wavelength; violet is shorter, but harder
to see), and so we will see white light in the entire center, since white is the mixture of all other colors!
That is, the inside of the rainbow is bright, then we have the bow in all the colors, and the outside is dark
- the light cannot exit that way, since the angle would be greater than the 42.4 degrees maximum for any
visible wavelength of light!
This next part should probably be seen in lecture - video is much better than still pictures for this expla-
nation. I have two pictures, where there should be 4-5 (so that we can see them being drawn one part at
a time).
Here, we have the sun coming in from our left, and rain to our right. If it were also raining to our left, the
light wouldnt reach the rain on our right side, and we wouldnt see a rainbow.
134
Light hits the leftmost marked raindrop (and all drops along the line between us and it), and we get the
cone (half-angle of about 42.4 degrees, for the outer ring, the red light) as shown above. However, we are
looking outside that cone, and so we see no light there.
It also hits the bottom marked drop, and we are then looking into the cone, and we see white light at that
location in the sky (and from all other drops on that line).
It also hits the middle marked drop, and we see red light from it - as we are looking right at the edge of
the cone of light.
Of course, all raindrops will be hit by light, and all will act in the same way - but only some will be
visible to us, creating a rainbow in that direction - the direction opposite the sun. We could never see
a rainbow towards the sun, since in order to create a rainbow, light needs to be reected (after some
refraction inside the drop) back towards the sun, and so if we are looking at rain between the sun and us,
the rainbow-colored light would be away from us, and never reach our eyes.
So say the sunlight comes in from an angle, and the sun is higher in the sky that previously.
The rain is still to our right.
We envision a line from our head to our shadow, which then is the line to the sun behind us. At at angle
of 42 degrees above that line, we should see the red in the raindow, with the other colors being slightly
lower, and having white light on the inside.
Because of this relationship, when the sun is low, the bow will be relatively high in the sky. The higher the
sun is, the lower it will become (as the 42.4 degrees from the line will be closer and closer to the ground).
When the sun is at 42 degrees or more in the sky, we cant see rainbows any more - unless the water is
right next to you (you can create rainbows with a garden hose, for example!).
If we return to the rst picture in this section on rainbows, with the reection and refraction angles for a
single raindrop, note that at point C, where the light is drawn as exiting by refraction, there will also be
some reection.
If we follow the line of this reection, but backwards - so that sunlight comes from above, shines down
towards point C, refracts in, and reects twice inside the drop before leaving, we will nd that the light
creates a secondary bow. All we said above applies to the primary.
The secondary bow is fainter than the primary (even faint enough to not be visible at times), but wider
and reversed in the color order.
If we do the math for this secondary bow, we will nd that the maximum angle for the red will be about
50.4 degrees, while for the blue, it will be about 53.5 degrees, giving a dierence of about 3.1 degrees,
compared to about 1.7 degrees for the primary.
However, because the sun is not a point source - it is about half a degree in the sky - the bows will be
slightly larger than these numbers of 1.7 and 3.1 degrees.
135
Note also that the blue maximum angle is now greater than that of the red, so the colors in the secondary
will be reversed, with violet/blue on the outside and red on the inside.
We can also see, looking at these angles, that the secondary bow will be about 10 degrees above the
primary, so we only need to look slightly above the primary bow to nd the secondary.
Now, on to answering 12 of the 15 questions asked prior to the lecture.
Red outside or inside? The red will be on the outside of the primary bow, inside the secondary.
Radius of the bow in degrees? About 42 degrees, for the primary.
Length? Depends on the suns location, where there is rain, etc. If the sun is high, the bow goes
downwards, and not much is left to see. The bow is part of a circle with a radius of 42 degrees, so
we can nd the length by knowing how much of it is visible.
Width (in degrees)? As mentioned above, a bit above the 1.7 degrees we nd from just looking at a
single raindrops angles.
Light intensity, inside vs outside? The light intensity will be greater inside, as all colors will be
present there (and appear as white light). None of this light can end up outside, so the outside
should be darker.
At what time of day can we see them? Well, the sun should preferably be fairly low in the sky, to
make a bow visible, so morning or afternoon is perhaps better than mid-day.
In what direction should we look? In the direction opposite to the sun. (In fact, there is a tertiary
bow in the direction of the sun, but it was only rst photographed in 2011, after people had tried
for over 100 years, so it is extremely dicult to capture. With your naked eye you would only see
the sun.)
Is there one or two bows? Two; there is a secondary. (Not counting the ones we essentially cannot
see.)
If two, where is the second? The secondary is about 10 degrees higher than the primary.
If two, what is the color sequence of the second? The secondary has its color order, when compared
to the primary.
What is the radius of the secondary? About 52 degrees.
Width of the secondary? About 3.5 degrees or so, based on the dierence in angles plus the suns
apparent size.
Those are the rst 12 questions; the remaining three have to do with polarization of the rainbows light,
which we will look at shortly. First, a picture from Newton himself:
136
Here, we can see how the sunlight (S) is reected and refracted to reach an observers (O) eyes, creating
the primary (lower) bow, and the secondary (higher).
Note how the light in the secondary goes the other way around the raindrop, and is subject to one more
reection.
Below, we see a supernumerary bow.
Note that the colors arent all in order, and that there are multiple green and blue-purple-ish areas.
This cannot be explained by the reection and refractions weve talked about previously. Instead, this is
137
because of the wave nature of light.
Interference between rays of light along slightly dierent paths, with dierent phase, cause constructive
interference (for in-phase waves) and destructive interference (for waves out of phase by about half a
wavelength). In the case of destructive interference, the waves cancel each other out.
This can be seen by maxing a graph of two waves, e.g. of sin + sin( + ) where varies; at = the
wave is completely cancelled out, and at = 0, the waves amplitude is doubled.
Diraction (the interference mentioned above) can also cause entirely white rainbows to appear. In this
case, the water droplets are extremely small, which (for complex reasons) causes the light to essentially
smear out, and so the bows colors are all mixed together, creating a box that appears partially or entirely
white.
This can also be combined with the destructive interference eects, causing a white bow with smaller dark
bows inside!
11.2.1 Other atmospherical optical phenomena
Next, we have some other atmospherical optical phenomena.
Around the sun, and around the moon, we can sometimes see a 22

halo.
They are the result of ice crystals in the atmosphere, and form complete circles around their source, with
an angular radius of 22 degrees, giving them their names.
22 degree halos are fairly common. There are also 46 degree halos, that are then of course larger in radius,
but much fainter and thus rare to see.
Then there are sun dogs, or more properly, parhelia (singular parhelion):
Sun dogs are always 22 degrees distant from the sun (or moon, in which case they are then called moon
dogs), i.e. at the same distance the 22 halo is. They are always at the same height at as the sun.
They can be seen anywhere in the world, year-round, though thats not to imply that they are always
visible.
Like the above mentioned halos, sun dogs are due to light refracted through ice crystals in the atmosphere.
11.2.2 Polarization of rainbows
We now return to the last three questions, all about the polarization of the light in rainbows.
Is light from rainbows polarized, and if yes, how strongly and in which direction?
138
The answer is yes, and very strongly; as for the direction, the light will be polarized along the bow: at
the leftmost part, the polarization will be up/down, curving along the bow, being horizontal at the bows
top, and then again turn vertical at the very end.
Looking again at the diagram of the single raindrop, we know that the actual bow shows up when 42
degrees, which happens when the angle of incidence is about 60 degrees.
The angle of refraction r is related to i via Snells law, as the light is refracted:
r = arcsin
_
sin i
n
2
_
So for n
2
1.333 in water, and i of 60 degrees, r 40.5 degrees.
At that point, B, where light is reected back towards the front (left side) of the raindrop, the Brewster
angle is given by

Br
= arctan
n
2
n
1
... where n
2
is the refractive index of the medium the light would be going to (the air, n
2
= 1), and n
1
the refractive index of the water, n
1
1.333. With those values, we nd
Br
37.88 degrees, which is of
course very close to the 40.5 or so degree angle of the light coming there.
Because light would be 100% polarized at the Brewster angle, the light will be not 100%, but very strongly
polarized.
11.3 Review for Exam 3
There was just about nothing reviewed that had not been covered previously, as expected, so I did not
take any notes watching this lecture.
139
Chapter 12
Week 12
12.1 Double slit interference and interferometers
Light can be seen as both waves and particles, under dierent circumstances. When light waves interfere
with each other, which this and the next lecture are all about, clearly we use the wave model. It doesnt
make much sense to think of two photons, each with the same energy, cancelling each other out to become
darkness. With waves, on the other hand, it does - at least from a mathematical viewpoint, where e.g.
sin(x) + sin(x + ) is zero everywhere.
A simple physical example of this is when we have two sources, producing waves with exactly the same
frequency.
In this example, we can think of the waves as two-dimensional waves, such as waves on the surface of a
body of water. However, the same principle applies to EM waves in three dimensions.
Say we are at point P, at t distance r
1
and r
2
away from the two sources, respectively.
If the waves from both sources arrive at the same time, in phase, they will add: the amplitude will become
twice as high, as both the mountains and the valleys of the waves will add. This is called construc-
tive interference: the two waves interfere in such a way that the net result is greater than either wave alone.
If they were to appear exactly 180 degrees out of phase, and are of the same amplitude, then there will
be points where the net amplitude is zero - the waves cancel each other out. We call this destructive
interference. Therefore, we need to have the waves dier by half a wavelength - which is the same as
saying 180 degrees out of phase. So the dierence between the distances to the two sources, at the point
we are, must be
[r
1
r
2
[ =
(2n + 1)
2
(for destructive interference)
... where n is an integer. This will then yield 0.5, 1.5, 2.5 etc.
140
To get constructive interference, the waves need to arrive in phase, so the distance must be a multiple of
the full wavelength. Thats easier:
[r
1
r
2
[ = n (for constructive interference)
We should perhaps call the above total constructive/destructive interference, or something to that eect,
since e.g. two waves, each of amplitude A might add to a wave with an amplitude of say 1.1A, which is
then also constructive interference. The same but in reverse applies to destructive interference: if the sum
of the two waves is smaller than each wave alone, then there is destructive interference.
When the sum of distances like this is a constant, so that r
1
+ r
2
=constant, that gives us the equation
for an ellipse, e.g.
x
2
2
2
+
y
2
3
2
= 1
... which graphs as an ellipse with height 3 and width 2 (semi-minor axis 2, semi-major axis 3). If we
switch to a dierence, so that we only change from a plus sign to a minus sign, we get a hyperbola instead.
The points (-2, 0) and (2, 0) are solutions to both equations, however.
In three dimensions, if we rotate the entire thing, we get a hyperboloid instead, which (as expected) has a
similar shape, but in three dimensions.
Lets have a look at how this might look.
(Note that the lines can also be mirrored on the other side! They were only drawn on one side, but they
are equally valid mirrored.)
We have the separation d between the two sources, which are in phase and at the same frequency/wavelength.
At the line where r
2
r
1
= 0, there will be a maxima in all cases. We can see this from the maxima
equation with n = 0: there will be a maxima there regardless of the wavelength.
In the case of 3 dimensions, the line will instead become a plane, perpendicular to the blackboard.
The yellow lines, where the destructive interference completely cancels out the waves, are called nodal lines,
or nodal surfaces in the three-dimensional case. The maxima (white lines) are then sometimes known as
anti-nodes, or simply maxima.
If we look at the distance between the nodal lines, right in between the two sources (so to the very left
of the long lines), the distance between two adjacent nodal (yellow) lines will be /2, and likewise for the
spacing between two adjacent anti-nodal lines.
Therefore, the approximate number of maxima or minima is given by
# maxima, # minima
d
/2
=
2d

... where d is the separation between the two sources.


141
Because the curves will be hyperbolas, which are asymptotic to a line (for example, y
2
x
2
= 1 is asymp-
totic to y = x in the second and fourth quadrants (x < 0, y > 0 and x > 0, y < 0), and to y = x in the
rst and third (x > 0, y > 0 and x < 0, y < 0).
Therefore, we can dene an angle between the origin (the center of the two sources), and the hyperbolic
line in the rst quadrant, at least assuming that we are looking from far away, where we can treat the
hyperbola as equal to its asymptote, without any signicant loss of precision.
In the approximation that r
1
d and r
2
d, i.e. the separation between the sources is negligible
compared tot he distance to them, we can make some simplications. First, we can think of the lines
between our point P to the sources as parallel, despite that they obviously cannot be exactly parallel,
since they meet up at P.
This means we can measure the angle as shown, which will then prove to be equal at many places, again
as shown.
r
2
r
1
, shown in pink, can be found via simple trigonometry:
sin =
r
2
r
1
d
r
2
r
1
= d sin
We can now easily nd the directions where we have constructive and destructive interference.
As we saw earlier, for constructive interference, we want r
2
r
1
= n With our newfound equality, that
means
sin
n
=
n
d
(for constructive interference)
As for destructive interference, we do the same in replacing r
2
r
1
with d sin and nd
sin
n
=
(2n + 1)
2d
(for destructive interference)
... adding an index n to the angles, as there will be multiple maxima angles and multiple minima angles.
Now that we know the angles, let us look at what happens if we project this onto a screen a distance L
away, where L is a very large distance (L d).
If we use x to denote distance on the screen, perpendicular the length L (thus a vertical distance), then
we will nd that
tan =
x
L
142
When is small, tan sin (as tan =
sin
cos
, and cos 1 for small angles).
We can then calculate where the maxima will be on the projection, i.e. where there is constructive inter-
ference on the screen, still at the distance L, as given by the x value. There will be multiple, as there are
multiple angles of maxima being projected.
We know that for constructive interference to occur, the waves must arrive in phase. That is, the dierence
is distance r
2
r
1
= d sin must be a multiple of the wavelength. What implication does that have for
the vertical distance? When x = 0, we will have a maxima - as we saw earlier (as = 0 where x = 0). As
x grows, as does . The relation is simple, when we are using radians for the angle:
x = L
For small angles, sin , which means that, using the above equivalence
sin
n
=
n
d
we have

n

n
d
(I am not sure about the above reasoning, but since tan sin for small angles, it should be true.
The answers are the same as given in lecture, so those are certainly correct.)
We then multiply that by L as we found above, and nd:
x
n

nL
d
(for constructive interference)
that tells us at what linear separations from the center line there will be maxima, at a distance L from
the sources.
Using the similar equivalence for the destructive interference angle, the x values where there are minima
are given by
x
n

(2n + 1)L
2d
(for destructive interference)
However, it is important to remember that these are approximations! They become less and less accurate
as grows, and since tan for certain angles, the approximation is truly awful for larger angles. In
such cases, we need to use the actual tangent instead of just the angle or the sine.
Anyhow, lets look at a dierent way of creating these two sources:
143
We shine bright light - laser light is excellent, as it is essentially monochromatic (only one wavelength) -
onto two very very small slits. According to the Huygens principle, these slits will then act as sources for
secondary waves, and will emit in-phase spherical waves, at the same frequency as the incoming light.
So above, we have the incoming light as plane waves, and on the other side of the two slits, we have two
sources of spherical waves, which will interfere at a screen a certain distance away.
We shine the laser beam, which has a diameter of about 3 mm, onto these slits, which are cut-out of a
black material. The separation between the two lines is d = 0.088 mm, or 8.8 10
5
m.
The wavelength of the laser light is 6328 ngstrm, which is 632.8 nm in units I personally prefer. With
the distance d = 8.8 10
5
m between the slits, and L 10 meters, we can calculate the x value for the
rst maximum (except for the one at x = 0, which is certainly always there if the sources are in phase).
x
1

L
d
7.2 cm
The result looks like this:
We indeed see that there are places where the two waves cancel each other out, and there is darkness.
The light intensity of the maxima is non-uniform, which will be talked about more next lecture. If the
slots were much thinner (much thinner than the separation between the two), the intensity would be more
uniform, but then less light would pass through the now-thinner slots, and so it is a tradeo between the
two. Clearly, visibility is important for a demonstration such as this one.
If we do this experiment with white light instead, there will probably not be locations of darkness. The
reason is that since white light is made up by tons of dierent wavelengths, and the location of max-
ima/minima is dependent upon the wavelength, each color of light would have its maxima and minima at
dierent locations in space (except for x = 0, as usual), and so the colors would smear out a bit, and there
would not be any well-dened locations where all light cancels out.
These equations are demonstrated in several dierent ways, showing the extreme similarities between dif-
ferent types of waves: sound waves ( 0.113 m), water waves (f 7 Hz), the red laser light ( 630
144
nm) and radar waves ( 3 cm).
Lets look at the radar example. We have
= 3 cm
d = 23 cm
L = 120 cm
What is x
1
, the rst maximum (except x = 0)?
First, we can calculate the angle, using
sin
1
=

d

1
= arcsin
_

d
_


d
(for small angles)
Using this small angle approximation, we can calculate x
1
as follows:
x
1

L
d
15.65 cm
If we ignore the small angle approximation, and use the more correct equation:
tan =
x
L
x = Ltan = Ltan
_
arcsin
_

d
__
x = 15.787 cm
12.2 Gratings and resolving power
Last lecture, we talked about interference between waves, due to two coherent light sources. This lecture
will talk about this when we have dozens to thousands of sources, rather than just two.
In the case where have many such thin slits (or sources of in-phase spherical waves of the same frequency),
say N such sources/slits, where the distance between two adjacent ones is d (d was previously the distance
between the two sources, where we only had those two), we will in fact nd the same angles and linear
locations of maxima:
sin
n
=
n
d

n
(small angle approximation)
So in other words, we will nd minima in the direction of
n

n
d
, and we can turn that into a linear
distance by multiplying it by the distance L to the screen. All this is assuming the angles are measured
in radians, and are fairly small.
An intuitive way of thinking about this result that if the top two sources costructively interfere, then
the second and third will also constructively interfere, as will the third and fourth, etc., as the distance
between two adjacent sources is always d.
As for the points where there is destructive interference, that is tricker, and for a full derivation, we are
told to have a look at MITs 8.03 course (Vibrations and Waves).
As for the result, with N sources, we will nd that in between each pair of maxima, we will have N 1
minima - i.e. places where there is complete destructive interference.
145
Note that with N = 2 as in the previous lecture, we indeed nd the result of having one minima in between
the maxima.
Here, we see how the interference pattern may look for N = 5, giving 4 minima in between each maxima.
Assuming the secondary peaks (the small bumps) are identical, we can calculate the angle between
0
= 0
radians and the rst minima, simply by dividing the distance to the next primary maxima by N:
=

Nd
(measured from the middle of a primary peak)
is then the width of a peak, in radians (from peak to the next minima).
The larger N is, the higher (but also narrower) the primary peaks will be. If N increases by a factor of
e.g. 3, the electric eld vector also increases by a factor of three. However, light intensity is given by the
Poynting vector, which is proportional to E
2
- and therefore, the intensity of the peaks is proportional to
N
2
.
The peaks will also become narrower by a factor of three, which restores balance and ensures that there
is no violation of the conservation of energy going on here.
As a demonstration, the same red laser as before is used, together with a grating (a transparent sheet with
lots of thin lines blocking light) of 2500 lines per inch, and with L = 10 meters. With the laser beams
diameter of 2 mm (hmm, it was given as 3 mm earlier!), we can calculate some values. The given values
are then
d = 10.16 m
= 6.3 10
7
m
L = 10 m
984.25
lines
cm
N 984.25
lines
cm
2 mm0.1
cm
mm
200 lines
We can then nd the rst-order maximum using the good old formula,

1


d
0.062 rad 3.55

The width of the peak, from the center to the next minima (so that the peak is perhaps really twice as
wide, in a way) is then given by above:
=

Nd
0.018

146
We can then multiply this angle (in radians!) by the distance to the screen, to nd the rough width in
meters, which turns out to be about 3 mm. The real with will probably be greater, as the laser beam used
is spreading out more than the 0.018 degrees we found above, and becomes the limiting factor.
As we saw with the double-slit interference, if we use white light instead, the peaks will spread out, since
their positions are proportional to the wavelength of the light. Blue light will have the peaks closer to-
gether than red light, and looking at the equation for , should also have thinner peaks.
If we do this with white light, we will see white at the center maximum - all wavelength have a maximum
at = 0, but at other places, the light will tend to spread out according to its wavelength.
Heres what we see when shining white light through (or on, as this is a reecting grating) a second grating,
with d 2.5 m:
As we would expect, there is white light at the center, where there is a maxima regardless of wavelength.
Further out, we see other colors, with blue always being the closest to the center (the angles for blue
maxima are always smaller than other colors of greater wavelength), and red farther away.
If we add in the red laser again, which is aimed exactly the same, we expect that to fall in the same places
where the red light from the white light ends up, and that is indeed what happens:
We can apply the same principles to light of other colors, of course. As an example, a demonstration
is given in the lecture using a neon light source, which looks red to the naked eye. Looking through a
grating, we can see discrete lines at particular wavelengths - many of which are indeed in the red, a few
in the orange, and a few in the green as well.
12.3 Single-slit diraction
Lets look at another dierent case, whereby we have a single slit. Even then, there will be interference,
and there will be places of total destructive interference (at least assuming coherent light, such as the red
laser weve been using).
For a single slit of width a, we will nd a maximum at = 0, as always. We will nd minima at
sin
n
=
n
a

n
... where the derivation is once again shown in the follow-up course, 8.03 Vibrations and Waves.
This looks suspiciously similar to the one we found for the two- and multiple-slit interference. As always,
in the small-angle approximation, we can just multiply the angle in radians by a distance L, to nd the
linear separation on a screen a distance L away:
x
n

Ln
a
I refer to the lecture and the book for more details about this.
Heres what the intensity will look like, as a function of x on some screen a distance L away:
147
So we nd a very broad maxima at the center, which ends at the rst minima, where the intensity is zero
After that, we nd the rst-order maxima, which is very weak compared to the center; the second order
is fainter yet, etc.
In other words, the width of the center maxima is then really just
L
a
- or twice that, if you prefer, since
it is symmetric around the other side of the zero.
Theres an extremely interesting phenomenon (not very) hidden in this simple equation. Since a is the
width of the slit, and the width of the peak is inversely proportional to a, making the slit narrower will
make the peak on the screen wider! Very non-intuitive.
We can make the peak extremely wide (but also faint, since less light can come through) by making the
slit very, very narrow.
What about if we dont have a slit, but a circular pinhole instead?
We get something similar to the above graph, except we rotate it around its axis, as we now have axial
symmetry. We would see a circular maximum at the center of a certain radius, followed by a ring of
darkness of greater radius, followed by a ring of the rst order maxima, etc.
For reasons not explained in the lecture, the radius (in radians) of
1
, the angular radius of the rst
minima, will be

1

1.22
a
We use the small-angle approximation here as well.
(The source of the 1.22 appears to be from the rst zero of a rst-order Bessel function of the rst kind -
something I have no experience at all with.)
If we have a pinhole, and we look at the image of two light sources through the pinhole - where the sources
are far apart. Two examples given are the headlight of a car, and two stars - far apart is relative, of
course, and angular separation is what really matters.
Each source would produce a circular diraction pattern. As we move the two sources closer and closer
together, the diraction patterns will also move closer together. The angular resolution is the smallest
angular separation where we could still say that we see two separate patterns, rather than having them
merge into one. For the denition of angular resolution, we take the light sources as equal in intensity,
and for simplicity we now assume that they are monochromatic.
Heres a drawing:
148
The commenly accepted denition is then that the center of the second (rightmost) pattern must be no
closer to the rst than that it is centered on the rst minima, i.e. at

1
1.22

a
If they are any closer together, we well tend to see it as a single source. We call this the Rayleigh criterion
of resolution. In other words, the Rayleigh criterion says that the separation between the two light beams
has to be larger than the above angle, or we will see it a as a single source. Note that it is a function of
a, the pinhole diameter, such that we can increase the resolution by making a larger.
This is also known as the diraction limit on angular resolution. Whether we have a pinhole, or a lens or
a concave mirror, the latter two which we might nd in a telescope or a regular camera, we can never beat
this limit.
Using this formula, we nd that the diraction limit for a 20 cm lens is about half an arc second for 500
nm light, while that for a 2.4 m lens is about 1/25 of an arc second (calculated by ignoring the factor of
1.22).
(A degree, or arc degree can be subdivided, like time, into 60 arc minutes, each of which is 60 arc seconds.
Thus 1 degree = 1/360 of a circle = 60 arc minutes = 3600 arc seconds.)
In other words, the larger the telescope we have, the better the angular resolution.
However, for Earth-based telescopes (on the ground), there is another, much more problematic limit: air
turbulence in the atmosphere. Air turbulence limits the resolution to about half an arc second, in a process
called astronomical seeing. Seeing is one reason that we have space telescopes: there is essentially no air
whatsover at the height of e.g. the Hubble space telescopes altitude of about 150 km, and so Hubbles
lenses are indeed diraction limited.
The angular resolution of a human eye is diraction limited to about half an arc minute (about 1/120
of a degree), calculated for a pupil size of about 4 mm and with 500 nm light (green). According to the
professor, however, most students will have vision worse than that, such that it is not diraction limited.
Nobody could possibly have better optical resolution, however, unless their pupil size was greater than the
4 mm used in the calculation.
12.4 Doppler eect and the Big Bang
The Doppler eect is the reason for the familiar eect where the pitch of an ambulance/police car/re
trucks siren changes as it travels past you.
When the sound transmitter is moving away from you, the pitch you hear is lower than the pitch that is
actually transmitted at the source. If the transmitter is coming towards you, the pitch instead increases.
This is given by the equation
f

= f
v
s
v
rec
v
s
v
tr
149
... where f is the original frequency, f

is the frequency received/percieved/heard, v


s
the speed of sound
in air, v
rec
the velocity of the receiver and v
tr
the velocity of the transmitter.
In this equation and in what follows, we take the velocities as positive when moving towards the right of
the blackboard, and negative when moving towards the left.
Note that the case of a stationary transmitter and a receiver moving away is not equal to the case of aa
stationary receiver where the transmitter is moving away!
If the transmitter is stationary, and is at say 440 Hz (the standard tuning in modern music is to the A440
note), and the receiver moves away at the speed of sound, then there will be no sound heard at all - the
receiver (just barely) outruns the sound waves entirely. The equation above gives 0 Hz for that case -
granted, it sounds weird talk about a 0 Hz sound, but since 0 Hz means no wave motion, I suppose that
is technically true.
On the other hand, if the transmitter is moving away at the speed of sound, while the transmitter is
stationary, what do we nd?
f

= 440
340 0
340 (340)
= 220 Hz
The frequency is halved, but we still hear the sound. If you nd that unintuitive, think about it a bit
more - its very clear that while the transmitter moves, the waves will still reach you as they always travel
at 340 m/s in your direction.
The equation also has a very curious eect, which can be hard to show in practice: if the transmitter is
moving away at twice the speed of sound, the sound (a piece of music, perhaps) will reach a stationary
observer in time and in tune, but backwards!
How do we intuitively explain the Doppler eect? It is actually quite simple, but is best shown using
animated graphics, which I cant really use here. I suggest looking it up online.
In short, when moving towards a transmitter (or the transmitter is moving towards you), the wave fronts
become compressed, and the wavelength becomes shorter (meaning that for sound, the pitch increases;
for light, it becomes blueshifted; more on that soon). When the distance between is increasing, the wave
fronts become separated from each other, and the wavelength becomes longer (sound: lower pitch; light:
redder color, or redshift).
12.4.1 The Doppler eect and light
Lets look more at the Doppler eect in light (and other EM radiation).
The derivation requires special relativity, but the result is given by
f

= f
_
1
1 +
_
1/2
where
=
v
c
where v is the relative velocity between the transmitter and receiver, and c is the speed of light.
If > 0, then the two are moving apart; if < 0 the two are approaching each other.
Much of the idea behind special relativity, as the name implies, is that motion is relative. There is no such
thing as an absolute reference frame, and therefore, there is only one term of velocity in the equation. It
does however, of course, matter whether the two are approaching or receding from each other.
The reason this doesnt apply to the sound waves above is that sound waves require a medium to travel
through, whereas electromagnetic do not.
150
In terms of wavelength, which is physically identical but perhaps more useful, we nd (as =
c
f
and thus

=
c
f

):

=
_
1 +
1
_
1/2
Note that the denominator and numerator have swapped places, as compared to the frequency-based equa-
tion.
The velocity v in the above equations is the radial velocity. If the velocity between the source and the
receiver is not is the straight line connecting the two, then there is an angle between the velocity vector
and the position vector between the tow. The velocity we need is then given by v cos , which is a result
weve seen many times before when using vectors. Only the component of the velocity in the direction of
the object matters.
Police often use radar guns, which can measure the speed of cars by reecting radar waves of a known
frequency/wavelength o cars, and measure the wavelength of the returning waves. The radial velocity of
the car can then be calculated.
12.4.2 Big Bang cosmology
Doppler shift is also used by astronomers to calculate the radial velocity of distant stellar objects. Most
stellar spectra consist of lines, spectral lines, of a discrete wavelength, from the atoms and molecules that
they consist of.
If we happen to know a certain spectral line, and we see the same line, but shifted in wavelength, we have
two names for that, as mentioned earlier:

> : redshift

< : blueshift
So if the actual wavelength, from the stellar object, is shorter than we would have expected, that means
the object is approaching us (the light is blueshifted), and vice versa if it is longer than expected (the light
is redshifted).
Strictly speaking, if a spectral line is at e.g. 1000 nm when stationary, and say 1100 nm when measured,
we still call that redshift, despite that it is technically moving away from the red (red being at 620-700
nm or so) now that the wavelength has surpassed that of red.
The same applies to blueshift: any decrease in wavelength is known as blueshift, and any increase is known
as redshift, regardless of the actual wavelengths involved.
If we know a certain spectral line, as above, and we can generate that same spectral line in our observatory,
we can compare the two, and via the equation above, nd and then nd v = c, and so we know the
relative radial velocity between us and the object (perhaps a star, or an entire galaxy).
Edwin Hubble, the astronomer after which the Hubble Space Telescope is named, discovered a curious
relation between the radial velocity of a galaxy and its distance.
The further away a galaxy is, the higher its redshift is. In other words, the further away it is, the faster
it is moving away from us! This fairly linear relation it known as Hubbles law, and the constant of pro-
portionality H or H
0
(the latter is commonly used for the current value, as it actually changes over large
time periods) is known as the Hubble constant. The value of this constant was most recently measured
(on March 21 2013) to be 67.80 0.77 km/s/Mpc (kilometers per second per megaparsec).
One parsec is about 3.26 light years, so a megaparsec is 3.26 10
6
light years, or about 3.1 10
19
km.
151
The value has uctuated over time, however, as measurements have gotten more precise. (Hubbles original
measurement was on the order of 500 km/s/Mpc!)
Hubbles law, in mathematical form, states that
v = H
0
d
where v is then measured in km/s, H
0
in km/s/Mpc and d in Mpc.
As an example, if we see an absorption line in the spectrum of a galaxy that is is 7% shifted towards the
red, we can calculate by solving the wavelength equation for :
=
(

)
2

2
(

)
2
+
2
If we plug in

= 1.07 (

> , or else its not redshifted but blueshifted), we nd


0.0675556
If we then use v = c, we nd
v 20266 km/s
We can then calculate the approximate distance using Hubbles law:
v = H
0
d
d =
v
H
0

20266
67.80
298.9 Mpc
One way to describe why this happens is to assume we are at the center of the universe, which was
formed from a single point, if we only turn back the clock. The galaxies that obtained the largest speed
at that initial explosion (the Big Bang) are not the furthest away from us.
This isnt strictly true, as the universe doesnt even have a center according to current theories on the
shape of the universe. (For a two-dimensional analogue, think about the surface of a balloon/sphere:
where on the surface is the center? The universe can be thought of in a similar way, except we have three
dimensions rather than two.)
Thats also ignoring that there were no galaxies right after the Big Bang. Still, lets assume this is true
for now.
We can then ask the question: if the universe had a beginning, where everything was together, as we would
see it if we simply ran time backwards (everything would get closer and closer to everything else), how
long ago did this happen?
According to data from two probes that monitor the Cosmic Microwave Background Radiation (CMBR),
the WMAP (W-map, the Wilkinson Microwave Anisotropy Probe) and the Planck satellite, essentially
the successor to WMAP, the universe is 13.7980.037 billion years old. Numbers in this order of magnitude
are widely accepted in the scientic community, and indeed there is plenty of proof, such as very, very old
known galaxies that are only a few hundred million years younger than the universe itself.
As a side note, the solar system and Earth are about 4.5 billion years old, while life on Earth (simple cells)
are thought to have existed for about 3.6 billion years.
12.5 Farewell special
I didnt take any notes for this lecture. While the content is certainly interesting, Im not sure exactly
what to write down for it; it must certainly be seen either way.
152
Chapter 13
Homework problems
13.1 Week 2
13.1.1 Problem 1: Motion of charged particle in electric eld
An electron of mass m
e
and charge q = e is injected horizontally midway between two very large
oppositely charged plates. The upper plate has a uniform positive charge per unit area + and the lower
plate has a uniform negative charge per unit area . You may ignore all edge eects. The particle has
an initial velocity v
0
= v
0
x.

a) What is the magnitude [and direction] of the electric eld between the plates?
I didnt bother solving this one manually, since I remembered the answer. Each innite plate has a eld
magnitude of

2
0
. The positive, top plate has a eld pointing outwards, while the bottom, negative plate
has a eld pointing inwards.
Since they are acting in the same direction between the plates, the answer is twice the above. Thus, the
electric eld between the two plates is

E =

0
( y)
b) What is the magnitude of the acceleration of the electron when it is between the plates?

F = ma, so a =

F
m
.

F is given by the electric eld strength times the magnitude charge e:

F =
e

0
Therefore,
[a[ =
e

0
m
e
c) What is the y-component of the position of the particle when it reaches the plane dened by x = L?
This one is a bit harder to calculate - unless my solution is a bad one.
We were given that the initial velocity is purely in the x direction, and the electric force (the only force
we are to calculate) acts strictly in the y direction. Therefore, the velocity in the x direction will in fact
be constant, so we can easily calculate the time t taken for the electron to reach x = L:
t =
L x
0
v
0
=
L
v
0
We can now use a fairly complex equation from mechanical physics - it can a least be complex with all
values lled with expressions:
y
final
= y
initial
+ v
initial
y
t +
1
2
a
y
t
2
y
initial
is given as 0, as is v
initial
y
, so that simplies our situation a lot:
153
y
final
=
1
2
a
y
t
2
We found the acceleration previously; we just need to multiply it by t
2
, where we found t =
L
v
0
:
y
final
=
e
2
0
m
e
_
L
v
0
_
2
That concludes problem 1!
13.1.2 Problem 2: Three plates capacitor
Three innite uniformly charged thin sheets are shown in the gure below. The sheet on the left at
x = d is charged with charge per unit area of 3 , The sheet in the middle at x = 0 is charged with
charge per unit area of +, and the sheet on the right at x = d is charged with charge per unit area of
2 . Find the x component of the electric eld E
x
in each of the regions 1, 2, 3, and 4.
The gure is simply three vertical lines, the sheets, and four regions. Region 4 is to the left of all of
the sheets, region 3 between the rightmost and the middle sheet, region 2 between the middle and the
rightmost sheet, and region 1 to the right of all the sheets.
This problem is very straightforward, as long as you know the electric eld of an innite plate to be
E =

2
0
.
We then calculate the electric eld due to each plate, and use superposition to add them together, keeping
the directions in mind for each region.
E
1
=
3

0
E
2
=

0
E
3
=
2

0
Region 1:

2
0
(2 + 1 3)
Region 2:

2
0
(3 + 1 + 2) = 0
Region 3:

2
0
(3 1 + 2)
Region 4:

2
0
(3 1 + 2)
Thats really all there is to it!
13.1.3 Problem 3: Electric eld, potential, and electrostatic potential energy
Point charges Q
1
, Q
2
, and Q
3
reside on three corners of a square with sides of 1 m; the distance from Q
2
to P
3
is 2m (see diagram).
(a) What is the electric potential, V , at P
1
? (Normalize the potential to be zero at and give your
answers in Volts). (Note: P
1
is located at the unoccupied corner of the square.)
Well, we begin by calculating the electric potetial V for each of the charges alone, using V =
Q
4
0
r
:
154
V
Q1
=
11 10
6
4
0
r
V
Q2
=
3 10
6
4
0
r
V
Q3
=
7 10
6
4
0
r
The question then asks for the potentials at P
1
, P
2
and P
3
. All we have to do is to add the three potentials
above, while substituting the distance to that point from each charge for r. That is a mess to write, but
easy to do, so Ill leave that to the reader. Its either as simple as 1 m or 3 m, or youll have to use the
Pythagorean theorem.
(b) Are there points or surfaces in space (other than innity) where V is zero?
Yes. We have both positive and negative charges, which means we will have both positive and negative
potentials at dierent locations in space. Potentials dont charge abruptly, so in an area where the po-
tential transitions from positive to negative or vice versa, there will be a point where it exactly equals zero.
(c) What is the electrostatic potential energy of the system? Express your answer in Joules.
To solve this, we add up the electrostatic potential energies of each pair of charges:
U
total
= U
12
+ U
13
+ U
23
Since U = qV , this is the potential at each point times the charge at that point, e.g.
U
12
= V
1
Q
2
U
13
= V
1
Q
3
U
23
= V
2
Q
3
For r, part of the potentials, we use the distance between the two charges in question.
The answer to the question is then U
total
.
(d) Suppose we release the three charges so that they can move freely in empty space. How much energy
is released in the form of kinetic energy?
This is the only question of the homework so far that I didnt solve on my own, even with the book and
Google. The correct answer is The question is not well dened, if Ive understood it correctly because
the answer depends on the order you release the charges in, and (I think?) the times you do so.
I still dont see how there can be anything but one correct answer, if you release them all exactly simul-
taneously, though, even if we havent learned how to calculate it.
13.1.4 Problem 4: Electric eld of a charged ring
A point-like negatively charged particle of mass m and charge q is initially released from rest from the
point P along the positive z-axis where the magnitude of the electric eld is largest.
The diagram shows a charged ring, with radius R and total charge Q, centered on the z axis and located
o the x-y plane. The point P has coordinates (0, 0, z), where we are to nd z:
(a) What is the distance from the point P to the origin? Express your answers in terms of the following
variables, if relevant: q, m, R, Q, and
0
.
155
Well, all we know about P is that the magnitude of the electric eld is the maximum there. Clearly, our
rst task is to nd the electric eld due to the ring, to begin with!
By symmetry arguments, along the z axis, the x and y components of the electric eld will cancel out, so
we only care about the magnitude of the z component, E
z
.
First, we divide the ring into tiny segment

d, each of which carries a charge dq =
Q
2R
(the charge
distribution is uniform, though the problem didnt actually state that).
Each of these contribute a small amount

dE to the total electric eld at the point we are interested in.
The distance between a point on the ring and P is given by r =

R
2
+ z
2
, via the Pythagorean theorem.
So far, we have (from Coulombs law)
dE =
dq
4
0
r
2
=
Q
8
2

0
r
2
R
We are only interested in the z component, dE
z
= dE cos (by vector decomposition), with being the
angle between the point P and d (if we draw a line between the two, the angle is what that line makes
with the z axis).
Via geometry, we can nd see cos =
z
r
. If we make this substitution, and then substitute the value for
r =

R
2
+ z
2
:
dE
z
=
Q
8
2

0
r
2
R
cos =
Q
8
2

0
r
2
R
z
r
dE
z
=
Qz
8
2

0
r
3
R
dE
z
=
Qz
8
2

0
R(

R
2
+ z
2
)
3
dE
z
=
Qz
8
2

0
R(R
2
+ z
2
)
3/2
Finally, we want to integrate this over the entire circle. That turns out to be an easy integral, because
everything above is a constant!
E
z
=
_
dE
z
d
The innitesimal segment d = Rd (arc length):
E
z
=
_
2
0
dE
z
Rd =
_
2
0
Qz
8
2

0
R(R
2
+ z
2
)
3/2
Rd
E
z
=
Qz
8
2

0
(R
2
+ z
2
)
3/2
_
2
0
d
E
z
=
Qz
4
0
(R
2
+ z
2
)
3/2
We want to know where the eld is at its maximum, so we start o by calculating the derivative of the
electric eld:
d
dz
_
1
4
0
Qz
(R
2
+ z
2
)
3/2
_
=
Q(R
2
2z
2
)
4
0
(R
2
+ z
2
)
5/2
For that, I thank Mathematica. We then set the derivative equal to zero, to nd extreme points; we know
that P
z
> 0, so we can discard the negative solution we nd. One solution remains:
156
P
z
=
R

2
... which answers part (a).
(b) What is the speed of the particle when it reaches the origin? Express your answers in terms of the
following variables, if relevant: q, m, R, Q, and
0
.
For this part, we want to know the potential dierence between P and the origin. One way of doing that
is to integrate the electric eld dot the direction vector. However, we know that they point in the same
direction ( z), so we can skip the dot product and do a regular integration with dz:
V
P
V
origin
=
_
0
P
z

E

dz =
Q
4
0
_
0
P
z
z
(R
2
+ z
2
)
3/2
dz =
_
Q
4
0
__
1
R

1

R
2
+ z
2
_
I admit, more Mathematica for that integral.
At least now the worst part is over. Only algebra remains.
We now know the potential dierence, and can very easily calculate the potential energy dierence by
multiplying the voltage dierence by q, the charge of the particle.
Via conservation of energy, the gain/loss in potential energy plus the gain/loss in kinetic energy must
equal zero:
U + K
e
= 0
The particle is released from rest, so K
e
starts at zero, and K
e
= K
e
. U is
U = V (q)
So what is K
e
? Its the kinetic energy the particle has when it reaches the origin. We use classical
mechanics for this, so
K
e
=
1
2
mv
2
We set the sum of the two equal to zero, and solve for the velocity v, which will be our answer:
U + K
e
= 0
V (q) +
1
2
mv
2
= 0
2V (q) + mv
2
= 0
mv
2
= 2V (q)
v
2
=
2V q
m
v =
_
2V q
m
Almost done! The only thing left is to put our ugly expression for V in there... Hold on:
v =

_
2
__
Q
4
0
__
1
R

1

R
2
+z
2
__
q
m
... actually, I lied. z isnt allowed in the answer, but we solved that the point in question has z =
R

2
.
157
v =

_
2
_
_
Q
4
0
_
_
1
R

1

R
2
+(
R

2
)
2
__
q
m
... and that can be our nal answer, if we choose to not clean it up.
13.1.5 Problem 5: Two spherical conductors
Two spherical conductors, A and B, are placed in vacuum. A has a radius r
A
= 25 cm and B of r
B
= 35
cm. The distance between the centers of the two spheres is d = 225 cm.
A has a potential of V
A
= 110 volts and B has a potential of V
B
= 40 volts.
(a) An electron is released with zero speed from B. What will its speed be as it reaches A? Express your
answer in m/sec.
The distance betwee is in fact superuous information! We can calculate this using only the potential
dierence:
V
B
V
A
= 150 V
V
A
V
B
= 150 V
(Well need both, as part (b) asks about a proton.)
The kinetic energy (in classical mechanics) is:
K
e
=
1
2
mv
2
We can solve that for v and use the equation to convert from potential energy to velocity:
K
e
=
1
2
mv
2
2K
e
m
= v
2
v =
_
2K
e
m
So with that done, we need to know K
e
and m. m is the mass, which for an electron is about 9.109 10
31
kg. K
e
can be found easily by the denition of the electron volt (eV):
1 eV = 1 e 1 V = 1.602 10
19
J
where e is the elementary charge. We have a potetial dierence of 150 volts, so
K
e
= 150 V e = 150 1.602 10
19
J
If we substitute that in the velocity equation, we get
v =
_
2 (150 1.602 10
19
)
9.109 10
31
7.264 10
6
m/s
... which is our answer for (a).
(b) A proton is released with zero speed from A. What will its speed be as it reaches B? Express your
answer in m/sec.
158
The charge of a proton is exactly the opposite of the electron, i.e. the magnitude is the same. Since the
question is now reversed (A to B instead of B to A), the only thing we need to charge from the answer
above is the mass. A protons mass is about 1.673 10
27
, so:
v =
_
2 (150 1.602 10
19
)
1.673 10
27
1.69 10
5
m/s
Due to the much greater mass, the proton arrives with a much lower velocity than the electron.
(c) We now change the potential of B to V
B
= +25 volts. For this new conguration, what is the ratio of
the speed of the electron (as it arrives at A) and the speed of the proton (as it arrives at B)?
To solve this one, we recalculate the potential dierence to be of magnitude 85 volts, and substitute that
in our equations:
v
electron
=
_
2 (85 1.602 10
19
)
9.109 10
31
5.464 10
6
m/s
v
proton
=
_
2 (85 1.602 10
19
)
1.673 10
27
1.275 10
5
m/s
The ratio, and answer, is then
v
electron
v
proton

5.464 10
6
1.275 10
5
42.85
... which answers (c), and we are done!
13.1.6 Problem 6: Two conducting hollow cylinders
Two conducting thin hollow cylinders are co-aligned. The inner cylinder has a radius R
1
, the outer has a
radius R
2
. Calculate the electric potential dierence V(R
2
) - V(R
1
) between the two cylinders. The inner
cylinder has a surface charge density of
a
= , where > 0, and the outer surface has a surface charge
density of
b
= 3.
The cylinders are much much longer than R
1
. Thus, you may ignore end eects and neglect the thickness
of the cylinders.
(a) What is the electric potential dierence between the outer cylinder and the inner cylinder V(R
2
) -
V(R
1
)?
Okay, lets get to work! We choose a co-axial Gaussian cylinder and place it in between the two cylinders,
such that it encloses the inner, but not the outer cylinder.
Because they are co-axial, we dont need the surface integral from Gausss law, but can use EA instead:
EA =
Q
ins

0
A is the surface area of our Gaussian surface 2r, where r is the radius and the length.
What is Q
ins
? Well, its the enclosed surface area of the inner cylinder, 2R
1
, times the charge density

a
= : 2R
1
. Thus we have
2rE =
2R
1

0
We can cancel out a few terms, and solve for E:
159
rE =
R
1

0
E =
R
1

0
r
This is the the electric eld for R
1
< r < R
2
, i.e. between the cylinders. Keep in mind that this is not
valid for the later sub-questions!
Now then. The question was about the potential dierence, so we integrate the eld dot the direction

dr,
from R
2
to R
1
:
V(R
2
) V(R
1
) =
_
R
1
R
2

R
1

0
r


dr =
R
1

0
_
R
1
R
2
dr
r
Because we know

dr and

E to both be radially outwards, we reduce the integral to a very, very simple
one.
_
R
1
R
2
dr
r
= ln r

R
1
R
2
= ln R
1
ln R
2
= ln
_
R
1
R
2
_
Lets add in the rest, the constant stu we had before the integral:
V(R
2
) V(R
1
) =
R
1

0
ln
_
R
1
R
2
_
That answers question (a)!
(b) What is the magnitude of the electric eld outside the cylinders, r > R
2
?
Well, lets do what we did for (a), since that worked a treat. Th dierence here is that Q
ins
has changed,
but the calculation is the same other than that. Where we had 2R
1
, we now have a density that
depends on R
1
and R
2
as well as two sigmas. Lets start over with Gausss law:
EA =
Q
ins

0
A is still the area of the Gaussian surface, so A = 2r. Now, lets nd Q
ins
:
Q
A
= R
1
2
Q
B
= 3R
2
2
Q
ins
= Q
A
+ Q
B
= (3R
2
R
1
) 2
Put it all together:
2rE =
(3R
2
R
1
) 2

0
rE =
(3R
2
R
1
)

0
E =
(3R
2
R
1
)

0
r
... which answers (b).
160
(c) What is the electric potential dierence between a point at a distance r = 2R
2
from the symmetry axis
and the outer cylinder V(2R
2
) - V(R
2
)?
Well need to integrate the electric eld again.
V(2R
2
) V(R
2
) =
_
R
2
2R
2
(3R
2
R
1
)

0
r


dr =
(3R
2
R
1
)

0
_
R
2
2R
2
dr
r
We have already solved that integral: it is ln
_
R
2
2R
2
_
. Thus, we have
V(2R
2
) V(R
2
) =
(3R
2
R
1
)

0
ln
_
R
2
2R
2
_
... which answers (c), and we are done!
13.1.7 Problem 7: Speed of an electron
Finally, the last problem of the week. Writing these down in this L
A
T
E
Xdocument has taken a lot of time
- Ive essentially re-solved them from scratch, having only a few scratch equations left from when I solved
them the rst time. That is surely an excellent way to learn and truly understand, though!
An electron is projected, with an initial speed of v
i
= 1.74e +05 m/sec, directly towards a proton that is
essentially at rest. If the electron is initially a great distance from the proton, at what distance from the
proton is its speed instantaneously equal to twice its initial value? Express your answer in meters.
My rst instinct was that more information must be required, but as expected (they wouldnt ask unan-
swerable questions!), we can solve this. Fairly easily, too, once we know how.
We know the initial velocity, and so we can calculate the initial kinetic energy. We also know the nal
velocity - twice the original one - and can calculate the nal kinetic energy. Via conservation of energy,
the dierence is the two is the change in electric potential energy U, which we can relate the the electric
potential V of the proton, to nd the distance between them.
So, lets begin. First, the initial kinetic energy of the electron is
K
e
initial
=
1
2
m
e
v
initial
2
K
e
final
=
1
2
m
e
v
final
2
=
1
2
m
e
(2v
initial
)
2
U =
1
2
m
e
(2v
initial
)
2

1
2
m
e
v
initial
2
=
1
2
m
e
_
(2v
initial
)
2
v
initial
2
_
Lets now nd the electric potential due to the proton. That one is easy - we know the formula:
V =
Q
p
4
0
r
The distance between the two is initially great, so we treat it as if the electric potential energy is 0 to
begin with. Thus, U = U. We also know that U = V q, where V is the electric potential due to the
proton, and q is the charge of the electron.
Therefore, we can now set the sum of our two expressions for U equal to zero, and solve for r. The charge
of the electron is q = e, so we get:
1
2
m
e
_
(2v
initial
)
2
v
initial
2
_
+
e Q
p
4
0
r
= 0
Messy, but if we solve for r and substitute Q
p
= +e, we get:
161
r =
e
2
6
0
m
e
v
initial
2
If we stick our values in there, we get
r =
(1.602 10
19
)
2
6 (8.854187 10
12
)(9.109 10
31
)(1.74 10
5
)
2
5.556 10
9
m
That is, about 5.56 nanometers! After travelling for a very long distance, it only reaches twice its initial
speed less than a picosecond before they collide. Wow.
Finally, last question of the week:
In your opinion do we need to worry about the special theory of relativity for the speeds given in this
problem?
The answer is no. The speed is only about 0.1% of the speed of light, so relativistic eect are negligible.
Even at 10% of the speed of light, the relativistic eects are fairly small.
13.2 Week 3
13.2.1 Problem 1: Spherical capacitor
A capacitor consists of two concentric spherical shells. The outer radius of the inner shell is a = 0.6 mm
and the inner radius of the outer shell is b = 2.96 mm.
(a) What is the capacitance C of this capacitor? Express your answer in Farads.
Lets begin by stating that capacitance C =
Q
V
, where V is the potential dierence between the two
spheres. Q is a given, so if we nd V , we should be able to solve this problem easily. To nd V , we can
integrate the electric eld between the two radii, so lets begin by nding the electric eld!
We know since before that a spherical shell has the same electric eld as a point charge would, as long
as youre outside the shell; and we know that the electric eld inside the larger sphere, due to the larger
sphere, is zero.
Therefore the electric eld for a < r < b is given by the charge on the inner sphere a only:

E =
Q
4
0
r
2
r
V
a
V
b
=
_
b
a

E

dr =
Q
4
0
_
b
a
dr
r
2
r
We know that the electric eld from a sphere is radially outwards and thus always parallel to dr, so we
can ignore the vectors and the dot product, and simply integrate
1
r
2
and evaluate at the limits:
V
a
V
b
=
Q
4
0
_
b
a
dr
r
2
=
Q
4
0
_

1
r

b
a
_
=
Q
4
0
_
1
a

1
b
_
The latter part of the result can be simplied:
1
a

1
b
=
b a
ab
And therefore, the reciprocal, which well need very soon, is
1
1
a

1
b
=
ab
b a
162
Now then. Back to our potential dierence result. Now that we have V (the potential dierence V
a
V
b
)
we can divide Q by that, which will get rid of the Q and ip everything else upside down, so we get:
C =
Q
V
= 4
0
ab
b a
We were asked to answer numerically, for a = 0.6 mm and b = 2.96 mm:
C = 4
0
(0.6 10
3
)(2.96 10
3
)
2.96 10
3
0.6 10
3
8.37299 10
14
F
(b) Suppose the Maximum possible electric eld at the outer surface of the inner shell before the air starts
to ionize is E
max
(a) = 3.010
6
Vm
1
. What is the maximum possible charge on the inner sphere? Express
your answer in Coulombs.
Well, the magnitude of the electric eld is (again) given by:
E =
Q
4
0
r
2
If we solve the equation for Q and substitute E = 3 10
6
V/m and r = a = 0.6 10
3
m we should get the
answer:
Q = 4
0
r
2
E = 4
0
(0.6 10
3
)
2
(3 10
6
) 1.2059 10
10
C
(c) What is the maximum amount of energy stored in the capacitor? Express your answer in Joules.
The energy stored can be calculated in these ways:
Energy stored =
1
2
CV
2
=
1
2
QV =
1
2
Q
2
C
Since we know Q and C, we prefer the latter form, and so our answer is
1
2
Q
2
C
8.6838 10
8
J
(d) When E(a) = 3.0 10
6
V m
1
what is the absolute value of the potential dierence between the shells?
Express your answer in Volts.
We have the equation from the potential dierence from question (a), we have the charge of a at this
E-eld strength from question (b), and we have the value of all the constants required to nd the answer:
[V [ =

Q
4
0
b a
ab

_
1.2059 10
10
4
0
_
(2.96 0.6) 10
3
(2.96 10
3
) (0.6 10
3
)

1435.1 V
13.2.2 Problem 2: Coaxial cylinders
A very long cylindrical capacitor consists of two thin hollow conducting cylinders with the same axis of
symmetry. The inner cylinder has a radius a, the outer one has a radius b. You may ignore end eects.
There is a gure that shows the inner cylinder with charge +Q and the outer with charge Q.
(a) What is the capacitance per unit length?
Well, we need to do roughly the same thing as we did with the sphere. First, we want to nd the electric
eld.
Lets set up a Gaussian cylinder with a < r < b. The total charge on the cylinder is Q, so the charge per
unit length is =
Q

. Via Gausss law, we have that


163
EA =
Q
enc

0
A is the surface area, so:
E2r =
Q
enc

0
E =
Q
enc
2r
0
The enclosed charge is the charge per unit length times the length, so:
E =

2
0
r
E =

2
0
r
The cancels - anything else would be bizarre; surely the length we choose for the Gaussian cylinder cant
change the electric eld strength.
We carry out the good old integration to nd the potential dierence:
V
a
V
b
=
_
b
a

E dr =

2
0
_
b
a
dr
r
=

2
0
ln
_
b
a
_
We know that C =
Q
V
, and Q = :
C =

[V
a
V
b
[
=
2
0

ln
_
b
a
_
The question asked for the capacitance per unit length, so the disappears.
C =
2
0
ln
_
b
a
_
(b) Now consider the limit where b is very close to a. Express b as a + ; where

a
1. What is the
capacitance per unit length in that limit? Hint: in that limit you can use the following approximation:
ln (1 + x) x
Express your answer in terms of a, and
0
.
OK, well, we can rewrite with b = and see how that looks:
C =
2
0
ln
_
a+
a
_ =
2
0
ln
_
1 +

a
_
Following the hint, we then ignore the 1 and the whole logarithm:
C
2
0
a

... which is our nal answer for (b).


164
13.2.3 Problem 3: The eect of a dielectric medium on capacitance
Consider a capacitor made of two square plates of side . The distance between the two plates is d.
(a) We insert a dielectric of dielectric constant K > 1 and width a distance x (as in the diagram). What
is the total capacitance of this arrangement?
We should be able to treat this like two separate capacitors. From the diagram shown, it seems to me as
if we then need to add the capacitance of the two, as they would be connected in parallel. Lets try that
and see what happens.
The plates have area , but are then split up into two regions. Sketch it up on paper and it will make a
lot of sense. The capacitor with the dielectric is then made of sides x and , so A
dielec
= lx while the one
without the dielectric is ( x) =
2
lx, which certainly looks sensible.
We have the good old parallel plate capacitor equation:
C =
A
0
d
K
Lets call the capacitor without the dileectric C
1
, and the capacitor with the dielectric C
2
:
C
1
=
(
2
x)
0
d
C
2
=
x
0
d
K
C
total
= C
1
+ C
2
=
(
2
x)
0
d
+
x
0
d
K =
0
_
(
2
x)
d
+
x
d
K
_
(b) The capacitor is now connected to battery which provides a dierence of potential V
0
across the ca-
pacitor. What is the energy stored in the capacitor?
The energy stored is
1
2
CV
2
, so we can mostly copy and paste from the previous answer:
1
2

0
_
(
2
x)
d
+
x
d
K
_
V
0
2
(c) While the battery is still connected to the capacitor, we now move the dielectric slab a bit further in
between the plates, increasing x by an amount . What is the change in the energy stored in the capacitor?
Well, lets see. Lets rst calculate the increase in capacitance. First, the new total will be:
C
new
=
0
_
(
2
(x + ))
d
+
(x + )
d
K
_
If we subtract the two and simplify, we get the dierence in capacitance:
C =

0
(K 1)
d
(The notation becomes slightly awkward as C is not times C, but rather the change in C. Same for
E below.)
We can then use the half-C V squared formula to nd the extra energy stored in C:
E =
1
2
CV
2
=
1
2

0
(K 1)
d
V
0
2
165
That answers part (c), or (c1). Then come the highly related follow-ups, which are unlabeled, so lets call
them (c2) and (c3). First, (c2), with my label:
(c2) What is the work done by the battery while we push the dielectric slab in from its original position
x to x + ? (Make sure you have the correct sign!)
Uh oh! Well, intuitively, I entered the same as for (c1), since it seemed like it would make sense for the
batterys work to be the same as the added energy. Nope!
The correct answer is twice of (c1), so just remove the half factor in the front. The answer to the next
question adds some insight...
(c3) What is the work done by us while we push the dielectric slab in? (make sure you have the correct sign!)
The answer here is the negative of (c1). That is, we do negative work, equal in magnitude to the added
charge on the capacitor. The battery, meanwhile, must provide both the added energy stored in the ca-
pacitor and cover up the negative work we do!
The reason this happens is that due to electrostatic forces, the dielectric is pulled into the capacitor.
13.2.4 Problem 4: Coaxial cable with dielectric
A certain coaxial cable consists of a copper wire, radius a, surrounded by a concentric copper tube of
inner radius c. The space between is partially lled (from b out to c) with material of dielectric constant
K. The goal of this problem is to nd the capacitance per unit length of this cable. You may neglect edge
eects.
Note that there is space a < r < b which is vacuum! Thus, we can not simply calculate a single
electric eld and use that everywhere!
Note that for technical reasons, we use the symbol for charge per unit length, rather than the more
typical . Do not get confused, is not a length!
Argh, okay then. Well see how many times I screw that one up before I remember.
a1) Assume that the copper wire has uniform positive charge per unit length and the copper tube has
uniform negative charge per unit length on its inner surface . Calculate the radial component of the
electric eld in the region 0 < r < a.
Well, since there is no mention of a applied potential dierence, and this is conductor, this is the likely
the easiest sub-question of this week: the answer is zero.
a2) Calculate the radial component of the electric eld in the region a < r < b.
Frankly, Ive already done that in earlier examples, so Ill just nick the answer from there. The equation
looks like the one for electric potential due to a point charge, except Q is replaced by the charge per unit
length, and its multiplied by two. So:
E
r
=

2
0
r
Calculate the radial component of the electric eld in the region b < r < c.
In other words, inside the dielectric. Well, we just divide the above by the dielectric constant. No more,
no less.
166
E
r
=

2
0
Kr
Calculate the radial component of the electric eld in the region r > c.
Ah, nally some work to do. Well, not really. We apply Gausss law and nd no net enclosed charge (+
and per unit length, respectively, for the two cylinders), so the electric eld outside is also zero. Too easy.
(b1) What is the potential dierence V(b) - V(a) (be careful about sign)?
Well, is there any work this time around, then? Yes! There is an electric eld in the region, that we can
integrate. It is, as per above:
E
r
=

2
0
r
As usual, the eld is radial and the

dr vector we would use too, so this is a simple integral:
V
b
V
a
=
_
a
b

2
0
r
dr =

2
0
_
a
b
dr
r

a
b
=

2
0
ln
_
a
b
_
(b2) What is the potential dierence V(c) - V(b) (be careful about sign)?
Clearly, we need to integrate just as above, but with the eld that is also divided by K (see above), and
from c to b instead of b to a. We can thus add K to the solution and rename some variables:
V
c
V
b
=
_
b
c

2
0
Kr
dr =

2
0
_
b
c
dr
r

b
c
=

2
0
K
ln
_
b
c
_
What is the magnitude of the potential dierence |V(c) - V(a)|?
Uh oh! We have two dierent electric elds between those points, so we need to integrate twice. Or
integrate zero times, perhaps, because we already have. Lets see then, we want
V
c
V
a
=
_
a
c

E

dr
... but there is no one

E between those points, so we integrate in two parts:
[V
c
V
a
[ =

_
b
c

E
cb


dr +
_
a
b

E
ba


dr

We just solved those integrals:


[V
c
V
a
[ =

2
0
K
ln
_
b
c
_
+

2
0
ln
_
a
b
_

Lets factor out that term:


[V
c
V
a
[ =

2
0
_
1
K
ln
_
b
c
_
+ ln
_
a
b
_
_

Now we just need to take the absolute value of that expression, since the absolute value bars wont be
accepted as part of the answer.
is the magnitude of the charge per unit length, so that must be positive; all the other constants are
positive. Only the natural logs could cause this to become negative... so when are they zero? They are
when the argument is less than one. Since a < b < c, what can we say about the ratios inside? It seems
they will both become less than one!
167
Therefore, we need to negate the whole thing (or otherwise turn it positive, by mucking around with
signs/order inside):
[V
c
V
a
[ =

2
0
_
1
K
ln
_
b
c
_
+ ln
_
a
b
_
_
(c) What is the capacitance per unit length?
If we temporarily ignore this problems variable names, and use sensible ones:
C =
Q
V
Q =
where is the charge per unit length, and is the length. V is the big chunk we found above, the potential
dierence between the inside and the outside, so to speak.
In this problem, is charge per unit length. Due to this confusion, I will temporarily use x to mean length,
such that Q = x. We then nd the answer to our problem by dividing this Q rst by V above, and then
by x to get per unit length. If it seems nonsensical to introduce a variable only to divide it out, I do so to
us C =
Q
V
as-is. Well then, lets go:
C =
x


2
0
_
1
K
ln
_
b
c
_
+ ln
_
a
b
__
First, lets divide out x. Easy enough, it just disappears from the top there. After that, note that
cancels out - the capacitance doesnt depend on the charge per unit length. So remove the x, turn into
1, and factor out that fraction within a fraction. If we factor out the a fraction from the bottom, it turns
upside-down, so we now have:
C = 2
0
1
_
1
K
ln
_
b
c
_
+ ln
_
a
b
__
Ill leave it at that. Still a bit messy, but better than the original expression.
13.2.5 Problem 5: Capacitor network
Three parallel plate capacitors (C
1
, C
2
and C
3
) are in series with a battery of 100 V. C
1
= 2500F;
C
2
= 2C
1
, C
3
= 3C
1
.
(a) What is the potential dierence over each capacitor and how much charge is on each capacitor?
To solve this problem, we begin by stating an important rule: the charge on capacitors in series is the
same for each capacitor. The magnitude of charge must always be equal on to adjacent plates, whether in
the same capacitor or not, so Q
1
= Q
2
= Q
3
. With that in mind, we can do some equation juggling.
C
i
=
Q
V
i
Q = C
i
V
i
V
i
=
Q
C
i
We also know that the potential dierence of the three must add up to the 100 volts of the power supply,
so we set up that equation:
Q
C
1
+
Q
C
2
+
Q
C
3
= 100
168
Q
C
1
+
Q
2C
1
+
Q
3C
1
= 100
If we solve that equation for Q, we get
Q =
600
11
C
1
So given our values for this problem,
Q =
3
22
0.1363636 C
Since the charge is equal for all three capacitors, this answers 3 out of 6 questions.
Next up is nding the potential dierence across each capacitor. Now that we know Q (and the given C
i
),
this is easy.
V
1
=
Q
C
1
=
3
22

1
2500 10
6
V
2
=
Q
C
1
=
3
22

1
2 2500 10
6
V
3
=
Q
C
1
=
3
22

1
3 2500 10
6
(b) We now connect a 4th parallel plate capacitor (C
4
= 4C
1
) to the battery. One side of the capacitor
is connected to the negative side of the battery, the other side is connected to the positive side of the battery.
Now C
4
is in parallel with C
1
+ C
2
+ C
3
.
Solving for C
4
should be easy, since its essentially independent on the others. We know C = 4C
1
=
4 2500F and V = 100V , so we know Q and V for that one.
Using the same logic, the other capacitors are unaected by the addition of a fourth capacitor parallel to
the battery, so we simply copy and paste our answers for those three capacitors!
(c) We make one more change: We still have our 4 capacitors as above, but we now place a dielectric
(K = 3) between the plates of the 4th capacitor. What is the potential dierence across C4 and what is
the charge on it?
Well, a voltage source is conneted, so V cannot change and will remain at 100 volts.
Because V cannot change, and C must change (C
new
= KC
old
), the charge will also increase by a factor
of K via Q = CV :
Q
4
= KC
old
V = 3 4 2500F
13.2.6 Problem 6: Resistances of conducting wires
Three conducting straight wires (though insulated from each other) are each 2 m long. They are electrically
connected to each other only at their ends points A and B. AB = 2 meter.
The wires are cylindrical. One is made of copper, one of aluminum and one of iron. Their radii are 2, 3
and 4 mm, respectively.
a) What is the ohmic resistance between point A and B? Express your answer in Ohms. (you might need
to look up the resistivity of these elements).
Lets start out by nding the symbolic answer; after that, we can look up the resistivity and calculate the
numerical answer.
First o, when connecting multiple resistors in parallel, the formula to use for the equivalent resistance is:
169
R
equiv
=
1
1
R
1
+
1
R
2
+
1
R
3
+
Looking about for a useful equation to use, I found Pouillets law:
R =

A
where is the resistivity in ohm-meters, the conductors length and A its cross-sectional area. We can
then begin to nd the resistance of each wire alone, using the above formula.
l
Cu
= l
Al
= l
Fe
= 2 m
A
Cu
= (2 10
3
)
2
m
2
A
Al
= (3 10
3
)
2
m
2
A
Fe
= (4 10
3
)
2
m
2
Substistuting in some resistivity values, we nd that
R
Cu
= 1.68 10
8
2
(2 10
3
)
2
R
Al
= 2.82 10
8
2
(3 10
3
)
2
R
Fe
= 1.0 10
7
2
(4 10
3
)
2
The total resistance (which is lower than the resistance of each individual conductor, since there are
multiple ways for the current to ow) is then
R
total
=
1
1
R
Cu
+
1
R
Al
+
1
R
Fe
0.000887591 0.888 m
We now attach at B a straight copper wire which is in electrical contact with the 3 wires at B. This 2nd
copper wire is identical to the one between A and B. It runs from B to C. The distance AC is 4 m.
(b) What is the ohmic resistance between A and C? Express your answer in Ohms.
The resistance of the wire is of course the same as R
Cu
above, so we take R
total
and add another R
Cu
,
since the current must pass through the sum of the resistances this time (they are in series).
R
AC
= R
total
+ R
Cu
0.00356139
We now turn AC into a near perfect circle. A is very close to C but it is not touching C.
(c) What is the ohmic resistance between A and C now? Express your answer in Ohms.
Well, if its not touching, then nothing has changed. Whether the conductor is a straight line or a near-circle
does no dierence, so the answer is same as in part (b).
13.2.7 Problem 7: Resistor network
Ugh, this was more than a pain... I wont actually list the full solution. For nding the currents, I used 1
KCL equation at node B and 2 KVL equations, around each of the two smaller loops (the ABMGA and
BDNMPB).
For node B, I noted that
I
1
= I
4
+ I
2
170
For the KVL equations, I went clockwise, adding resistor voltage drops if the current was in the same
direction I was going, otherwise subtracting. For voltage sources, I added if I hit the + side rst, and
otherwise subtracted.
I
1
R
1
I
4
R
4
V
2
I
1
R
5
+ V
1
= 0
V
3
I
2
R
2
I
2
R
3
+ I
4
R
4
+ V
2
= 0
After solving this system of three equations, you get a RIDICULOUSLY complex expression for each
current, with roughly 14-17 terms per equation. If we substitute the values from the problem into those,
we get the currents:
I
1
=
2442
29431
I
2
=
723
1549
I
3
=
11295
29431
Note that I
3
is negative, indicating the real direction is the opposite of what the problem assumed, so that
current goes downwards.
After that, youre asked to nd three potential dierences, which was even more of a pain, at least using
my approach. Id probably solve this using the node method and superposition, if I did this again.
13.3 Week 4
Theres no howework this week due to the upcoming midterm exam!
13.4 Week 5
13.4.1 Problem 1: Lorentz Force
An electron has velocity components v
x
= +100 m/s, v
y
= 80 m/s, and v
z
= +75 m/s. It enters a
region of uniform B-eld with components B
x
= 10
3
T, B
y
= 7 10
4
T and B
z
= 5 10
4
T.
What are the components of the force acting on the electron? Express your answer in Newtons.
Since no electric eld is mentioned, well assume that Lorentz force here means the magnetic force alone,
so
F
B
= q(v

B)
This problem could probably be solved in a single line of Mathematica, which is what Ill do, for the most
part. Some quick analysis rst, though. Since the force is given by the cross product of the velocity and
the magnetic eld, only components which are perpendicular to each other will be part of the answer.
Indeed, the cross product, in components, is dened as
F
x
, F
y
, F
z
= q v
x
, v
y
, v
z
B
x
, B
y
, B
z
= q B
z
v
y
B
y
v
z
, B
x
v
z
B
z
v
x
, B
y
v
x
B
x
v
y

Very, very ugly... but note how all multiplied quantities are perpendicular to each other. If we use this,
and stick our numbers in, including for q, we get
171
F
x
= 2.003 10
21
F
y
= 4.005 10
21
F
z
= 1.602 10
21
... which is marked as being correct.
13.4.2 Problem 2: Motion of a charged particle in magnetic eld
An ion of charge q and mass m, is accelerated from rest by a potential dierence of V = 25 kV. The
particle then enters a magnetic eld region (with strength B = 0.01 T) where the B-eld is uniform and
perpendicular to the ions velocity. The ion then travels on a circular path with radius R = 2 m.
(a) Write an expression for the mass, m, of the ion in terms of q, B, R and V . Note that the answer
should be a formula, not a number.
We have our good old radius equation, which is
R =

2mV
qB
2
We get what we need if we solve it for m:
R
2
=
2mV
qB
2
R
2
qB
2
= 2mV
m =
qB
2
R
2
2V
(b) Now suppose we have another particle: a positively ionized deuteron (the deuteron mass is 3.3410
27
kg) accelerated through the same voltage and then entering the same magnetic eld. What is the radius
of its trajectory? Express your answer in meters.
This ought to be even easier, we simply put the numbers into the formula we had above, before we solved
it for m:
R =

2mV
qB
2
=

2(3.34 10
27
)(25 10
3
)
(1.602 10
19
)(0.01)
2
3.22869 m
(c) How long does it take the deuteron to move around a full circle once. Express your answer in seconds.
Looking back at the lecture notes, a useful formula is
T =
2m
qB
... which is independent of the velocity, very cool. We need to multiply by to take care of relativistic
correction, but that should not be an issue as an electron (with a much lower mass) with 25 keV is non-
relativistic, so this particle must be, too.
Plugging in the values, we get
T
2 3.34 10
27
1.602 10
19
0.01
1.30998 10
5
s
172
13.4.3 Problem 3: Cyclotron
Consider a deuteron in a cyclotron with eld strength 0.5 T. The deuteron is accelerated twice per rota-
tion by a potential of V = 25 kV.
(a) If the radius of the cyclotron is 2 meter, what is the maximum energy of the deuteron? Express your
answer in Joules (the deuteron mass is 3.34 10
27
kg)
I will again assume that relativistic corrections are not necessary. We can use the radius equation again,
but solve for the velocity, and use that to gure out the energy.
R =
mv
qB
v =
qRB
m

(1.602 10
19
)(2)(0.5)
3.34 10
27
4.7964 10
7
m/s
The energy is then given by
E =
1
2
mv
2

1
2
3.34 10
27
(4.7964 10
7
)
2
3.8419 10
12
J
(b) Starting from a negligibly small velocity, how many full rotations does the deuteron need before it
reaches this maximum energy?
We can gure out the potential dierence required to accelerate it; E = qV , so V =
E
q
:
V =
3.8419 10
12
1.602 10
19
23.982 10
6
V
Divide that by the 50 kV per rotation, and we get 479.64 480 rotations.
(c) What is the time it takes for the deuteron to make one complete rotation when its energy is about
500 keV and when it is about 5 MeV? Ignore possible relativistic eects. Express your answer in seconds.
We have the time formula since before, so
T =
2m
qB
m and q are the same as in the previous question, but B is now 0.5, so:
T
2(3.34 10
27
)
(1.602 10
19
)(0.5)
2.61995 10
7
s
Since the time is independent of velocity, and thus kinetic energy, this is the answer for both questions!
13.4.4 Problem 4: Rectangular current loop
A current I travels counterclockwise through a closed copper wire loop which has the shape of a rectangle
with sides a and b.
What is the magnitude of the magnetic eld at the center, C, of the rectangle? Express your answer in
terms of a, b, I and
0
.
Well, the illustration is hardly required; it only labels the sides, essentially. a is the left/right sides, while
b are the top/bottom sides. Not that it really matters.
Anyway. Due to symmetry, we only need to calculate the contribution due to one a and one b side, multiply
each by two to account for the other, duplicate side, and add the results.
Using the right-hand curl rule, its clear that the eld is additive: all four sides will contribute to a
173
magnetic eld that comes out of the page at the center of the rectangle.
Well, thats about as far as we can go qualitatively. What about some numbers? Or, at least, some
equations? Well have to use Biot-Savarts law for this one.
Remember that

B =

0
4
_
I
r
2
(

d r)
If we draw a diagram, for which I recommend you watch the week 5 problem solving videos, we get
something like this (screen capture from the video):
As shown there, if we call the distance to the point we measure at d, and the distance along the axis x,
we get r =

d
2
+ x
2
via the Pythagorean theorem. We also get that sin =
d
r
, which comes in useful.
The reason we have a sine in there is because of the cross product

d r = [

d[sin, since the magnitude


of r is 1 by denition. If we make that substitution of the sine in the shown [dB[ equation, while also
substituting the square root-value for r, we get the nal equation shown:
[

B[ = 2
_
0.5
0

0
4
Id
dx
(x
2
+ d
2
)
3/2
=

0
2
Id
_
0.5
0
dx
(x
2
+ d
2
)
3/2
The integral is
_
dx
(x
2
+ d
2
)
3/2
=
x
d
2

x
2
+ d
2
(+ C)
... where is the length of the wire segment. We now use this equation to nd the magnetic eld due to
a and b, multiply each by 2 (the 2 in the equation has a dierent origin: the result without in is only for
one-half of the length) and add them up, and we should have our nal answer.
Lets now look at the rectanglular loop we had again. d is the distance to the center, measured from the
middle of the wire. For the b part, that distance is then
a
2
, while it would be
b
2
for the a part. What is
then x (and dx)? Well, thats the coordinate along the axis of the point we are at on the wire.
Lets try to calculate the magnitude of the magnetic eld due to one a (short) side alone. We use the
integral result, and make sure not to forget about the constants we moved in front, and evaluate the
integral at 0.5 and 0. We also multiply the whole thing by 2, to take care of both wire segments:
174
[B
a
[ = 2

0
2
I
b
2
_
_
x
(
b
2
)
2
_
x
2
+ (
b
2
)
2
_
_
x=0.5
x=0
After some painful algebra and simplication, the above is
[B
a
[ =
2aI
0
b

a
2
+ b
2
If we do the same for the b parts, it would be fairly shocking if the result was anything else than the same
thing with a and b swapped, so
[B
b
[ =
2bI
0
a

a
2
+ b
2
And, nally, the net answer is the sum of the two. Lets factor out the common parts rst:
B =
2
0
I

a
2
+ b
2
_
a
b
+
b
a
_
Finally, we can simplify that a bit further; if we rst combine the fractions to the right, we get
a
b
+
b
a
=
a
2
+ b
2
ab
B =
2
0
I

a
2
+ b
2
a
2
+ b
2
ab
=
2
0
I(a
2
+ b
2
)
ab

a
2
+ b
2
Because
a
2
+ b
2

a
2
+ b
2
=

a
2
+ b
2
(think about right triangles, with c
2
= a
2
+b
2
and c =

a
2
+ b
2
; it then essentially says that
c
2
c
= c), this
becomes
B =
2
0
I

a
2
+ b
2
ab
... and we are nally done. Prior to this problem, I thought this weeks homework was uncharacteristically
easy. That changed a bit!
13.4.5 Problem 5: Resistor network
Consider this circuit with the following arrangement:
[Diagram of a cube, with a battery connected to opposite corners of the cube.]
Each edge of the cube is a resistor with resistance R = 10 (there are a total of 12 resistors).
What is the equivalent resistance R
eq
from one corner of the cube to the diagonally opposite corner? (Hint:
think about a steady current I owing from the battery, how does the current split as it reaches and then
moves through the wires making up the cube?)
I had a look at the recommended reading material (helpful content from the book), and... the second
listed item had the answer. Uh, I feel like I just cheated! I looked it through, and sure enough, the problem
is indeed identical, and the books answer is correct for this problem.
Heck, there shouldve been a spoiler warning there!
175
Because of the graphical layout of the problem, this is explained simply in the book, while an explanation
here would be harder to follow. For that reason, I refer to the book (or a search engine, which will no
doubt nd many solutions) for this problem.
For what its worth, the answer is
R
eq
=
5
6
R
13.4.6 Problem 6: Coaxial current loops
Two concentric circular loops have radii R
1
= 9 cm and R
2
= 26 cm. They lie in the same horizontal
plane. The direction of the two currents are opposite. Seen from above the current through the outer loop
is I
2
= 15 A clockwise, the one through the inner loop is I
1
= 10 A counterclockwise.
What is the magnitude [and direction] of the magnetic eld at the center of the loops? Express your
answer in Teslas.
The problem diagram helpfully also shows two unit vectors; z upwards, and r radially outwards, with the
current loops centered in the coordinate system.
Via Biot-Savart:
B =

0
4
_
I
r
2
(

d r)
The cross product is [

d[[ r[ sin = d, since the two are always perpendicular, and the magnitude of the
unit vector is 1:
B =

0
I
4r
2
_
d
The integral of d will come out to be 2r, so
B =

0
I
4
2r
r
2
=

0
I
2r
We add the magnetic elds due to each current loop alone, keeping direction in mind (the els oppose
each other in the middle):
B =

0
2
_
I
1
R
1

I
2
R
2
_
3.3564024 10
5
T
13.4.7 Problem 7: Parallel plate capacitor
A parallel plate capacitor has plates of area A and separation d. We connect the capacitor to a battery
of V volts. We then disconnect [the battery] when the capacitor is fully charged.
(a) How much energy is stored in the capacitor? Express your answer in terms of the following variables
if needed, A, V , d and
0
.
Capacitance is given by
C =
A
0
d
(times , but theres no mention of a dielectric, and we cant use it in our answer either). Stored energy is
U =
1
2
CV
2
=
AV
2

0
2d
176
Alternatively, we can use the eld energy formula:
U =
_
1
2

0
E
2
dV
_
dV = Ad, because the electric eld is zero everywhere else. Thus the integral, which is over all space,
is just over the volume enclosed by the capacitor plates. Thus we have
U =
1
2

0
E
2
Ad
V = Ed, so E =
V
d
U =
1
2

0
_
V
d
_
2
Ad
U =
AV
2

0
2d
Unsurprisingly, we get the same answer either way!
We now increase the plate separation by a small amount dz.
(b) What now is the Electric eld in the capacitor? Express you answer in terms of the following variables
if needed, A, V , d, dz and
0
.
Remember that the battery has been disconnected! V can and will vary as we separate the plates! The
charge Q on the plates is xed, however. Q = CV ; C will go down as d increases, and so V must go up
to compensate.
What about the electric eld? E =
V
d
, but also E =

0
. The latter makes it clear that the electric eld
must stay the same, so the answer is
E =
V
d
(since we arent allowed to use or Q in the answer).
(c) What now is the energy stored in the capacitor? Express you answer in terms of the following variables
if needed, A, V , d, dz and
0
.
Okay; so energy stored is given by many dierent equations:
U =
1
2
CV
2
=
1
2
Q
2
C
=
1
2
QV =
AV
2

0
2d
Since all must obviously be true, we can pick the one(s) that suits us. From the rst, it looks as if the
stored energy has gone up, as V has increased by the same factor C has decreased, but V is squared.
From the second, it also looks like its gone up. Same for the third. And fourth. So clearly, we did work
in separating the plates, and the energy has increased.
To what?
Looking at U =
1
2
QV , we know that Q in constant and V increases linearly with the distance. Therefore
U must also increase linearly, so
U
new
= U
old

d + dz
d
177
U
new
=
AV
2

0
2d

d + dz
d
=
AV
2

0
(d + dz)
2d
2
(d) What force do I have to apply on one plate (the other plate is xed), to separate the plates by the
amount dz? Express you answer in terms of the following variables if needed, A, V , d, dz and
0
.
At rst glance, this confuses me a bit. I guess the force required is an innitesimal amount higher than
the force pulling the plates together, but how do we specify that tiny extra bit? Lets analyze it, though.
Force times distance is work, so if we gure out the work required, that should help us out. Okay.
Well, the plates attract each other via their electric elds. However we must be careful to not use the sum
of their two electric elds; objects cant exert a force on themselves, so only the force from one plate on
the other will matter.
I will imagine that we move the bottom plate downwards, and that the bottom plate has negative charge
on it.
The electric eld that the bottom plate is in, due to the top plate, is
E =

2
0
Since the plates contribute equally to the total eld of

0
=
V
d
, the eld due to the top plate should be
E
top
=
V
2d
The magnitude of the force pulling the bottom plate upwards is then
F =
V
2d
Q
We can nd Q by knowing C and U:
U =
1
2
Q
2
C
2UC = Q
2
Q =

2UC
Now, clearly, if we try to nd the work by multiplying this by the distance moved, and then divide away
the distance to get force, we get back what we have. So lets substitute Q in there:
F =
V
2d

2UC =
V
2d
_
2(
AV
2

0
2d
)(
A
0
d
)
=
V
2d

2
_
A
2
V
2

2
0
2d
2
_
=
V
2d
_
AV
0
d
_
=
AV
2

0
2d
2
... and that is indeed accepted as the correct answer!
178
13.5 Week 6
13.5.1 Problem 1: Amperes law in action
A long coaxial cable consists of two concentric conductors. The inner conductor is a cylinder with radius
R
1
, and it carries a current I
0
uniformly distributed over its cross section. The outer conductor is a
cylindrical shell with inner radius R
2
and outer radius R
3
. It carries a current I
0
that is also uniformly
distributed over its cross section, and that is opposite in direction to the current of the inner conductor.
Calculate the magnetic eld

B as a function of the distance R from the axis.
Well, the problem name says what we should do.
First, lets think about what will happen with the magnetic eld inside the outer shell, due to the outer
shell.
We choose an Amperean circle with radius r = R
2
, and an open surface that is just the circles area, and
get
2rB
inside
=
0
_
I
pen
+
0

d
dt

E

dA
_
Since we are only considering the eld due to the outer shell, I
pen
is zero, since the surface is smaller than
the shell. What about the electric ux? That is also zero - the electric eld is zero inside! Even if it
werent, there would be zero net ux through the Amperean surface, since the electric eld is radially
outwards (+ r) from the cylinder, and in the same plane as the circle. And even if that were not the case,
the ux should be constant, so its derivative would be zero. In short: the magnetic eld inside due to the
shell must be zero, so we can ignore it for the calculations where r < R
2
.
Lets start looking at the questions.
What is the direction and magnitude of the magnetic eld for 0 < r < R
1
? Express your answer in terms
of I
0
, R
1
, R
2
, R
3
, r and
0
.
Inside the conductor, in other words. Well, we use Amperes law again, as usual.
The electric ux will, again, be zero in our circle, so we can use the old, simplied version of the law, as
the second term is known to be zero either way. The left-hand side is just the integral around the circle,
as usual, so we start out with
2rB =
0
I
pen
What is I
pen
, though? Since r < R
1
, it will be less than the full current. Namely, it will be the ratio of
the areas times the full current:
2rB =
0
r
2
R
2
1
I
0
B =

0
I
0
2
r
R
2
1
The direction is given by the right-hand rule, so it will be in the + direction.
What is the direction and magnitude of the magnetic eld for R
1
< r < R
2
?
We barely need to re-solve it; instead, consider that I
pen
= I
0
, and so we only need to solve the rst
equation we had with that substitution:
179
2rB =
0
I
0
B =

0
I
0
2r
The direction is unchanged.
What is the direction and magnitude of the magnetic eld for R
2
< r < R
3
?
Ah, now the outer shell begins to matter. The magnetic eld due to the inner wire will be exactly the
same as the previous answer, however!
Once again, the second term in the law will be zero - there will be no magnetic ux penetrating the surface
we choose, so the change in magnetic ux will certainly also be zero. We set up the simple one instead:
2rB
shell
=
0
I
pen
Ah, that I
pen
again... Well, this time, it will be the ratio of the area of the part of the shell inside r, to
the full shells area. Also, its negative, since its in the opposite direction as before. That is
I
pen
= I
0
r
2
R
2
2
R
2
3
R
2
2
For the top term, we have the area of the circle with radius r, minus the area of R
2
which is the void
inside. Same for the outer part. Simplied, and substituted into Amperes law, we have
2rB
shell
=
0
r
2
R
2
2
R
2
2
R
2
3
I
0
... keeping the minus sign in mind. So, the magnetic eld due to the shell, inside the shell, is
B
shell
=
0
r
2
R
2
2
2r(R
2
2
R
2
3
)
I
0
The direction of this is then in the + direction as well, since we used the minus sign for the current.
The net magnetic eld inside the shell is the above, plus the old one:
B =
0
r
2
R
2
2
2r(R
2
2
R
2
3
)
I
0
+

0
I
0
2r
(for R
2
< r < R
3
)
Simplied, we get
B =
I
0

0
(r
2
R
2
3
)
2r (R
2
2
R
2
3
)
(for R
2
< r < R
3
)
And nally:
What is the direction and magnitude of the magnetic eld for r > R
3
?
This one is the easiest, actually!
To see why, imagine an Amperean circle, with its usual simple at surface attached to it. The circle is
larger than the cable (center + shell), and so I
pen
= +I
0
I
0
= 0! The I
pen
term is zero, and the electric
ux is zero (and thus the ux change is zero), for the same reasons weve seen before.
The net eld outside is zero!
180
13.5.2 Problem 2: Intuition breaks down
(a) In the diagram above the small semi-circle in the wire above R1 indicates that it is not in contact
with the horizontal wire which it appears to intersect due to the 2-dimensional projection. Additionally,
the horizontal wire is continuous, not broken.
Calculate [V
1
[ in terms of the emf E, R
1
, R
2
.
Oh boy, non-conservative elds and weirdness. Lets see how this goes.
We can NOT use Kirchos rule here, which is the entire point. We can however nd the current, and use
Ohms law to nd voltage drops, so lets see... I think were to assume that I
1
= 0 and I
2
= 0 since there
is no mention about the volt meter resistance, etc. Lets see if that assumption gives us correct answers.
I =
E
R
1
+ R
2
The magnitude of the voltage on V
1
would then be that current times R
1
... except that the voltmeter
is also wound around the changing ux, so we get a +E, as there will be an induced EMF inside the
voltmeter as well, separately from the current loop I.
[V
1
[ =
E
R
1
+ R
2
R
1
+E
They then ask for [V
2
[; that is trivial: it is the current times the resistance, nothing more, nothing less.
Why? Because the voltmeter isnt part of a loop where there is changing magnetic ux, so we get nothing
but basic circuit analysis rules.
[V
2
[ =
E
R
1
+ R
2
R
2
Lastly, they want the ratio of the two, so we divide them and get
[V
1
[
[V
2
[
= 1 +
2R
1
R
2
Now wrap that same wire not once around the circuit but 100 times before you connect it again to the
D-side. Without realizing it, in doing so you have been building some kind of a transformer (we will
discuss transformers later in the course).
(b) Calculate [V
1
[ [and [V
2
[, and their ratio] in terms of the 1 loop emf E ,R
1
, R
2
.
Cool. Its immediately obvious that [V
2
[ doesnt change, as that is still connected directly, and there will
be no extra EMF there. There is also no extra EMF that increases the current through R
1
, so only the
voltmeters extra EMF will change, from E to 100E. That indeed gives the correct answers, yay! The
three answers are then
[V
1
[ =
E
R
1
+ R
2
R
1
+ 100E
181
[V
2
[ =
E
R
1
+ R
2
R
2
[V
1
[
[V
2
[
= 100 +
101R
1
R
2
13.5.3 Problem 3: Helmholtz coils
A current I = 2 Amperes ows in a circular single loop coil of radius R = 10 cm.
(a) Evaluate the magnitude of the magnetic eld (in Tesla) at a point P on the axis of the coil, and a
distance z = 5 cm away from the plane of the coil itself.
Magnetic eld due to a circular current loop
Superposition is valid for magnetic elds, so we can start out by making the calculation for a single ring.
Unfortunately, that is not easy by itself - and the rest problem gets a lot worse from there. I suggest using
some form of computer math software for the rest of the problem... I use Mathematica, but there are
plenty of other alternatives, many (cost-)free and/or free+open source.
This part isnt too hairy mathematically, though.
For a better explanation than mine below, watch this video: http://www.youtube.com/watch?v=lN296gUXkl4
Since the solution (both the process and the nal equation) are in the book, watching that is hardly any
more cheating than reading the textbook is. In fact, I could simply use the books equation, but that
wouldnt make me learn, so I derive it myself as far as possible.
Now, then. Well have to use Biot-Savart, to begin with.

dB =

0
4
I
r
3
(

d r)
This would all be easy for z = 0, which we have done for a previous homework. Since we cant make that
simplication, r is no longer just the radius of the circle, but the distance from a point on the circle to
P, which is at height z above the center of the loop. This probably needs to be drawn on paper to make
complete sense.
What else? Well, the direction of r also changes; its no longer simply the vector pointing inwards towards
the middle of the circle. However, as long as we stay on the z-axis, via symmetry arguments, this should
all cancel out.
That is, the elds direction, if we stay along the z-axis, should be simply + z.
Drawing this out and using the Pythagorean theorem, we nd that the distance r to the point is given by
r =

R
2
+ z
2
This makes an angle with the xy-plane.
One more thing. If we look at the symmetry of this problem, its clear that only the z component isnt
cancelled out (by the symmetry above). Therefore, are only interested in the z component of the eld
We then have
dB
z
=

0
4
I
(R
2
+ z
2
)
3/2
(d[r[) cos
where the cosine comes from vector decomposition, to get the z-component.
Via geometry, we can see that
cos =
R
r
=
R

R
2
+ z
2
182
Therefore we have
dB
z
=

0
4
I
(R
2
+ z
2
)
3/2
d[r[
R

R
2
+ z
2
dB
z
=

0
4
I
(R
2
+ z
2
)
3/2
d

R
2
+ z
2
R

R
2
+ z
2
dB
z
=

0
4
IR
(R
2
+ z
2
)
3/2
d
We can nally integrate that:
B
z
=
_

0
4
IR
(R
2
+ z
2
)
3/2
d =

0
4
IR
(R
2
+ z
2
)
3/2
_
d
B
z
=

0
2
IR
2
(R
2
+ z
2
)
3/2
Thus, we can answer part (a) by plugging in the values given.
Two circular loops
Now consider two N-turn circular coils of radius R, each perpendicular to the axis of symmetry, with
their centers located at z =

2
. There is a steady current I owing in the same direction around each
coil, as shown in the gure below.
The image shows a similar situation to what we had, then, but with two coils, above and below z = 0,
where the old one was.
(b) Assuming the N = 100, I = 2 Amperes, R = 10 cm, = 2 cm, calculate the magnitude of the
magnetic eld (in Tesla) at a distance z = 0.5 cm from the midpoint between the centers of the two coils.
The eld due to N loops is simply N times stronger, as each loop simply adds to the rest. (I suppose we
can think of it as integrating over the circle N times, so we integrate from 0 to N 2, which then just
spits out a factor N.)
With that in mind, and superposition, this part should be easy enough. When using the equation we just
derived, the z value is really the distance from the center of the ring, which isnt at z = 0 in the coordinate
system any longer. We need to adjust it slighly; same goes for the circle below. Ill call the top one 1 and
the bottom one 2:
B
z1
=

0
2
NIR
2
(R
2
+ (

2
z)
2
)
3/2
B
z2
=

0
2
NIR
2
(R
2
+ (

2
z)
2
)
3/2
The sum is then
B
z
=

0
NIR
2
2
_
1
(R
2
+ (

2
z)
2
)
3/2
+
1
(R
2
+ (

2
z)
2
)
3/2
_
If we just plug in the numbers given now, and sum the two answers, we get the answer they want.
(c) Evaluate the rst derivative with respect to z of the magnetic eld calculated in part (b) using the
same values for the parameters. Express your answer in Tesla per meter.
183
Goodness, this is why I recommend using math software. Im not dierentiating that - the z is hidden
inside two exponents, in the denominator of a fraction - and they want the second derivative later on, too!
The rst derivative is given by
dB
z
dz
= 4
0
NIR
2
_
6(l 2z)
((l 2z)
2
+ 4R
2
)
5/2

6(l + 2z)
((l + 2z)
2
+ 4R
2
)
5/2
_
The second derivative is given by
d
2
B
z
dz
2
= 4
0
NIR
2
_
60(l + 2z)
2
((l + 2z)
2
+ 4R
2
)
7/2

12
((l + 2z)
2
+ 4R
2
)
5/2

12
((l 2z)
2
+ 4R
2
)
5/2
+
60(l 2z)
2
((l 2z)
2
+ 4R
2
)
7/2
_
What a friggin nightmare! My heart goes out to the students that solved this on their own. (I just noticed
that these equations are in the book. Im sure some worked it out manually, though.)
Once again, though, plugging in the values gives the correct answer.
The rest of the questions are for the same problem, with R = l = 10 cm and z = 0, so we just plug in the
numbers to solve those, the point being that the rst and second derivatives are both exactly zero in that
case.
13.5.4 Problem 4: Spinning loop in a magnetic eld
A square loop (side L) spins with angular frequency in eld of strength B. It is hooked to a load R.
(a) Write an expression for current I(t) in terms of B, L, R, and t.
Well, we can nd the EMF using Faradays law. Its the negative of the change in magnetic ux through
the loop. So lets rst nd the ux through the loop; we have a planar surface, and a uniform, constant
magnetic eld of strength B (since there is no other information). Magnetic ux is B-eld strength times
area, so

B
= L
2
Bcos
... where is the angle between B and the normal to the surface.
Since we are rotating the loop, changes with time according to the angular frequency :
=
0
+ t
We can choose
0
= 0 to simplify, so we do that, and then dierentiate and negate it to nd the EMF:

B
= L
2
Bcos t
d
B
dt
= L
2
B sin t
E(t) =
d
B
dt
= L
2
B sin t
The current is then this, divided by the resistance R:
I(t) =
L
2
B sin t
R
(b) How much work is done by the generator per revolution? Express your answer in terms of B,L,R and
(enter omega for ).
184
The work done will be equal to the heat dissipated in the resistor, so we could integrate that power over
the time of one rotation.
=
2
T
, so T =
2

The power is given by I


2
R:
P(t) =
_
L
2
B sin t
R
_
2
R =
L
4
B
2

2
sin
2
t
R
The work is the integral, then, from t = 0 to t = T:
W =
_
T
0
L
4
B
2

2
sin
2
t
R
=
L
4
B
2

2
R
_
T
0
sin
2
t
When we integrate the sine, and substitute T =
2

, the result of the integral is simply


, so we get:
W =
L
4
B
2

R
(c) To make it twice as hard to turn (twice as much work), what factor would you have to multiply the
resistance R?
Its clear from the equation just above that, since no other variable is dependent or R, we simply need to
halve R, so the answer is
1
2
.
13.5.5 Problem 5: Loop in a magnetic eld
A rectangular wire loop of dimensions a b, with a = 3.5 cm and b = 4.5 cm, is pulled parallel to side a
at a constant speed v = 4.3 m/sec into a region where the magnetic eld, [

B[ = 0.125 T, is uniform and


perpendicular to the loop. The loop enters the magnetic eld at time t = 0.
(a) Calculate the rate at which the magnetic ux through the loop is changing. Express your answer in
Tesla
m
2
sec
.
I wouldve preferred an illustration, but all right. I assume that pulled parallel to side a means there is
some external force in the same direction as the line a is in. That makes the area of the loop exposed to
the magnetic eld ba(t). The ux is

B
= b a(t)B
The rate of change is then
d
B
dt
= b
da
dt
B = bvB
(b) What is the magnitude of the induced EMF at time t?
This confuses me, since they are expecting a numerical answer... The magnitude of the EMF is the same as
the rst answer, though, which is accepted. I suppose the point they want to make is that it is independent
upon t? However, as soon as the loop has entered the magnetic eld fully, surely the ux would become
constant.
185
13.5.6 Problem 6: Electrodynamic tether
A L = 3 mile long conducting cable tethers a satellite to the International Space Station. The Cable
moves through the Earth magnetic eld of about B = 0.02 Gauss at a speed of v = 8 km/sec. The cable
points radially from the Space Station to the satellite. The ISS moves Eastwards and the magnetic eld
is Northward.
What is the potential dierence between the two ends of the tether?
This is really nothing more than a simple motional EMF problem. Unless Im missing something (despite
having the answer right), most of the work is in the unit conversion, really. Most of this solution is restat-
ing facts about motional EMF.
Say we have a conductive bar moving rightwards, in a magnetic eld going into the page. The free electrons
in the bar experience a magnetic force upwards:

F
B
= q(v

B)
Positive charge gathers closer to the top, and negative charge towards the bottom.
This produces an electric eld inside the bar, from positive to negative as usual, which forces positive
charges down, and negative charges up, with force

F
E
= q

E
In an static situation of constant magnetic eld and velocity, well nd an equilibrium where the forces
are equal:
q(v

B) = q

E
v

B =

E
We evaluate the cross product, which is simply vB, since the angle between them is 90 degrees, so the
sin term in the cross prodcut is one.
E = vB
Given that the electric eld should be constant (since v and B are constant), the potential dierence is
then given by V = Ed, where d is the distance, here
V = vB
Going back to our problem, we can now solve for the potential dierence:
V = vB = (8000
m
s
)(0.02 10
4
tesla)(3 miles
1609 m
1 mile
) = 77.232 V
Which end has the higher potential? If the velocity is eastwards, and the B-eld northward, then the
magnetic force is upwards (outwards from the surface), so the positive charge gathers at the far end, so
the potential is highest at the end the farthest away from the Earth.
13.5.7 Problem 7: Motor
A simple motor has N = 250 turns of wire on a coil measuring 12 cm 4 cm. Its resistance is R = 10
ohms. The magnetic eld is B = 1 tesla.
(a) How much current (in Ampere) in the coil is needed to produce a maximum torque of 40 N m?
186
Motors arent my strong side - they havent been talked about a whole lot in the course, and I didnt take
8.01 prior to this course, so Im not super-familiar with torque... so I had a look in the suggested reading.
The torque on the loop is given by
[[ = IAB
... where A is the area of the loop.
In this case, the current will be given by NI, as the multiple windings simply multiply the torque. So, we
have
[[ = NIAB
I =
[[
NAB
Now we just need to stick the numbers in, and nd I =
100
3
= 33.3333 amperes.
(b) What maximum EMF (in Volts) is needed to drive the motor at 50 Hz if the current is constant?
Lets see. There will be an induced EMF due to the changing magnetic ux through the loop, as it rotates.
The current through the loop (due to the external source) in the books drawing is counterclockwise, so
lets assume that it will be at one instant, just for visualization.
What about the induced EMF, in which direction will that one be? In the one that opposes the change
in the external ux, which turns out to be in the opposite direction, so the induced EMF will counteract
the external one the power supply provides.
What will the induced EMF be?
The ux through the surface is

B
= NABcos
Since it rotates, as weve done before, we set = t, and then dierentiate the ux, and negate it, to nd
the EMF:
d
B
dt
= NAB sin t
E =
d
B
dt
= NAB sin t
The EMF maximum will then occur when sin t = 1, so the maximum induced EMF will be NAB.
= 2f = 2 (50 Hz) = 100, so the maximum induced EMF is
E = (250)(4 cm)(12 cm)(1 T)(100 rad/s) 377 V
Heres the interesting point. The above is not only the maximum induced EMF, but actually also the
answer to the question.
Motors and generators are strongly related. If we think of this motor as a generator, the maximum EMF
generated at angular frequency is the same as the maximum EMF required to turn the motor at the
same angular frequency .
Im still unsure of exactly why this is the case, though. Im hoping next weeks talk about inductance and
phase lag help a bit.
187
13.5.8 Problem 8: Auroral zone
This problem has a very long introduction, with a derivation of the equation used, etc, and several gures.
I suggest reading that rst!
... The solar wind ram pressure causes the magnetic eld of the earth to terminate at about 10R
E
on the
sunward side of the earth (see gure).
The auroral zone is dened by the last eld line from the Earth that returns to the Earth. If the eld
of the earth extends no further than 10R
E
in the sunward direction, at what angle (in degrees) in the
picture above does the last eld line that returns to the earth on the sunward side leave the polar regions
of the earth at 1R
E
, assuming that the elds are always described by r = R
0
sin
2
? In the drawing, this
is equivalent to asking what is the angle with respect to the vertical of a line from the center of the earth
to the point that the red curve intersects the light blue circle.
Admittedly, I rst solved this one by graphing it, and will solve it more mathematically below.
In polar coordinates, r = 1 gives a circle; if we choose our coordinate system such that R
E
= 1, this then
represents the Earth.
The eld lines are then plotted by the equation given, r = R
0
sin
2
, with various values for R
0
. When
R
0
= 10, we get that last eld line the questions asks about.
When weve graphed both on the same plot, we can simply zoom in and read o the angle. Well, almost -
polar coordinates dene angles as 0 on the +x axis, and so to nd the answer - the angle from the vertical
axis, we need to shift it. Additionally, this angle will be in radians, while they requested degrees.
Anyhow, I read it o as 0.3975 from the horizontal axis; with the 90 degree =

2
conversion, and then
from radians to degrees, the answer is approximately
_

2
0.3975
_
180

rad
18.45

It turns out the mathematical solution isnt very much harder. We just set up
R
E
= 10R
E
sin
2

... since those are the r coordinates where the lines meet, and solve the equation. The solution pops out:
10 sin
2
= 1
sin
2
=
1
10
sin =
1

10
= arcsin
1

10
18.43495

Keep in mind that


sin
2
= (sin )(sin )
if that part was unclear.
13.6 Week 7
13.6.1 Problem 1: Magnetic energy of a solenoid
Compare the total magnetic energy in two solenoids, each with N turns, area A, and current I, but length
x and 2x. Denote U
x
the total magnetic energy in the solenoid with length x, and U
2x
the one in the
188
solenoid with length 2x. What is U
2x
/U
x
?
We calculated the magnetic eld energy density back in this weeks lecture notes, and since it featured a
fairly ugly integral, a circuit to solve etc., I wont redo it all here, as I would if it were simpler.
Instead, we found the inductance L of a solenoid and the eld energy density U to be
L = r
2
N
2


0
U =
B
2
2
0
The B-eld inside the solenoid can be derived relatively easily using Amperes law, which we have done
previously (week 5), so I will do the quick-and-dirty version here. There is no displacement current through
the Amperean surfacae we choose, so we can use the simplied Amperes law:
_

B

d =
0
I
pen
Ba =
0
Ia
N

B =
0
I
N

We then want to square that, and then divide by 2


0
, to nd the energy density:
U =
1
2
0
_

0
IN

_
2
=

0
I
2
N
2
2
2
The we have used is the length of the solenoid, which then is to be x or 2x.
Since the B-eld is assumed to be constant inside a solenoid, and 0 outside, the total magnetic energy is
the above times the volume, A, where we substitute in the value for :
U
x
=

0
I
2
N
2
2x
2
Ax
U
2x
=

0
I
2
N
2
2(2x)
2
A2x
Finding the ratio U
2x
/U
x
, we rst get rid of all the terms that are the same in both expressions, and get
U
2x
U
x
=
A2x
(2x)
2
_
Ax
x
2
=
A2x
4x
2

x
2
Ax
=
2Ax
3
4Ax
3
With all that cancels, we nd
U
2x
U
x
=
1
2
Since the total energy is proportional to
1

(the U
x
= ... equation above), this does appear to make sense.
The energy density is proportional to
1

2
, but we then multiply that by a term to nd the total energy.
189
13.6.2 Problem 2: Displacement current
Consider the process of charging a parallel plate capacitor with circular plates of radius R = 5 cm sepa-
rated by a distance d = 0.2 cm. At some time t
1
, the capacitor is being charged with a current I = 0.035 A.
Consider a point P on the plane which is equally distant from the two plates and is a distance r = 0.068
m away from the axis of the capacitor.
(a) Calculate the magnitude of the magnetic eld (in Tesla) at a point P at time t
1
during the charge of
the capacitor.
We will want to use the (almost) full version of Amperes law for this one. (Next week we will nally have
the fully complete version.)
_

B

d =
0
_
I
pen
+
0

d
dt


E

dA
_
We choose our Amperean loop as a circle with radius r, concentric with the capacitor plates, and centered
in the distance between the plates as mentioned. We must then attatched an open surface to our closed
loop (the circle), and we choose a at surface in the plate of the circle (think of it as the circles area) for
simplicity.
The left-side integral becomes B2r; I
pen
= 0 since the surface is in in mid-air and no real current can pass
through it. The rate of change of the electric ux through the surface will however be nonzero! Replacing

E with its magnitude


Q

0
R
2
and the area the ux goes through by R
2
(capital R! r > R, so the ux
only goes through part of r
2
!) we get, with all our changes:
B2r =
0

d
dt
_
Q

0
R
2
R
2
_
The dot product inside the integral turns into a regular multiplication, and the integral is trivial since the
E-eld is uniform, so all innitesimal products

E

dA are the same.
In taking the time derivative, we nd that the only non-constant is Q, which will change over time, so
d
dt
_
Q

0
R
2
R
2
_
=
R
2

0
R
2
d
dt
(Q) =
1

0
dQ
dt
=
I

0
where I is the current in the wire to (or from) the capacitor, not I
pen
which we established was zero.
Putting it all together and solving for B:
B2r =
0

0
B =

0
I
2r
For the values given,
B =
(4)(10
7
)(0.035)
20.068
=
2(10
7
)(0.035)
0.068
1.0294 10
7
T
13.6.3 Problem 3: RL circuit
Consider a circuit with an RL series with L = 0.09 H and R = 0.05 Ohm. At t = 0 the circuit is connected
to a battery which provides V
0
= 12 V.
(a) How long does it take to the current to equal a fraction 0.95 of the steady state current?
We could set up and solve a dierential equation, but theres little in point in doing that, since we found
the solution already in this weeks lectures. The current is governed by the equation
190
I(t) =
V
R
_
1 exp
_
Rt
L
__
The steady state current is then just
V
R
, as the inductance will have no eect for the steady state.
Thus, we want to nd when the rst equation equals 0.95 times the second:
V
R
_
1 exp
_
Rt
L
__
= 0.95
V
R
1 exp
_
Rt
L
_
= 0.95
exp
_
Rt
L
_
= 0.05
exp
_
Rt
L
_
= 0.05
Rt
L
= ln 0.05
t =
L
R
ln 0.05
For the values given, t 5.392 seconds.
(b) What is the energy stored (in Joules) in the magnetic eld when the current equals a fraction 0.95 of
the steady state current?
The energy stored is given by
U =
1
2
LI
2
Since I at that time equals 0.95
V
R
,
U =
1
2
L
_
0.95
V
R
_
2
2339.28 J
(c) What is the total energy delivered (in Joules) by the battery up to the time t found in part (a)?
How much energy (in Joules) has been dissipated in the resistor?
The total energy delivered is the energy stored (in the ideal inductor) plus the energy burned in the resistor
up to that time, the integral of I
2
R from t = 0 to t = (answer a).
_
5.392318
0
I
2
R dt =
_
5.392318
0
V
2
R
2
_
1 exp
_
Rt
L
__
2
R dt =
V
2
R
_
5.392318
0
_
1 exp
_
Rt
L
__
2
dt
I didnt see an easy way to get that integral down to something simple, so I solved it with Mathematica.
Simplied, and keeping the
V
2
R
in front in mind, I found

V
2
_
L
_
e

2RT
L
4e

RT
L
+ 3
_
2RT
_
2R
2
with T being the upper integration limit; T = 5.392318 for this problem. Putting the values in there, we
nd that the energy burned is 8265.796 joules.
That is then the answer for question (c2), while the answer to (c1) is (b) plus (c2).
(The total energy delivered is the energy stored plus the energy burned in the resistor.)
191
13.6.4 Problem 4: RL circuit
(Yep, the problem name is the same as the previous one!)
Consider the following circuit. We have R
1
= 11 Ohms, R
2
= 15 Ohms, R
3
= 13 Ohms, V = 7 Volts,
L = 0.09 H.
At t = 0 the switch is closed.
(a) Immediately after the switch is closed, what are currents I
1
, I
2
, I
3
?
Well, we dont need dierential equations nor regular equations for this one. I
1
= I
3
=
V
R
1
+R
3
. I
2
= 0
because of the inductor.
(b) What are the three currents after the switch has been closed for a long time?
Again simple: we now treat the inductor as a short-circuit with R = 0, and again we just have Ohms law
with a few parallel resistors.
I
1
= I
2
+ I
3
is clear, since the current through I
2
and I
3
must go through R
1
rst.
We can set up two loop equations for the outer and lower loops, keeping in mind that we need Faradays
law for the lower one. For the outer loop, we start in the bottom-right corner and go clockwise, simply
subtracting drops in potential, and adding climbs (the battery); the sum is then 0, Kirchhos rule applies.
I
1
R
1
I
3
R
3
+ V = 0
For the lower loop, we use Faradays law! The left-hand side will be the sum of the

E

d potential
dierences, while the right-hand side will be the negative of the time derivative of the ux. We start in
the same corner. In this method, we instead think about the E-elds direction and currents direction
when deciding the same. For the same direction we add, opposite directions we subtract. We nd:
I
1
R
1
+ I
2
R
2
+ 0 V =
d
B
dt
Remember that E = 0 inside the inductor, so it doesnt contribute to the left side! Instead, we use the
denition of inductance LI =
B
to get
I
1
R
1
+ I
2
R
2
V = L
dI
dt
We then have three equations and three unknowns. Because the current has stabilized,
dI
dt
= 0, so that
term disappears, and our dierential equation simplies:
192
I
1
= I
2
+ I
3
I
1
R
1
I
3
R
3
+ V = 0
I
1
R
1
+ I
2
R
2
V = 0
Solving the system, we get (I used Mathematica):
I
1
= I
2
+ I
3
I
2
=
R
3
V
R
1
R
2
+ R
1
R
3
+ R
2
R
3
I
3
=
R
2
V
R
1
R
2
+ R
1
R
3
+ R
2
R
3
The switch is now opened again.
(c) What are the three currents after the switch is reopened?
The question is slightly unclear, but the answers work if we calculate the answer for t = 0
+
where it was
opened at t = 0. That is, an innitesimal amount of time later, so that the inductor current is unchanged.
Note that the circuit changes. S, V and R
1
are now disconnected/left open, so the circuit is now L, R
2
and R
3
in series. Thus I
3
= I
2
, with I
2
unchanged from our answer above.
I
1
= 0, since it is disconnected.
13.6.5 Problem 5: Opening a switch on an RL circuit
Even more RL circuits!
The LR circuit shown in the gure contains a resistor R
1
and an inductance L in series with a battery of
emf E
0
= V
0
. The switch S is initially closed. At t = 0, the switch S is opened, so that an additional very
large resistance R
2
(with R
2
R
1
) is now in series with the other elements.
(a) If the switch has been closed for a long time before t = 0, what is the steady current I0 in the circuit?
Express your answer in terms of, if appropriate, V
0
, R
1
, R
2
and L.
Well, that parts crazy easy. I =
V
0
R
1
, since everything else is a short circuit.
(b) While this current I
0
is owing, at time t = 0, the switch S is opened. Write the dierential equation
for I(t) that describes the behavior of the circuit at times t > 0. Solve this equation (by integration) for
I(t) under the approximation that V
0
= 0. (Assume that the battery emf is negligible compared to the
total emf around the circuit for times just after the switch is opened.)
Express your answer in terms of the initial current I
0
, R
1
, R
2
, t and L.
Well, thats clearly a lot harder! What the problem means is then that V
0
is replaced by a short circuit,
while the switch of R = 0 is replaced by the resistance R
2
. So the circuit is R
1
, R
2
and L all in series,
with nothing else to worry about.
I would normally just apply the solution to the dierential equation directly, but since they insist...
Starting at the bottom-right again (why not?), applying Faradays law; leftside closed line integral, right
side change in ux:
IR
2
+ IR
1
= L
dI
dt
Not having taken a dierential equations class, hmm... Solve by integration, they say. Im not sure how
to do that, and itd feel a bit like cheating to just skip to the known solution, so Ill attempt the method
193
taught in 6.002x (circuits and electronics). First, we try write it in a dierent form. We set R = R
1
+R
2
,
to simplify things.

L
R
dI
dt
I = 0
First, we nd the particular solution, which is any solution that satises the equation. If we choose I = 0,
that works, so I
p
= 0.
For the homogeneous solution, we need a function such that the derivative of I
h
is a constant
L
R
times
itself. The exponential function has this property, so we guess (yes, really) that might be a solution, only
that we have two unknown coecients. We choose I
h
= Ae
bt
:
L
R
dI
h
dt
+ I
h
= 0
L
R
d
dt
_
Ae
bt
_
+ Ae
bt
= 0
L
R
Abe
bt
+ Ae
bt
= 0
We can divide out Ae
bt
at this point:
L
R
b + 1 = 0
We can now nd b:
L
R
b = 1
b =
R
L
Well, that looks familiar.
The answer to the dierential equation is the sum of the particular and homogeneous solutions, we we add
the two. However, our particular solution was 0, so the answer is just the homogeneous solution:
I(t) = Aexp
_

R
L
t
_
(using exp(x) = e
x
notation, to make it readable)
We can nd the value of A by substituting an initial condition, I(0) =
V
0
R
(by analysis of the circuit), so
we substitute in that value for I and t = 0:
A =
V
0
R
The exponential turns into 1, so we nd the answer at once. The full solution, with A and b known, is
then
I(t) =
V
0
R
exp
_

R
L
t
_
... which we really knew already, having seen the answer in lecture. At least now we have derived it
ourselves, too.
The task was to express this answer, using I
0
for the initial current, I
0
=
V
0
R
=
V
0
R
1
+R
2
(remember that we
set R = R
1
+ R
2
earlier):
I
0
exp
_

_
R
1
+ R
2
L
_
t
_
194
(c) Using your results from part b), nd the value of the total emf around the circuit (which from Fara-
days law is LdI/dt) just after the switch is opened. Express your answer in terms of, if required, V
0
,
R
1
, R
2
and L.
Well, I suppose we should dierentiate the above (since we want to know
dI
dt
and we know I), and multiply
it by L, then.
d
dt
I
0
exp
_

_
R
1
+ R
2
L
_
t
_
=
I
0
(R
1
+ R
2
)
L
exp
_

_
R
1
+ R
2
L
_
t
_
The EMF is then the above times L, so

E = I
0
(R
1
+ R
2
) exp
_

_
R
1
+ R
2
L
_
t
_
The answer does make intuitive sense - the current running times the total resistance ought to equal the
total EMF!
At t = 0, the above simplies down to

E = I
0
(R
1
+ R
2
)
Time to be careful! I originally substituted in I
0
=
V
0
R
1
+R
2
and got

E = V
0
, several times over. However,
R
2
was just added - the circuit hasnt had time to react to that, yet. So instead,
I
0
=
V
0
R
1

E =
V
0
R
1
(R
1
+ R
2
) = V
0
+
V
0
R
2
R
1
= V
0
_
1 +
R
2
R
1
_
How reasonable is your assumption in part b) that V0 could be ignored for times just after the switch is
opened?
Well, considering that we got the answer while assuming that, it appears it was a fairly great assumption.
d) What is the magnitude of the potential drop across the resistor R
2
at times t > 0, just after the switch
is opened? Express your answers in units of V
0
assuming R
2
= 100R
1
.
It would have to be the current, answer (b), times R
2
:
V
R2
= I
0
exp
_

_
R
1
+ R
2
L
_
t
_
=
V
0
R
1
exp
_

_
R
1
+ R
2
L
_
t
_
R
2
With R
2
= 100R
1
:
V
R2
= 100V
0
exp
_

_
R
1
+ R
2
L
_
t
_
The times t > 0 part is a bit confusing; they are really asking for the value JUST after the switch is
opened, i.e. at t = 0
+
. Thus the exponential term disappears (with t = 0, the whole term becomes 1),
and V
R2
= 100V
0
, or, in units of V
0
as they ask for, just 100.
195
13.6.6 Problem 6: Self-inductance of a toroid
A coil consists of N = 10
5
turns of wire wrapped uniformly around a plastic torus. The inside radius of
the torus is r
0
= 0.21 m and the outer radius is r
1
= 0.24 m. Thus, each winding has a diameter d = 0.03
m. What is the self-inductance (in H) of this coil? Work in the approximation d r
0
.
To nd the self-inductance, we rst apply Amperes law to nd the B-eld inside. Once we have that,
LI =
B
can be used to calculate the inductance.
Lets see. We begin by choosing the Amperean circle they have drawn for us (the dashed line), with the
radius r, to which we attach the at open surface that it bounds, r
2
. The B-eld inside is approximately
constant at all points along the line, so we start out with
B2r =
0
_
I
pen
+
0

d
dt

E

dA
_
If I is constant, I see no reason for there to be any change in electric ux. (If I varies, Im not sure.)
We get rid of the displacement current term to get
B2r =
0
I
pen
B =

0
2r
I
pen
What is I
pen
, however? Thats a bit trickier than with the straight solenoid... Or is it? The length of the
toroid must be the circumference of our Amperean circle, centered in the toroid, so = 2r. Other than
that, the number of times it penetrates should be the same - the number of loops per unit length, times
the length, so
I
pen
= I
N

= NI
In that case,
196
B =
NI
0
2r
This turns out to be the correct B-eld for a toroid! All right, lets try to nd the ux, so we can nd L.
As can be seen above, the B-eld is not constant inside, so we really need to integrate this over the area,
to nd the correct ux.
However, we are told to [w]ork in the approximation d r
0
.
We can write the B-eld as
B =
NI
0
2(r
0
+ )
Factoring out r
0
from the sub-expression in the denominator gives us
B =
NI
0
2r
0
(1 + /r
0
)
Now, since r
0
, the /r
0
term becomes far less than the 1, so we neglect it, and nd
B
NI
0
2r
0
If we use that B-eld approximation, we see that in the approximation, it is now constant over the cross-
sectional area. We can now nd the ux using
B
= BA, with A being the cross-sectional area, and then
multiply by the number of windings to nd the total ux:

B

NI
0
2r
0

_
d
2
_
2

B

NId
2

0
8r
0
The inductance is then given by
L =
N
B
I

N
2
d
2

0
8r
0
6.73 H
This result is consistent with that of a a solenoid, which shouldnt come as a huge shock, seeing that a
toroid is essentially a wrapped-up solenoid.
We also see that the toroid has a fairly huge inductance, which also shouldnt be entirely surprising given
its size and 10
5
windings!
13.6.7 Problem 7: RL circuit
(Yes, really, RL circuit!)
A circuit consists of a self inductor of L = 0.003 H in series with a resistor R
1
= 5 Ohm. Parallel to these
is a resistor R
2
= 10 Ohm. A battery of V
0
= 9 volt is driving the circuit.
Current has been running for 10 minutes.
(a) How much energy (in Joules) is now stored in the self-inductor?
The time constant is less than a millisecond, so the 10 minutes means steady-state has been reached long
ago.
We treat the inductor as having zero resistance, so the current through it is I
1
=
V
0
R
1
. Power stored is
given by U =
1
2
LI
2
, so we can easily nd the answer using that combination.
197
U =
1
2
L
_
V
0
R
1
_
2
= 0.00486 J
(b) How much power (in Watt) is then generated by the battery into the circuit? (ignore internal resistance
of the battery).
The inductor no longer draws any power, so the answer is
P =
V
2
R
1
+
V
2
R
2
24.3 W
The connection to the battery is now broken (so that the battery is not connected to the circuit anymore).
(c) How long will it take (in seconds) for the current through R1 to be reduced by 50%?
We derived the equation for this earlier on in this weeks homework.
First, the circuit will change to be R
1
, R
2
and L in series.
Second, the current will die out following an exponential decay, given by
I(t) = I
0
exp
_

R
L
t
_
If we set that equal to 0.5I
0
, and solve:

R
L
t = ln 0.5
t =
L
R
ln 0.5 1.386 10
4
s
keeping in mind that R = R
1
+ R
2
. This also answers (e), since the circuit is now a series circuit.
(d) How long does it take (in seconds) till the energy stored in the self-inductor has been reduced by
50%?
Energy stored is
U =
1
2
LI
2
=
1
2
L
_
I
0
exp
_

R
L
t
__
2
=
1
2
LI
2
0
exp
_
2
R
L
t
_
We set that equal to 0.5 times the current stored energy,
1
2
LI
2
0
:
1
2
LI
2
0
exp
_
2
R
L
t
_
=
1
4
LI
2
0
exp
_
2
R
L
t
_
=
1
2
2
R
L
t = ln(2)
t =
L
R
ln(2)
2
6.93 10
5
s
Since we solved part (e) above (same answer as for part c), so we are now done!
13.7 Week 8
Theres no homework this week, as the second midterm is next weekend.
198
13.8 Week 9
13.8.1 Problem 1: RC circuit
In the following circuit V = 9 V, R = 25 Ohm, C = 0.0025 Farad.
At t=0 we start charging the capacitor with no charge initially on the capacitor.
(a) Calculate the time t (in seconds) when the potential across the capacitor is V/2.
The circuit is a simple series RC circuit, so I hardly see a need to draw it. The capacitor voltage and
current (the latter is the same for the entire circuit, of course) is given by
V
C
(t) = V
_
1 exp
_

t
RC
__
I(t) =
V
R
exp
_

t
RC
_
To nd the answer, we set the potential V
C
equal to V/2, and solve the exponential for t:
V
_
1 exp
_

t
RC
__
=
V
2
1 exp
_

t
RC
_
=
1
2
exp
_

t
RC
_
=
1
2

t
RC
= ln
1
2
t
RC
= ln 2
t
RC
= ln 2
t = RC ln 2
We plug our values in, and get t 0.04332 seconds.
(b) How much energy (in Joules) has the power supply generated between t = 0 and the time when the
potential across the capacitor reaches V/2?
There are several ways to solve this. We could nd the energy stored in the capacitor (
1
2
CV
C
2
) and the
energy dissipated in the resistor (
_
t
0
I
2
R dt), or we could integrate the power sources power over time,
_
t
0
V I dt. Ill choose the later approach, since it should be slightly easier, and it is also the approach that
immediately answers the question, without invoking conservation of energy etc.
We have the integral
E =
_
t
0
V I(t)dt = V
_
t
0
V
R
exp
_

t
RC
_
dt =
V
2
R
_
t
0
exp
_

t
RC
_
dt
E = V
2
C
_
1 exp
_

t
RC
_
E 0.101247 J
199
13.8.2 Problem 2: RC circuit
In the following circuit R
1
= 3 Ohm, R
2
= 10 Ohm, C = 0.003 Farad, V = 5 Volts.
At time t = 0 we connect the power supply to the circuit. At t = 0 the capacitor is uncharged.
(a) What is the time constant (in seconds) to charge up the capacitor?
The time constant is given by = RC, where R is the equivalent resistance of the circuit, as seen by
the capacitor. We can nd that by shorting out the voltage source (replacing it with ideal wire), and
calculating the resistance by doing series/parallel analysis.
Doing so, we nd that all current to the capacitor must go through R
1
. It can go then either go through
R
2
or R
1
at the bottom; they are in parallel. So we have
R
eq
= R
1
+ (R
2
[[R
1
) = R
1
+
R
2
R
1
R
2
+ R
1
5.308
therefore
= R
eq
C 5.308 0.003 0.015924 s
(b) What is the electric potential over the capacitor when it is fully charged?
When it is fully charged, the current through the top resistor is zero, and the voltage drop there is zero.
Thus the capacitors voltage is in parallel with V + V
R1
, and the two must be equal.
The current through the lower loop is I =
V
R
1
+ R
2
=
5
13
A, and the voltage drop across R
1
therefore
15
13
volts. Thus
V
C
= V V
R1
3.846 V
(c) Calculate the current delivered by the power supply as a function of time and evaluate it (in Ampere)
for t = 0.6, where is the value obtained in part (a).
The function needs to have two parts: one that decays with time, and one that doesnt.
The latter must either be a constant
5
13
, the current when the capacitor is fully charged (see (d) below),
or move up to that value as time passes.
It turns out that the two parts are the
5
13
and the capacitor current (the dierence between the current
when its completely uncharged and acts like a short, and when its charged and acts like an open) times
the exponential, that is
I(t) = I() + (I(0) I()) exp
_

_
Therefore
I(t) =
5
13
+ (
V
R
eq

5
13
) exp
_

_
Using the questions values for all variables and constants:
I(t) =
5
13
+ (
5
5.308

5
13
) exp (0.6) 0.6905 A
(d) What is the current delivered by the power supply when the capacitor is fully charged?
200
Well, the capacitor is an open circuit at that point, so its simply the current in the lower loop,
I =
V
R
1
+ R
2
=
5
13
A
(e) How much energy (in Joules) is in the capacitor when it is fully charged?
U =
1
2
CV
C
2
0.0222 J
... using the result for its voltage from (b).
13.8.3 Problem 3: RLC circuit
In the following circuit R = 5 ohm, L = 0.02 H, C = 0.0065 Farad and V = 13 Volts. The capacitor is
initially uncharged.
(a) What is the current delivered by the battery immediately after the switch is closed?
The inductor will act like an open, while the capacitor will act as a short circuit. The current will be at a
maximum through the capacitor, so I = V/R = 13/5 A.
(b) What is the the current delivered by the battery a long time after the switch is closed?
Now the inductor is a short, so the same answer applies.
What is the potential dierence across the capacitor a long time after the switch is closed?
0, because the inductor is a short circuit in parallel with it.
13.8.4 Problem 4: An LRC circuit
A circuit contains a self-inductance L in series with a capacitor C and a resistor R. This circuit is driven
by an alternating voltage V = V
0
sin t. We have L = 0.015 H, R = 80 , C = 510
6
F, and V
0
= 40 volts.
(a) What is the value (in radians/seconds) of the resonance frequency,
0
?
The resonance frequency for a series RLC circuit is given by

0
=
1

LC
so
0
3651.48 rad/sec.
(b) Consider three separate cases for which = 0.25
0
, =
0
, and = 4
0
respectively. For each case
calculate the the peak current I
0
in Amperes.
To do this, we need to nd the equation that governs the current in the circuit. Finding that is not a
trivial task, so I will refer to the known result from lecture/the book instead:
I(t) =
V
0
_
R
2
+ (L
1
C
)
2
sin(t )
where
tan =
L
1
C
R
201
Pay attention to that we use the sine function, rather than the usual cosine, as the problem description
states the driving voltage is a sine. The only dierence is in the phase, so the old result should certainly
still be valid, however.
I didnt realize this until I was essentially 100% sure I calculated the energy stored in the inductor correctly.
Turns out I did, if the driving voltage had been a cosine!
So, with that in mind, the peak current in the term that multiplies the sine, so we just need to stick our
values in there, and nd
I
0
0.181467 A (for = 0.25
0
)
I
0
0.5 A (for =
0
)
I
0
0.181467 A (for = 4
0
)
Note how at =
0
, the resonance frequency, the reactance is cancelled out and the peak current is
V
0
R
.
(c) Find the energy UC(t) and the energy UL(t) stored in the capacitor and in the inductor, respectively,
at time t
1
= 0.0003 seconds for =
0
. Express your answers in Joules.
OK. How should we approach this? We have an expression for the current in the circuit, and the current
must be the same through all elements at all times. Therefore, we can use the current expression and
U =
1
2
LI
2
to answer the second question:
U
L
=
1
2
LI
2

1
2
L0.444566
2
0.0014823 J
We can nd U
C
trivially by knowing a fact about RLC circuits: the energy stored in the capacitor plus
the energy stored in the inductor is always the maximum energy stored in either. So there is a time where
the total stored energy in the circuit is
1
2
CV
2
C
, and at that time U
L
= 0. At another time, the inductors
stored energy is at its maximum, which is when the capacitors energy is 0.
Between the extremes, the sum of their stored energies is always the maximum stored energy, which we
nd easily with
1
2
LI
0
2
(since, as stated, U
C
= 0 at that time). Therefore,
U
C
=
1
2
LI
0
2
U
L
= 0.001875 0.00148229 = 0.00039271 J
We can also nd this by integrating the capacitors voltage:
V
C
(t) =
1
C
__
t
1
0
I(t)dt
_
+ V
C
(0)
V
C
(t) =
1
C
_
t
1
0
V
0
sin t
R
dt + V
C
(0) =
V
0
RC
_
t
1
0
sin t dt + V
C
(0)
since we are at resonance, and the inductive and capacitive reactances cancel, so that only the resistor
contributes to limiting the current, which is in phase with the driving voltage.
Solving it, we nd
V
C
(t) =
V
0
RC
1 cos(t
1
)

+ V
C
(0) 14.8528 + V
C
(0)
Well, what is V
C
(0)? Its not zero; keep in mind that the equations we are working with are only true
at steady state, so the capacitor doesnt necessarily have to be uncharged at t = 0; the circuit must have
been running prior to t = 0.
One way to nd it is to use the knowledge that the current is 0 (because the driver is a sine, and it is in
202
phase with the current), which means the capacitors voltage is at a maximum (or else the current would
not be zero). Therefore, the capacitors voltage is
V
C
(0) = I
max
1
C

0.5
3651.48 5 10
6
27.386 V
(via essentially Ohms law, using the impedence).
We can then apply the potential energy formula:
U
C
=
1
2
(5 10
6
)(14.8528 27.386)
2
0.0003927 J
... though this was more complex than the solution that uses the knowledge that U = U
C
+U
L
is constant.
13.8.5 Problem 5: Design a ute
A ute can be regarded as a tube open at both ends. It will emit a musical note if the utist excites a
standing wave in the air column inside the tube.
The lowest musical note that can be played on a ute is C (261.7 Hz). What must be the length of the
tube? Assume that the air column is vibrating in its fundamental mode.
Interesting! But very simple. We use the equation from lecture,
f
1
=
v
2L
v 340 m/s, the speed of sound in room-temperature air. L is the length of the ute, which we want to
nd. So we rearrange it, and get
L =
v
2f
1
With v = 340 m/s and f
1
= 261.7 Hz, we nd L = 0.65 meters.
13.8.6 Problem 6: Width of resonance peak
Consider the following RLC circuit: (standard series RLC circuit)
with V = 3 Volts, R = 25 Ohms, C = 0.006 Farad, L = 0.085 Henry. Dene the frequencies as the
frequencies such that the absolute value of the current of the circuit I0() equals
1
2
I
0max
.
What is the dierence =
+

? Express your answer in radians/sec.


Well, the problem doesnt actually specify the driving waveform, except that theres a sine wave on the
voltage source in the diagram, so Ill take it to be V cos t. The phase shouldnt matter for this question
anyhow.
Given that, we have this:
I(t) =
V
0
_
R
2
+ (L
1
C
)
2
cos(t )
where
tan =
L
1
C
R
If we set
X = L
1
C
203
Z =

R
2
+ X
2
to reduce the clutter, we have
I(t) =
V
0
Z
cos(t arctan
X
R
)
If Z = R, we have the maximum possible current (at resonance). Therefore we should have half the
maximum current at twice that impedance, Z = 2R (where half the impedance comes from the L and C).
_
R
2
+ (L
1
C
)
2
= 2R
The equation has four solutions, but two are negative (with the same absolute value as the other two), so
we nd

+
= 513.247 rad/s

= 3.82035 rad/s
= 509.4267 rad/s
(Im sure we could do this with less math, but I used Mathematica for the equation solving here; I couldnt
nd a really simple way to solve this; solving

R
2
+ X
2
= 2R for X, and then the denition of X =

3R
which we nd from the rst equation gives us two frequencies, the absolute values of which are correct,
but one is negative.)
The resonance frequency is

0
=
1

LC
44.281 rad/s
We can nd the resonance frequency as the geometric mean of the half-current frequencies:

0
=

44.281 rad/s
13.8.7 Problem 7: Standing wave
Consider the following standing wave, (all units are SI)
z = 0.2 sin(0.4y) cos(350t)
(a) what is the wavelength?
The equation is of the form
z = Asin(ky) cos(t)
so we nd the wavelength as =
2
k
=
2
0.4
.
(b) what is the angular frequency in radians/sec?
= 350, see above.
(c) what is the frequency (f) in Hz?
f =

2
=
350
2
204
(d) The values of y where the displacement is always zero, are of the form y = a + bn, where n=0, 1,
2, etc.
What are a and b?
Lets see. The wavelength is =
2
k
=
2
0.4
, and this happens at multiples of half that. At y = 0,
sin(0.4 0) = 0, so there is a node there. Therefore a = 0, so that when n = 0, a + bn = 0.
b must then be half the wavelength (above).
(e) The values of t where the displacement at all values of y equals to zero are of the form t = c + dn,
where n=0, 1, 2, etc.
What are c and d?
So what this is saying that if we plot the wave, it will be a straight line through the y axis (x axis on a
x-y plot) twice per cycle, at times t = c + dn.
This should happen when cos t = 0. Solving that, we nd
t =

2
+

n
which nicely translates into c and d.
13.8.8 Problem 8: Traveling wave
Consider the following traveling wave (all units are SI).
x = 0.05 cos(3.5y + 126t)
(a) What is the wavelength in m?
Here we have the form
x = Acos(ky + t)
The wavelength is again =
2
k
.
To save time writing: the other questions are for , f, the waves direction, the speed of propagation v
and the amplitude.
is given in in the t; f =

2
; the direction is y; minus because there is a plus between ky and t (the
direction is opposite that sign) and y because we have ky.
The speed of propagation is found as
= vT
so
v =

T
= f
Finally, the amplitude is the number multiplying the cosine, so A = 0.05.
205
13.8.9 Problem 9: Lightly damped undriven circuit
On to what looks like the weeks hardest problem.
Consider the RLC circuit shown in the gure below in which R
2
< 4L/C (underdamped).
Assume that at t = 0 , the charge on the capacitor has its maximum value.
(a) The dierential equation obeyed by the potential across the capacitor can be written in the following
form:
d
2
V
dt
2
+ a
dV
dt
+ bV = 0
Express a [and b] in term of, if necessary, R, L, C.
OK. We start by writing down the circuit in terms of voltage, which means using a few equations used for
capacitors and inductors:
I
C
(t) = C
dV
dt
I
L
(t) =
1
L
_
t
0
V
L
(t)dt + V
L
(0)
Using Kirchhos bastardized loop rule (the one that holds, but confuses the physics for many),
V
R
(t) + V
C
(t) + V
L
(t) = 0
RI(t) +
_
1
C
_
t
0
I(t)dt + I(0)
_
+ L
dI
dt
= 0
The current through the circuit can be writtes as I = C
dV
dt
, where V is the potential across the capacicitor
(V = V
C
):
RC
dV
dt
+
_
1
C
_
t
0
C
dV
dt
dt + I(0)
_
+ LC
d
2
V
dt
2
= 0
I(0) = 0, because the capacitor is fully charged at that time, so the I(0) term disappears, and the integral
of the derivative simplies:
RC
dV
dt
+ V + LC
d
2
V
dt
2
= 0
Re-order:
LC
d
2
V
dt
2
+ RC
dV
dt
+ V = 0
Remove coecient from the second derivative:
d
2
V
dt
2
+
R
L
dV
dt
+
1
LC
V = 0
So we see then, that
a =
R
L
b =
1
LC
These values have other often used names, a = 2 and b =
2
0
. Here we see that 2 = (which we also
havent discussed) is the bandwidth, and
0
=
1

LC
is the resonance frequency of the circuit.
206
In the circuit above, R = 251 Ohm, L = 0.07 H, C = 2 10
6
F.
(b) What is the decay constant (in seconds) according to which the charge on the capacitor is decaying?
It seems to me this would be easier if we have an equation for Q, so lets solve a dierential equation for
Q instead, and see what happens.
Going clockwise with the current, starting at the top-left corner, using Faradays law:
I(t)R + 0
Q
C
= L
dI
dt
I(t)R
Q
C
+ L
dI
dt
= 0
Writing I(t) in terms of
dQ
dt
:
R
dQ
dt

Q
C
L
d
2
Q
dt
2
= 0
... keeping the sign in mind, as the current is clockwise, but
dQ
dt
< 0. We can multiply by -1 throughout,
and sort:
Q
C
+ R
dQ
dt
+ L
d
2
Q
dt
2
= 0
We solve this for the initial conditions Q(0) = Q
0
and Q

(0) = 0 (as the current will be zero if the


capacitors fully charged at the moment). We nd a complex solution, however, the beginning of it is
Q(t) = Q
0
exp
_

Rt
2L
_
_
...
_
Using the standard form of
exp
_

_
we nd
=
2L
R
which gives the correct answer. We could also have watched the problem solving videos where this decay
constant is given.
(c) The times at which the total energy stored in the RLC circuit is exclusively of electric nature can be
written as
t = a + bn n = 0, 1, 2, 3 ...
What are a and b?
This happens when all the energy is in the capacitor (stored in an electric eld), which happens when
I = 0.
Finding the equation governing the current may not be easy, but we dont have to. We know that I(0) = 0,
and by denition, the zero crossing occur with a period
T
2
, with T given by T =
2

.
(There are two zero crossings per period of a sinusoid.)
207
isnt easy to nd either (Mathematicas solution of the dierential equation doesnt contain it in an
easy-to-nd manner, at least), so I will use the equation shown in the problem solving video:
=
_
1
LC

R
2
4L
2
So we have
T =
2

0.003170045
a = 0
b
0.003170045
2
a = 0 because we know a zero crossing is at t = 0, so when n = 0, bn = 0, and a + bn = 0.
13.9 Week 10
13.9.1 Problem 1: Traveling electromagnetic waves
Consider two examples of a plane, monochromatic, electromagnetic wave traveling in a homogeneous
medium. The electric eld vector is given in the two cases by
case (1) E
x
= 0; E
y
= 0; E
z
= 50 sin(4.71x + 9.12 10
8
t)
case (2) E
x
= 0; E
z
= 0; E
y
= 55 sin(4.71x 1.05 10
9
t)
where [

E[ is measured in V/m, t in sec, and x in m. For each case, answer the following questions:
(a) What is the propagation direction of the wave?
The direction of propagation is given by the term inside the sine; in the rst case, we have the form
sin(kx +t) which implies movement in the negative x direction. In the second case, the opposite is true.
They then ask for the wavelength and wave number. The wave number we use is dened as k =
2

, so
=
2
k
, with k being 4.71 in this case, so that answers the four sub-questions here.
After that, they ask for the frequency. That is given by f =

v
, but v ,= c here, and we do not know it, so
we have to nd some other way to calculate it. We can use , the term multiplying t in the sine.
= 2f, so f =

2
, which answers these two questions.
Next up, the propagation velocity in meters per second (which we couldve used above). We can relate
that to the wavelength and frequency: f = v.
Next question: index of refraction. n =
c
v
, so we can pretty much take 3 10
8
m/s and divide it by two
previous answers. Easy.
After that, they want the corresponding equations for

B = B
0
sin(kx
t
)

d, where we are to give them


the direction, B
0
, k and , with the correct sign.
They are in phase with the electric eld, so we just copy the contents of the sine above.
As for the direction,

E

B = v, so we have z

B = x for the rst case. That means that

B = y via the
right-hand rule.
For the second case, we have y

B = + x, which means

B = + z.
What about magnitudes? In a vacuum,
208
B
0
=
E
0
c
...but this is not in a vacuum. I assume we can replace c by v and be done with it, and that indeed gives
us green checkmarks, using v from above instead of c.
Finally, the last sub-question: what is the time-averaged Poynting vector (magnitude and direction) for
x = y = z = 3 for case 2?
The Poynting vector is given by

S =

E

B

0
, and its time-averaged magnitude by S =
E
0
B
0
2
0
.
Thus the direction is given by dir

S =

E

B = y z = x.
The magnitude is simply found using the formula above - we have E
0
and B
0
already, and know
0
=
410
7
.
The coordinates do not enter into this, because this is a plane wave.
13.9.2 Problem 2: A standing electromagnetic wave
A wave solution to Maxwells Equations is given by

E = E
0
x cos(9.5z) cos(1.84 10
11
t) where z is mea-
sured in centimeters and t in seconds.
(a) What is the wavelength (in meter) of the wave?
Ugh, centimeters. Well, lets look at this the right way, and not divide by 100 where we should really
multiply, etc.
The wavelength is =
2
k
, where k is given in radians per meter.
We are given k in rad/cm, so we need to multiply it by 100 to get rad/m:
=
2
9.5 100
They then ask for the frequency f =

2
and index of refraction n =
c
v
of the medium.
f =

2
=
1.84 10
11
2
Hz
v =

k
=
1.84 10
11
9.5 100
1.936 10
8
m/s
n =
c
v
=
3 10
8
1.936 10
8
(c) The associated magnetic eld can be written as

B = B
0
f
1
(kz) f
2
(t)

d
What is the direction, the value of B
0
, k, and what are f
1
and f
2
?
The direction must be y, because the wave is propagating in the z direction (though this is a standing
wave, so the movement is cancelled out), and the E-eld is in the x direction.
We should nd that x

B = z, so

B = y for that to be true.
The magnitude can be found via B
0
=
E
0
v
, where v =
c
n
, so we have the values we need to answer that.
k and have the same values as for the E-eld, we can just copy and paste them (but watch out for the
units of k being in cm, so we must multiply it by 100 here as well).
f
1
and f
2
are both the sine function. The E-eld uses cosines, and the B-eld is 90 degrees out of phase
in a standing wave (see the problem solving videos).
209
Use the following trigonometric identity
2 cos cos = cos( + ) + cos( )
to express the standing wave as a superposition of two traveling waves. Then, obtain the magnetic elds
associated to those traveling electric eld, and combine them together to get the magnetic eld associated
with the standing electric eld.
Well, we have

E = E
0
x cos(9.5z) cos(1.84 10
11
t)
that we need to match up with the identity, and split up. Clearly, = 9.5z and = 1.84 10
11
t. The
identity has a factor of two prior, and our equation has the factor E
0
... so it looks like we can write it as

E = E
0
x cos(9.5z) cos(1.84 10
11
t)

E =
E
0
2
x
_
cos(9.5z + 1.84 10
11
t) + cos(9.5z 1.84 10
11
t)
_
So the standing E-eld can be written as the above sum. Writing the two traveling waves as two equations,
we nd
E
tr1
=
E
0
2
x cos(9.5z + 1.84 10
11
t)
E
tr2
=
E
0
2
x cos(9.5z 1.84 10
11
t)
... with
tr1
signifying traveling wave 1, etc. Next, they want us to nd the magnetic elds of those waves.
We now have two traveling waves, so we use B
0
= E
0
/v, with v 1.936 10
8
m/s as we found earlier.
The B-eld should now be in phase, so we nd
B
tr1
=
E
0
2v
y cos(9.5z + 1.84 10
11
t)
B
tr2
=
E
0
2v
y cos(9.5z 1.84 10
11
t)
Note the minus sign in B
tr1
, which is absolutely crucial! Without it, we dont nd

E

B = v, but v
instead, which is of course incorrect. That screws up our entire result, too.
We can combine the result using the identity
cos( ) = cos cos sin sin
B
tr1
=
E
0
2v
y
_
cos(9.5z) cos(1.84 10
11
t) sin(9.5z) sin(1.84 10
11
t)
_
With less confusing minus signs (by distributing the one in front):
B
tr1
=
E
0
2v
y
_
cos(9.5z) cos(1.84 10
11
t) + sin(9.5z) sin(1.84 10
11
t)
_
B
tr2
=
E
0
2v
y
_
cos(9.5z) cos(1.84 10
11
t) + sin(9.5z) sin(1.84 10
11
t)
_
We now see that when we add these two, the cosine terms cancel, and the sine terms add:

B =
E
0
2v
y 2 sin(9.5z) sin(1.84 10
1
1 t)

B =
E
0
v
y sin(9.5z) sin(1.84 10
1
1 t)
210
This is indeed what we need to get part (c) correct (above).
(d) What is the time-averaged Poynting vector for x = y = 3, z =

3? Express you answer in joules per


square meters per second.
We barely need to do math to solve this one - the answer is 0 for all components, because this is a standing
wave! The time-averaged Poynting vector will be zero in such a case, since there is no net movement.
However, lets have a quick look at the math that proves this.

S =

E

B

0
=
(E
0
cos(9.5z) cos(1.84 10
11
t))(
E
0
v
sin(9.5z) sin(1.84 10
11
t))

0
(

E

B)
Thats way too messy, so I will calculate the result with and instead:
S
0
=
E
2
0
v
cos cos sin sin
What is the time average of the trigonometric mess on the right? Looking at the functions of time (), we
have one cosine and one sine, essentially sin t cos t. Because the argument is the same for both functions,
we get the shape of a sinusoid, with lower amplitude than plain sin t. Indeed, its equivalent to
1
2
sin(2t),
and the time-average of that is zero! Since we are multiplying all the terms together, and one of them has
a time-average of 0, the time average of the Poynting vector is also zero.
13.9.3 Problem 3: E-M waves - Maxwells equations, and the speed of light
We discussed in lectures that traveling Electromagnetic waves in vacuum of the form

E = E
0
x cos(kz t)

B = B
0
y cos(kz t)
satisfy all 4 Maxwells equations. In lectures, I showed that an application of the generalized Amperes Law
(closed loop surrounding area A2, see below), leads to: B
0
= E
0

0
c, and I mentioned that independently
it follows from an application of Faradays Law that B
0
= E
0
/c. Combining these two results then leads
to the fantastic result that the speed of light in vacuum c =
1

0
. I want you to show that Faradays
Law indeed leads to the result B
0
= E
0
/c. You can show this by choosing a similar special area as we did
in lectures:
We dene the normal of the surface A1 in the gure above to point in the + y direction. In the following,
assume that =1 m, E
0
= 1 V/m, f = 610 10
12
Hz, where f is the frequency of oscillation.
211
(a) We dene f
1
(t) =
_

E

d, where
_

E

d is the closed loop integral taken along the contour of the area A1
in the gure above. Evaluate the function f
1
(t) in Volts for the following value of t: t = 2.210
16
seconds.
Wow, thats a lot of text! Well, most of the intro text is irrelevent for solving the problem (its a nice
background, but we dont need to memorize it!).
The E-eld is in the x direction, so we can think of it as a sinusoid drawn upon the gure above, in the
z x plane, going left to right.
We go counterclockwise for the line integral. The reason for this is that the problem denes the surface
normal as + y, so according to the right-hand rule, with the thumb pointing out of our screens, we curl
the ngers counterclockwise.
The top and bottom segments are both perpendicular to the E-eld, and so their dot products are both
zero.
The rightmost segment has

d up, and the E-eld up, so that contributes a positive term +E(/4, t).
The leftmost segment has

d down, but the E-eld up, so that contributes a negative term E(0, t). Thus
we have:
_

E

d = E(/4, t) E(0, t)
= E
0

_
cos(
k
4
t) cos(0 t)
_
= E
0

_
cos(

2
t) cos(t)
_
= E
0
(sin(t) cos(t)) 0.0817035 V
(b) Consider the function f
2
(t) =
B
(t). Following the method used in Lecture 27, calculate the function
f
2
(t), and evaluate it (in Volts seconds) for the following time t: t = 2.2 10
16
seconds.
The method used in lecture was to integrate over the area shown in the gure above, so lets try that from
memory (it should be obvious how to do this anyway).
Since the B-eld is in the + y direction, we need to do this on the area perpendicular to that, A1, or we
will surely nd
B
= 0.
This being a plane wave, the B-eld is constant in the xy plane, but not in the z plane. We can thus
integrate it, with being the length of the side where its constant, and dz the width, integrating from
z = 0 to z = /4.
By the way, the area is perpendicular to the B-eld, so we need not worry about any cos here.

B
=
_
/4
0
Bdz = B
0

_
/4
0
cos(kz t)dz

B
=
B
0

k
(cos(t) + sin(t))
We plug in the values, with B
0
=
E
0
c
(which we are trying to prove!), = 2f and k =
2f
c
, and nd

B
3.68366 10
16
T/m
2
.
Next they want the negative of the derivative, i.e. the change in ux, that is, Faradays law.

d
B
dt
=
B
0

k

d
dt
(cos(t) + sin(t))

d
B
dt
=
B
0

k
( sin(t) + cos(t))
212

d
B
dt
0.0817035 V
We nd the same result as in part (a), which shouldnt be surprising, as Faradays law states
_

E

d =
d
B
dt
... and all weve done is to calculate the left and right sides of the equation separately!
However, we calculated B
0
using
E
0
c
, so at least for these numbers, we have proven the two to be equal.
Can we prove it generally? Lets see, if we equate the symbolic answers for the line integral and the
(negative) ux derivative:
E
0
(sin(t) cos(t)) =
B
0

k
( sin(t) + cos(t))
Factor out and cancel :
E
0
(sin(t) cos(t)) =
B
0

k
(sin(t) + cos(t))
Factor out 1 on the right-hand side:
E
0
(sin(t) cos(t)) =
B
0

k
(sin(t) cos(t))
We can now cancel the trig terms completely:
E
0
=
B
0

k
= B
0

k
= B
0
c
Success!
E
0
= B
0
c
B
0
=
E
0
c
13.9.4 Problem 4: Polarized radiation
Write down the electric eld and associated magnetic eld in vacuum for a traveling plane wave with the
following properties. The amplitude of the electric vector is E
0
and the frequency is . The radiation is
linearly polarized in the y-z plane at an angle of /4 with respect to the y-axis, and it is traveling in
the +x direction. The electric and magnetic eld can be written in the following way:

E = E
0
cos(t + (1)
b
E
kx)(c
E
x + d
E
y + e
E
z)

B = cos(t + (1)
b
B
kx)(c
B
x + d
B
y + e
B
z)
There are two options for the direction of . What are the parameters for both electric and magnetic
eld? Give answers for the two possible options:
They then want values for b
E
, c
E
, d
E
and e
E
, for = /4 from the +y direction (rotated towards +z) and
for = /4 from the +y direction (rotated towards -z).
Lets look at the rst case rst. But even before that, what do they even mean?
If the radiation were linearly polarized in y, that would mean that the E-vector is described by

E = E
0
y
times some sine or cosine. That is, the x and z components are exactly zero.
Since this wave is polarized in the yz plane, with an angle with respect to the x axis, we will have a zero
x component, but nonzero y and z components. (However considering the angle of 45 degrees, I assume
the y and z components will be equal in magnitude!)
213
The direction of the polarization must always be perpendicular to the direction of propagation, which it
is: the wave propagates in the +x direction (in phase with the B wave), while the polarization is in the
yz plane.
(The B-eld is perpendicular to the E-eld, as usual, and is in phase, as usual. However, we dene linear
polarization via the E-vector only, and we will ignore the B-eld for now, and nd the E-eld rst.)
Okay, so lets look at this thing. There should be a minus sign in the cosine, since the wave is propagating
in the plus x direction. So
b
E
= 1
c
E
= 0
... as c
E
is the x component, which is zero as the polarization is in the yz plane. What remains is

E = E
0
cos(t kx)(d
E
y + e
E
z)
Two unknowns remain. We know that the direction we want is part y, and part z. Equal parts, since the
direction is exactly 45 degrees from the y axis (with 90 degrees, it would be in the z plane only).
So that essentially leaves us with one unknown, since the values should be equal (at least in magnitude).
When we combine unit vectors, we get this result: a vector (of non-unit length! the magnitude will be
larger than 1) pointing in a direction that is neither y nor z, but something in between, is y + z. This
principle applies here, as well.
In order to get the unit magnitude of such a vector, we need to divide it by its magnitude. So what is the
magnitude?
Lets ignore that third dimension, and think of this as a 2D problem. We draw a coordinate system, and
set up a vector from the origin to (1, 1) - 1 away from the y axis, 1 away from the z axis. What is the
distance to the origin? r =

1
2
+ 1
2
=

2. So that unit vector, lets call it r (just picking something), is


r =
y + z

2
I hope this doesnt get too confusing. The above was just the explanation of the principle. Because the
magnitude of the E-eld must be E
0
(at its maximum), the magnitude of the vector components we nd
must be 1, or we will scale it upwards (or downwards)! That is,
[d
E
y + e
e

E[ = 1
must hold for our chosen values. We need
_
d
E
2
+ e
E
2
= 1 and d
E
= e
E
. It looks fairly apparent that the
two values that makes this relation hold are (plus the b
E
and c
E
values we already found):
b
E
= 1
c
E
= 0
d
E
=
1

2
e
E
=
1

2
So that answers the rst part. Next up: the B-eld. It will be perpendicular to the E-eld, so we can
work from that. b
B
= 1, so that it the travel direction and phase match up with the E-eld.
Here, there is no magnitude specied in the equation given, so rather than having our vector components
add to a magnitude 1, they must add to B
0
=
E
0
c
.
214
What will the direction be? If the E-vector is in the direction of y + z, and the B-eld is plus or minus 90
degrees from that, and

E

B = x must hold... What do we nd?
_
y + z

2
_


B = x
Honestly, Im not sure how to solve this mathematically, but its easy to nd more intuitively, at least as
far as the direction is concerted. The B-eld should be perpendicular to the E-eld weve found, so there
are two possible choices that may work out with also being perpendicular with the propagation direction.
One is that we rotate it in the same direction as we did with the E-eld, from

4
to

2
+

4
=
3
4
. The
other is that we go in the other direction. What happens in the rst case? Drawing a coordinate system
(which partly helps with 3D visualization, but not entirely), it seems as that this one is the one that will
work (keep in mind that the right-hand rule should apply!).
So now, we want one part + z and one part y, and then we correct for the magnitude by dividing by

2 (so that the net vector has magnitude 1), and nally multiply by the
E
0
c
that we want as the net
magnitude of the wave:
bB = 1
c
B
= 0
dB =
E
0
c

2
eB =
E
0
c

2
When we multiply all this out, the magnitude becomes
E
0
c
, as we want. the negative in d
B
is to get it in
the y direction, i.e. downwards (the way Ive drawn this, +x is to the right of the paper, +y upwards in
the plane of the paper, and +z out of the paper).
Next up: case 2, where the rotation is in the other direction. The E-eld is rst out.
Clearly, the x component must still be zero, as we are in the yz plane still. The direction of propagation
has not changed, so b
E
= b
B
= 1.
Looking at the old intuitive directions, we now want part of z and part of + y for the E-eld vector, so
that it points upwards, inwards using the coordinate system sketch mentioned above. The magnitudes
are the same as before, though, so we just throw a negative on the e
E
value, but other than that, the
E-eld vector is unchanged from before.
b
E
= 1
c
E
= 0
d
E
=
1

2
e
E
=
1

2
And nally, the B-eld. That is now located where the old E-eld was; rotated 90 degrees from the /4
we had, it is now at +/4 degrees rotation, just as the rst E-eld. So we just copy and paste the values
from there... except that we need to multiply them by
E
0
c
for the magnitudes!
b
B
= 1
215
c
B
= 0
d
B
=
E
0
c

2
e
B
=
E
0
c

2
13.9.5 Problem 5: Polarization of electromagnetic radiation
(a) Describe the polarization state of the plane E-M waves represented by the following equations for the
electric eld

E(x, t)(E
x
= 0 in all three cases):
case(1):
E
y
= E
0
sin(kx t), E
z
= 4E
0
sin(kx t)
The two components are in phase, so this is linear polarization.
Linear polarization is characterized by wave components that are in phase, such that the resulting wave
is conned to a plane.
Circular polarization is has wave components 90 degrees out of phase, which causes the resulting vector
to rotate and trace out circles.
Elliptical polarization is like circular, except the phase or magnitudes are such that an ellipse is traced,
rather than a circle.
If you have not already seen such videos, I highly recommend searching for videos that show linear and
circular polarization via 3D graphics software. Its very helpful to understand the concepts.
case(2):
E
y
= E
0
cos(kx + t), E
z
= E
0
sin(kx + t)
Here we have a 90 degree phase dierence (sine versus cosine), but with equal magnitudes. That gives us
circular polarization.
case(3)
E
y
= 2E
0
cos(kx t + /2), E
z
= 2E
0
sin(kx t)
We can rewrite the cosine as a minus sine:
E
y
= 2E
0
sin(kx t), E
z
= 2E
0
sin(kx t)
In phase, equal magnitude: linear polarization.
(b) For case (1), where E
y
= E
0
sin(kx t), E
z
= 4E
0
sin(kx t), the magnetic eld can be written in
the following way:

B = sin(kx + (1)
b
B
t)(c
B
x + d
B
y + e
B
z)
In other words: B
x
= c
B
sin(kx + (1)
b
B
t), B
y
= d
B
sin(kx + (1)
b
B
t), Bz = e
B
sin(kx + (1)
b
B
t)
What are the values of the dierent parameters in the above equation for

B?
Heh. Not that easy to read, but its no harder than the previous homework problem.
b
B
= 1, because we want that minus sign, to keep it in phase with and in the same direction as the E-eld.
The wave propagates in the x direction, with the E-eld having only y and z components. The B-eld also
only has y and z components, though its still perpendicular to the E-eld, of course.
As in the previous problem we did with this, the x component must be zero.
Unlike the previous problem, we cant simply try to get the B-eld to be
E
0
c
in magnitude, because E
0
is
not the magnitude of the E-eld. We need to nd that magnitude rst!
216

E = E
x
x + E
y
y + E
z
z

E = sin(kx t)(E
0
y + 4E
0
z)
So the magnitude will be given by
[

E[ = [E
0
y + 4E
0
z[ =
_
E
0
2
+ (4E
0
)
2
=
_
17E
0
2
=

17E
0
So B
0
=
E
0

17 c
. However the E-eld has dierent magnitudes for the dierent components, unlike last
time.
The E
y
components should have a B-eld component perpendicular to it of the same magnitude, I would
guess. So B
z
=
E
y
c
=
E
0
c
.
The same thing goes for the E
z
component, which turns into B
y
as we rotate it 90 degrees. Thus we nd
b
B
= 1
c
B
= 0
d
B
=
4E
0
c
e
B
=
E
0
c
...which is marked as correct! Intuition proved useful once again!
13.9.6 Problem 6: Poynting vector
The Poynting vector associated with a plane electromagnetic wave is described as follows:

S = e
0
E
0
2
c cos
2
(ky t) y
with k = 433 rad/m and = 1.31 10
11
rad/sec. The direction of the electric eld oscillates along the x
axis.
(a) One possible vector description of the electric and magnetic eld in the plane wave associated with
this Poynting vector is

E(y, t) = E
0
x cos(ky + (1)
a
E
t)

B(y, t) = (b
B
x + c
B
y + d
B
z) cos(ky + (1)
a
B
t)
Determine the parameters in these expressions for

E and

B that correspond to the given Poynting vector.
Even more of this... Im getting a bit tired of it, to be honest.
Well, the direction of the Poynting vector is

S =

E

B, which in this case should equal y. With

E = x,
the B-eld should then be in the z direction for the cross product to work out:
x ( z) = y
So now we at least know the B-elds direction, z.
What about the cosine signs, i.e. a
E
and a
B
?
The Poynting vector should be propagating in the same direction as the wave, which means the wave
should propagate in the + y direction, as the Poynting vector does. Therefore, we want the minus signs
inside the cosines, and a
E
= a
B
= 1 causes that to happen.
217
As for the component values, that is pretty easy. We have found the direction as z, so there must be a
negative z components, and 0 x and y components. The one nonzero component is really simple, as the
magnitude of that will be the magnitude of of net vector
0
=
E
0
c
, so we nd
a
E
= 1
a
B
= 1
c
B
= 0
d
B
= 0
e
B
=
E
0
c
... with a negative to get it in the negative z direction, as mentioned above.
Could a dierent pair of electric and magnetic elds have this same Poynting vector?
Hmm, well, lets see. The direction must be

E

B = + y; can we get that using other values that

E = x
and

B = z? Well, yes: negate both the E-eld and the B-eld, and it works out... But does that really
get us a dierent pair of elds? They oscillate back and forth, and we just changedt he phase...
(The answer is that it is possible, but Im not sure the above is enough to conrm that.)
(b) What is the wavelength in meters [and period T, in seconds] of the wave?
This should be second nature by now.
=
2
k
=
2
433
m
T =
1
f
=
1
/2
=
2

=
2
1.30 10
11
s
13.9.7 Problem 7: Intensity of the sun
At the upper surface of the earths atmosphere, the time-averaged magnitude of the Poynting vector is
referred to as the solar constant and is given by
_

_
=1.35 10
3
W/m
2
.
(a) If you assume that the suns electromagnetic radiation is a plane sinusoidal wave, what is the magnitude
of the electric eld in V/m?
Digging up the denion of the time-averaged Poynting vector:
S =
1
2
E
0
B
0

0
=
1
2
E
0
2

0
c
So we can set that latter thing equal to the value we have, and solve:
1.35 10
3
=
1
2
E
0
2

0
c
2.7 10
3

0
c = E
0
2
E
0
=

2.7 10
3
410
7
3 10
8
1008.89 V/m
What is the magnitude of the magnetic eld in T?
We can nd this as B
0
= E
0
/c, knowing that equivalence. Or, we could use the other equivalence (which
is really found by using B
0
= E
0
/c anyway!):
218
1.35 10
3
=
1
2
(1008.89)(B
0
)
410
7
2.7 10
3
410
7
= (1008.89)(B
0
)
B
0
=
2.7 10
3
410
7
1008.89
3.36 10
6
T
(b) What is the time-averaged power (in Watt) radiated by the sun? The mean sun-earth distance is
r
es
= 1.5 10
11
m.
We usually do these calculations the other way around, but this shouldnt be a problem.
The intensity at the Earth is inversely proportional to the distance, as the sun radiates (roughly) equally
in all directions:
I
sun

1
4r
es
2
= 1.35 10
3
I
sun
= 1.35 10
3
4r
es
2
3.81 10
26
W
13.9.8 Problem 8: Snells law in action: ber optics!
An optical ber is a exible, transparent ber devised to transmit light between the two ends of the ber.
Complete transmission of light is achieved through total internal reection. This problem aims to calculate
the minimum index of refraction n of the optical ber necessary to obtain total internal reection for every
possible incidence angle.
(a) Express sin , where the angle is dened in the gure above, in terms of the incidence angle and
the index of refraction n of the optical ber. Evaluate this function for n = 1.55 and = 84

. Take the
index of refraction of air to be 1.
Lets begin by writing down Snells law, in two forms.
n
1
sin
1
= n
2
sin
2
sin
1
sin
2
=
n
2
n
1
Lets think this through a bit before we go further.
The beam enters and leaves (if it does leave at the top) on non-parallel sides, so it will not leave parallel
to the incident light, unless n = 1.
Looking at the triangle drawn,

2
+ + =
=

2

... since the sum of angles in a triangle must be 180 degrees, i.e. radians.
We can then attempt to nd as a function of , using Snells law:
sin
1
sin
2
=
n
2
n
1
sin
sin
= n
sin = nsin
219
sin =
sin
n
= arcsin
_
sin
n
_
So that means that
=

2
arcsin
_
sin
n
_
We want to nd the relationship between and , however, so we cant stop here!
We have another refraction here, with as the incident angle:
sin
1
sin
2
=
n
2
n
1
sin
sin
=
1
n
sin
_

2
arcsin
_
sin
n
__
sin
=
1
n
We ip the whole thing, to nd sin , and apply sin(

2
x) = cos x:
sin
sin
_

2
arcsin
_
sin
n
__ = n
sin = ncos
_
arcsin
_
sin
n
__
... which answers the rst part!
(b) The condition on n for total internal reection of all beams entering the ber is achieved when = 90

or unphysical (i.e. sin 1) for all values of . Determine the smallest value of n that satises that con-
dition.
Well, they essentially set up an equality and told us to solve it, so lets go.
ncos
_
arcsin
_
sin
n
__
1
Ugh, the cosine and arcsine are a bit annoying... This would be so easy without the n inside. Ah well.
I didnt know the exact identity, but I did know that cos(arcsin(x)) could be written in a dierent way,
using some square root. Indeed:
cos(arcsin(x)) =

1 x
2
So we can rewrite our inequality:
n

1
_
sin
n
_
2
1
n
_
1
sin
2

n
2
1
n
2
_
1
sin
2

n
2
_
1
2
n
2
sin
2
1
220
n
2
1 + sin
2

n
_
1 + sin
2

We need to nd the smallest n for which this holds true for all . The right-hand side can only move
between

1 + 0 and

1 + 1 =

2, so n =

2 is the smallest value of n for which this is always true,


which solves the problem!
13.10 Week 11
Theres no homework this week, due to the third midterm.
13.11 Week 12
13.11.1 Problem 1: Primary rainbow
Combine = 4r 2i with Snells Law to express the angle as a function of i and n only.
(a) Evaluate (i, n) for the following values (give your answers in degrees): i = 5

, n = 1.55
We use Snells law to accomplish this, as mentioned in the question. The incident light has angle i, and
the refracted light (at point A) has angle r, which is a function of i according to Snells law.
n
1
sin i = n
2
sin r
Using n
1
= 1 for air, and n
2
= n from the problem statement:
sin i = nsin r
sin r =
sin i
n
r = arcsin
_
sin i
n
_
Therefore, (i, n) is
(i, n) = 4 arcsin
_
sin i
n
_
2i
Evaluating the three angles is now easy (just keep track of using radians and degrees correctly!).
(b) For a given index of refraction n, determine the value of the incidence angle i

that maximizes , and


the correspondent value of
max
(give your answers in degrees). All answers should be positive angles.
The following identity to dierentiate the inverse function f
1
may be useful:
221
d f
1
(x)
dx
=
1
d f(z)
dz

z=f
1
(x)
Okay, so they want us to dierentiate (i, n) with respect to i and nd its maxima. The formula is clearly
intended for nding the derivative of the arcsine function. Using f
1
(x) = arcsin(x), f(x) = sin(x) (since
the arcsine is the inverse of the sine) and
d
dz
sin(z) = cos(z):
d
dx
arcsin(x) =
1
cos(arcsin(x))
That doesnt answer the entire thing, however. We still need to apply the chain rule, not to mention we
cant forget about the constant multiplier and the 2i. All in all, we have
(i, n)
i
= 4

i
arcsin
_
sin i
n
_
2
The (partial) derivative of the arcsine is then just the chain rule: the derivative of the outside with respect
to the inside, times the derivative of the inside with respect to i (in this case) is how I remember it best.
We know the rst part already, and the second part is just the the derivative of sin i times the constant
one over n inside. Keeping all this in mind:
(i, n)
i
= 4
1
cos
_
arcsin
_
sin i
n
__
cos i
n
2 =
4 cos i
ncos
_
arcsin
_
sin i
n
__ 2
Frankly, this is a bit painful, and as it is allowed, I will use Mathematica to substitute in n and set the
whole thing equal to zero. Doing so (the second part, leaving n as is), I nd
2 arctan
_
_

_
n
2
+ 2
4 n
2
2

n
2
1
(n
2
4)
2
_
_
... there has got to be a better way to solve this... But the above does give correct answers.
For n = 1.55, we nd i

= 46.8634

, while for n = 1.3 we nd i

= 61.3418

.
We then plug those values into (i, n) and nd
(46.8634

, 1.55) = 18.6158

(61.3418

, 1.3) = 47.1321

(c) What is the angle of incidence i of light on a spherical raindrop that will lead to the red in the pri-
mary rainbow (give your answers in degrees)? The index of refraction of red light in water is n
red
= 1.331.
The red in the rainbow is where
max
is for red light, so we just plug n = 1.331 into the above formula, to
and then plug that into (i, n) as above.
i

= 59.5267

(59.5267

, 1.331) = 42.3698

222
13.11.2 Problem 2: Polarization of primary rainbow
The average index of refraction of water is n=1.336. (a) What is the Brewster angle
Br
when light reects
in water o air (i.e. at point B) (give your answer in degrees)?
Easy, assuming we know the denition of the Brewster angle. It is

Br
= arctan
n
2
n
1
... with n
1
being the index of refraction of the current medium (the water), and n
2
of the medium we
reect o (the air, n
2
= 1). Thus

Br
= arctan
1
1.336
36.815

(b) Assume that i is the incident angle which gives the largest value of as found in Problem 1. Then, r
is both the refracted angle at Point A and the angle of incidence of the light ray at point B in the gure
above. What is r
Br
(give your answer in degrees)?
Well, we return to our formulae for i

and r(i, n) with n = 1.336:


i

= 59.2362

r(59.2362

, 1.336) =
arcsin(sin 59.2362

)
1.336
= 40.0291

r
Br
= 40.0291

36.815

= 3.2141

(c) Which of the following sentences do you think is most accurate about the light from the primary
rainbow?
(The choices are completely unpolarized, weakly linearly polarized and strongly linearly polarized.)
Strongly linearly polarized, as the angle of incidence for the reection is very close to the Brewster angle. If
the angles were equal, the light would be 100% polarized; instead, it will be fairly close to 100% polarized.
13.11.3 Problem 3: Glassbow
What would be the radius (
max
, in degrees) of a glass bow? The glass beads have an index of refraction
n = 1.5. We spread them out on the ground and we observe a glass bow as the sun is high in the sky.
The maximum for the angle of incidence, i

, would be i

(n) = 49.797

, and (49.797

, 1.5) = 22.8415

, and
so the answer is 22.8415

. Piece of cake, now that we have the previous problems gured out!
223
13.11.4 Problem 4: Secondary rainbow
All answers should be in degrees.
(a) Using the result of Lecture 31 Question 3, and Snells law, derive and evaluate = (i, n) for the sec-
ondary rainbow for the following values of the incidence angle i and index of refraction n: i = 6

, n = 1.35.
The result from question 3 is that
(i, r) = 2 + 2i 6r
is then given as = from the gure.
(i, r) = + 2i 6r
We use Snells law to get rid of the r:
n
1
sin i = n
2
sin r
sin i = nsin r
sin r =
sin i
n
r = arcsin
_
sin i
n
_
Nothing new there. Combining the results:
(i, n) = + 2i 6 arcsin
_
sin i
n
_
We can plug in the values and nd the rst three answers. All three answers are in the range 90

> >
180

.
(b) For a given index of refraction n, determine the value of the incidence angle i

that minimized , and


the correspondent value of
min
.
224
This is in a way the same thing as we did for the primary rainbow earlier on. Because the secondary bow
is ipped inside-out, the minimum angles are where we will nd the colors; all colors are allowed outside
(there will be while light outside the secondary).
We could dierentiate it again, but I will simply use Mathematicas FindMinimum function here, for the
various values.
For n = 1.35:
i

= 71.2981

min
= 55.249

For n = 1.5:
i

= 66.7163

min
= 86.8651

(c) What is the angle of incidence i of light on a spherical raindrop that will lead to the red (n = 1.331)
in the secondary rainbow?
This question is again exactly like the two previous, so we do the same thing for n = 1.331:
i

= 71.9073

= 50.3651

13.11.5 Problem 5: Diraction pattern


Light of a red laser (wavelength = 650 nm) goes through a narrow slit which is only a = 2 microns
wide. After the light emerges from the slit, it is visible on a screen that is L = 5 meters away from the
slit.
What is the approximate width on the screen (in cm) of the bright central spot? Here width is dened as
the distance between the center and the rst minimum.
If we have the formula we found in lecture, this is very easy.
x
n

Ln
a
m
n = 1, L = 5m, = 650 10
9
m, and the answer should be in cm, so
x
n

5 650 10
9
2 10
6

100 cm
1 m
162.5 cm
Note: this solution uses the small angle approximation, which was good enough to be marked as correct,
but it is actually more than 5% o the correct answer, due to the relatively large angle (I didnt pay
attention to that, since my rst try was marked as correct). For a more correct answer, we use the formula
we found this week, that is
x = Ltan = Ltan
_
arcsin
_

d
__
225
13.11.6 Problem 6: Optical resolution of the human eye
A car with its 2 headlights on is approaching you at night. Approximately how close (in meters) does the
car have to be to you so that you can distinguish the two headlights? Assume that the diameter of your
pupils is 6.5 mm, the distance between the two lights is 1.5 m, and the wavelength of the light is 550 nm.
Again, this should be fairly easy. There will be diraction associated with each source on its own, and
they need to be further apart, in angular distance, than the Rayleigh criterion limit of
sin = 1.220

d
where d = 6.5 mm, the diameter of the lens.
sin = 0.10323 10
3

(For such small angles, the sine of the angle and the angle in radians are essentially equivalent.)
OK, so we have that but we need the actual distance in meters.
We draw the situation on paper: an isosceles triangle (from our position (right in front of the car) to the
two headlights.
Because it is symmetric, we can split it in two, and have /2 radians between one headlight and the cars
center. One side (the long, non-hypotenuse side, adjacent to ) of that triangle is then the distance d
between us and the car, and we nd
tan

2
=
1.5/2
d
d tan

2
= 1.5/2
d =
1.5/2
tan(/2)
= 0.75 cot(/2) 14531 m
Wow, thats insanely far! This only counts the diraction limiting, though... Id be surprised if this was
actually true in a real situation, given non-ideal eyesight and all that. Hmm.
226
Chapter 14
Exam problems
14.1 Midterm 1
14.1.1 Problem 1: Electric eld on the surface of a conductor
The electric eld at point A on the surface of a conductor is 49 10
3
V/m. What is the surface charge
density (C/m
2
) at that point?
Lets call the surface charge density . The total charge for that tiny area at the surface is Q = A.
We choose a Gaussian pillbox that is located such that the end cap is located just above the surface.
Via Gausss law
EA =
A

0
A on the left side is the area of the pillbox, and A on the right side the area of the surface. Of course, we
choose them to be equal, so they cancel:
E =

0
Thus
= E
0
We know E at that point, so
= 49 10
3
8.854187 10
12
4.3385 10
7
C/m
2
We can sanity check the units of this. The electric eld is in V/m,
0
is in F/m = (C/V)/m and in
C/m
2
:
C
m
2
? =
V
m

F
m
C
m
2
? =
V
m

C
V m
C
m
2
=
C
m
2
227
14.1.2 Problem 2: Non-conducting charged planes
Two parallel non-conducting sheets of area A = 4m
2
are separated by a distance d = 0.007 m. They carry
equal but opposite surface charge densities of = 7 10
6
C/m
2
.
(a) What is the electric eld (V/m) between the plates? (ignore end eects)
We could use a Gaussian pillbox to solve this for an innite plane (since we are to ignore end eects), but
the result is so easy to remember that I wont feel guilty for using it. The electric eld between two such
plates is
E =

0
... half of which (

2
0
) is from the top plate, and half from the bottom.
So the answer for (a) is
E =
7 10
6
8.854187 10
12
790586.41974 V/m
(b) What is the total energy (Joules) contained in the electric eld between the sheets?
We can use U =
1
2
Q
2
C
here.
Q is given by A = 4 = 28 10
6
C.
C is given by the geometry, C =
A
0
d
=
4
0
0.007
5.059535 10
9
F. Thus,
U =
1
2
(28 10
6
)
2
5.059535 10
9
0.077477 J
(c) A conducting sheet with the same area as the two nonconducting sheets and with thickness h = 1 mm
(0.001 m) is inserted between the two sheets. What is the potential dierence (Volts) between the sheets
with this conducting sheet in place?
Uh oh. Well, the potential dierence without the sheet in place, since the electric eld is uniform, is simply
V = Ed 5534.105 V.
If we insert a dielectric, an induced electric eld will appear, that cancels out part of the original eld; the
net eld is reduced by a factor of , the dielectric constant for the material. However, thats for dielectrics
- and only when they completely ll the void between the two plates, too.
It turns out the answer is that since the sheet is a conductor, and the electric eld inside it is zero, we can
simply integrate over the rest of the distance. That is, we add up the potential dierence from the top
plate to the conductor, add the potential dierence across the conductor (zero), and then the potential
dierence from the bottom of the conductor to the bottom plate.
Since the eld is uniform, V = Ed. The distance d to use is then the distance between the plate minus
the thickness of the sheet (where the eld is zero):
V = E(d h)
V =

0
(d h) =
7 10
6
8.854187 10
12
(0.006 0.001) 4743.5 V
228
14.1.3 Problem 3: Incandescent bulbs
An incandescent light bulb A consumes P
A
= 21 W when it is connected to a V = 126 V battery. Another
incandescent light bulb B consumes P
B
= 253 W when it is connected to a V = 126 V battery. We place
the two light bulbs in series. You may ignore the internal resistor in the power supply and you may also
assume (for simplicity) that the resistance of the light bulbs are independent of the temperature.
We connect these two light bulbs in series to the V=126 V battery.
(a, b) How much power (in W) is used by bulb A (and B)?
Ah, nally something thats not intimidating at all to me. Exams always freak me out a bit...
Well, power is given by P = V I = I
2
R =
V
2
R
. We know P and V, so well use the last equation and solve
it for R.
P =
V
2
R
R =
V
2
P
We can now calculate the bulb resistances:
R
A
=
126
2
21
= 756
R
B
=
126
2
253
62.750988
We put them in series with the battery of 126 volts, and get a current:
I =
V
R
A
+ R
B
0.15389296 A
We can then use I
2
R to nd the power dissipated in each bulb.
P
A
= (0.15389296)
2
756 17.90438 W
P
B
= (0.15389296)
2

126
2
253
1.486135 W
All of a sudden the previously weak bulb is the brightest! Lets sanity check the answers, since this is an
exam.
The total power must equal the power put out by the battery, V I 126 0.15389296 = 19.93 watts.
Indeed it does.
(c, d) What will be the potential dierence across bulb A (and B)?
Well, V = IR and we know I and R. Easy.
V
A
= 0.15389296 756 = 116.343 V
V
B
= 0.15389296
126
2
253
9.6569 V
Again, we can sanity check: the sum must equal the batterys voltage, and it does.
229
14.1.4 Problem 4: Circuit
(a) Find all the currents in the circuit above, where R
1
= 16 Ohm, R
2
= 1 Ohm,R
3
= 3 Ohm, V
1
= 10 V,
V
2
= 1 V.
Doh. I had to insert a picture for this one. Excuse the size/image quality, if it looks bad. It looks good
here, though.
We can use loop analysis to solve for the currents in this problem. After that, we can calculate voltage
drops across the resistors.
Lets begin by working symbolically.
First, we set up an equation for the left loop, starting at point A, going clockwise around (as an arbitrary
choice). We add when we move over an increase in potential, and subtract when we move over a decrease:
I
1
R
1
V 1 I
2
R
2
+ V 2 = 0
For the right loop, starting at the middle node (where V2, R2 and R3 meet), going clockwise:
R
2
I
2
+ I
3
R
3
= 0
We now have two equations, but three unknowns (I
1
, I
2
and I
3
). We can relate the three with a KCL
equation, at the middle node mentioned above:
I
1
+ I
2
= I
3
I
1
enters from the left, and I
2
from above, while I
3
leaves to the right. Alternatively, we can sum currents
entering a node and enter the leaving current as a minus, but we would get the same equation, only with
everything on one side and set equal to zero.
This gives us the nasty answers:
I
1
=
(R
2
+ R
3
)(V
1
V
2
)
R
1
R
2
+ R
1
R
3
+ R
2
R
3
I
2
=
R
3
(V
1
V
2
)
R
1
R
2
+ R
1
R
3
+ R
2
R
3
I
3
=
R
2
(V
1
V
2
)
R
1
R
2
+ R
1
R
3
+ R
2
R
3
I
1
=
36
67
A
I
2
=
27
67
A
230
I
3
=
9
67
A
We can now calculate the voltage drops over each resistor:
V
R1
= I
1
R
1
=
36
67
16 =
576
67
V
V
R2
= I
2
R
2
=
27
67
1 =
27
67
V
V
R3
= I
3
R
3
=
9
67
3 =
27
67
V
Adding up the voltage drops along the left loop gives us
1 +
27
67
+ 10
576
67
which equals zero.
For the right loop, the sum is also zero, as V
R2
= V
R3
.
Our answers make sense!
(b) The potential at point G is V(G)=0. What is the potential at point A (in Volts)?
Well have to add up the voltage drops we encounter when moving from G to A.
V
R3
+ V
2
=
27
67
+ 1 =
94
67
V
If we add V
R1
and subtract V
1
from that, we get zero, for the outer loop, so all of our answers satisfy KVL,
an excellent sign that they are correct. At the very least, its not as sign that theyre wrong!
14.1.5 Problem 5: Point charge in a hollow conducting sphere
A very thin hollow conducting sphere has a radius of R = 0.14 m. The center of the sphere is at S. A
charge of q = 810
6
C is present at point P which is located d = 0.06 m from S. Assume that this charge
stays at this position (thus it cannot move). We draw a line from S to P and extend it beyond the radius
of the sphere.
What is the Electric eld strength (V/m) at point D which is located on the line from SP at a distance of
= 0.76 m from S (thus 0.7 m from P)?
Aha! While this looks like it could be very scary due to the non-centered charge, the fact that the shell is
conducting and a sphere makes it almost trivial!
The positive charge inside will cause the inside surface of the sphere to hold a charge of q to cancel
the E-eld inside the conductor out. This charge will be non-uniformly distributed, because the charge is
o-center.
However, the sphere was neutral to begin with, so there will be a charge of +q on the outside.
This charge will, amazingly, be uniform. We know from earlier on - or could re-calculate it using Gausss
law - that a charged sphere produces an electric eld exactly as a point charge at the center would. Thus,
we can use the good old point charge formula:
E =
Q
4
0
r
2
where Q = +q is the same as the charge inside the sphere, and r is the distance between the center (not
the surface!) and the point D.
231
E =
8 10
6
4
0
(0.76)
2
124481.34 V/m
Thats our nal answer. Since the conductor acts as a Faraday cage, the position of the charge inside is
completely irrelevant! As long as the charge is inside the cavity, the external electric eld will be the same
- after it has settled down (which happens extremely quickly after a change in position).
14.1.6 Problem 6: Charges on an equilateral triangle
A charge of q
1
= 2 10
6
C is located at corner A of an equilateral triangle. A charge q
2
= 4 10
6
C
is located at corner B and a charge of q
3
= 8 10
6
C is located at corner C. The distance AB = 0.26 m
(the other two sides of the triangle are the same).
The point P is located exactly in the middle of q1/point A and q3/point B, along the x axis, and exactly
below q2/point B.
(a) What is the x-component of the Electric eld (in V/m) at point P which is located half way between
A and C?
Before we start crunching numbers, lets think about this (which is always a good strategy)! The electric
eld from each point charge alone is always radially outwards (or inwards, for negative charges). Thus, the
x-component of the electric eld from a charge directly above you (x
charge
= x
point
) will be zero! Therefore,
we can treat this as a one-dimentional problem, and completely ignore q
2
.
Therefore, the x component is given by
E
x
=
q1
4
0
(AP)
2
+
q3
4
0
(PC)
2
AP = PC =
AC
2
=
0.26
2
= 0.13 m.
So, accounting for directions (q1 is negative, and its eld is towards the left at point P; q2 is positive, its
force is also towards the left at point P):
E
x
=
2 10
6
4
0
(0.13)
2
+
_
8 10
6
4
0
(0.13)
2
_
5.318085 10
6
V/m
(b) What is the y-component of the Electric eld (in V/m) at point P?
Similarly as before, due to the location of the point and the charges, we can now completely ignore charges
q
1
and q
2
, so this should be a piece of cake. All we need to do is to nd the distance BP, and plug it into
Coulombs law.
BP is equal to the height of the equilateral triangle. As I dont remember the formula for that, I drew it
up on paper. Via Pythagoras theorem, we get
h
2
+
_
d
2
_
2
= d
2
h
2
=
4d
2
4

d
2
4
h =

3
2
d
Thus
BP =

3
2
0.26 m
232
E
y
=
4 10
6
4
0
(0.5

3 0.26)
2
709077.12957
Note that since the eld is radially outwards from the charge, and P is located below the charge, the eld
will be downwards, and so the y-component must be negative. The distance from the charge is greater
than that of the charges in the previous problem, so we expect the magnitude of the eld to be quite a bit
lower, which it is.
(c) What is the Electric Potential (in Volts) at point P?
Work, work! Well, at least potential is not a vector, so we dont have directions and such to think about.
The potential at a point is the sum (superposition) of the potential due to each charge alone, as with
electric eld strength. We just add the potential due to each charge alone (V =
Q
4
0
r
) :
V
P
=
1
4
0
_
q1
AP
+
q2
BP
+
q3
CP
_
V
P
=
10
6
4
0
_
2
0.13
+
4
0.5

3 0.26
+
8
0.13
_
574470.61 V
(d) What is the Electric Potential Energy (conguration energy) in Joules of this system of 3 charges?
Last one. Well, for this problem anyway. Say the potential at innity is zero. We start out with nothing,
and so the potential energy is zero. We add charge q
1
, and the potential energy is still zero - but the
potential is not. We add a second charge, and the potential energy is now V
1
q
2
. We add the third charge,
and get a potential energy of V
1
q
2
from before, plus V
1
q
3
+ V
2
q
3
from adding the new charge. That is:
U =
q
1
q
2
4
0
r
12
+
q
1
q
3
4
0
r
13
+
q
2
q
3
4
0
r
23
Looking at the diagram, r
12
= r
13
= r
23
= 0.26 m, so we can factor out that, too:
U =
1
4
0
0.26
(q
1
q
2
+ q
1
q
3
+ q
2
q
3
)
U =
10
6
10
6
4
0
0.26
((2)(4) + (2)(8) + (4)(8)) 0.27661247 J
14.1.7 Problem 7: Capacitor network
Two ideal plate capacitors in air, with capacitance C
1
= 4 10
6
Farad and C
2
= 4 10
6
Farad, are
connected to a battery of voltage V = 10 Volts as shown schematically below. The plates are separated
by a distance d = 0.003 m.
(The diagram is too simple to reproduce: it has two capacitors, C
1
to the left and C
2
in the middle,
connected in parallel with a voltage source. The upper plates are connected to the positive side of the
voltage source.
(a) What is the total charge residing on the upper plate of capacitor C1 (in Coulombs)? Make sure you
have the correct sign.
Q = CV , so this one is simply
Q = 4 10
6
10 = 4 10
5
C
233
(b) What is the total charge residing on the lower plate of capacitor C2 (in Coulombs)? Make sure you
have the correct sign.
Note the lower plate, and of C
2
. The magnitude is the same as above (C
1
= C
2
, as given), but the lower
plate will have the negative charge, so we negate the above answer.
(c) What is the electric eld (direction and magnitude) between the plates of capacitor C1 (direction and
magnitude)?
V = Ed for a parallel plate capacitor, so E =
V
d
. V and d are given, so
E =
10
0.003
= 3333.33 V/m
The direction is downwards (from positive to negative).
We now disconnect the battery but leave the two capacitors connected together as shown. After discon-
necting the battery, we ll the the entire air gap of the capacitor C
2
with a dielectric ( = 4).
(d) Now, what is the total charge residing on the upper plate of capacitor C1 (in Coulombs)? Make sure
you have the correct sign.
OK, so lets think a bit. The battery is disconnected, so the potential dierence between the plates is no
longer xed. It must however be the same for both capacitors.
The charge is trapped on the capacitors, but we have two, and they are still connected, so while Q
1
+ Q
2
must clearly equal the initial total charge on them, it should be able to move around between the two.
After a very long struggle I solved this one. I rst wrote down a few things that simply can not change,
period. V
1
= V
2
due to the connection, so I simply call that V :
C
1
=
Q
1
V
C
2
=
Q
2
V
Q
1
+ Q
2
= Q
initial
= V (C
1
+ C
2
)
C
2new
= C
2
K
C
1new
= C
1
C
1
simply can not change; it is given by
A
0
d
, and none of those change, so it must be constant.
With the same argument, C
2
must increase by a factor , as the above is true for C
2
as well, except wi
insert a dielectric which always increasese capacitance by .
The voltage can and will change, but the total charge cannot. With all this in mind, we set up and solve
this system:
Q
1
+ Q
2
= Q
init
C
1
=
Q
1
V
4 C
1
=
Q
2
V
... with C
2
= 4C
1
to avoid adding a fourth line. We also have the constraint that V > 0.
This gives us a solution of
234
Q
1
=
Q
init
5
Q
2
= 4Q
1
V =
Q
1
C
1
After we plug in Q
init
= V (C
1
+ C
2
) = 8 10
5
C, we get
Q
1
=
8
5
10
5
=
1
62500
C
Q
2
=
32
5
10
5
=
1
15625
C
V =
8
5
10
5

1
4 10
6
= 4 V
They now ask us for Q
1
, Q
2
, and the electric eld between each set of plates. The electric eld is E =
V
d
for both capacitors, so we plug in V = 4 and get our answer.
14.2 Midterm 2
14.2.1 Problem 1: RL circuit
The switch in the circuit below has been open for a long, long time, R
1
= 4.0 Ohm, R
2
= 5.0 Ohm,
L = 0.045 H, V = 5 V. The internal resistance of the battery is negligibly small.
Determine the currents I
1
, I
2
, I
3
(in Ampere) in the resistors and in the self-inductor at the moment
(a) just after the switch is closed:
OK, so looking at the circuit (which is not reproduced here), if the switch has been open for a long time,
then the current everywhere is zero.
Just after closing the switch, the inductor current will still be zero, and so I
3
= 0.
Because that part of the circuit is essentially an open circuit, we have a series circuit with the battery, R
1
and R
2
:
I
1
= I
2
=
V
R
1
+ R
2
=
5
9
A
I
3
= 0
(b) a long time after the switch is closed.
A long time after, the current in the inductor will be at a maximum, limited only by the battery and R
1
.
R
2
will be completely bypassed by the inductor, which now looks like a short circuit, and so I
2
= 0, and
I
1
= I
3
.
I
1
= I
3
=
V
R
1
=
5
4
A
I
2
= 0
235
14.2.2 Problem 2: Non-conservative elds
Two voltmeters, V
right
and V
left
, each with an internal resistance of 10
6
are connected through wires of
negligible resistance (see the circuit below). The + side of both voltmeters is up as shown. A changing
magnetic eld is present in the shaded area.
At a particular moment in time V
right
reads 0.7 Volt (notice the - sign).
(a) What, at that moment, is the magnitude of the induced EMF (in Volts) in the circuit?
OK. The circuit diagram is very simple: two voltmeters, side by side, connected to each other. So both
have the plus side upwards, and then the + sides are connected at the top, and the - sides connected at
the bottom. Theres a changing magnetic ux in the middle of the loop.
Lets not fool ourselves here; we will use Faradays law, and map out the circuit based on the electric elds
in it, not based on Kirchhos rules, which dont apply here!
If we start at the bottom node, going clockwise, we rst encounter V
right
, which measures 0.7 volts,
meaning the current though it goes into the bottom and out of the top, or else it would display a positive
value. Thus, the current is going counterclockwise in the loop.
We now know the voltage across the voltmeter (which is not really a voltage drop, though, but an induced
EMF), and the resistance, so
I =
V
R
I =
0.7
10
6
= 7 10
7
A
Since the total resistance of the loop is 2 10
6
ohms, and Ohms law holds, the total EMF is
V = IR = 7 10
7
2 10
6
= 1.4 V
Question (a) wants the magnitude of this, so 1.4 volts.
(b) At that moment in time, what is the reading of V
left
? (make sure you have the correct sign!)
That voltmeter must have a a current of the same magnitude through it, but it is positive (as it goes into
the positive terminal), and so
V
left
= IR = 7 10
7
10
6
= 0.7 V
14.2.3 Problem 3: Bainbridge mass spectrometer
A Bainbridge mass spectrometer is shown in the gure. A charged particle with mass m, charge
[q[ = 4.8 10
19
C and speed v = 3 10
6
m/s enters from the bottom of the gure and traces out
the trajectory shown in the elds shown. The only electric eld E = 10 10
3
V/m is in the region where
the trajectory of the charge is a straight line.
(a) When the particle is moving through the rst (straight-line) segment of its trajectory, what is the
magnitude of the magnetic eld B in Tesla?
The particle moves towards the left in a magnetic eld pointing into the page, so it must be positively
charged. The E-eld is towards the right, and so for the particle to move in a straight line, the magnitude
of the electric force and the magnetic force must be equal.
236
The E-eld is uniform between the plates, at 10 10
3
V/m (given above); alternative usits would be N/C,
which will be more useful here. The electric force on the particle is thus
F
E
= qE = q 10 10
3
= 4.8 10
19
10 10
3
= 4.8 10
15
N
... and the direction is, then, to the right. The strength of the magnetic force on the particle must equal
this, only that the magnetic force will be towards the left. The magnetic force is given by
F
B
= q(v B) = qvB
... since the B-eld is perpendicular to v, sin = 1, and we get a nice, simple equation. Solving for the
B-eld strength, we nd
F
B
= qvB
B =
F
B
qv
Since F
B
must be equal to the electric force (in magnitude), we nd
B =
F
E
qv
=
4.8 10
15
4.8 10
19
3 10
6
= 0.0033 T
as the answer to (a).
(b) The charge hits the left wall of the spectrometer at a vertical distance h = 0.157 m above where it
entered the upper region and a horizontal distance L = 0.374 m to the left of where it entered the upper
region (see sketch). What is the radius r of the trajectory in m?
Well, rst, lets have a guess at the rough size of the answer. Via the sketch, r is slightly larger than L/2
(L is a bit smaller than the diameter of the circle the particle would trace). So if we nd r 0.2 then
that would look about right.
We know that
R =
mv
qB
... but we dont know R and we dont know B (the B-eld above is for another region of the device!), so
we cant use that as-is.
Looking at the diagram, we can simply use the Pythagorean theorem to nd r:
h
2
+ (L r)
2
= r
2
h
2
+ L
2
2Lr = 0
h
2
+ L
2
= 2Lr
r =
h
2
+ L
2
2L
We stick our values in and nd r 0.219953 m.
(c) The mass of the particle can be determined using the radius r, the charge q, the speed v, and the
magnetic eld B
0
. Using a value of B
0
= 0.2 T, evaluate the mass of the particle in kg. (Note that the
magnitude of the eld in the curved section, B
0
, is NOT the same as the magnitude in the straight section,
B, found in part a).
237
Now we can use the radius equation! We just need to solve it for m.
R =
mv
qB
0
m =
qRB
0
v
With our values, m 7.0385 10
27
kg.
14.2.4 Problem 4: RL circuit
In the circuit shown in the gure below, the switch closes at t = 0, R = 2.0 Ohm, E = 7 V, L = 0.085 H.
(a) What are the currents (in A) through the two bottom branches at t = 0
+
(just after the switch is
closed)?
As in the previous (very similar) circuit, I
1
= 0 at this point, because the inductor wont allow any current
to pass through without at least some time passing.
I
2
is then depends only upon the EMF and the resistances 2R + R = 3R, so
I
2
=
E
3R
=
7
6
A
(b) What are the currents (in A) through the two bottom branches at a much later time t ?
In this case, the current through the inductor has reached its maximum, and it is a short circuit. Therefore
I
2
= 0, and
I
1
=
E
2R
=
7
4
A
14.2.5 Problem 5: Magnetic eld of a loop
A current I = 4.9 A ows around a continuous path that consists of portions of two concentric circles of
radii a and a/2, respectively, where a = 1 cm, and two straight radial segments. The point P is at the
common center of the two circle segments.
Calculate the components of the magnetic eld (in T) at point P.
Okay. So neither of the straight wire segments contributes to the B-eld at point P, so we can neglect
them completely, and focus on calculating the B-eld due to the two arcs.
The eld ought to be in the + z direction alone, so I would expect the x- and y-components to be zero.
More on that later.
Well, lets see then. We use Biot-Savart, which says that

dB =

0
4
I
r
2
(

d r)
... where

d is a segment along the current-carrying wire, always aligned with the currents direction.
Since we need to calculate the B-eld due to two arcs, and then vectorially add the two, we will do this in
a general way, and solve it for an arc of radius r, and then substitute in r = a and r = a/2 later on.
Since an arc is part of a circle, we have the wonderful properly that [

d r[ = d, since they are always


exactly perpendicular, and the magnitude of r is 1 by denition. The direction is given by the cross
product, and will be z (out of the page) for all d.
238

d r = [

d[[ r[sin = d 1 1 = d z
So we now have

dB =

0
4
I
r
2
d z
r is the distance between all points on the wire, and point P, so we dont need to make any substitutions
there. (We will later, when we are done with the integration.)
Via arc length formulas we nd that d = rd, so

dB =

0
4
I
r
2
rd z
Thus, we can integrate:

B =
_
arc

dB =
_

0

0
4
I
r
2
rd z =

0
4
I
r
_

0
d z

B =

0
4
I
r
z =

0
I
4r
z
This resembles the B-eld in the center of a full circle - in fact, its exactly half of that! So that makes
sense.
Lets now make the substitutions, then. We have one arc with radius r = a/2 where the current goes
upwards (in the plane of the page, so + y at the center), which makes the B-eld go out of the page.
Another arc of radius r = a has the current going down at the center, which contributes to a B-eld going
into the page.
Since the current is the same, and the one with radius a/2 is closer to the point, the net B-eld will point
out of the page (the closest one will win). So we have then, that

B =

0
I
2a
z +

0
I
4a
( z) =

0
I
2a
z

0
I
4a
z

B =

0
I
4a
z
So the components, and our answers, are
B
x
= 0 T
B
y
= 0 T
B
z
=
410
7
4.9
4 0.01
0.000153938 T
14.2.6 Problem 6: Magnetic eld of a current-carrying ribbon
Consider a thin, innitely long conducting ribbon that carries a uniform current density j (current per
unit area). The width of the ribbon is w and its thickness s is extremely small (s w). P is a point in
the plane of the ribbon, at a large distance (x s) from the ribbon edge. (See the gure below)
What is the magnitude of the magnetic eld B (in T) at point P for the following values of w , j, s and
x? w = 6 cm; s = 0.1 cm; j = 1 A/m
2
and x = 16 cm.
Oh my. Well... Lets see. If we divide the cross-sectional area s w into pieces of s dr (dr to avoid
ambiguity with dw and w - r is the distance between the piece and the point P) and integrate the B-eld
due to each current j s dr , will that be enough? Or do we need to do a double integral where we consider
239
ds and dr?
Since s x, and s is described as extremely small, I will in fact assume that we ignore the thickness,
and treat it as essentially 0. That means that we wont bother calculating the B-eld due to the top layer,
the bottom layer, and an innite amount of layers in between, but indeed treat it as a 1-dimensional set
of currents, each at a dierent distance from the point P, with the closest being r = x and the furthest
being r = x + w.
Since the question only asks for the magnitude, as a single number, we wont have to worry about direc-
tions and such.
We can nd each current i (I wont call it di since we wont integrate it to nd the full current) as
i = j dA = j s dr
At this point, I started working with Biot-Savart... and realized that it wont work. I got stuck when
I thought I could treat the system as an innitely short (depth-wise) wire, and realized that I need to
integrate d in the direction in which is is 0... Doh!
Well, Amperes law can work instead, actually. To use it, we treat the system as a set of innite wires,
stacked width-wise. So each wire has area s dr, but since s is very small, we can neglect it (via the problem
description), and we get a set of innite wires of zero radius, where Amperes law applies just ne for
the really rectangular shape.
Lets then look at the B-eld due to one such wire. Clearly, it will be equal at all equidistant points
from the wire, so that we can use Amperes law and get
_

dB

d = dB2r. Since there is no changing
electric ux (and point P is in empty space), we use the simple version of Amperes law:
dB2r =
0
I
pen
What is I
pen
here? Well, we choose the Amperean circle with the radius that equals the distance to the
point P, and attach a at surface to that, so it is the full current j s dr though the tiny wire segment.
That means that we have
dB2r =
0
j s dr
dB =

0
j s dr
2r
Well, the dimensions work out to be tesla, so this looks promising.
We can now integrate the above, with r (the distance between the wire and P) x to x + w.
B =
_
x+w
x

0
j s dr
2r
=

0
j s
2
_
x+w
x
dr
r
B =

0
j s
2
ln
_
x + w
x
_
This does indeed give a green checkmark for the answer. With the given values we nd B 6.36907 10
11
tesla.
14.2.7 Problem 7: Magnetic eld of a rotating charged sphere
A spherical shell of radius R carries a uniform surface charge density (charge per unit area) . The center
of the sphere is at the origin and the shell rotates with angular velocity (in rad/sec) around the z-axis
(z = 0 at the origin). Seen from below, the sphere rotates clockwise. (See the gure below)
240
(a) Calculate the magnitude of the total current (in A) carried by the rotating sphere for the following
values of , and R:
= 5 10
4
C/m
2
, = 8 rad/s and R = 1 m.
I managed to get a correct answer on this on the test, despite using incorrect methods to solve it. Now
that Ive realized two things I did wrong, Ill try again and see whether I can nd the correct answers (not
just something close, but the actual answer!) or not.
First, we will model this as a set of circular (with an innitesimal height) current loops. That is, we cut
the sphere up in the x-y plane, such that only extremely thin slices remain. If the radius of the sphere is
R, and we call the radius of each circle r, we will nd via trigonometry that
r = Rsin
where is the polar angle (the angle from the z axis). Clearly, each circle will have a dierent radius,
depending on where it is on the sphere. Only the one centered at the origin (the great circle) will have
r = R, while ones closer to the poles will approach 0 radius.
Thats one dimension; to nd an area element, we need to multiply that by a d or something or another.
Well, we know that arc length is given by Rd, so that the circumferance of the sphere (or of the spheres
great circle, rather) is 2R; for half the sphere, its simply R, etc. So the angle in radians times the
radius.
It really isnt any harder than that, so we have
dA = 2r(Rd)
= 2R
2
sin d
As a sanity check, lets integrate that over the entire sphere, and see if we nd A = 4R
2
. If we dont, it
is clearly incorrect.
A =
_

0
2R
2
sin d
= 2R
2
_

0
sin d
= 4R
2
Excellent! Let us continue with the physics.
First o, part (a): the total current. This can be found easily: current is charge moving past a point
per unit time, and the period of each ring of current will be the same, as they are rotating together.
Therefore, the total current is the total charge, 4R
2
, divided by the period T =
2

, or
Q = 4R
2
T =
2

I =
Q
T
=
4R
2

2
= 2R
2
= 0.008 A
We can also do this the proper way to prove that the above is correct. In this case, we use the area
element dA we found, and integrate it over the surface area, while multiplying by , and dividing by the
period.
An innitesimal current di is given by
241
di =
dA
T
=
dA
2
=
(2R
2
sin d)
2
= (R
2
sin d)
Integrating that over the sphere, the total current is
I =
_

0
R
2
sin d
= R
2
_

0
sin d
= 2R
2
... which is exactly the equation we found above, so our answer of 8 mA looks correct!
(b) Calculate the magnitude of the magnetic eld B(z) (in T) that is generated by the circular current of
the rotating shell at a point P on the z-axis for the following values of ,, z and R:
(same values, plus z = 2.3 m)
Okay. Since we are considering the currents as circular loops with an innitesimal height, we can use
Biot-Savart to nd the B-eld along the axis of one such loop. We have done that before, so I will simply
refer to the result. It says that
B
z
=

0
2
IR
2
(R
2
+ z
2
)
3/2
Since we are integrating a whole set of such loops, I will refer to that as dB instead. In the above, R refers
to the radius of the circular loop, which is given by little r = Rsin in our case.
z is the distance from the center of the current loop to the point P where we nd the B-eld (lets call
that P
z
, as in the z-coordinate of point P), so we need to substitute in an expression for that distance for
z. We get
dB =

0
2
di r
2
(r
2
+ (P
z
z)
2
)
3/2
Substituting in the value for di we found, we have
dB =

0
2
(R
2
sin d) r
2
(r
2
+ (P
z
z)
2
)
3/2
Ugh, so now we have both z and as variables, plus r (which is Rsin ). Still, two is one too much. We
need to rewrite in terms of z or vice versa; I chose to use z = Rcos (conversion from cylindrical to
spherical coordinates) here. That gives us, if we also replace r = Rsin at the same time,
dB =

0
2
(R
2
sin d) (Rsin )
2
((Rsin )
2
+ (P
z
Rcos )
2
)
3/2
What a mess! We can do some of the squares, to simplify:
dB =

0
2
R
4
sin
3
d
(R
2
sin
2
+ (P
z
Rcos )
2
)
3/2
Hardly pretty, but we can actually integrate this now; we only have constants and one variable, . Of
course, doing this integration manually would be madness. Doing it symbolically, using software, would
probably also be madness. We can do a numerical integration, however, by giving the software the values
for all the constants, and telling it to integrate numerically. With Mathematica, this is done by setting
the constants, and then using the NIntegrate function.
242
Using NIntegrate, or something similar, we evaluate the following integral:
B =
_

0

0
2
R
4
sin
3
d
(R
2
sin
2
+ (P
z
Rcos )
2
)
3/2
With constants moved outside the integral:
B =

0
R
4
2
_

0
sin
3
d
(R
2
sin
2
+ (P
z
Rcos )
2
)
3/2
Solving the integral gives us B 2.75419755 10
10
T, which is correct! (Not only accepted as correct i.e.
within 5%, but as far as the answer is given, looks to be exactly correct.)
14.3 Midterm 3
14.3.1 Problem 1: RC Circuit
Here we go.
The circuit below consists of three identical resistors each with resistance R = 21 Ohm, two identical
batteries with emfs E = 17 V, and a capacitor with capacitance C = 551F. The capacitor is initially
uncharged at t = 0.
The rst sub-questions ask for i
1
, i
2
, i
3
and the capacitors voltage after a very very long time t RC.
Well, the capacitor will act as an open circuit after such a long time. Thus, the entire left loop will at as
if it is disconnected, so i
1
= 0. We can redraw the circuit with only the right loop, and nd an extremely
simple circuit:
i
1
= 0
i
2
= i
3
=
E
2R
=
17 V
2 21
0.40476 A
For the rest, we can do turn the circuit into its Thevenin equivalent.
First, we nd the open-circuit voltage for the port where the capacitor is located. That is, we remove the
capacitor, and then analyze the circuit that remains.
The left-hand loop becomes disconnected, and the voltage is decided by the right-hand loop alone.
We can either use KVL around the loop, or simply see that the circuit is a voltage divider, with R
1
=
R
2
= R, so the voltage across the middle resistor is half of E. In addition to that, we have the leftmost
voltage source, which is upside down (the way I think of it), that is, its voltage is positive downwards.
We go around the left loop (never mind that theres a hole where the capacitor was, we include that),
clockwise, starting in the bottom middle (below the center resistor):
i
3
R E + V
TH
= 0
V
TH
= E + i
3
R
V
TH
= 17 + 8.5 = 25.5 V
V
TH
= 25.5 V
Next, we need to nd the equivalent resistance into that port. We turn o the voltage sources by turning
them into shorts (current sources, if any, are turned into open circuits), and redraw the circuit. It becomes
obvious that the equivalent resistance is
R
eq
= R + (R[[R) = R + (
R
2
2R
) = 1.5R
243
So we can now analyze a much simpler circuit: a series circuit of a voltage source (25.5 volts), a resistor
(1.5 21 = 31.5 ) and the capacitor.
After a long time, clearly, the capacitors voltage will equal the batterys, or there would still be a current
running. Therefore,
V
C
(t ) = 25.5 V
(d) Assuming that the capacitor starts uncharged, how long will it take (in seconds) for the voltage across
the capacitor to reach 3/4 of its maximum value?
We can use the same circuit, and the equation
V
C
(t) = V
TH
_
1 exp(
t
R
TH
C
)
_
Set it equal to three-quarters V
TH
and solve for t...
V
TH
_
1 exp(
t
R
TH
C
)
_
=
3
4
V
TH
_
1 exp(
t
R
TH
C
)
_
=
3
4
exp(
t
R
TH
C
) =
1
4

t
R
TH
C
= ln
1
4
t = R
TH
C ln
1
4
t = R
TH
C ln 4 0.02406 s
14.3.2 Problem 2: RLC circuit
A solenoid has N = 470.0 turns, length d = 10 cm , and radius b = 0.4 cm, (b d). The solenoid is
connected via a switch, S
1
, to an ideal voltage source with electromotive force E = 5 V and a resistor with
resistance R = 49 Ohm . Assume all the self-inductance in the circuit is due to the solenoid. At time
t = 0, S
1
is closed while S
2
remains open.
(a) When a current I = 1.02 10
2
A is owing through the outer loop of the circuit (i.e. S
1
is still closed
and S
2
is still open), what is the magnitude of the magnetic eld inside the solenoid (in Tesla)?
All right. So we dont need to do any circuit analysis for this step, but we do need to calculate the B-eld
inside the solenoid (and then the inductance). We could use a known result, but I will derive it using
Amperes law.
We choose an Amperean loop, in the form of a rectangle, placed such that one side (length c) is inside
the solenoid; the other sides dimension is irrelevant, if it is long enough that the second length c side is
outside the solenoid.
The basic version of Amperes law (we wont need the displacement current term, as the change in electric
ux will be 0) tells us
_

B

d =
0
I
pen
In a solenoid, the B-eld is approximately constant inside, and approximately 0 outside. It will be com-
pletely parallel to the length c side, but completely perpendicular to the other side. Therefore, only one
244
side of the loop contributes to the left side: the one inside.
We attach a at surface to this loop, to nd the current that penetrates it. The current may/will penetrate
multiple times, since N is large. It will penetrate N times, multiplied by the ratio of the loops length to
the solenoids:
I
pen
= IN
c
d
The left-hand side gave us c for the line integrals length:
cB =
0
IN
c
d
B =
0
I
N
d
This result looks familiar, except d is usually called .
We can now answer the rst sub-question, of the B-eld:
B =
0
I
N
d
= 410
7
1.02 10
2

470
0.1
6.024 10
5
T
(b) What is the self-inductance L of the solenoid (in H)?
The denition of self-inductance is the ux, divided by the current:
L =

B
I
Which also means that

B
= LI
d
B
dt
= L
dI
dt
(keeping in mind to multiply by N to nd the total ux, as needed).
In other words, we need to nd the ux inside the inductor. To do so, we attach a surface to the inductors
loop - a very weird surface (a helical surface), but we can nd the ux easily:

B
= NAB
... where A is the cross-sectional area of the solenoid, and N of course the number of windings.

B
= N b
2

0
I
N
d
L =

B
I
= b
2

0
N
2
d
L 0.000139533 H
(c) What is the current (in A) in the circuit a very long time (t L/R) after S
1
is closed?
At that point, I = ER (since the inductor has no resistance in this model).
I =
E
R
=
5
49
0.102 A
(d) How much energy (in J) is stored in the magnetic eld of the coil a very long time (t L/R) after
S
1
is closed?
U =
1
2
LI
2

1
2
0.000139533
_
5
49
_
2
7.26432 10
7
J
245
For the next part, assume that a very long time (t L/R) after the switch S
1
was closed, the voltage
source is disconnected from the circuit by opening switch S
1
. Simultaneously, the solenoid is connected to
a capacitor of capacitance C = 851F by closing switch S
2
. Assume there is negligible resistance in this
new circuit.
(e) What is the maximum amount of charge (in Coulombs) that will appear on the capacitor?
Hmm. We will have an LC circuit, since the rest is disconnected.
Since there is stored energy to begin with (in the inductor), and no energy is lost to resistance, the circuit
should oscillate forever.
I dont see an easier way to solve this (actually, for this question, we could probably use U =
1
2
Q
2
C
, but we
need some other solution for the next quetion), so lets look at setting up the dierential equation using
Faradays law:
_

E d =
d
B
dt
First, we go around the circuit starting at the top (arbitrarily), going clockwise - in the direction of the
initial current.
The inductor doesnt contribute to the line integral, as there can be no electric eld inside it, having no
resistance.
For the capacitor, we nd V
C
= Q/C (C = Q/V , V C = Q, V = Q/C); charge will build up on the bottom
plate for this current direction, and so the E-eld is is the same direction we are moving, and
Q
C
= L
dI
dt
We can use I =
dQ
dt
:
Q
C
= L
d
2
Q
dt
2
We use the initial conditions Q(t = 0) = 0 (no charge on the capacitor) and Q

(t = 0) = I
0
(initial current)
to solve the dierential equation, and nd (I used Mathematica)
Q(t) = I
0

LC sin
_
t

LC
_
That can easily answer the question, since the sine will move between -1 and +1. The maximum charge is
Q
max
= I
0

LC =
E
R

LC
5
49

0.000139533 851 10
6
3.516 10
5
C
(f) How long does it take (in s) after S
1
is opened and S
2
is closed before the capacitor rst reaches its
maximum charge?
That would be when sin
t

LC
= 1, or when
t

LC
= /2.
t

LC
=

2
t =

LC
2
t 0.000541281 s
BTW (they didnt ask for it), the current is given by the derivative of the charge, of course, so
I(t) = I
0
cos
_
t

LC
_
246
14.3.3 Problem 3: Energy ow of a capacitor
Consider a charging capacitor made out of two identical circular conducting plates of radius a = 9 cm.
The plates are separated by a distance d = 10 mm (see gure below, note that d a). The bottom plate
carries a positive charge
Q(t) = +Q
0
_
1 +
t
T
_
with Q
0
= 2 10
6
C and T = 0.006 sec, and the top plate carries a negative charge Q(t). The current
through the wire is in the positive

k-direction. You may neglect all edge eects.
(a) Calculate the components of the electric eld (in V/m) inside the capacitor for t = 2T.
Three questions (this being one) want the answers as components in r,

and

k, so they clearly want us
to use cylindrical coordinates for this one.
As for the electric eld, that is given by
E =

0
... in the direction from the positive to the negative charge, so +

k in this case. can be found by dividing


the total charge by the plate area:
=
Q
A
=
Q
a
2
So
E =

0
=
Q

0
a
2
For the change at t = 2T, using the equation, we nd
Q(t = 2T) = Q
0
_
1 +
2T
T
_
= 3Q
0
We nd then,
E
r
= 0 V/m
E

= 0 V/m
E
k
=
Q

0
a
2
=
3Q
0

0
(0.09)
2
2.66298 10
7
V/m
... that seems insane, as it is several (not far from 9) times higher than the maximum possible E-eld in air,
before breakdown occurs... I keep getting the same answer though, after having double- and triple-checked
my calculations! Lets see...
C =
A
0
d

(0.09
2
)(8.854187 10
12
)
0.01
2.25 10
11
F
V =
Q
C
=
6 10
6
2.25 10
11
266666.67 V
Well, the results are in line with everything else. I still nd them bizarre, though!
I actually posted about this on the wiki, to ask whether unphysical values could be valid, and the answer
was yes, they didnt choose the values to avoid the breakdown limit. So the above is most likely correct
247
after all.
(b) Calculate the components of the magnetic eld B(r, t) (in Teslas) at time t = 2T inside the capacitor
at a distance r = 0.9 cm from the central axis of the capacitor.
First, as a sanity check, lets calculate the B-eld outside the wire. Using Amperes law (without the
displacement current term),
B2r =
0
I
B =

0
I
2r
B 2.222e 5 I
The current is given by the derivative of Q at t = 2T, which is constant at Q
0
/T (where T = 0.006, given
early in the problem), so I 3.333 10
4
A.
B 7.405 10
9
T
... for the B-eld outside the wire. I do believe this is equal to the answer (it turned out it wasnt!), but I
will calculate it using the displacement current term inside the plates, to do things right.
The equation we need is then
_

B

d =
0
_
I
pen
+
0
d
E
dt
_
I
pen
= 0 since we will be mid-air (or mid-vacuum). We again to around a circle with radius r = 0.9 cm, so
B2r =
0

0
d
E
dt
B =

0

0
2r
d
E
dt
OK, time to calculate the change in electric ux. First, the ux itself. Since r is much smaller than the
plate radius, we are inside the plates. Only the ux through our Amperean loop matters, not the ux
through the entire plate area.
We attach a at surface to our Amperean loop, and nd

E
= EA =

0
r
2
=
Q

0
a
2
r
2

E
=
Qr
2

0
a
2
So the bigger our loop, the stronger the ux. Makes sense. Its also inversely proportional to the plate
area, as the E-eld depends on the charge density - lower density (larger plate area) equals less ux per
area. That also makes sense.
We then need to nd the derivative of this. Well, thats extremely easy - we have a bunch of constants
multiplying Q(t), so we get the same constants times dQ/dt = I:
d
E
dt
=
Ir
2

0
a
2
And so the B-eld is
B =

0

0
2r
Ir
2

0
a
2
=

0
Ir
2a
2
248
... which matches a result we found in lecture many weeks ago.
With our values, we nd
B 7.40667 10
11
T
... which is certainly not equal to the value we calculated previously. The values are equal at the very
edge of the plates (r = a), but not inside. (This equation cant describe anything outside the plates, so
we cant check thete without recalculating.)
Interesting. Still, lets move on, as this certainly should be correct, despite not being what I expected.
The direction of the magnetic eld will be azimuthal, i.e. in the direction

. With the current going
upwards, we use the right-hand rule to nd the B-eld curling counterclockwise, as seen from above. Im
not sure about the sign, but it appears to be positive if curling counterclockwise.
(The unit vector is positive inwards, to the right of the drawing, but to the left of the

k and r vectors.
Slightly confusing, but it appears common to have azimuth positive when counterclockwise seen from
above.)
(c) Calculate the components of the Poynting vector

S(r, t) (in W/m
2
) at time t = 2T between the plates
at a distance r = 0.9 cm from the central axis of the capacitor.

S =

E

B

0
Looks like we need to do a cross product in cylindrical coordinates. The Poynting vectors direction will
be given by

S =

k

= r
because

k = r
The magnitude will be the product of the E- and B-elds, divided by
0
, as weve seen:
[

S[ =
EB

0
=
QIr
2
2
a
4

0
For our values, at r = 0.9 cm:
[

S[ =
(3 2 10
6
)(3.333 10
4
)(0.9/100)
2
2
0.09
4

0
1569.57 W/m
2
The direction is mentioned above, and the other components will be zero. Keep the direction in mind
however, as the value must be entered as negative!
(d) What is the ow of energy (in W) into the capacitor at time t = 2T?
This is the Poynting vector, times the surface area that the Poynting vector works over, 2ad (r = a at
the outer edge, so we need to take the dierent B-eld in mind and cannot use the previous numerical
values!):
dU
dt
= SA =
QIr
2
2
a
4

0
2ad

r=a
=
QId
a
2

0
88.766 W
(e) How fast is the energy stored in the electric eld changing (i.e. what is the rate of change in W)
within the capacitor at time t = 2T?
Hmm. The stored energy per unit volume is given by
249
U
E
=
1
2

0
E
2
=
1
2

0
2
U
E
=
1
2

0
U
E
=
1
2
(Q/a
2
)
2

0
U
E
=
1
2
Q
2
/
2
a
4

0
U
E
=
1
2
Q
2

2
a
4
We multiply by the volume of the capacitor, a
2
d, to nd the total stored energy:
U =
1
2
Q
2

2
a
4
a
2
d
U =
1
2
Q
2
d

0
a
2
The rate of change in energy stored is then the derivative of that:
dU
dt
=
QdI

0
a
2
For our values, we nd
dU
dt
=
(3 2 10
6
)(0.01)(3.333 10
4
)
(8.854187 10
12
)(0.09)
2
88.766 W
This answers both (d) and (e), though we calculated it as the answer to (e). Thus, the energy ow into
the capacitor is equal to the growth of the stored energy. That makes quite a bit of sense!
14.3.4 Problem 4: Electromagnet with small air gap
An electromagnet has a steel core (
M
2500) with an approximately circular cross sectional area of
6.0 cm
2
. The radius of the magnet is 12.0 cm; there is a small air gap of only 2.4 mm (see sketch). The
current through the magnets N = 140-turn coil is 20 A.
What will the magnetic eld strength be inside the air gap? Express your answer in Tesla. Assume that
the magnetic eld is azimuthal (i.e. points around the circle formed by the steel) everywhere and d the
radius of the electromagnet.
The magnetic eld lines will approximately follow the circle (which is also stated in the description), even
in the gap, since magnetic eld lines must be continuous and connect back. They dont have a starting
or ending point.
We use Amperes law around the magnet:
_

B

d =
M

0
I
pen
Choosing our Amperean loop to be a circle, concentric with radius of the magnet, and attaching a at
surface to it, we nd I
pen
= NI.
As for the line integral, we need to calculate it in two parts: in the steel, and in the air part. Due to the
massive dierence in permeability, ignoring the air part is not an option.
250
_

B

d inside the steel is roughly the B-eld times the length, so that contributes with a term B(2Rd)
(since there is no steel in that last part).
For the air, the same deal applies, except the permeability is waaay lower, and we need to adjust for that:
_

B

d = B
M
d
All in all, we have
B((2R d) + d
M
) =
M

0
NI
B =

M

0
NI
(2R d) + d
M
B 1.30287 T
An alternate way to look at Amperes law is to rearrange it:
_
1

B

d =
0
I
pen
This is of course the same thing, but now its easy to see that we need to use the dierent permeability
values in the separate line integral parts, one with a high
M
and one with
M
1.
14.3.5 Problem 5: RLC Circuit
A circuit consists of a resistor of R = 3 , a capacitor of C = 6F, and an ideal self-inductor of L = 0.06
H. All three are in series with a power supply that generates an EMF of 16 sin(t) Volt. The internal
resistance of both the power supply and the inductor are negligibly small. The system is at resonance.
(a) What is the time averaged power (in Watts) generated by the power supply?
At resonance, the circuit acts exactly as though the inductor and capacitor werent there (in steady state,
that is). The time-averaged power is thus the same as it would be if we had a voltage source-resistor circuit.
The current will be given by I(t) =
V (t)
R
(since we are at resonance), so
P = V I =
_
(16 sin t)
_
16 sin t
3
__
P =
16
2
3

sin
2
t
_
The time-average of a sine-square function is one-half, which means that
P =
16
2
6
42.667 W
This should be half of what we would nd using the DC method of P =
V
2
R
=
16
2
3
, which it indeed is!
We decrease the frequency of the power supply to a value for which the reactance (
1
C
L) becomes
equal to R; the maximum EMF remains 16 V.
(b) What now will be the time averaged power (in Watts) generated by the power supply?
We will now have to use the full equation for the RLC circuit:
I(t) =
V
0

R
2
+ X
2
sin(t )
251
where
= arctan
_
X
R
_
With X = R (because X = L
1
C
, the negative of what we were given), we nd = /4, and
therefore
I(t) =
16

3
2
+ 3
2
sin(t + /4)
I(t) =
16
3

2
sin(t + /4)
P(t) = (16 sin t)
16
3

2
sin(t + /4)
P(t) =
16
2
3

2
sin(t + /4) sin t
Now we just need to nd the time average. We can nd it by integrating the power over one period, and
then dividing by the period:
P =
1
T
_
T
0
V (t)I(t) dt
T =
2

, so
P =

2
_
T
0
16
2
3

2
sin(t + /4) sin t dt
P =
16
2

2
_
T
0
sin(t + /4) sin t dt
P =
16
2

2
_

2
_
P =
16
2
12
21.333 W
So the power is exactly half of what it were earlier. Quite a lot of work to nd a factor of one half!
14.3.6 Problem 6: Electromagnetic wave
The magnetic vector of a plane electromagnetic wave is described as follows:

B = B
m
x sin(12 rad/m z 3.60e9 rad/sec t)
where B
m
> 0. This represents the full magnetic eld, so that B
y
= B
z
= 0.
(a) What is the wavelength of the wave, in meters?
(b) What is the frequency f of the wave, in cycles per second?
(c) In which direction does this wave propagate?
OK. We can use some simple formulas for these few.
=
2
k
=
2
12
=

6
m
f =

2
=
3.6 10
9
2
Hz
252
As a sanity check, f = c (in a vacuum), which is true here.
The direction of propagation is + z, because we have a kz in the sine, and a minus t; the direction is
opposite the sign.
(d) The associated electric eld

E(x, t) can be written as

E = A
0
sin(k
x
x + k
y
y + k
z
z t) m
where k
x
, k
y
, k
z
and are all positive (or zero).
Determine m, A
0
(in V/m), k
x
, k
y
, k
z
(in rad/m) and (in rad/s), assuming B
m
= 9.73 10
7
T.
The EM wave propagates in the + z direction, and so the E-eld must do the same. Therefore k
x
= k
y
= 0.
It needs to be in phase with the B-eld, so k
z
= 12 rad/m and = 3.6 10
9
rad/s.
E
0
= B
0
c
, only that here we call them by other names, so
A
0
= B
m
c = 9.73 10
7
3 10
8
291.9 V/m
And lastly, the direction.

E

B = v must be true, so in this case,

E x = z; therefore the E-eld must
be in the y direction, so we need to set A
0
to a negative value (since there are only x, y and z choices
for the direction).
(e) What is the time-averaged Poynting ux associated with this wave, assuming B
m
= 9.73 10
7
T?
The Poynting vector is in the direction of propagation (since it is given by

S =

E

B

0
). As for the
magnitude, we will have to take the time average of the above, and nd
_

S
_
=
A
0
B
m
2
0
z
14.3.7 Problem 7: Radiation pressure on the Earth
The average energy ux in the sunlight on the Earth is
S = 1.4 10
3
W/m
2
You might need to use some of the following constants:
Distance from earth to the sun = 150 10
9
m
R
Earth
= 6.4 10
6
m
R
Sun
= 7.0 10
8
m
G = 6.67 10
11
m
3
/(kg s
2
)
M
Earth
= 5.97219 10
24
kg
M
Sun
= 1.9891 10
30
kg
(a) What force (in N) does the pressure of light exert on the Earth? Assume that all the light striking
the Earth is absorbed.
We can nd the radiation pressure by dividing the time-averaged Poynting vector by c:
253
S
c
=
1.4 10
3
3 10
8
= 4.6667 10
6
W/m
2
m/s
= 4.6667 10
6
Pa
We can then multiply this by a number , which depends upon whether the radiation goes through ( = 0
for a fully transparent material), is absorbed ( = 1 for 100% absorption) or reected ( = 2 for 100%
reection). In this case, = 1, so the number above is the one we need.
We then multiply this by the Earths cross-sectional area (as if it were a circle, facing the Sun) - I previ-
ously (during the lecture sequence) solved this by integrating it as a cross product, which gave me exactly
the 0.02 N that was marked as correct in that case, but it turned out the correct answer was found using
R
2
, and the answer was then rounded up to 0.02.
Therefore, Ill use the R
2
method here, of course.
F = pA = 4.666 10
6
(6.4 10
6
)
2
6.004 10
8
N
Wow, thats much more than I would have guessed. Well, the Earths radius is huge compared to the
other example of an asteroid with a 100 meter radius, so it makes sense. Not to mention that it goes with
the radius squared.
(b) What is the gravitational force (in N) that the Sun exerts on the Earth? (Think about how that
compares to the force due to the pressure of light. Does your answer make sense?)
Time for Newton to shine.
F
g
= G
m
1
m
2
r
2
= 6.67 10
11
(5.97219 10
24
)(1.9891 10
30
)
(150 10
9
)
2
3.52155 10
22
N
Whoa, that was also larger than I would have expected. Still, it should be way higher than the radiation
pressure, and indeed it is.
254
14.4 Final Exam
14.4.1 Problem 1: Loop in a Magnetic Field
An innite straight wire carrying a current I = 10 A owing to the right is placed above a rectangular
loop of wire with width w = 6 cm and length L = 15 cm, as shown in the gure below. The distance from
the innite wire to the closest side of the rectangle is h = 2.3 cm. The loop of wire has resistance R = 0.36
Ohm.
(a) What is the magnitude (in Tesla) of the magnetic eld due to the innite wire at the point P in the
rectangular loop, a distance r=4.7 cm from the wire (see gure).
Okay. Well, rst, we can completely ignore the rectangular loop for this rst question. All we need to
know is the current and the distance.
Using Amperes law, we can nd the B-eld at point P:
_

B

d =
0
I
B2r =
0
I
B =

0
I
2r
0.0000425532 T
The B-eld is circular. It points into the screen at point P (and at all other points below the wire), and
out of the screen above the wire.
(b) Calculate the magnitude of the magnetic ux (in Tesla m
2
) through the rectangular loop due to the
magnetic eld created by the innite wire.
Okay, we need to be careful here. The eld will be constant along the L dimension, as all points on such
a line is equidistant from the wire. However, the distance from the wire will matter greatly, of course. We
need to integrate the B-eld over the rectangle. Calling the ux , an innitetesimal amount of ux d
is given by d = BL dr, where dr is an innitesimal distance in the direction of r. The total ux is given
by the integral over the area,
=
_
S
d =
_
h+w
h

0
I
2r
L dr
We can move most things out of the integral, since the distance r is the only thing that isnt constant:
=

0
IL
2
_
h+w
h
1
r
dr =

0
IL
2
ln
_
h + w
h
_
255
3.85004 10
7
T m
2
As a quick sanity check, what happens if we just multiply the B-eld at P by the area of the rectangular
loop (i.e. treat the B-eld as constant throughout the loop)? The result should be somewhere in the same
order of magnitude as the answer we just found. Doing that, we nd 3.83 10
7
Tesla m
2
, which is quite
reassuring! It looks like our integral is correct.
(c) Suppose the current in the innite wire starts increasing in time according to I = bt, with b = 60
Amps/sec. What is the magnitude (in Amps) of the induced current in the loop? Neglect any contribution
to the magnetic ux through the loop due to the magnetic eld created by the induced current.
The magnitude of the current will be given by
[I[ =
1
R
d
B
dt
which is really just Ohms law, using Faradays law to nd the magnitude of the induced EMF (which
is equal to the rate of change of the magnetic ux through the loops surface). The magnetic eld is
linearly proportional to the current in the wire, and the magnetic ux through the loop is likewise linearly
proportional to the B-eld (and thus the current).
Weve already found (or
B
; I chose the shorter name earlier, but they are the same here):

B
=

0
IL
2
ln
_
h + w
h
_
Which variables here will change with regards to time? Assuming the loop doesnt move (h is constant),
only I can change. Therefore, when we take the time derivative, we nd
d
B
dt
=
dI
dt

0
L
2
ln
_
h + w
h
_
Weve given I = bt; b is constant, so the time derivative is
dI
dt
= b = 60 A/s. Therefore, the above
expression with
dI
dt
= b A/s is the magnitude of the induced EMF. We only need to divide that by the
resistance of the loop to nd the current:
I
ind
= b

0
L
2R
ln
_
h + w
h
_
I
ind
6.4167 10
6
A
The direction of this current will be such that it opposes the increasing ux from the wires B-eld. The
ux from the wires B-eld increases into the page, so the our current must be induced to create ux out
of the page, which means counterclockwise according to the right-hand rule.
14.4.2 Problem 2: Non-conducting innite sheet and innite parallel slab
We have an innite, non-conducting sheet of negligible thickness carrying a uniform surface charge density
+ = 2.0010
6
C/m
2
and, next to it, an innite parallel slab of thickness D = 10 cm with uniform volume
charge density = 1.60 10
5
C/m
3
(see sketch). Note that the slab has a negative charge. All charges
are xed in place and cannot move.
256
Assuming the innite sheet is in the x-y plane, and z points as shown in the gure, calculate the compo-
nents of the electric eld at the locations listed below. Since we do not ask for the direction separately,
these components could be positive or negative.
Give all of your answers in Volts/m.
(a) a distance h = 3 cm above (i.e. in the + z direction) the positively charged sheet.
Because the sheet is innite in both x and y, those components must be zero for all questions, by symmetry.
Only the z components wont be cancelled out, so only those can be nonzero.
The rst step, after guring that part out, is to nd the E-eld due to the sheet, and due to the slab (and
also due to the slab, inside the slab).
We start with the sheet; I will do the full (but quick) derivation using Gausss law as practice.
We use a cylinder (pillbox) of height 2r, centered on the charged sheet, so that r of the cylinder is above
the sheet, and r below. The area of the cylinders end caps is A.
(Pretend that the slab isnt there, since we are calculating the E-eld due to the sheet alone!)
Using the same symmetry argument as above, there can be no ux through the rest of the cylinder; only
through the end caps. We have Gausss law:
_

E

dA =
Q
encl

0
The area in question is the entire surface area of the cylinder. However, as mentioned, there will be no
ux through the side, so only the end cap area is relevant. Because the E-eld is constant over the area,
we get
E(2A) =
Q
encl

0
as the ux through the two end caps. The charge enclosed is given by Q = A, and is a given, so
E(2A) =
A

0
2E =

E =

2
0
( z) (due to the sheet alone)
The direction, since the sheet is positively charged, will be outwards. That is, upwards (+ z) above the
sheet, and downwards (- z) below the sheet.
Next up, we calculate the E-eld due to and outside the slab.
Again, we ignore that the sheet is there, since we want to calculate the E-eld due to the slab alone.
257
We use a Gaussian pillbox again, again with end-cap area A (per cap). We again center the pillbox,
and make it 2r high, such that 2r > D (the thickness). That is, the cylinder is higher than the slab and
encloses the entire thickness of the slab, so to speak.
Using the same arguments as before, we will end up here again:
E(2A) =
Q
encl

0
The enclosed charge is now given by the volume V = AD, times the charge density (charge per unit
volume) :
E(2A) =
AD

0
E =
D
2
0

E =
D
2
0
( z) (due the slab alone, outside the slab)
As the thickness decreases to where it no longer matters, this starts to look like the equation we have for
the thin sheet, which we would expect.
Finally, we do the calculation for the E-eld inside the slab, due to the slab alone.
Once again we use the pillbox, but this time we choose it such that r is much smaller than D, so that
it is completely inside the slab. We center it in the slab; the E-eld should be exactly zero at the exact
center, then grow linearly until the edge (since the charge density is constant) and then stay the same
until r , since the E-eld due to an innite plane/slab doesnt decay with distance.
Using Gausss law, we arrive at this step again:
E(2A) =
Q
encl

0
The charge enclosed is now given only by the size (volume) of our pillbox, V = 2rA (2r being the height
of the pillbox):
E(2A) =
2rA

E =
r

0
( z)
So to summarize:

E =

2
0
( z) (due to the sheet alone)

E =
D
2
0
( z) (due the slab alone, outside the slab)

E =
r

0
( z) (due to the slab alone, inside the slab)
Lets not forget that the slab is negatively charged!
Next, lets gather together expressions for the E-elds above both, inside the slab, and below both.
Above both, the E-eld due to the sheet will point upwards (it is positively charged), and the E-eld due
to the slab point downwards (it is negatively charged).
The net E-eld above is the superposition of the both, that is, the dierence between the two, in this case:
258

E = (

2
0
z) + (
D
2
0
( z))

E = (

2
0
z) (
D
2
0
z)
Does this make intuitive sense? If wins, the eld points upwards; if D wins, it points downwards.
Sounds good to me, so lets simplify it down:

E =
D
2
0
z (above the sheet and slab)
Next up, we calculate the E-eld below everything.
Below both, the sheets E-eld will point downward, and the slabs upward.

E = (

2
0
( z)) + (
D
2
0
z)

E = (
D
2
0
z) (

2
0
z)

E =
D
2
0
z (below the sheet and slab)
Again, we check to make sure that it makes sense. If the sheet wins, the net eld is now negative (down-
wards), and it is positive (upwards) if the slab wins. This indeed makes sense, given their charge polarities.
We note also that this is the opposite/negative of the E-eld above both, which also makes sense.
Almost there... We need to nd the combined E-eld inside the slab, as well. Well, lets see.
(This is really a bit of overengineering: we dont need this equation to solve it, we can just calculate it for
a particular location, since the answer should be a number, not an expression.)
There should be no sharp transitions when moving through the slab. At the very top, the E-eld should
be the sum of that of the sheet and that outside (above) the slab. At the center, the slabs E-eld is zeroed
out, and only the sheets should remain.
At the very bottom, we should have the same value as below both. And this with a linear transisition.
Lets have a look at this equation:

E =
r

0
( z) (due to the slab alone, inside the slab)
r will go from +0.5D to 0.5D as we move from top to bottom. We need to express this as d, a distance
below the sheet (above the slab), which can then go from 0 at the top, to D at the bottom (and 0.5D in
the middle).
r = 0.5D d seems to t nicely. Thus, we have

E =
(0.5D d)

0
( z) (due to the slab alone, inside the slab)
We can now also take care of the direction properly. Since the slab is negatively charged, the direction
will be downwards above the center, and upwards below the center.

E =
(0.5D d)

0
( z) (due to the slab alone, inside the slab)
We can nally calculate the net E-eld inside the slab, by simply adding to this the E-eld due to the
sheet, which is downwards here (below the sheet):

E =
(0.5D d)

0
( z) +

2
0
( z)
259

E =
(0.5D d)

0
( z)

2
0
z (inside the slab, total)
Again, a sanity check: for d = 0, the E-eld due to the slab is downwards, at it should be, as is the E-eld
due to the sheet. For d = D, at the bottom of the two, the slabs eld turns positive:
E =
D
2
0
( z)
... which is the same equation that we found earlier. So we nally have the full set of equations:

E =
D
2
0
z (above the sheet and slab)

E =
(0.5D d)

0
( z)

2
0
z (inside the slab)

E =
D
2
0
z (below the sheet and slab)
We can now plug in our values, and enter the answers as the z components, keeping all other components
zero. Keep the signs in mind!
(a) a distance h = 3 cm above (i.e. in the +z direction) the positively charged sheet.
The distance doesnt matter, but we plug in the values and nd
E
z
= 22588.2 V/m
> D so this makes sense. Also, with = D the answer is zero, which again makes sense.
(b) inside the slab at a distance d = 1 cm below (i.e. in the z direction) the positively charged sheet.
Note that d < D so that this is also a distance 9 cm above the bottom of the slab.
We plug it in again, and nd
E
z
= 185223 V/m
Both E-elds should be downward at this location, so a strong downwards value makes sense.
(c) a distance H = 22 cm below (i.e. in the z direction) the charged sheet. Note that H > D so that
this is also a distance 12 cm below the bottom of the slab.
Again, the distance doesnt matter, but we plug the values in:
E
z
= 22588.2 V/m
The answer is the negative of the one we found above the sheet, as it should be.
14.4.3 Problem 3: RLC circuit
A circuit is composed of a capacitor C = 2F, two resistors both with resistance R = 67 Ohm, an inductor
L = 0.03 H, a switch S, and a battery V = 5 V. The internal resistance of the battery can be ignored.
260
Initially, the switch S is open as in the gure above and there is no charge on the capacitor C and no
current owing through the inductor L. At t = 0 we close the switch.
Dene the current through the inductor to be positive if it ows through the inductor and then through
the resistor and therefore down in the drawing. Similarly, dene the current through the capacitor to be
positive if it ows down in the drawing.
What is the current through the inductor (in Amps) at t = 0 (i.e. just after the switch is closed)?
There are 6 questions, the above is the rst; they want the current through both inductor and capacitor,
at t = 0, two times in between, and at t , for a total of 6 questions.
Now, the ones for t = 0 and t are easy, but the others might just be quite hard, instead.
However, I noticed that for the inductor, they ask for t =
L
R
, and for the capacitor, for t = RC.
If we can treat the two as separate circuits, this would be almost trivial. Because they are in parallel with
the battery, which decides the voltage/EMF across L + R as V , and the voltage across C + R as V , I
believe that we indeed can treat them as separate RC and RL circuits.
Lets do the calculations as if they were separate. Since they are in parallel, the voltage across L + R
needs to be the same as the one across the C + R, which has a few possible implications. However, they
are also in parallel with the battery, which will always decide the voltage even if separate. With that in
mind, lets do the simple calculations for series RL and series RC circuits:
First, for the RC part:
I(t) =
V
R
exp
_

t
RC
_
Plugging in the values, we nd
I(t = 0) =
5
67
0.074627 A
I(t = 1.34 10
4
) = 0.0274537 A
I(t ) = 0 A
The values make sense. The second answer should be about the rst times 1/e (as it has decayed over
exactly one time constant), and it is.
261
For the RL part:
I(t) =
V
R
_
1 exp
_

tL
R
__
I(t = 0) = 0 A
I(t = 4.48 10
4
) = 0.0471878 A
I(t ) =
5
67
A
In this case, the second answer should be (1 1/e) times the nal one, and it is.
If these were two separate circuits, I would be very condent in these answers. As is, however, Im only
fairly condent in them.
However, considering that the 12th question on the exam, which is slightly similar, is ungraded in part
due to its diculty (it was included as a special challenge), I would assume that this one is easier, and
so I submitted these answers, and they were marked as correct.
14.4.4 Problem 4: Plane electromagnetic wave
The electric and magnetic elds of a plane electromagnetic wave are given by the formulas below, where
z and are both in meters and t is in seconds. This wave is traveling through a medium whose index of
refraction is 1.2.

E = 8.6 x sin(
2

z + 5.91 10
16
t) V/m

B =

B
0
sin(
2

z + 5.91 10
16
t) Tesla
(a) In what direction does the wave propagate?
The direction of propagation is given by the sign inside the sin function. In this case, it is z due to the
+z inside the sin.
(b) What is the magnitude and direction of

B
0
(assuming B
0
is positive)?
E
0
= B
0
v, where v is the propagation velocity. We can nd v using the refractive index, as the denition
of n (here 1.2) is that
1.2 =
c
v
1.2v = c
v =
c
1.2
Therefore, since E
0
= 8.6 V/m, and since we divide by v we ip the fraction upside down:
B
0
= 8.6
1.2
c
3.44 10
8
T
The direction must be such that

E

B = v, that is, x

B = z.
Via the right-hand rule, that means that

B = y.
(c) What is the wavelength of the wave (in meters)?
262
One way of nding the wavelength is by using f = v. This means that =
v
f
.
f is the frequency, which can be found from using f =

2
. All in all, we have
=
2v
f
2.65786 10
8
m
14.4.5 Problem 5: Loop in a magnetic eld
A rigid rectangular wire loop (above) of width W = 36.1 cm and length = 99 cm falls out of a region
with a magnetic eld. The dashed line in the gure above is at z = 0 and z is positive downward as shown.
The magnetic eld is constant with magnitude [

B[ = 0.940 T out of the page for z < 0 (above the z = 0


line) and zero for z > 0 (below the z = 0 line). The loop has mass M = 0.810 kg and total resistance
R = 1.4 . You may ignore any self-magnetic eld generated by the loop itself.
At t = 0, the rigid wire loop is at rest, the current I in the loop is 0, and the bottom of the loop is at
z = 0. At later times t, the loop carries current I(t) and is moving with a speed V (t) =
d
dz
z(t) downward,
where z(t) is the distance to the bottom edge of the loop as shown on the gure. Use g = 9.81 m/s
2
for
the acceleration due to gravity. (Note that downward acceleration is in the + z direction).
In all of these questions, assume that z(t) < so that part of the loop is still in the region with magnetic
eld.
(a) What is the total magnetic ux through the loop (in Tesla m
2
) when z(t) is 83 cm? Include only
the magnetic ux associated with the external eld

B (i.e. ignore the ux associated with the magnetic
self-eld generated by the current in the wire loop). Note that you do not need to calculate or know at
what time the loop is at this location.
Okay, lets see... We have the gravitational force pulling the loop downwards at all times.
In addition, there will be a magnetic force on the loop, due to the induced current.
The magnetic ux through the loop will constantly decrease, since it starts out 100% inside the magnetic
eld, and ends up outside it.
The eld is out of the page, so the induced current will be counterclockwise to attempt to keep the ux
through the loop constant.
Lets then look at the magnetic force on the three sides of the loop that are still in the magnetic eld.
The magnetic force on a current-carrying wire is given by I

L

B, where L is the directional vector along
which the current I ows, and

B the external B-eld.
For the top segment, the force is upwards. For the left side, it is to the left, and for the right side, to the
right. Those two cancel out by symmetry, so the magnetic force is
263

F
B
= IWB( z) N (upwards)
There is then the downwards force by gravity. Since

F = ma and a = 9.81 z m/s,

F
g
= 9.81 m/s
2
0.81 kg = 7.95 z N
We know W and B, but need to calculate I in order to be able to nd a numeric value for the upwards force.
First, lets calculate the (external) ux through the surface of the loop, which will also answer the rst
question.
The magnetic ux
B
is given by the area inside the B-eld, times the B-eld strength:

B
= BW( z(t))
For z(t) = 83 cm, this is

B
= (0.940 T)(36.1 cm)(99 cm83 cm) = 0.0542944 T m
2
(b) Using Faradays Law and Ohms Law, nd the magnitude (in Amps) of the induced current I(t) in
the bar at the time when V = 9.00 m/sec. Note that you do not need to calculate or know at what time
the loop has this speed.
We need to calculate the time derivative of the ux, which via Faradays Law gives us the induced EMF.
We then divide that by the loop resistance, which will give us the magnituded of the current.

B
= BW( z(t))

B
= BW BWz(t)
d
B
dt
= BWV (t)
The negative of this is the induced EMF. Were asked for the magnitude, so we dont really care about
minus signs, but still. We divide this by the loop resistance to nd the current:
I(t) =
BWV (t)
R
2.18147 A
... where the approximation is for V (t) = 9 m/s, as they asked for in part (b).
(c) Which way does the current ow around the loop, clockwise or counterclockwise?
Counterclockwise; we answered that long ago.
(d) What is the total magnetic force (in Newtons) on the rigid wire loop when V=9.00 m/sec? Again,
ignore any eects due to the self magnetic eld.
Now that we know the current, we can calculate this.
[

F
B
[ = IWB = (2.18147 A)(36.1/100 m)(0.940 T) = 0.74026 N
The direction is upwards, as stated earlier.
(e) What is the magnitude of the terminal speed (in m/sec) of the loop (i.e. the speed at which the loop
will be moving when it no longer accelerates)?
264
It stops accelerating when the net force is zero. That happens when the magnetic force upwards equals
the gravitational force downwards, in other words when
I(t)WB = 7.95
BWV (t)
R
WB = 7.95
B
2
W
2
V (t)
R
= 7.95
V (t) =
7.95R
B
2
W
2
Plugging in the values on the right-hand side, we nd that V (t) = 96.6552 m/s.
That looks far from a possible value, since the net acceleration must be at most 9.81 m/s
2
(due to gravity)!
That is, it must have been accelerating for about 10 seconds or so, at which point it would clearly have
fallen out of the B-eld long ago!
Still, I checked using dimensional analysis, and the result seems both sensibly calculated and in the correct
units, so I decided to use one of my three attempts, and I got green checkmarks for everything, including
the last part. Phew!
(I still believe the result to be impossible in practice, but there was a similar question on the previous
exam, where the E-eld in a capacitor was way higher than possible in air without breakdown, due to the
random values assigned to each student.)
14.4.6 Problem 6: Charged particles in a magnetic eld
Two charged particles initially are traveling in the positive x direction, each with the same speed v, and
enter a non-zero magnetic eld at the origin O. The magnetic eld for x > 0 is constant, uniform and
points out of the page. In the magnetic eld, their trajectories both curve in the same direction, but
describe semi-circles with dierent radii. The radius R
2
of the semi-circle traced out by particle 2 is bigger
than the radius R
1
of the semi-circle traced out by particle 1, and
R
2
R
1
= 4.25. Note that the drawing is
not to scale, so the ratio of radii may not be represented accurately. Assume that the particles are sent
through at dierent times so that they do not interact with each other.
265
(a) Assume that the two particles have the same mass m, but have dierent charges, q
1
and q
2
. What is
the ratio
q
2
q
1
?

The two particles must be positively charged, or the magnetic force of qv



B would have sent them in
the other direction. That wasnt asked for, however!
The radius of a charged particles trajectory in a constant magnetic eld is given by
R =
mv
qB
Since we know that
R
2
R
1
= 4.25, and m, v and B are all the same for both particles, we can substitute in
the values for R. We get a 4-level equation, which I write as two levels instead:
R
2
R
1
=
mv
q
2
B
/
mv
q
1
B
R
2
R
1
=
1
q
2
/
1
q
1
R
2
R
1
=
q
1
q
2
4.25 =
q
1
q
2
q
2
q
1
=
1
4.25
Thus, q
2
is smaller than q
1
. That makes sense, because if it were larger, it would trace out a smaller radius
than R
1
, due to the then larger magnetic force on it.
(b) How much energy did the particle with charge q
2
gain traversing the region with magnetic eld?
Well, thats an easy one. None whatsoever - the magnetic force cant change its energy, so the answer is
zero.
(c) Assume that instead of dierent charges as shown in the gure, the two particles have the same charge
q, but have dierent mass m
1
and m
2
. What is the ratio
m
2
m
1
?
R
2
R
1
=
m
2
v
qB
/
m
1
v
qB
R
2
R
1
=
m
2
m
1
m
2
m
1
= 4.25
14.4.7 Problem 7: Diraction pattern
Light from a coherent monochromatic laser of wavelength = 4.30 10
5
m is incident on two slits, both
of width a, separated by a distance d. The slits are placed a distance L away from a screen where L d
and L a. The diraction/interference intensity is plotted on the gure below as a function of sin()
(i.e. the X axis is sin() where varies from 90

to +90

).
266
Based on the plot of intensity versus sin() in the gure above,
(a) what is a in m?
Lets see... This combined diraction+interference pattern might be the weakest point of my knowledge
of the entire 8.02x material.
I recall that the interference pattern is the one with many small peaks, while the diraction pattern is
the envelope. That is, the diraction pattern (not really drawn as-is) has one large peak, from about
0.5 < sin < 0.5, with one secondary peak on each side of that.
For single-slit diraction, we have the rst minima at
sin =

a
(destructive interference, single-slit diraction)
267
The rst maximum is at the center, sin = = 0, and has a width of sin()

a
, which is the rst
minima.
For double-slit interference, we have one maximum at the center, as always, and other maxima at
sin
n
=
n
d
(constructive interference, double-slit interference)
sin
n
=
(2n + 1)
2d
(destructive interference, double-slit interference)
So the diraction causes the envelope, which has a width of (from 0 to the rst minima) which looks
to be about sin = 0.43 or something like that. So we have
0.43 =

a
a =

0.43
=
4.30 10
5
0.43
= 10
4
m
... which is incorrect! (The value for d below was correct, however.)
I tried again with sin 0.5, and got it correct! Thats a bit disappointing in a way, though: my method
was correct, but the estimation was o. Since its not marked explicitly on the graph, Im sure several
students will get this one wrong twice (there are two tries for this question) based on estimation errors alone.
As for the interference part of it all, we have the rst-order maximum at approximately sin = 0.12 (I
hope the grader allows for a wide margin of error!), so we have
0.12 =

d
d =

0.12
=
4.30 10
5
0.12
= 3.5833 10
4
m
I had another quick look in the book (since this is an open book exam; I only use it to check smaller stu,
though), to nd the horrible equation for the intensity plot. It is
I = I
0
cos
2
(d sin /)
_
sin(a sin /)
a sin /)
_
2
I plotted I/I
0
using this function, to see whether the plot was anything like this, and it looked fairly
similar... However I had as the x axis, while the book had sin , so an exact comparison wasnt possible.
Im not sure how to plot with sin as the x axis in Mathematica, so I skipped that.
14.4.8 Problem 8: Capacitor washers
A capacitor consists of two identical metal washers of inner radius a = 1.4 mm and outer radius b = 5.1
mm, separated by a distance d = 0.19 mm (note that b d). The capacitor is initially uncharged at t = 0
and we then begin to charge it using wires outside the plates that are not shown in the gure. Suppose
at time t the charge on the lower washer is Q(t) = st with s = 19C/sec and that this charge is spread
uniformly over the area of the bottom washer.
The top washer has an identical, but negative, charge Q(t), also distributed uniformly. Throughout
this problem you may assume that the electric eld is non-zero only inside the capacitor over the range
a < r < b, where r is the cylindrical radial distance.
You may also assume that, in the regions where the electric eld is non-zero, it points exactly perpendicular
to the surface of the washers (i.e. upward as shown) and has the same magnitude at all locations (but the
magnitude varies with time).
268
(There is a note that there was an error in this problem up until 14:10 GMT, and that people who tried
to solve it prior to that, but failed, now need to use the new random values shown. This didnt apply to
me, as I started on this problem at 17:20 GMT.)
(a) What is the magnitude (in Volts/m) of the electric eld for a < r < b at time t = 5sec?
Interesting! I wonder how accurate the solutions will be given all the strictly incorrect assumptions of a
constant/non-fringing E-eld, etc. Still, without those assumptions, I wouldnt have a clue how to solve
this - probably with a computer EM simulator. I digress; lets look at the problem.
Given that the E-eld is constant between the plates, and the plates have no mentioned thickness, its
clear that we are to use the result for the E-eld due to a changed innite plane here (that we derived
earlier), only that we use two of them, just as inside a regular circular or rectangular capacitor.
That is,
E =

0
We need to nd , though. First, lets nd it in terms of Q, after which we will nd it as a function of
time.
=
Q
A
, as usual. What is a? The area of a ring is given by the area of a circle with radius b, minus the
area of a circle with radius a (the hole in the middle):
A = b
2
a
2
So therefore,
=
Q
b
2
a
2
and
E =
Q

0
(b
2
a
2
)
As soon as we nd Q(t) at t = 5s, we can calculate the answer to (a).
Q(t) = st = 19
C
sec
t
Q = 9.5 10
11
C at t = 5s. We plug that in, and nd
E 142007 V/m
That sounds high, but it turns out to be less than 30 volts, given the small separation between the two
washers.
269
Oh, that was the next question!
(b) What is the magnitude (in Volts) of the electrostatic potential dierence between the washers at the
same time t = 5sec?
V = Ed
V = (142007 V/m)(0.19 10
3
m) 26.981 V
(c) What is the capacitance (in Farads) of this arrangement of conductors?
By denition,
C =
Q
V
As a symbolic answer,
C =
s t
d s t

0
(b
2
a
2
)
=
s t
0
(b
2
a
2
)
d s t
C =

0
(b
2
a
2
)
d
Very nice. Another denition of capacitance is
C =
A
0
d
... which is exactly what we found above!
We plug in the numbers, and nd
C 3.52095 10
12
F
Very tiny, as expected with almost no area, and no dielectric.
(d) What is the magnitude of the magnetic eld (in Tesla) inside the capacitor at a radius r = 3.4 mm
and at time t = 5s?
Ah, thats a fairly painful calculation for a mere sub-question!
I could look back in my notes to nd an equation for the B-eld, but I will re-derive it as I have done with
most things so far.
Well need to use the Ampere-Maxwell law:
_

B

d =
0
_
I
pen
+
0
d
E
dt
_
We choose our Amperean loop to be a circle, concentric with the washers, of course. We center it exactly
between the plates (though that should not matter), and attach a at surface to it.
Zero real current will penetrate that surface, since it is mid-air. There will however be a displacement
current, that is, a change in electric ux, as the capacitor is being charged.
The left-hand side of the equation should be easy, since we will assume the B-eld to be constant along
the circle. We simplify the equation to
B(2r) =
0

0
d
E
dt
270
B =

0

0
2r
d
E
dt
Next up, we calculate the electric ux
E
. Lets not forget that there is a big hole in the center of the
capacitor, unlike previous times weve done a similar calculation!

E
= EA =
st

0
(b
2
a
2
)
A
What is A here, then? Again, lets be careful. We have a ring; the outer radius is r, the radius of the
Amperean loop, which is smaller than b (the outer radius of the capacitor) and larger than a (the inner
radius).
The inner radius is a. Thus A = r
2
a
2
, and

E
=
st

0
(b
2
a
2
)
(r
2
a
2
)

E
=
st

0
(b
2
a
2
)
(r
2
a
2
)

E
=
st(r
2
a
2
)

0
(b
2
a
2
)
Next, we take the time derivative. The charge is Q = st, so the time derivative of that is I = s. s is given
to us in microcoulombs per second, which is the same as microamperes.
Alternatively, we can just take the derivative with respect to t of the above expression, and we get the
same result. That is,
d
E
dt
=
s(r
2
a
2
)

0
(b
2
a
2
)
We can then calculate the B-eld, since this term was the only thing missing:
B =

0

0
2r
s(r
2
a
2
)

0
(b
2
a
2
)
B =

0
s
2r
r
2
a
2
b
2
a
2
B 4.46129 10
10
(for r = 3.4 mm)
Time for a sanity check! If we use this method to calculate the B-eld for r = b, and also use
B =

0
I
2r
(B-eld for an innite wire)
... the results should be identical. We dont even need to put numbers in there, because when you set
r = b, it simplies down to that instantly! Awesome.
(e) What is the magnitude of the Poynting vector (in Watts/m
2
) at radius r = b and at time t = 5s?
The Poynting vector

S is given by

S =

E

B

0
Making the substitutions (also, keep in mind that I = s),

S =
1

0
_
Q

0
(b
2
a
2
)
__

0
I
2r
_
271

S =
s
2
t
2
0

2
r(b
2
a
2
)
But r = b (for this sub-question):

S =
s
2
t
2
0

2
b(b
2
a
2
)
We can plug the numbers in to nd
[

S[ 84.2003 W/m
2
Another sanity check: we have calculated values for E (constant) and B (at a dierent location), so if we
multiply those together and divide by
0
, we should nd something not way far o. Doing that, we get
142007 4.46 10
10
410
7
50.4 W/m
2
. A bit o, but the B-eld is greater further out, and so the Poynting
vector will be greater there as well.
As for the direction, radially inward makes a lot of sense: energy is entering the capacitor.
We can show that mathematically, however. The direction is given by the

E

B term, which turns out
to be z = r.
(e) What is the magnitude of the total energy per second (in Watts) owing into or out of the outer side
of the capacitor (i.e. at a radius r = b) at time t = 5s?
We need to nd the surface area (of an imaginary surface) of the capacitors outside. That is, the side
of a cylinder with radius b and height d.
A = 2bd
We multiply that by the Poynting vector, and nd
dU
dt
=
s
2
d t

0
(b
2
a
2
)
W (J/s)
As a number, we nd
dU
dt
5.12646 10
4
W (J/s)
14.4.9 Problem 9: Circuit
Two identical parallel plate capacitors lled with air, with capacitances C
1
= 4F and C
2
= 5F, are
connected to a battery of voltage V = 10 V (see the circuit below). The capacitor plates are separated by
a distance d
1
= 5 mm for C
1
and d
2
= 7 mm for C
2
. Note that the two plate separations may be dierent
although shown as the same size in the drawing.
(a) What is the electric eld (magnitude and direction) between the plates of each capacitor?
272
So they are identical, but dier in capacitance and plate separation? Hmm... Strange wording.
In fact, if we solve for the area, they dier in area too! How the heck are they identical, exactly?!
Ah well, I will assume that this is simply an oversight, and that the word identical should really be
removed...
Well then. We know that
E
1
=

1

0
E
2
=

2

0
... but we do not know the charge density, nor do we know the charge.
We can also nd it as
E
1
=
V
d
1
E
2
=
V
d
2
V is the same for both capacitors, as they are connected in parallel. Since there is a battery connected,
which decides V , we can just plug the numbers in at this point.
E
1
= 2000 V/m
E
2
1428.57 V/m
Both E-elds are pointed downward, since the positive plate is on top.
We ll the entire air gap of C
1
, the capacitor on the left, with a dielectric having = 4.0 while keeping
the battery connected.
(b) What now is the electric eld (direction and magnitude) between the plates of each capacitor?
We must not forget (or miss) that the battery is still connected! The potential dierence will still be V
across each capacitor; that simply cannot change!
The E-eld inside is constant at all places inside, with or without a dielectric. Some equations that always
apply are
C =
Q
V
V =
Q
C
V = Ed
C =

0
A
d

Because d is xed and V is xed, E must be xed at
E =
V
d
for both capacitors. That is
E
1
=
V
d
1
= 2000 V/m
273
E
2
=
V
d
2
1428.57 V/m
... exactly as before. This is because the battery is connected, which forces V to stay unchanged.
(c) Suppose that we disconnect the battery before lling the entire air gap of C
1
, the capacitor on
the left, with a dielectric having = 4.0. When disconnecting the battery, we leave the two capacitors
connected to each other as shown. After the dielectric is put in (and the battery remains discon-
nected), what would the electric eld be between the plates of each capacitor (direction and magnitude)?
What happens now is that V can change (but must still be the same for both capacitors), while Q on the ca-
pacitors can vary. The total change must still equal the total change that was there to begin with, however.
We know that
Q = CV
So
Q
1
= C
1
V = 4 10
5
C
Q
2
= C
2
V = 5 10
5
C
... before the dielectric is inserted! Thus, Q
initial
= 9 10
5
C; this is the total charge on both capacitors,
which must be held constant once the battery is removed.
A new set of equations that must hold true now:
Q
1
+ Q
2
= Q
initial
C
1new
=
Q
1
V
= C
1
C
2
=
Q
2
V
Because of the inserted dielectric, C
1new
= C
1
, where C
1
is the old, original capacitance.
We can combine the above equations, and solve for V :
V (C
1new
+ C
2
) = Q
initial
V =
Q
initial
C
1new
+ C
2
We now have that in terms of variables we know. Finally, the electric eld is given by
E =
V
d
so that
E =
Q
initial
d(C
1new
+ C
2
)
E =
Q
initial
d(C
1
+ C
2
)
Plugging in the numbers, keeping in mind that d varies between the capacitors:
E
1
=
6000
7
857.143 V/m
274
E
2
=
30000
49
612.245 V/m
The potential dierence between the plates is V 4.286 V after equilibrium has been reached.
14.4.10 Problem 10: An LC circuit
(a) In the LC circuit shown in the gure, the current is in the direction shown and the charges on the
capacitor have the signs shown. At this time,
a) the current is increasing and the charge on the positive plate is increasing.
b) the current is increasing and the charge on the positive plate is decreasing.
c) the current is decreasing and the charge on the positive plate is increasing.
d) the current is decreasing and the charge on the positive plate is decreasing.
e) the current and the charge on the positive plate are constant in time.
f) none of the above
Well, since Im taking the time this exam (since these are probably the last physics problems Ill solve in a
few months) to re-derive everything, lets try to derive, with some help from Mathematica, the equations
governing the LC circuit.
Starting at the upper node, we go left (along the current) and apply Faradays law. First, the left-hand
side:
_

E

d =
d
B
dt
Q
C
=
d
B
dt
Q
C
is positive, because we moved along the direction of the electric eld. The inductor doesnt have an
electric eld (we treat it as ideal), and only contributes to the right-hand side. Given that, by denition

B
= LI
Q
C
=
d(LI)
dt
L is a constant, so we move it out:
275
Q
C
= L
dI
dt
We can write it all in terms of Q, by making the substitution I =
dQ
dt
:
Q
C
= L
d
2
Q
dt
2
Thats about all we can do. We could rewrite it to have all terms on the left side, but since Im not going
to solve this dierential equation manually, it hardly matters.
I solved it using the boundary conditions Q(0) = Q
0
and Q

(0) = 0, that is: initial charge Q


0
on the
capacitor, and zero current at t = 0. The current through a capacitor is maximum while it is being
charged, but zero when the charge (and voltage across it) is at the maximum.
The solution is
Q(t) = Q
0
cos
_
t

LC
_
Based on the denition C =
Q
V
, V =
Q
C
:
V
C
(t) =
Q
0
C
cos
_
t

LC
_
Finally, the current is given by the time derivative of the charge:
I(t) =
Q
0

LC
sin
_
t

LC
_
Using
I
0
=
Q
0

LC
I(t) = I
0
sin
_
t

LC
_
Lets think about signs for a second. I recall getting this backwards previously, and I had a quick look in
the book (this is an open-book exam after all); indeed, they use I =
dQ
dt
here. Lets see if that makes
more sense.
In our equations, when Q
0
has just reached its maximum and starts decreasing, that means that, using
the signs in the gure above, the current is clockwise, which is indeed negative as shown. Ill stick with
the convention I just found, as that seems less confusing than ignoring the arrows as drawn.
OK, so with these equations in mind, is the rst question now easier to answer?
What do we know from the picture? The current is owing into the positive plate (counterclockwise).
This must mean that the charge is increasing, since that would deposit positive charge on the top plate.
That leaves options (a), (c) and (f), none of the above.
Is the current increasing or decreasing? Weve established that the charge must be increasing (on the
positive plate). If we plot the two (just plot y = cos(x) for the charge and y = sin(x) for the current)
we see that when the charge is increasing (that is, is above y = 0 and has positive slope), in such a plot
between x = and x =
3
2
, the current is decreasing.
That also makes intuitive sense, if we know how these circuits work. As the charge is zero and starts
increasing, the current is at a maximum. The charge keeps increasing, and the current decreasing, until
the charge reaches a maximum.
276
Thus, the answer is (c).
For the following questions, assume that the inductance is 0.05 H, the capacitance is 4 F, and the max-
imum magnitude of the current in the circuit is 90 milliAmps.
(b) How many times per second will the magnitude of the current have this maximum value?
Lets see. First, they ask about the magnitude of the current, so this maximum happens more than once
per cycle!
I(t) sin
_
t

LC
_
The term multiplying the sine doesnt charge the period in any way, so we can ignore it for now. All we
want to know is how often this function is either +1 or 1, which is when the currents magnitude is at a
maximum.
This is twice per period, so we can calculate the frequency, and just double that.
The angular oscillation frequency is

0
=
1

LC
We can convert this to a regular frequency in hertz using f =

2
:
f =

2
=
1
2

LC
We wanted twice this value though, so the answer is
# maxima per second = 2f =
1

LC
711.763 Hz
(c) The energy stored in the capacitor oscillates up and down as a function of time. What is the magni-
tude of the dierence (in Joules) between the largest and smallest amounts of energy ever stored in the
capacitor?
I dont like how that is phrased... It oscillates between 0 and a maximum, so I assume the answer is the
same as the maximum energy stored, then?
In an LC circuit, all energy sloshes back and forth between the inductor and capacitor. This means that
we dont have to calculate the capacitors energy, but we can instead calculate the inductors and use that
as the answer.
Doing so is easier, because they gave us I
0
= 90 mA.
U
max
=
1
2
LI
0
2
2.025 10
4
J
Lets verify that, though - this is the nal exam!
We can calculate the maximum stored energy in the capacitor using U =
1
2
Q
2
C
.
The maximum charge can be found by using the given value of I
0
.
I
0
=
Q
0

LC
Q
0
= I
0

LC 4.02492 10
5
C
Therefore,
277
U
max
=
1
2
(4.02492 10
5
)
2
4 10
6
2.025 10
4
J
Looks good!
(d) What is the time averaged power (in Watts) owing into or out of the self inductor over 3 full cycles?
Dene power into the inductor as positive.
In one cycle, a certain amount of energy will both enter and leave, which nets 0 energy per cycle. Thus
the time-average power over any amount of full cycles is zero.
14.4.11 Problem 11: Dielectric sphere
Consider a dielectric spherical shell with inner radius 1 cm and outer radius R = 15 cm. The spherical
shell is lled with dielectric with electric permittivity = 4.0 and there is a point charge q = 2.3 C
exactly at its center, which we consider the origin of our reference frame.
Calculate the direction and the magnitude of the electric eld for the following points:
(a) What is the direction [and magnitude] of the electric eld at the point whose distance from the origin
is 11 cm?
To clarify, there is a (very thick) shell (almost sphere, but there is a spherical void in the center) made
of a dielectric with the given value of , and a small spherical hole in the center, with a radius of only 1 cm.
First, using Coulombs law, the E-eld due to the point change, in a vacuum, is

E =
q
4
0
r
2
r (in vacuum)
... where r is the unit vector pointing radially outwards from the charge. The above equation is only valid
inside a vacuum, however. At the point where they ask for the E-eld, we are inside the dielectric.
A dielectric lowers the E-eld below that of the vacuum eld, by a factor . That is, inside the dielectric:

E =
q
4
0
r
2
r (in the dielectric)
Plugging in the numbers, we nd

E 427094 r V/m
(b) What is the direction [and magnitude] of the electric eld at the point whose distance from the origin
is 15 cm?
Again, plugging in the numbers, we nd E 229682 V/m... IF this is still inside the dielectric, otherwise
times that! Ugh.
I tried submitting that answer, and it came back wrong. I tried again with times the value (E 918728
V/m), and that was correct, so the intention was that 15 cm is exactly outside the dielectric, not inside
exactly on the edge.
(c) What is the total net charge (in Coulombs) contained in the dielectric (i.e. not counting the point
charge) inside a radius r of 11 cm?
This is the only question I missed on this exam, so this answer is written after looking at the stas
answers.
My rst try was zero, because charge cant ow freely inside the dielectric, but that was incorrect.
278
I then realized that induction occurs to create a negatively charged layer on the inside surface, and a
positive charged layer on the outside of the dielectric, so my second try was the negative of the charge at
the center, which was also incorrect, and left me with a wrong answer. Doh!
The correct answer is that the dielectric constant matters, such that
Q
induced
= Q
center
_
1
1

_
Since the centered charge is positive, the induced charge will, as mentioned above, be negative. In other
words,
Q
induced
1.725 10
6
C
(d) What is the total net charge (in Coulombs) contained in the dielectric (i.e. not counting the point
charge) inside a radius r of 15 cm?
This one is zero; the induced charge cancels out in the dielectric (charge cant come from nowhere; induced
negative charge means induced positive charge elsewhere).
14.4.12 Problem 12: Parallel RLC
SPECIAL CHALLENGE: Note that this problem is NOT worth any points! This problem is included as
a special challenge to allow students to test their detailed understanding on Faradays Law as applied to
circuits. You will be able to see whether your answers are correct but this problem is not worth any points.
Consider the following RLC circuit.
At t = 0 there is a charge Q
0
on the capacitor.
(a) Faradays Law for the loop on the left side in the gure above can be written as
a
1
Q + b
1
I
1
+ c
1
I
2
= d
1
dI
1
dt
+ e
1
dI
2
dt
with a
1
negative, and for the loop on the right side
a
2
Q + b
2
I
1
+ c
2
I
2
= d
2
dI
1
dt
+ e
2
dI
2
dt
with b
2
negative, where Q is the charge on the capacitor.
279
Express all the coecients above (a
1
, b
1
, c
1
, d
1
, e
1
, a
2
, b
2
, c
2
, d
2
, e
2
in terms of R, L, C and natural
constants.
This looks scary, but many of the coecients will be zero, from the looks of it.
Still, scored question or not, having solved the rest of the exam (expect for 11c which I missed), I will of
course give this a try.
To get a
1
negative, we go in the clockwise direction, as the arrow suggests. We nd

Q
C
+ I
1
R I
2
R = 0
That is,
a
1
=
1
C
b
1
= R
c
1
= R
d
1
= 0
e
1
= 0
For the right loop, we need to use Faradays law, since there is an inductor present. We also want to get
b
2
negative, so we again go clockwise for that to happen. Starting at the bottom:
I
1
R + I
2
R =
d
B
dt
The last term, the change in magnetic ux, is given by the denition of inductance, and the current I
2
through the inductor:
I
1
R + I
2
R = L
dI
2
dt
To get this result in term of the coecients, we need
a
2
= 0
b
2
= R
c
2
= R
d
2
= 0
e
2
= L
(b) What is
dQ
dt
equal to? Your answer can include only one or both of I
1
and I
2
and numbers.
The change in charge of the capacitor is equal to the current through it - whether it is positive or negative
is the only thing we really need to consider.
Say
dQ
dt
> 0, so that the charge is increasing. The way I would see this is that the current is counterclock-
wise, so that charge increases on the top plate, here marked as positive.
Therefore, the answer should be I
1
(it is opposite of the marked current).
(c) The charge on the capacitor obeys the following dierential equation:
d
2
Q
dt
2
+ a
dQ
dt
+ bQ = 0
280
Express a and b in terms of, if necessary, R, L and C.
Lets try to apply KCL for the top node, summing the current leaving (going downwards). Because the
voltage across all elements must be the same (they are in parallel), I call the voltage across each element
V
C
(t), i.e. the voltage across the capacitor.
C
dV
C
(t)
dt
+
V
C
(t)
R
+
1
L
_
t
0
V
C
(t)dt = 0
(assuming I
L
(0) = 0, or the integral needs to be from )
Because V
C
(t) =
Q(t)
C
, we can rewrite this:
C
d
dt
_
Q(t)
C
_
+
Q(t)
RC
+
1
L
_
t
0
Q(t)
C
dt = 0
Q

(t) +
Q(t)
RC
+
1
LC
_
t
0
Q(t)dt = 0
The units of each term works out to be amperes, so it looks like were still on the right track. However,
we dont want that integral, so we take the time derivative of both sides of the equation.
Q

(t) +
Q

(t)
RC
+
Q(t)
LC
= 0
Awesome, we now have the equation in the form we need! We can rewrite the derivatives using Leibniz
notation to make it even more obvious what a and b are:
d
2
Q
dt
2
+
1
RC
dQ
dt
+
1
LC
Q(t) = 0
So
a =
1
RC
b =
1
LC
(d) Now consider an RLC circuit built connecting in series a resistor R

, a capacitor C and an inductor


L; and a RLC circuit built connecting in parallel a resistor R, a capacitor C and an inductor L. Find the
expression of the resistance R

in terms of R, C, L such that the charge on the capacitor obeys the same
equation derived in (c).
Okay, so rst o, we need to derive the dierential equation for the charge on a capacitor in a series RLC
circuit.
I drew one on paper, in a simple rectangle with nothing on the left side, the resistor on top, capacitor on
the right, and inductor on the bottom - not that it matters, but knowing makes it easier to follow along.
I apply Faradays law, starting at the very left, going clockwise:
_

E

d =
d
B
dt
I(t)R

+
Q(t)
C
+ 0 = L
dI
dt
We can rewrite I(t) as Q

(t), which also goes for the right-hand side:


R

dQ
dt
+
1
C
Q(t) = L
d
2
Q(t)
dt
2
281
R

dQ
dt
+
1
C
Q(t) + L
d
2
Q(t)
dt
2
= 0
Divide by L throughout and sort, higher-order derivatives rst:
d
2
Q(t)
dt
2
+
R

L
dQ
dt
+
1
LC
Q(t) = 0
So the two equations are, for the parallel and series RLC circuits, respectively:
d
2
Q(t)
dt
2
+
1
RC
dQ
dt
+
1
LC
Q(t) = 0
d
2
Q(t)
dt
2
+
R

L
dQ
dt
+
1
LC
Q(t) = 0
What we need to do is choose R

such that we get


1
RC
multiplying the rst derivative. Looks easy enough:
R

=
L
RC
... which concludes this problem, this exam, and this entire course!
I hope you enjoyed the course as much as I did, and thanks for reading!
282

You might also like